VWCE modulates amino acid-dependent mTOR signaling and coordinates with KICSTOR to recruit GATOR1 to the lysosomes

The mechanistic target of rapamycin complex 1 (mTORC1) is a crucial regulator of cell growth. It senses nutrient signals and adjusts cellular metabolism accordingly. Deregulation of mTORC1 has been associated with metabolic diseases, cancer, and aging. Amino acid signals are transduced to mTORC1 through sensor proteins and two protein complexes named GATOR1 and GATOR2. In this study, we identify VWCE (von Willebrand factor C and EGF domains) as a negative regulator of amino acid-dependent mTORC1 signaling. Knockdown of VWCE promotes mTORC1 activity even in the absence of amino acids. VWCE interacts with the KICSTOR complex to facilitate the recruitment of GATOR1 to the lysosomes. Bioinformatic analysis reveals that expression of VWCE is reduced in prostate cancer. More importantly, overexpression of VWCE inhibits the development of prostate cancer. Therefore, VWCE may serve as a potential therapeutic target for the treatment of prostate cancers.

While indeed there is value in elucidafing mechanisms by which KICSTOR coordinates GATOR1 complex funcfion, in my opinion the model presented here is too preliminary to be published in its current form.My concerns are outlined below.
1) I am concerned by the lack of aftenfion given to the molecular features of the VWCE protein itself, and whether it is plausible that it could perform the funcfions suggested in the model.For example, VWCE contains a predicted N-terminal signal sequence, which likely directs it to the ER lumen upon translafion.Furthermore, it has predicted EGF-like domains, which contain several cysteines capable of disulfide bonds, features that almost exclusively exist in lumenal or extracellular proteins / regions of proteins.Therefore, the likelihood that VWCE exists cytoplasmically, which the model as presented would require, seems very low.As it stands, the authors did not present any evidence that VWCE exists on the cytoplasmic face of lysosomes.If the authors believe that VWCE is in fact a cytoplasmic-facing protein, then they must provide a compelling argument in favour of this and ideally provide experimental evidence to support it -e.g., protease treatment of immuno-isolated lysosomes.If VWCE is targefted to the lysosome lumen and nonetheless performs the funcfions suggested in the paper, this needs to be explicitly discussed, and some sort of 'inside-out' mechanism would need to be invoked.
2) Why was immunofluorescence of the VWCE-FLAG construct expressed in cells never shown?Was this experiment aftempted?
3) What were the criteria that were used to idenfify the 'candidate interacfing proteins' from the BioPlex dataset listed in Ext Fig 1?It is stated that these are proteins that can 'interact with at least two known components of the mTORC1 pathway' but it is not defined what consfitutes an 'interacfion' and what components of the mTORC1 were considered.Please provide details.
4) Some more experimental detail could be provided with respect to starvafion / resfimulafion experiments.For example, the amino acid-free media that was used for starvafion does not contain glucose -was glucose added or were cells starved of glucose as well as amino acids?When serum starvafion was tested, was this using dialyzed FBS, and was this in standard or amino acid-free DMEM? 5) In many instances the differences observed by Western blot are subtle, yet the conclusions drawn by the authors are absolute.The conclusions would be more convincing if data from independent (> / = 3) Western blot experiments were quanfified and summarized.
6) The immunofluorescence images are small and difficult to interpret.If possible, please make these images larger, especially the magnified insets.

Reviewer #3 (Remarks to the Author):
The manuscript by Zhao et al describes VWCE, a novel regulator of the mTORC1 signaling acfivity.Mostly via protein interacfion and cell imaging experiments, the authors provided data supporfing a model where VWCE interacts with the KICSTOR complex on the lysosome to recruit GATOR1 and suppress mTORC1 acfivity.The authors also presented funcfional data demonstrafing a tumor-suppressing role of VWCE in prostate cancer cells.

Suggesfions for revision:
In Fig. 3b and Extended Fig. 3a-c, bands sfill show significant interacfions between GATOR2 component and GATOR1 or KICSTOR component (e.g.MISO and DEPDC5, MISO and SZT2) under the VWCE knockdown condifion.Therefore, I am not sure if the authors can conclude that interacfions with GATOR2 complex is disrupted by VWCE knockdown.If anything, the effect seems to be subtle.
In Fig. 3c, the decrease of NPRL2 under siVWCE is convincing, but the decrease of NPRL3 is subtle.How about the other component DEPDC5?It is not shown in the figure.Also, in Fig. 3d-e, only NPRL2 decrease is shown.Is siVWCE affecfing specifically NPRL2 or the whole GATOR1 complex?
The authors state that "VWCE may serve as a potenfial therapeufic target for the treatment of prostate cancer".Besides the in vitro colony formafion assay, xenografting of prostate cancer cells overexpressing VWCE onto nude mice may befter support this point.
The authors should discuss why VWCE exert different effects on prostate cancer and liver cancer, and how it may shed light onto the molecular funcfions of VWCE.
Other minor points: Fig. 2g.Why is LAMP2 expression so low in the whole cell lysates?

Reviewers' comments:
We thank all the reviewers for the encouraging remarks on our work.We are particularly grateful for the constructive comments that have helped us to improve the study.

Reviewer #1:
By knockdown of VWCE, they showed that S6K remains phosphorylated despite absence of amino acids, indicating VWCE is a negative regulator of mTORC1.Next, they performed epistasis analysis and found that VWCE associates with KICSTOR and is downstream of GATOR2 but upstream of GATOR1 and Rag.By lysosome immune-capture, they found that VWCE localizes to this compartment in an aa-insensitive manner but is independent of KICSTOR.They also found that VWCE is required for tethering GATOR1 to KICSTOR at the lysosomes.They next analyzed different cancers and found that VWCE expression is low in prostate cancers and that overexpression prevented colony formation and diminished mTORC1 signaling.Based on these findings, they concluded that "VWCE modulates aa-dependent mTOR signaling and coordinates with KICSTOR to recruit GATOR1 to lysosomes." The studies are interesting and identify another protein that is involved in mTORC1 regulation by amino acids via lysosomal localization.The data are mostly robust.However, most of the analyses are done on overexpressed VWCE.The knockdown of VWCE (protein expression) was also not verified in all of the experiments.Thus it is not clear how relevant is VWCE interaction with KICSTOR and its regulation of mTORC1 signaling at endogenous levels.The mechanisms as to how VWCE mediate KICSTOR recruitment of GATOR1 were also only superficially addressed.Lastly, the role of VWCE as a possible tumor suppressor is interesting but was not fully interrogated either.Hence, the studies are highly preliminary.

Reviewer #2:
The authors present a model in which von Willebrand factor C and EGF domain-containing protein (VWCE) acts as a negative regulator of mTORC1 activation by amino acids through coordination of KICSTOR function.VWCE was identified as a potential regulator of mTORC1 regulation by amino acids by mining data from the published BioPlex interaction proteomics dataset.Using RNA interference and a variety of biochemical and cell biology approaches the authors suggest that VWCE coordinates KICSTOR function, through an unknown mechanism, to recruit GATOR1 to the lysosome surface.By mining The Cancer Genome Atlas they identify cancers with decreased VWCE expression, which include prostate cancers, and use the Cancer Cell Line Encyclopedia to select prostate cancer cell lines for functional analysis of VWCE.Consistent with their model, increasing expression of VWCE in prostate cancer lines restored mTORC1 amino acid sensitivity and inhibited anchorage-independent cell growth.
While indeed there is value in elucidating mechanisms by which KICSTOR coordinates GATOR1 complex function, in my opinion the model presented here is too preliminary to be published in its current form.My concerns are outlined below.
1) I am concerned by the lack of attention given to the molecular features of the VWCE protein itself, and whether it is plausible that it could perform the functions suggested in the model.For example, VWCE contains a predicted N-terminal signal sequence, which likely directs it to the ER lumen upon translation.Furthermore, it has predicted EGF-like domains, which contain several cysteines capable of disulfide bonds, features that almost exclusively exist in lumenal or extracellular proteins / regions of proteins.Therefore, the likelihood that VWCE exists cytoplasmically, which the model as presented would require, seems very low.As it stands, the authors did not present any evidence that VWCE exists on the cytoplasmic face of lysosomes.If the authors believe that VWCE is in fact a cytoplasmic-facing protein, then they must provide a compelling argument in favour of this and ideally provide experimental evidence to support it -e.g., protease treatment of immuno-isolated lysosomes.If VWCE is targetted to the lysosome lumen and nonetheless performs the functions suggested in the paper, this needs to be explicitly discussed, and some sort of 'inside-out' mechanism would need to be invoked.
We thank the reviewer for the constructive comments.We have followed the reviewer's advice to treat immuno-isolated lysosomes with trypsin.Our results indicated that lysosomal VWCE can be digested by trypsin, demonstrating that VWCE is localized on the cytosolic surface of lysosomes (Extended Data Fig. 2j).
2) Why was immunofluorescence of the VWCE-FLAG construct expressed in cells never shown?Was this experiment attempted?
Following the reviewer's advice, we conducted an immunostaining assay.Our findings revealed that VWCE is localized both in the nucleus and the cytosol.Within the cytosol, VWCE is constitutively present on lysosomes, regardless of the availability of amino acids (Extended Data Figs. 2h,i).
3) What were the criteria that were used to identify the 'candidate interacting proteins' from the BioPlex dataset listed in Ext Fig 1?It is stated that these are proteins that can 'interact with at least two known components of the mTORC1 pathway' but it is not defined what constitutes an 'interaction' and what components of the mTORC1 were considered.Please provide details.
Based on the reviewer's advice, we've included detailed information in the revised manuscript.Details of the bait components of mTORC1 and the candidate prey proteins are provided in Supplementary Information Table 1.4) Some more experimental detail could be provided with respect to starvation / restimulation experiments.For example, the amino acid-free media that was used for starvation does not contain glucose -was glucose added or were cells starved of glucose as well as amino acids?When serum starvation was tested, was this using dialyzed FBS, and was this in standard or amino acid-free DMEM?
We appreciate the reviewer for highlighting these points and guiding us to enhance the clarity of our study.The media used for amino acid starvation was supplemented with 4.5 g/L glucose.Meanwhile, the media used for serum starvation was standard DMEM without FBS.Detailed descriptions have been added to the 'Cell Treatments' section of the Methods." 5) In many instances the differences observed by Western blot are subtle, yet the conclusions drawn by the authors are absolute.The conclusions would be more convincing if data from independent (> / = 3) Western blot experiments were quantified and summarized.
In line with the reviewer's suggestions, we have quantified the Western blots, which have not yielded significantly noticeable effects.The results have been summarized based on three independent experiments (Figs. 3b,c;4f and Extended Data Figs. 2b;6b,d).
6) The immunofluorescence images are small and difficult to interpret.If possible, please make these images larger, especially the magnified insets.
We have followed the reviewer's advice to enlarge the magnified insets.

Reviewer #3:
The manuscript by Zhao et al describes VWCE, a novel regulator of the mTORC1 signaling activity.Mostly via protein interaction and cell imaging experiments, the authors provided data supporting a model where VWCE interacts with the KICSTOR complex on the lysosome to recruit GATOR1 and suppress mTORC1 activity.The authors also presented functional data demonstrating a tumor-suppressing role of VWCE in prostate cancer cells.

Suggestions for revision:
In Fig. 3b and Extended Fig. 3a-c, bands still show significant interactions between GATOR2 component and GATOR1 or KICSTOR component (e.g.MISO and DEPDC5, MISO and SZT2) under the VWCE knockdown condition.Therefore, I am not sure if the authors can conclude that interactions with GATOR2 complex is disrupted by VWCE knockdown.If anything, the effect seems to be subtle.
We appreciate the reviewer for posing this question.To strengthen our conclusion, we have quantified these interaction data and compiled the results from three independent experiments (Fig. 3b, c, and Extended Data Figs.4b-g).In the revised manuscript, we have also amended the language by substituting "disrupt" with "impair".
In Fig. 3c, the decrease of NPRL2 under siVWCE is convincing, but the decrease of NPRL3 is subtle.How about the other component DEPDC5?It is not shown in the figure.Also, in Fig. 3d-e, only NPRL2 decrease is shown.Is siVWCE affecting specifically NPRL2 or the whole GATOR1 complex?
In an effort to further understand how liver cancers might overcome the tumor-suppressive effect of high VWCE expression levels, we conducted the following analysis.First, we identified 17 transcription factors (TFs) co-expressing with VWCE in both prostate and liver cancers, based on the TCGA database (please see below, Fig. 1a).We then analyzed the expression levels of these 17 TFs in five selected cancer cell lines using the CCLE database (please see below, Fig. 1b).We selected seven TFs (HNF4A, GATA4, ARID3A, CEBPA, VTN, NR1H4, and HHEX), whose expression levels showed the highest correlation with VWCE, for promoter binding analysis.ChIP-seq data from the Encode project demonstrated strong binding of HNF4A, GATA4, and ARID3A to the VWCE promoter region (please see below, Fig. 1c).Knockdown of these three TFs resulted in downregulated VWCE expression, suggesting that HNF4A, GATA4, and ARID3A are potential VWCE TFs in liver cancer cells (please see below, Fig. 1d, e).However, knockdown of these three TFs inhibited the growth of multiple liver cancer cell lines, including HepG2 and Huh7, according to the DepMap database (please see below, Fig. 1f).The oncogenic phenotype of HNF4A, GATA4, and ARID3A might be driven by other oncogenes under the influence of these three TFs.In such a scenario, VWCE might merely be a passenger whose suppressive effect is overwhelmed in liver cancers.This may explain why the overexpression of VWCE did not inhibit the growth of liver cancers.
We believe the aforementioned analysis falls beyond the scope of the current manuscript.Therefore, we have only provided these results in the point-by-point response and have briefly discussed them in the revised manuscript.If the reviewer feels it is necessary to include these data within the main manuscript, we are more than willing to make those adjustments.Other minor points: Fig. 2g.Why is LAMP2 expression so low in the whole cell lysates?
We thank the reviewer for raising this question.The seemingly low expression of LAMP2 is due to the short exposure time during that specific immunoblotting experiment.We have repeated the LAMP2 immunoblotting using the same samples and extended the exposure time, as shown in Fig. 2g.Fig. 2h.Immunoblotting of SZT2 needs to be shown to confirm its efficient knockdown.
We appreciate the reviewer's suggestion.However, an anti-SZT2 antibody is not commercially accessible.The knockdown efficiency of SZT2 was validated by a decrease in the SZT2 mRNA level as determined by the qPCR experiment (Extended Data Fig. 2k) and constitutive activation of mTORC1 during amino acid starvation (Extended Data Fig. 2l).Fig. 4g.Suppression of mTORC1 activity by VWCE overexpression in the DU145 cell line is not obvious.Similarly, in Fig. 4i, mTORC1 lysosomal localization was not altered significantly in DU145 cells.
We thank the reviewer for raising this question.The DU145 cell line exhibits lower sensitivity to amino acid availability.In the revised manuscript, we have prolonged the amino acid starvation period to 2 hours and followed it with a 15-minute amino acid restimulation, as mentioned in the legend of Extended Data Fig. 6.We found that overexpression of VWCE also inhibited the activation and lysosomal localization of mTORC1 in DU145 cells (Extended Data Figs. 6a,b,h,l).The immunoblotting assays were quantified and compiled with data from three independent replicates.
The authors have added new data to support their conclusion.The authors have thoroughly addressed this reviewer's comments.
Reviewer #2 (Remarks to the Author): This is a resubmission of paper that I previously reviewed.While the authors have made several improvements based on comments of the three reviewers, some aspects of the paper I remain dissafisfied with.For example, a significant concern I voiced in my original review was about the plausibility that VWCE, a protein will all the hallmarks of a secreted protein, is localized to the cytoplasmic face of lysosomes.Early in the manuscript, when the authors first introduce VWCE and propose to invesfigate its role in mTORC1 funcfion, they need to pre-empt our scepficism by describing the molecular features of VWCE and explaining why they nonetheless believe a role for this protein in mTORC1 biology is plausible.I appreciate the effort that the authors have made in addressing my concern and performing protease treatment of lyso-IP'd lysosomes as I suggested.However, given that they went to the considerable effort to FLAG-tag the endogenous VWCE gene, I really wish they had performed this assay with their 'endogenous' VWCE-expressing cells rather than using cells with 'exogenous' stably-expressed VWCE.I think it remains a disfinct possibility that FLAG-tagged VWCE expressed by an exogenous protomer may not behave exactly like the endogenous protein.
On that note, an important concern raised by another reviewer was the relafive importance of endogenous VWCE more generally to the proposed model.I think this remains a valid concern and the treatment of this in the manuscript could be improved.For example, I am confused by Extended Data Figure 2a as it is presented.If this is a 'knock-in' experiment, where the endogenous VWCE gene has been FLAG-tagged, then this needs to be more explicitly explained in the manuscript text or Figure legend, etc.Furthermore, how do the authors know that the red arrow is poinfing to VWCE-FLAG?What are the extremely prominent bands below the alleged VWCE-FLAG band?It would have been nice to see an anfi-FLAG Western blot of whole cell lysates (not immunoprecipitated samples) for both 'WT' and 'knocked-in' cells (with molecular weight markers included) to convince readers that this red arrow is in fact poinfing at VWCE.Furthermore, demonstrafing that this band is also sensifive to VWCE siRNA would improve our confidence.

Fig. 2h .
Fig. 2h.Immunoblofting of SZT2 needs to be shown to confirm its efficient knockdown.

Fig. 4g .
Fig. 4g.Suppression of mTORC1 acfivity by VWCE overexpression in the DU145 cell line is not obvious.Similarly, in Fig. 4i, mTORC1 lysosomal localizafion was not altered significantly in DU145 cells.

Figure 1
Figure 1.a) Co-expression of transcriptional factors with VWCE in liver and prostate cancers.An analysis of genes co-expressed with VWCE in PRAD and LIHC was carried out using the TCGA database.A total of 17 transcription factors (TFs) with a correlation greater than 0.4 are indicated in red.b) The expression heatmap of VWCE and the TFs in five selected prostate or liver cancer cell lines, based on the CCLE dataset.c) HNF4A, GATA4, and ARID3A bind to the VWCE promoter region.The binding of these TFs to the VWCE promoter region was analyzed using ChIP-seq data from the Encode project.d) Knockdown of HNF4A, GATA4, or ARID3A results in the downregulation of VWCE mRNA and protein levels in liver cancer cells.HepG2 and Huh7 cells stably expressing the indicated shRNAs were subjected to RT-qPCR (N=3) or immunoblotting assays.e) The knockdown efficiency of HNF4A, GATA4, or ARID3A was measured by RT-qPCR (N=3).f) According to the DepMap database, knockdown of HNF4A, GATA4, or ARID3A suppresses the growth of several liver cancer cell lines.Data are mean ± s.d.*P<0.05;**P<0.01;***P<0.001;****P<0.0001(unpaired Student's t-test).