The signaling cascade of induction and maintenance of ES cell diapause

Nutrient deficiency during pregnancy in numerous animal species can induce the state of embryonic diapause. Diapause is characterized by changes in protein and gene expression that minimize the organism’s reliance on external energy sources and ensure survival. Remarkably, the systematic changes associated with diapause appear to spare the gene expression program that supports embryonic cells’ maintenance in the pluripotent state. The phenomenon of the differentiation “freeze” during diapause can be reproduced in vitro. Mimicking nutrient deficiency by pharmacological inhibition of mTOR induces the diapause-like state in ES cells without affecting ES cell pluripotency. We discovered a connection between mTOR signaling and the chromatin-bound bromodomain and extra-terminal (BET) transcriptional regulator BRD4, showing a key role of BET-protein in the induction of diapause-like state in ES cells. mTOR inhibition rapidly and negatively impacts BRD4 binding to chromatin, which is associated with changes in gene expression that can contribute to diapause. Conversely, pharmacological inhibition of BET-protein circumvents the diapause dependence on mTOR inhibition and causes the diapause-like state. BET-repressed diapause-like ES cells retain the undifferentiated pluripotent state, which is associated with upregulation of a functionally linked group of genes encoding negative regulators of MAP kinase (MAPK) signaling and inactivation of MAP kinase. The transcriptional switch-off of MAP kinase following chronic BET inhibition imitates the transcriptional de-repression of MAP kinase negative regulators in response to mTOR inhibition. Mechanistically, suppression of mTOR or BET-protein leads to a profound decline in Capicua transcriptional repressor (CIC) at promoters of key negative regulators of MAP kinase. The discovered mTOR-BRD4 axis in the induction of diapause and the rapid transcriptional shut-off of differentiation program is likely to play a major role in the maintenance of embryonic diapause in vivo, as well as in controlling of the undifferentiated state of various types of stem cells during diapause-like metabolic dormancy.


Introduction
Embryonic diapause is a distinct organism phenotypic state that occurs in response to a limited nutrient supply to newly conceived embryos in numerous species of animals, including mice 1 .3][4][5] Remarkably, diapauseassociated translational and transcriptional changes have no evident impact on the differentiation potential of the embryos that, upon restoration of the nutrient supply, resume normal growth and give rise to healthy progeny. 1,6The mechanism by which diapause embryos can maintain the undifferentiated potential for extended periods is unknown.
Diapause in vivo can be phenotypically imitated in vitro by nutrient deprivation of ex vivo isolated mouse blastocysts 3 or cultured embryonic stem (ES) cells 3,4 .The eukaryotic cell response to nutrient supply is governed by the mTORC1/2 protein complexes 7 .
Accordingly, pharmacological suppression of mTOR imitates nutrient deprivation and causes a diapause-like state in vitro 3,4 .This diapause-inducing effect of mTOR inhibition has been mostly linked to translational changes 3,8 .We identi ed a novel link between the activity of mTOR and gene transcription via BRD4, a member of the bromodomain and extra-terminal (BET) protein family that plays a key role in gene regulation 9,10 .The mTOR-BRD4 axis acts as reinforcer of the negative effect of nutrient deprivation on gene transcription.Conversely, long-term suppression of BET circumvents diapause dependence on mTOR and induces the diapause-like state in ES cells.

Results
mTOR inhibition has a rapid and strong negative effect on BRD4 subnuclear organization and chromatin binding Treatment of ES cells with Torin1, which inhibits both mTORC1 and mTORC2 (hereafter mentioned as mTOR), alters the BRD4 organization into the supramolecular nuclear domains (BRD4 + speckles) and BRD4 association with speci c gene targets (Fig. 1).Treatment with Torin1 results in a rapid reduction in the BRD4-containing nuclear speckle numbers (Fig. 1a) without affecting the overall BRD4 expression levels (Extended Data Fig. 1a).The Torin1 effect on BRD4 speckles is within the range of the effect of the pan-bromodomain inhibitor I-BET151 (Fig. 1b).The I-BET affects the bromodomain-mediated BET binding to the acetylated histone H4 and, to minor degree, to other acetylated nuclear proteins 11 .The mTOR inhibition has no effect on histone H4 acetylation genome-wide (Extended Data Fig. 2a, b) and at the residues that support BRD4 binding (Extended Data Fig. 2c-f).This nding suggests that mTOR-mediated effect on BRD4-chromatin association and formation of the speckles may re ect changes in BRD4 or chromatin that preclude the BRD4 binding despite unaltered acetylated histone H4 levels.
In agreement with the negative impact of mTOR inhibition on BRD4 speckles, Torin1 treatment causes rapid and signi cant decline in BRD4 abundance at the promoters/ transcriptional start sites (TSS) of a large fraction (n = 2,386) of the protein coding genes expressed in the ES cells (n = 12,859) (Fig. 1c, Supplementary Table 1).The most pronounced decline of BRD4 binding is con ned to Integrated Stress Response (ISR) genes that support cells response to different environmental stresses, including nutrient deprivation 12 (Fig. 1d and Extended Data Fig. 1b).Many of the ISR genes that display a signi cant decline or even loss of BRD4, are controlled by the ATF4 transcription factor 13,14 , a key regulator of the stress response (Fig. 1d) 12,15,16 .

Decline of BRD4 binding correlates with impaired transcription of BRD4-bound genes
The decline of BRD4 levels correlates with a rapid decline in gene transcription as judged by nascent gene transcriptional rate quanti ed by using metabolic incorporation of 4-thiouridine (4sU) in newly transcribed RNA (TT-seq) 17 (Fig. 1e, left and middle panel and Extended Data Fig. 1c).Accordingly, changes in transcription precede a decline in the corresponding mRNA levels via total RNA sequencing (Fig. 1e, right panel and Extended Data Fig. 1d).Overall, a strong correlation between decline in BRD4 enrichment and transcriptional changes (Fig. 1f) suggests a possible causal role for impaired BRD4 binding in setting up a transcriptional program that supports cells response to mTOR inhibition that can lead to diapause.The ISR genes become activated at the early stages of nutrient de ciency and help cells to cope with the damaging effect of the stress [18][19][20] .It is well established that inhibition of mTOR following the initial response to nutrient de ciency helps to modulate the stress response levels by affecting translation of proteins that drive and support the stress 21,22 .Our observations show that mTOR suppression can mitigate stress response and hence promote cell survival not only via translational but also transcriptional mechanisms.The early timing of transcriptional response of ISR genes following mTOR inhibition suggests a direct link between mTOR signaling and transcription factors that drive ISR.

Pharmacological inhibition of bromodomain-dependent BET-protein binding to chromatin triggers diapause-like state
The ability of mTOR to affect BRD4 suggested the possibility of circumventing diapause dependence on mTOR inhibition by inhibiting the BET proteins directly.The exposure of ES cells to incrementally increased concentration of I-BET151, the pharmacological inhibitor of the bromodomain-containing BET proteins 23 results in the generation of ES cells, hereafter de ned as I-BET resistant (I-BETR), that can exist in the presence of I-BET at a concentration prohibitive for the bromodomain-dependent chromatin binding (Fig. 2a, Extended Data Fig. 3a, b).The I-BETR ES cells maintain the expression of naïve ES cell marker alkaline phosphatase (AP), are smaller, and grow at a considerably slower rate than the control ES cells (Fig. 2a, b).The metabolic state of I-BETR ES cells is characterized by reduced levels of oxidative phosphorylation as judged by lower basal and maximal oxygen consumption rates (Fig. 2c, upper panel).The declined oxidative phosphorylation capacity is coupled with reduced levels of glycolysis as judged by reduced basal and maximal extracellular acidi cation rates before and after adding the mitochondrial electron transport chain inhibitors (Fig. 2c, lower panel).Finally, the I-BETR cells have decreased biosynthetic activity, as indicated by reduced overall and de novo RNA and protein synthesis (Fig. 2d, e).The described features of I-BETR ES cells are consistent with the previously described phenotypes of ES cells that were rendered diapause by mTOR inhibition or nutrient deprivation.Several lines of evidence suggest that acquiring the diapause-like state by I-BETR ES cells re ects cell adaptation rather than a selection of the rare ES cell variant.First, the progression towards the high concentration of I-BET was associated with minimal cell attrition at each step of increasing I-BET concentration (Extended Data Fig. 3a, b).Second, removal of I-BET reverses fully and quickly the diapause-like phenotype of I-BETR ES cells (Fig. 2b, c).Finally, after a short-term (12-14hours) removal of I-BET, the I-BETR ES cells could generate chimeras upon injection into the C57BL/6J mice-derived blastocysts (Fig. 2f).This nding underscores the unaltered pluripotent capacity of I-BETR ES cells and is consistent with the reversibility of diapause.
Transcriptional reprogramming of I-BET resistant ES cells is consistent with diapauselike state It is well established that chronic exposure to BET inhibitors in tumor cells can promote resistance to BET inhibitors 24,25 .This resistance is associated with adaptive changes in transcriptional control of the RNA Pol II transcribed genes in a fashion that reduce reliance of BET overall or bromodomain-dependent BET binding to chromatin 26 .Additionally, the BET inhibitor-resistant tumor cells undergo reprogramming of the cell kinome that adds to the complexity of adaptive processes that confer BET inhibitor resistance 27,28 .In the ES cells, the adaptive changes to BET inhibition append the transcriptional program characteristic for the control ES cells (Fig. 3a, Supplementary Table 2).At the height of these changes are cellular processes and biochemical pathways that have been previously linked to embryonic of ES cell diapause append 3,5 (Fig. 3b, c, Supplementary Table 3, 4).mTOR inhibition leads to rapid transcriptional activation of negative regulators of MAP kinase The broad changes in gene expression and metabolism in I-BETR cells have a negative impact on gene program that support ES cell differentiation into various lineages.The expression of differentiation-promoting genes in the I-BETR ES cells was as low as in ES cells treated with the mix of GSK3β and MAP kinase inhibitors (2i) 29 (Fig. 3d).While some of the pluripotency supporting genes were down-regulated as well as in I-BETR ES cells, this effect was less uniform as compared the differentiation speci c genes (Fig. 3d).The striking similarity between the impact of 2i and the long-lasting BET inhibition suggested a possible suppression of MAP kinase signaling during diapause.Indeed, a comparative analysis of gene expression patterns between I-BETR ES cells and previously described Myc-de cient ES cells revealed negative regulators of MAP kinase (MAPK) signaling pathway dual-speci city protein phosphatase Dusp4, Dusp6, and Spry4 among commonly affected genes 30,31 .The transcriptional activation of negative regulators of MAP kinase signaling is among the earliest events that occur in response to mTOR inhibition in ES cells.Inhibition of mTOR upregulates genes encoding Dusp1, Dusp4, Dusp6 as well as members of Sprouty family signaling antagonist Spry1, Spry2 and Spry4 (Fig. 4a, Supplementary Table 5).Increased transcription of Dusp4, Dusp6 and Spry4 is accompanied by a rise of the corresponding protein levels and a concomitant decline of phosphorylated ERK (pERK) levels (Fig. 4b, upper panel, Extended Data Fig. 4a) following short-term inhibition of mTOR and maintained in the long-term mTOR inhibition induced diapause ES cells (Fig. 4b, lower panel, Extended Data Fig. 4b).

Transcriptional activation of negative regulators of MAP kinase ensures the pluripotency during diapause
The transcriptional switch-off of MAP kinase occurs also in I-BETR ES cells, and Dusp6, Dusp4 and Spry4 are upregulated in I-BETR ES cells as compared to control ES cells (Fig. 4c, d).This transcriptional upregulation is associated with a profound decline in Erk activity (Fig. 4d, e).This shut-off the MAP kinase is critical for the I-BETR ES cell maintenance at the naïve state (Fig. 4f).The siRNA-mediated knockdown of Dusp4, Dusp6, or Spry4 or the combined knockdown of all three genes led to the differentiation of I-BETR ES cells but did not substantially affect control ES cells, as measured by alkaline phosphatase expression and changes in ES cell morphology (Fig. 4f).

mTOR inhibition rapidly reduces the levels of CIC transcriptional repressor at promoters of negative regulators of MAP kinase
What mechanism can contribute to the transcriptional activation of negative regulators of MAP kinase in response to mTOR or during the diapause?In ES cells or cells of other types, the negative regulators of MAP kinase are transcriptionally repressed by Capicua transcriptional repressor (CIC) 32 .The signal-induced activation of MAP kinase leads to dissociation of the Capicua transcriptional repressor (CIC) from promoters of genes encoding negative regulators of MAP kinase followed by these genes' upregulation and inactivation of MAP kinase signaling 33 .This feedback mechanism ensures maintenance of MAP kinase signaling at physiological levels 33 .Our data show that the CIC is involved in the suppression of MAP kinase signaling in response to the diapause-induced signals.Pharmacological inhibition of mTOR in ES cells triggers a major decline in CIC occupancy at the promoters of genes encoding negative regulators of MAP kinase that coincides chronologically with the upregulation of these genes' expression (Fig. 5a, b).The CIC was absent at the promoters of Dusp4 and Dusp6 genes in the diapause-like I-BETR ES cells (Fig. 5c).These data suggest a key role of mTOR-BRD4-CIC axis in maintenance of ES differentiation during diapause.

Discussion
The described mTOR-and BET-controlled mechanism of diapause ES cell maintenance in an undifferentiated state has several implications.Our studies explain the high pluripotency potential of ES cells derived from the diapause embryos 34,35 .The metabolic dormancy is a hallmark of tissue resting stem cells or tumor stem cells [36][37][38][39] .It is possible that natural or arti cial, i.e. druginduced, suppression of BET binding to chromatin can play a de ning role in setting up the transcriptional program responsible for the dormancy during stemness and vice versa.The metabolic dormancy or diapause-like state render cell resistance to adverse environmental conditions 40 .It is plausible that chronic I-BET treatment may endow some of the differentiated cells with enhanced resistance to toxic impacts.In agreement with this model, our unpublished data (SV, TZ, AT, AS) show a major protective effect of I-BET and associated metabolic dormancy on mouse neurons in vitro and in vivo.The induction of dormant or diapause-like state and concomitant support of stemness and protection against toxic impacts suggest a possible application of BET inhibitors for the purpose of tissue protection in vitro, e.g.organ preservation, or in vivo during conditions such as neurodegeneration or conditions that affect the stem cell pool.In this respect the protective effect of BET inhibition on type I diabetes in mice 41 may re ect not only the suppression of T cell-driven in ammation but also direct enhancement of the pancreatic stem cell resistance to toxic impacts.

Generation of I-BET resistant (I-BETR) ES cells and chimerism
Parental CY2.4 albino B6 ES cells derived from B6(Cg)-Tyrc-2J/J embryos were constantly maintained on the mitomycin treated mouse embryonic broblasts (MEFs).Parental ES cells were rendered to I-BET resistant (I-BETR) with incrementally increase the concentration of I-BET151, starting from 250nM and ending with 4,000nM (4μM).The I-BETR ES cells were maintained one passage after removal of MEFs on the 0.5% gelatin treated plate before chimerism assay.After 12-14 hours of withdraw of I-BET from the culture medium (described before), the I-BETR ES cells were used for morula aggregations.
In brief, C57BL/6J inbred stock was used as embryo donors for aggregation with CY2.4 albino I-BETR ES cells and as pseudo pregnant surrogates.Embryos were collected at E2.5 from super-ovulated C57BL/6J females.Zona pellucida of embryos were removed by using acid Tyrode's solution (T1788, Sigma).I-BETR ES cell colonies were treated with 0.05% Trypsin-EDTA for getting single cell suspension and 10-20 cells was aggregated with each zona-free embryo.Recombined chimeric embryos were cultured overnight in micro drops of EmbryoMax advanced KSOM embryo medium (MR-101-D, Sigma) covered with mineral oil (9305, Fuji lm) at 37 °C in 6% CO 2 .In the next morning, morulae and/or blastocysts were transferred into the uteri of pseudo pregnant E2.5 C57BL/6J surrogates.Chimerism was judged at birth by the presence of black eyes and later by the coat pigmentation.

Analysis of STED imaging
Identi cation of the speckles: in brief, the individual .lifles were imported in the Cellpro ler pipeline, followed by splitting the images based on the different channels (STED for Brd4 staining and Picogreen for nuclear staining).The Picogreen image was used to identify the nuclei, inclusion criteria were set to be 450-pixel units diameter min.and 2000 max.Objects outside the range were discarded, and identi ed objects were labelled as "nuclei".In a next step, a gaussian lter was applied on the STED image, and the earlier identi ed nuclei were used as object to mask the foci in the gaussian ltered STED image.Next, the foci within this mask were identi ed as speckles, and the following criteria were used: typical diameter of objects min.6, max 24-pixel units; objects outside the diameter range or touching the border were discarded.The threshold strategy used was "global" and the thresholding method "otsu", three class thresholding was applied, and pixels in middle intensity were assigned to the background.
The threshold smoothing scale was 0, threshold correction factor 1.2 and lower and upper bounds on threshold were 0 and 1.No log transformation was included before thresholding, and the method to distinguish clumped objects and draw lines between clumped objects was based on intensity.The size of the smoothing lter for de-clumping was automatically calculated, and the value for local maxima that were closer than the minimum allowed distance was set to 4. The setting for ll holes in identi ed objects was set to "after both thresholding and de-clumping".Additional steps were included to measure intensity, identify macromolecular speckles, measure object size, shape, lters to select based on intensity etc., but these have no effect on the identi cation of the speckles.Determination of signi cant differences: the data were transferred from the excel le in a GraphPad Prism le, where the graphs were generated, and statistical analysis was performed.Determination of signi cant differences was done using an Ordinary One-Way ANOVA with Dunnett's multiple comparison's test, both for the Torin1 and controls (I-BET151 and d-BET6) graphs.
Western blotting 1 × 10 6 cells were collected and lysed on ice for 1 hour in 100 μl low salt (150mM NaCl) RIPA buffer (89900, Thermo Fisher Scienti c) containing 2 × Protease Inhibitor Cocktail (P8340, Sigma-Aldrich), 1 × Phosphatase Inhibitor Cocktail Set I (524624, Millipore), 331nM TSA (1406, Tocris) and 2µl of Benzonase Endonuclease (70664, Millipore).7.7 µl of 5M NaCl was add for every 100 µl of RIPA buffer to increase the salt concentration to 500mM.Cell lysate was then incubated on a spinning wheel in the cold room (4°C) for 1 hour, following with spinning down at 12,000g for 15min to separate the supernatant as the whole cell lysate.For western blotting, the whole cell lysate was mixed with 1 × NuPAGE LDS Sample Buffer (NP0007, Thermo Fisher Scienti c) and 1 × NuPAGE Sample Reducing Agent (NP0004, Thermo Fisher Scienti c) and denatured at 70°C for 10 minutes.50,000 cells derived lysate was loaded into 12% Bis-Tris Protein gels (NP0341BOX, Thermo Fisher Scienti c) and separated at 150 volts for 45-50 minutes.Protein gels then transferred onto Trans-Blot Turbo Mini 0.2 µm nitrocellulose membranes (1704158, Bio-Rad) and stained with ponceau S Staining Solution (A40000279, Thermo Fisher Scienti c) to ensure equal transfer.Membranes were sequentially incubated with primary antibodies and secondary antibodies conjugated with horseradish peroxidase antibodies.
After Incubation with SuperSignal West Dura Extended Duration Substrate (34075, Thermo Fisher Scienti c), membranes were scanned by using ChemiDoc Imaging System (12003153, Bio-Rad) and the data was analyzed by Image Lab (Version 6.1).Western blotting for the investigation of pERK1/ERK2 and pERK1/ERK2 levels in the cells were performed by using stripping method.
Membranes for pERK1/ERK2 detection was stripped with Restore PLUS Western Blot Stripping Buffer (46430, Thermo Fisher Scienti c) for 15min at room temperature and blocked for 3 hours with 5% BSA TBST.Membranes were then reused for ERK1/ERK2 detection.Western blotting for the investigation of histone acetylation levels were also performed by using stripping method.In brief, membranes for detection of the histone acetylation modi cations were stripped with Restore PLUS Western Blot Stripping Buffer (46430, Thermo Fisher Scienti c) for 15min at room temperature and blocked for 3 hours with 5% BSA TBST.
Membranes were then reused for histone H4 detection.
For HRP conjugated secondary antibodies used for western blotting, Goat polyclonal anti-Rabbit IgG (H+L) Secondary Antibody (31460, Thermo Fisher Scienti c) and Goat polyclonal anti-Mouse IgG (H+L) Secondary Antibody (31430, Thermo Fisher Scienti c) were applied in the study.

Analysis of nascent transcription or de-novo protein synthesis
Total nascent transcription (5-Ethynyl Uridine, 5-EU) or de novo translation (O-propargyl-puromycin, OPP) were assessed in ES cells using Click-iT RNA Alexa Fluor 488 Imaging Kit and Click-iT Plus OPP Alexa Fluor 647 Kit respectively according to the manufacturer's instructions (C10329 and C10458, Thermo Fisher Scienti c).For analysis nascent RNA synthesis, ES cells were incubated for 1 hour in ES medium supplemented with 0.5 mM 5-EU.To measure de novo protein synthesis, ES cells were incubated for 30 minutes in ES medium supplemented with 10 μM OPP.Cells treated with 1mg/ml of Actinomycin D (15021, Cell Signaling) for 2 hours and 50µg/ml of Cycloheximide (239763, Millipore) for 4 hours were used as the negative controls for the assays.The collected cells were xed for 15 min at room temperature in PBS supplemented with 3.7% paraformaldehyde (15710, Electron Microscopy Sciences) following by permeabilized in PBS supplemented with 1% BSA (A7906, Sigma-Aldrich) and 0.1% Saponin (SAE0073, Sigma-Aldrich) for 10 min at room temperature.The copper(I)-catalyzed alkyne-azide cycloaddition (CuAAC) was performed by using the indicated kits respectively.Samples were analyzed on the BD LSRII ow cytometry.Data were analyzed using the FlowJo software (Tree Star, Ashland, OR) and showed similar variance.
Analysis of intracellular pERK1/2 level 1 × 10 6 ES cells were collected and pelleted by spinning for 3min at 300g.Cells were resuspended in 1 ml 1 × DPBS (14190144, Thermo Fisher Scienti c).Formaldehyde (28908, Thermo Fisher Scienti c) was added to obtain a nal concentration of 4% and the cells were xed for 10 min at room temperature.Cells were then chilled on ice for 2min and washed twice with 1 ml 1× DPBS.Cells next were permeabilized by resuspending in 200 µl of 0.1% Saponin (SAE0073, Sigma-Aldrich) in 3% BSA DPBS and incubated for 15 min at room temperature.Permeabilization buffer was discarded, and cells were resuspended in 100 µl of 0.1% Saponin in 3% BSA DPBS containing 1:200 diluted Anti-pERK1/ERK2 (9101S, Cell Signaling), and incubated for 30 min at room temperature.Primary antibody was discarded, and cells were resuspended in 100 µl of 0.1% Saponin in 3% BSA DPBS containing 1:1,000 diluted Goat anti-Rabbit IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor 488 or Alexa Fluor 647 (Catalog # A-11008 or A-21244, Thermo Fisher Scienti c) and incubated for 30 min at room temperature.Cells were washed twice by using 200 µl of 0.1% Saponin in 3% BSA DPBS following by one time washing with 1 × DPBS.Cells were eventually resuspended in 100 µl of 1× DPBS and analyzed on the BD LSRII ow cytometry.Data were analyzed using the FlowJo software (Tree Star, Ashland, OR) and showed similar variance.
Quantitative RT-PCR (qRT-PCR) Indicated ES cells were freshly collected, and total RNA was extracted by using Trizol reagent (15596026, Thermo Fisher Scienti c) according to the manufacturer's instructions.RNA was then treated with RNase free DNase set (79254, Qiagen) and cDNA was synthesized using reagents supplied with the rst strand cDNA synthesis kit (11483188001, Roche).Quantitative real-time PCR was performed using SYBR Green I (04707516001, Roche) on a LightCycler 480 instrument (Roche).Primer sequences shown in Table S6.

Alkaline phosphatase (AP) staining
Alkaline Phosphatase staining was performed using the Stemgent® Alkaline Phosphatase Staining Kit II (Stemgent, 00-0055) according to the manufacturer's instructions.In detail, 5 × 105 ES cells were seeded in a well of the 6-well plate a day before the experimental day.ES medium was aspirated, and cells were washed with 2 ml of 1 × PBST (1 × DPBS with 0.05% v/v of Tween20).
Fix solution was added and incubated at room temperature for 2 minutes.Fix solution was then aspirated, and the xed cells were washed with 2 ml of 1 × PBST.1.5 ml of freshly prepared AP substrate solution was added in each well and incubated at room temperature for 12 minutes.Reaction was stopped by aspirating the AP Substrate Solution and the cells were washed twice with 2 ml of 1 × PBS.Cells were covered with 1 × PBS and directly used for imaging by using DMiL LED inverted tissue culture microscope (11521266, Leica) with Flexacam C1 camera.LAS X software was used for the data analysis.

Seahorse metabolic assays
Control or I-BETR CY2.4 albino mouse embryonic stem (ES) cells were seeded in a Seahorse XF96 Cell Culture Microplate with the density of 5 × 10 4 cells/well.Cells were cultured in medium with or without glucose overnight before the assay.For measuring OCR and ECAR rates in the I-BETR ES cells, medium without I-BET was applied to the I-BETR cells for indicated length of time.
Culture media were exchanged to Agilent Seahorse XF Base Medium (102353-100, Agilent) 1 hour prior to the assay.OCR rates were measured by Agilent Seahorse XF Cell Mito Stress Test Kit (103015-100, Agilent) under basal conditions and after the sequential addition of 1 μM Oligomycin, 0.5 μM FCCP (carbonyl cyanide p-(tri uromethoxy) phenylhydrazone), 0.5 μM Rotenone and Antimycin A for the nal concentrations according to the manufacturer's instructions.ECAR rates were measured by Agilent Seahorse XF Glycolysis Stress Test Kit (103020-100, Agilent) under basal conditions and after the sequential addition of 0.5 μM Rotenone/Antimycin A and 50 mM 2-DG for the nal concentrations according to the manufacturer's instructions.Both the OCR and ECAR assays were performed by using Agilent Seahorse XFe96 Analyzer.All values were calculated per well and normalized to cell number for each experiment.Each experiment was normalized to the average of all controls across all the experiments performed (n=3).The metrics of OCR assay were calculated according to the following equations: Basal Respiration = and Maximum Respiration = .As proton e ux from live cells comprises both glycolytic and mitochondrial-derived acidi cation, for the ECAR assay, inhibition of mitochondrial function by Rotenone and Antimycin A enables calculation of mitochondrial-associated acidi cation.Subtraction of mitochondrial acidi cation to total proton e ux rate results in glycolytic proton e ux rate.
RNAi-mediated knockdown of Dusp4, Dusp6 and Spry4 ES cells are trypsinized, counted and diluted in ES media without antibiotic to a density of 2.5 × 10 5 cell/ml. 1 × 10 5 ES cells were seeded in a 6-well plate containing feeders and incubated in at 37°C with 5% CO 2 overnight.On the next day, mix 2.5 µl of 20 uM SMARTpool siRNA oligonucleotides (Dharmacon) targeted to mouse Dusp4, Dusp6, Spry4 or a scrambled siRNA control (diluted in RNase-free buffer) with 197.5 µl of serum-free media in a microfuge tube (Tube 1) and incubate for 5 min at room temperature.In another tube, mix 4 µl of DharmaFECT® 1 with 196 µl of serum-free media (Tube 2), mix and incubate for 5 min at room temperature.Transfer the contents of Tube 1 into Tube 2, mix gently, and incubate for another 20 min at room temperature.
Remove ES media without antibiotic from the wells and add 400 µl of transfection medium to each well.After a 30 min incubation at room temperature, top up the volume to 2ml for each well by using ES media without antibiotic (~1.6ml).Mock-transfected cells were given DharmaFECT 1 transfection reagent only.Incubate the cells in a CO 2 incubator for 48 hours and targeted gene expression was assessed by qRT-PCR.The order information for SMARTpool siRNA targeting of mouse Dusp4, Dusp6 and Spry4 are as follows: L-061306-00-0005, L-040050-00-0005 and L-059172-01-0005.The order information for mock non-targeting siRNA pool is D-001810-10.

RNA-seq library preparation and analysis
Generation and sequencing of RNA-seq samples Three or four replicates were used for samples in RNA-seq assays.Freshly collected cells were used for total RNA extraction by using TRIzol reagent (Thermo Fisher Scienti c) according to the manufacturer's instructions.Samples were either spiked-in or not with RNA Spike-In Mix (4456740, Thermo Fisher Scienti c) following manufacturer's recommendations.Ribosomal and mitochondrial RNA was removed, and library preparation were performed by using 250 ng total RNA in all samples based on TruSeq Stranded Total RNA Ribo-Zero Gold library prep kit for Illumina (RS-122-2301, Illumina).Samples were sequenced at Genomics Resource Center, The Rockefeller University on Illumina NextSeq500.

Generation of Transcript Annotations
The Get Gene Annotation (GGA) pipeline 43 (https://github.com/AdelmanLab/GetGeneAnnotation_GGA; https://doi.org/10.5281/zenodo.5519927)was used to generate cell-type speci c gene annotations using PRO-seq (control) and RNA-seq (+/-1 µM Torin1) data generated in mESCs.Brie y 5' end of PRO-seq reads were used to call transcription start sites (TSSs) and assign a dominant TSS for each gene.RNA-seq data was then quanti ed by kallisto (version 0.45.1) 44and used to assign a transcript end site (TES) location for each gene.These consensus gene annotations were used for all subsequent analyses.
Duplicates were also removed using STAR.Gene counts were generated using the featureCounts function of the Rsubread package (version 2.0.1) 46using the GGA annotations described above.Per-gene normalized counts values were then quanti ed using DESeq2 (version 1.26.0) 47and differentially expressed genes identi ed.Given comparable spike-recovery values between samples, normalization was performed using DESeq2 generated size factors.

Ontology analysis
For I-BETR RNA-seq comparisons, differentially expressed genes called by DESeq2 (Adjusted p-value < 0.001 and fold change > 2) were ranked by fold change.The top 225 upregulated genes were run through EnrichR 48 (https://maayanlab.cloud/Enrichr/)and unique MSigDB Hallmark and Reactome categories passing signi cance thresholds (p-value < 0.001).The same process was repeated for the 200 most downregulated genes ranked by fold change.
TT-seq library preparation and analysis TT-seq library preparation CY2.4 albino B6 ES cells were plated in T175 tissue culture asks one day prior to treatment.On the day of treatment, Torin1 was added to a nal concentration of 1 μM to ES media.For control treatment, an equal volume of vehicle was added to media for the same length of treatment.500 mM 4sU (T4509, Millipore) was added to the samples exactly at 10min to the end of the given length of treatments.Cells were rinsed with room-temperature PBS and harvested using trypsin and quenched with cold DMEM + 10% FBS.After an additional wash with PBS, cells were counted and resuspended in 1 mL Trizol per 2 × 10 6 cells.Samples were frozen immediately at -80°C until all the samples were ready for the downstream steps.
Prior to addition of chloroform to the lysates, samples were spiked-in with 5% cell counts based 4sU-labeled Drosophila S2 Trizol lysate (cells labeled with 4sU for 2 h and resuspended a concentration of 10 million cells/ml in Trizol).RNA was then isolated per the manufacturer's protocol.The aqueous phase was precipitated by addition of 2.5 volumes of 100% ethanol with 1.4mM DTT, incubation at -20°C for 2h.Pellets were collected by centrifugation at 20,000g for 30 min at 4°C and washed twice with 500 mL of 75% ethanol before resuspension in 180 mL of nuclease free water.Aliquots were removed for quanti cation by spectrophotometry and analysis of RNA integrity by Agilent TapeStation 4200 using RNA high sensitivity tapes.Samples with RIN > 9.5 were used for further processing.
Total RNA from previous step was treated to remove residual DNA by using RNeasy Micro Kit (Qiagen, 74004) combined with Ampli cation Grade DNase I (18068015, Invitrogen) and Superase-In RNase inhibitor (AM2694, Thermo Fisher Scienti c).In detail, 40 µl of 1M DTT was added into 1mL of RLT buffer prior to the procedures.200 µl of RNA sample was mix with 700 µl of RLT buffer and 500 µl of 100% ethanol.Samples were loaded onto a MinElute column and spun for 1min at 3,500g.700 µl of RW1 was then added and spun the column for 2min at 14,000g. 5 µl of DNase I and 35 µl of RDD mixture was then added onto the column and incubate at room temperature for 30 min.After incubation, 660µl of RW1 was added and column was spun for 1min at 14,000g.Wash the column with one time of 500 µl of RW1 buffer followed with one time of 500 µl of RPE buffer.Next, 500 µl of 80% ethanol was added and then spun the column for 2min at 14,000g.Repeat the last step once more and open the lid of the tubes for 5min at room temperature.30 µl of RNase free water was added onto the column and incubate at room temperature for 2min and then spun at 200g for 2min.Add another 22 µl of RNase free water onto the column and repeat the same incubation following with spinning.Eventually the RNA samples were eluted by spun down for 2min at 14,000g and the concentration and quality of the samples were detected by Nanodrop and TapeStation respectively.Samples with RIN > 9.5 were used for further processing.
RNA samples were then chemically fragmented by addition of 20 mL cold 5 × fragmentation solution (375 mM Tris-HCl (pH 8.3), 562.5 mM KCl, 22.5 mM MgCl 2 ) and incubation at 94°C for 3.0 minutes.At the end of the fragmentation time, RNA was placed immediately on ice and 25 mL of cold 250 mM EDTA was added.RNA was precipitated by addition of 1/10 volume of 5 M NaCl, 2.5 volumes of 100% ethanol and incubation at -20°C overnight.RNA samples were pelleted, washed, quanti ed, and analyzed again as described above.The TapeStation results should show bulk of the signal at 200-1,000nt with the peak size around 450-600nt.
Fragmented RNA was biotinylated as described in Duffy et al. (2015) with the following modi cations: the biotinylation reaction was performed in a total volume of 200 µl and allowed to incubate for 45 min in the dark.Excess biotin was removed using chloroform:isoamyl alcohol and Phase-Lock-Gel (Heavy/High density) tube (Qiagen, 129056) were used to separate organic and aqueous phases.Biotinylated RNA was resuspended in 100 µl of nuclease-free water and aliquots taken to use as the total RNA input fraction.In parallel, Dynabeads MyOne Streptavidin C1 (Invitrogen, 65001) were prepared for binding to render them RNasefree: for each sample, 25 µl of beads were used and treated in batch to render them RNase free.The beads were incubated 10 min in the decon solution (100 mM NaOH, 50 mM NaCl) and placed on a magnetic stand.Then the beads were washed as follows: resuspending the beads fully for each wash, twice with 500 µl of 100 mM NaCl, twice with 500 µl of 1 × TTseq wash solution (100 mM Tris-HCl (pH 7.4), 10mM EDTA, 1 M NaCl, 0.05% Tween 20 in nuclease free water to which 1 µl SuperaseIN RNase Inhibitor (AM2694, Thermo Fisher Scienti c) per 1mL solution is added prior to use), once in 500 µl of 0.3 × TTseq wash solution, and nally resuspended in 52 µl /sample of 0.3 × TTseq wash solution and 1 µl/sample Superase-In RNase inhibitor (Invitrogen, AM2696).
Biotinylated RNA was heated at 65°C for 5min, placed on ice for 2 min, and mixed with 50 µl of prepared beads.Samples were STED imaging of BRD4 cells were cultured and plated on Nunc Lab Tek II Chamber Slides (Thermo Scienti c, 154534PK).Cells were xed with 4% paraformaldehyde (Electron Microscopy Sciences, 15710) in DPBS for 15 minutes at RT and then washed twice with DPBS.Cells then were permeabilized with DPBS + 0.2% Triton (PBST) and blocked with 2% normal goat serum (Thermo Scienti c, 50197Z) in PBST for 1 hour at RT. Cells were directly incubated with primary antibody BRD4 (1:500, Ab84776, Abcam) in 2% normal goat serum in PBST overnight at 4°C.Cells were washed twice with PBST and incubated with Alexa Fluor 488-conjugated goat antirabbit antibody (1:500, Thermo Scienti c) containing 2% normal goat serum in PBST for 1 hour at RT. Cells were washed twice with PicoGreen (Invitrogen, P11496) containing DPBS and cover-slipped by using Prolong Gold antifade reagent (Invitrogen, P36930) and dried overnight.Imaging was performed by using the Zeiss LSM 780 Confocal Microscope (Zeiss, Oberkochen, DE).I-BET151 (10µM, 1 hour) and d-BET6 (100nM, 1 hour) were used as the positive control for the experiment.