henhaobio
Sign in

Abstract

Mammalian biology unfolds over time, within tissues and organs opaque to our eyes and instruments. We applied DNA Typewriter, a sequential molecular recorder, to record the cell lineage of a mouse over nearly two weeks of development. From one embryo, we reconstruct a time-calibrated, parsimony-supported, zygote-rooted phylogeny of 1.28M transcriptionally-profiled cells. A burst of editing unequivocally marks the daughters of the first cleavage, which serve as inline replicates. We quantify clonal dominance arising during gastrulation. Tree siblings share cell type far above chance; heterotypic siblings mark terminal differentiations. Temporal sweeps of clade co-occurrence recover a dated hierarchy of cell-type couplings; imputed labels for internal nodes recapitulate known state paths. A lineage-anchored ontogeny of mammalian development, long out of reach, is coming into view.

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 2 (5 Oct 2026) · CC BY 4.0

A prime-editing recorder in one mouse embryo yields a dated, zygote-rooted lineage tree of 1.28M cells at E13.5, enabling clonal dominance, sibling-fate and cell-type coupling analyses.

Destructive sampling severs each cell from its past, so time-series atlases can infer a cell-type ontogeny only in broad strokes and not lineage relationships. DNA Typewriter had previously been shown only in cultured cells and monoclonal gastruloids, with phylogenies of up to about 50,000 cells.

Key results

  1. ShownA time-calibrated, zygote-rooted phylogeny of 1,281,141 cells from one E13.5 embryo, each tip carrying a transcriptome and one of 138 cell types. About 10% of the embryo is sampled.

    655,701-cell backbone (neighbor-joining, least-squares dating) plus 625,440 cells placed by nearest-relative attachment. 78% (A) and 80% (B) of internal branches carry at least 1 supporting edit; 64% and 68% carry at least 2. · Recording dynamics shape dating and branch…; Distance-based placement doubles the tree to…; Phylogenetic reconstruction of the backbone tree; Internal node support

  2. ShownRecording was not at a constant rate. A burst of editing at the 2-cell stage marks the two first-cleavage daughters, followed by a slowdown and a recovery around gastrulation.

    Founder states carry 23 (A) and 19 (B) edits; about 20-edit burst, slowdown, then recovery near E6 peaking at about 9 edits/cell/day. 55 ± 5 of 66 monomers written in 4,371 fully genotyped cells. The slowdown magnitude is an upper bound. · Single-nucleus RNA-seq of a DNA Typewriter mouse…; Recording dynamics shape dating and branch…; Note S1. Preliminary characterization of four…

  3. ShownClonal dominance arises during gastrulation: lineages defined at E7.0 contribute far more unequally to E13.5 than a neutral pure-birth null predicts.

    3,849 E7.0 founders; the 170 largest account for 50% of tips versus 720 under the Yule null; Gini 0.80 versus 0.50 for the null. Similar on the backbone tree and in blastomere A and B subtrees. · A wave of clonal dominance arises during…; Founder definition; Null model

  4. SuggestedLineage predicts fate at fine and coarse scales. Tree siblings share cell type far above chance, and clade co-occurrence yields a dated hierarchy of cell-type couplings. Absolute dates are method-sensitive.

    51% of sibling pairs share a cell type (9.4-fold over permutation null); 37 progenitor-to-post-mitotic pairings at least 3-fold at 1% FDR; 129 coupling depths spanning E8.25-E13.25. Backbone-only rank order kept (Spearman 0.86); median date shift 0.75 days. · Tree siblings share fate far in excess of…; A hierarchy of fate couplings dated from…; Backbone tree; Within-clone null

  5. SuggestedHeuristic imputation of cell types for internal nodes recovers many known differentiation steps, though the imputed paths are short and not independently validated.

    926,852 of 1,054,879 internal nodes labelled; 275 qualifying state paths, 98% recovered in both blastomeres. Of 360 testable steps, 160 (44%) are adjacent in a 474-edge curated transition graph (15.7-fold over chance). · Ancestral state imputation recovers known…; Ancestral state imputation; Traceback path; Traceback evaluation

Methods, data and reproducibility

Prime-editing tapes (PEmax, epegRNAs, circular TAPE-1 transcripts) were delivered to mouse zygotes by pronuclear injection with piggyBac. Edit states and transcriptomes were read from one E13.5 embryo by sci-RNA-seq3. A dated tree was built by neighbor-joining, least-squares dating and placement, then analysed with permutation nulls.

  • Embryo #3, E13.5, sci-RNA-seq3 transcriptomes with circtape genotypes (11 integrations) · mouse · n = 1,584,848 nuclei after QC · GSE341627
  • Bulk amplicon sequencing of tapes from four additional DNA Typewriter embryos (E14.5, E18.5 x2, E19.5) · mouse · n = 4 embryos (4 of 8 recovered showed editing) · GSE341627
  • Reference time-resolved mouse developmental scRNA-seq atlas used for label transfer and ancestral imputation · mouse · n = 11.4M nuclei, 74 embryos; 1,645,424 cells from 7 timepoints used for annotation
Methods & tools
PEmax, epegRNA, TAPE-1 / circtape, sci-RNA-seq3, DecentTree RapidNJ, LSD2, Scanpy, Scrublet, STAR, fastp, NextCell, ineq (R)
Reagents & environment
Versions stated: LSD2 v2.4.4, BAT v2.11.1, ape v5.7.1, ineq v0.2-13, Scanpy v1.6.0, STAR v2.6.1d, scipy v1.13, trim_galore v0.6.5, Scrublet v0.1.
Reproducibility gaps
Code and intermediate files (Newick trees) are said to be at a project GitHub site and NextCell, but no URL is printed in the text. Constructs are from Addgene or the corresponding author. No compute details given.
Limitations (6)
  • Only one embryo was profiled at single-cell resolution; replication relies on comparing blastomere A and B subtrees. Effects of integration-site gene disruption, such as reduced Chd3, cannot be assessed. Results are preliminary pending independent animals. · stated by the authors
  • The strict molecular clock is violated: early dates are least reliable, dates are underestimated toward the tips, and editing is slower in hematopoietic cell types. Absolute coupling dates should not be overinterpreted. · stated by the authors
  • Placed cells are only reliably in the correct local neighborhood, not necessarily at the correct depth. Accuracy depends on retained tape number and edit count. · stated by the authors
  • Ancestral state imputation is heuristic, short-path and not independently validated. It is scored against a curated transition graph from the group's own earlier atlas. · stated by the authors
  • The tree samples about 10% of the embryo's cells. Of 11 tapes, many are saturated by E13.5, and only 54% of cell by tape entries were genotyped. · stated by the authors
  • The four additional embryos were characterized only by bulk amplicon sequencing, so the dynamics reproducibility claim lacks single-cell support. · noted in this summary

Read this if you work on lineage tracing or developmental cell-type ontogeny and want a dense, dated in vivo mouse phylogeny with its confidence and clock caveats.

Citations
3
FWCI
6.24
Field percentile
top 3%

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Similarity Map
SelectedSame fieldCross-field
THISLiu '26Qiu '24Chan '19Colgan '26Seidel '26Regalado '25
64% semantic matchDevelopmental Biology
Yunzheng Liu, Jihye Park, Laura Amaya+17 more·bioRxiv·Oct 12, 2026