Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.
Learn more at Review Commons
Reply to the reviewers
Review report for 'Sterols regulate ciliary membrane dynamics and hedgehog signaling in health and disease', Lamazière et al.
Reviewer #1
In this manuscript, Lamazière et al. address an important understudied aspect of primary cilium biology, namely the sterol composition in the ciliary membrane. It is known that sterols especially play an important role in signal transduction between PTCH1 and SMO, two upstream components of the Hedgehog pathway, at the primary cilium. Moreover, several syndromes linked to cholesterol biosynthesis defects present clinical phenotypes indicative of altered Hh signal transduction. To understand the link between ciliary membrane sterol composition and Hh signal transduction in health and disease, the authors developed a method to isolate primary cilia from MDCK cells and coupled this to quantitative metabolomics. The results were validated using biophysical methods and cellular Hh signaling assays. While this is an interesting study, it is not clear from the presented data how general the findings are: can cilia be isolated from different mammalian cell types using this protocol? Is the sterol composition of MDCK cells expected to the be the same in fibroblasts or other cell types? Without this information, it is difficult to judge whether the conclusions reached in fibroblasts are indeed directly related to the sterol composition detected in MDCK cells. Below is a detailed breakdown of suggested textual changes and experimental validations to strengthen the conclusions of the manuscript.
We would like to thank the reviewer for their helpful comments
Major comments:
- It appears that the comparison has been made between ciliary membranes and the rest of the cell's membranes, which includes many other membranes besides the plasma membrane. This significantly weakens the conclusions on the sterol content specific to the cilium, as it may in fact be highly similar to the rest of the plasma membrane. It is for example known that lathosterol is biosynthesized in the ER, and therefore the non-presence in the cilium may reflect a high abundance in the ER but not necessarily in the plasma membrane.
The reviewer is correct that we compared the sterol composition of the primary ciliary membrane to the average of the remaining cellular membranes. We agree that this broader reference fraction contains multiple intracellular membranes, including ER- and Golgi-derived compartments, and therefore does not isolate the plasma membrane specifically. We would like to emphasize that our study did not aim to compare the cilium directly to the plasma membrane, nor did we claim that the comparison was in any way related to the plasma membrane. It is also worth noting that previous studies in other ciliated organisms have reported a higher cholesterol content in cilia compared to the plasma membrane, suggesting that the two membranes may not be compositionally identical despite their continuity. However, we concur that determining the sterol composition of the MDCK plasma membrane would provide valuable context and enable a comparison with the membrane continuous with the ciliary membrane. Hence, we are willing to try isolating plasma membrane in the same cellular contexts.
- While the protocol to isolate primary cilium from MDCK cells is a valuable addition to the methods available, it would be good to at least include a discussion on its general applicability. Have the authors tried to use this protocol on fibroblasts for example?
Thank you for the reviewer's positive comment on the value of the ciliary isolation protocol. Indeed, we have attempted to apply the same approach to other ciliated cell types, namely IMCD3 and MEF cells. In the case of IMCD3 cells, we were able to isolate primary cilia using the same general strategy; however, we are still refining the preparation, as the overall yield is lower than in MDCK cells and the amount of material obtained is currently insufficient for comprehensive biochemical analyses. With MEF (fibroblast) cells, the procedure proved even more challenging, as the yield of isolated cilia was extremely low. This difficulty is likely due to the shorter length of fibroblast cilia and to their positioning beneath the cell body, which probably makes them more resistant to detachment. Overall, these observations suggest that while the protocol can be adapted to other cell types, its efficiency depends on cellular architecture. We have added a discussion of these aspects in the revised manuscript to clarify the method's current scope and limitations (lines 492-502).
- Some of the conclusions in the introduction (lines 75-80) seem to be incorrectly phrased based on the data: in basal conditions, ciliary membranes are already enriched in cholesterol and desmosterol, and the treatment lowers this in all membranes.
We agree, this was modified in the revised manuscript (lines 75-80).
- There seems to be little effect of simvastatin on overall cholesterol levels. Can the authors comment on this result? How would the membrane fluidity be altered when mimicking simvastatin-induced composition? Since the effect on Hh signaling appears to be the biggest (Figure 5B) under simvastatin treatment, it would be interesting to compare this against that found for AY9944 treatment. Also, the authors conclude that the effects of simvastatin treatment on ciliary membrane sterol composition are the mildest, however, one could argue that they are the strongest as there is a complete lack of desmosterol.
We thank the reviewer for these insightful comments. Regarding the modest overall effect of simvastatin on cholesterol levels, we would like to note that MDCK cells are an immortalized epithelial cell line with high metabolic plasticity. Such cancer-like cell types are known to exhibit enhanced de novo lipogenesis, particularly under culture conditions with ample glucose availability. This compensatory lipid biosynthesis can partially counterbalance pharmacological inhibition of the cholesterol biosynthetic pathway. Because simvastatin acts upstream in the pathway (at HMG-CoA reductase), its inhibition primarily reduces early intermediates rather than fully depleting end-product cholesterol, explaining the relatively mild changes observed in total cholesterol content.
Concerning desmosterol, we agree with the reviewer that its complete loss under simvastatin treatment is a striking finding that deserves further discussion. Interestingly, our data show that simvastatin treatment produces the strongest inhibition of pathway activation (as measured by SMO activation), but the weakest effect on signal transduction downstream of constitutively active SMOM2. This dichotomy suggests that the absence of desmosterol may preferentially affect the activation step of Hedgehog signaling at the ciliary membrane, without equally impacting downstream propagation. We have expanded the Result section to highlight this potential role of desmosterol in the activation phase of Hedgehog signaling and to contrast it with the effects observed under AY9944 treatment (lines 463-469).
It is not clear to me why the authors have chosen to use SAG to activate the Hh pathway, as this is a downstream mode of activation and bypasses PTCH1 (and therefore a potentially sterol-mediated interaction between the two proteins). It would be very informative to compare the effect of sterol modulation on the ability of ShhN vs SAG to activate the pathway.
Our study aims to demonstrate that the sterol composition of the ciliary membrane plays an essential role in the proper functioning of the Hedgehog (Hh) signaling pathway, comparable in importance to that of oxysterols and free cholesterol. Because ShhN itself is covalently modified by cholesterol, and Smoothened (SMO) can be directly activated by both oxysterols and cholesterol, we reasoned that using a non-native SMO agonist such as SAG would allow us to specifically assess defects arising from alterations in membrane-bound sterols. In this way, pathway activation by SAG provides a more direct readout of the functional contribution of ciliary membrane sterols to SMO activity, independent of potential confounding effects related to ShhN processing, secretion, or PTCH1-mediated regulation.
- The conclusions about the effect of tamoxifen on SMO trafficking in MEFs should be validated in human patient cells before being able to conclude that there is a potential off-target effect (line 438). Also, if that is the case, the experiment of tamoxifen treatment of EBP KO cells should give an additional effect on SMO trafficking. Also, could the CDPX2 phenotypes in patients be the result of different cell types being affected than the fibroblast used in this study?
We agree that carrying the proposed experiment would be a good way to assess a potential off-target effect. However, such validation is beyond the scope of the present study, as this comment on off-target effect was aimed primarily to propose a mechanistic hypothesis to explain the differences observed in Hedgehog pathway activation between patient-derived fibroblasts and tamoxifen-treated MEFs. We leaned towards this hypothesis because drug treatments are known for their overall variable specificity, but we agree other hypotheses are possible, and among them the difference in cell type, as both are fibroblasts but from different origin. We rephrased this passage in the revised manuscript (lines 447-448 ).
Regarding the reviewer's third point, we fully agree that the CDPX2 phenotype in patients is unlikely to arise solely from fibroblast dysfunction. Nevertheless, fibroblasts are the only patient-derived cells currently available to us, and they provide a useful model for assessing ciliary signaling. It is reasonable to expect that similar defects could occur in other, more physiologically relevant cell types.
- For the experiments with the SMO-M2 mutant, it would be useful to show the extent of pathway activation by the mutant compared to SAG or ShhN treatment of non-transfected cells. Moreover, it will be necessary to exclude any direct effects of the compound treatment on the ability of this mutant to traffic to the primary cilium, which can easily be done using fluorescence microscopy as the mutant is tagged with mCherry.
The SmoM2 mutant is indeed a well-characterized constitutively active form of Smoothened that has been extensively studied by us and others. It is well established that this mutant correctly localizes to the primary cilium and robustly activates the Hedgehog pathway in MEFs (see Eguether et al., Dev. Cell, 2014 or Eguether et al, mol.biol.cell, 2018). In our study, we have already included supporting evidence for pathway activation in Supplementary Figure S1b, showing Gli1 expression levels in untreated MEFs transfected with SmoM2, which illustrates the extent of its activation compared to ligand-induced conditions.
In line with the reviewer's recommendation, we will additionally include microscopy data showing SmoM2 localization in MEFs treated with the different sterol modulators. These data should confirm that the observed effects are not due to altered ciliary trafficking of the mutant protein but instead reflect changes in downstream signaling or membrane composition.
Minor comments:
Line 74: 'in patients', should be rephrased to 'patient-derived cells'
This was modified in the revised manuscript
Figure 2A: What do the '+/-' indicate? They seem to be erroneously placed.
We apologize for the oversight, the figures initially submitted with the manuscript inadvertently included some earlier versions, which explains several of the discrepancies noted by the reviewers. This issue has been corrected in the revised submission, and all figures have now been updated to reflect the finalized data.
Figure 2B: no label present for which bar represents cilia/other membranes
We apologize for the oversight, the figures initially submitted with the manuscript inadvertently included some earlier versions, which explains several of the discrepancies noted by the reviewers. This issue has been corrected in the revised submission, and all figures have now been updated to reflect the finalized data.
Figure 2C: this representation is slightly deceptive, since the difference between cells and cilia for lanosterol is not significantly different as shown in figure 2A.
This representation has been removed in the revised figures.
Figure 3A: it would be useful to also show where 8-DHC is in the biosynthetic pathway.
This has been modified in the revised figures.
Line 373: the title should be rephrased as it infers that DHCR7 was blocked in model membranes, which is not the case.
This has been modified in the revised manuscript.
Lines 377-384: this paragraph seems to be a mix of methods and some explanation, but should be rephrased for clarity.
We believe the technical information within this paragraph are useful for the understanding of the reader. We would rather leave as is unless recommended by other reviewers or editorial staff.
Line 403: 'which could explain the resulting defects in Hedgehog signaling': how and what defects? At this point in the study no defects in Hh signaling have been shown.
This has been modified in the revised manuscript.
Figure 4D: 'd' is missing
We apologize for the oversight, the figures initially submitted with the manuscript inadvertently included some earlier versions, which explains several of the discrepancies noted by the reviewers. This issue has been corrected in the revised submission, and all figures have now been updated to reflect the finalized data.
Line 408: SAG treatment resulted in slightly shorter cilia: this is not the case for just SAG treated cilia, but only for the combination of SAG + AY9944. However, in that condition there appears to be a subpopulation of very short cilia, are those real?
This is correct, this is not the case for untreated cilia, but the short population is real, not only in AY9944 but also in Tamoxifen and Simvastatin. Again, the relevance and significance of minor cilia length change is unclear and we are not trying to draw any other conclusion from this than saying that the ciliary compartment is modified.
Figure 5b: it would be good to add that all conditions contained SAG.
This has been modified in the revised figures.
Figure 5D: Since it is shown in Fig 5C that there are no positive cilia -SAG, there is no point to have empty graphs in Fig 5D on the left side, nor can any statistics be done. Similarly for 5K.
We think this is still worth having in the figure. As the reviewer noted in one of his next comment, there are cases where Smoothened or Patched can be abnormally distributed (see also Eguether et al, mol biol cell, 2018). This shows that we checked all conditions for presence or absence of Smo and that there is no signal to be found. We would rather leave it as is unless asked otherwise by editorial staff.
Figure 5E: it is not clearly indicated what is visualized in the inserts, sometimes it's a box, sometimes a line and they seem randomly integrated into the images.
We apologize for the oversight - the figures initially submitted with the manuscript inadvertently included some earlier versions, which explains several of the discrepancies noted by the reviewers. This issue has been corrected in the revised submission, and all figures have now been updated to reflect the finalized data.
Figure 5H: is this the intensity in just SMO positive cilia? If yes, this should be indicated, and the line at '0' for WT-SAG should be removed. I am also surprised there is then ns found for WT vs SLO, since in WT there are no positive cilia, but in SLO there are a few, so it appears to be more of a black-white situation. Perhaps it would be useful to split the data from different experiments to see if it consistently the case that there is a low percentage of SMO positive cilia in SLO cells.
Yes, as in the rest of figure 5, the fluorescence intensity of Smo is only taken into account in SMO positive cells. This is now indicated in figure legend (lines 890, 898, 903 ). As for Smo positive, this is a good suggestion. We checked and for cilia in non-activated SLO patients, there are 8 positive cilia over a total of 240 counted cilia, mainly from one of the experiments. We could remove the data or leave as is given that the result is not significant.
Fig S1: panels are inverted compared to mentioning in the text.
We apologize for the oversight, the figures initially submitted with the manuscript inadvertently included some earlier versions, which explains several of the discrepancies noted by the reviewers. This issue has been corrected in the revised submission, and all figures have now been updated to reflect the finalized data.
Methods-pharmacological treatments: there appear to be large differences in concentrations chosen to treat MDCK versus MEF cells - can the authors comment on these choices and show that the enzymes are indeed inhibited at the indicated concentrations?
We thank the reviewer for this important comment. The concentrations of the pharmacological treatments were optimized separately for MDCK and MEF cells based on cell-type-specific tolerance. For each compound, we used the highest concentration that produced no detectable cytotoxicity or morphological changes. These conditions ensured that the treatments were effective (as seen by changes in sterol composition in MDCK cilia and Hh pathway phenotypes in treated MEFs) and compatible with cell viability and ciliation. Although we did not directly assay enzymatic inhibition in each case, the selected concentrations are consistent with those previously reported to inhibit the targeted enzymes in similar cellular contexts.
Compound
Typical Concentration Range in Mammalian Cell Culture
Typical Exposure Duration
Example Cell Types
Representative Peer-Reviewed References
AY9944 (DHCR7 inhibitor)
1-10 µM widely used; 1 µM for minimal on-target effects; 2.5-10 µM for robust sterol shifts
24-72 h; some sterol studies up to several days
HEK293, fibroblasts, neuronal cells, macrophages
Kim et al., J Biol Chem, 2001 - used 1 µM in dose-response experiments.; Haas et al., Hum Mol Genet, 2007 - 1 µM in cell-based assays.; Recent macrophage sterol study - 2.5-10 µM to induce 7-DHC accumulation.
Simvastatin (HMG-CoA reductase inhibitor)
0.1-10 µM common; 1-10 µM most widely used for robust pathway inhibition
24-72 h
Diverse mammalian lines, including liver, fibroblasts, epithelial cells
Bytautaite et al., Cells (2020) - discusses common in-vitro ranges (1-10 µM).; Mullen et al., 2011 - used 10 µM simvastatin, noting it is a standard in-vitro concentration.
Tamoxifen (modulator of sterol metabolism)
1-20 µM; 1-5 µM for mild/longer treatments; 10-20 µM in cancer/cilia signaling studies
24-72 h (longer treatments often at 1-5 µM)
MDCK, MEFs, MCF-7, diverse epithelial lines
Schlottmann et al., Cells (2022) - used 5-25 µM in sterol-related cell studies.; MCF-7 literature - 0.1-1 µM for estrogenic signaling, higher (5-10 µM) for metabolic/sterol pathway effects.; Additional cancer cell work indicating similar ranges.
This information has been clarified in the revised Methods section (lines 222-224).
(optional): it would be interesting to include a gamma-tubulin staining on the cilium prep to see if there is indeed a presence of the basal body as suggested by the proteomics data.
Thank you, we will try this.
There are many spelling mistakes and inconsistencies throughout the manuscript and its figures (mix of French and English for example) so careful proofreading would be warranted. Moreover, there are many mentionings of 'Hedgehog defects' or 'Hedgehog-linked', where in fact it is a defect in or link to the Hedgehog pathway, not the protein itself. This should be corrected.
We thank the reviewer for noting these issues. We apologize for the inconsistencies observed in the initial submission, as mentioned previously, some of the figures inadvertently included earlier versions, which may have contributed to the errors identified. All figures have now been carefully revised and updated in the resubmitted manuscript.
Regarding the text, we are surprised to hear about the spelling inconsistencies, as the manuscript was professionally proofread prior to submission (documentation can be provided upon request). Nevertheless, we have conducted an additional round of thorough proofreading to ensure consistency throughout the text and figures.
Finally, we have corrected all instances of "Hedgehog defects" or "Hedgehog-linked" to the more accurate phrasing "Hedgehog pathway defect" or "Hedgehog pathway-linked," as suggested by the reviewer throughout the manuscript.
Reviewer #1 (Significance (Required)):
The study of ciliary membrane composition is highly relevant to understand signal transduction in health and disease. As such, the topic of this manuscript is significant and timely. However, as indicated above, there are limitations to this study, most notably the comparison of ciliary membrane versus all cellular membranes (rather than the plasma membrane), which weakens the conclusions that can be drawn. Moreover, cell-type dependency should be more thoroughly addressed. There certainly is a methodological advance in the form of cilia isolation from MDCK cells, however, it is unclear how broadly applicable this is to other mammalian cell types.
We would like to thank the reviewer for their helpful comments and we appreciate the reviewer's recognition of the relevance and timeliness of studying ciliary membrane composition in the context of signaling regulation. We fully acknowledge that our comparison was made between the primary ciliary membrane and the total cellular membrane fraction, which encompasses multiple intracellular membranes. Our intent, however, was to obtain a global overview of how the ciliary membrane differs from the average membrane environment within the cell, thereby highlighting features that are unique to the cilium as a signaling organelle. This approach provides valuable baseline information that complements, rather than replaces, future targeted comparisons with the plasma membrane. As mentioned in this reply, we aim at carrying out these experiments before publication. Regarding cell-type dependency, we concur that ciliary lipid composition may vary between cell types, reflecting differences in their functional specialization. Our method was intentionally established in MDCK cells, which are epithelial and highly ciliated, to ensure sufficient yield and reproducibility. We have initiated trials with other mammalian cell types, including IMCD3 and MEF cells, and while yields remain limited, preliminary results indicate that the approach is adaptable with further optimization. Thus, our current work establishes a robust and reproducible proof of concept in a mammalian model, providing the first detailed sterol fingerprint of a mammalian primary cilium.
We believe this constitutes a significant methodological and conceptual advance, as it opens the way for systematic exploration of ciliary lipid composition across diverse mammalian systems and pathological contexts.
Reviewer #2 (Evidence, reproducibility and clarity (Required)):
Overview
Accumulating evidence suggests that sterols play critical roles in signal transduction within the primary cilium, perhaps most notably in the Hedgehog cascade. However, the precise sterol composition of the primary cilium, and how it may change under distinct biological conditions, remains unknown, in part because of the lack of reproducible, widely accepted procedures to purify primary cilia from mammalian cultured cells. In the present study, the authors have designed a method to isolate the cilium from the MDCK cells efficiently and then utilized this procedure in conjunction with mass spectrometry to systematically analyze the sterol composition of the ciliary membrane, which they then compare to the sterol composition of the cell body. By analyzing this sterol profiling. the authors claim that the cilium has a distinct sterol composition from the cell body, including higher levels of cholesterol and desmosterol but lower levels of 8-DHC and & Lathosterol. This manuscript further demonstrates that alteration of sterol composition within cilia modulates Hedgehog signaling. These results strengthen the link between dysregulated Hedgehog signaling and defects in cholesterol biosynthesis pathways, as observed in SLOS and CDPX2.
While the ability to isolate primary cilia from cultured MDCK cells represents an important technical achievement, the central claim of the manuscript - that cilia have a different sterol composition from the cell body - is not adequately supported by the data, and more rigorous comparisons between the ciliary membrane and key organellar membranes (such as plasma membrane) are required to make this claim. Moreover, although the authors have repeatedly mention that the ciliary sterol composition is "tightly regulated" there is no evidence provided to support such claim. At best, the data suggest that the cilium and cell body may differ in sterol composition (though even that remains uncertain), but no underlying regulatory mechanisms are demonstrated. In addition, much of the 2nd half of the paper represents a rehash of experiments with sterol biosynthesis inhibitors that have already been published in the literature, making the conceptual advance modest at best. Lastly, the link between CDPX2 and defective Hedgehog signaling is tenuous.
We would like to thank the reviewer for their helpful comments
Major comments
Figure 1.
C) Although the isolation of cilium from the MDCK cells using dibucaine treatment seems to be very efficient, the quality control of their fractionation procedure to monitor the isolation is limited to a single western blot of the purified cilia vs. cell body samples, with no representative data shown from the sucrose gradient fractionation steps. Given that prior studies (including those from the Marshall lab cited in this manuscript) found that 1) sucrose gradient fractionation was essential to obtain relatively pure ciliary fractions, and 2) the ciliary fractions appear to spread over many sucrose concentrations in those prior studies , the authors should have included the comparison of the fractionation profile from the sucrose gradient while isolating the primary cilium. This additional information would have further clarified and supported the efficiency of their proposed method.
We thank the reviewer for their insightful comments regarding the quality control of our ciliary fractionation. We would like to clarify several important methodological aspects that distinguish our approach from those used in the studies cited (including those from the Marshall lab). In the cited work, the authors used a continuous sucrose gradient ranging from 30 % to 45 %, which allowed visualization of the distribution of ciliary proteins across the gradient. In contrast, we employed a discontinuous sucrose gradient (25 % / 50 %) optimized for higher recovery and reproducibility in our hands. In our preparation, the primary cilia consistently localize at the interface between the 25 % and 50 % layers. We systematically collect five 1 mL fractions from this interface and use fractions 1-3 for downstream analyses, as fractions 4-5 are typically already depleted of ciliary material. This targeted collection ensures good enrichment and low contamination, while avoiding unnecessary dilution of the limited ciliary sample. We also note that the prior studies the reviewer refers to were optimized for proteomic analyses, and therefore used actin as a marker of contamination from the cell body. In our case, the downstream application is lipidomic profiling, for which such protein-based contamination markers are not directly informative, since no reliable lipid marker exists to differentiate between organelle membranes. For this reason, we limited the protein-level validation to a semi-quantitative assessment of ciliary enrichment using ARL13B Western blotting, which robustly reports the presence and enrichment of ciliary membranes. Finally, to complement this targeted validation, we performed proteomic analysis followed by Gene Ontology (GO) Enrichment Analysis using the PANTHER database. This analysis evaluates the overrepresentation of proteins associated with ciliary structures and functions relative to the background frequency in the Canis lupus familiaris proteome. The resulting enrichment profile confirms that the isolated material is highly enriched in ciliary components and somewhat depleted of non-ciliary contaminants, thereby serving as an unbiased and global assessment of sample specificity and purity. We believe that, together, these methodological choices provide a rigorous and quantitative validation of our fractionation efficiency and support the robustness of the cilia isolation protocol used in this study.
-
D) The authors presented proteomic data for the peptides analyzed from the isolated cilia in the form of GO term analysis; however, they did not provide examples of different proteins enriched within their fractionation procedure, aside from Arl13b shown in the blot. Including a summary table with representative proteins identified in the isolated ciliary fraction, along with the relative abundance or percentage distribution of these proteins, would make the data more informative.
We thank the reviewer for this valuable suggestion. As mentioned in the manuscript, our proteomic dataset includes numerous hallmark components of the cilium, such as 18 IFT proteins, 4 BBS proteins, and several Hedgehog pathway components (including SuFu and Arl13b), as well as axonemal (Tubulin, Kinesin, Dynein) and centrosomal proteins (Centrin, CEPs, γ-Tubulin, and associated factors). This composition demonstrates that the isolated fraction is highly enriched in bona fide ciliary components while retaining a small proportion of basal body proteins, which is expected given their physical continuity. Importantly, our dataset shows a 70% overlap with the ciliary proteome published by Ishikawa et al. and a 41% overlap with the CysCilia consortium's list of potential ciliary proteins, which supports both the specificity and reliability of our isolation procedure. Regarding the suggestion to present relative protein abundances, we would like to clarify that defining "relative to what" is challenging in this context. The stoichiometry of ciliary proteins is largely unknown, and relative abundance normalized to total protein content can be misleading, as ciliary structural and signaling components differ greatly in copy number and membrane association. For this reason, we chose to highlight in the text proteins such as BBS and IFTs, which are known to be of low abundance within the cilium; their detection supports the depth and specificity of our proteomic coverage. In addition, we performed an unbiased Gene Ontology (GO) Enrichment Analysis using the PANTHER database, which provides a systematic and quantitative overview of the biological processes and cellular components overrepresented in our dataset relative to the canine proteome. This analysis with regard to purity wa already discussed in the submitted manuscript discussion. To further address the reviewer's comment, we will include as a supplemental table in the revised manuscript, a summary table listing representative ciliary proteins identified in our fraction, including those overlapping with the CysCilia (Gold ans potential lists), CiliaCarta and Ishikawa/Marshall proteomes. This addition should make the dataset more transparent and informative while preserving scientific rigor.
Figure 2.
The authors represented the comparison of sterol content within the cilia versus whole cell (as cell membranes). Since different organelles have a very diverse degree of cholesterol contents within them, for instance plasma membrane itself is around 50 mol% cholesterol levels while organelles like ER have barely any cholesterol. Thus, comparing these two samples and claiming a 2.5-fold increase in cholesterol levels is misleading. A more appropriate comparison would be between isolated primary cilia and isolated plasma membranes (procedures to isolate plasma membranes have been described previously, e.g., Naito et al., eLife 2019; Das et al, PNAS 2013. The absence of such controls makes it difficult to fully validate the reported magnitude of sterols enrichment in cilia relative to the cell surface.
As already discussed above for reviewer 1, we would like to emphasize that our study did not aim to compare the cilium directly to the plasma membrane, nor did we claim that the comparison was in any way related to the plasma membrane. Our intent, was to obtain a global overview of how the ciliary membrane differs from the average membrane environment within the cell, thereby highlighting features that are unique to the cilium as a signaling organelle. This approach provides valuable baseline information that complements, rather than replaces, future targeted comparisons with the plasma membrane. However, we concur that determining the sterol composition of the MDCK plasma membrane would provide valuable context and enable a comparison with the membrane continuous with the ciliary membrane. Hence, we are willing to try isolating plasma membrane in the same cellular contexts, and we thank the reviewer for the proposed literature.
Also, because dibucaine was used here to isolate MDCK cilia, a control experiment to exclude possible effects of the dibucaine treatment on sterol biosynthesis would be helpful.
Thank you for this comment, we will verify this point by quantifying by GC-MS the sterol content of whole MDCK cells with and without 15 minutes-dibucaine treatments.
Figure 3.
Tamoxifen is a potent drug for nuclear hormone receptor activity and thus can independently influence various cellular processes. As several experiments in the later sections of the manuscript rely on tamoxifen treatment of cells, it is important that the authors include appropriate controls for tamoxifen treatment, to confirm that the observed effects do not stem from effects on nuclear hormone receptor activity. This would ensure that the observed effects can be confidently attributed to the experimental manipulation rather than to the intrinsic effects of tamoxifen.
The reviewer is right, tamoxifen, like many drugs, has pleiotropic effects in different cell processes. Aware of this possible issue, we turned to a genetic model creating a CRISPR-CAS9 mediated knock down of EBP, the enzyme targeted by tamoxifen. We showed in figure 5 that the results between tamoxifen treated cells and CRIPSR EBP cells were in accordance with one another, showing that, for hedgehog signaling, the effect of tamoxifen recapitulates the effect of the enzyme KO.
Figure4.
The authors present the results of spectroscopy studies to analyze generalized polarization (GP) of liposomes in vitro , but only processed data are shown, and the raw spectra are not provided. The authors need to present representative spectra to enable the readers to interact the raw data from the experiments.
This has been added to new supplemental figure 1 and corresponding figure legend (lines 898-904)
Figure5.
B) The experiment shown Gli1 mRNA levels following treatment with inhibitors of cholesterol biosynthesis, but similar findings have already been reported previously (e.g., Cooper et al, Nature Genetics 2003; Blassberg et al, Hum Mol Genet 2016), and the present results do not provide a significant conceptual advance over those earlier studies.
We thank the reviewer for this comment and for highlighting the importance of earlier studies on Hedgehog (Hh) signaling and cholesterol metabolism. While we fully agree that confirming and extending established findings has intrinsic scientific value, we respectfully disagree with the assertion that our work does not provide conceptual novelty.
The seminal work by Cooper et al. (Nature Genetics, 2003) indeed laid the foundation for linking sterol metabolism to Hedgehog signaling, and we cite it as such. However, that study was conducted in chick embryos, a model that is relatively distant from mammalian systems and human pathophysiology. Moreover, their approach relied heavily on cyclodextrin-mediated cholesterol depletion, which is non-specific and extracts multiple sterols from membranes (discussed in this article lines 512-516). In contrast, our study employs pharmacological inhibitors targeting specific enzymes in the sterol biosynthetic pathway, thereby allowing us to modulate distinct steps and intermediates in a controlled and mechanistically informative manner. We also extend these analyses to patient-derived fibroblasts and CRISPR-engineered cells, providing direct human and genetic validation of the observed effects. Importantly, we complement these cellular studies with biochemical characterization of isolated ciliary membranes from MDCK cells, enabling a direct assessment of how specific sterol alterations affect ciliary composition and Hh pathway function - an angle not addressed in prior work.
Regarding Blassberg et al. (Hum. Mol. Genet., 2016), we agree that part of our findings recapitulates their observations on SMO-related signaling defects, which we view as an important confirmation of reproducibility. However, their study primarily sought to distinguish whether Hh pathway impairment in SLOS results from 7-DHC accumulation or cholesterol depletion, concluding that cholesterol deficiency was the main cause. Our results expand on this by demonstrating that perturbations extend beyond these two sterols, and that additional intermediates in the biosynthetic pathway also impact ciliary membrane composition and signaling competence. Furthermore, our experiments using the constitutively active SmoM2 mutant show that Hh signaling defects are not restricted to SMO activation per se, revealing a broader disruption of the signaling machinery within the cilium.
Finally, neither of the above studies examined CDPX2 patient-derived cells or the consequences of EBP enzyme deficiency on Hh signaling. Our finding that this pathway is altered in this genetic context represents, to our knowledge, a novel link between CDPX2 and Hedgehog pathway dysfunction.
Taken together, our work builds upon and extends previous findings by integrating cell-type-specific, biochemical, and patient-based analyses to provide a more comprehensive and mechanistically detailed view of how sterol composition of the ciliary membrane regulates Hedgehog signaling.
In addition, the authors analyze the effect of these inhibitors on SAG stimulation, but the experiment lacks the control for Gli mRNA levels in the absence of SAG treatment. Without this control, it is impossible to know where the baseline in the experiment is and how large the effects in question really are.
Below, we provide the data expressed using the ΔΔCt method (NT + SAG normalized to NT - SAG), which more clearly illustrates the magnitude of the effect in question. As similar qPCR-based Hedgehog pathway activation assays in MEFs have been published previously (see Eguether et al., Dev. Cell 2014; Eguether et al., Mol. Biol. Cell 2018), our goal here was not to re-establish the assay itself but to highlight the comparative effects across experimental conditions. In addition, one of the datasets was obtained using a new batch of SAG, which exhibited stronger pathway activation across all conditions (visible as higher overall expression levels). To ensure valid statistical comparisons across experiments and to focus on relative rather than absolute activation, we therefore chose to present the data as fold change values, which provides a more robust and statistically consistent measure for cross-condition analysis.
J-K) The data represented in these panels for SAG treatment as fraction of Smo and its fluorescence intensity for the same sample appears to be inconsistent between the two graphs. Under SAG treatment for EBP mutants shows higher Smo fluorescence intensity while Smo positive cilia seems to be less than the wild type control cells. If the number of Smo+ cilia (quantified by eye) differs between conditions, shouldn't the quantification of Smo intensity within cilia show a similar difference?
We thank the reviewer for this careful observation. The apparent discrepancy arises because the two panels quantify different parameters. In panel (j), we counted the percentage of cilia positive for SMO (i.e., cilia in which SMO was detected above background). In contrast, panel (k) reports the fluorescence intensity of SMO, but this measurement was performed only within the SMO-positive cilia identified in panel (j). This distinction has now been explicitly clarified in the figure legend, as also suggested by Reviewer 1.
Taken together, these two analyses indicate that although fewer cilia display detectable SMO accumulation in the EBP mutant cells, the amount of SMO present within those cilia that do recruit it is comparable to wild-type levels (as reflected by the non-significant difference in fluorescence intensity). This interpretation helps explain the partial functional preservation of Hedgehog signaling in this condition and contrasts with cases such as AY9944 treatment, where both the number of SMO-positive cilia and the SMO intensity are reduced.
-
I) The rationale for using SmoM2 in the analysis of cholesterol metabolism-related diseases such as SLOS and CDPX2 is unclear. The SmoM2 variant is primarily associated with cancer rather than cholesterol biosynthesis defects and its relevance either of these disorders is not immediately apparent.
We thank the reviewer for this pertinent observation. We fully agree that SmoM2 was originally identified as an oncogenic mutation and is not directly associated with cholesterol biosynthesis disorders. However, our rationale for using this mutant was mechanistic rather than pathological. SmoM2 is a constitutively active form of SMO that triggers pathway activation independently of upstream components such as PTCH1 or ligand-mediated regulation.
By using SmoM2, we aimed to determine whether the signaling defects observed under conditions that alter sterol metabolism (e.g., treatment with AY9944 or tamoxifen) occur upstream or downstream of SMO activation. The results demonstrate that, even when SMO is constitutively active, the Hedgehog pathway remains impaired under AY9944 treatment-and to a lesser extent with tamoxifen-indicating that these sterol perturbations disrupt the pathway beyond the level of SMO activation itself. In contrast, cells treated with simvastatin maintain normal pathway responsiveness, reinforcing the specificity of this effect.
This experiment is therefore central to our study, as it reveals that sterol imbalance can hinder Hedgehog signaling even in the presence of an active SMO, providing new insight into how membrane composition influences downstream signaling competence.
Minor corrections
-
Line 385 seems to be a bit confusing which mentions cilia were treated with AY9944 - do the authors mean that cells were been treated with the drugs before isolation of cilia, or were the purified cilia actually treated with the drugs?
Thank you, this has been modified in the revised manuscript
The authors should add proper label in Figure 2 panel b for the bars representing the cilia and cell membranes.
We apologize for the oversight, the figures initially submitted with the manuscript inadvertently included some earlier versions, which explains several of the discrepancies noted by the reviewers. This issue has been corrected in the revised submission, and all figures have now been updated to reflect the finalized data.
Panels in Figure S1 should be re-arranged according to the figure legend and figure reference in line 450.
We apologize for the oversight, the figures initially submitted with the manuscript inadvertently included some earlier versions, which explains several of the discrepancies noted by the reviewers. This issue has been corrected in the revised submission, and all figures have now been updated to reflect the finalized data.
Legend for the Figure S1b should be corrected as data sets in graph represents 7 points while technical replicates in legend shows 6 experimental values.
Thank you, this has been modified in the revised manuscript
The labels for drug in Figure 3 and 5 should be corrected from tamoxifene to tamoxifen and simvastatine to simvastatin.
We apologize for the oversight, the figures initially submitted with the manuscript inadvertently included some earlier versions, which explains several of the discrepancies noted by the reviewers. This issue has been corrected in the revised submission, and all figures have now been updated to reflect the finalized data.
Reviewer #2 (Significance (Required)):
In the present study, the authors have designed a method to isolate the cilium from the MDCK cells efficiently and then utilized this procedure in conjunction with mass spectrometry to systematically analyze the sterol composition of the ciliary membrane, which they then compare to the sterol composition of the cell body. By analyzing this sterol profiling. the authors claim that the cilium has a distinct sterol composition from the cell body, including higher levels of cholesterol and desmosterol but lower levels of 8-DHC and & Lathosterol. This manuscript further demonstrates that alteration of sterol composition within cilia modulates Hedgehog signaling. These results strengthen the link between dysregulated Hedgehog signaling and defects in cholesterol biosynthesis pathways, as observed in SLOS and CDPX2.
While the ability to isolate primary cilia from cultured MDCK cells represents an important technical achievement, the central claim of the manuscript - that cilia have a different sterol composition from the cell body - is not adequately supported by the data, and more rigorous comparisons between the ciliary membrane and key organellar membranes (such as plasma membrane) are required to make this claim. Moreover, although the authors have repeatedly mention that the ciliary sterol composition is "tightly regulated" there is no evidence provided to support such claim. At best, the data suggest that the cilium and cell body may differ in sterol composition (though even that remains uncertain), but no underlying regulatory mechanisms are demonstrated. In addition, much of the 2nd half of the paper represents a rehash of experiments with sterol biosynthesis inhibitors that have already been published in the literature, making the conceptual advance modest at best. Lastly, the link between CDPX2 and defective Hedgehog signaling is tenuous.
We thank the reviewer for this detailed summary and for acknowledging the technical advance represented by our method for isolating primary cilia from MDCK cells. However, we respectfully disagree with several aspects of the reviewer's assessment of our work.
As we elaborated in our responses to earlier comments, particularly regarding Figure 5, we disagree with the characterization of part of our study as a "rehash", a somewhat derogatory word, of previously published experiments. Our approach differs from earlier studies by relying on specific pharmacological modulation of defined enzymes in the sterol biosynthesis pathway, rather than using non-specific agents such as cyclodextrins, and by linking these manipulations to direct biochemical measurements of ciliary sterol composition. This strategy allows, for the first time, a targeted and physiologically relevant examination of how specific sterol perturbations affect Hedgehog signaling.
Regarding our statement that ciliary sterol composition is "tightly regulated," we acknowledge that we have not yet explored the underlying molecular mechanisms of this regulation. Nevertheless, the experimental evidence supporting this statement lies in the variation of ciliary sterol composition across multiple treatments that strongly perturb cellular sterols. Despite broad cellular changes, the ciliary sterol profile remains very resilient for some parameters, an observation that, in our view, strongly supports the idea of a selective or regulated process maintaining ciliary sterol identity. This conclusion does not depend on comparison with other membrane compartments.
We also respectfully disagree that the observed differences between cilia and the cell body (which doesn't equal to plasma membrane) are "uncertain." The consistent enrichment in cholesterol and desmosterol, combined with the relative depletion in 8-DHC and lathosterol, were detected across independent replicates using robust lipidomic profiling and are statistically supported. These findings are, to our knowledge, the first quantitative demonstration of a sterol fingerprint specific to a mammalian cilium.
Finally, while we agree that the mechanistic link between CDPX2 and defective Hedgehog signaling warrants further exploration, the data we present, combining pharmacological inhibition (tamoxifen), CRISPR-mediated EBP knockout, and SMOM2 activation assays, all consistently indicate a functional impairment of the Hedgehog pathway under EBP deficiency. This is further reinforced by clinical reports describing Hedgehog-related phenotypes in CDPX2 patients. We therefore believe that our work provides a solid experimental and conceptual basis for connecting EBP dysfunction to Hedgehog signaling defects.
In summary, our study introduces a validated and reproducible method for mammalian cilia isolation, provides the first detailed sterol composition profile of primary cilia, and establishes a functional link between ciliary sterol imbalance and Hedgehog pathway modulation. We believe these findings represent a meaningful conceptual advance and a valuable resource for the field
Reviewer #3 (Evidence, reproducibility and clarity (Required)):
Lamaziere et al. describe an improved protocol for isolating primary cilia from MDCK cells for downstream lipidomics analysis. Using this protocol, they characterize sterol profile of MDCK cilia membrane under standard growth conditions and following pharmacological perturbations that are meant to mimic SLOS and CDPX2 disorders in humans. The authors then assess the impact of the same pharmacological manipulations on Shh pathway activity and validate their findings from these experiments using orthogonal genetic approaches. Major and minor concerns that require attention prior to publication are outlined below.
We would like to thank the reviewer for their comments
Major
1.Since the extent of contamination of the cilia preps with non-cilia membranes is unclear, and variability between replicates is not reported, it makes interpretation of changes in cilia membrane sterol composition in response to pharmacological manipulations somewhat difficult to interpret. Discussing reproducibility of cilia sterol composition between replicates (and including corresponding data) could alleviate these concerns to some extent.
We thank the reviewer for this comment. We would like to clarify that variability between replicates is indeed reported throughout the manuscript. In Figures 2 and 3, all data are presented as mean {plus minus} SEM, as indicated in the figure legends. Specifically, the data in Figure 2 are derived from six independent experiments, reflecting the central dataset used for comparative analyses, while the data in Figure 3 are based on three independent experiments.
We also note that the overall variability between replicates is low, further supporting the reproducibility of our ciliary sterol composition measurements. This consistency across independent biological replicates provides confidence that the differences observed between cilia and the cell body are robust and not due to stochastic contamination or technical variation.
2.An abundant non-ciliary membrane protein (rather than GAPDH) may be a more appropriate loading control in Fig. 1C.
This is a valuable comment and we will find a non-ciliary membrane protein to complement this experiment.
3.Fig. 2b - which bar corresponds to cells and which one to cilia? What do numbers inside bars represent? Please label accordingly.
We apologize for the oversight, the figures initially submitted with the manuscript inadvertently included some earlier versions, which explains several of the discrepancies noted by the reviewers. This issue has been corrected in the revised submission, and all figures have now been updated to reflect the finalized data.
4.Fig. 3b-d, right panels - please define what numbers inside bars represent
Thank you, this was done in the revised manuscript. The numbers are reports of absolute quantification.
5.The font in Figs 2, 3, and 4 is very small and difficult to read. Please make the font and/or panels bigger to improve readability.
We did our best to enlarge font despite space limitations, but we are willing to work with editorial staff to improve readability as suggested.
6.It would help to have a diagram of the key steps in the cholesterol synthesis pathway for reference early in the paper rather than in figure 3.
We thank the reviewer for his comment, but we don't understand why this would be helpful as we only use sterol modulators involving the pathway's enzyme in fig3. We are open to discussion with editorial staff about moving it up to fig2. If they feel this is needed
7.The authors need to discuss why/how global inhibition of enzymes (e.g. via AY9944 treatment) in a cell could cause reduction in cholesterol levels only in the cilium and not in other cell membranes (see also point 1). Yet, tamoxifen treatment lowers cholesterol across the board.
We thank the reviewer for these insightful comments. Regarding the modest overall effect of simvastatin on cholesterol levels, we would like to note that MDCK cells are an immortalized epithelial cell line with high metabolic plasticity. Such cancer-like cell types are known to exhibit enhanced de novo lipogenesis, particularly under culture conditions with ample glucose availability. This compensatory lipid biosynthesis can partially counterbalance pharmacological inhibition of the cholesterol biosynthetic pathway. Because simvastatin acts upstream in the pathway (at HMG-CoA reductase), its inhibition primarily reduces early intermediates rather than fully depleting end-product cholesterol, explaining the relatively mild changes observed in total cholesterol content. . This has been added in a new paragraph in the revised manuscript (lines 371-378).
8.Fig. 5c, g, and j - statistical analyses are missing and need to be added in support of conclusions drawn in the text of the manuscript.
Thank you, this has been done in the revised manuscript
9.The decrease in the fraction of Smo+ cilia observed in EBP KO cells is mild (panel j, no statistics), and there is possibly a clone-specific effect here as well (statistical analysis is needed to determine if EBP139 is indeed different from WT and whether EBP139 and 141 are different from each other). Similarly, Smo fluorescence intensity after SAG treatment (panel k) is the same in WT and EBP KO cells, while there is a marked difference in intraciliary Smo intensity after tamoxifen treatment. The author's conclusion "...we were able to show that results with human cells aligned with our tamoxifen experiments" (line 436) should be modified to more accurately reflect the presented data. Ditto conclusions on lines 440-442, 530-531. In fact, it is the lack of Hh phenotypes in CDPX2 patients that is consistent with the EBP KO data presented in the paper.
We thank the reviewer for this detailed comment. We have now performed the requested statistical analyses and incorporated them into the revised manuscript.
The new analyses confirm that both EBP139 and EBP141 CRISPR KO clones show a statistically significant reduction in the fraction of Smo⁺ cilia compared to WT cells. They also reveal that the two clones differ significantly from each other, consistent with the expected clonal variability inherent to independently derived CRISPR lines.
Despite this variability, several lines of evidence support our conclusion that the EBP KO phenotypes align with the effects observed after tamoxifen treatment:
1- Directionally consistent reduction in Smo⁺ cilia:
Although the magnitude of the decrease differs between clones, both clones display a significant reduction compared to WT, paralleling the reduction observed in tamoxifen-treated cells. This directional consistency is the key point for comparing pharmacological and genetic perturbations.
2-Converging evidence from SmoM2 experiments:
Tamoxifen treatment also reduces pathway output in the context of SmoM2 overexpression. This supports the interpretation that both EBP inhibition (tamoxifen) and EBP loss (CRISPR KO) impair Hedgehog signaling at the level of ciliary function, albeit more mildly than AY9944/SLOS-like perturbations.
3-Interpretation of Smo intensity (panel k):
As clarified in the revised text, the fluorescence intensities in panel K correspond only to cilia that are Smo-positive. The absence of a difference in intensity therefore does not contradict the observed reduction in the number of Smo⁺ cilia. Rather, it explains why the phenotype is milder than that observed for SLOS/AY9944: when Smo is able to enter the cilium, its enrichment level is comparable to WT.
4- Clinical relevance for CDPX2:
While Hedgehog-related phenotypes in CDPX2 patients may be milder or under-reported, several documented features, such as polydactyly (10% of cases), as well as syndactyly and clubfoot, are classically associated with ciliary/Hedgehog signaling defects. This clinical pattern is consistent with the milder yet detectable defects we observe in EBP KO cells.
Minor
•Line 310: 'intraflagellar' rather than 'intraciliary' transport particle B is a more conventional term
We agree that intraflagellar is more conventional than intraciliary, but in this case, this is how the GO term is labeled in the database. In our opinion, it should stay as is.
-
Fig. 2c - typos in the color key, is grey meant to be "cells" and blue "cilia"? Individual panels are not referenced in the text
This panel has been removed thanks to comment from reviewer 1 and 3 finding it misleading.
-
Lines 357-358: "Notably, AY9944 treatment led to a greater reduction in cholesterol content as well as a greater increase in 7-DHC and 8-DHC in cilia than in the other cell membranes" - the authors need to support this statement with appropriate statistical analysis
We respectfully believe there may be a misunderstanding in the reviewer's concern. In all cases, our comparisons are made between treated vs. untreated conditions within each compartment (cell bulk vs. ciliary membrane), and the statistical significance of these differences is already reported as determined by a Mann-Whitney test. In every case, the changes observed are greater in cilia than in the cell body. The statement in the manuscript simply summarizes this quantitative observation. However, if the reviewer feels that an additional statistical test directly comparing the magnitude of the two compartment-specific changes would strengthen the claim, we are willing to include this analysis. Alternatively, if preferred, we can remove the sentence entirely, as the comparison is already clearly visible in Figure 3b.
-
Line 473 - unclear what is meant by "olfactory cilia are mainly sensory and not primary". Primary cilia are sensory.
We agree, primary cilia are sensory, but still different from cilia belonging to sensory epithelia like retina photoreceptors or olfactory cilia. Nevertheless, this statement was modified in revised manuscript
-
Line 551: 'data not shown'. Please include the data that you would like to discuss or remove discussion of these data from the manuscript.
The data is not shown because there is nothing to show, as we discussed in that sentence, use of cholesterol probe resulted in the disappearance of primary cilia altogether. We are willing to work with editorial staff to find a better way of expressing this idea.
Reviewer #3 (Significance (Required)):
Overall, the manuscript expands our knowledge of cilia membrane composition and reports an interesting link between SLOS and Shh signaling defects, which could at least in part explain SLOS patients' symptoms. The findings reported in the manuscript could be of interest to a broad audience of cell biologists and geneticists.
We would like to thank the reviewer for his recognition of the importance of this work