Ancestral secretory programs underlie the evolution of morphological innovations across Spiralia

Reviewer #1 (Public review):

Summary:

This manuscript examines the evolution of molluscan shells using single-cell analyses of the adult mantle of Crassostrea gigas and compares these data with previous datasets from embryonic and larval stages of this species and other spiralians. The authors provide support for a scenario in which secretory cells are broadly conserved across spiralians, and the incorporation of lineage-restricted genes contributes to the evolution of molluscan shells.

Strengths:

High-quality datasets for mantle tissue in Crassostrea gigas and thorough comparisons with existing datasets for this species and other spiralians. Balanced discussion.

Weaknesses:

No major weaknesses. The analyses follow fairly standard approaches in the field that have been previously applied and developed in similar systems.

Reviewer #2 (Public review):

Summary:

Bai et al. present in their study three single-cell RNA seq datasets derived from gastrulae, trochophores, and adults of the bivalve Crassostrea gigas. While a dataset on the oyster trochophore has already been published previously (Piovani et al. 2023), the gastrula and adult datasets have not been published yet. The authors conclude that cell types secreting the oyster shell valves use a genetic repertoire that is also used by epithelial and secretory cell types of very different spiralians, such as annelids, chaetognaths and flatworms.

Strengths:

The study provides new single-cell datasets from multiple developmental stages of an oyster, offering a valuable resource for the field. It takes a broad comparative approach using state-of-the-art techniques across diverse animal groups and addresses an important question regarding the origin and evolution of shell-forming cell types.

Weaknesses & suggestions to improve the manuscript:

(1) Validation of cell types

Cell type identities are not convincingly validated. Although the authors cite previous studies (l. 92), the referenced marker genes are largely not used, and the cited works do not provide sufficient spatial validation. Without in situ data, the inferred locations of cell types (e.g. Figure 2A) are not supported. Spatial validation of marker genes (e.g. via HCR) is essential, particularly for a study addressing shell field evolution. In addition, the gastrula dataset is not meaningfully analyzed, and its inclusion remains unclear.

(2) Robustness of cell type classification

Several proposed cell types may not represent distinct entities (not individuated) but rather reflect over-clustering. Marker genes are often not specific and are shared across clusters (e.g. Sec1/Sec2), making it difficult to distinguish cell types reliably.

(3) Comparative analysis of secretory cells

The comparative framework is not sufficiently supported. Secretory cells are highly diverse, and without proper validation, their comparison across taxa is not meaningful. The transcription factor analysis is limited, as only a few genes are shared and many are inconsistently expressed (Figure 3E). The conclusion of a conserved regulatory program across spiralians is therefore overstated.

(4) Clarity and interpretation of results

Results are at times difficult to follow and remain superficial. Marker genes are insufficiently annotated (especially for Crassostrea), and comparisons across taxa lack functional interpretation. Unvalidated and heterogeneous cell types are grouped together, and transcriptional similarities are overinterpreted. Overall, key conclusions are not adequately supported by the presented data.

Reviewer #3 (Public review):

Summary:

This manuscript by Bai et al. reports single-cell transcriptomics of the oyster mantle to elucidate the respective contributions of ancient conserved programmes and lineage-specific genes to the origin of the molluscan shell. The authors compare their dataset with other oyster larval datasets as well as data from other organisms (annelids, chaetognaths) and find evidence of evolutionary conservation and functional similarity with secretory cell types. They also observe that cells involved in secreting the larval skeleton express predominantly recent genes, whereas the adult skeleton-secreting programme is evolutionarily more conserved.

Strengths:

The manuscript is well written and clearly presented, and the results are interesting, particularly the distinction between larval and adult skeleton secretion, which is placed in a thoughtful evolutionary context.

Weaknesses:

(1) My main concern is that the authors rely primarily on previous studies for the experimental and functional characterisation of the identified cell types. The cited papers (Piovani, 2023 and de la Forest Divonne et al., 2025) deal with distinct stages or tissues (larvae and hemocytes, respectively), which limits their direct relevance. The authors also cite other papers for in situ expression data; it would be helpful to summarise somewhere (e.g. in a table) which genes have been experimentally characterised and what their expression domains are, or alternatively to provide HCR or in situ staining on the mantle. For instance, what is the rationale for the claim that proliferative cells give rise to the mantle? The trajectory inference approach used (Monocle) would likely yield a similar result regardless of the reference cell type, so additional justification is needed.

(2) More broadly, I find that the functional properties of the identified cell types and their relationship to the expressed genes deserve more detailed discussion. For example, at L100, several genes are mentioned, but their functional roles are not discussed. Similarly, the basis for annotating the proliferative cells is not explained. How was gene orthology assessed? Throughout the manuscript, vertebrate-style gene names are used without explicitly establishing orthology status in oyster, which should be addressed.

(3) More detail is needed on the methods and quality control for the single-cell data. The authors should clarify that the platform used (BMKMANU) is a droplet-based technology comparable in principle to Drop-seq. BMKMANU is not widely used in the field. How does it compare to 10x Genomics in terms of sensitivity and cell recovery? The authors appear to use the 10x Chromium cellranger pipeline for data analysis, which suggests compatibility, but this should be stated explicitly. Additionally, no information is provided on the number of sequencing runs or biological replicates, nor on how reproducible the results are across samples.

(4) A limitation of the phylostratigraphic analysis is that it is restricted to mantle tissue, making it difficult to place the results in a whole-organism context. How do the age profiles of mantle-expressed genes compare to those of more evolutionarily conserved tissues, such as the nervous system? I appreciate the methodological and experimental constraints, but this is a genuine limitation of the study. The authors could at least discuss it explicitly, and ideally consider generating a broader single-cell atlas of the oyster to provide this comparative baseline.

(5) Have the authors considered the potential importance of lineage-specific gene duplication? It is well established that spiralians, including oysters, have undergone extensive lineage-specific duplication of transcription factors such as homeobox genes, and many structural shell-associated proteins may similarly have been duplicated. This could be relevant to interpreting both the phylostratigraphic results and the expansion of secretory gene families.

Author response:

Public Reviews:

Reviewer #1 (Public review):

Summary:

This manuscript examines the evolution of molluscan shells using single-cell analyses of the adult mantle of Crassostrea gigas and compares these data with previous datasets from embryonic and larval stages of this species and other spiralians. The authors provide support for a scenario in which secretory cells are broadly conserved across spiralians, and the incorporation of lineage-restricted genes contributes to the evolution of molluscan shells.

Strengths:

High-quality datasets for mantle tissue in Crassostrea gigas and thorough comparisons with existing datasets for this species and other spiralians. Balanced discussion.

Weaknesses:

No major weaknesses. The analyses follow fairly standard approaches in the field that have been previously applied and developed in similar systems.

We thank the reviewer for the positive evaluation of our work. We are encouraged that the reviewer finds our conclusions balanced and the analyses appropriate. Although no major concerns were raised, we will incorporate clarifications and improvements prompted by the other reviewers to further strengthen the manuscript.

Reviewer #2 (Public review):

Weaknesses:

(1) Validation of cell types

Cell type identities are not convincingly validated. Although the authors cite previous studies (l. 92), the referenced marker genes are largely not used, and the cited works do not provide sufficient spatial validation. Without in situ data, the inferred locations of cell types (e.g. Figure 2A) are not supported. Spatial validation of marker genes (e.g. via HCR) is essential, particularly for a study addressing shell field evolution. In addition, the gastrula dataset is not meaningfully analyzed, and its inclusion remains unclear.

We thank the reviewer for this important comment regarding cell type validation. In the previous version of the manuscript, we provided a detailed compilation of referenced marker genes from previous studies in Supplementary File 2. It is possible that, due to an incorrect or unclear reference in the main text, this information was not readily accessible. We will correct and clarify these citations in the revised manuscript to ensure that these resources are clearly presented.

We agree that spatial validation would provide important support for cell type identities. In the revised version, we will strengthen this aspect by selecting more specific marker genes for each SEC cluster and performing fluorescence in situ hybridisation (FISH) to validate their spatial localization.

Regarding the gastrula dataset, our original intention was to investigate the developmental transition of shell gland-related cell populations from gastrula to trochophore stages. However, following the reviewer’s suggestion and considering the limited interpretability of the gastrula dataset in its current form, we agree that its inclusion does not substantially strengthen the study. We therefore plan to remove the gastrula dataset from the revised manuscript, and instead focus on the trochophore stage as a representative developmental stage for larval shell formation, enabling a clearer comparison between larval and adult shell-forming cell populations. We note that this change does not affect the main conclusions of the study. In addition, we will curate a refined set of experimentally supported marker genes, and provide an updated supplementary table summarizing detailed information, including cell type annotations, literature sources, and experimental validation methods.

(2) Robustness of cell type classification 

Several proposed cell types may not represent distinct entities (not individuated) but rather reflect over-clustering. Marker genes are often not specific and are shared across clusters (e.g. Sec1/Sec2), making it difficult to distinguish cell types reliably.

In the revised manuscript, we will refine marker gene selection by prioritizing genes with higher specificity and stronger discriminatory power to improve the robustness of cell type identification. To further support cell identity assignment, we will select representative marker genes for SEC clusters and perform FISH to validate their spatial localization. These revisions will lead to a more robust and conservative interpretation of cell populations.

(3) Comparative analysis of secretory cells

The comparative framework is not sufficiently supported. Secretory cells are highly diverse, and without proper validation, their comparison across taxa is not meaningful. The transcription factor analysis is limited, as only a few genes are shared and many are inconsistently expressed (Figure 3E). The conclusion of a conserved regulatory program across spiralians is therefore overstated.

We agree that secretory cell types are highly diverse across spiralians and that cross-species comparisons require careful interpretation. In the revised manuscript, we will adopt a more cautious framework, highlight partial conservation of regulatory program alongside functional convergence in secretory processes. We also will strengthen the comparative framework by integrating functional annotations, which may provide complementary support beyond individual gene overlaps. Importantly, we will improve the reliability of oyster SEC annotations through FISH-based spatial validation, thereby increasing confidence in cross-species comparisons. These revisions will provide a more balanced and biologically grounded interpretation of secretory cell evolution across spiralians.

(4) Clarity and interpretation of results

Results are at times difficult to follow and remain superficial. Marker genes are insufficiently annotated (especially for Crassostrea), and comparisons across taxa lack functional interpretation. Unvalidated and heterogeneous cell types are grouped together, and transcriptional similarities are overinterpreted. Overall, key conclusions are not adequately supported by the presented data.

In the revised manuscript, we will re-evaluate marker gene annotations to ensure support from existing experimental evidence. For SEC populations, we will validate representative markers using FISH. We will also expand the functional annotation of marker genes and strengthen cross-species comparisons. In addition, we will substantially revise the Results and Discussion sections to improve clarity and depth, reduce overinterpretation of transcriptional similarities, and ensure that all conclusions are more tightly aligned with the strength of the supporting evidence.

Reviewer #3 (Public review):

Weaknesses:

(1) My main concern is that the authors rely primarily on previous studies for the experimental and functional characterisation of the identified cell types. The cited papers (Piovani, 2023 and de la Forest Divonne et al., 2025) deal with distinct stages or tissues (larvae and hemocytes, respectively), which limits their direct relevance. The authors also cite other papers for in situ expression data; it would be helpful to summarise somewhere (e.g. in a table) which genes have been experimentally characterised and what their expression domains are, or alternatively to provide HCR or in situ staining on the mantle. For instance, what is the rationale for the claim that proliferative cells give rise to the mantle? The trajectory inference approach used (Monocle) would likely yield a similar result regardless of the reference cell type, so additional justification is needed.

We agree that our reliance on previous studies for functional and experimental characterization requires clearer justification and integration. In the revised manuscript, we will compile a new supplementary table summarizing marker genes with available experimental validation, including their associated cell types, literature sources, and experimental methods. For SEC populations, we will select representative marker genes and perform FISH to validate their spatial localization, thereby providing independent support for cell identity.

Regarding trajectory inference, we agree that methods such as Monocle are sensitive to assumptions. We will clarify the rationale for root cell selection, test alternative root assignments to assess robustness, and revise our interpretation to avoid strong lineage claims. Rather than stating that proliferative cells give rise to mantle cells, we will describe the observed trajectory as being consistent with a potential developmental relationship, while acknowledging that this does not constitute direct evidence of lineage progression.

(2) More broadly, I find that the functional properties of the identified cell types and their relationship to the expressed genes deserve more detailed discussion. For example, at L100, several genes are mentioned, but their functional roles are not discussed. Similarly, the basis for annotating the proliferative cells is not explained. How was gene orthology assessed? Throughout the manuscript, vertebrate-style gene names are used without explicitly establishing orthology status in oyster, which should be addressed.

We thank the reviewer for this important comment. In the revised manuscript, we will expand the functional interpretation of key genes by incorporating available literature and, where possible, functional annotations. We will also clarify the basis for cell type annotation and explicitly describe the criteria used, including for proliferative cell populations (e.g. cell proliferation-associated markers).

Regarding gene annotation, gene names in oyster were assigned based on sequence similarity searches against the eggNOG database. In the revised manuscript, we will provide a comprehensive supplementary table linking gene IDs to their annotations, along with the corresponding database sources. In addition, we will clearly describe how orthology relationships were assessed, including the methods and criteria used (e.g. sequence similarity searches and orthology databases). Throughout the revised manuscript, we will ensure that the use of vertebrate-style gene names is accompanied by appropriate annotation information and does not imply unsupported one-to-one orthology relationships.

(3) More detail is needed on the methods and quality control for the single-cell data. The authors should clarify that the platform used (BMKMANU) is a droplet-based technology comparable in principle to Drop-seq. BMKMANU is not widely used in the field. How does it compare to 10x Genomics in terms of sensitivity and cell recovery? The authors appear to use the 10x Chromium cellranger pipeline for data analysis, which suggests compatibility, but this should be stated explicitly. Additionally, no information is provided on the number of sequencing runs or biological replicates, nor on how reproducible the results are across samples.

In the revised manuscript, we will expand the Methods section to provide a clearer and more detailed description of the experimental and analytical procedures. BMKMANU is a droplet-based single-cell RNA-seq platform, conceptually comparable to Drop-seq and similar in principle to 10x Chromium. We will also explicitly state that the data generated are compatible with the Cell Ranger pipeline, which was used for downstream processing and analysis. Although BMKMANU is less widely used than 10x Genomics platforms, it has been successfully applied in several recent studies (e.g. Li et al., 2024: https://doi.org/10.1007/s11427-023-2548-3; Li et al., 2025: https://doi.org/10.1038/s41559-025-02642-6; Wei et al., 2024: https://doi.org/10.1038/s41467-024-46780-0), demonstrating its applicability for single-cell transcriptomic analyses across different biological systems. Regarding platform performance, based on technical information provided by the manufacturer, BMKMANU shows comparable sensitivity and cell capture efficiency to 10x Genomics platforms (http://www.biomarker.com.cn/zhizao/dg1000danxibao). In this study, the mantle sample was obtained from a single individual oyster and processed in a single sequencing run, without batch effects introduced by multiple runs. We will clearly state this in the revised manuscript. In addition, we will provide detailed quality control metrics, including the number of cells retained, gene detection rates, and filtering criteria.

(4) A limitation of the phylostratigraphic analysis is that it is restricted to mantle tissue, making it difficult to place the results in a whole-organism context. How do the age profiles of mantle-expressed genes compare to those of more evolutionarily conserved tissues, such as the nervous system? I appreciate the methodological and experimental constraints, but this is a genuine limitation of the study. The authors could at least discuss it explicitly, and ideally consider generating a broader single-cell atlas of the oyster to provide this comparative baseline.

We agree that restricting the phylostratigraphic analysis to mantle tissue represents a limitation when attempting to place our findings in a whole-organism evolutionary context. In the revised manuscript, we will explicitly acknowledge this limitation and expand the Discussion to address how gene age profiles in mantle tissue may differ from those in more evolutionarily conserved tissues. In particular, we will clarify that the enrichment of younger, lineage-specific genes observed in shell-forming cells may reflect tissue-specific functional specialization, and therefore should not be directly generalized to other cell types.

We acknowledge that a broader single-cell atlas spanning multiple tissues would provide an important comparative baseline for interpreting gene age patterns across the organism. While generating such a dataset is beyond the scope of the present study, we will highlight this as an important direction for future research.

(5) Have the authors considered the potential importance of lineage-specific gene duplication? It is well established that spiralians, including oysters, have undergone extensive lineage-specific duplication of transcription factors such as homeobox genes, and many structural shell-associated proteins may similarly have been duplicated. This could be relevant to interpreting both the phylostratigraphic results and the expansion of secretory gene families.

We thank the reviewer for this insightful suggestion. Lineage-specific gene duplication is likely to play an important role in shaping both transcription factor repertoires and shell-associated gene families in spiralians, including oysters. In the revised manuscript, we will incorporate a discussion of lineage-specific duplication, particularly in relation to transcription factors and biomineralization-related proteins. We will also, where feasible, explore its potential contribution to our observations and highlight how such duplications may drive the expansion and diversification of secretory gene families.