Mechano-osmotic signals control chromatin state and fate transitions in pluripotent stem cells

Abstract

Acquisition of specific cell shapes and morphologies is a central component of cell fate transitions. Although signalling circuits and gene regulatory networks that regulate pluripotent stem cell differentiation have been intensely studied, how these networks are integrated in space and time with morphological changes and mechanical deformations to control state transitions remains a fundamental open question. Here we focus on two distinct models of pluripotency, preimplantation inner cell mass cells of human embryos and primed pluripotent stem cells, to discover that cell fate transitions associate with rapid, compaction-triggered changes in nuclear shape and volume. These phenotypical changes and the associated active deformation of the nuclear envelope arise from growth factor signalling-controlled changes in cytoskeletal confinement and chromatin mechanics. The resulting osmotic stress state triggers global transcriptional repression, macromolecular crowding and remodelling of nuclear condensates that prime chromatin for a cell fate transition by attenuating repression of differentiation genes. However, while this mechano-osmotic chromatin priming has the potential to accelerate fate transitions and differentiation, sustained biochemical signals are required for robust induction of specific lineages. Our findings uncover a critical mechanochemical feedback mechanism that integrates nuclear mechanics, shape and volume with biochemical signalling and chromatin state to control cell fate transition dynamics.

Fig. 1. Exit from primed pluripotency is associated with mechano-osmotic remodelling of the nucleus.

a, Representative images (from five embryos) and quantification of nuclear volume in human preimplantation stage embryos stained for DAPI, Gata6, Nanog and LaminB1. b, Nuclear volume of Gata6-high cells in the ICM (scale bars, 50 and 5 µm; n = 20 (Gata high), 16 (Nanog high) nuclei pooled across 5 embryos; mean ± s.d.; Mann–Whitney). c, Representative images of human preimplantation stage embryos stained for DAPI, Nanog and F-Actin (phalloidin). Note actin-rich bleb-like structures with corresponding nuclear deformation (scale bars, 20 and 10 µm; images representative of 3 embryos). d, Representative images and quantification of nuclear volume and phosphorylated p38 (p-p38) in Gata6-positive and Gata6-negative ICM cells from human blastoids generated from naive hiPS cells stained for Gata6, Oct3/4 and pp38 (scale bar, 100 µm; n = 6 blastoids representing 33 (Gata6 high) and 76 (Gata6 low) nuclei for volumes and 26 (Gata6 high) and 126 (Gata6 low) for p-p38 intensity, respectively; paired t-test). e, Representative top views (x–y), 3D reconstructions and cross-sections (z), of Sox2-GFP-tagged hiPS cells undergoing ectodermal differentiation for the indicated timepoints (representative of 3 independent experiments; scale bars, 15 µm). f, Quantification of nuclear height from hiPS cells undergoing ectodermal differentiation for the indicated timepoints (n = 3 independent experiments with 360 (t = 0), 591 (t = 8), 656 (t = 24), 1,086 (t = 48) total nuclei/timepoint; Kruskal–Wallis/Dunn’s). g, Quantification of nuclear volume from hiPS cells undergoing trilineage differentiation for the indicated timepoints (n = 3 independent experiments with 903, 946, 1,771 and 1,636 (ectoderm 0, 8, 24 and 48 h, respectively); 334, 437, 816 and 741 (mesoderm 0, 8, 24 and 48 h, respectively); and 393, 416, 709 and 835 (endoderm 0, 8, 24 and 48 h, respectively) total nuclei/condition; Kruskal–Wallis/Dunn’s). h, Representative immunofluorescence images of Sox2-GFP-tagged hiPS cells on 2D micropatterns treated with BMP4 for the indicated timepoints and stained for Brachyury and Nanog. Note radial pattern of differentiation at 48 h (scale bars, 100 µm; n = 3 independent experiments). i,j, Representative snapshots (i) and quantification of nuclear deformation and Sox2 intensity dynamics (j) from live imaging videos of mosaic micropatterns with Sox2-GFP and LaminB1-RFP hiPS cells treated with BMP4 for the indicated timepoints. Note transient compaction of colony, accompanied by nuclear deformation before appearance of radial differentiation pattern (scale bars, 100 µm; n = 9 gastruloids pooled across 4 independent experiments; mean ± s.d.). k,l, Representative immunofluorescence images (k) and quantification (l) of LaminB1-RFP-tagged hiPS cells on 2D micropatterns treated with BMP4 for the indicated timepoints of maximal colony compaction and stained for p-p38 and YAP. Note transient YAP and p38 activation at pattern centres as well as sustained YAP and p38 activation within edge cells (scale bars, 100 µm; n = 9 (9 h)/10 (rest) 2D gastruloids pooled across 3 independent experiments with 1,278, 1,918 and 3,203 (6 h centre, mid and edge, respectively); 1,249, 1,881 and 3,135 (7 h centre, mid and edge, respectively); 1,388, 2,155 and 3,590 (8 h centre, mid and edge, respectively); 1,363, 2,054 and 3,421 (9 h centre, mid and edge, respectively); and 1,260, 1,898 and 3,750 (10 h centre, mid and edge, respectively) total nuclei/condition; minimum-to-maximum box plots show 75th, 50th and 25th percentiles; ANOVA/Dunnett’s). a.u., arbitrary units; diff, differentiation; nuc, nucleus; cyto, cytoplasm. Source data

Fig. 2. Mechano-osmotic nuclear remodelling is a rapid response to FGF2 removal.

a, Quantification of change in nuclear volume upon exposure to culture medium/growth factors indicated (n = 4 independent experiments with 953 (Pluripotency; Pluri), 1,157 (Basal), 851 (FGF2) and 882 (TGF-β1) nuclei per condition; minimum-to-maximum box plots show 75th, 50th and 25th percentiles; Kruskal–Wallis/Dunn’s). b, Representative projections of nuclear envelope fluctuations as a function of time upon pluripotency factor removal and adding back specific growth factors. Note that removal of pluripotency factors triggers fluctuations that can be rescued by adding back TGF-β1 and FGF2 (scale bars, 5 µm; n = 301 (Pluri), 268 (Basal), 320 (Basal + FGF), 382 (Basal + TGF-β1) and 386 (Basal + FGF2/TGF-β1) nuclei pooled across 3 independent experiments; ANOVA/Dunnett’s). c, Representative snapshots and line scans of live imaging videos of LaminB1-RFP-tagged hiPS cells in basal medium, stained with FastAct and memGlow to label actin and plasma membrane, respectively. Left: perinuclear actin rings surrounding nuclei and intercellular cavities corresponding nuclear deformation. Right: blebs derived from a mitotic cell deforming the nucleus of a neighbouring cell (scale bars, 10 µm; images representative of 5 videos). d, Quantification of nuclear fluctuations from cells in basal medium with or without inhibitor treatments as indicated (n = 812 (Basal), 267 (Calyculin A (Calyc)), 357 (CytochalasinD (CytoD)), 522 (CytoD + Nocodazole (Nocod.)) and 998 (ATP-depleted) nuclei per condition pooled across 3 independent experiments; ANOVA/Dunnett’s). e,f, A schematic of the experimental outline, representative images (e) and quantification (f) of nuclear envelope fluctuations in cells compressed (Comp) in pluripotency or basal medium for timepoints indicated. Note decreased fluctuations in pluripotency condition and an increase in basal medium (scale bars, 10 µm; n = 3 independent experiments with 318, 337, 355 and 207 (Pluripotency 0, 5, 15 and 30 min, respectively) and 350,188, 320 and 151 (Basal 0, 5, 15 and 30 min, respectively) total nuclei per condition; ANOVA/Fischer’s). g, Quantification of nuclear volume dynamics from of Sox2-GFP-tagged hiPS cells live imaged directly after a media change into pluripotency or basal medium, followed by compression. Line represents median volume and individual dots are average colony volumes at indicated timepoints (n = 10 colonies per condition pooled across 6 independent experiments). h, AFM force indentation experiments of iPS cell nuclei within 20 min of media switch. Note increased elastic modulus of cells in basal media conditions, restored by adding FGF2 (n = 69 (Pluri), 71 (Basal), 76 (Basal + FGF2), 85 (Basal + TGF-β1) and 74 (Basal + FGF + TGF-β1) nuclei pooled across 5 independent experiments; Kruskal–Wallis/Dunn’s). i, Representative tracks of nucGEM particles. Colours represent average rate of diffusion per tracked particle (scale bars, 5 µm). j,k, Quantification of mean squared displacement (MSD) versus lag time (tau τ/s per nucGEM particle (j) and nucGEM diffusion (D_eff) and diffusivity exponent β (k) (n = 260 (Pluri) and 370 (Basal) cells pooled across 4 independent experiments; mean ± s.d.; Kruskal–Wallis/Dunn’s). l,m, Representative snapshots of live imaging and quantification of HALO-tagged endogenous YAP localization (l) and nuclear height (m) in cells compressed to 5 µm height in pluripotency or basal medium. Note YAP nuclear entry in pluripotency condition but not in basal medium upon compression (scale bars, 30 µm; l, n = 3 independent experiments with 127 (Pluri) and 101 (Basal) total cells per condition; m, n = 56 cells (Pluri uncompressed), 42 (Basal uncompressed), 34 (5 µm Pluri), 47 (5 µm Basal), 37 (3 µm Pluri) and 41 (5 µm Basal) cells pooled across 3 independent experiments; ANOVA/Friedman). Source data

Fig. 3. Nuclear flattening primes chromatin for spontaneous differentiation.

a, Representative top views, side views and 3D reconstructions of LaminB1-RFP-tagged hiPS cells subjected to compression (scale bars, 10 µm; images representative of six independent experiments). b, UMAP of scRNA- and scATAC-seq from hiPS cells subjected to compression for timepoints indicated. c, A heatmap of predicted regulons enriched in compressed cells from SCENIC+ analyses of the multiome data. d, A schematic of experimental outline for genome-wide mapping of H3K27ac changes. e,f, Heatmap (e) and metaplot (f) analysis of mean H3K27ac levels at active promoters and predicted active enhancer regions. Note reduction in H3K27ac enrichment at promoters across all conditions and at enhancers in cells compressed in basal medium or exposed to hypertonic shock. g, UpSet plot showing an overlap of enhancers decommissioned in compression and hypertonic shock conditions. h, Venn diagram and Reactome pathway enrichment of compression-specific and shared decommissioned enhancers as defined in g. i, A schematic of the experimental outline for the quantification of the nascent transcriptome. 4sU, 4-thiouridine. j, Quantification of RNA synthesis across conditions from TTseq. Note reduced synthesis across all conditions compared with pluripotency medium condition (n = 3 biological replicates per condition). k, Quantification of total changes in nascent RNA production across conditions relative to the pluripotency medium condition (n = log₂FC of 19,288 genes computed from 3 biological replicates; Tukey’s box plots show 75th, 50th and 25th percentiles). l, A heatmap of z scores from altered nascent RNA levels of relevant transcripts from TTseq quantified by DESeq2. Note increased levels of IEGs specifically in cells compressed in pluripotency medium while key pluripotency and growth factor regulators are repressed. Comp, compression; Hyper, hypertonic; Rec, recovery; pluri, pluripotency. Source data

Fig. 4. Mechano-osmotic signals control kinetics of lineage commitment.

a,b, Schematic of the experimental outline (a) and PCA plot (b) of bulk RNA-seq in cells subjected to compression (Comp) or hypertonic (Hyper) shock and recovery in the indicated media conditions. Note the divergence in transcriptomic responses to hypertonic shock and mechanical compression in pluripotency medium, and their convergence in basal medium. c, A heatmap of the top variable genes from bulk RNA-seq in conditions indicated. Note the increase in differentiation gene expression in cells compressed in basal medium. d, Transcription factor binding enrichment analysis from genes upregulated in the bulk RNA-seq for the indicated conditions. e, Representative examples of gene expression changes across the conditions. Note increased expression of differentiation genes in basal compressed and hypertonic shock conditions (n = 2 biological replicates (Comp)/3 biological replicates rest, pooled; mean ± s.d.). f,g, Schematic of experimental outline (f), representative images and quantification (g) of hiPS cells immunostained for Oct4 and Pax6 after compression or hypertonic shock. Note the increased differentiation in compressed cells or cells exposed to hypertonic shock (scale bars, 75 µm; n = 3 independent experiments with 900 (control); 1,190 (compression); 949 (hypertonic) total nuclei per condition; ANOVA/Kruskal–Wallis). PC, principal component; basal med, basal media; pluri, pluripotency. Source data

Fig. 5. Osmotic pressure controls CBX2 condensation to gate gene repression.

a, Heatmap and Euclidian distance dendrogram of differentially abundant phosphosites quantified by mass spectrometry in cells subjected to compression (Comp) or hypertonic (Hyper) stress. b, Distance-based clustering of phosphosites and GO-term analyses show changes specific or common to the specific stresses. c, Example heatmaps of differentially abundant phosphoproteins from b. d, Representative snapshots of live imaging and quantification of CBX2 condensation dynamics from hiPS cells with an endogenously tagged CBX2 allele. Note the rapid dissolution and re-establishment of condensates upon removal of pluripotency (pluri) factors or exposure to axial compression or hypertonic shock. Arrows mark dissolving condensates; dotted arrows mark newly formed condensates (scale bars, 20 µm; n = 136 (Pluri), 172 (Basal), 99 (Comp); 146 (Hyper) total nuclei tracked over time and pooled across 3 independent experiments; mean ± s.e.m.; two-way-ANOVA/Tukey’s). e, Heatmap and Euclidian distance dendrogram of differential CBX2 occupancy quantified by CUT&Run in cells subjected to removal of pluripotency factors (basal), compression (comp) or compression in basal medium, normalized to pluripotency condition. Asterisks mark transcription factors that control differentiation. f, Reactome analysis of genes in clusters 1 and 4 implicate metal-binding genes with reduced CBX2 occupancy in both basal medium and basal medium compression condition, whereas differentiation genes show reduced CBX2 in compression in basal medium. g, Representative tracks of genes with altered CBX2. h, Model of how intranuclear and cytoskeletal forces influence hiPS cell exit from pluripotency. Under conditions with pluripotency growth factors (GFs), nuclear mechanics are maintained and differentiation is prevented under volumetric stress, restoring pluripotency gene expression. In the absence of pluripotency GFs, osmotic stress leads to nuclear envelope fluctuations and CBX2 condensation, priming chromatin for a cell state transition. This ultimately causes derepression of CBX2 target genes, facilitating exit from pluripotency. Source data

Extended Data Fig. 1. Mechano-osmotic changes of the nucleus accompany exit from pluripotency.

(a-c) Representative top views (x-y), 3D reconstructions and cross sections (z), and quantification of nuclear height from of Sox2-GFP-tagged hiPSCs undergoing ectodermal (a), mesodermal (b), or endodermal (c) differentiation for the indicated time points (scale bars 15 µm; n = 3 independent experiments with 348, 496, 841, 765 (mesoderm 0, 8, 24, 48 h, respectively); 435, 661, 800, 835 (ectoderm 0, 8, 24, 48 h, respectively) total cells/condition; ANOVA/ Dunnett’s). (d) Quantification of lineage marker intensity from immunofluorescence stainings in hiPSCs undergoing ectodermal (left), endodermal (middle), or mesodermal (right) differentiation for the indicated time points (scale bars 15 µm; n = 3 independent experiments with 5093, 1257, 3913, 8290, 11345 (ectoderm SOX1); 1257, 3255, 5751, 8290, 11309 (ectoderm OCT4); 1362, 3967, 4680, 6542, 8450 (Endoderm OCT4); 1362, 3967, 4680, 6542, 6410 (endoderm SOX17); 841, 3089, 4897, 5368, 6189 (mesoderm Brachyury); 842, 3089, 4897, 5368, 6186 (mesoderm OCT4) total cells/0, 24, 48, 72, 96 h conditions, respectively; mean±SD). (e) Representative snapshots from live imaging data and quantification of nuclear volume from LaminB1-RFP hiPSCs treated with Calyculin A for 15 min. Note compaction of colony and nuclear deformation at colony edges (scale bars 50 µm; n = 4 independent experiments with 546 (DMSO) and 682 (Calyculin A) total nuclei/condition; Student’s t-test). (f) Representative images and quantification of p-p38 from hiPSCs treated with Calyculin A for 15 min. Note increased p38 activation at colony edges (scale bars 100 µm; n = 3 independent experiments with 14 (DMSO) and 17 (Calyculin A) total colonies/condition; Student’s t-test). Source numerical data are available in source data. Source data

Extended Data Fig. 2. Mechanical signals control YAP activity while osmotic signals control nuclear rheology.

(a) Osmolarity measurements of the various media conditions (mean ± SD; n = 3 independent measurements). (b) Representative images and quantification of p-p38 from hiPSCs 5 min after exchanging pluripotency maintenance medium with pluripotency, basal or basal + FGF2 medium for 5 min. Note moderate but consistent activation of p38 activation in cells in basal but not in basal + FGF2 medium (scale bars 50 µm; n = 5 independent experiments; RM-ANOVA/ Fischer’s). (c) Representative snapshots of live imaging (x/y), optical cross sections (z), and quantifications from LaminB1-RFP-tagged hiPSCs before (Pre) and after (Hyper) hypertonic shock. Note progressive decline in nuclear volume. Line represents median volume and individual dots are average colony volumes at indicated timepoints (scale bars 10 µm; n = 3 independent experiments with 80, 166, 137 nuclei/experiment tracked over the time; Mann-Whitney). (d) Quantification of nuclear envelope fluctuations in cells exposed to hypertonic shock and subsequent washout in pluripotency or basal medium for time points indicated. Note attenuations of fluctuations upon hypertonic shock in both conditions and recovery to more abundant fluctuations in basal medium (n = 428 (Pluri Baseline), 331 (Pluri Hyper 30), 338 (Pluri Rec 0), 400 (Pluri Rec 10), 372 (Pluri Rec 30), 466 (Basal Baseline), 307 (Basal Hyper 30), 392 (Basal Rec 0), 424 (Basal Rec 10), 317 (Basal Rec 30) cells pooled across 3 independent experiments; Kruskal-Wallis/Dunn’s). (e) AFM force indentation experiments of iPS cell nuclei within 20 min of media switch or hypertonic shock. Note data is reproduced from Fig. 1m but with additional condition of hypertonic shock (n = 69 (Pluripotency), 71 (Basal), 76 (Basal+FGF2), 85 (Basal+TGF-β1), 74 (Basal+FGF + TGF-β1), 60 (Hyper Pluripotency) nuclei pooled across 5 independent experiments; ANOVA/Kruskal-Wallis). (f) AFM force indentation experiments of iPS cell nuclei treated with Cytochalasin D in pluripotency or basal medium (n = 14 and 17 nuclei for Pluripotency and Basal conditions, respectively, pooled across 3 independent experiments). (g) Representative snapshots and quantification of live imaging of YAP-Halo-tag hiPSCs during hypertonic shock (scale bars 10 µm; n = 3 independent experiments with 132 total cells tracked across time). (h) Representative snapshots of live imaging and quantification from YAP-Halo-tag hiPSCs during 3 µm compression in pluripotency or basal medium. Note comparable activation of YAP in both conditions (scale bars 30 µm; n = 3 independent experiments representing 255 (Basal), 201 (Pluripotency) cells/condition tracked across time). (i) Representative snapshots of live imaging and quantification from YAP-Halo-tag hiPSCs transfected with NLS-EGFP and compressed to 3 µm height. Note anticorrelated dynamics of YAP and EGFP where YAP nuclear localization is enhanced upon compression whereas EGFP is not (scale bars 50 µm (left panel), 30 µm (right panel); n = 31 cells pooled across 3 independent experiments). All error bars mean±SD. Source numerical data are available in source data. Source data

Extended Data Fig. 3. Analyses of transcriptional and chromatin changes in single cells in response to mechano-osmotic stimuli.

(a) Quantification of open chromatin from scATACseq of hiPSCs compressed (Comp) for 5 or 30 mins with or without 24 h recovery (Rec) as well as uncompressed controls (CNL). Note reduced accessibility at transcription start sites (TSS) at 30 min compression. (b) Violin plots of computed apoptosis, necrosis, starvation and hypoxia scores from single cell RNAseq across all conditions. (c) UMAP of cell cycle stage distribution from single cell RNAseq across all conditions. (d) Heatmap of ChromVAR analysis from scATACseq of hiPSCs compressed for 30 mins with or without 24 h recovery. (e) Volcano plot of differential gene expression between 30 min compression and CNL. (f) Violin plots of POU5F1 and SOX2 gene expression across all conditions (Wald/IHW). Source data

Extended Data Fig. 4. Changes in enhancer and promoter activity in response to mechano-osmotic stimuli.

(a) Heatmaps of pseudo-bulk ATAC-seq, H3K4me3 and H3K27ac Cut & Tag at annotated H3K27ac peaks, further classified into active promoters and putative active enhancers, illustrating expected enrichment of histone modifications and chromatin accessibility at promoters and enhancers. Log2 fold change (FC) heatmaps highlight H3K27ac changes at active promoters and enhancers across different conditions compared to the pluripotency medium condition. (b) Reactome pathway enrichment of decommissioned enhancers specific to hypertonic shock (Fisher’s exact /Benjamini-Hochberg). Source numerical data are available in source data. Source data

Extended Data Fig. 5. Changes in transcription in response to mechano-osmotic stimuli.

(a) Spike-in normalized MA plot contrasting CompBasal to CompPluri nascent transcript states. Loess local regression line in blue. Selected statistically significant upregulated genes are labeled. (b, c) Spike-in normalized volcano plot contrasting CompPluri to Pluri (b) and CompBasal to Basal (d) conditions (Wald/IHW). (d) Representative western blots from RNAPII-S2P from cells exposed to compression or hypertonic/hypotonic shocks in the media compositions and time points indicated. Note decrease in RNAPII-S2P at 30 min compression in pluripotency medium but already at 5 min in basal medium (n = 3 independent experiments). (e) Representative western blots from phosphorylated p38 (pp38) from cells exposed to compression in the media compositions and time points indicated. Note increase pf pp38 at 30 min compression in pluripotency medium but already at 5 min in basal medium and suppression of compression mediated activation of p38 by addition of FGF2 but not with TGF-β1 into basal medium (n = 3 independent experiments). Hypertonic (HR) and hypotonic (HO) shocks are used as controls for c, d. Unprocessed blots are available in source data. Source data

Extended Data Fig. 6. Long-term transcriptional changes in response to mechano-osmotic stimuli.

(a) Venn diagram and Z-score heatmap analyses of specific and overlapping genes from bulk RNA sequencing in 24 h upregulated in response to compression in basal medium or pluripotency medium when compared to their respective media conditions. Note that compression in basal medium leads to more genes being upregulated and the specific genes representing differentiation genes. Compression in pluripotency medium leads to upregulation of cytoskeletal and stress genes. (b) Gene set enrichment analyses of differentially expressed genes from bulk RNA sequencing in 24 h basal medium versus 30 min compression + 24 h recovery in basal medium (left panel) and 24 h basal medium versus 30 min hypertonic shock + 24 h recovery in basal medium (right panel). Note enrichment of gene sets involved in morphogenesis and metal ion homeostasis in both conditions. (c) Representative images of Pax6 staining and quantification of LaminB1-RFP hiPSCs exposed to 30 min basal medium (Control), 30 min basal medium + 10 min hypotonic shock (Hypotonic), 30 min compression (Compression) and 30 min compression +10 min hypotonic shock (Compression+Hypo) in basal medium followed by 96 h of spontaneous differentiation in basal medium. Note delayed differentiation in both shock conditions (scale bars 50 µm; n = 3 independent experiments with 33844 (Basal), 34959 (Hypo), 35860 (Comp), 27615 (Comp+Hypo) total cells/condition; ANOVA/Dunnett’s). (d) Schematic representation of experimental outline, representative images of Oct4 and Pax6 staining and quantification of cells exposed to 30 min compression or 30 min hypertonic shock in pluripotency medium for 96 h. Note delayed differentiation in both shock conditions (scale bars 75 µm; n = 3 independent experiments with 673 (control), 564 (Compression), 421, 832 (Hypertonic) total cells/condition; Kruskal-Wallis/Dunn’s, ns=not significant). Source numerical data are available in source data. Source data

Extended Data Fig. 7. Analyses of CBX2 condensation dynamics and its role in pluripotency exit.

(a) RT-qPCR from cells exposed to 30 min compression in indicated media conditions followed by 24 h recovery in the same media conditions with and without YAP and ERK inhibitors as indicated. Note increased metallothionine (MT2A) and differentiation gene (LHX5, PAX6, SOX21) expression in basal medium and 30 min compression in basal medium conditions (CompBasal) and lack of rescue with YAP/ERK inhibition. Cells compressed in pluripotency medium (CompPluri) show increased differentiation gene expression upon inhibition of YAP/ERK (n = 3 independent experiments; mean ±SD). (b) Representative images and quantification of CBX2 clustering at the nuclear periphery (scale bars 5 µm; left graph: n = 3 independent experiments with 1450 (Pluripotency), 1091 (Basal), 1181 (PluriComp), 1093 (BasalComp) total nuclei/condition; Student’s t-test; right graph: n = 5 independent experiments with 100 nuclei/condition; mean ±SD). (c) KEGG pathway analysis (left) and CheEA consensus Transcription Factor (TF) prediction (right) from CBX2 peaks differentially occupied in basal/compression in basal medium conditions (Fisher’s exact /Benjamini-Hochberg). (d) Representative images and quantification of CBX2 levels after CRISPRi depletion (scale bars 100 µm; n = 3 independent experiments with 6539, 5597 total cells for CNL and CBX2 CRISPRi conditions, respectively; Student’s t-test). (e) Representative images and quantification of Pax6 levels after CBX2 CRISPRi depletion and 96 h of spontaneous differentiation in Basal medium (scale bars 100 µm; n = 3 independent experiments with 21562, 25843 total cells per CNL, CBX2 CRISPRi conditions, respectively; Student’s t-test). Source numerical data are available in source data. Source data