SPRED2: A Novel Regulator of Epithelial-Mesenchymal Transition and Stemness in Hepatocellular Carcinoma Cells

Round 1

Reviewer 1 Report

The authors want to clarify the role in hepatocellular carcinoma of Spred2, a negative regulator of ERK1/2 previously studied in other human cancers. They state in detail the role of this protein and how its downregulation in HCC provides an activation of ERK1/2 pathway, following EMT and stemness promotion. The paper is reasonably comprehensive, and this reviewer has appreciated the further deepening in 3D models, which more closely resemble the in vivo cellular environment, compared to 2D models.

In general, the results are clearly presented and the conclusion are properly driven. Higher resolution images in WB and IHC should be included. Moreover, the results should be better described in more detail.

- In the paper the authors always talk about ERK, rather than ERK1/2. This should be replaced consistently

-In materials and methods, I suggest a technical insight in 4.13. The IHC procedure is lightly presented.

- I suggest putting the molecular weight in all the images of western blot.

- In figure 1F they talk about an accelerated gap closure in SPRED2 Knock Out cells but I’m a little concerned about that. Do they use an inhibitor for proliferation while performing the experiments? They study the gap closure at 24 and 48 hours: if they don’t use an inhibitor of proliferation, they can’t exclude that the effect is a consequence of increased cell proliferation, rather than motility. These experiments should be rerun in the presence of a mitotic inhibitor. Moreover, the increase of pERK1/2 in figure 1B seems to support an increase cell proliferation.

- In figure 1G they perform a 24h Matrigel-invasion assay. Do they use an inhibitor of proliferation? About that I have the same comment as figure 1F.

- In line 171 they say PD98059 treatment decreases the expression of their three stemness pluripotency factors: Nanog, c-Myc and KLF4, which are represented in figure 4B. I think KLF4 WB image isn’t representative of this result.

- In the legend to Figure 5, line 228, 'ells' should be 'cells'.

Author Response

(Point-by-point responses)

First, we would like to thank reviewers for reviewing our manuscript and providing constructive comments. Those comments were very helpful to improve our manuscript. Our point-by-point responses are presented below.

Comments to Reviewer 1:

The authors want to clarify the role in hepatocellular carcinoma of Spred2, a negative regulator of ERK1/2 previously studied in other human cancers. They state in detail the role of this protein and how its downregulation in HCC provides an activation of ERK1/2 pathway, following EMT and stemness promotion. The paper is reasonably comprehensive, and this reviewer has appreciated the further deepening in 3D models, which more closely resemble the in vivo cellular environment, compared to 2D models. In general, the results are clearly presented, and the conclusion are properly driven.  

1. Higher resolution images in WB and IHC should be included. Moreover, the results should be better described in more detail.

[Our response] We have replaced the original images with those in a higher resolution. We now describe the results in more detail.

2. In materials and methods, I suggest a technical insight in 4.13. The IHC procedure is lightly presented.

[Our response] In response to the comment, we describe the immunostaining methods in more detail in Materials and Methods (line 697-706).

3. I suggest putting the molecular weight in all the images of western blot.

[Our response] In response to the comment, we added the molecular weight of each protein in the figures. The molecular weight of each band can also be found on all images of western blotting in the supplementary data.

4. In figure 1F they talk about an accelerated gap closure in SPRED2 Knock Out cells but I’m a little concerned about that. Do they use an inhibitor for proliferation while performing the experiments? They study the gap closure at 24 and 48 hours: if they don’t use an inhibitor of proliferation, they can’t exclude that the effect is a consequence of increased cell proliferation, rather than motility. These experiments should be rerun in the presence of a mitotic inhibitor. Moreover, the increase of pERK1/2 in figure 1B seems to support an increase cell proliferation.

[Our response] The reviewer’s comment is reasonable. In response to the comment, we newly performed a scratch assay with 18-hour incubation time. As you can see in the figure here, at this time point, there was  no difference in cell proliferation between SPRED2KO and WT cells by CCK-8 assay. We did not use a mitotic inhibitor because it also restricts cell movement. We have replaced the data in Figure 1F and modified the related sentences in the text (line 649-650).

5. In figure 1G they perform a 24h Matrigel-invasion assay. Do they use an inhibitor of proliferation? About that I have the same comment as figure 1F.

[Our response] We agree with the reviewer’s comment. For the same reason for cell proliferation study, we newly performed a matrigel-invasion assay with 18-hour incubation time and obtained similar results to those with 24-hour incubation. We have replaced the data (Figure 1G) and modified the related sentences in the text (line 656-658).

6. In line 171 they say PD98059 treatment decreases the expression of their three stemness pluripotency factors: Nanog, c-Myc and KLF4, which are represented in figure 4B. I think KLF4 WB image isn’t representative of this result.

[Our response] Thank you for pointing this out. We replaced the KLF4 image with a new image that shows a clear decrease in KLF4 level in Figure 4B.

7. In the legend to Figure 5, line 228, 'ells' should be 'cells'.

[Our response] Thank you for pointing out the misspelling. We have corrected the mistake.

Author Response File: Author Response.pdf

Reviewer 2 Report

This article describes the role of endogenous SPRED2 in inhibiting the induction of EMT and cancer cell stemness. I have only a few points to raise. First, in the abstract the authors do not report any of the results about EMT. They only state, at the end of the abstract, that downregulation of SPRED2 promotes EMT. They should add a few words on this aspect, perhaps at the expense of cancer cell stemness. Second, when the authors address the enhanced cancer cell stemness following downregulation of SPRED2 they do so only on the basis of in vitro results. There is no formal proof that these cells have, actually, increased tumor-initiating capacity in vivo. Could the authors at least discuss this point or show results of some in vivo experiments? Third, the authors describe that downregulated SPRED2 leads to increased proliferation of the cells, while SPRED2 OE diminishes proliferation of the cells. This is an important aspect. In fact, in most, but not call cases, EMT has been related to diminished proliferation of tumor cells. The authors should discuss this point. Fourth, and perhaps most importantly, the authors do not report any result about their cells undergoing a partial or complete EMT in response to SPRED2 downregulation. As the authors know this is now considered a crucial aspect in  EMT biology and they should show some results on it.

Author Response

(Point-by-point responses)

First, we would like to thank reviewers for reviewing our manuscript and providing constructive comments. Those comments were very helpful to improve our manuscript. Our point-by-point responses are presented below.

Comments to Reviewer 2:

This article describes the role of endogenous SPRED2 in inhibiting the induction of EMT and cancer cell stemness. I have only a few points to raise. This is a well-written, well-presented manuscript and the data (with minor exceptions) supports their conclusions. 

1. First, in the abstract the authors do not report any of the results about EMT. They only state, at the end of the abstract, that downregulation of SPRED2 promotes EMT. They should add a few words on this aspect, perhaps at the expense of cancer cell stemness.

[Our response] We now clearly state the results about the induction of EMT by SPRED2  deletion as follows: “SPRED2-knockout (KO)-HepG2 cells displayed an elongated spindle shape with increased cell migration/invasion, and cadherin-switching, features of epithelial-mesenchymal transition (EMT).” (line 15-17)

2. Second, when the authors address the enhanced cancer cell stemness following downregulation of SPRED2 they do so only on the basis of in vitro results. There is no formal proof that these cells have, actually, increased tumor-initiating capacity in vivo. Could the authors at least discuss this point or show results of some in vivo experiments?

[Our response] We appreciate the comment. In our in vivo transplantation model, tumor developed more frequently after implantation of SPRED2-KO cells than WT cells (Figure S3), suggesting an increased tumor-initiating capacity by SPRED2 deletion in vivo.

 3. Third, the authors describe that downregulated SPRED2 leads to increased proliferation of the cells, while SPRED2 OE diminishes proliferation of the cells. This is an important aspect. In fact, in most, but not all cases, EMT has been related to diminished proliferation of tumor cells. The authors should discuss this point.

[Our response] Thank you very much for the valid comment. As pointed out by the reviewer, our results showed that the loss of SPRED2 promoted both cell proliferation and EMT, while overexpression of SPRED2 hampered cell proliferation and EMT. As you can see in Figure 2B, STAT3 was activated by SPRED2 deficiency and repressed by SPRED2 overexpression. This may explain our results, since STAT3 promotes cell proliferation in addition to EMT induction. We added this explanation in the discussion section (line 484-490) with 2 new references (Ref # 44; Neito MA. Science 342;1234850, 2013 and Ref# 45; Sun X, et al. Cancer lett 262:201-213, 2008).

 4. Fourth, and perhaps most importantly, the authors do not report any result about their cells undergoing a partial or complete EMT in response to SPRED2 downregulation. As the authors know this is now considered a crucial aspect in EMT biology and they should show some results on it.

[Our response] Thank you for pointing out this very important point. In response to the comment, we investigated the protein levels of E-cad, N-cad and Snail on day 0 and day 14 in 3D culture condition. As expected, cadherin switching and increased Snail expression was observed on day 14, as compared to on day 0. We presented the data in Figure 5C and described the findings in the text (line 286-289).

 

Author Response File: Author Response.pdf

Reviewer 3 Report

Gao et al. adressess the role of the protein SPRED2 in HCC cells. First they demonstrated that reduced levels of SPRED2 in 3 distinct HCC cell lines lead to increased ERK phosphorylation and elevated proliferation. This result was further confirmed by CRISPR mediated KO in HepG2 cells. The KO also leads to enhanced cell migration in a scratch assays. The authors conclude that SPRED2 is involved to suppress cell prolifation, EMT and tumorgenicity in HCC. Via KO and overexpression experiments, they show that SPRED2 regulates signalling pathways such as STAT3. Regarding EMT they show a change of expression of N-Cadherin and E-cadherin, supporting that SPRED2 regulates EMT pathways in HCC cells.

Next they showed that SPRED2 influences cancer cell stemness, by measuring spere formation and spherical colongy formation assays. Further the authors show that SPRED2 KO decreases the number of apoptotic cells upon cisplatin treatment, indicating enhanced resistance, which is typical for cancer stell cells. This may be linked to the proteins MDR1 nd MRP1, which show increased expression in the KO cells. Furthermore, it is demonstrated that pluripotency markers such as Nanog, c-myc and KLF4 influenced by the SPRED2 levels – they become stronger expressed in the KO and weaker expressed in the OE cells. Together these results suggest that presence of SPRED2 in HCC cells down regulates cancer cell stemness.

Further the manuscript demonstrated the culturing HepG2 cells in 3D culture conditions leads to decreased expression of SPRED2. Reversely, returning the cells to 2D conditions leads to increased expression of SPRED2 again. Similar effects were observed for ERK activation and pluripotency markers. Lastly the investigated patient material, as well as publicly available data from TCGA and the protein atlas. The authors claim that this analysis confirms that high SPRED2 expression is better for patient survival and leads to a reduced stemness.

Overall, this is a worthwhile study. The main weakness of this manuscript is that is largely descriptive without any explanation by which molecular mechanisms SPRED2 is influenced in cellular processes. Also, most experiments where performed only with HepG2 cells and only with one KO clone. Thus, a further validation in additional cell lines and clones would be important to further substantiate the findings. The study may provide the ground for further exploring the role of SPRED2 in HCC.

However, there are several technical issues, why I currently cannot recommend a publication of this manuscript in its current form.

 

Major points:

1.      I am confused about Figure 6A. As far as I know TCGA has only collected tumor samples, and no adjacent samples. If at all they have non-tumor samples. Where are the 50 adjacent samples from? When looking in Gepia (http://gepia.cancer-pku.cn/detail.php?gene=SPRED2), there appears no big difference of SPRED2 expression in tumor versus non tumor HCC samples (here as LIHC).

2.      I in figure 3B the authors compared 12 tumor samples with only 3 normal samples from the proteinatlas. With this low numbers of samples, I don’t think it make much sense to make a comparison. Also, the staining strength refers to “hepatocytes” for normal samples, while for tumor samples it refers to “tumor cells” I am not convinced one can compare these samples. I would recommend removing this graph.

3.      I am further confused about Figure 6C. When using the mentioned km-plotter tool

a.      I found that there are only 370 samples for HCC available (and not > 4000)

b.      I found no strong correlation of SPRED2 expression with patient’s survival.

c.      The same is true when using the GePIA tool or the survival curves in the proteinatlas.

Thus, the authors should recheck whether 6A and 6C graphs are correct and revise the manuscript accordingly. For me it appears, that the TCGA data suggest that high expression of SPRED2 is linked to worse survival, which is the opposite to what the manuscript says.

4.      I do not understand why the authors used taqman probes for Yeast for MDR1 and MRP1. Even worse the mentioned Yeast Taqman probe Sc04113969_s1 refers to the Yeast “mitochondrial 37S ribosomal protein MRP1”, which has absolutely nothing to do with the human Multidrug resistance-associated protein 1 MRP1 (also named ABCC1). The authors should have used Hs00184500_m1 (for MDR1, ABCB1) and Hs01561483_m1 (for MRP1, ABCC1). This major mistake precludes a publication of the current manuscript. I would recommend redoing the experiment with the correct probes and resubmit the manuscript.

 

Minor points:

1.      In Figure 6G, it is not clear what is plotted against what. It appears that the X-axis is the SPRED2 expression, and the Y-axis is Nanog, c-Myc, KLF4. This should be properly labelled.

2.      From the text, it is not fully clear why 3D culturing is a suitable procedure the investigate stemness. Perhaps this can be a bit further elaborated in the results part.

3. The discussion may profit from speculations about the molecular mechansims how SPRED2 influences the cells. 

 

 

Author Response

(Point-by-point responses)

First, we would like to thank reviewers for reviewing our manuscript and providing constructive comments. Those comments were very helpful to improve our manuscript. Our point-by-point responses are presented below.

Comments to Reviewer 3:

Overall, this is a worthwhile study. The main weakness of this manuscript is that is largely descriptive without any explanation by which molecular mechanisms SPRED2 is influenced in cellular processes. Also, most experiments were performed only with HepG2 cells and only with one KO clone. Thus, a further validation in additional cell lines and clones would be important to further substantiate the findings. The study may provide the ground for further exploring the role of SPRED2 in HCC.

[Our response] Thank you very much for the thoughtful comment. We showed in this study, by deleting and over-expressing SPRED2, that endogenous SPRED2 is a key molecule that controls EMT and stemness in HCC cells via downregulating the ERK1/2 pathway. Upregulation of the ERK1/2 pathway leads to the activation of genes involved in cell migration and proliferation. Increased STAT3 activation in SPRED2-KO cells appears to be another mechanism controlling EMT and stemness. We showed in Figure 6S in our original manuscript that IL-6 mRNA expression was upregulated in SPRED2-KO cells after cytokine simulation. Since epidermal growth factor (EGF) plays a role in the progression of liver cancer, we examined EGF mRNA expression and found that SPRED2-KO cells, compared to WT cells, expressed a higher level of EGF mRNA without stimulation, and its expression was much higher  when stimulated (Figure S6). EGF receptor is a receptor tyrosine kinase and the signaling pathway downstream of EGF receptor could also be further activated in the absence of SPRED2. EGF and IL-6 activates STAT3 in different cell types, including HepG2 cells (Ref #42; Wang Y et al, Proc Natl Acad Sci USA 110:16975-16980, 2013., Ref #43; Zhong Z, et al, Science 264:95-98, 1994). Thus, SPRED2 can regulate EMT directly through ERK1/2 activation and indirectly through STAT3 activation caused by increased cytokine production. We added the new data (EGF mRNA expression in Figure S6) and state above discussion in the text (line 478-482). Above new references (Ref# 43) were also cited.

Regarding the validation of results obtained with HepG2 cells, we used three HCC cells (HepG2, Hep3B and HLE)  to begin this study. We showed that SPRED2-knockdown with a SPRED2-specific siRNA (Figure S1) enhanced the phosphorylation of ERK1/2 phosphorylation (Figure 1B), and cell proliferation (Figure 1C) in all three cell lines. Cadherin switching, and augmented Snail expression were also detected in SPRED2-deleted HepG2, Hep3B and HLE cells (Figure S5).

Since the level of SPRED2 and the effects of SPRED2 knockdown were highest in HepG2 cells, we attempted to generate SPRED2-KO cells using HepG2 cells. We established 7 individual KO clones and found that SPRED2 was successfully deleted in two clones B5 and E1, and both clones exhibited increased ERK1/2 activation and proliferation. Clone E1 was subsequently chosen for the rest of this study. Therefore, we believe that our findings are not specific to just one clone. We now state this fact in the text (line 84-92) and present the data in Supplementary data (Figure S2B, C).

Major points:

 1.  I am confused about Figure 6A. As far as I know TCGA has only collected tumor samples, and no adjacent samples. If at all they have non-tumor samples. Where are the 50 adjacent samples from? When looking in Gepia (http://gepia.cancer-pku.cn/detail.php?gene=SPRED2), there appears no big difference of SPRED2 expression in tumor versus non tumor HCC samples (here as LIHC).

[Our response] We apologize for the confusion. We downloaded the original data set from TCGA database using “liver hepatocellular carcinoma” as the keyword. The data set included 371 HCC and 50 adjacent. Data from tumor tissues and adjacent non-tumor tissues (not normal tissue) are independent from each other. Therefore, we deleted this graph.

 2. In figure 3B the authors compared 12 tumor samples with only 3 normal samples from the protein atlas. With this low numbers of samples, I don’t think it make much sense to make a comparison. Also, the staining strength refers to “hepatocytes” for normal samples, while for tumor samples it refers to “tumor cells” I am not convinced one can compare these samples. I would recommend removing this graph.

[Our response] Thank you very much for the kind suggestion. We deleted this graph.

 3. I am further confused about Figure 6C. When using the mentioned km-plotter tool

   a. I found that there are only 370 samples for HCC available (and not >4000)

  b. I found no strong correlation of SPRED2 expression with patient’s survival.

c. The same is true when using the GePIA tool or the survival curves in the protein atlas.

Thus, the authors should recheck whether 6A and 6C graphs are correct and revise the manuscript accordingly. For me it appears, that the TCGA data suggest that high expression of SPRED2 is linked to worse survival, which is the opposite to what the manuscript says.

[Our response] Again, we apologize for the confusion. Among 371 HCC cases, 82 cases were selected that could be followed for 2 years, including those who died. The SPRED2 expression of these 82 cases was dichotomized, and a survival curve was drawn with the high 50% as SPRED2-high and the low 50% as SPRED2-low. We replaced the graph and revised the related sentences (line 328-339, 717-721).

 4.  I do not understand why the authors used Taqman probes for Yeast for MDR1 and MRP1. Even worse the mentioned Yeast Taqman probe Sc04113969_s1 refers to the Yeast “mitochondrial 37S ribosomal protein MRP1”, which has absolutely nothing to do with the human Multidrug resistance-associated protein 1 MRP1 (also named ABCC1). The authors should have used Hs00184500_m1 (for MDR1, ABCB1) and Hs01561483_m1 (for MRP1, ABCC1). This major mistake precludes a publication of the current manuscript. I would recommend redoing the experiment with the correct probes and resubmit the manuscript.

[Our response] Thank you for pointing out the mistake. We used Hs00184500_m1 for MDR1 (ABCB1) and Hs02514106_s1 for MRP1 (ABCC1), and it was a simple mistake. We corrected the Table 2 accordingly.

Minor points:

 1. In Figure 6G, it is not clear what is plotted against what. It appears that the X-axis is the SPRED2 expression, and the Y-axis is Nanog, c-Myc, KLF4. This should be properly labelled.

[Our response] We apologize for not labeling it correctly. Both X and Y axes are as you indicated. X-axis is now properly labelled in the graphs.

2. From the text, it is not fully clear why 3D culturing is a suitable procedure the investigate stemness. Perhaps this can be a bit further elaborated in the results part.

[Our response] Thank you for pointing this out. We have added the following sentence to the Results section. “3D tissue culture models closely resemble the natural environment for cells compared to 2D culture models, providing more physiologically useful information which may allow for a better understanding of cancer cell biology.” (line 277-279)

 3. The discussion may profit from speculations about the molecular mechanisms how SPRED2 influences the cells.

[Our response] Thank you very much for the important comment. SPRED2 is demonstrated to inhibit the Ras-ERK1/2 pathway by interacting with Raf. It is well known that the Ras-ERK1/2 can be activated by the activation of receptor tyrosine kinases, but Ras also plays a role in signaling pathways downstream of other receptors. Therefore, deletion of SPRED2 likely leads to the activation of various signaling pathways and previously unknown phenotypes. In response to the comment, we have added a following paragraph in the Discussion (line 514-564). A new reference was cited (Ref #50; Gimple RC and Wang X. Front Oncol 9:965,2019)

“We have been investigating the role of SPRED2 in inflammatory responses and cancer by analyzing human data and performing experiments using human cell lines and SPRED2 KO mice. However, the exact role of SPRED2 in the cell signaling is still unclear. It is known that SPRED2 inhibits the Ras-ERK1/2 signaling pathway by interacting with Raf. In general, Ras is thought to be a signaling molecule downstream of receptor tyrosine kinases (RTKs), but it also plays a role in other signaling pathways. Therefore, deletion of SPRED2 likely activated many signaling pathways in which Ras plays a role. It is necessary to continue studies to better understand the role of this molecule in cancer biology.”

 

 

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The authors have satisfactorely addressed the indicated criticisms.

Reviewer 2 Report

The authors have made significant efforts in order to address the points that I have raised. The manuscript looks now fine for me

Reviewer 3 Report

I thank the authors for adequately revising the manuscript.