Human spinal cord organoids reveal cell intercalation as a conserved mechanism for secondary neurulation

Reviewer #1 (Public review):

Summary:

In this manuscript, Blanco-Ameijeiras et al. present an organoid-based model of the caudal neural tube that builds upon established principles from embryonic development and prior organoid work. By systematically testing and refining signaling conditions, the authors generate caudal progenitor populations that self-organize into neuroepithelia with molecular features consistent with secondary neurulation. Bulk-RNA sequencing supports the emergence of caudal neural identities, and the authors further examine cellular features such as apico-basal polarity and interkinetic nuclear migration. Finally, they provide evidence for a conserved, YAP-dependent mechanism of tube formation specific to secondary neurulation. The manuscript provides valuable methodological resources, including troubleshooting guidance that will be especially useful for the field. While this work represents a significant advance toward modeling human spinal cord development - particularly the process of secondary neurulation - the claims of complete caudalization and full AP-axis representation require additional experimental support and clarification.

Strengths:

(1) Methodological clarity and transparency: The first figure and accompanying text provide an exemplary explanation of protocol optimization and troubleshooting. This transparency - showing approaches that failed as well as those that succeeded - sets a high standard for reproducibility and will be highly beneficial to laboratories aiming to adopt or build upon this model.

(2) Testing across multiple cell lines: Multiple hPSC and hiPSC lines were evaluated, strengthening the robustness and generalizability of the reported protocol.

(3) Biological relevance: The focus on secondary neurulation fills a notable gap in current human organoid models of spinal cord development. The identification of YAP-dependent mechanisms in tube formation is a valuable insight with potential translational relevance.

(4) Resource creation: The detailed parameters and signaling regimes will serve as a resource for the spinal cord and organoid communities.

Weaknesses:

(1) The manuscript over-interprets bulk RNA-seq data to make strong claims on the organoid AP patterning and caudalization. Bulk sequencing provides population-level averages and cannot confirm that individual organoids represent discrete AP levels. To support claims of generating every AP identity, the authors must perform staining or in situ hybridization for HOX genes on individual organoids. Further, the current interpretation of CDX2 as marking "very distal" identity is inaccurate in vitro; CDX2 marks caudal progenitors across the spinal cord axis. The language should be revised accordingly.

(2) The claim of being the first organoid system to model secondary neurulation overlooks prior work showing HOXC9 in human organoids (Xue et al., Nature 2024; Libby et al., Development 2021), which would reflect the beginning of secondary neurulation. While this system may indeed be the first isolated secondary neurulation organoid model that expresses HOXD9/10 - a meaningful advance - bulk RNA-seq alone is insufficient to support the exclusivity of this claim. Additional single-organoid-level spatial analyses (via immunofluorescence of in situ hybridisation) and frequency quantification of regional identities are required to fully characterize the system.

(3) Similarly, as written, there are overstatements taken from the bulk RNA sequencing to determine dorsal-ventral identity. Although dorsal markers are present, the dataset also contains ventral-associated genes (PAX6, SP8, NKX6-1, NKX6-2, PRDM12). To claim a "dorsal-only" identity, the authors should perform PAX7 immunostaining to demonstrate dorsalization of the entire organoid tissue.

(4) The studies identifying YAP as a key driver of lumen fusion in Figure 6 are important and should be extended to the apical organoid system to demonstrate that this is truly a feature of secondary neurulation.

Reviewer #2 (Public review):

Summary:

In this study, Blanco-Ameijeiras and colleagues present the use of stem cells to create human spinal cord organoids that recapitulate anterior-posterior identity, with a large focus on posterior fates. In particular, the authors show robust transcriptional landscape specification that reflects certain anterior-posterior spinal cord development.

Recapitulation of spinal cord development is essential to understand the fundamentals of developmental defects in a systematic manner. This work provides a broad approach to test certain aspects of neural tube morphogenesis, particularly posterior and dorsal identities. Perhaps the shorter protocol is an interesting upgrade for current standards, and the mechanical interpretation provides good proof of concept work that aligns with the need to better understand neural tube mechanobiology.

Strengths:

The manuscript addresses a major gap by focusing on posterior spinal cord identity and secondary neurulation, a phase that is less well captured by existing neural tube organoid models (although some do recapitulate that). The manuscript situates the approach within vertebrate development and human embryology.

Morphometric quantifications are well described and provide a dynamic interpretation of cell-level interpretation, and that is a true strength of the work. This is important to develop important metrics that can later be used to compare modulations and pathway disruption.

The protocols are well described and documented.

Weaknesses:

Some key data lacks proper quantification to robustly support the claims. For example, it is not clear how many organoids in total are counted in Figure 1D to derive the % of organoids expressing certain markers (e.g. SOX2 or BRA).

Some claims are overstated. In the manuscript, the organoids show primarily dorsal and posterior identities under the current conditions, yet the discussion sometimes reads as if a more complete dorsoventral recapitulation is achieved. Therefore, one can either demonstrate ventral patterning (e.g., SHH / FOXA2) or reduce the claims about spinal cord identity, which, given the results, are more specific to a particular region.

The mention of anterior organoids seems to distract the reader from the important work, which primarily focuses on the posterior identity. Further, it is not understood why SOX2 identity is reduced by Day 7 in Figure 1D. Since SOX2 in the manuscript is considered a neural marker (although also pluripotency along with NANOG, etc.), a further explanation should be provided. The author should also test the presence of PAX6, which is one of the earliest neuroectoderm markers in humans (Zhang X. et al., Cell Stem Cell 2010).

The authors position the work as a substantial addition to the field. The work is very much welcomed; however, some claims align with an interpretation that leads the readers to understand a novelty that is beyond the work presented here. For example, in certain instances in the intro, the manuscript conveys that this work consists of the first recapitulation of spinal cord fates anterior or posterior, while other works (Rifes P. Nature Cell Biology 2020, Xue X. Nature 2024) recapitulate dorsoventral and anterior-posterior patterning and identity (albeit not of secondary neurulation) through controlled gradients of WNT and RA activity. To clearly position the importance of this work, the intro should focus on secondary neurulation and posterior identities.

In a similar fashion, the claim that "Importantly though, to our knowledge these are the first neural organoids exhibiting a robust spinal cord transcriptome identity" is not very well understood when other neural tube organoid systems (including spinal cord identities) have been exhaustively profiled at the single cell level (Rifes P. Xue X. Abdel Fattah A.). Further explanation is therefore needed.

The mechanical angle is important and adds to the large body of research that traces NT morphogenesis to mechanics. However, the YAP localization images can be much improved. Lower magnification images are needed to show the entire organoid to robustly convince the reader of the correct and varying localization of the YAP protein. The authors should also check for YAP-associated genes in their bulk RNA sequencing.

The quantification of the YAP analysis in a total of 23 and 18 cells in the two conditions and in 7 organoids is by no means enough to draw a conclusion about YAP localization, and an increase in the number of cells is needed. Moreover, the use of dasatinib as an inhibitor for YAP is great, but there is no evidence shown that in this culture system, the inhibitor actually inhibits YAP. As such, IF images are required to confirm cytosolic YAP. Additionally, the authors can try other inhibitors (such as verteporfin) since most inhibitors are broadband.

Given the mechanically oriented conclusions, other relevant works have shown posteriorized and ventralized neural tube organoids using RA and SHH activation, which were also mechanically stimulated via actuation, such as work done from the Ranga lab (Nature comm. 2021/2023). Although not strictly related to YAP, the therein molecular profiling, mechanical stimulation, lumen measurements, and NTD-like phenotype using PCP-mutated genes make these important relevant mentions since the current work adds important aspects with YAP analysis.

Reviewer #3 (Public review):

Summary:

In this manuscript, Blanco-Ameijeiras and collaborators describe the 3D differentiation of human pluripotent stem cells into the posterior spinal cord. The authors first test the exposure of different combinations of extrinsic signals to generate human neural organoids with distinct antero-posterior identities, as shown by bulk transcriptome analysis. They show that neural organoids, whether anterior or posterior, display tissue architecture, organisation and dynamics resembling the in vivo situation. Increasing the size of initial cell aggregates leads to the formation of a single lumen through a multi-lumen stage and a process of cell intercalation, mimicking the situation that they recently described for chick secondary neurulation (Gonzalez-Gobartt et al. Dev Cell. 2021 PMID: 33878300). The authors go on to show that, as in chick, YAP is involved in the resolution of multiple lumens into a single lumen. They conclude that their human organoid approach faithfully models human secondary neurulation, which may be instrumental in unravelling the mechanisms of human neural tube defects.

Strengths:

Overall, this is an important study demonstrating that lumen formation in human spinal organoids recapitulates key aspects of secondary neurulation observed in animal models. This organoid approach may be instrumental in unravelling the mechanisms of human neural tube defects.

Weaknesses:

The significance of the findings is tempered by several limitations. While the authors show convincing evidence that organoids undergo lumen formation with similar morphological, cellular and molecular features as seen in chick in their previous work (Gonzalez-Gobartt et al. Dev Cell. 2021 PMID: 33878300), whether this is linked to their caudal spinal cord identity is unclear.