Inhibition of fatty acid amide hydrolase prevents pathology in neurovisceral acid sphingomyelinase deficiency by rescuing defective endocannabinoid signaling

Abstract Acid sphingomyelinase deficiency (ASMD) leads to cellular accumulation of sphingomyelin (SM), neurodegeneration, and early death. Here, we describe the downregulation of the endocannabinoid (eCB) system in neurons of ASM knockout (ASM‐KO) mice and a ASMD patient. High SM reduced expression of the eCB receptor CB 1 in neuronal processes and induced its accumulation in lysosomes. Activation of CB 1 receptor signaling, through inhibition of the eCB‐degrading enzyme fatty acid amide hydrolase (FAAH), reduced SM levels in ASM‐KO neurons. Oral treatment of ASM‐KO mice with a FAAH inhibitor prevented SM buildup; alleviated inflammation, neurodegeneration, and behavioral alterations; and extended lifespan. This treatment showed benefits even after a single administration at advanced disease stages. We also found CB 1 receptor downregulation in neurons of a mouse model and a patient of another sphingolipid storage disorder, Niemann–Pick disease type C (NPC). We showed the efficacy of FAAH inhibition to reduce SM and cholesterol levels in NPC patient‐derived cells and in the brain of a NPC mouse model. Our findings reveal a pathophysiological crosstalk between neuronal SM and the eCB system and offer a new treatment for ASMD and other sphingolipidoses.

(Note: With the exception of the correction of typographical or spelling errors that could be a source of ambiguity, letters and reports are not edited. Depending on transfer agreements, referee reports obtained elsewhere may or may not be included in this compilation. Referee reports are anonymous unless the Referee chooses to sign their reports.) The current manuscript by Bartoll and co-workers reveals that endocannabinoid system, in particular CB1 receptor, is downregulated upon sphingomyelin accumulation, as for example observed in the acid sphingomyelinase knockout (ASM-KO) mice. Application of the fatty acid amide hydrolase (FAAH) inhibitor restored CB1 signaling and correlated with the reduced levels of sphingomyelin (SM), reduced inflammation, neurodegeneration and prolonged life span. Data presented in this manuscript suggest that modulating CB1 signaling may provide benefits for treatment of acid sphingomyelinase deficiency (ASMD). This is an interesting and timely report, as ASMD is a fatal and incurable disease. However, the results presented in the manuscript do not fully support major conclusions and additional lines of evidence should be included. As anticipated by the authors, mechanistic link between CB1 and sphingomyelin has already been established. Although this link has now been explored upon ASAD, the manuscript does not provide additional mechanistic information on how this complex regulatory loop is regulated and whether there are regional or cell-type specific differences that may be of relevance for disease pathology and therapeutic treatment. Major issues 1) It is very difficult to judge protein levels of CB1 from the presented immunofluorescent images (Fig 1D). This result should be confirmed by Western blot analysis. 2) Immunofluorescent analysis of a single ASMD patient is not conclusive. It is hard to estimate whether the aggregation signal observed in the NPA patient is indeed specific, and if so, this may not be a common pattern. MAP2 signal is very low, even in the control, and almost absent from the NPA patient. Severe neurodegeneration would also interfere with the interpretation of the CB1 levels.
3) The authors suggest that cellular misdistribution of CB1 occurs in ASM-KO neurons and upon sphingomyelin supplementation. It would be beneficial to include lower magnification images showing more than a single neuron. Aggregation phenotype of CB1 is more obvious in Fig 2C and F, compared to Fig 2B and E. Can the cellular misdistribution phenotype of CB1 also be observed in ASM-KO neurons in vivo? 4) Lysosomal accumulation of CB1 has been observed in ASM-KO or upon SM supplementation. The authors discuss that lysosomal degradation may contribute to the reduced levels of CB1 (discussion, page 12). Would inhibition of lysosomal degradation in cultured primary neurons (for example upon addition of SM) effect CB1 levels and its cellular distribution? Along these lines, can normal cellular distribution of CB1 be restored upon treatment of ASM-KO with SMase? 5) Western blot image of CB1 in Fig 2I does not support the reported 45% increase in CB1 protein levels. 6) The authors report a body weight gain in ASM-KO mice treated with PF and claim that no differences were observed in the WT. However, differences can also be observed upon PF application in the WT (Fig 4A). No statistics was included to support the rescue phenotype. 7) Histopathological analysis of the ASM-KO cohort treated longer with PF and included into the survival analysis ( Fig 4F) has not been performed and could add additional, therapeutically relevant information when compared with the treatment efficacy over 8 weeks. 8) Figure 5 should be improved to have the analysis more consistent. AEA and NSM levels were monitored in the hippocampal extracts only (Fig 5A and B). SM analysis was done for the cerebellum, hippocampus and the cortex (hippocampus was indeed less affected). Further immunohistological analysis only revealed changes in the cerebellum. This makes it difficult for a reader to follow the logic. It would be helpful to have AEA, NSM, CB1 and SM levels plus histological examination for the same brain region (or at least the one affected most). Showing only Lamp1 staining in Fig 5D is not beneficial. Cellular marker or at least dapi images should be included to judge what the authors aim to present. Similar to comment #3, also in Fig 5F and G it would be beneficial to have a larger overview to better judge the GFAP and Iba1 pathology. 9) Is the treatment with PF able to correct the cellular distribution phenotype of CB1? 10) Acute treatment is not offering any additional information beyond toxicity assessment. SM levels were only slightly reduced in the cortex. However, the neuroinflammatory effect was also seen in the hippocampus where no SM reduction could be detected. The authors should consider if the statement on page 11 "...and also revealed efficacy after a single administration of high PF doses." is fully supported by the data. 11) I appreciate the efforts towards examining CB1 signaling in other diseases where SM accumulates. However, if the authors wish to study CB1 signaling in NPC, more evidence should be provided, including Western blot analysis of CB1 across different brain regions and their corresponding SM levels. Moreover, brain material and cultured cells from more than 1 NPC patient should be included to make this result conclusive. Also in Fig 7 more consistency would be helpful for a reader. Fig 7A shows CB1 levels in the cerebellum of an NPC1-deficient mouse model, human tissue reveals CB1 staining of a hippocampus and cultured cells of an NPC patient treated with PF were only examined for the levels of SM (CB1 analysis is missing). Analysis of other phenotypes such as cellular distribution of CB1, lysosomal morphology or levels of cholesterol upon PF treatment would provide additional information to judge the therapeutic potential of CB1 modulation in NPC disease. Minor issues 1) Authors should discuss differences between CB1 (neuronal) and CB2 (immune cells) expression patterns that may result in different signaling networks in neurons vs microglia. Potential common (or diverse) mechanism of FAAHi effect in different nervous system cells could be discussed. 2) Authors should consider improving figure legends such as in Fig 4A and F to distinguish more easily between different conditions. 3) Consider re-phrasing "...dual impact" on page 11 of the discussion. 4) Reference Nr 50 on page 15 of the discussion should be corrected according to the common reference style.
Referee #3 (Remarks for Author): The authors describe the importance of the eCB system in the development of severe neurological disorders like ASMD and propose a new therapeutic strategy based on the modulation of eCB system by acting on the CB1 receptor for rescuing the pathological storage of SM, inflammation and behavioral abnormalities in the ASM-KO mouse model. The invitro study performed by using cultured hippocampal neuronal cells represents an optimal method to validate the role of CB1 receptor and its direct link with the SM accumulation. Starting from this invitro validation, the authors tried first to confirm in ASM-KO mice the role of CB1 receptors in the brain pathology development and after to test the efficacy of a therapeutic approach based on the use of FAAH inhibitors for rescuing the neuropathology in this animal model and also in NPC human cell line. While the study is of potential interest some data should be improved to better validate the effectiveness of this strategy for the treatment of neuropathology in ASMD mouse model. Major Comments: - Figure 1 and Results: The fluorescence intensity measurement doesn't represent a quantitative analysis of CB1 protein. The authors should perform the quantitative analysis of CB1 by WB experiments in the same brain regions used for qPCR experiments.
-Results page 5: Since the CB1 receptors are well abundant in the prefrontal cortex, it should be interesting if the authors indicate in text and in the figure legend of Fig.1C which cortical regions have been analyzed and if they performed all the experiments in the same mouse cortical regions.
- Figure 1D: the author should improve the resolution of cerebellum images in which it is difficult to appreciate the signal of CB1 receptor.
- Figure 1E: In the IF staining on cerebellum, the MAP2 signal is almost undetectable in the slides of both CTR and NPA patients and it is also complicated to understand the morphology of this brain region. The author should improve the IF experiments in order to better analyze the decrease of CB1 signal in the cerebellum of NPA patients. They could also try to perform a labeling with Calbindin/CB1/TOPRO markers.  Figure 2A: In order to quantify the expression levels of CB1 receptor, the authors should perform the WB analysis on WT and ASM-KO hippocampal cells. The immunofluorescence experiments represent only a semi-quantitative analysis able to support the WB data. Moreover, since the authors explain in the results that the main pathological hallmark of the ASMD is represented by the increase of SM levels, it is important to quantify the amount of sphingomyelin in both WT and ASM-KO neuronal cells.
- Figure 2C: Since the measure of CB1 fluorescent intensity has been already quantified in the figure  2B, the authors should replace the CB1 quantification graph in Fig 2C with another one in which they show the LAMP1 quantification as marker of autophagy impairment. Moreover, the authors showed an increase of CB1 in the cell body of ASM-KO cells. Since this increase is associated with an increase of SM levels, the author should show a quantification of SM levels among WT and ASM-KO cells.
- Figure 2D-2E: The authors could show the qPCR and the IF experiments in WT and also ASM-KO cell lines. Furthermore, they should perform the WB analysis for CB1 receptor with the relative protein quantifications in the following neuronal cell lines: WT, WT+40uM SM and ASM-KO.
- Figure 2F: Since the authors quantified in both panels 2E and 2F the CB1 signal using the same cell lines, they should replace the figure 2E with figure 2F.
- Figure 2G, 2H, 2I: The authors should also perform the experiments in the WT cell lines as positive control.
- Figure 3A-3E: the authors should also perform the experiments in the WT neuronal cell line as CTR cell line.     Since it is difficult to appreciate to LAMP1 localization and the cerebellum morphology, the authors should perform a co-labeling with LAMP1 markers and DAPI.
Moreover, since the authors demonstrated an increase in the LAMP1 and CB1 colocalization in ASM-KO cells, they should also perform a co-labeling experiment with LAMP1 and CB1 markers in cerebellum and hippocampus of treated mice. This data should support the biochemical experiments performed in the same animals. Figure 5F-5G: the authors show the IBAI and GFAP staining in the cerebellum of treated mice. It is difficult to appreciate the cerebellum morphology, probably because they used a different magnification respect to the previous images. In order to better understand the signal of both markers, the author should provide images with higher resolution and lower magnification. Moreover, they could provide the inflammation analysis (IBA1 and GFAP markers) in the hippocampus of treated mice. Figure 6A: the authors should perform a tunnel assay in a representative brain region of KO mice treated with higher doses of PF in order to exclude any toxic effect on the neuronal cells. Figure 6C: the authors performed the IF experiments with IBA marker in order to evaluate the microglia activation in the brain regions of KO mice treated with different concentration of PF. In order to better appreciate the morphology of brain regions and IBA distribution, the authors should perform immunofluorescence experiments in the mouse brain regions with IBA1 marker and DAPI.
Page 11: To understand the role of CB1 in the neuropathology development, the authors should improve the invitro studies in NPC fibroblasts by: -Quantifying the CB1 receptor in WT and NPC fibroblast cell line from patients with WB experiments.
-Treating the NPC and WT fibroblast cell line with SMase and evaluating the SM and CB1 levels upon the treatment.
-Performing IF analysis for CB1 and LAMP1 markers in order to evaluate the CB1 co-localization with lysosomes and a possible block of autophagy in NPC fibroblast cell line.
Moreover, in order to quantify the levels of CB1 receptors they should perform WB experiments in the brain samples and also Immunofluorescence staining for CB1 receptor in the hippocampus of WT and NPC1 nmf164 mice. To test the therapeutic effectiveness of the PF treatment in NPC1 nmf164 mouse models, the authors could perform a short term study with the best dosage of PF and analyze the CB1 and SM levels in the brain samples of treated mice.

Minor comments:
FigureS1A: the authors should explain how they calculated the fluorescence intensity in the hippocampus region of treated mice and if they normalized the data respect to the brain areas or to the number of cells analyzed.
-Page 6: In the results it has been described a change in CB1 distribution in the ASM-KO neuronal mouse cell line the. The authors should comment this part in the results and in the discussion.   Figure 7A, 7B: The authors should show immunofluorescence images at low magnification in order to better appreciate the CB1 reduction and distribution in hippocampus NPC1 nmf164 mice.

POINT-BY-POINT ANSWER TO REFEREES
Please find below the Point-by-Point answer to referees in which we detailed the experiments and modifications done in our manuscript EMM-2019-11776. These are highlighted in the main text (underlined and yellow labelling) and in the following new Figures: Figure 1D,E,F,G,H; Figure 2B,D,E,H,J,N; Figure 4A; Figure 5C,D,E,F; Figure 7A,C,D,E,F,G,H and new Appendix Supplementary Figures S1, S2, S3, S4, S5, and S7.

Referee #1
We thank this referee for considering our report interesting and timely. We are grateful for his/her careful evaluation and queries that have been addressed as follows: Major points 1) It is very difficult to judge protein levels of CB1 from the presented immunofluorescent images (Fig 1D). This result should be confirmed by Western blot analysis.
We have performed Western blot analysis to quantify CB 1 levels in cerebellar, hippocampal and cortical extracts. No significant differences were observed although a tendency to reduction was found in the cerebellum ( Figures 1D and S1). This moved us to improve the original immunofluorescence analysis and checked in detail the cell-type specific expression of CB 1 by co-labelling with neuronal, astrocytic and microglia markers in the cerebellum (the most affected area in the disease). CB 1 levels were significantly reduced in the Purkinje cells (identified by calbindin staining) as well as the co-localization of CB 1 with these neurons as indicated by a diminished Mander´s coefficient ( Figure 1E). However, CB1 levels were not significantly reduced in astrocytes (identified by GFAP staining) or microglia (identified by F4/80) ( Figure  1F,G). While the co-localization studies revealed an unchanged CB 1 level in microglia it was notably increased in astrocytes, probably due to the higher number of these cells in the cerebellum of ASM-KO compared to WT mice ( Figure 1F,G). These results, together with the reduced levels of CB1 observed in neurons of the cerebellum and medium bulb in the ASMD patient ( Figures 1H and S2) as well as the results in cultured hippocampal neurons from the ASM-KO mice (Figures 2A,B), lead us to conclude that reduction in CB 1 in ASMD mainly affects neuronal cells compared to glia cells. This may explain why we see no significant differences in CB 1 levels when monitored by Western blot in total extracts. This is also discussed on pages 6 and 13 of the revised text.
2) Immunofluorescent analysis of a single ASMD patient is not conclusive. It is hard to estimate whether the aggregation signal observed in the NPA patient is indeed specific, and if so, this may not be a common pattern. MAP2 signal is very low, even in the control, and almost absent from the NPA patient. Severe neurodegeneration would also interfere with the interpretation of the CB1 levels.
We agree with the reviewer that analysis of a single ASMD patient is not conclusive. However, being such a rare disease tissue samples from ASMD patients, especially from the brain, are almost impossible to obtain. We are indeed very grateful to the Wylder Nation Foundation for sharing with us the only brain tissue they had available.

18th Jul 2020 1st Authors' Response to Reviewers
We believe the results from this single individual are worth showing, particularly since information on human patients is so limited in the literature. Nevertheless, to address this concern, in the revised text we no longer highlight the aggregation pattern and now stress that the results in the human samples are not conclusive since they are derived from only one patient (page 6). To rule out that the decreased CB 1 levels observed in the NPA patient are due to severe neurodegeneration we have performed co-labelling with calbindin (for Purkinje cells of the cerebellum) and with MAP2 (for neurons of the medium bulb) and chose for CB 1 quantification only those neurons showing similar integrity in the control and NPA samples. We include these images and quantification in the Figures 1H and S2, which confirmed the significant reduction of CB 1 levels in neurons of the NPA patient.
3) The authors suggest that cellular misdistribution of CB1 occurs in ASM-KO neurons and upon sphingomyelin supplementation. It would be beneficial to include lower magnification images showing more than a single neuron. Aggregation phenotype of CB1 is more obvious in Fig 2C and F

, compared to Fig 2B and E. Can the cellular misdistribution phenotype of CB1 also be observed in ASM-KO neurons in vivo?
We agree with this referee, and a similar comment from referee 3, that the way we depicted the results in the original panels C, F and B, E was redundant and confusing. Following the suggestion of referee 3 we have now merged panel C with F and B with E in Figures 2C and 2I, respectively. However, we believe that the high magnification images provide a better illustration of the CB 1 cellular misdistribution than low magnification ones. This, together with the graphs included in the figure showing the quantification of CB 1 associated fluorescence and misdistribution in at least 30 neurons per culture in three different cultures move us to kindly ask this referee to keep the original high magnification images. To address the query about CB 1 misdistribution in vivo we have quantified the degree of co-localization of CB 1 and the lysosomal marker LAMP1 in Purkinje cells of the cerebellum of WT and ASM-KO mice. In agreement with the in vitro data we observed increased localization (quantified by the Mander´s coefficient) of CB 1 in lysosomes in the ASM-KO compared to WT mice ( Figure 2D).

4) Lysosomal accumulation of CB1 has been observed in ASM-KO or upon SM
supplementation. The authors discuss that lysosomal degradation may contribute to the reduced levels of CB1 (discussion, page 12). Would inhibition of lysosomal degradation in cultured primary neurons (for example upon addition of SM) effect CB1 levels and its cellular distribution? Following the suggestion of this referee, we have quantified CB 1 levels under conditions of lysosomal function inhibition. We have done so in WT neuronal cultures in which SM was added or not in the presence or absence of the lysosomal inhibitor Bafilomycin. Lysosomal inhibition prevented the SM-induced reduction of CB 1 levels by 61% ( Figure 2J). This result supports the concept that lysosomal accumulation and degradation of CB 1 upon high SM levels contributes to CB 1 reduction. Along these lines, can normal cellular distribution of CB1 be restored upon treatment of ASM-KO with SMase? Yes, SMase treatment reduced the aberrantly high co-localization of CB 1 with lysosomes in ASM-KO cultured neurons ( Figure 2N). Fig 2I does not support the reported 45% increase in CB1 protein levels.

5) Western blot image of CB1 in
We now show a more representative WB example of the reported 45% mean increase ( Figure 2M). However, as we also indicate in the text, this result did not reach statistical significance (page 7).
6) The authors report a body weight gain in ASM-KO mice treated with PF and claim that no differences were observed in the WT. However, differences can also be observed upon PF application in the WT (Fig 4A). No statistics was included to support the rescue phenotype. We have performed statistical analysis of the slopes of the weekly weight data. This analysis indicated a significant difference between the vehicle treated ASM-KO mice with respect to the other three groups (vehicle treated WT mice and PF treated WT and ASM-KO mice) ( Figure 4A). (Fig 4F) has not been performed and could add additional, therapeutically relevant information when compared with the treatment efficacy over 8 weeks. We apologize for not being able to perform the histopathological analysis this referee suggests since we did not collect tissue samples from the mice devoted to the survival analysis. Figure 5 should be improved to have the analysis more consistent. AEA and NSM levels were monitored in the hippocampal extracts only (Fig 5A and B). SM analysis was done for the cerebellum, hippocampus and the cortex (hippocampus was indeed less affected). Further immunohistological analysis only revealed changes in the cerebellum. This makes it difficult for a reader to follow the logic. It would be helpful to have AEA, NSM, CB1 and SM levels plus histological examination for the same brain region (or at least the one affected most). We apologize for the confusing presentation of the data in the original figure. Following the reviewer's suggestion we now present AEA, NSM, CB 1 and SM levels plus histological examination all from the cerebellum, which is the most affected area in the disease ( Figure 5). Showing only Lamp1 staining in Fig 5D is  We now show co-labelling of CB 1 with DAPI in the cerebellum ( Figure 5E) and colabelling of CB 1 with Lamp1 in Purkinje cells ( Figure 5F). This has allowed us to quantify lysosomal area in these cells and to determine that PF treatment reduced the aberrant high co-localization of CB 1 with these organelles in the ASM-KO mice as indicated by changes in the Mander´s coefficient ( Figure 5F).

8)
10) Acute treatment is not offering any additional information beyond toxicity assessment. SM levels were only slightly reduced in the cortex. However, the neuroinflammatory effect was also seen in the hippocampus where no SM reduction could be detected. The authors should consider if the statement on page 11 "...and also revealed efficacy after a single administration of high PF doses." is fully supported by the data. We have corrected this statement accordingly. Fig 7 more consistency would be helpful for a reader. Fig 7A shows CB1 levels in the cerebellum of an NPC1-deficient mouse model, human tissue reveals CB1 staining of a hippocampus and cultured cells of an NPC patient treated with PF were only examined for the levels of SM (CB1 analysis is missing). Analysis of other phenotypes such as cellular distribution of CB1, lysosomal morphology or levels of cholesterol upon PF treatment would provide additional information to judge the therapeutic potential of CB1 modulation in NPC disease. We thank the referee for this suggestion, which encouraged us to deepen our analysis of NPC. We have focused on the cerebellum, which is a most affected brain area in the disease. As for the ASM-KO mouse we did not find significant changes in CB 1 levels analyzed by Western blot in cerebellar extracts of NPC nmf164 compared to WT mice ( Figure 7A). However, immunofluorescence analysis showed a significant 32% CB 1 protein reduction in the Purkinje cells ( Figure 7B). As with the NPA patient it is very difficult to obtain brain tissue form NPC patients. Still, we believe the findings in the human scenario, while not conclusive, are worth showing. To be more consistent with the mouse data we have analyzed CB 1 levels in the Purkinje cells of the cerebellum of the control and NPC-affected children finding a significant reduction in the latter ( Figure 7C). To satisfy this referee query we have analyzed several phenotypes in control and NPC cultured fibroblasts treated or not with PF. PF treatment in NPC fibroblasts reduced SM and cholesterol levels, and diminished the aberrantly high colocalization of CB 1 in lysosomes ( Figures 7D, E, F). Besides extending the analysis in the cultured fibroblasts, we have performed an acute in vivo treatment with PF. A single oral administration of the high dose (5mg/kg) of PF reduced, after 48 hours, the levels of SM and cholesterol in the cerebellum of NPC nmf164 mice ( Figure 7G) and diminished inflammation as indicated by the lower area of microglia in the PF treated mice ( Figure  7H).

Minor issues 1) Authors should discuss differences between CB1 (neuronal) and CB2 (immune cells) expression patterns that may result in different signalling networks in neurons vs microglia. Potential common (or diverse) mechanism of FAAHi effect in different nervous system cells could be discussed.
This is now discussed on page 13. Fig 4A and F to distinguish more easily between different conditions. The figure legends have been improved.

2) Authors should consider improving figure legends such as in
3) Consider re-phrasing "...dual impact" on page 11 of the discussion. Dual impact has been rephrased to "multiple impacts".

4) Reference Nr 50 on page 15 of the discussion should be corrected according to the common reference style.
We thank this reviewer for the careful revision of our paper. The style of these references has been corrected.

Referee #3
We thank this referee for the comments, for finding the study of interest, and for acknowledging the in vitro study in cultured neuronal cells as an optimal method to validate the role of CB 1 receptor and its direct link with the SM accumulation. We explain below how we have addressed his/her queries: Major Comments: - Figure 1

and Results: The fluorescence intensity measurement doesn't represent a quantitative analysis of CB1 protein. The authors should perform the quantitative analysis of CB1 by WB experiments in the same brain regions used for qPCR experiments.
Western blot analysis of CB 1 levels in the cerebellum is shown in the Figure 1D. For the sake of simplicity, and following the recommendation of referee 1, we have focused on this brain area in all panels of Figure 1, since it is the most severely affected in the disease. WB analysis of CB1 levels in the hippocampus and cortex is shown in Figure  S1. The additional analysis of cell type expression of CB 1 by immunofluorescence, which uncovered the specific CB 1 reduction in neurons and not in glial cells ( Figures  1E,F,G), may explain why we do not see differences in the levels of this receptor by Western blot of total extracts. This is discussed in the revised text. We performed all experiments in the prefrontal cortex. This is now indicated in the legend of Figure S1 and in page 6.
- Figure 1D: the author should improve the resolution of cerebellum images in which it is difficult to appreciate the signal of CB1 receptor. We now provide with better resolution images of CB 1 co-labelled with different cellular markers such as Calbindin (for Purkinje cells), GFAP (for astrocytes) and F4/80 (for microglia) in the cerebellum of WT and ASM-KO mice (Figures 1E,F,G). We also show the quantification of CB 1 intensity and its degree of co-localization (Mander´s coefficient) with each cell type (Figures 1E,F,G).
- Figure 1E: In the IF staining on cerebellum, the MAP2 signal is almost undetectable in the slides of both CTR and NPA patients and it is also complicated to understand the morphology of this brain region. The author should improve the IF experiments in order to better analyze the decrease of CB1 signal in the cerebellum of NPA patients. They could also try to perform a labelling with Calbindin/CB1/TOPRO markers. We have improved the MAP2 staining in the medium bulb of the control and NPA patients ( Figure S2). Since MAP2 is not a good marker for neurons in the cerebellum we have followed this reviewer suggestion and performed triple labelling with CB 1 , the specific Purkinje cell marker calbindin, and TOPRO ( Figure 1H). Quantification of CB 1 reduction in the Purkinje cells of the NPA patient compared to the control child is added in the graphs of Figures 1H and S2. We now provide WB analysis of CB 1 levels ( Figure 2B) showing a 66% reduction in the ASM-KO compared to WT cultured hippocampal neurons. We have also measured the SM levels confirming a 78% increase in the ASM-KO neurons ( Figure 2E).
- Figure 2C: Since the measure of CB1 fluorescent intensity has been already quantified in the figure 2B, the authors should replace the CB1 quantification graph in Fig 2C  with

another one in which they show the LAMP1 quantification as marker of autophagy impairment. Moreover, the authors showed an increase of CB1 in the cell body of ASM-KO cells. Since this increase is associated with an increase of SM levels, the author should show a quantification of SM levels among WT and ASM-KO cells.
We apologize for the redundancy in the original figure and thank this reviewer for the suggestion. We have now merged panels B and C in the Figure 2C in which we also include the colocalization of CB1 with Lamp1 in WT and ASM-KO cultured cells quantified by the Mander´s coefficient. We have also quantified the accumulation of Lamp1 as marker of autophagy impairment in the ASM-KO neurons. Although we do not include the data in the figure, since it confirms those published in our previous work (Gabande-Rodriguez et al., Cell Death Diff 2014), Lamp1-associated intensity and the area of lysosomes increased by 64% and 74%, respectively in the ASM-KO neurons. We now provide the quantification of SM levels showing that they are 78% higher in ASM-KO neurons compared to WT ( Figure 2E). In addition we show that the increased co-localization of CB 1 and Lamp1 also occurs in vivo in the Purkinje cells of the cerebellum of ASM-KO compared to WT mice ( Figure 2D).
- Figure 2D-2E: The authors could show the qPCR and the IF experiments in WT and also ASM-KO cell lines. Furthermore, they should perform the WB analysis for CB1 receptor with the relative protein quantifications in the following neuronal cell lines: WT, WT+40uM SM and ASM-KO. We now show qPCR results and WB analysis of CB 1 in WT and ASM-KO neuronal cultures (Figures 2A and 2B) and in WT cultures treated or not with SM ( Figures 2G  and 2H).
- Figure 2F: Since the authors quantified in both panels 2E and 2F the CB1 signal using the same cell lines, they should replace the figure 2E with figure 2F. As with the original panels B and C we apologize for the redundancy and have merged the original panels E and C in the Figure 2I in which we quantified not only the levels of CB 1 but also the Mander´s coefficient of CB 1 and Lamp1 co-localization in WT neuronal cultures treated or not with SM.
- Figure 2G, 2H, 2I: The authors should also perform the experiments in the WT cell lines as positive control. These experiments are now shown in Figure S3 and mentioned in page 8.
- Figure 3A-3E: the authors should also perform the experiments in the WT neuronal cell line as CTR cell line. These experiments are now shown in Figure S5 and mentioned in page 9.  Figure 3C: The authors should directly quantify of CB1 receptor levels by WB experiments in WT and ASM-KO cells treated with AEA with or without the GW4869 Figure 3F:

In order to both evaluate the neuronal morphology and CB1 neuronal distribution, the authors should perform a Co-IF experiments with MAP2 and CB1 receptor markers on ASM-KO treated cells and WT cells. The authors should also quantify the levels of CB1 in ASM-KO cells by WB analysis upon the treatment with FAAH inhibitors.
The evaluation of expression levels of CB 1 receptor by WB in ASM-KO neurons treated with AEA with or without GW4869, with the different FAAH inhibitors, or with PF and SR-PF are now shown in the Figure S4 and mentioned in pages 8-9.  Figure 5D, 5E: The authors show in the panel D, the LAMP1 staining in cerebellum samples of treated mice. Since it is difficult to appreciate to LAMP1 localization and the cerebellum morphology, the authors should perform a co-labelling with LAMP1 markers and DAPI. Moreover, since the authors demonstrated an increase in the LAMP1 and CB1 colocalization in ASM-KO cells, they should also perform a colabelling experiment with LAMP1 and CB1 markers in cerebellum and hippocampus of treated mice. This data should support the biochemical experiments performed in the same animals. Figure 5F-5G: the authors show the IBAI and GFAP staining in the cerebellum of treated mice. It is difficult to appreciate the cerebellum morphology, probably because they used a different magnification respect to the previous images. In order to better understand the signal of both markers, the author should provide images with higher resolution and lower magnification. Moreover, they could provide the inflammation analysis (IBA1 and GFAP markers) in the hippocampus of treated mice.
Following the reviewer recommendations, and to avoid the confusing mixture of results in different brain areas, we have focused our quantifications in the cerebellum that is the most affected region in the disease. Thus, we have analyzed in this brain area all the parameters requested for this referee in the hippocampus. NSM levels have been analysed in the cerebellum by WB in the Figure 5C. We also provide with intensity analysis of CB 1 in the cerebellum ( Figure 5E) and with co-localization analysis of Lamp1 and CB 1 in Purkinje cells of WT and ASM-KO quantified by Mander´s coefficient (Figure 5F). GFAP and Iba1 immunofluorescence analysis in the cerebellum remain as in the original figure (Figures 5H and 5I). Figure 6A: the authors should perform a tunnel assay in a representative brain region of KO mice treated with higher doses of PF in order to exclude any toxic effect on the neuronal cells.
To determine cell toxicity we have used cleaved caspase3 staining as a tunnel assay to detect apoptotic cells, in the cerebellum of WT and ASM-KO mice treated with the different PF doses. These results confirm the lack of significant toxicity of PF treatments, with the highest dose showing great variability, and are shown in the Figure  S7 and mentioned in page 12. Figure 6C: the authors performed the IF experiments with IBA marker in order to evaluate the microglia activation in the brain regions of KO mice treated with different concentration of PF. In order to better appreciate the morphology of brain regions and IBA distribution, the authors should perform immunofluorescence experiments in the mouse brain regions with IBA1 marker and DAPI. DAPI staining made less clear the Iba1 labelling, which is the focus on Figure 6C. Therefore, we kindly ask this referee to keep the images as Iba1 single staining. Detection of CB 1 by WB is not straightforward. We were able to set specific conditions to do so in the mouse samples but none of the antibodies tested worked to detect CB 1 in the human fibroblasts. As in the ASM-KO mice, quantification of CB 1 levels by WB of total cerebellar extracts from NPC1 nmf164 mice did not show significant differences compared to age-matched WT mice ( Figure 7A), likely because the CB 1 reduction is specific to neurons. However, by immunofluorescence we observed a significant 32% reduction of CB 1 levels in the Purkinje cells of the cerebellum of NPC1 nmf164 mice ( Figure 7B) (page 13).
-Performing IF analysis for CB1 and LAMP1 markers in order to evaluate the CB1 colocalization with lysosomes and a possible block of autophagy in NPC fibroblast cell line.
We performed co-labelling of CB 1 and Lamp1 in control and NPC fibroblasts. The quantification of the Mander´s coefficient indicated an increased co-localization of CB 1 and Lamp1 in the NPC compared to the control fibroblasts that was diminished by PF treatment ( Figure 7F) (page 13).
-To test the therapeutic effectiveness of the PF treatment in NPC1 nmf164 mouse models, the authors could perform a short term study with the best dosage of PF and analyze the CB1 and SM levels in the brain samples of treated mice. Following this referee's suggestion we conducted a short-term study by administering a single dose of PF (5mg/kg) to NPC1 nmf164 mice. After 48 hours we observed a reduction in SM and cholesterol levels in the cerebellum of the PF-treated mice compared to the vehicle-treated ( Figure 7G). Moreover, the acute PF treatment diminished the inflammation as indicated by the reduction in the area of microglia ( Figure 7H) (page 13).

Minor comments:
FigureS1A: the authors should explain how they calculated the fluorescence intensity in the hippocampus region of treated mice and if they normalized the data respect to the brain areas or to the number of cells analyzed. We now clarify this issue in the methods. In the original studies fluorescence intensity was normalized to the brain area. In the new immunofluorescence studies performed in specific cell types fluorescence intensity was calculated per cell area.   We apologize for the discrepancy that has now been corrected Figure 7A, 7B: The authors should show immunofluorescence images at low magnification in order to better appreciate the CB1 reduction and distribution in hippocampus NPC1 nmf164 mice. Following the referee's suggestion, and to avoid mixed data from different brain areas, we have now focused the study on the cerebellum, which is the most affected brain area in the disease. CB 1 levels and distribution have been analyzed by immunofluorescence in the Purkinje cells of NPC1 nmf164 mice ( Figure 7B) and of the NPC patient ( Figure  7C).
1st Aug 2020 1st Revision -Editorial Decision 1st Aug 2020 Dear Dr. Ledesma, Thank you for the submission of your manuscript to EMBO Molecular Medicine. We have now heard back from the two referees whom we asked to re-evaluate your manuscript.
You will see that while the 2nd referee is now satisfied, the 1st one still isn't and we decided to give you another chance to address the following issues: -statistical significance must be provided (exact n and p-values, not a range, along with the statistical test used) -CB1 misdistribution in vivo & LAMP1 staining -immunohistochemistry data in human specimen -and finally, we would like you to also try at least to confirm the IFA data with biochemical experiment using positive and negative controls that would eventually attest to the population heterogeneity/cell specificity should this attempt be unsuccessful.
Please revise your article as requested and provide a point-by-point letter. According to the mature of the revision, we may reserve the right to ask the referee to evaluate the new data.
I look forward to seeing a revised form of your manuscript as soon as possible. I appreciate the efforts that authors put into performing additional experiments that were requested and improving the manuscript quality. However, few concerns regarding this study still remain. The major remaining concern is that immunofluorescent data could to a larger extent not be validated by other means such as biochemical experiments. The authors argue with cell specificity of their effects, but taking into account that neuronal population is not a minor population in brain extracts, one would have expected effects to be detectable by biochemical analysis as well. Apart from this issue, statistical significance is missing in several experiments, questioning the robustness of the data. Furthermore, CB1 misdistribution in vivo relies on LAMP1 staining that shows an unusual pattern in the ASM KO that differs between the presented figures (Fig 2D and Fig 5F), making data interpretation difficult. Is the reduction in Fig 7B specific for CB1 (would calbindin staining of the same image still reveal preserved Purkinje neurons, was the Purkinje cell marker included into this analysis)?. In addition, immunohistochemistry data in human specimens still remain hard to interpret, beside the fact that only 1 patient has been analyzed.
Referee #3 (Comments on Novelty/Model System for Author): The revised version of the manuscript has clearly showed the clinical potential of a new therapeutic approach for the treatment of severe neurological pathologies like the acid sphingomyelinase deficiency (ASMD). The in vitro and in vivo models used are adequate for studying and validating the efficacy of the treatment for ASMD and other sphingolipidoses. Moreover, the new data shown have increased the quality and medical impact of the article.
Referee #3 (Remarks for Author): The revised version of the manuscript from Bartoll et al., entitled: "Inhibition of Fatty Acid Amide Hydrolase Prevents Pathology in Neurovisceral Acid Sphingomyelinase Deficiency by Rescuing Defective Endocannabinoid Signaling" has been significantly improved. The authors accurately have addressed all the requests and suggestions previously indicated. In particular, they thoroughly performed the new in-vivo and in-vitro experiments providing an indepth description of the showed data. In addition, it is worth noting that the authors notably ameliorated the quality of images and improved the discussion.
As a conclusion, this new version of manuscript is suitable for the publication in EMBO Molecular Medicine.
Below I detail the experiments and modifications done, which are highlighted in the revised text and in the modified Figure 2D and the new Appendix Supplementary Figure S2. 1.The major remaining concern is that immunofluorescent data could to a larger extent not be validated by other means such as biochemical experiments. The authors argue with cell specificity of their effects, but taking into account that neuronal population is not a minor population in brain extracts, one would have expected effects to be detectable by biochemical analysis as well.
To attest cell population heterogeneity and imbalance in the ASM-KO compared to WT mice we now provide with the quantification of astrocytes and microglia by immunofluorescence and also by Western blot using specific cell markers in cerebellar extracts (New Supplementary Figure S2). These data confirm that the population of astrocytes and microglia, where we do not find reduced CB1, increase by 6-fold and 4-fold, respectively (as indicated by immunofluorescence). In contrast neuronal population diminishes ( Figure 5F). These results strongly support that the presence of remarkable astrogliosis and microgliosis, could prevent detecting the neuronal specific reduction of CB1 in the biochemical experiments (now mentioned in page 6).
2. Statistical significance is missing in several experiments, questioning the robustness of the data. Exact n and p-values along with the statistical test used is now provided in the figure legends for all significant values. Due to the Covid19 outbreak during the time of revision of this manuscript we were obliged to drastically reduce mouse colonies and cell culture work. This, together with the low fertility of NPC mice and the difficulty to grow enough cells from the human NPC fibroblast line, prevented the analysis of a larger sample size. This may explain why in several experiments in Figure 7 the clear trend we observe in the data does not reach statistical significance. (Fig 2D and Fig 5F), making data interpretation difficult. We realized that the brightness in the images shown in figure 2D was higher than in those in Figure 5F. This might have been confusing for the referee. We now show images with the same brightness settings that evidence the similarities in the patterns in both figures. Indeed, Lamp1 staining show an unusual pattern in the ASM-KO brains compared to WT since lysosomes are enlarged due to lipid 12th Aug 2020 2nd Authors' Response to Reviewers