The essential role of YAP O-GlcNAcylation in high-glucose-stimulated liver tumorigenesis

O-GlcNAcylation has been implicated in the tumorigenesis of various tissue origins, but its function in liver tumorigenesis is not clear. Here, we demonstrate that O-GlcNAcylation can enhance the expression, stability and function of Yes-associated protein (YAP), the downstream transcriptional regulator of the Hippo pathway and a potent oncogenic factor in liver cancer. O-GlcNAcylation induces transformative phenotypes of liver cancer cells in a YAP-dependent manner. An O-GlcNAc site of YAP was identified at Thr241, and mutating this site decreased the O-GlcNAcylation, stability, and pro-tumorigenic capacities of YAP, while increasing YAP phosphorylation. Importantly, we found via in vitro cell-based and in vivo mouse model experiments that O-GlcNAcylation of YAP was required for high-glucose-induced liver tumorigenesis. Interestingly, a positive feedback between YAP and global cellular O-GlcNAcylation is also uncovered. We conclude that YAP O-GlcNAcylation is a potential therapeutic intervention point for treating liver cancer associated with high blood glucose levels and possibly diabetes.


Results:
The transcriptional activity of TEAD is the indicator of YAP activity This statement probably should read activity of TEAD is one of the indicators of YAP activity. For growth, yes, for total transcriptional output, probably not.   Figure 1D, the data suggest that p-YAP levels are not different between HCC and normal tissues (it is the level of YAP that changes along with global O-GlcNAcylation). This seems to correlate with 2B in which inhibition of OGA correlates with decreased p-YAP but increased YAP and expression of downstream CTGF. (1) PugNAc is not an "activator" of O-GlcNAcylation, it is an inhibitor of OGA.  : Suggest that KD of OGT has reduction in OGT activity and yields increase in apoptotic cells. But, these phenotypes were reversed when YAP was over expressed. The claims for this figure need additional experiments. (1) What does addition of YAP over expression alone look like for these samples? (2) Overexpression of another protein would also be important as a control here to confirm that these findings are specific to YAP over expression. (3) In A wherein YAP is over expressed, it does not appear that there is an increase in YAP between the Mock and o/e. Does it just return to baseline expression of YAP? Is there a change in p-YAP as well?
In summary, this is a nice piece of work that could have important implications if vital experiments can be performed to test some of the ideas contained therein.
Reviewer #2 (Remarks to the Author): Minor comments: 1. Lines 32-33: YAP, CREB, and TEAD need to be fully spelled as they appear in the manuscript for the first time. Furthermore, lines 269 and 280 need to be revised to correct the syntax.
Major comments: 1. Line 59: The authors provide intriguing data about a new mechanism of tumorigenesis linked to HCC associated with greater YAP and O-GLCNAcylation expression in tissue samples. Over 200 human HCC tissues using TMA analysis were screened. However, a part from the name of the institution where the tissues were collected, no further information about the patients was provided in the manuscript. Because one of the key findings positively links YAP expression to glucose levels in the cell lines Bel-7402 and SMMC-7721 ( Figure S4A-B), it would be important to know etiology and metabolic background, especially the glycaemic levels of a representative number of patients. This information would improve the clinical impact of the findings provided in vitro and in the mouse model.
2. Line 64: The authors choose a broad spectrum of HCC cell lines and compare them with one non-cancerous hepatocyte line. A more representative number of non-cancerous hepatocyte cells should be included.
3. Line 338: The authors conclude that YAP, via O-GlcNAcylation, is crucial in liver tumorigenesis, particularly in diabetes-associated liver tumorigenesis. This conclusion is based on the studies conducted in mice models, whose diabetic condition is achieved by streptozoicin treatment (stated in line 241-242). However, the authors do not provide data showing the levels of insulin in the mice after treatment, nor they show any histopathology concerning this part of the study. Therefore, this conclusion lacks supporting information.
Based on these comments, we recommend that this manuscript be revised extensively. In its current state, the manuscript is not suitable for publication in Nature Communications.
Reviewer #3 (Remarks to the Author): The manuscript reports a novel modification of YAP proto-oncogene protein, namely O-GlcNAcylation (OGNAc) on Threonine 241 that is located within the conserved region of the second WW domain of YAP 1-2 isoform. Importantly, YAP that is decorated with OGNAc on Thre-241 has a higher stability and pro-proliferative activity.
The study is original, well designed and clearly presented. The quality of the data is good and the burgeoning interest in the Hippo-YAP pathway and the not well understood role of YAP for propensity in gene amplification and causality in HCC should make this report of interest to a wide readership.
The Reviewer and two of his/her Colleagues who are experts in sugar metabolism suggest in in concert the following changes to improve the manuscript.
1. It is likely, even from the visual inspection of the 3D structure of YAP WW domain in complex with its PPxY-containing cognate ligand that Thre 241 when decorated with a large sugar should disrupt YAP interaction with its cognate ligands, in particular with LATS (but also with AMOT, PTPN14) and result in nuclear localization of YAP to drive transcription of genes that induce proliferation. To document the mechanism of the YAP-OGNAc-Thre-241, it would be important to show that either isolated second WW domain of YAP with OGNAc-Thre-241 in vitro or YAP-OGNAc-Thre-241 in cello show reduced propensity for interaction with LATS -PPxY peptide and better yet with LATS protein.
2. Please discuss better a correlation between YAP expression and global expression of OGNAc proteins in HCC, in particular, if the localization of YAP-Thre-OGNAc-241 in HCC as determined by IHC, for example, is changed as expected.
3. If we are correct in our interpretation, in 4.3. Since TAZ, a YAP paralogue does not have a second WW domain in vertebrates (except fish) and also YAP 1-1 isoform does not have a second WW domain, one could discuss why YAP is such a prevalent oncogene for liver cancer, compared to TAZ. The cloning of TAZ was by Mike Yaffe and his team at MIT and the report could be referenced in the discussion of TAZ. Kanai et al., (2000) EMBO, J., 19, page 6778. This is a solid paper that establishes a causal relationship between Yap O-GlcNAcylation, lack of Yap phosphorlyation, Yap stabilization, and tumorogenesis in HCC. 1. Better data should be obtained from the human HCCs with documentation of Yap O-GlcNAcylation and relationship to some objective criteria such as TEAD transcription, proliferation etc. 2. The mechanism for the inverse correlation between Yap O-GlcNAcylation and Yap phosphorlyation at two distant sites should be determined. #. The role of hyperglycemia in HCC is probably overstated. There is a small but significant increase in HCC in type I diabetes which is modeled by treatment with streptazoticin. The real increase is in type 2 diabetes with hyperinsulinemia, fatty liver, insulin resistance, and usually obesity.

Results:
The transcriptional activity of TEAD is the indicator of YAP activity. This statement probably should read activity of TEAD is one of the indicators of YAP activity. For growth, yes, for total transcriptional output, probably not.

Answer to Question #1.2
We agree with this suggestion from this reviewer, and have changed our statement more accurate and reasonable. The new statement is: "Transcriptional activity of TEAD is one of the most important indicators of YAP activity." (Page 5) TEAD is required for YAP-induced cell growth, oncogenic transformation, and epithelial-mesenchymal transition [Zhao, et al., 2008]. Disruption of the TEAD-YAP complex suppresses the oncogenic activity of YAP [Liu, et al., 2012]. Hence, we believe testing TEAD transcription activity is a good way to indirectly reflect the activity of YAP.

Answer to Question #2
We chose transcription factor TEAD as the control. The reasons are listed below: 1) TEAD is one of the most important YAP-dependent transcription factors, and its transcription activity relies on YAP; 2) The aim of the present study is to investigate the correlation between O-GlcNAcylation and YAP and its contribution to liver tumorigenesis.
Exclude the possibility that TEAD is also can be affected by O-GlcNAcylation is necessary to support the conclusion that the effects of O-GlcNAcylation on liver tumorigenesis are mainly via a YAP-dependent manner, but not via its YAP-dependent transcription factors.
The data in new Figure.

Answer to Question #3
Because to the best of our knowledge, no tissue microarray regarding non-cancerous and cancerous liver cell lines is commercially available, we

Answer to Question #4
Firstly, we mentioned in the manuscript that PuGNAc is an inhibitor of OGA (Page 5).
In new Figures. 2a-b, the experiment using GlcNAc alone has been performed.
We found treatment of GlcNAc alone can stimulate luciferase activity from pUAS-LUC/TEAD-Gal4 system (new Figure.  In the left panel of new Figure 2d, the "-" symbols have been aligned to a line.
The symbol indicating statistical significance of Figure. 2e has been corrected.

Question #5
Figure 4: Suggest that KD of OGT has reduction in OGT activity and yields increase in apoptotic cells. But, these phenotypes were reversed when YAP was over expressed. The claims for this figure need additional experiments. (1) What does addition of YAP over expression alone look like for these samples?
(2) Overexpression of another protein would also be important as a control here to confirm that these findings are specific to YAP over expression. (3) In A wherein YAP is over expressed, it does not appear that there is an increase in YAP between the Mock and o/e. Does it just return to baseline expression of YAP? Is there a change in p-YAP as well?

Answer to Question #5
In new Figure. 4a-d, addition of YAP over expression alone has been supplemented. YAP overexpression could increase cell proliferation and colony formation capacity, but inhibit apoptosis. However, its contributions to these phenotypes were not significant in liver cancer cells according to our statistics. These might due to YAP has already over-expressed or over-activated in liver cancer cells Wang, et al., 2015], the effects caused by addition of extra-YAP might not be obvious enough.
c-Fos is also an onco-protein that has been found to be overexpressed in liver cancer [Yuen, et al., 2001]. In new Supplementary Figure. Figure. S3b-d). However, the impaired transformative phenotypes could not be reversed by simultaneous overexpression of c-Fos ( Figure. S3b-d).
Thereby, we here confirm that the data from new Figure. 4 are specific to YAP over expression.
As shown in new Figure. 4a, overexpressing YAP alone could lead to a significant increase of YAP expression. However, due to the facts that knockdown of OGT reduces protein stability of YAP, simultaneous overexpression of YAP in Bel-7402 and SMMC-7721 cells with OGT knocked down was unable to reach to the levels in cells overexpressed with YAP alone.
In order to reverse the YAP level to the baseline from "Mock", we adjusted YAP levels by overexpressing YAP to the levels similar to the "Mock" as much as possible.
The p-YAP was also tested, and the data are shown in new Figure.

Question #1
Lines 32-33: YAP, CREB, and TEAD need to be fully spelled as they appear in the manuscript for the first time. Furthermore, lines 269 and 280 need to be revised to correct the syntax.
Answer to Question #1.1 YAP, CREB, and TEAD have been fully spelled as "Yes-associated protein" (Page 1), cAMP-response element binding protein" and "TEA domain transcription factor", respectively when they firstly appear in the manuscript (Page 2).
The syntax in lines 269 and 280 has been corrected as follows: Line 269: Next, we evaluated the expression of genes involved in the HBP.
Line 280: We found that YAP-induced mRNA expression and promoter activities of OGT, Nudt9 and SLC5A3 genes could be enhanced by simultaneous overexpression of either TEAD or CREB (Fig. 7c), further confirming that the HBP is transcriptionally regulated in a YAP-dependent manner.

Answer to Question #2
The liver cancer tissue microarray slides were purchased from U.S. Biomax with diabetes compared to those without diabetes. By contrast, the p-YAP levels were much lower in liver cancer patients with diabetes compared to those without. These data suggest glycaemic levels are positively associated with the levels of YAP while negative associated with the p-YAP.

Question #3
Line 64: The authors choose a broad spectrum of HCC cell lines and compare them with one non-cancerous hepatocyte line. A more representative number of non-cancerous hepatocyte cells should be included.

Answer to Question #3
In addition to the hepatocyte line, HL-7702, another non-cancerous hepatocyte line, THLE-3 was added to compare the differences between hepatocyte and liver cancer cells. The data are shown in the new Figure.

Question #4
Line 338: The authors conclude that YAP, via O-GlcNAcylation, is crucial in liver tumorigenesis, particularly in diabetes-associated liver tumorigenesis.
This conclusion is based on the studies conducted in mice models, whose diabetic condition is achieved by streptozoicin treatment (stated in line 241-242). However, the authors do not provide data showing the levels of insulin in the mice after treatment, nor they show any histopathology concerning this part of the study. Therefore, this conclusion lacks supporting information.

Answer to Question #4
The levels of serum-insulin along with serum-glucose from mice treated with either saline or streptozocin were examined, and the data are shown in the new Supplementary Figure Figure. S5n). However, no significant changes of phenotype between the treatments of saline and streptozocin were seen in the colon, suggesting the effects caused by streptozocin might be liver-specific.

Question #1
It is likely, even from the visual inspection of the 3D structure of YAP WW domain in complex with its PPxY-containing cognate ligand that Thre 241 when decorated with a large sugar should disrupt YAP interaction with its cognate ligands, in particular with LATS (but also with AMOT, PTPN14) and result in nuclear localization of YAP to drive transcription of genes that induce proliferation. To document the mechanism of the YAP-OGNAc-Thre-241, it would be important to show that either isolated second WW domain of YAP with OGNAc-Thre-241 in vitro or YAP-OGNAc-Thre-241 in cello show reduced propensity for interaction with LATS -PPxY peptide and better yet with LATS protein.

Answer to Question #1
Thank you for your constructive suggestions. We have tried several times to synthesize the peptide of second WW domain of YAP, and this peptide contains potential O-GlcNAc site, Thr241. Unfortunately, it's very difficult to decorate this peptide at Thr241. Also, we are unable to synthesize the YAP protein with modification of O-GlcNAcylation at Thr241 in vitro, although we have tried a lot of times. For the above reasons, the only way to address this problem asked by this reviewer is to stimulate O-GlcNAcylation of YAP and see whether the YAP-LATS interaction is affected before and after mutation of Thr241 in liver cancer cells.
In new Figure 5d, stimulation of O-GlcNAcylation by combined treatment of PuGNAc and GlcNAc inhibited LATS1 binding to WT-YAP, however; this inhibitory effect was not significant when Thr241 was mutated. We also noticed that mutation of Thr241 in the YAP protein significantly increased YAP-LATS1 binding at basal level.
In new Figure 6d, the similar results were observed before and after treatments of high glucose. Figure Collectively, we propose that O-GlcNAcylation of YAP at Thr241 within the WW domain is accompanied by simultaneous reduction of the binding between YAP and LATS.

Question #2
Please discuss better a correlation between YAP expression and global expression of OGNAc proteins in HCC, in particular, if the localization of YAP-Thre-OGNAc-241 in HCC as determined by IHC, for example, is changed as expected.

Answer to Question #2
In DISCUSSION section, a better discussion on the correlation between YAP and global O-GlcNAcylation has been merged into the current version of manuscript, and is shown as follows in Page 20: YAP stimulates tumorigenesis in liver cancer [Netsirisawan, et al. 2015;Sodi, et al., 2015]. O-GlcNAcylation is an important posttranslational modification of proteins, and plays pro-oncogenic roles in several types of cancer, including liver cancer [Ferrer, et al., 2014;Guo, et al., 2013;Huang, et al., 2013].
O-GlcNAcylation of onco-proteins Jun proto-oncogene (c-Jun) and Tribbles pseudokinase 2 (TRIB2) has been recently reported to stimulate liver tumorigenesis [Qiao et al., 2016;Yao et al., 2016]. Interestingly, both c-Jun and TRIB2 are nuclear proteins. In the current study, similar to c-Jun and TRIB2, we also found that both O-GlcNAcylated proteins and O-T241-YAP are primarily expressed in the nucleus. Therefore, we speculate that O-GlcNAcylation is more likely to occur in the nucleus, and this modification may easily enhance the expression and function of nuclear proteins, such as YAP. However, whether and how O-GlcNAcylation tends to occur in the nucleus is still unclear and needs to be further investigated.
In addition, we have developed an antibody that can specifically recognize O-GlcNAcylation of YAP at Thr241 (anti-O-T241-YAP). We used this antibody to evaluate O-GlcNAcylation of YAP in the same liver cancer microarray that has been used in new Figure.

Question #3
If we are correct in our interpretation, in Fig. 6C, the data may indicate that there is still another O-GluNAc site in YAP? Please elaborate further on this point.

Answer to Question #3
Firstly, we agree with your opinion. We have discussed this point in the DISCUSSION section (Page 23). In new Figure.  The following references could be added if space allows. YAP WW domain was recently shown to be modified by Tyr phosphorylation in breast cancer models. This modification changes the ability of YAP WW domain to form complexs. Please consider discussing this report.

Answer to Question 4.1
The following context has been added into the DISSCUSSION section (in Page 22): Interestingly, Li et al. [Li, et al., 2016] reported that phosphorylation of tyrosine188 (Y188) in the YAP1-2 isoform stimulates YAP1-induced cellular transformation. Mutation of Y188 [especially replacement of Y to phenylalanine (F)] leads to a higher affinity of YAP for binding to its upstream negative regulators for cytoplasmic retention 51. Like Thr241, the Y188 site is also located in the conserved aromatic core of the second WW domain of

Question #4.3
Since TAZ, a YAP paralogue does not have a second WW domain in vertebrates (except fish) and also YAP 1-1 isoform does not have a second WW domain, one could discuss why YAP is such a prevalent oncogene for liver cancer, compared to TAZ. The cloning of TAZ was by Mike Yaffe and his team at MIT and the report could be referenced in the discussion of TAZ. Kanai et al., (2000) EMBO, J., 19, page 6778.

Answer to Question #4.3
The following context has been added into the DISSCUSSION section (in Page 22-23): The WW domain containing transcription regulator 1 (TAZ), a paralogue of YAP, has a structure similar to that of YAP. Unlike YAP, all the isoforms of TAZ in human cells do not have the second WW domain [Kanai, et al., 2000].
Comparatively speaking, YAP is a more prevalent oncoprotein in liver cancer.
However, the function of TAZ in liver cancer is limited. We speculate that O-GlcNAcylation on the second WW domain of YAP plays an important role in promoting liver tumorigenesis, and this finding also supports the notion that YAP is more important than TAZ, because TAZ has only one WW domain that might not be O-GlcNAcylated. A serial of studies [Komuro, et al., 2003;Oka, et al., 2012] have also demonstrated that both YAP1-1 and YAP1-2 are present in liver tissue, and YAP1-2 has stronger transactivation activity compared to that of YAP1-1 [Komuro, et al., 2003]. Therefore, it is also not difficult to conclude that O-GlcNAcylation of the YAP1-2 proportion at its second WW domain might enhance the YAP pro-tumorigenic function contributed by both

Question #1
Better data should be obtained from the human HCCs with documentation of Yap O-GlcNAcylation and relationship to some objective criteria such as TEAD transcription, proliferation etc.

Answer to Question #1
Because TEAD transcription is usually tested by a pUAS-LUC/TEAD-Gal4 system Tang, et al., 2015], which is cell-based, thereby it's difficult to test TEAD transcription activity directly in tissues. CTGF is a well-established TEAD controlled gene [Zhao, et al., 2008], and its expression can indirectly reflect the transcription activity of TEAD. Due to the above reasons, we evaluated CTGF expression by IHC in the same liver cancer samples whose levels of YAP and O-GlcNAc have already been tested (new Figure. 1a). The Ki67, a well-known proliferation marker, is also tested in the same samples (new Figure.  The mechanism for the inverse correlation between Yap O-GlcNAcylation and Yap phosphorlyation at two distant sites should be determined.

Answer to Question #2
As shown in the new Figure 5d, stimulation of O-GlcNAcylation by treatment of PuGNAc and GlcNAc inhibited LATS1, the key kinase that can phosphorylate YAP at Ser127, binds with WT-YAP. However, this inhibitory effect was much reduced when the Thr241, the potential O-GlcNAc site, was mutated in both Bel-7402 and SMMC-7721 cells. We also noticed that mutation of Thr241 significantly increased YAP-LATS1 binding at basal level (new Figure. 5d).
Moreover, we excluded that stimulation of O-GlcNAcylation influences phosphorylation of LATS1 (new Supplementary Fig. S4o), suggesting reduction of YAP-LATS1 binding might not due to alteration of LATS modification, instead, might due to the changes of YAP, especially modification at Thr241.
Interestingly, the Thr241 site is located within one WW domain in the YAP protein, and it's well known that the WW domain is essential for LATS1 binding with YAP [Hao, et al., 2008;Oka, et al., 2008]. These data suggest that O-GlcNAcylation of YAP at the Thr241 site within WW domain might prevent LATS1 interact with YAP for further phosphorylation.

Question #3
The role of hyperglycemia in HCC is probably overstated. There is a small but significant increase in HCC in type I diabetes which is modeled by treatment with streptazoticin. The real increase is in type 2 diabetes with hyperinsulinemia, fatty liver, insulin resistance, and usually obesity.

Answer to Question #3
Thank you for your suggestion. We have revised the manuscript according to your suggestion. But we still believe that hyperglycemia is one of the most important risk factors, which may cause high occurrence of liver cancer. In type 2 diabetes, hyperglycemia is still one of its major characteristics. Type 2 diabetes is associated with increased risk of liver cancer as numerous studies have reported [Chen et al., 2015;Qiao et al., 2014;Yang et al., 2013]. In one meta-analysis [Chen et al., 2015], authors have found that hepatitis C virus (HCV)-infected or cirrhotic patients with concomitant presence of type 2 diabetes are more likely to develop liver cancer than those without diabetes.
Type 2 diabetes is also reported associated with poor prognosis of liver cancer [Qiao et al., 2014]. Thereby, as a common characteristic of both type 1 and type 2 diabetes, hyperglycemia might increase risk of the occurrence of liver cancer.