研究用化学诱导全能干细胞构建小鼠胚胎发育模型
Modelling mouse embryogenesis from chemically induced totipotent stem cells
清华大学等团队在 Nature 发表研究，用化学诱导全能干细胞（ciTotiSCs）构建小鼠胚胎模型 ciToti-EMs，可模拟从着床前到 E8.5 阶段的胚胎发育。
该研究用化学诱导全能干细胞构建小鼠胚胎模型，并系统比对自然胚胎各阶段转录组，可供了解胚胎模型构建与评估方法。
译文尚不完整，完整内容请切换到原文。
数据可用性
评估本研究结果和结论所需的全部数据均在本文及其补充信息中提供。本研究中产生的所有独特且稳定的试剂可在完成材料转移协议后向通讯作者索取。本研究产生的scRNA-seq数据已存入NCBI基因表达综合数据库（GEO），登录号为GSE292862。mm10-2020-A参考基因组下载自https://www.10xgenomics.com/support/software/cell-ranger/downloads。用于比较分析的已发表数据集包括2C胚胎数据（GEO：GSE45719）、着床前和着床后EM（GEO：GSE243925；GSE135701；GSE208681；GSE189425；GSE294766；GSE287909；基因组序列归档：CRA016958；以及国家基因组科学数据中心BioProject：PRJCA027661）和小鼠NE（GEO：GSE134240；GSE135701；GSE186068；GSE100597；GSE109071；GSE117542；以及EMBL-EBI BioStudies：E-MTAB-6967）。源数据随本文提供。
代码可用性
用于scRNA-seq数据分析和可视化的代码可在GitHub获取（https://github.com/hwakwu/Wu-et-al.-2026-ciToti-EM）。
参考文献
Yang, Y. et al. Derivation of pluripotent stem cells with in vivo embryonic and extraembryonic potency. Cell 169, 243–257 (2017).
Yang, J. et al. Establishment of mouse expanded potential stem cells. Nature 550, 393 (2017).
Shen, H. et al. Mouse totipotent stem cells captured and maintained through spliceosomal repression. Cell 184, 2843–2859 (2021).
Yang, M. et al. Chemical-induced chromatin remodeling reprograms mouse ESCs to totipotent-like stem cells. Cell Stem Cell 29, 400–418 (2022).
Xu, Y. et al. Derivation of totipotent-like stem cells with blastocyst-like structure forming potential. Cell Res. 32, 513–529 (2022).
Li, R. et al. Generation of blastocyst-like structures from mouse embryonic and adult cell cultures. Cell 179, 687–702(2019).
Sozen, B. 等。小鼠干细胞自组织形成扩展潜能的类囊胚。Dev. Cell 51, 698–712 (2019)。
Zhang, P. 等。从剪接体抑制的小鼠全能性卵裂球样细胞高效生成类囊胚结构。Sci. China Life Sci. 66, 423–435 (2023)。
Luo, Y. 等。利用全能性细胞来源的类囊胚探索衰老对着床和早期胚胎发育的影响。J. Adv. Res. 68, 115–129 (2025)。
Peng, B. 等。小鼠全能性卵裂球样细胞模拟从合子基因组激活到着床后的胚胎发生。Cell Stem Cell 32, 391–408 (2025)。
Ren, Y. 等。基于连续全能性样细胞的胚胎模型再现小鼠从合子基因组激活到原肠胚形成的胚胎发生。Nat. Cell Biol. 27, 1993–2006 (2025)。
Li, H. 等。通过化学诱导的胚胎创始细胞实现涵盖器官发生的小鼠胚胎发生完整模型。Cell 188, 5912–5930 (2025)。
Yilmaz, A. 等。仅由幼稚 ESC 和 iPSC 无转基因生成小鼠原肠胚形成后全胚胎模型。Cell Stem Cell 32, 1545–1562 (2025)。
Liu, K. 等。通过双向多能干细胞模拟原肠胚后发育。Cell Res. 35, 954–969 (2025)。
Hu, Y. 等。通过确定的化学混合物诱导小鼠全能干细胞。Nature 617, 792–797 (2023)。
Tarkowski, A. K. 小鼠卵分离卵裂球的发育实验。Nature 184, 1286–1287 (1959)。
Zhu, M. & Zernicka-Goetz, M. 哺乳动物胚胎自组织原理。Cell 183, 1467–1478 (2020)。
Rivron, N. C. 等。仅由干细胞生成的类囊胚结构。Nature 557, 106–111 (2018)。
Harrison, S. E. 等。组装胚胎干细胞与胚外干细胞以在体外模拟胚胎发生。Science 356, eaal1810 (2017)。
Sozen, B. 等。胚胎干细胞与两种胚外干细胞类型自组装为原肠胚样结构。Nat. Cell Biol. 20, 979–989 (2018)。
Amadei, G. 等。可诱导的干细胞衍生胚胎在体外捕获小鼠形态发生事件。Dev. Cell 56, 366–382 (2021)。
Tarazi, S. 等。由小鼠幼稚ESC在体外生成的原肠胚后合成胚胎。Cell 185, 3290–3306 (2022)。
Amadei, G. 等。胚胎模型完成原肠胚形成至神经胚形成和器官发生。Nature 610, 143–153 (2022)。
Solter, D. 从畸胎癌到胚胎干细胞及其后：胚胎干细胞研究史。Nat. Rev. Genet. 7, 319–327 (2006)。
Posfai, E. 等。使用日益严格的标准评估全能性。Nat. Cell Biol. 23, 49–60 (2021)。
Macfarlan, T. S. 等。胚胎干细胞潜能随内源性逆转录病毒活性波动。Nature 487, 57–63 (2012)。
Ishiuchi, T. 等。下调复制依赖性染色质组装可诱导早期胚胎样细胞。Nat. Struct. Mol. Biol. 22, 662–671 (2015)。
Zhang, Y. 等。表皮生长因子在扩展多能干细胞重建类囊胚过程中诱导出类似人类胚胎的滋养外胚层谱系转录组。Cell Prolif. 55, e13317 (2022)。
Seong, J. 等。上胚层诱导因子在体外捕获小鼠滋养外胚层干细胞并塑造类囊胚以用于子宫内着床。Cell Stem Cell 29, 1102–1118 (2022)。
Zenker, J. 等。扩展的肌动蛋白环拉链式闭合小鼠胚胎以形成囊胚。Cell 173, 776–791 (2018)。
Nichols, J. & Smith, A. 幼稚态与始发态多能性状态。Cell Stem Cell 4, 487–492 (2009)。
Yanes, O. 等。代谢氧化调控胚胎干细胞分化。Nat. Chem. Biol. 6, 411–417 (2010)。
Pijuan-Sala, B. 等。小鼠原肠胚形成及早期器官发生的单细胞分子图谱。Nature 566, 490–495 (2019)。
Aguilera-Castrejon, A. 等。小鼠胚胎从原肠胚形成前到晚期器官发生的体外发育。Nature 593, 119–124 (2021)。
Richardson, B. & Lehmann, R. 引导原始生殖细胞迁移的机制：来自不同生物体的策略。Nat. Rev. Mol. Cell Biol. 11, 37–49 (2010)。
Ohta, H. 等。小鼠原始生殖细胞样细胞的体外扩增重现了表观遗传空白状态。EMBO J. 36, 1888–1907 (2017)。
Büttner, M., Miao, Z., Wolf, F. A., Teichmann, S. A. & Theis, F. J. 用于评估单细胞 RNA-seq 批次校正的检验指标。Nat. Methods 16, 43–49 (2019)。
Tran, H. T. N. 等。单细胞 RNA 测序数据批次效应校正方法的基准测试。Genome Biol. 21, 12 (2020)。
Zhang, L. 等。单细胞分析揭示结肠癌中髓系靶向治疗的机制。Cell 181, 442–459 (2020)。
Li, J. 等。小鼠和人类原始生殖细胞中可及染色质的准确注释。Cell Res. 28, 1077–1089 (2018)。
Canu, G. & Ruhrberg, C. 首血：造血祖细胞的内皮起源。Angiogenesis 24, 199–211 (2021)。
Capellera-Garcia, S. 等。通过直接谱系转换定义红细胞生成所需的最小因子。Cell Rep. 15, 2550–2562 (2016)。
Dang, D. 等。识别通用巨噬细胞生物标志物的计算方法。Front. Physiol. 11, 275 (2020)。
Epelman, S., Lavine, K. J. & Randolph, G. J. 组织巨噬细胞的起源与功能。Immunity 41, 21–35 (2014)。
Meilhac, S. M. & Buckingham, M. E. 构成哺乳动物心脏的细胞谱系的部署。Nat. Rev. Cardiol. 15, 705–724 (2018)。
Meilhac, S. M. 等。构成心脏的心脏细胞谱系。Cold Spring Harb. Perspect. Med. 4, a013888 (2014)。
Hirschy, A. 等。发育中小鼠心脏心肌细胞骨架结构的建立。Dev. Biol. 289, 430–441 (2006)。
Epelman, S. 等。胚胎期和成年期来源的常驻心脏巨噬细胞在稳态和炎症期间通过不同机制维持。Immunity 40, 91–104 (2014)。
Stary, M. 等。壁内胚层分泌的 SPARC 促进体外早期心肌发生。Exp. Cell. Res. 310, 331–343 (2005)。
Nascone, N. & Mercola, M. 内胚层与心脏发生：新见解。Trends Cardiovasc. Med. 6, 211–216 (1996)。
Ohinata, Y. 等。通过 Blimp1-mVenus 和 stella-ECFP 双转基因报告系统对小鼠原始生殖细胞发育进行全面的非侵入性可视化。Reproduction 136, 503–514 (2008)。
Zhou, Q. 等。胚胎干细胞来源生殖细胞在体外完成减数分裂。Cell Stem Cell 18, 330–340 (2016)。
Garcia, M. D. 等。用于培养小鼠胚胎的大鼠血清制备。Cold Spring Harb. Protoc. 2011, prot5593 (2011)。
Takahashi, M. 等。适用于哺乳动物全胚胎培养的大鼠血清制备。J. Vis. Exp. 90, e51969 (2014)。
Nosaka, Y. 等。在无卵巢体细胞条件下由小鼠胚胎干细胞生成生发泡卵母细胞。Dev. Cell 60, 2976–2994 (2025)。
Deng, Q. 等。单细胞 RNA-seq 揭示哺乳动物细胞中动态、随机的单等位基因表达。Science 343, 193–197 (2014)。
Qiu, C. 等。小鼠胚胎发生过程中细胞轨迹的系统重建。Nat. Genet. 54, 328–341 (2022)。
Mohammed, H. 等。小鼠早期原肠胚形成过程中转录异质性和细胞命运决定的单细胞图谱。Cell Rep. 20, 1215–1228 (2017)。
Cheng, S. 等。单细胞 RNA-seq 揭示小鼠早期发育过程中多能性转变和 X 染色体动态的细胞异质性。Cell Rep. 26, 2593–2607 (2019)。
Chan, M. M. 等。哺乳动物胚胎发生的分子记录。Nature 570, 77–82 (2019)。
Stuart, T. 等。单细胞数据的全面整合。Cell 177, 1888–1902 (2019)。
Korsunsky, I. 等。使用 Harmony 对单细胞数据进行快速、灵敏且准确的整合。Nat. Methods 16, 1289–1296 (2019)。
Young, M. D. 等。来自人类肾脏的单细胞转录组揭示肾肿瘤的细胞身份。Science 361, 594–599 (2018)。
Sala, L. 等。MUSCLEMOTION：一种用于量化体外和体内心肌细胞及心肌收缩的多功能开源软件工具。Circ. Res. 122, e5–e16 (2018)。
van Meer, B. J. et al. Quantification of muscle contraction in vitro and in vivo using MUSCLEMOTION software: from stem cell-derived cardiomyocytes to zebrafish and human hearts. Curr. Protoc. Hum. Genet. 99, e67 (2018).
Acknowledgements
We thank the staff at the Cell Biology Facility, Center of Biomedical Analysis at Tsinghua University for assistance with confocal microscopy, and the Laboratory Animal Research Center at Tsinghua University for support with mouse procurement and experiments; M. Saitou and Q. Zhou for providing BVSC mES cells; K. He and Q. Tao for discussions; P. Tan for helping the initial exploration of bioinformatic analysis; and X. Jiang, Z. Guan, X. Du and J. Yang for experimental help.
Funding
This work was supported by the New Cornerstone Investigator Program (to S.D.), the Beijing Natural Science Foundation (JQ22016 to T.M.), the Natural Science Foundation (325B2030 to H.W.), the National Key R&D Program of China (2022YFA1103704 to S.D.) and the Tsinghua-Peking Center for Life Sciences (to S.D.).
Author information
Author notes
These authors contributed equally: Yanyan Hu, Hanzhi Yu, Nana Huang
Authors and Affiliations
New Cornerstone Science Laboratory, School of Pharmaceutical Sciences, Tsinghua University, Beijing, China
Hua Wu, Hanzhi Yu, Menghan Gao, Xinghan Du, Yuxia Zhang, Xiaoqing Xu, Zhimin Liu, Junsong Yuan, Wei Zhou, Qiliang Shao, Dan Wang, Tianhua Ma & Sheng Ding
Department of Medical Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, USA
Yanyan Hu
Howard Hughes Medical Institute, Chevy Chase, MD, USA
Yanyan Hu
Tsinghua-Peking Center for Life Sciences, Tsinghua University, Beijing, China
Hanzhi Yu, Xinghan Du, Junsong Yuan, Qiliang Shao & Sheng Ding
Department of Obstetrics and Gynecology, Peking University Third Hospital, Beijing, China
Nana Huang, Xiya Sun & Yangyu Zhao
National Clinical Research Center for Obstetrics and Gynecology (Peking University Third Hospital), Beijing, China
Nana Huang, Xiya Sun & Yangyu Zhao
National Center for Healthcare Quality Management in Obstetrics, Beijing, China
Nana Huang, Xiya Sun & Yangyu Zhao
State Key Laboratory of Female Fertility Promotion, Department of Obstetrics and Gynecology, Peking University Third Hospital, Beijing, China
Nana Huang, Xiya Sun & Yangyu Zhao
Beijing Key Laboratory of Reproductive Endocrinology and Assisted Reproductive Technology, Beijing, China
Nana Huang, Xiya Sun & Yangyu Zhao
CRE Life Institute, Beijing, China
Lu Shi
Academy for Advanced Interdisciplinary Studies, Peking University, Beijing, China
Xue Lu
Authors
- Hua Wu
- Yanyan Hu
- Hanzhi Yu
- Nana Huang
- Lu Shi
- Menghan Gao
- Xinghan Du
- Xue Lu
- Yuxia Zhang
- Xiya Sun
- Xiaoqing Xu
- Zhimin Liu
- Junsong Yuan
- Wei Zhou
- Qiliang Shao
- Dan Wang
- Yangyu Zhao
- Tianhua Ma
- Sheng Ding
Contributions
H.W., T.M. and S.D. conceived the project and designed the experiments, with T.M. and S.D. providing overall supervision. H.W. performed most of the experiments and bioinformatic analyses. H.W., Y.H., Y. Zhao, T.M. and S.D. interpreted the data. N.H., H.Y., X.D. and X.S. assisted in establishing the conditions for ex utero embryogenesis. N.H., H.Y. and X.S. helped to process sera. H.Y. assisted with the embryo sectioning and immunostaining. Y. Zhang carried out the mouse breeding and embryo collection. M.G. and Q.S. contributed to the in vitro maturation experiments of PGCs. L.S., X.L. and M.G. assisted in conducting the bioinformatics analyses, with additional interpretation contributed by J.Y.; X.X. and Z.L. assisted with ES cell line establishment. D.W. provided assistance with the reagent preparation. H.W., Y.H., W.Z., T.M. and S.D. wrote the manuscript with input from all of the authors.
Corresponding authors
Correspondence to Tianhua Ma or Sheng Ding.
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Nature thanks Isabelle Migeotte, who co-reviewed with Anchel De Jaime Soguero; Manu Setty; and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Extended data figures and tables
Extended Data Fig. 1 Induction, molecular characterization, and formation of blastocyst-like structures from ciTotiSCs.
a, Representative images of mESCs carrying MERVL-tdTomato and OCT4-GFP reporters cultured in 2i/L, TAW, or TAWa medium. Lower panel, heat maps showing their relative expression of classical totipotency- and pluripotency-associated genes. Scale bars, 200 μm. Representative images from 3 independent experiments. b, Heatmaps showing relative expression of totipotency- and pluripotency-associated genes in middle 2C-embryos (M2C), late 2C-embryos (L2C), mESCs, and ciTotiSCs. c, Expression profiles (z-scores) of maternal and zygotic genome activation (ZGA) gene programs. d, 2D density scatter plot of individual cells based on Pluripotency and Totipotency z-scores. e, Violin plots of Maternal (left) and ZGA (right) module scores. n = 12,419, 8,236 and 10,248 cells for mESC, TAW and TAWa, respectively, derived from one scRNA-seq sample per group. Violin plots show the distribution of individual values. Centre lines indicate the median, boxes indicate the interquartile range, and whiskers extend to 1.5× the interquartile range. f, ECDF plots showing the Maternal & ZGA combined score (left) and transcriptional distance to the ideal state (right). g, Schematic depicting the generation of blastocyst-like structures from ciTotiSCs (left; ACLEB: A83-01, CHIR, LPA, EGF, BMP4; HF: heparin, FGF4), and a bright-field microscopy (BFM) image of collected ciToti4-EMs (right). Scale bar, 200 μm. Representative image from 3 experiments. h, Time-course BFM images of ciTotiSC aggregation in microwells (d1-d4). Scale bar, 200 μm. i, Immunofluorescent images of ciToti structures and blastocysts stained for SOX2 (yellow), GATA3 (magenta), and DAPI (cyan). Scale bars, 200 μm. n = 36 ciToti3-EMs, 23 ciToti4-EMs, and 16 blastocysts from 2 experiments. j,k, Efficiency of ciToti-EM formation at day 4 after varying induction durations (j, n = 14,400 day 0-aggregates, 3 experiments; data are presented as mean values ± s.d.) or A-366 concentrations (k). l, Blastulation efficiency following treatment with different cytokines and small molecules. m, Time-lapse BFM images and quantification of rhythmic blastocoel inflation/deflation cycles. Scale bar, 100 μm. n, Immunofluorescent images (left) of ciToti1-4 EMs and natural blastocysts (E3.5/E4.5) stained for SOX2 (yellow), OCT4 (magenta), and DAPI (cyan), and quantification of SOX2+OCT4+ cells (right). Scale bars, 50 μm. n = 23 (ciToti1-4), 12 (E3.5), 10 (E4.5) from 4 experiments. o, Immunofluorescent images showing the spatial distribution of NANOG (red), OCT4 (green), and GATA6 (magenta) in representative ciToti4-EMs. Yellow and cyan arrows indicate Gata6+ cells in ICM and TE, respectively. Scale bars, 50 μm. n = 47 ciToti4-EMs (28 for case #1, 13 for case #2, 6 for case #3) from 2 experiments. p, Immunofluorescent images of ciToti4-EMs and E4.5 natural embryos (NEs) stained for SOX2 (yellow), SOX17 (green), and GATA3 (magenta). White arrows indicate SOX17+ cells in ICM. Scale bars, 50 μm. n = 15 ciToti4-EMs and 9 E4.5-NEs from 2 experiments. q, Co-immunostaining for polar/mural trophectoderm markers CDX2 (yellow) and CK18 (magenta). Scale bars, 50 μm. n = 7 ciToti4-EMs and 4 E4.5-NEs from 2 experiments. r, Immunofluorescent images (left) of ciToti4-EMs stained for CDX2 (magenta) and E-CAD (green) highlighting squamous TE morphology (white arrows indicate stretched TE cells; Scale bars, 50 μm; n = 5 EMs from 2 experiments), and quantification (right) of the maximum to minimum diameter ratio (\({d}_{\max }/{d}_{\min }\)) for TE and ICM cells (n = 20 TE cells, 20 ICM cells examined over 2 experiments; data are presented as mean values ± s.d.).
Extended Data Fig. 2 Single-cell transcriptomic landscape of pre-implantation ciToti-EMs and developmental characterization of during peri-implantation.
a, UMAP plots of scRNA-seq data from ciToti structures (ciToti0-4) and natural blastocysts (E3.5, E4.5), coloured by unsupervised cluster affiliations. EPI, epiblast; EPI-p, EPI progenitor; TE, trophectoderm; TE-p, TE progenitor; PrE, primitive endoderm; PrE-p, PrE progenitor. b, Expression levels and proportions of cells expressing indicated marker genes across specified clusters and stages. c, Density ridge plots illustrating the distribution of lineage scores within the respective EPI/EPI-p, TE/TE-p, and PrE/PrE-p populations across developmental stages. Dashed arrows highlight the progressive shift in score distributions, indicating the gradual maturation of lineage-specific identities during development. d,e, Fraction of totipotent-like cells (d) and overall lineage composition (e) across the indicated stages. f, Average expression of selected marker genes within intermediate cell clusters in ciToti4-EMs, E3.5 NEs, and E4.5 NEs. g,h, Single-cell trajectory mapped along Components 1 and 2, coloured by inferred pseudotime with a grey arrow indicating developmental direction (g), and highlighting the transitional positioning of the intermediate cell population (blue arrowhead) (h). i, Number and percentage of intermediate cells assigned to TE, EPI, PrE, and NA identities via reference mapping. j,k, Violin plots showing the Euclidean distances from individual intermediate cells to the TE, EPI and PrE centroids in PCA space using the full combined ciToti1–4 dataset (j; n = 2,748 intermediate cells) and in DDRTree space after proportional random downsampling of the combined dataset to 8,000 cells (k; n = 433 intermediate cells). Within each panel, distances to all three centroids were calculated for the same set of intermediate cells. Violin plots show the distributions of individual values. Centre lines indicate the median, boxes indicate the interquartile range, and whiskers extend to 1.5× the interquartile range. l, BFM image of ciToti5-EMs in suspension culture. Scale bar, 100 μm. Representative image from 3 experiments. m, Immunofluorescence images of a ciToti5-EM stained for SOX17 (magenta), CDX2 (red), GATA3 (yellow), and DAPI (green). Scale bars, 50 μm. n = 6 ciToti5-EMs from 2 experiments. n, Sequential BFM images and schematic annotations depicting the transition to egg cylinder-like structures. PAC, pro-amniotic cavity; EPI, epiblast; ExE ecto, extraembryonic ectoderm; PrE, primitive endoderm; TE, trophectoderm. Scale bar, 100 μm. Representative images from 3 experiments. o-q, Characterization of pro-amniotic cavity (PAC) formation by confocal fluorescence imaging of ZO-1 (magenta), SOX2 (yellow), F-ACTIN (grey), and DAPI (cyan) (o), corresponding orthogonal views (p), and fluorescence intensity profiles (q), showing a PAC enclosed by SOX2+ EPI cells with apical F-ACTIN and ZO-1 enrichment. Scale bars, 50 μm. n = 5 ciToti5-EMs from 2 experiments. r,s, Apical-basal polarity of the PAC in ciToti5-EMs (r) and E5.0 NEs (s), stained for PODXL (magenta), F-ACTIN (yellow), and DAPI (cyan), with magnified views highlighting rosette-like apical organization. Scale bars, 30 μm. n = 23 ciToti5-EMs and 4 E5.0 NEs from 3 experiments. t-v, Immunofluorescence analysis of ciToti-EMs stained for SOX2 (yellow), GATA3 (magenta), and DAPI (cyan), shown as 3D reconstruction (t), orthogonal views (u), and intensity profiles (v), showing segregation of SOX2+ EPI and GATA3+ primitive extraembryonic ectoderm (pExE). Scale bars, 50 μm. n = 2 ciToti5-EMs and 2 ciToti6-EMs from 2 experiments. w, Immunofluorescence images comparing E5.0 NEs and ciToti-EMs for PODXL (magenta) and DAPI (green). Arrowheads highlight PODXL localization at apical surfaces of internal cavities and the outer surface of EMs and NEs. Scale bars, 50 μm. n = 23 ciToti5-EMs, 7 ciToti6-EMs, and 4 E5.0 NEs from 3 experiments.
Extended Data Fig. 3 ciToti-EMs progressively acquire post-implantation embryonic features resembling those of NEs up to the E8.5 stage.
a, Schematics (top) and corresponding bright-field microscopy (BFM) images (bottom) illustrating the developmental progression from pre- to post-implantation-like stages. Key compartments are outlined: trophectoderm (TE, magenta dashed line), epiblast (EPI, yellow dashed line), cavities, extraembryonic ectoderm (ExE ecto), visceral endoderm (VE), and amniotic cavity (AC). ICM-p, ICM progenitor; TE-p, TE progenitor. Scale bars, 100 μm. n = 3,065 ciToti3-EMs, 2,615 ciToti4-EMs, 242 ciToti7-EMs, 107 ciToti8-EMs from 8 experiments. b, Representative BFM image of ciToti7-EMs, with egg cylinder-like structures highlighted by yellow outlines. Scale bar, 200 μm. n = 76 EMs from 3 experiments are shown. c, Immunofluorescence images of a ciToti7-EM showing primitive streak (PS, white dashed line) formation. A-P, anterior-posterior axis. PH, phase contrast. Scale bars, 100 μm. n = 2 ciToti7-EMs from 2 experiments. d, Immunofluorescence images showing EPI (SOX2, yellow) and ExE ecto (AP2γ, magenta) compartmentalization in a ciToti7-EM. Scale bar, 100 μm. n = 2 for ciToti7-EMs from 2 experiments. e, Immunofluorescence images comparing luminal cavity formation in ciToti7-EM and E6.5 NE. Scale bars, 100 μm. n = 4 ciToti7-EMs from 2 experiments. n = 6 E6.5 NEs from 2 experiments. f, Co-immunostaining confirming proper compartmentalization of ExE ecto and EPI in ciToti7-EMs. Scale bars, 100 μm. n = 4 ciToti7-EMs from 2 experiments. n = 4 E6.5 NEs from 2 experiments. g, Immunofluorescence images comparing VE and parietal endoderm (PE) among in utero E6.5 NEs, ex utero E6.5 NEs, and ciToti7-EMs. Scale bars, 50 μm. n = 5 ciToti7-EMs from 2 experiments. n = 5 ex-utero E6.5 NEs from 2 experiments. n = 4 in-utero E6.5 NEs from 2 experiments. h, BFM images of multiple ciToti8-EMs (#1-8) maintaining distinct embryonic and extraembryonic compartments. Scale bars, 100 μm. n = 8 ciToti8-EMs from 3 experiments are shown. i, Co-immunostaining demonstrating gastrulation progression (Brachyury/FOXA2). Scale bars, 100 μm. n = 2 ciToti8-EMs from 2 experiments. n = 4 E7.5 NEs from 2 experiments. j, Immunofluorescence images highlighting the development of extraembryonic tissues (HAND1/NR2F2). Dashed boxes indicate the magnified subregions (1-4). Scale bars, 100 μm. White arrows indicate extraembryonic mesoderm (ExEM)-like cells. n = 2 ciToti8-EMs from 2 experiments. n = 5 NEs from 3 experiments. k, Dot plot showing expression levels and proportions of cells expressing selected genes in indicated tissues analysed by scRNA-seq. l,m, Roller culture system (l) and immunofluorescence images of the ectoplacental cone (EPC)-yolk sac connecting region in ciToti10-EMs (m). Scale bar, 100 μm. n = 4 ciToti10-EMs from 2 experiments in (m). n,o, BFM images of partially and fully dissected ciToti10-EMs (n) and fully dissected E8.5 NEs (o). Scale bars, 200 μm. p, Quantification of total ciToti-EM area at different developmental time points (n = 10 individual EMs per time point, 5 experiments). M, mean value. Data are presented as mean values ± s.d. q, Absolute developmental efficiency normalized to Day 0 starting aggregates. Dark grey: empirical observations (ciToti1-4). Light grey: Day 0-normalized extrapolated estimates (ciToti7-12) calculated based on the Day 4 morphological selection rate. Data are mean ± s.d. Indicated n and x represent total number of examined structures and independent experiments, respectively. r, Dorsal view of immunofluorescence images for neuroectoderm (OTX2) and somites (HOXB4). Scale bars, 200 μm. n = 3 EMs from 3 experiments; n = 4 NEs from 2 experiments. s, Confocal images highlighting PAX6+ neural folds (arrowheads) in the anterior region. A-P, anterior-posterior axis. Scale bars, 100 μm. n = 2 ciToti10-EMs from 2 experiments. n = 2 E8.5 NEs from 2 experiments. t, Tissue sections and immunofluorescence images highlighting internal gut and neural fold (NF) architectures. Scale bars, 200 μm. n = 4 sections for both EMs and NEs from 2 experiments, respectively. u, Immunofluorescence images co-stained for A-P regional specification markers (OTX2/HOXB4/CDX2). Insets show magnified views of the somite (S). D-V, dorsal-ventral axis. Scale bars, 200 μm. n = 2 ciToti10-EMs and n = 2 E8.5 NEs from 2 experiments.
Extended Data Fig. 4 Emergence of primordial germ cells, extraembryonic tissue specification, and transcriptome-wide similarity to natural embryos in ciToti-EMs.
a, Immunofluorescence images of a ciToti8-EM co-stained for BLIMP1 (yellow), STELLA (magenta), and DAPI (grey). White arrowheads indicate BLIMP1+STELLA+ primordial germ cells (PGCs). Scale bars, 50 μm. n = 4 ciToti8-EMs from 2 experiments. b, Overlay of bright-field and fluorescence images tracking the spatial emergence and migration of BLIMP1-Venus+ PGCs (green, arrowheads) across ciToti7, ciToti8, and ciToti10-EMs. TB, tail bud; NF, neural folds; Hg, hindgut. Scale bars, 200 μm. n = 7 EMs from 2 experiments. c, Schematic of in vitro dissociation and subsequent expansion/maturation of PGCs on a feeder layer. d, Immunofluorescen images of in vitro expanded PGC colonies. Top/middle: Epigenetic reprogramming (low H3K9me2, high H3K27me3; green) in Stella-CFP+ (SC, magenta) cells. Bottom: Late maturation markers DAZL (green) and DDX4 (red) in SC+ cells. Dashed lines outline PGC colonies. Scale bars, 20 μm. Representative images from 2 independent experiments. e, Time-course IF images (day 9, 11, 13) of in vitro PGC colonies showing progressive BLIMP1 (yellow) downregulation and STELLA (magenta) upregulation. Scale bars, 20 μm. Representative images from 2 independent experiments. f, UMAP plots (left) and corresponding cell type proportions (right) of scRNA-seq data from ciToti10-EMs and E8.5 NEs. g, Expression levels and proportions of cells expressing selected genes across indicated tissues from ciToti10-EMs and E8.5 NEs. h,i, UMAP plots (h) and corresponding cell type fractions (i) of extraembryonic tissues (amnion, allantois, yolk sac, parietal endoderm) from ciToti10-EMs and E8.5 NEs. j, Pearson correlation heatmap detailing transcriptome-wide similarity across extraembryonic tissues. k, Expression levels and proportions of selected genes in the annotated extraembryonic clusters from ciToti10-EMs and E8.5 NEs. l, Feature plots of selected marker genes for amnion, allantois, yolk sac, and parietal endoderm mapped on UMAPs shown in (h). m, Schematic summary and Pearson correlation analysis of transcriptome-wide similarity between staged ciToti-EMs (ciToti4, 7, 8, 10) and corresponding natural embryos (E3.5, 6.5, 7.5, 8.5).
Extended Data Fig. 5 A supervised learning framework for comparative transcriptomic profiling of integrated NE datasets.
a, Schematic of the supervised learning framework. An annotated reference atlas integrating published scRNA-seq datasets (left) trains a logistic regression classifier with elastic net regularization (middle, combining L1 and L2 penalties for robust feature selection) to project and assign quantitative transcriptomic similarity scores to unannotated query NE data (right). b, UMAPs of the integrated single-cell transcriptomic landscape spanning from blastocyst (E3.5/E4.5) to E8.5. Upper panels are coloured by data source (integrating datasets generated in this study with public data), and lower panels are coloured by corresponding cell-type annotations. c, Quantitative assessment of cell-type fidelity. Heatmaps display the average predicted similarity scores (0-1) evaluated by the classifier established in (a), with columns representing clusters from queried datasets and rows representing reference cell types. Z_2022, Zernicka-Goetz_2022; W_2019, Weissman_2019; H_2022, Hanna_2022; H_2025, Hanna_2025.
Extended Data Fig. 6 Integrated transcriptomic landscape and systematic benchmarking of EMs against NEs.
a, UMAP visualizations of a comprehensive single-cell transcriptomic dataset integrating NEs, ciToti-EMs and a collection of recently reported EMs across developmental stages equivalent to E4.5-E8.5. The upper panels are coloured by data source, illustrating the integration and overlap between natural reference data (red, NEs) and various stem cell-derived models. The lower panels display the same embeddings coloured by cell-type annotation, providing a high-resolution map to assess the extent to which EMs recapitulate natural lineage specification. b, Heatmap displaying global, embryo-level integration performance metrics. Scores were computed after Harmony-based batch correction to assess the overall mixing of each EM with the corresponding natural reference (E4.5-E8.5). To visualize relative performance, values are plotted as differential scores (\(\varDelta ={\text{Metric}}_{\text{ciToti}-\text{EM}}-{\text{Metric}}_{\text{other Model}}\)) between ciToti-EM and other models. Notably, for kBET, scores are negated differences (\(-\varDelta \)) to match directional consistency (i.e., higher values indicate lower rejection rates/better mixing). c, Cell-type-resolved differential performance heatmaps. The analysis stratifies integration quality by specific lineages (rows) across different developmental stages (columns). d, Dot plots showing the quantitative distribution of cell types across developmental timepoints (E4.5-E8.5 equivalents). Datasets are grouped into NEs and EMs to facilitate direct comparison. H_2022_EM_day8, day 8 EMs reported by Hanna et al. (2022).
Extended Data Fig. 7 Quantitative assessment of cell type fidelity in EMs using a supervised logistic regression classifier.
a, Global similarity prediction heatmaps for ciToti-EM (this study) and other published EMs, using the supervised logistic regression classifier illustrated in Extended Data Fig. 5a. Datasets were projected onto the natural reference model. Rows represent reference natural cell types, and columns represent corresponding clusters from the indicated EMs. The colour scale indicates prediction probability, highlighting the extent to which each model recapitulates specific natural lineages. The dashed frame highlights a focused comparison of cardiac and blood lineages. A subset of heatmaps from late-stage models highlights prediction scores for cardiac lineages (i.e., Cardiac mesoderm, Cardiomyocytes) and blood lineages (i.e., Blood progenitors, Erythroid). b,c, Dot plots displaying expression levels and proportions of the indicated genes within erythroblasts (b) and cardiomyocytes (c) from ciToti10, H_2022_EM_Day8, ETiX8 EMs, and E8.5-NEs.
Extended Data Fig. 8 Transcriptomic characterization of PGCs and the emergence of hematopoietic lineages in ciToti-EMs.
a, Pearson correlation of global transcriptomes across early mouse embryos (2- to 16-cell stage), ICM of E3.5 NEs, EPI of E4.5 NEs, PSAE (primitive streak and adjacent ectoderm) from E6.5 NEs, and PGCs from E7.5, E8.5 NEs, and ciToti10-EMs. b,c, Principal component analysis (PCA) of the transcriptomes in (a), showing 3D clustering (b) and 2D overlap (c) of ciToti10-EM PGCs and natural E8.5 PGCs. d, Volcano plot showing differential gene expression between E8.5 PGCs and ciToti10 PGCs. Differential expression was assessed using the MAST hurdle-model test implemented in Seurat FindMarkers, with Bonferroni correction for multiple comparisons. Genes with |log2(fold change)| > 1 and P < 0.05 were considered differentially expressed. e, Expression profiles of conserved PGC marker genes identified as not significantly changed in (d). f, Expression levels and the proportions of cells expressing selected genes in PGCs from ciToti10-EMs and E8.5 NEs. g, UMAPs of staged ciToti-EMs (ciToti4, 7, 8, 10) highlighting the emergence and progression of blood lineages (dashed boxes). h,i, Subclustered UMAPs of hematopoietic lineages (vascular endothelium, erythroblasts, epithelia, HEP, and macrophages) from ciToti10-EMs and E8.5 NEs, coloured by sample origin (h) and cluster identity (i). j, Cell type fractions for cells in each cluster relative to all clusters shown in (i). k, Blood collection in ciToti10-EMs. Em, embryo body. Scale bars: (i) and (ii), 200 μm; magnified erythroblasts, 20 μm. Representative images from 3 independent experiments. l, Fluorescence images of isolated ciToti10-EM blood cells stained with membrane-specific NBD-PE (red) and DAPI (blue). Scale bars, 10 μm. Representative images from 2 independent experiments. m, BFM and fluorescence images of later-stage blood cells in ciToti12-EMs. NBD-PE (red), DAPI (blue). Scale bars, 10 μm. Representative images from 2 independent experiments. n, Time-course images of CFU assays (days 4–10) showing characteristic erythroblast (Ery) and macrophage (Mac) colonies derived from hematopoietic progenitors from ciToti10-EMs and E8.5 NEs. Scale bars, 100 μm. Representative images from 2 independent experiments.
Extended Data Fig. 9 Structural, functional, and transcriptomic maturation of E10.5-equivalent cardiac tissues in ciToti-EMs.
a, UMAPs of scRNA-seq data from cardiac tissues of ciToti10- and ciToti12-EMs. FHF, first-heart-field; SHF, second-heart-field; A-CMs, atrial cardiomyocytes; V-CMs, ventricular cardiomyocytes; OFT, outflow tract; CNCCs, cardiac neural crest cells; EC, endothelial cells. b, H&E staining and immunofluorescence images (MHC, yellow; DAPI, blue) of cardiac sections in ciToti10-EMs. A-P, anterior-posterior; D-V, dorsal-ventral. Scale bars, 100 μm. n = 2 ciToti10-EM sections from 2 experiments. H, heart; TB, tail bud; NF, neural fold; Hg, hindgut. c, Normalized trace of spontaneous contraction of a beating heart-like structure in a ciToti10-EM. d, Bright-field image of a ciToti12-EM body dissected from the extraembryonic compartment. H, heart; TB, tail bud; NF, neural fold. Scale bars, 200 μm. n = 3 ciToti12-EMs from 3 experiments. e, Masson’s trichrome staining of cardiac sections from ciToti12-EMs and E10.5 NEs. Magnified insets (i, ii for ciToti12; iii, iv for E10.5) detail a thin epicardial layer (blue arrows) and trabeculated myocardium. Scale bars, 50 μm. n = 2 sections for both EMs and NEs from 2 experiments, respectively. f,g, Immunofluorescence images of cardiac sections from ciToti12-EMs and E10.5 NEs stained for WT1 (magenta), TNNT2 (green), CD31 (red) (f, Scale bars, 100 μm), and WNT5a (green), MSX1 (red), TNNT2 (magenta) (g, Scale bars, 50 μm). Green arrowheads in (g) indicate WNT5a expression in the outflow tract (OFT). DAPI is in white. n = 2 sections for both EMs and NEs from 2 experiments, respectively. h, Immunofluorescence images of cardiac sections stained for atrial marker MYL7 (green), atrioventricular canal (AVC) marker MSX1 (red), pan-myocardial marker TNNT2 (magenta), and DAPI (white/grey). V, ventricle; A, atrium. Dashed boxes (1-3 for ciToti12; 4-6 for E10.5) correspond to magnified views highlighting regionalized expression. Scale bars, 50 μm. n = 2 sections for both EMs and NEs from 2 experiments, respectively. i, Immunofluorescence images characterizing ventricular chamber maturation, stained for HEY2 or NPPA (green) with HAND1 (magenta) and DAPI (white/grey), highlighting outer compacting myocardium (compaction) and inner trabecular network. Scale bars, 50 μm. n = 2 sections for both EMs and NEs from 2 experiments, respectively. j, Immunofluorescence images of cardiac sections from ciToti12-EMs and E10.5 NEs stained for TNNT2 (green), CD31 (red), and DAPI (white/grey). Scale bars, 100 μm. n = 2 sections for both EMs and NEs from 2 experiments, respectively. k, Immunofluorescence images of the sinus venosus (SV) stained for CDH5 (green), SOX17 (magenta), and DAPI (white/grey). Arrowheads highlight their colocalization in the SV endothelial lining. Scale bars, 200 μm. n = 2 sections for both EMs and NEs from 2 experiments, respectively. l, Immunofluorescence images confirming blood lineages within the cardiac chamber of ciToti12-EMs, co-stained for F4/80 (yellow, macrophage/Mac), TNNT2 (green, myocardium), Haemoglobin (red, erythroblasts/Ery), and DAPI (white/grey). Scale bars, 30 μm. n = 4 sections for both EMs from 2 experiments, respectively. m, Feature plots showing expression patterns of lineage-specific marker genes across corresponding sub-populations shown in Fig. 5g. n, Proportion of cells assigned to specific cardiac and hematopoietic lineages in developing hearts from ciToti12-EMs (purple) and E10.5 NEs (grey).
Extended Data Fig. 10 Transcriptomic and regulatory similarity of ciToti4-EM-derived epiblast cells to NE counterparts relative to previously reported blastoid models.
a, UMAP visualization of re-clustered Epiblast (EPI) and inner cell mass (ICM) cells extracted from ciToti4-EMs, NEs (E3.5, E4.5), and published blastoid models (EPSC-blastoid, mTBLC-blastoid, EPSC-S4). Two distinct populations are identified: an “EPI” cluster (predominant in NEs) and an “EPI-like” cluster (an alternative EPI-like state found primarily in previously reported blastoid models). b, Dot plot showing the expression levels of selected marker genes across the “EPI” and “EPI-like” populations in the indicated samples. Genes are categorized by function, including pluripotency regulation, late epiblast specification (E4.5-specific), epigenetic modulation, metabolic regulation (ROS processing, mitochondrial function), trophectoderm (TE) and primitive endoderm (PrE) specification, and somatic differentiation. c, Stacked plot quantifying the proportion of EPI and EPI-like cells in each sample. d, UMAP projection based on single-cell gene regulatory network (regulon) activity inferred by SCENIC, revealing the underlying regulatory landscape of the integrated dataset. e, Violin plots showing per-cell cosine distances between regulon activity profiles of the indicated EMs and those of natural E3.5 (left) or E4.5 epiblasts (right). n = 1,348, 1,094, 639, 550, 134 and 88 cells for EPSC-blastoid, ciToti4-EM, E4.5, E3.5, EPSC-S4 and mTBLC-blastoid, respectively, with one scRNA-seq sample per group. Centre lines indicate the median, boxes indicate the interquartile range, and whiskers extend to 1.5× the interquartile range. No statistical hypothesis testing was performed. f, Feature plots displaying the regulon activity (AUC scores) for key developmental programs. Numbers in parentheses (e.g., ‘63 g’) indicate the number of target genes in each transcription factor regulon.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Wu, H., Hu, Y., Yu, H. et al. Modelling mouse embryogenesis from chemically induced totipotent stem cells. Nature (2026). https://doi.org/10.1038/s41586-026-11164-5
Received: 06 May 2025
Accepted: 14 September 2026
Published: 14 October 2026
Version of record: 14 October 2026
DOI: https://doi.org/10.1038/s41586-026-11164-5
来源：Nature 干细胞主题 · nature.com