Developmental signals control chromosome segregation fidelity during pluripotency and neurogenesis by modulating replicative stress

Abstract

Human development relies on the correct replication, maintenance and segregation of our genetic blueprints. How these processes are monitored across embryonic lineages, and why genomic mosaicism varies during development remain unknown. Using pluripotent stem cells, we identify that several patterning signals-including WNT, BMP, and FGF-converge into the modulation of DNA replication stress and damage during S-phase, which in turn controls chromosome segregation fidelity in mitosis. We show that the WNT and BMP signals protect from excessive origin firing, DNA damage and chromosome missegregation derived from stalled forks in pluripotency. Cell signalling control of chromosome segregation declines during lineage specification into the three germ layers, but re-emerges in neural progenitors. In particular, we find that the neurogenic factor FGF2 induces DNA replication stress-mediated chromosome missegregation during the onset of neurogenesis, which could provide a rationale for the elevated chromosomal mosaicism of the developing brain. Our results highlight roles for morphogens and cellular identity in genome maintenance that contribute to somatic mosaicism during mammalian development.

Fig. 1. Embryo patterning signals regulate chromosome segregation fidelity in PSCs.

a Schematic of signalling-driven patterning and the hypothesised role in CIN. b Chromosome segregation analyses in mESCs and hiPSCs upon treatment for 16 h with different pathway-specific activating signals (agonists), inhibiting signals (antagonists), or small molecule compounds. Data are mean and P-value from one-way ANOVA analyses with multiple comparisons with Dunnet corrections of n = 3–6 biological experiments with >100 anaphases per condition in each experiment. An example of a DKK1-treated hiPSC in anaphase is shown. DAPI stains the DNA and CENPC marks kinetochores. Scale bar = 10 μm. c, d Chromosome missegregation (n = 3-4 biological replicates) from (b) and representative qRT-PCR analyses of the WNT target gene AXIN2 (c) or the BMP target gene ID1 (d) in hiPSCs upon treatment for 16 h with the indicated compounds and proteins (the experiment was repeated three times). Data are plotted as mean ± s.d. of n = 3 technical replicates. P-values from one-way ANOVA analyses with multiple comparisons with Dunnet corrections from the indicated groups, **P < 0.01, ***P < 0.001. e Schematic of the signalling axes driving gastrulation in mammalian embryos. Note that Chordin and Noggin do not establish the anterior visceral endoderm in mammals, but are required for subsequent anterior patterning (see also Supplementary Data 1). f Chromosome segregation analyses in hiPSCs upon co-treatment with the indicated signals promoting anteriorisation or posteriorisation during mammalian gastrulation. Data are mean ± s.d. of n = 3 biological replicates, with >100 anaphases analysed in each condition per replicate). Source data for all experiments are provided as a Source data file. P-values from one-way ANOVA analyses with multiple comparisons and Dunnet corrections from the indicated groups. From left to right: **P = 0.0025, **P = 0.0011. a, e Created with BioRender.com released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license.

Fig. 2. WNT, BMP, and FGF signalling regulate chromosome segregation fidelity in PSCs.

a Simplified scheme of the WNT, BMP, and FGF signalling pathways highlighting the proteins and modulators used in this work. b–d Chromosome segregation analyses in hiPSCs treated as indicated for 16 h. Data are mean ± s.d. of b n = 4 biological replicates (except n = 3 in FGF2 condition), c n = 3 biological replicates (left panel) and n = 3–6 biological replicates (right panel), and d n = 4 biological replicates with >100 mitotic cells analysed per condition in each replicate. P-values from one-way ANOVA analyses with multiple comparisons and Dunnet corrections are indicated as *P < 0.05, **P < 0.01, ***P < 0.001, or n.s. (P > 0.05, not significant). e Schematic of the epistasis interactions between WNT, FGF and BMP signalling in chromosome segregation fidelity. f Analyses of the chromosome gains in Giemsa staining and M-FISH experiment analyses are shown. Data represent a representative M-FISH experiment with n = 40 cells analysed per condition, and a total of n = 3 biological replicates of Giemsa staining with the following cells counted per condition (in total): Control (n = 208 cells), DKK1 (n = 229 cells), FGF2 (n = 79 cells) and Noggin (n = 179 cells). g M-FISH examples of hiPSCs treated as indicated. Source data for all experiments are provided as a Source data file. a, e Created with BioRender.com released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license.

Fig. 3. WNT, BMP, and FGF converge into the regulation of DNA replication stress.

a GO term enrichment analysis of common transcripts differentially upregulated by DKK1, FGF2, and Noggin in hiPSCs after 16 h treatment. Expression analyses were performed by single-cell RNA sequencing of >75 cells per condition. After normalizing the data for differences in library size, the ‘FindMarkers’ function with the ROC test was used to determine differentially expressed genes (DEGs) between the different treatment conditions. We selected DEGs with power > 0.25 for the subsequent GO analysis. P-values from GO enrichment analyses are provided directly from DAVID’s EASE score: a one-sided Fisher Exact P-value for gene-enrichment analysis. Data is provided in Supplementary Data 2. b–g DNA combing experiments in hiPSCs treated as indicated for 3 h and labelled with consecutive pulses of CIdU and IdU as shown in (b). Each experiment of every subfigure was replicated at least twice, and the data displayed corresponds to a representative experiment, where is shown the mean ± s.d. of: control (n = 127), Aphidicolin (n = 95), DKK1 (n = 150), FGF2 (n = 135), and Noggin (n = 153) single forks measurements in (c) and control (n = 42), Aphidicolin (n = 59), Aphidicolin + WNT3A (n = 44), Aphidicolin + GSK3i (n = 123) and Aphidicolin + BMP4 (n = 53) single fork measurements in (f). In d, g inter-origin firing distance was assessed. In d control = 46, Aphidicolin = 60, DKK1 = 22, FGF2 = 48, Noggin = 19 inter-origin distances were measured, and in g control (n = 49), Aphidicolin (n = 49), Aphidicolin + WNT3A (n = 31), Aphidicolin + GSK3i (n = 26) and Aphidicolin + BMP4 (n = 59) inter-origin firing distances were measured. In b examples of fork progression and origin (O) are shown. In e a scheme of the replication stress cascade targeted in (f, g) is shown. P-values from one-way ANOVA analyses with multiple comparisons and Dunnet corrections from the indicated groups, ***P < 0.001 or in (f) n.s. > 0.05. h Proposed roles of WNT, BMP, and FGF signalling in DNA replication. i Pipeline for single-cell EdU sequencing. Note that hiPSCs were treated first for 3 h, followed by two 15-min pulses of EdU separated by 1 h. j Replication forks per cell obtained by scEdU-seq in control or DKK1 (3 h) treated hiPSCs from a single biological experiment. Data corresponds to the number of forks per cell in Control (n = 894 cells) and DKK1 (n = 888 cells) after sequencing analysis. Single cells were ranked for their relative position in the S-phase (x-axis) according to their fork distribution pattern across different chromosomes, as previously described. k Replication forks per chromosome per cell, as described in (j). Source data for all experiments are provided as a Source data file, with the exception of (j, k), which is included in the data repository GSE271478. h, i Created with BioRender.com released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license.

Fig. 4. WNT, BMP, and FGF affect DNA damage in S-phase and chromosome segregation fidelity in mitosis through their roles in DNA replication.

a–f Accumulation of γ-H2AX foci (a–d) and phospho-RPA (e, f) during S-phase in hiPSCs treated for 3 h as indicated. HU, hydroxyurea; CPT, camptothecin. Data are median fluorescence intensity (MFI) of EdU⁺ nuclei of a representative experiment, performed at least three independent times (n = 3 biological replicates) with >250 cells per condition in each experiment. MFI at EdU⁺ nuclei of: control (n = 616 cells), DKK1 (n = 571 cells), FGF2 (n = 526 cells), Noggin (n = 478), and Aphidicolin (n = 460) in (b), control (n = 1824 cells), Aphidicolin (n = 2394 cells), Aphidicolin + WNT3A (n = 1523 cells), Aphidicolin + GSK3i (n = 1549 cells), and Aphidicolin + BMP4 (n = 1743 cells) in (c), control (n = 250 cells), HU (n = 355 cells), HU + WNT3A (n = 529 cells), HU + GSK3i (n = 373 cells), and HU + BMP4 (n = 448 cells) in (d), control (n = 414 cells), CPT (n = 417 cells), CPT + WNT3A (n = 689 cells), CPT + GSK3i (n = 582 cells), and HU + BMP4 (n = 657 cells) in (f) were analysed. g Schematic summarising the signalling functions of WNT, BMP, and FGF upon DNA replication stress. P-values from one-way ANOVA analyses with multiple comparisons with Dunnet corrections from the indicated groups, ***P < 0.001, in (d) n.s. = 0.993, in (f) n.s. = 0.217. h, i Mitotic microtubule plus-end assembly rates measured by EB3-GFP tracking during prometaphase in hiPSCs. Cells were treated as indicated (16 h of total treatment) and arrested in mitosis using dimethylenastrone for 2 h prior to imaging. Experiments were replicated at least two times and the data shown corresponds to a representative experiment. Data are mean ± s.d. of average growth rates of 20 microtubules/per cell, where microtubule polymerization rate was measured for at least four consecutive time points (every 2 s). Each dot in the figure corresponds to one cell; control = 12 cells, Aphidicolin = 15 cells (left panel), control = 14 cells, control + dNs = 11 cells, DKK1 = 13 cells, DKK1 + dNs = 12 cells, FGF2 = 13 cells, FGF2 + dNs = 14 cells, Noggin = 13 cells, Noggin + dNs = 12 cells. P-values from one-way ANOVA analyses with multiple comparisons with Dunnet corrections are indicated as ***P < 0.001, or not significant (n.s. = 0.974), for single cells of a representative experiment. j–l Chromosome segregation analyses in hiPSCs treated as indicated for 16 h. Data are mean ± s.d. of n = 4 biological replicates (j), n = 3 biological replicates (k), and n = 3 biological replicates (l) with >100 mitotic cells per condition and per replicate. P-values from one-way ANOVA analyses with multiple comparisons with Dunnet corrections are indicated as *P < 0.05, **P < 0.01, ***P < 0.001, or in (j) not significant (n.s. = 0.07) for independent experiments. m Proposed model for cell signalling regulation of DNA replication and chromosome segregation fidelity. Scale bars = 10 μm. Source data for all experiments are provided as a Source data file. g, m Created with BioRender.com released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license.

Fig. 5. WNT, BMP, and FGF do not impact chromosomal stability after human early lineage specification.

a Summary of chromosome segregation analyses across different embryonic lineages obtained upon differentiation of hiPSCs as further described in Supplementary Fig. 5a. Data show fold changes in chromosome missegregation rates upon DKK1 treatment (WNT inhibition) or aphidicolin versus control untreated conditions, and P-values from two-tailed t-tests (treated vs control) of 3–7 biological replicates with >100 anaphases analysed in each condition, also shown in Supplementary Fig. 5. Data are displayed in a lineage tree depicting relative developmental position (A–P) in time of each lineage. NPC, neural progenitor cell. b–d Chromosome segregation analyses in hiPSCs undergoing differentiation towards primitive streak (P.S.) (b) or neuroectoderm (c) as indicated in Supplementary Fig. 6a, and treated for 16 h as indicated. Data are mean ± s.d. of b n = 3 biological replicates, c n = 3 biological replicates and d n = 4-5 biological replicates with >than 100 anaphases quantified per condition and per replicate). Note that GSK3i/WNT3A were exchanged to DKK1 during the differentiation towards primitive streak to ensure proper WNT inhibition. In d hiPSCs were cultured with DKK1 in E8 media for 4 days, and either kept for another 16 h with DKK1 or exchanged towards control (−), WNT3A or GSK3i in E8 media. P-values from two-way with Tukey corrections (b, c) or one-way with Dunnet corrections (d) ANOVA analyses with multiple comparisons are indicated as **P < 0.01, ***P < 0.001, or n.s. (P > 0.05, not significant). e Representative Western blot analyses (of n = 2 biological replicates) of hiPSCs undergoing differentiation towards primitive streak (Left = day 0, right = day 2 of differentiation as shown in Supplementary Fig.7a). The molecular weight markers are indicated in kDa. f, g Chromosome segregation analyses in hiPSCs, primitive streak-like (P.S.), lateral mesoderm (L.M.) and paraxial mesoderm (P.M.) like cells treated as indicated for 16 h. Data are mean ± s.d. of f n = 3 biological replicates and g n = 3 biological replicates with >100 anaphases quantified per condition and per replicate) APH, 50 nM aphidicolin. ATMi, 3 μM AZD0156. P-values from one-way ANOVA analyses with multiple comparisons and Dunnet corrections are indicated as *P < 0.05, **P < 0.01, ***P < 0.001, or n.s. (P > 0.05, not significant). h Schematic of the functional interaction between signalling roles in chromosome segregation and cell fate. Source data for all experiments are provided as a Source data file. h Created with BioRender.com released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license.

Fig. 6. WNT and FGF display antagonistic roles in genome stability during human in vitro neurogenesis.

a, b In vitro specification of hiPSCs into cortical neural progenitors and neurons. NIM neural induction media, NPC neural progenitor cell, B-tub III beta tubulin III. c, d Chromosome segregation analyses in hiPSCs, expanding hiNPCs (10 days) and differentiating hiNPCs (16 days) as indicated in (a), and treated with DKK1 or FGF2 16 h before harvesting. Data are mean ± s.d. of (c) n = 3 biological replicates and (d) n = 3-4 biological replicates with >50–100 anaphases quantified per condition in each replicate). P-values from two-way ANOVA analyses with multiple comparisons and Tukey corrections are indicated as *P < 0.05, **P < 0.01, ***P < 0.001, or n.s. (P > 0.05, not significant). e DNA combing experiments in differentiating hiNPCs treated as indicated for 3 h, and labelled with consecutive pulses of CIdU and IdU. Data are mean ± s.d. of the fork speed measurement in Control=83 forks, DKK1 = 112 forks and FGF2 = 119 forks, corresponding to a representative experiment after biological replication (n = 3). P-values from one-way ANOVA analyses with multiple comparisons with Dunnet corrections are indicated, from left to right as **P = 0.0052 and **P = 0.0011. f, g Accumulation of γ-H2AX foci in differentiating hiNPCs treated as indicated. A representative experiment from three independent experiments is shown. Data are median of the fluorescence intensity (MFI) in control 267 nuclei, DKK1 = 206 nuclei and FGF2 = 247 nuclei. P-values from one-way ANOVA analyses with multiple comparisons with Dunnet corrections are indicated, from left to right as **P = 0.0017 and **P = 0.002. h–j Chromosome segregation analyses in differentiating hiNPCs (16 days) treated as indicated for 16 h. Data are mean ± s.d. of n = 3-4 biological replicates (h–j) with 50–100 mitotic cells analysed per condition in each replicate. P-values from one-way ANOVA analyses with multiple comparisons with Dunnet corrections are indicated as *P < 0.05, **P < 0.01, ***P < 0.001. Scale bars = 10 μm. Source data for all experiments are provided as a Source data file.

Fig. 7. A tug-of-war between WNT and FGF regulates chromosome segregation fidelity in mouse neural progenitors.

a–c Immunofluorescence analyses of the activated WNT (LRP6) and FGF (FGFR1) receptors in the ventricular zone of E12.5-16.5 embryonic mouse brains. Quantification of phospho-LRP6 (S1490) (b) and phospho-FGFR1 (Y154) (c) in NPCs from the ventricular zone of mouse embryos is shown for three developmental stages of cortical neurogenesis. Data are mean ± s.d. of the relative fluorescence intensity normalised to Nestin in apical neural progenitors of n > 5 brain cryosections from n = 3 embryos of each condition. P-values from one-way ANOVA analyses with multiple comparisons with Dunnet corrections are indicated as ***P < 0.001, or n.s. (P > 0.05, not significant). d Chromosome segregation analyses in ex vivo cultured NPCs from E12.5 and E14.5 mouse embryos, treated as indicated for 16 h. Data are mean ± s.d. of chromosome missegregation rates from n = 3 biological replicates with >50 anaphases per condition in each replicate. In each experiment, mNPCs dissociated from n > 3 mouse embryo brains were pooled together for seeding. P-values from two-way ANOVA analyses with multiple comparisons and Tukey corrections are indicated as **P < 0.01, ***P < 0.001, or n.s. (P > 0.05, not significant). e, f Accumulation of γ-H2AX foci in S-phase of E14.5-derived NPCs treated for 3 h as indicated. Data are the MFI in n > 200 EdU+ nuclei from a representative experiment out of three biological replicates. P-values from one-way ANOVA analyses with multiple comparisons with Dunnet corrections are indicated as ***P < 0.001. g Chromosome segregation analyses in ex vivo cultured NPCs from E14.5 mouse embryos, treated as indicated for 16 h. Data are chromosome missegregation rates plotted as mean ± s.d. from n = 3 biological replicates with 50–100 anaphases analysed per condition and per replicate. P-values from one-way ANOVA analyses with multiple comparisons with Dunnet corrections are indicated as ***P < 0.001. h Schematics of in utero ventricular injections of PBS (control) or recombinant DKK1 in E13.5 mouse embryos, later sacrificed at stage E14.5. i, j Chromosome segregation analyses in the ventricular zone of PBS (Control) or DKK1 injected mouse E14.5 embryos. Data are mean ± s.d. of three injected embryos per condition (>10 cryosections per embryo). P-values from a two-tailed t-test are indicated as *P = 0.029. k A tug-of-war between WNT and FGF controls chromosome segregation fidelity in NPCs, and might underlie the high levels of chromosome missegregation occurring during neurogenesis. Scale bars = 10 μm. Source data for all experiments are provided as a Source data file. h, k Created with BioRender.com released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license.

Fig. 8. Model of proposed roles of patterning signals in chromosome segregation fidelity during human lineage specification.

In pluripotent stem cells, the WNT, BMP, and FGF pathways form part of an ATM-dependent signalling rheostat that modulates DNA replication stress during S-phase, which in turn regulates microtubule dynamics and chromosome segregation fidelity in the subsequent mitosis. WNT signalling sits at the helm of this regulatory network by protecting pluripotent stem cells from chromosome missegregation upon different sources of DNA replication stress, including by other patterning signals. The capacity of investigated extracellular signals to influence chromosome segregation fidelity is largely lost after exit from pluripotency and specification into the three germ layers following the withdrawal of ATM signalling as a first responder during DNA replication stress, but remerges during neurogenesis. In particular, we find that FGF signalling induces high levels of chromosome missegregation of neural progenitors committed to neurogenesis. Figure 8 was created with BioRender.com and released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license.