PIK3CA-related overgrowth spectrum (PROS) zebrafish models reveal pan-lineage developmental dysregulation

  1. Hannah Brunsdon
  2. Nuoya Wang
  3. Micha Sam Brickman Raredon
  4. Ralitsa R Madsen
  5. Robert K Semple
  6. E Elizabeth Patton

  • Curated by eLife

    eLife logo

    eLife Assessment

    This is an important study that establishes a zebrafish model of PIK3CA-related overgrowth syndrome. The imaging characterization of the mesodermal, particularly vascular, lesions of the model is compelling. The scRNA-Seq analysis is convincing, revealing key perturbations in the PIK3CA-mutation model, although deeper investigation of the exact mechanism leading to the lesions, as well as validation at different time points, could further strengthen the findings. This work will be of interest to medical biologists working on PROS, and potentially to a broader audience interested in non-cell-autonomous signaling of PIK3CA and its implications in other diseases.

Reviewed by

Read the full article See related articles

Discuss this preprint

Start a discussion What are Sciety discussions?

Listed in

Log in to save this article

Abstract

Post-zygotic gain-of-function PIK3CA mutations arising during embryonic development cause disorders collectively known as the PIK3CA -related overgrowth spectrum (PROS). This ranges from minor, localized overgrowth to devastating multi-tissue overgrowth. Disease severity is widely attributed to a combination of PIK3CA genotype, affected cell type, and developmental timing of mutation acquisition. However, in PROS this explains neither the biased pattern of overgrowth - disproportionately affecting mesoderm and neuroectoderm-derived tissues - nor the typical low mutation burden within areas of extensive tissue overgrowth. Indeed, growing evidence suggests PROS mutations cause overgrowth non-cell-autonomously, although mechanisms of this are poorly understood. Here, we develop mosaic zebrafish models of PROS with overexpression of orthologous hotspot pik3ca mutations ( pik3ca PROS ) to visualize and examine the effects of mutated cells on early development, in whole live animals. Reminiscent of PROS, we observe a spectrum of embryonic vasculature malformations (VMs), accompanied by larval muscle and bone overgrowth. Surprisingly, VMs only rarely expressed pik3ca PROS in constituent endothelial cells, with pik3ca PROS -expressing cells often closely abutting malformations instead. Single-cell transcriptomics of pik3ca PROS mosaic zebrafish prior to VM onset revealed that most pik3ca PROS cells are relatively immature and developmentally inert or constitute a small minority of mesodermal-derived cell types. Despite this constriction, global changes to cell fate were evident, alongside pervasive, pan-lineage abnormalities of gene expression, and rewiring of predicted ligand-receptor communication between lineages. We propose that targeting signals that indirectly propagate overgrowth through non-cell autonomous mechanisms - as well as PI3K activation itself - is worthy of therapeutic investigation.

Significance statement

Patchy, or mosaic, activating mutations in PIK3CA cause asymmetric human overgrowth due to aberrant hyperactivation of phosphoinositide 3-kinase (PI3K) signaling. Overgrowth prominently affects blood vessels and may be severely debilitating. While causation by PIK3CA mutations is clear, the explanation for the extent and pattern of associated overgrowth is not. Leveraging the power of zebrafish models for observation of early development, we now provide evidence for widely pervasive developmental effects of PIK3CA mutations extending beyond transcriptionally and phenotypically affected cells and lineages. This suggests potential therapeutic value of targeting secondary effects of PIK3CA activation as well as the activated PI3K itself.

Article activity feed

  1. eLife

    eLife Assessment

    This is an important study that establishes a zebrafish model of PIK3CA-related overgrowth syndrome. The imaging characterization of the mesodermal, particularly vascular, lesions of the model is compelling. The scRNA-Seq analysis is convincing, revealing key perturbations in the PIK3CA-mutation model, although deeper investigation of the exact mechanism leading to the lesions, as well as validation at different time points, could further strengthen the findings. This work will be of interest to medical biologists working on PROS, and potentially to a broader audience interested in non-cell-autonomous signaling of PIK3CA and its implications in other diseases.

  2. eLife

    Reviewer #1 (Public review):

    Summary:

    Brunsdon et al. present a zebrafish model of mosaic PIK3CA activation to investigate mechanisms underlying PIK3CA-related overgrowth spectrum (PROS), with a particular focus on non-cell-autonomous mechanisms of tissue overgrowth. The study is timely and addresses an important gap in the understanding of how mosaic activation of PI3K signaling leads to tissue-specific developmental abnormalities.

    Using a Tol2-based mosaic expression system combined with single-cell transcriptomics, the authors provide evidence suggesting that mutant PIK3CA-expressing cells influence surrounding wild-type tissues through indirect signaling mechanisms, contributing to vascular malformations and tissue overgrowth.

    Overall, the work presents an interesting and potentially impactful model for studying mosaic PIK3CA-driven overgrowth and non-cell-autonomous signaling mechanisms. However, several aspects require clarification, additional controls, and improved presentation to strengthen the mechanistic conclusions and overall impact of the study.

    Strengths:

    This study addresses an important and timely question by investigating the mechanisms underlying mosaic PIK3CA activation in the context of PROS, a condition for which developmental mechanisms remain poorly understood. The use of a mosaic zebrafish model is particularly appropriate, as it closely reflects the mosaic nature of PIK3CA mutations observed in patients and allows the investigation of non-cell-autonomous effects.

    Another major strength of the study is the integration of single-cell transcriptomics, which provides valuable insight into potential signaling pathways involved in indirect tissue overgrowth and offers a rich dataset for hypothesis generation. The authors also propose an interesting conceptual framework in which PI3K-activated cells influence surrounding tissues through paracrine signaling, which could have broader implications beyond PROS and contribute to understanding mosaic developmental disorders more generally.

    Finally, the work has potential translational relevance, as identifying mechanisms driving mosaic PI3K activation and non-cell-autonomous signaling could inform future therapeutic strategies for PROS and related conditions.

    Weaknesses:

    Despite these strengths, several aspects of the study require clarification and additional experimentation.

    Major comments:

    (1) The Tol2-based system results in mosaic overexpression of mutant PIK3CA in the presence of endogenous wild-type PIK3CA, making it difficult to determine how co-expression of WT and mutant proteins influences the observed phenotypes. While mosaic expression is relevant to PROS, a complementary approach in which endogenous PIK3CA is knocked out prior to introducing mutant variants would allow clearer interpretation of mutant-specific effects.

    (2) The authors do not clearly describe the validation of editing or integration efficiency. It would be important for the authors to clarify whether sequencing was performed to confirm integration, to quantify the proportion of mosaic expression, and to measure transgene expression levels. These controls would strengthen confidence in the model and interpretation of the results.

    (3) The manuscript would benefit from rescue experiments to strengthen causal conclusions. It remains unclear whether the phenotypes induced by PIK3CA PROS variants can be rescued, either through expression of wild-type PIK3CA, pharmacological inhibition of PI3K signaling, or assessment of developmental reversibility. Such experiments would strengthen the link between PI3K activation and the observed phenotypes.

    (4) The authors propose candidate signaling molecules mediating non-cell-autonomous effects downstream of PI3K hyperactivation; however, these conclusions remain speculative, as no functional validation is provided. Testing selected candidate mediators identified in the RNA-seq dataset would significantly strengthen the mechanistic conclusions.

  3. eLife

    Reviewer #2 (Public review):

    In this manuscript, Burnsdon et al. aim to study PIK3CA-related overgrowth spectrum (PROS) by establishing a mosaic zebrafish model with overexpression of pik3ca carrying hotspot mutations, coupled with an mScarlet+ reporter. Using fluorescence microscopy, the authors demonstrated that overexpression of pik3ca with a number of hotspot mutations led to mesodermal and particularly vascular malformations in the zebrafish model. Interestingly, they found a paucity of mScarlet+ mutant cells in the vascular lesions, consistent with the finding of low PIK3CA mutation burden in PROS tissue. Such data suggest a non-cell-autonomous effect of PIK3CA mutation. Following this logic, the authors performed single-cell RNA-Sequencing on zebrafish overexpressing WT pik3ca and mutant pik3ca at 19 hpf, and demonstrated widespread transcriptomic perturbations across multiple lineages, including lineage frequencies, key cell pathways, and cell-cell interactions. Importantly, they demonstrate that mScarlet+ cells carrying mutant pik3ca cluster separately from other cell types, do not demonstrate clear lineage identity, and have a general downregulation in signaling components.

    Overall, the conclusions in the manuscript are well-supported by the presented data. The imaging studies are particularly convincing. The transcriptomic analysis generated a list of potential pathways to further investigate and potentially target with future therapeutic interventions. Importantly, this study provides a valuable in vivo model of PROS that: 1) recapitulates key features of PROS (e.g., multiple mesodermal defects, paucity of mutation burden in lesions suggesting non-cell-autonomous interactions); 2) is scalable; and 3) offers direct visualization of lesion development, compatible with time-course live imaging. This model will be valuable to further understand PROS and potentially study other diseases where the PIK3CA pathway is altered (e.g., certain cancers).

    The following are not necessarily weaknesses of the data, but rather suggestions where the manuscript could be further strengthened:

    (1) The model recapitulates the variability of mesodermal lesions in PROS. It would be valuable to utilize this model to further study factors that are associated with the development of more severe lesions (e.g., by comparing samples with more severe lesions to those unaffected despite carrying the mutations, Figure 1F).

    (2) ScRNA-seq analysis could be enriched with a comparison between cells overexpressing mutant pik3ca vs. those overexpressing WT pik3ca.

    (3) In the scRNA-Seq analysis, it is curious that the C0 cluster, enriched with mScarlet+ cells, is found to have downregulated signaling interactions (Fig. 5C), yet exerts a widespread non-cell-autonomous effect. Meanwhile, there is also a noticeable loss of certain lineages (e.g., notochord, Figure 4E) and related cell-cell interactions (e.g., notochord-related interaction, Figure 5A). A deeper exploration of the basis of the non-cell-autonomous effect would be valuable.

    (4) The scRNA-Seq analysis was performed at one time point (19 hpf). Additional analysis (not necessarily by scRNA-Seq) at other time points to study whether findings at 19 hpf are persistent throughout development or undergo dynamic changes (e.g., cell fate/state of mSc+ mutant cells) would be helpful.

    (5) The scRNA-Seq analysis provides a valuable list of perturbed interactions that could be targeted by future therapeutic approaches. Validation of the scRNA-Seq findings with protein-level analysis, and studying the effect of targeting some of the pathways on the disease phenotype, would offer valuable data for the community.

  4. eLife

    Reviewer #3 (Public review):

    Summary:

    The study "PIK3CA-related overgrowth spectrum (PROS) zebrafish models reveal pan-lineage developmental dysregulation" presents important findings that extend significantly beyond a single subfield, bridging developmental biology, vascular medicine, and cancer-related PI3K signalling. By developing mosaic zebrafish models of PROS and combining live imaging with single-cell transcriptomics, the authors provide compelling evidence for a non-cell-autonomous mechanism of tissue overgrowth, a conceptual shift with meaningful therapeutic implications.

    Strengths:

    The evidence is overall convincing, with methodology appropriate and well-validated relative to the current state of the art; the integration of multiple approaches (in vivo modelling, scRNA-seq, ligand-receptor inference) strengthens the central claims. However, some aspects of the proposed non-cell-autonomous signalling mechanisms remain partly correlative, and direct functional validation of the rewired ligand-receptor interactions would further consolidate the conclusions.

    Weaknesses:

    The transgenic overexpression approach chosen by the authors represents a well-established and effective strategy for generating mosaic models in zebrafish. However, this approach introduces notable limitations: the lack of control over transgene dosage and unknown integration sites may generate non-physiological effects, potentially confounding the interpretation of key findings.

    The authors are certainly aware that alternative approaches (though technically more demanding) could be considered in future studies to further strengthen the model. For instance, a CRISPR/Cas9-mediated knock-in of the pik3ca-PROS allele at the endogenous locus (retaining upstream native regulatory elements with only a minimal promoter in the construct, co-expressed with a fluorescent reporter via P2A) could allow even more physiological, lineage-restricted expression while enabling direct visualisation of mutant cells. Mesodermal specificity could potentially be further refined by driving mosaic Cas9 expression under a pan-mesodermal tbx promoter, restricting editing to the relevant lineage while simultaneously marking mutant cells fluorescently, thus even more closely mimicking the post-zygotic mutational events characteristic of PROS. As a complementary strategy, blastula transplantation experiments using pik3ca-PROS donor cells (ideally co-expressing a distinct fluorescent marker such as mCherry) into fli1:GFP transgenic hosts could provide a powerful and technically consolidated approach to directly visualise and quantify non-cell-autonomous effects on host vasculature, with precise control over mutant cell burden. This combinatorial framework, separating donor mutant cells from host tissue in a two-colour imaging setup, could be particularly compelling for validating the ligand-receptor rewiring predicted by single-cell transcriptomics in future investigations.

    These reflections are offered in the spirit of prospective methodological development and do not diminish the value of the current work, which opens a valuable new avenue for therapeutic investigation, suggesting that targeting indirect overgrowth-propagating signals, alongside PI3K inhibition, deserves serious consideration.

  5. eLife

    Author response:

    eLife Assessment

    This is an important study that establishes a zebrafish model of PIK3CA-related overgrowth syndrome. The imaging characterization of the mesodermal, particularly vascular, lesions of the model is compelling. The scRNA-Seq analysis is convincing, revealing key perturbations in the PIK3CA-mutation model, although deeper investigation of the exact mechanism leading to the lesions, as well as validation at different time points, could further strengthen the findings. This work will be of interest to medical biologists working on PROS, and potentially to a broader audience interested in non-cell-autonomous signaling of PIK3CA and its implications in other diseases.

    We are delighted that the Editors and Reviewers consider the work of value and that it is interesting to a broad audience. We also appreciate and take on board the areas that the reviewers identify for improvement, and their suggestions on how this could be achieved.

    There are two major pieces of work suggested by the reviewers which we plan to carry out for this manuscript. The first of these is an additional scRNA-seq experiment at a later developmental stage when vascular malformations are established. Through comparison between pik3caPROS, pik3caWT and no-pik3ca injected controls, this would help answer if the global lineage and transcriptional dysregulation observed at 19 hpf persists over time, and if the largely inert ‘C0’ cluster of PROS mScarlet+ cells changes during development (Reviewer 2 comment 3).

    Secondly, we are already optimising rescue experiments with the specific Pik3ca inhibitor alpelisib, which is currently used as a therapy for PROS. Some troubleshooting has been required for the best delivery method and concentration for this to rescue vascular malformations in embryos, and to cause measurable decreases in PI3K signalling at the protein level through Akt and S6 pathways.

    Public Reviews:

    Reviewer #1 (Public review):

    Summary:

    Brunsdon et al. present a zebrafish model of mosaic PIK3CA activation to investigate mechanisms underlying PIK3CA-related overgrowth spectrum (PROS), with a particular focus on non-cell-autonomous mechanisms of tissue overgrowth. The study is timely and addresses an important gap in the understanding of how mosaic activation of PI3K signaling leads to tissue-specific developmental abnormalities.

    Using a Tol2-based mosaic expression system combined with single-cell transcriptomics, the authors provide evidence suggesting that mutant PIK3CA-expressing cells influence surrounding wild-type tissues through indirect signaling mechanisms, contributing to vascular malformations and tissue overgrowth.

    Overall, the work presents an interesting and potentially impactful model for studying mosaic PIK3CA-driven overgrowth and non-cell-autonomous signaling mechanisms. However, several aspects require clarification, additional controls, and improved presentation to strengthen the mechanistic conclusions and overall impact of the study.

    We thank Reviewer 1 for their support of our work, and constructive and helpful comments.

    Strengths:

    This study addresses an important and timely question by investigating the mechanisms underlying mosaic PIK3CA activation in the context of PROS, a condition for which developmental mechanisms remain poorly understood. The use of a mosaic zebrafish model is particularly appropriate, as it closely reflects the mosaic nature of PIK3CA mutations observed in patients and allows the investigation of non-cell-autonomous effects.

    Another major strength of the study is the integration of single-cell transcriptomics, which provides valuable insight into potential signaling pathways involved in indirect tissue overgrowth and offers a rich dataset for hypothesis generation. The authors also propose an interesting conceptual framework in which PI3K-activated cells influence surrounding tissues through paracrine signaling, which could have broader implications beyond PROS and contribute to understanding mosaic developmental disorders more generally.

    Finally, the work has potential translational relevance, as identifying mechanisms driving mosaic PI3K activation and non-cell-autonomous signaling could inform future therapeutic strategies for PROS and related conditions.

    Weaknesses:

    Despite these strengths, several aspects of the study require clarification and additional experimentation.

    Major comments:

    (1) The Tol2-based system results in mosaic overexpression of mutant PIK3CA in the presence of endogenous wild-type PIK3CA, making it difficult to determine how co-expression of WT and mutant proteins influences the observed phenotypes. While mosaic expression is relevant to PROS, a complementary approach in which endogenous PIK3CA is knocked out prior to introducing mutant variants would allow clearer interpretation of mutant-specific effects.

    PROS/CLOVES patients co-express endogenous wild-type and mutant PIK3CA in affected cells, which in turn constitute only a small proportion of cells in affected tissues (Madsen et al. 2018). As our intent was strictly to model human PROS/CLOVES (an aim informed by support from and close collaboration with the CLOVES Syndrome Community, a key patient advocacy group), we designed our model to reflect this as closely as possible. It is not clear to us what translational end would be served by expressing mutants in a null background, interesting though this may be. Given our transgenic strategy, we did experiment with overexpressing wildtype pik3ca as a control for some experiments to test whether overexpression of pik3ca itself drives overgrowth phenotypes, without the presence of hotspot PROS mutations (Figure 3D, Supplementary Figure 1A). We found that ubiquitous or mesodermal overexpression of pik3caWT did not cause vascular malformations or cause the ectopic fli1:eGFP endothelial cell phenotype observed when overexpressing pik3caPROS variants. While not precisely addressing the reviewer’s comment, this adds to evidence that increased expression of wildtype pik3ca does not confound the observed gain of function phenotype in the PROS model.

    (2) The authors do not clearly describe the validation of editing or integration efficiency. It would be important for the authors to clarify whether sequencing was performed to confirm integration, to quantify the proportion of mosaic expression, and to measure transgene expression levels. These controls would strengthen confidence in the model and interpretation of the results.

    We used secondary transgenesis markers, such as the cardiac reporter cmlc2:GFP, as a visual readout of integration efficiency and confirmation of integration – for example, embryos with >50% of GFP+ heart cells indicates that Tol2 transgenesis has occurred efficiently and so these would be included in an experiment, whereas the presence of only 1 or 2 green cardiac cells would suggest the levels of transgene in the embryo would be negligible and so this would be excluded from the experiment. Independently of this reporter, we showed an upregulation of pik3ca transcript in PROS mosaics compared to control by scRNA-seq (Figure 4D, Supplementary Figure 4A) confirming the transgene produces a measurable upregulation of pik3ca.

    We agree that it would be optimal to quantify the transgene expression and copy number for each individual embryo. However, for experiments where phenotypes are scored, hundreds of embryos are injected each time. Therefore, although it would be valuable to quantify the transgene expression and transgene copy number in terms of finding its correlation to phenotype severity, it is not feasible to do this at this scale. In the future, we would like to refine our model to include more sophisticated inducible transgenic models, with stable integration sites to control for integration site/copy number variation. However, for this manuscript, the priority as set out by our charity funders was to generate and characterise a pik3caPROS model that could rapidly test different patient hotspot alleles as well as tissue-specific promoter drivers. Thus, we chose this simpler model for now, but we would be very interested in continuing this work with a more refined model for one or two mutations (See Reviewer comment 1).

    This heterogeneity in transgene dosage and expression levels will inevitably have introduced ‘noise’ into our data. We can account for this somewhat by large numbers of embryos injected per experiment and reproducibility across populations of zebrafish between experiments. We also note that this strategy reflects the heterogeneity in human PROS, with disease mosaicism, presentation, and severity being highly variable from person to person. Therefore, we don’t necessarily see this as a drawback for our current approach.

    (3) The manuscript would benefit from rescue experiments to strengthen causal conclusions. It remains unclear whether the phenotypes induced by PIK3CA PROS variants can be rescued, either through expression of wild-type PIK3CA, pharmacological inhibition of PI3K signaling, or assessment of developmental reversibility. Such experiments would strengthen the link between PI3K activation and the observed phenotypes.

    We agree this is an exciting direction and a great next step for this research to take. This work is currently ongoing, using the specific Pik3ca inhibitor alpelisib, and optimizing treatment conditions to ensure our experimental readouts are meaningful. Through phenotype scoring we do see a significant rescue in the severity of vascular malformations in PROS mosaic embryos. However, we didn’t feel this work was ready for the initial submission because (1) the concentrations we must add to the zebrafish medium by immersion are far higher than the doses needed for inhibition of PI3K signalling in human cell lines and (2) we do not see an obvious decrease in pAkt or pS6 levels by western blot analyses of embryos at alpelisib doses of up to 100 μM, for either short or long term exposure. This drug is poorly soluble in water, and so we are also experimenting with introducing it to embryos intravenously.

    (4) The authors propose candidate signaling molecules mediating non-cell-autonomous effects downstream of PI3K hyperactivation; however, these conclusions remain speculative, as no functional validation is provided. Testing selected candidate mediators identified in the RNA-seq dataset would significantly strengthen the mechanistic conclusions.

    We thank the reviewer for this suggestion, and it is indeed a long-term aim of our work to find better treatments for PROS by combining inhibition of PI3K signalling with other candidate mediators to treat overgrowth. Our scRNA-seq experiments suggest that Notch, Wnt and Ephrin signalling pathway components may contribute to disease, and so a lot of potential for treatment strategies. After we have optimised treatment with alpelisib to rescue our disease phenotype in line with current mammalian models (see response to Comment 3 above), then we will start to look at other candidate mediators alone or in conjunction with alpelisib. However, given the challenges we are facing with the alpelisib treatment, we may need to develop this work in a subsequent study.

    Reviewer #2 (Public review):

    In this manuscript, Brunsdon et al. aim to study PIK3CA-related overgrowth spectrum (PROS) by establishing a mosaic zebrafish model with overexpression of pik3ca carrying hotspot mutations, coupled with an mScarlet+ reporter. Using fluorescence microscopy, the authors demonstrated that overexpression of pik3ca with a number of hotspot mutations led to mesodermal and particularly vascular malformations in the zebrafish model. Interestingly, they found a paucity of mScarlet+ mutant cells in the vascular lesions, consistent with the finding of low PIK3CA mutation burden in PROS tissue. Such data suggest a non-cell-autonomous effect of PIK3CA mutation. Following this logic, the authors performed single-cell RNASequencing on zebrafish overexpressing WT pik3ca and mutant pik3ca at 19 hpf, and demonstrated widespread transcriptomic perturbations across multiple lineages, including lineage frequencies, key cell pathways, and cell-cell interactions. Importantly, they demonstrate that mScarlet+ cells carrying mutant pik3ca cluster separately from other cell types, do not demonstrate clear lineage identity, and have a general downregulation in signaling components.

    Overall, the conclusions in the manuscript are well-supported by the presented data. The imaging studies are particularly convincing. The transcriptomic analysis generated a list of potential pathways to further investigate and potentially target with future therapeutic interventions. Importantly, this study provides a valuable in vivo model of PROS that: 1) recapitulates key features of PROS (e.g., multiple mesodermal defects, paucity of mutation burden in lesions suggesting non-cell-autonomous interactions); 2) is scalable; and 3) offers direct visualization of lesion development, compatible with time-course live imaging. This model will be valuable to further understand PROS and potentially study other diseases where the PIK3CA pathway is altered (e.g., certain cancers).

    We thank Reviewer 2 for their careful reading and support of our manuscript, and their helpful suggestions.

    The following are not necessarily weaknesses of the data, but rather suggestions where the manuscript could be further strengthened:

    (1) The model recapitulates the variability of mesodermal lesions in PROS. It would be valuable to utilize this model to further study factors that are associated with the development of more severe lesions (e.g., by comparing samples with more severe lesions to those unaffected despite carrying the mutations, Figure 1F).

    This is a very interesting question, and something that we have wondered ourselves. The clinical observation that PROS mutations cause pathology in mesodermal-derived tissues suggests that there is a lineage permissivity of PROS mutations. We plan to perform additional scRNA-seq experiments on later stage embryos (aligned with Figure 1) and hope to incorporate comparison of embryos with more severe lesions to those unaffected despite carrying pik3caPROS mutations.

    (2) ScRNA-seq analysis could be enriched with a comparison between cells overexpressing mutant pik3ca vs. those overexpressing WT pik3ca.

    The scRNA-seq experiment presented in this paper was limited by funding constraints at the time, and so we focussed on choosing samples that were likely to yield the most meaningful data. Ideally, we would have included a WT overexpression control in addition to an injected no-pik3ca control, however as we did not observe any phenotypes associated with mosaic pik3caWT transgenic embryos (Supplementary Figure 1A, Figure 3D), we chose to not include this condition. We are grateful for subsequent funding that will allow us to perform a scRNAseq experiment at a later timepoint, detailed below, where we plan to include this control.

    (3) In the scRNA-Seq analysis, it is curious that the C0 cluster, enriched with mScarlet+ cells, is found to have downregulated signaling interactions (Fig. 5C), yet exerts a widespread noncell-autonomous effect. Meanwhile, there is also a noticeable loss of certain lineages (e.g., notochord, Figure 4E) and related cell-cell interactions (e.g., notochord-related interaction, Figure 5A). A deeper exploration of the basis of the non-cell-autonomous effect would be valuable.

    Thank you for this important comment. We agree that this finding is very interesting and warrants further investigation, although a definitive answer may be too difficult for this current revision. Using conventional differential expression analyses on our scRNA-seq data (such as was used in Figure 4), we could not find significant upregulation of many genes and pathways, and CellChat and NICHES analyses did suggest that signalling between C0 and other clusters was weak. Nevertheless, using the Decoupler package, we did find significant upregulation of some footprint signatures enriched in mScarlet+ vs - cells in PROS mosaics (Supplementary Figure 4B) including PI3K and EGFR (as one would expect), but also apoptosis and UV response suggesting that overexpression of pik3caPROS may cause cellular stress. Using NICHES, we also found Myc, Notch, Wnt and Ephrin ligand-receptor pairs to be upregulated in PROS mosaic C0 sending and receiving interactions compared to controls, which would be candidates for validating in subsequent studies (Supplementary Figure 4C). We will be interested to determine if C0 like cells are present in older embryos in our scRNA-seq analysis, and if they have similar signalling activity.

    (4) The scRNA-Seq analysis was performed at one time point (19 hpf). Additional analysis (not necessarily by scRNA-Seq) at other time points to study whether findings at 19 hpf are persistent throughout development or undergo dynamic changes (e.g., cell fate/state of mSc+ mutant cells) would be helpful.

    We agree that the inclusion of a later timepoint in our scRNA-seq experiment would be valuable in answering a lot of our questions about the fate of C0 cells and the persistence of the transcriptional dysregulation, including non-cell autonomous interactions that we see at 19 hpf. As mentioned above, we were constrained by time and funding for the original experiment but are now in a position to add to this work and address this point.

    (5) The scRNA-Seq analysis provides a valuable list of perturbed interactions that could be targeted by future therapeutic approaches. Validation of the scRNA-Seq findings with proteinlevel analysis, and studying the effect of targeting some of the pathways on the disease phenotype, would offer valuable data for the community.

    Thank you for this comment. We agree that this an essential next step to take and is also a priority for our patient advocates. As mentioned above (Reviewer 1, point 4), we would like to be confident that alpelisib is on-target in our system first, and then we very much want to identify new therapeutic venues to explore in this pre-clinical space.

    Reviewer #3 (Public review):

    Summary:

    The study "PIK3CA-related overgrowth spectrum (PROS) zebrafish models reveal panlineage developmental dysregulation" presents important findings that extend significantly beyond a single subfield, bridging developmental biology, vascular medicine, and cancerrelated PI3K signalling. By developing mosaic zebrafish models of PROS and combining live imaging with single-cell transcriptomics, the authors provide compelling evidence for a noncell-autonomous mechanism of tissue overgrowth, a conceptual shift with meaningful therapeutic implications.

    We thank Reviewer 3 for their time and thoughtful comments considering our work.

    Strengths:

    The evidence is overall convincing, with methodology appropriate and well-validated relative to the current state of the art; the integration of multiple approaches (in vivo modelling, scRNAseq, ligand-receptor inference) strengthens the central claims. However, some aspects of the proposed non-cell-autonomous signalling mechanisms remain partly correlative, and direct functional validation of the rewired ligand-receptor interactions would further consolidate the conclusions.

    Weaknesses:

    The transgenic overexpression approach chosen by the authors represents a well-established and effective strategy for generating mosaic models in zebrafish. However, this approach introduces notable limitations: the lack of control over transgene dosage and unknown integration sites may generate non-physiological effects, potentially confounding the interpretation of key findings.

    Thank you for this important comment. We agree that there are limitations in our current model, and we are working to refine it such that we have temporal as well as spatial control over the expression of pik3caPROS.

    Our funding for the start of this study came from the CLOVES Syndrome community charity, and in collaboration with them, we decided that for this work, our priority was to understand more about the disease mechanisms at disease onset, and also to be able to test multiple pik3ca hotspot mutations that affect patients. One question for families is if the pik3ca hotspot mutations contribute differently to patient overgrowths. Our data here suggests that all mutations are able to promote overgrowth equally, and that differences between disease presentation in patients likely reflects the timing and cellular origins of the mutation.

    As a side note, together with CLOVES Syndrome community, we also felt that we wanted to test actual patient mutations, rather than artificial hyperactivated variants of Pik3ca such as the widely used p110a* allele (Hu et al. 1995; Venot et al. 2018), which can inform important mechanisms about pathway dysregulation, but less about actual patient-specific disease mutations.

    The authors are certainly aware that alternative approaches (though technically more demanding) could be considered in future studies to further strengthen the model. For instance, a CRISPR/Cas9-mediated knock-in of the pik3ca-PROS allele at the endogenous locus (retaining upstream native regulatory elements with only a minimal promoter in the construct, co-expressed with a fluorescent reporter via P2A) could allow even more physiological, lineage-restricted expression while enabling direct visualisation of mutant cells. Mesodermal specificity could potentially be further refined by driving mosaic Cas9 expression under a pan-mesodermal tbx promoter, restricting editing to the relevant lineage while simultaneously marking mutant cells fluorescently, thus even more closely mimicking the postzygotic mutational events characteristic of PROS. As a complementary strategy, blastula transplantation experiments using pik3ca-PROS donor cells (ideally co-expressing a distinct fluorescent marker such as mCherry) into fli1:GFP transgenic hosts could provide a powerful and technically consolidated approach to directly visualise and quantify non-cell-autonomous effects on host vasculature, with precise control over mutant cell burden. This combinatorial framework, separating donor mutant cells from host tissue in a two-colour imaging setup, could be particularly compelling for validating the ligand-receptor rewiring predicted by single-cell transcriptomics in future investigations.

    These reflections are offered in the spirit of prospective methodological development and do not diminish the value of the current work, which opens a valuable new avenue for therapeutic investigation, suggesting that targeting indirect overgrowth-propagating signals, alongside PI3K inhibition, deserves serious consideration.

    Thank you for these excellent suggestions and feedback. We are keen to try to generate fish that more closely align with what is happening in patients. Two challenges we have faced include:

    (1) In our hands, the pik3ca promoter itself is not strong enough to drive fluorophore expression to an extent that we can observe fluorescent PROS cells in zebrafish. As a control, after we saw no fluorescence attempting to knock-in fluorophores at the 5’ end of endogenous pik3ca, we tried making a transgenic using various lengths of pik3ca promoter regions driving GFP expression. Despite having stable integration of the transgene shown by a secondary transgene reporter inherited through to F1 generation, we could not visualise GFP/mNeonGreen expression at any stage of development.

    (2) A drawback of the IRES approach we used here is that the fluorophore expression levels will be lower than using a short cleavable peptide sequence such as P2A. Unfortunately, the critical kinase region (and location of the orthologous hotspot codon 1048) is located only a few amino acids from the stop codon, and we found that the function of Pik3ca was likely impeded by the addition of several extra amino acids after the P2A cleaves itself.

    Despite these challenges, we hope to be able to generate models in future with more precise control over mutant cell burden.

    References

    Hu Q, Klippel A, Muslin AJ, Fantl WJ, Williams LT. 1995. Ras-dependent induction of cellular responses by constitutively active phosphatidylinositol-3 kinase. Science 268: 100102.

    Madsen RR, Vanhaesebroeck B, Semple RK. 2018. Cancer-Associated PIK3CA Mutations in Overgrowth Disorders. in Trends in Molecular Medicine, pp. 856-870. Elsevier Ltd.

    Venot Q, Blanc T, Rabia SH, Berteloot L, Ladraa S, Duong JP, Blanc E, Johnson SC, Hoguin C, Boccara O et al. 2018. Targeted therapy in patients with PIK3CA-related overgrowth syndrome. Nature 558: 540-546.

  6. Version published to 10.64898/2025.12.03.691507 on bioRxiv