Nur77 suppresses hepatocellular carcinoma via switching glucose metabolism toward gluconeogenesis through attenuating phosphoenolpyruvate carboxykinase sumoylation

Gluconeogenesis, an essential metabolic process for hepatocytes, is downregulated in hepatocellular carcinoma (HCC). Here we show that the nuclear receptor Nur77 is a tumour suppressor for HCC that regulates gluconeogenesis. Low Nur77 expression in clinical HCC samples correlates with poor prognosis, and a Nur77 deficiency in mice promotes HCC development. Nur77 interacts with phosphoenolpyruvate carboxykinase (PEPCK1), the rate-limiting enzyme in gluconeogenesis, to increase gluconeogenesis and suppress glycolysis, resulting in ATP depletion and cell growth arrest. However, PEPCK1 becomes labile after sumoylation and is degraded via ubiquitination, which is augmented by the p300 acetylation of ubiquitin-conjugating enzyme 9 (Ubc9). Although Nur77 attenuates sumoylation and stabilizes PEPCK1 via impairing p300 activity and preventing the Ubc9-PEPCK1 interaction, Nur77 is silenced in HCC samples due to Snail-mediated DNA methylation of the Nur77 promoter. Our study reveals a unique mechanism to suppress HCC by switching from glycolysis to gluconeogenesis through Nur77 antagonism of PEPCK1 degradation.

of Nur77 leads to only a small increase in PEPCK. It might be that the large overexpression leads to increased nuclear Nur77 and hence transcriptional activation of PEPCK. This needs to be addressed. Perhaps also use the zinc finger binding mutant in the presence of PEPCK shRNA (similar to figure 2G). In addition, what happens to the protein and RNA expression of the other Nur77 targetes The data on curcumin is concerning. It is unclear why it was included as it really does not add to the story and actually is quite contradictory. Numerous studies have shown that curcumin has antidiabetic, antigluconeogenic activity. This is odds with the data shown.
HepG2 are considered to be non tumorigenic. However, the authors show xenograft data with these cells using 2X106 cells. More details need to be provided as to how they were able to get these cells to form xenografts.
Huh7 in general do not express PEPCK protein (although there is abundant RNA). Therefore it is unclear how the authors were able to show PEPCK in this cell line and especially the changes observed. Indeed, in several figures, PEPCK expression is absent in these cells, eg. Supplemental figure 2F vs 2G. While the overexpression might explain some of this data, this is problematic.
According to the data, there is no PEPCK in HCC tumor tissue. Based on work done several decades ago, while perhaps decreased, there is PEPCK in liver. This is observed in the Oncomine and TCGA as well.
The authors do no discuss the use of NEM. If it is to show the sumoylation of PEPCK, why is it not used in other figures. In addition, how biologically relevant is such a small amount of sumoylation in light of the large amount of unsumoylated PEPCK. Indeed, the decreases shown in figure 3D are difficult to see, although this could be a result of such small figures (See note below regarding figures). In addition, the lower panel of figure D cannot be compared to the upper figure since Ubc and Sumo were not coexpressed. Rather anacardic acid should have been included with Sumo and Ubc9.
The experiments showing PEPCK inhibits cell proliferation should also be performed as a xenograft experiment.
Almost all the experiments showing sumoylation of PEPCK are following IP, with the exception of supplementary figure 5A and 5B, in which PEPCK was overexpressed (in a cell with PEPCK). Therefore, can the sumoylation be observed in cells without ectopic expression of PEPCK. Also can sumoylated PEPCK be observed in the absence of pulling down one of the transfected proteins. What about blotting for endogenous PEPCK vs the flag epitope with and without IP. Indeed in figure 3C, the authors A) show no change in Flag PEPCK in lysate, but also cut the blot so it cannot be determined whether there is sumoylated PEPCK in these lysates.
A more clear idea of control of PEPCK expression by Nur77 over time without CHX would be more informative (similar to figure 5A, but without CHX, not just time zero, indeed, although the blots cannot be compared since they don't look they are on the same blot, it does not look like there is a difference or if anything, PEPCK levels seem lower in cells with Nur77 at time 0.
Other issues: Reviewing the manuscript was very difficult since the size of the figures was so small. In addition, the manuscript is need of significant copy editing.
A number of statements that are made are overly stated Eg. "anacardic acid markedly stimulated..." there was a small change-not markedly.
It is unclear how the authors can make the statement on line 215. Both ALLN and MG132 will inhibit proteosomal degradation. Therefore, this raises the questions as to how ALLN did not have a similar effect as MG132.
Reviewer #2 (Remarks to the Author): In the here present study authors evaluate importance of the nuclear receptor Nur77 and PEPCK1 enzyme as potential tumor suppressors in hepatocellular carcinoma (HCC). By utilizing different approaches and models (in vitro, in vivo, human samples) the authors aim to explore the effects of Nur77 downregulation on HCC development. Further, they delineate a new regulatory role for Nur77 for PEPCK1 sumoylation and subsequent ubiquitination.
Key findings: -Majority of the HCC patients show downregulation of Nur77 in the tumor tissue compared to the normal liver. Patient expressing low amount of Nur77 have close association with poor clinical prognosis.
-Increased level of Nur77 induces tumor suppressive effect in HCC by promoting gluconeogenesis, preventing glucose uptake and depleting intracellular ATP.
-Nur77 prevents sumoylation and degradation of PEPCK1, a rate-limiting enzyme of gluconeogenesis. Stability of PEPCK1 is dependent on sumoylation of Lys124.
-Snail transcription factor is suppressing Nur77 gene expression by binding to E-boxes on Nur77 promoter.
Overall, the study focuses on an important topic, namely the relevance of key regulators of gluconeogenesis for hepatocarcinogenesis. The study is technically well performed. However, some of the mechanistic findings need to be extended and utilized cell lines should be consistently applied across the experiments. A potential p53-dependent role should further be explored (Zhang et al. PNAS 2014). Major comments -Given the impact of PEPCK1 on T-cell function (Ho et al. Cell 2015) and the importance of the chronic inflammatory liver disease for HCC development, pre-neoplastic stages of hepatocarcinogenesis should be explored in the context of the here suggested study. Further, clinic-pathological information for the investigated specimens should be added. Further, the data for the Nur77 expression in the investigated HCC cohort should be shown.
-Throughout the manuscript the utilized of cell lines are inconsistently applied (e.g. Figure 1 and 2). This should be extended, in particular for the investigation where only hepatoma cell line (and 293T cells) was used (e.g. Figure 4). Further, it would be interesting to include untransformed hepatocytes as a negative control. Furthermore, given the dominant role of p53 (also suggested in the introduction) on gluconeogenesis and the high incidence of p53 mutations in HCC, it might be interesting to explore this issue. This could be achieved, e.g. by adding the p53-null cell line Hep3B. Of note, SK-Hep1 is a cell line of endothelial origin (Heffelfinger et al. 1992) and should not be used synonymously in the context with HCC cell line. The cell line should be replaced.
-The authors suggest a direct regulatory effect of Nur77 on PEPCK1 (Supplemental Figure 2). While this is certainly interesting, mechanistic investigations confirming the suggested direct regulatory effect are missing. This should be extended to confirm the Nur77-PEPCK1 regulatory role for HCC. Further, despite the potential regulation of PEPCK1 it remains unclear how Nur77 affects the tumorigenic potential of HCC cells, e.g. by regulating WNT signaling (Chen et al. Gut 2012). In line with this, impact of Nur77 on sumoylation and the suggested mechanism of operation are not striking and most pronounced in non-hepatoma 293T cells ( Figure 5). Quantification and statistical evaluation is needed. Similar, statistical analyses of (e.g. Supplemental Figure 5A) seem necessary, since different base line levels of PEPCK1 expression are shown.
-Given the plethora of different curcumin targets, the mechanisms of Nur77 upregulation by curcumin are unclear and should be evaluated. Since curcumin is known to inhibit cell growth of hepatoma cells by affecting multiple molecular targets, specificity for the suggested findings should be confirmed by a more targeted approach.
-Promotor methylation of Nur77 should be confirmed in the cell lines ( Figure 6). Rational for Snail selection should be explained in more detail.
Minor comments -The study by Wurmbach et al. includes 4 stages of HCC. Please explain the differences in the here presented study. Please also explain how the statistical analyses shown in the Figure have been derived.
-The effect of different Nur77 levels on proliferative status in individual tumors (e.g. by Ki67 and PCNA) should be confirmed.
-The authors suggest that Nur77 is driven by non-genomic events. Please extend the explanation of this hypothesis. Maybe data supporting that Nur77 is not impaired by genetic alterations (e.g. by using publically available repositories) can be included. -It is unclear how the doses for the used compounds were established in the respective cell lines. Experimental data or suitable references should be provided. -Number of replicates and statistical analyses should be added wherever necessary.
Reviewer #3 (Remarks to the Author): Bian et al's manuscript demonstrates how Nur77 as a tumor suppressor suppresses hepatocellular carcinoma. They show that Nur77 is a tumor suppressor of HCC and affects gluconeogenesis via interacting with PEPCK1. They also show that PEPCK1 is a SUMOylated protein and SUMOylation affects PEPCK1 stability. More interestingly, they find that p300 enhances PEPCK1 sumoylation through acetylating UBC9. Increasing Nur77 expression by compound curcumin could suppress PEPCK1 sumoylation and HCC. This manuscript is quite complicated but of interesting, especially they find that p300 acetylation of UBC9 enhances PEPCK1 sumoylation, which will be a novel regulation mechanism for sumoylation.
Major: 1. Authors showed that sumoylation down-regulated PEPCK1 stability, which would decrease gluconeogenesis and increase tumorgenesis in HCC. If it is true, it should be detected that PEPCK1 is highly-sumoylated in HCC samples compared to in normal liver tissue, which would strongly support the link between PEPCK1 SUMOylation and tumor suppressor function of Nur 77. Additionally, how sumoylation affect PEPCK1 stability? 2. It has been shown that sumoylation of PEPCK1 decreases tumorigenesis in HCC. Therefore, sumoylation mutant of PEPCK1 would rescue the tumorigenesis in Nur77-silenced HCC. They should do that to conclusively make statement that Nur 77 suppress HCC via PEPCK1 sumoylation.
3. It is of interesting that p300 acetylation UBC9 increases PEPCK1 sumoylation. Does p300-UBC9 regulation affect global sumoylation? It seems that acetylated UBC9 has more affinity to bind to PEPCK1 than un-acetylated form. They should prove it by binding experiment. 4. The authors mention that hepatic expression of Nur77 can elevate gluconeogenesis through transcriptional upregulation of gluconeogenic genes, including G6pc, Fbp1 and Fbp2, and Eno317. They also show that mutation of Nur77 binding to DNA affect the growth of SMMC-7721 cells (Supplemental Figure 2B), suggesting that the transcription activity of Nur77 also engage the tumorigenesis. How to comment the contribution of Nur77-PEPCK1 non-transcription activity and Nur77 transcription activity to HCC?
Minor: 1. In Fig 3a, Is there any changes in PEPCK1 sumoylation in liver with or without DEN/CCL4 treatment? 2. In Fig 4, Does knock-down p300 affect PEPCK1 expression as well as sumoylation in HCC cells? 3. In Fig 4a (bottom), overexpression of p300 decreased endogenous PECPK1 expression in 293T and HepG2, did affect the sumoylation of endogenous PEPCK1? 4. The author mentioned in the discussion that the third mechanism of Nur77 regulating PEPCK1 is attenuating p300-induced PEPCK1 acetylation, is there any evidence for it? 2788-96), the following results indicated that insulin or IGF signaling was not involved in Nur77-mediated HCC suppression. In the DEN/CCl4 model, Nur77-KO mice developed HCC at a higher rate than WT mice; however, insulin and IGF1 levels were equal between these groups of mice ( Supplementary Fig. 1h). Similarly, in the HFD/STZ model, in which mice were intraperitoneally injected with the β cell toxin streptozotocin (STZ) at 2 days of age, the insulin and IGF1 levels in Nur77-KO mice were also comparable with those in WT mice ( Supplementary Fig. 1h), unlike the reported HFD model (i.e., no injection with STZ.

The authors need to show Nur77 and PEPCK expression in the HCC tumors at RNA and protein level. It would also be useful to show Ubc9 and p300 levels.
Nur77 and PEPCK mRNA levels and Ubc9 and p300 mRNA and protein levels in clinical HCC samples are now provided in Figure 1a, Figure 2b and Supplemental Figure 4i-j, respectively.

HepG2 have similar levels and LO2.
To further verify the lower expression of Nur77 in HCC cell lines, we used three additional liver cancer cell lines and displayed their mRNA and protein expression levels in Supplementary Fig. 1i. Consistent with our previous results, Nur77 expression was substantially lower in HCC cells than in the untransformed hepatocyte L02 cells. Although HepG2 and BEL-7402 expressed relatively higher Nur77 levels compared with other HCC cells, Nur77 mRNA and protein levels in these two cells were still lower than in L02 cells.
According to the suggestion from Reviewer #2, we removed the Sk-hep-1 cell line. The Sk-hep-1 cell line is derived from the ascitic fluid of a patient with adenocarcinoma of the liver; it is a cell line of endothelial origin (In Vitro Cell Dev Biol. 199;28A(2):136-42).
Therefore, it is not suitable to be used in conjunction with HCC cell lines.

Previous work shows that PEPCK promotes the TCA cycle, this would lead to increased NADH and FADH2, and hence ETC and oxygen consumption (Burgess et al 2004). Indeed, overexpression of PEPCK in muscle increases the oxidative capacity of the muscle (Hanson and Hakimi 2008). The authors need to reconcile these discrepancies .
Burgess and colleagues reported that liver-specific knockout of PEPCK1 in mice induces the accumulation of TCA cycle intermediates, which results in inhibition of the TCA cycle and oxygen consumption (J Biol Chem. 2004;279 (47) (3):297-308). Therefore, PEPCK1 knockdown in cancer cells may not accumulate TCA cycle intermediates, which is also reported by another paper (Mol Cell. 2015; 60(4):571-83). In our study, PEPCK1 knockdown in HCC cells leads to the inhibition of glucose production and the promotion of glucose utilization, which finally results in enhanced oxygen consumption in the TCA cycle ( Supplementary Fig. 2l). This interpretation has been added to the Discussion section.

The authors claim in methods that they examined glycolytic capacity. However, the technique they used appears to be the mitochondrial stress test, which is not the protocol for glycolytic capacity. Rather, the glycolytic stress is usually used. At what point are the authors measuring basal ECAR and OCR. Is this subtracting out non-mitochondrial respiration, or simply the initial ECAR and OCR prior to oligomycin injection. Perhaps the authors should show the Seahorse tracing.
We apologize for not clearly explaining the OCR and ECAR measurement procedure. In fact, we have performed the glycolysis stress test for ECAR measurement and the mitochondrial stress test for OCR measurement. The ECAR bar values represent the glycolytic capacity, which is attained by subtracting the non-glycolytic acidification from the maximum values. The OCR bar values represent the ATP production-related oxygen consumption, subtracting out non-mitochondrial respiration and proton leak from the basal values. The seahorse tracing curves are now shown in Supplemental Figures 2a, 2m, and 2o.
The procedure for OCR and ECAR measurement has also been rewritten in the Methods section.

PEPCK is elevated in the liver of obese mice. This would seem to be at odds with the data presented. Therefore, PEPCK should lead to protection against HCC?
It has been reported that PEPCK1 is elevated in the liver of obese mice (J Biol Chem. Therefore, enhanced PEPCK should protect against HCC.

Almost all the data showing sumoylated PEPCK is following immunoprecipitation of endogenous PEPCK and blotting for PEPCK or tagged PEPCK and blotting for tag. What about endogenous PEPCK under conditions of increased sumoylation. Most of the endogenous PEPCK blotting shows changes in PEPCK by Nur77 or curcumin. However, differences are difficult to discern perhaps because figures are difficult to read (see below).
To answer the question of "What about endogenous PEPCK under conditions of increased sumoylation," we added an experiment using HepG2 cells, in which the endogenous sumoylation of PEPCK1 could also be detected in the presence of NEM, an inhibitor of de-sumoylation ( Figure 3c). NEM was thus used to detect PEPCK1 sumoylation hereafter. We further verified that in different HCC cell lines, increased endogenous sumoylation of PEPCK1 by transfection of p300 was associated with decreased endogenous PEPCK1 protein level ( Figure 4b). In clinical HCC samples, increased sumoylation accompanied decreased PEPCK1. Conversely, there is no sumoylation, but high PEPCK1 levels, in para-carcinoma samples ( Figure 3b). Thus, higher sumoylation indeed corresponds to lower endogenous PEPCK1 levels in vivo.
In our study, PEPCK1 sumoylation promoted its ubiquitin-proteasomal degradation (Figure 3h), suggesting that endogenous sumoylated PEPCK1 is more labile and difficult to detect without immunoprecipitation.
Unlike other modifications, such as ubiquitination, sumoylation is described as "low-level, big-effect." Few proteins are quantitatively sumoylated, either constitutively or upon receiving their respective upstream signals (J. Proteome Res. 2015Res. , 14:2385. Instead, most targets appear to be modified to a small percentage at steady state. However, the fact that a small pool of sumoylated protein causes the dramatic effects that have been assigned to sumoylation has been well recognized (EMBO J. 2002(EMBO J. , 21:1456Mol. Cell. 2003, 11:1043-1054Mol. Cell. 2006, 24:341-354). As Ruth Geiss-Friedlander and Frauke Melchior mentioned in their review (Nature Reviews Molecular Cell Biology. 2007, 8:947-956), sumoylated targets can undergo rapid cycles of modification and demodification.
Although this equilibrium might lie on the side of the unmodified form, the whole pool of a given protein might be affected by sumoylation in a short window of time (Cell Metabolism. 2014, 20: 603-613). For example, steady-state sumoylation is usually less than 5%, and most transcription factors and co-regulators become significantly activated when the SUMO acceptor Lys residue has been mutated to an Arg residue (Mol. Cell. 2014, 13: 611-617).
Therefore, in most reports, the sumoylation level (particularly the endogenous level) is lower, but its outcomes are extremely diverse and important, including the localization of proteins and the activities of enzymes and target proteins.

Furthermore the large induction of Nur77 leads to only a small increase in PEPCK. It might be that the large overexpression leads to increased nuclear Nur77 and hence transcriptional activation of PEPCK. This needs to be addressed. Perhaps also use the zinc finger binding mutant in the presence of PEPCK shRNA (similar to figure 2G). In addition, what happens to the protein and RNA expression of the other Nur77 targetes.
Thank you for your questions and suggestion.  Figure 2p) efficiently inhibited clone formation in the control group, but not in PEPCK1-KD Huh7 cells. Together, these results further demonstrate that Nur77 has an effect on cell proliferation through PEPCK1 mediation that is independent of its transcriptional activity.

The data on curcumin is concerning. It is unclear why it was included as it really does not add to the story and actually is quite contradictory. Numerous studies have shown that curcumin has antidiabetic, antigluconeogenic activity. This is odds with the data shown.
Curcumin has various activities, including anticancer, antidiabetic, and antigluconeogenic activity. However, its mechanism of action is very complex. Based on the questions from you and Reviewer #2, the Editor has suggested deleting all of the curcumin data from the manuscript. We agree with this suggestion.

HepG2 are considered to be non tumorigenic. However, the authors show xenograft data with these cells using 2X106 cells. More details need to be provided as to how they were able to get these cells to form xenografts.
In the ATCC catalog, the HepG2 cell line is reported to be non-tumorigenic; however, many papers showed that HepG2

Therefore it is unclear how the authors were able to show PEPCK in this cell line and especially the changes observed. Indeed, in several figures, PEPCK expression is absent in these cells, eg. Supplemental figure 2F vs 2G. While the overexpression might explain some of this data, this is problematic.
Although the PEPCK1 protein level in Huh7 cells is relatively low, we could detect its expression after a longer exposure time by Western blotting using an anti-PEPCK1 antibody In Supplemental Figure 2j (original 2f), the PEPCK1 level in control group of Huh7 cells appears to be similar to that of SMMC-7721 cells. This finding is also due to the different exposure times. To avoid misunderstanding, we provided a new image, in which PEPCK1 expression in Huh7 and SMMC-7721 cells was shown on the same film with the same exposure time.

According to the data, there is no PEPCK in HCC tumor tissue. Based on work done several decades ago, while perhaps decreased, there is PEPCK in liver. This is observed in the Oncomine and TCGA as well.
PEPCK1 was expressed in clinical HCC tumor tissues but at substantially lower levels than in para-carcinoma tissues. Instead of the original image, we chose to present the same image with a relatively longer exposure time (Figure 2b).

authors do no discuss the use of NEM. If it is to show the sumoylation of PEPCK, why is it not used in other figures. In addition, how biologically relevant is such a small amount of sumoylation in light of the large amount of unsumoylated PEPCK. Indeed, the decreases shown in figure 3D are difficult to see, although this could be a result of such small figures (See note below regarding figures). In addition, the lower panel of figure D cannot be compared to the upper figure since Ubc and Sumo were not coexpressed. Rather anacardic acid should have been included with Sumo and Ubc9.
NEM is an inhibitor of de-sumoylation. Because sumoylated targets can undergo rapid cycles of modification and demodification, NEM was used in all sumoylation assays With regard to the question "such a small amount of sumoylation in light of the large amount of unsumoylated PEPCK," please see our previous explanation. Such phenotypes of lower sumoylation have also been observed in other reports (Cancer Cell. 2014, 25:748-761;Cell. 2007, 131:584-595). In particularly, the sumoylated PEPCK1 was more labile and was degraded by the ubiquitination pathway. Thus, sumoylated PEPCK1 is difficult to detect.
According to the suggestion, we performed a new experiment in which Sumo and Ubc9 were first transfected into different cell lines as indicated, and then, cells were treated with anacardic acid. As expected, anacardic acid treatment obviously increased endogenous PEPCK1 expression levels (Figure 3d, bottom). This result, together with the data in Figure   3d (middle), suggested that inhibition of both endogenous and Sumo/UBC9 overexpression-induced PEPCK1 sumoylation efficiently influenced PEPCK1 protein levels.

The experiments showing PEPCK inhibits cell proliferation should also be performed as a xenograft experiment.
Thank you for your suggestion. We have performed this xenograft experiment. As shown in Figure

Flag PEPCK in lysate, but also cut the blot so it cannot be determined whether there is sumoylated PEPCK in these lysates.
The sumoylation of endogenous PEPCK1 can be observed in cells ( The expression level of exogenous Flag-PEPCK1 is substantially stronger than that of endogenous PEPCK1, as revealed using an anti-PEPCK1 antibody (shown below).
Overexpression of UBC9 and SUMO1 enhances Flag-PEPCK1 sumoylation, which is easy to detect without IP, as shown in Supplementary Figure 3b  According to your suggestion, we now include a whole image in Figure 3c (now presented in Supplementary Figure 3a), in which the sumoylated PEPCK1 could also be observed in cell lysates.

A more clear idea of control of PEPCK expression by Nur77 over time without CHX
would be more informative (similar to figure 5A, but without CHX, not just time zero,

indeed, although the blots cannot be compared since they don't look they are on the same blot, it does not look like there is a difference or if anything, PEPCK levels seem lower in cells with Nur77 at time 0.
We agree with your viewpoint that "control of PEPCK expression by Nur77 over time without CHX would be more informative." To this end, we performed an experiment in which SUMO/Ubc9 and Nur77 were transfected together into cells, and we analyzed PEPCK levels at different times without CHX treatment. The result indicated that endogenous PEPCK1 levels could be elevated by Nur77 at different Nur77 transfection times (Figure 5c). Thus, PEPCK1 expression can be regulated by Nur77.
As you mentioned, the results in Supplementary Figure 5c  Other issues:

Reviewing the manuscript was very difficult since the size of the figures was so small. In addition, the manuscript is need of significant copy editing.
The Figures have been enlarged in the revised manuscript. The manuscript has also been edited by Nature Research Editing Service, and the relative certification has been included as below.

Eg. "anacardic acid markedly stimulated..." there was a small change-not markedly.
We have revised the overstated descriptions. Thank you for your helpful comment.

MG132 will inhibit proteosomal degradation. Therefore, this raises the questions as to how ALLN did not have a similar effect as MG132.
We apologize for mistakenly labeling "ALLM" as "ALLN." The compounds we used in Figure 3h were MG132 and ALLM. ALLM is a calpain inhibitor, while MG132 is a proteasome inhibitor.

Further, the data for the Nur77 expression in the investigated HCC cohort should be shown.
Thank you for your suggestion. To display the pre-neoplastic stages of hepatocarcinogenesis, we detected hepatic cirrhosis in DEN/CCl4-induced HCC mouse samples by PicroSirius Red staining, which showed that more hepatic fibrosis was detected in Nur77-KO mice than in WT mice (Supplemental Figure 1c). Consistently, Nur77-KO mice also developed more severe inflammatory status in this DEN/CCl4 model, as indicated by higher circulating IL-6 levels (Supplemental Figure 1d). Because inflammation and cirrhosis are considered major risk factors for hepatocarcinogenesis, our results suggested that Nur77 not only inhibits HCC development but also retards HCC initiation.
The clinic-pathological information and Nur77 immunoreactive scores in the investigated specimens are shown in Supplemental Table 2.

Figure 1 and 2). This should be extended, in particular for the investigation where only hepatoma cell line (and 293T cells) was used (e.g. Figure 4). Further, it would be interesting to include untransformed hepatocytes as a negative control. Furthermore, given the dominant role of p53 (also suggested in the introduction) on gluconeogenesis and the high incidence of p53 mutations in HCC, it might be interesting to explore this issue. This could be achieved, e.g. by adding the p53-null cell line Hep3B. Of note, SK-Hep1 is a cell line of endothelial origin (Heffelfinger et al. 1992) and should not be used synonymously in the context with HCC cell line. The cell line should be replaced.
In this manuscript, we mainly used HepG2 (higher Nur77 expression), SMMC-7721 and Huh7 (lower Nur77 expression) to analyze the biological functions of Nur77 and its effects on PEPCK1. 293T cells were mainly used for transfections due to their higher transfection efficiency. To make the utilized cell lines consistent, we added some critical experiments in hepatoma cell lines in Figure 1f, 3d, 4b, 5a and Supplementary Figure 2b, 2d, 3d, 4a, 4e in the revised version.
According to the suggestion, we also tested the functions of Nur77 in untransformed hepatocyte L02 cells. However, we unexpectedly found that in L02 cells, Nur77 was mainly in the nucleus (Supplemental Figure 2d). L02 cells displayed a resistance to Nur77-mediated repression of cell proliferation (Supplemental Figure 1k), although Nur77 showed similar regulatory effects on glucose metabolism in L02 cells (Supplemental Figure 2c) as in liver cancer cells (Figure 2a, bottom). In addition, Nur77 overexpression regulated the expression of its target gene E2F1 in L02 cells but not in liver cancer cell lines (Supplementary Figure   2e). These results suggest that the Nur77 regulatory mechanism in L02 cells is different from that in liver cancer cells.
A p53-null cell line Hep3B was used to analyze the p53 role in the current study. The results indicated that Nur77 overexpression still inhibited clone formation and regulated gluconeogenesis (Supplemental Figure 1m & 2b), similar to the effects observed in the HepG2 cell line (WT p53 expression). Therefore, Nur77 exerted its functions independent of p53 in our study.
According to the suggestion, we deleted the SK-Hep1 cell line and added three additional liver cancer cell lines in the revised version of the manuscript (Supplemental Figure 1i).

Nur77-PEPCK1 regulatory role for HCC. Further, despite the potential regulation of PEPCK1 it remains unclear how Nur77 affects the tumorigenic potential of HCC cells, e.g. by regulating WNT signaling (Chen et al. Gut 2012). In line with this, impact of Nur77 on sumoylation and the suggested mechanism of operation not striking and most pronounced in non-hepatoma 293T cells (Figure 5). Quantification and statistical evaluation is needed.
Similar, statistical analyses of (e.g. Supplemental Figure 5A)

seem necessary, since different base line levels of PEPCK1 expression are shown.
To confirm the Nur77-PEPCK1 regulatory role for HCC, we performed colony-formation assays, which showed that Nur77 and Nur77 2G only inhibited colony formation in controls, but not PEPCK1-knockdown HCC cells (Figure 2i, Supplementary   Figure 2p), suggesting a role of PEPCK1 mediation. Furthermore, the xenograft tumor assay in nude mice showed that Nur77 substantially inhibited the PEPCK1-expressing xenograft tumor growth, but slightly influenced the PEPCK1 K124R-expressing tumor growth ( Figure   5f), further emphasizing that the Nur77 repression of HCC depends on the sumoylation of PEPCK1 at Lys124.
To determine whether Nur77 also inhibits Wnt signaling to retard HCC growth, the plasmid dnTCF4, which has been shown to inhibit Wnt signaling activity (Mol Cell Biol. 1999;19(8):5696-706), was constructed. Nur77 could efficiently inhibit cell proliferation in both wildtype HCC cells and Wnt signaling-inactivated HCC cells that were transfected with dnTCF4 ( Supplementary Figure 1l), suggesting that Wnt signaling is not required for Nur77 to inhibit HCC in our study.
In this version, we deleted some of the experiments performed in 293T cells and supplemented additional experiments using HCC cell lines, including Figures 3c, 3d, 4b, 5a and 5c.
Necessary quantification and statistical evaluations have been included in Figure 3E, 3G, and Supplementary Figure 3e, 3g, 4d and 5c, according to our three parallel experimental results.

4-Given the plethora of different curcumin targets, the mechanisms of Nur77 upregulation by curcumin are unclear and should be evaluated. Since curcumin is known to inhibit cell growth of hepatoma cells by affecting multiple molecular targets, specificity for the suggested findings should be confirmed by a more targeted approach.
Curcumin has multiple molecular targets to exert its functions, and its functions in cancer are complex. According to the comments from you and Reviewer #1, the Editor has suggested deleting the content of the curcumin study. We agree with this suggestion. Figure 6).

Rational for Snail selection should be explained in more detail.
We added additional experiments to detect Nur77 promoter methylation in different HCC cell lines (Figure 6b, bottom).
Snail is a suppressive transcription factor that binds to E-box DNA sequences to silence

6-The study by Wurmbach et al. includes 4 stages of HCC. Please explain the differences in the here presented study. Please also explain how the statistical analyses shown in the Figure have been derived.
The raw Wurmbach liver data from Oncomine includes 6 different stages of liver tissues: normal liver, cirrhosis, liver cell dysplasia, stage 1 HCC, stage 2 HCC and stage 3 HCC. To study Nur77 expression in normal liver tissue vs. HCC, we applied the data from normal liver, stage 1 HCC, stage 2 HCC and stage 3 HCC and presented them in Supplementary Fig. 1a.
Significance was determined with one-way ANOVA followed by Tukey's post hoc test.

Ki67 and PCNA) should be confirmed.
To confirm the effect of different Nur77 levels on proliferative status, we analyzed PCNA protein levels in clinical samples. As expected, Nur77 and PCNA protein showed a clear negative correlation (Supplemental Figure 1b).

8-The authors suggest that Nur77 is driven by non-genomic events. Please extend the explanation of this hypothesis. Maybe data supporting that Nur77 is not impaired by genetic alterations (e.g. by using publically available repositories) can be included.
Nur77 not only acts as a transcription factor to positively or negatively regulate downstream gene expressions but also performs its non-genomic actions independent of its regulation of downstream target genes. This functional mode of Nur77 has been demonstrated recently by several reviews (Steroids. 2015 Mar;95:1-6;Expert Opin Ther Targets. 2012;16(6):573-85). In this study, we showed that Nur77 could not influence the transcription level of PEPCK1 (Supplementary Figure 5b), and Nur77 2G, a Nur77 mutant without transcriptional activity, could still inhibit HCC growth ( Supplementary Figure 2f & 2p). These results suggested that transcriptional activity is not involved in Nur77 inhibition of HCC.
Instead, Nur77 mainly inhibited HCC via interaction with PEPCK1 and stability of PEPCK1 protein through inhibition of its sumoylation. To avoid misunderstanding, we have rewritten the sentences on Page 9. The concentration of different compounds are now indicated in corresponding figure legends.

10-Number of replicates and statistical analyses should be added wherever necessary.
We have added this information to the figures as necessary. Thank you for your suggestion.

Authors showed that sumoylation down-regulated PEPCK1 stability, which would decrease gluconeogenesis and increase tumorgenesis in HCC. If it is true, it should be detected that PEPCK1 is highly-sumoylated in HCC samples compared to in normal liver tissue, which would strongly support the link between PEPCK1 SUMOylation and tumor suppressor function of Nur77. Additionally, how sumoylation affect PEPCK1 stability?
According to the suggestion, we added two in vivo sumoylation detection assays in clinical and mouse samples. In clinical para-carcinoma samples, PEPCK1 expression was higher than in carcinoma samples, while its sumoylation could be hardly detected in para-carcinoma samples (Figure 3b). In mouse samples, sumoylation was detected in DEN/CCl4-induced HCC samples, but not in normal liver tissues ( Figure 3a). As Nur77 expression was lower in carcinoma samples but higher in para-carcinoma samples (Figure 1a), and as Nur77 and PEPCK1 showed a positive correlation (Figure 2c), these results strongly support the link between PEPCK1 sumoylation and the tumor suppressor function of Nur77.
Moreover, the fact that Nur77 efficiently inhibited xenograft tumor growth when wildtype PEPCK1 was co-expressed, but only moderately inhibited tumor growth when PEPCK1 K124R was co-expressed (Figure 5f), further confirmed the suppressive function of Nur77 on HCC relied on its regulation of PEPCK1 sumoylation.
Our results indicated that sumoylation affects PEPCK1 stability through the ubiquitin-proteasome pathway, as MG132 but not ALLM treatment led to PEPCK1 accumulation in the presence of SUMO1/Ubc9 (Figure 3h, left), which was accompanied by increased endogenous ubiquitin targeting to PEPCK1 but not to PEPCK1 K124R (Figure 3h, middle). For comparison, increased ubiquitination was still detected on PEPCK1 K471&473R (Figure 3h, right), which is another indication that Lys124 is the only critical sumoylation site for PEPCK1 stability.

Therefore, sumoylation mutant of PEPCK1 would rescue the tumorigenesis in
Nur77-silenced HCC. They should do that to conclusively make statement that Nur 77 suppress HCC via PEPCK1 sumoylation.
To further verify Nur77 repression in HCC via PEPCK1 sumoylation, we performed xenograft experiments in nude mice. As shown in Figure 5f, Nur77 efficiently inhibited PEPCK1-expressed xenograft tumor growth, but only slightly influenced PEPCK1 K124R-expressed tumor growth, further suggesting that Nur77 suppressed HCC via regulation of HCC PEPCK1 sumoylation at Lys124.

p300-UBC9 regulation affect global sumoylation? It seems that acetylated UBC9 has more affinity to bind to PEPCK1 than un-acetylated form. They should prove it by binding experiment.
According to your suggestion, we detected global sumoylation in several HCC cells, and the results indicated that transfection of p300 could not affect the global sumoylation induced by Ubc9 (Supplemental Figure 4e). It appears that p300-mediated Ubc9 acetylation selectively influences PEPCK1 sumoylation.
The acetylation of Ubc9 at Lys65 by p300 facilitates its binding with PEPCK1, as revealed in Figure 4e, in which the transfection of p300 enhanced the PEPCK1-Ubc9 interaction, but not the PEPCK1-Ubc9 K65R interaction. However, PEPCK1 interacted with both Ubc9 and its mutant Ubc9 K65R with similar affinity (Supplemental Figure 4g) without the presence of p300, which further indicates that p300, through acetylating Ubc9 and facilitating acetylated Ubc9 binding to PEPCK1, could enhance PEPCK1 sumoylation.

The authors mention that hepatic expression of Nur77 can elevate gluconeogenesis through transcriptional upregulation of gluconeogenic genes, including G6pc, Fbp1 and
Fbp2, and Eno317. They also show that mutation of Nur77 binding to DNA affect the growth of SMMC-7721 cells (Supplemental Figure 2B), suggesting that the transcription activity of Nur77 also engage the tumorigenesis. How to comment the contribution of

Nur77-PEPCK1 non-transcription activity and Nur77 transcription activity to HCC?
It has been reported that Nur77 can transcriptionally activate the expression of several gluconeogenic genes in normal liver tissue (Nature medicine 2006, 12(9): 1048-1055).
However, Nur77 appears to lose most of its transcriptional activity in HCC cells, as revealed by the fact that Nur77 was mainly localized in cytoplasm (Supplementary Figure 2d). Nur77 only slightly enhanced the transcription of Fbp2 and Eno3, but could not influence G6pc and , it is likely that although the transcriptional activity of Nur77 may partially involve in HCC suppression, Nur77 mainly exerts its function to suppress HCC via protein-protein interaction, such as Nur77-PEPCK1. Minor:

In Fig 3a, Is there any changes in PEPCK1 sumoylation in liver with or without DEN/CCL4 treatment?
We have added this experiment. In mouse liver samples, sumoylation could be detected in DEN/CCl4-induced HCC, but not in normal liver tissues (Figure 3a).

In Fig 4, Does knock-down p300 affect PEPCK1 expression as well as sumoylation in HCC cells?
As shown in Supplementary Figure 4a, knockdown of p300 did affect PEPCK1 expression or sumoylation in transfection or endogenous experiments in different HCC cell lines. Fig 4a (bottom), overexpression of p300 decreased endogenous PECPK1 expression in 293T and HepG2, did affect the sumoylation of endogenous PEPCK1?

In
Overexpression of p300 increased endogenous PEPCK1 sumoylation, accompanied by decreased endogenous PEPCK1 expression, in different HCC cell lines (now presented in Figure 4b).