Write a comment

This manuscript describes a chemically defined, five-component cocktail (PRBJN) that rapidly and efficiently converts naïve human PSCs into 8-cell-like cells (ri8CLCs), reaching up to ~40% TPRX1-positive cells within 48 hours — a substantial improvement in speed and yield over previously reported approaches. The chemical screening strategy is systematic and well executed, and the transcriptomic and chromatin-accessibility characterization (bulk RNA-seq, ATAC-seq, and single-cell RNA-seq with reference mapping to human embryo datasets) provides reasonably convincing evidence that PRBJN-treated cells acquire an 8-cell/ZGA-like transcriptional identity. The functional data — accelerated blastoid formation, spontaneous trophoblast differentiation, and enhanced chimeric contribution to both ICM and TE in mouse embryos — are the most compelling and potentially high-impact aspects of the study, suggesting genuinely expanded developmental competence rather than a purely transcriptional phenocopy.

That said, several of the paper's central functional claims are not yet adequately supported by the data shown, and the manuscript would benefit substantially from additional controls before publication. Most importantly: (1) the immunofluorescence panel intended to demonstrate trilineage blastoid identity (Fig. 5M) does not match its own figure legend and does not show the markers described in the text, so the core claim of blastocyst-like lineage segregation is currently unsubstantiated; (2) nearly all functional assays (blastoid formation, trophoblast differentiation, chimera injection) use bulk, unsorted PRBJN-treated populations rather than sorted TPRX1+ cells, leaving open whether the reported phenotypes are attributable to the 8CLC state specifically or to direct pharmacological effects of the compounds; and (3) no genomic integrity data are provided despite the induction protocol combining a p53 agonist with broad chromatin-modifying compounds. These issues, together with several smaller inconsistencies in timing, controls, and comparisons to prior work detailed below, should be addressed. With these revisions, this could be a strong and useful contribution to the field; as it stands, the functional claims outpace the evidence provided.

Major issues

• 1. Blastoid lineage identity is inadequately supported, and Figure 5M does not match its own legend. The Figure 5M legend states that blastoids were stained for NANOG (epiblast), SOX17 (hypoblast/primitive endoderm), and GATA3-mKO2 (trophectoderm). However, the panel itself is labeled DAPI / KLF17 / GATA3-mKO2 / MERGE — neither NANOG nor SOX17 appears anywhere in the figure. The main text (line 299) likewise states that blastoids contain "inner cells expressing NANOG," which is not what is actually shown; KLF17 is an 8CLC/ZGA marker, not a canonical epiblast marker. Beyond this discrepancy, no hypoblast marker is shown at all, and GATA3 and KLF17 appear to co-localize in the ri8CLC panel rather than mark spatially distinct compartments. As it stands, the manuscript does not demonstrate that ri8CLC-derived blastoids contain the three lineages required to support the claim of faithful blastocyst-like structure. This needs to be corrected (matching stains to the legend) and, at minimum, supplemented with proper epiblast/hypoblast/TE triple staining before the claim can be considered supported.

  2. Functional lineage potential of ri8CLC-blastoids is untested. Can epiblast-, hypoblast-, and trophectoderm-derived cell lines actually be established from ri8CLC-blastoids upon appropriate seeding? Morphological cavitation and marker expression are necessary but not sufficient evidence of a bona fide blastocyst-like structure; functional derivation of the three lineages would substantiate the claim far more convincingly.

  3. The functional assays confound "8CLC state" with direct compound pharmacology, and this is not controlled for. All blastoid formation, trophoblast differentiation, and chimera injection experiments appear to use bulk, unfractionated PRBJN-treated cells (12–24 h), not sorted TPRX1+ populations. This leaves open the possibility that the enhanced developmental phenotypes are driven directly by PY60/Nutlin-3a/retinoic-acid signaling on cells that never passed through an 8C-like state, rather than by genuine transit through totipotency. Sorting TPRX1+ vs. TPRX1– fractions and testing each for blastoid/trophoblast/chimera competence would be the key experiment to resolve this, and its absence is a significant gap in the causal argument the paper makes.

  4. No genomic or cellular integrity data are provided for a demonstrably stressful induction protocol. PRBJN combines a p53 activator (Nutlin-3a/MDM2i) with two KDM inhibitors that broadly remodel chromatin (over 52,000 ATAC-seq regions altered). Given this, and given the paper's stated ambition as a scalable induction platform, the absence of karyotyping, DNA-damage marker analysis (e.g., γH2AX), or comet assay data is a meaningful omission that should be addressed before the platform is presented as broadly usable.

  5. Persistence and sortability of the TPRX1+ state are not characterized. Can TPRX1+ cells be isolated and maintained, or do they rapidly revert/lose the 8C signature once sorted? This bears directly on both the platform's practical utility and on resolving concern #3 above (whether developmental competence tracks with 8CLC identity specifically, or with bulk treatment).

  6. Treatment duration is inconsistent across assays without justification. The manuscript uses at least three different treatment windows for its key readouts — 12 h (blastoid formation, chimera injection), 24 h (scRNA-seq, trophoblast differentiation), and 48 h (bulk RNA-seq/TPRX1 sorting, where the strongest 8C transcriptional signature is reported). No rationale is given for why each assay uses a different timepoint, and it is not clear that "peak 8CLC identity" (48 h) is the state actually being tested functionally (12–24 h). This should be explicitly justified or reconciled.

  7. No direct comparison against Li et al. (spliceosome inhibition) or Mazid et al. induction protocols. The scRNA-seq reference-mapping analysis does include the Mazid dataset (Fig. 4F–G), but only as a low-dimensional embedding/label-transfer comparison — not a direct correlation of induction signatures, as the authors do for DUX4 vs. PY60 (Fig. 2F). Given that PRBJN's chemical logic (chromatin/p53/RA-based) is entirely distinct from spliceosome inhibition (Li et al.) or the Mazid protocol, a quantitative transcriptomic comparison of the induced states would substantially strengthen (or complicate) the claim that these represent convergent, or distinct, routes to the same 8C-like identity.

  8. Can ri8CLC reproduce Li et al.'s fully spontaneous blastoid formation? Since PRBJN already renders MEK inhibition, Nodal/Activin inhibition, and LIF dispensable, and LPA remains the only required PALLY component, it is a natural extension to test whether LPA can also be omitted — matching Li et al.'s report of fully spontaneous blastoid formation. This seems a low-effort, high-value control given the omission experiments the authors already performed (Fig. 5L).

  9. Epigenetic characterization is limited to chromatin accessibility. ATAC-seq and motif enrichment are informative but shallow relative to the paper's mechanistic claims about chromatin barriers. DNA methylation and histone modification data (e.g., H3K4me3, given the JQKD82/KDM5 mechanism proposed in the Discussion) would substantially strengthen the mechanistic narrative, though this is lower priority than the functional/validation gaps above.

  10. Possible confound between p53/stress response and genuine ZGA-like signature. The TP53-KO experiment (Fig. S2) shows that Nutlin-3a's TPRX1-inducing effect is p53-dependent, but does not establish whether the "ZGA-like" signature seen with the full PRBJN cocktail is specifically an 8C-like program or partly a generic p53/stress-response transcriptional signature. Comparing GSEA specificity with and without Nutlin-3a in the cocktail would help distinguish these possibilities.

Minor issues

• Lines 64–65: Add a sentence or two explicitly linking "chromatin-associated processes" to the specific epigenetic regulator classes screened, to better motivate the screen's design for readers.

• Line 85 / Fig. S1E: The KLF17 result appears inconsistent with the other 8CLC markers shown — worth flagging or explaining in the text.

• Line 89: Specify which naïve culture condition was used for the primary screen.

• Choice of culture conditions: Given the number of established naïve conditions, the manuscript would benefit from briefly justifying the choice of HENSM for both screening and PALLY-based blastoid formation, since the PALLY protocol was not originally developed in HENSM.

• Figure 1A: With multiple HDAC inhibitors among the hits, consider labeling them explicitly so readers can distinguish why compounds with stronger raw TPRX1+ induction were not prioritized over PY60.

• Figure 4A: Report the percentage of cells in cluster 4, and indicate whether it is comparable to the %TPRX1+ measured by FACS.

• Figure 4: Report the quantitative overlap between TPRX1+ cells and cluster 4 assignment.

• Line 247: The phrase "cluster 4 and TPRX1-positive cells" is ambiguous — clarify whether this refers to two distinct but overlapping groups, or a single population.

• Line 258: "Predominantly" somewhat overstates the label-transfer result as shown; consider softening or quantifying precisely.

Competing interests

The author declares that they have no competing interests.

Use of Artificial Intelligence (AI)

The author declares that they did not use generative AI to come up with new ideas for their review.