Prevention of High Glucose-Mediated EMT by Inhibition of Hsp70 Chaperone

Round 1

Reviewer 1 Report

In this work, Nikotina et al explore the progression of carcinomas through the process of epithelial-mesenchymal transition (EMT). They focus on the role of the chaperone Hsp70 in EMT by monitoring cell migration, wound healing, and expression of EMT markers under high and no glucose, and in the presence of different Hsp70 inhibitors. The results indicate that either a knockdown of Hsp70 or inhibition of Hsp70 under conditions of high glucose disturbs EMT. This work is important, as chaperones are novel targets of disease therapies.

The paper is nicely explained and the data is clear. However, below are a few minor comments:

Authors should define CIM (line 99,183)

Can the authors provide more information of how to interpret the values for the xCELLigence data? What would be a minimal/maximal value and what would it mean?

The authors may want to reiterate the EMT markers mentioned in the introduction (line 216)

There seemed to be some inconsistencies in the layout of the data presented. For example, sometimes E-cadherin expression is shown by microscopy (Fig 2,3), measured as gene fold expression (fig 4), or shown as a western blot (Fig 5). Additionally, figure 5 measures vimentin, but this was not a measure in any other experiment. It would be easier to compare the effects of different inhibitors on EMT markers if the data was shown similarly.

I appreciate that in these studies, inhibition or knockdown of Hsp70 has a positive effect. However, Hsp70 is a hub in the cell and coordinates many different activities. Are there any growth defects or functionality issues that are observed when inhibiting Hsp70?

Author Response

First of all, let us thank all three reviewers for their overall positive assessment of our work and for their valuable comments that helped make our work better. Then we answer the reviewers' questions one by one, and we really hope that our answers will be received favorably.

Reviewer 1

Q1: Authors should define CIM (line 99,183)

R1: «СIM-plates” is the commercial name proposed by Hoffmann-La Roche. Since these plates are used to study invasion and migration, I dare to suggest that «CIM» means «Cell Invasion & Migration».

Q2: Can the authors provide more information of how to interpret the values for the xCELLigence data? What would be a minimal/maximal value and what would it mean?

R2: In this study, we used only СIM plates, with the help of which we investigated the migration of DLD1 cells affected by Hsp70 inhibitors and inducers under hyperglycemia conditions. The starting point in these experiments is «1», which is the minimum value. If the value of the Cell Index increases, this means that cells crawl through the pores of the membrane separating upper and lower chambers and the device records the change in impedance. Unfortunately, there is no way to force 100% of cells to migrate, so it is impossible to determine the maximum number that the device will show. We can only compare the cell index of cells in different conditions and to conclude that the higher Cell Index the more cells possess migration capacity.

Q3: The authors may want to reiterate the EMT markers mentioned in the introduction (line 216)

R3: The EMT markers Snail, Slug and Twist as well as E-cadherin and vimentin are reiterated in Introduction (line 47).

Q4: There seemed to be some inconsistencies in the layout of the data presented. For example, sometimes E-cadherin expression is shown by microscopy (Fig 2,3), measured as gene fold expression (fig 4), or shown as a western blot (Fig 5). Additionally, figure 5 measures vimentin, but this was not a measure in any other experiment. It would be easier to compare the effects of different inhibitors on EMT markers if the data was shown similarly.

 R4: This question is not easy to answer. We tried to present the most representative data, and since we had nice blots with E-cadherin, we preferred to show them and not to duplicate the data obtained using confocal microscopy. With vimentin, it turned out that after our student completed his bachelor's degree about the effect of PES on EMT (these results were included in the article), there were no working antibodies left, and new ones did not come to date. We are sure that these data support the main message of the m/s, so we would ask the reviewer not to insist on their removal.

Q5:I appreciate that in these studies, inhibition or knockdown of Hsp70 has a positive effect. However, Hsp70 is a hub in the cell and coordinates many different activities. Are there any growth defects or functionality issues that are observed when inhibiting Hsp70?

R5. Of course, we checked how the suppression or increase of HSP70 affects cell growth, and we got the result that the more Hsp70, the faster the cells grow, and the smaller, the slower. We did not insert this data into the main text, because there is no particular novelty in it. But we show these to you.

Reviewer 2 Report

Nicotine and co-authors presented a descriptive work showing that the inactivation of HSF1 or HSP70 prevents the EMT in colorectal cancer cell lines. The connection between heat shock proteins and EMT is very controversial. It was shown before that both depletion and overexpression/activation of HSF1/HSP70 may prevent or stimulate EMT. Therefore, this submission could be interesting to continue the ongoing discussion in the HSPs-EMT field.
The author used high glucose to induce EMT in cells and three methodological approaches to inactivate HSP70: 1) genetic depletion of HSP70 with shRNA; 2) chemical inactivation of one of the main transcriptional regulators of HSP70, HSF1; 3) chemical inhibition of mitochondrial HSP70-p53 interaction. The EMT was validated mainly by QPCR analysis of EMT markers SNAIL, TWIST, Slug, and E-cadherin immunofluorescence.
The HSF1 part is quite confusing. Fig3a shows the wrong quantification of HSF1 and HSP70 levels after CL-43 treatment without heat shock. Obviously, there is no difference between treated and untreated cells, especially, taking into consideration significant differences in loading control marker tubulin. CL-43 without any doubts inhibit activation of HSF1 and HSP70 induction during heat shock, but without heat shock, CL-43 had no effect on HSF1, and the following analysis of EMT after CL-43 treatment may be explained by non-specific actions of the chemical drug. It is strongly advised to remove Figure 3 and corresponding text from the manuscript. HSF1 may directly regulate EMT and the levels of EMT markers (e.g. Snail -Liu D. et al. 2016). The data with HSF1 inhibitor in the present form does not help to support the conclusion on HSP70 effects on EMT. 
The primary CRC cell line from the patient HCC-9 should be better characterized, data presented in the manuscript, or experiments with HCC-9 should be removed from the paper. Primary tumor cell cultures are frequently a mixture of epithelial tumor cells and mesenchymal stromal cells. The observed changes of EMT markers may be explained by changes in stroma/tumor cells ratio rather than EMT. The paradox of increased E-cadherin after Glu+CL-43 treatment in comparison with control non-treated cells in Fig4d indicates that CL-43 may affect stromal mesenchymal cells proliferation or cell death in cultivating cells from CRC patient biopsy.
The main drawback of the manuscript is the absence of analysis of the main regulators of EMT, ZEB1/2. The addition of this analysis will significantly improve the quality of the manuscript and may help to find the mechanism behind HSP70 regulation of EMT through HSP70 chaperoning of Zeb1.
Minor points.
1)    Authors mentioned that only HSPA1A expression increased in tumors. Usually, both HSP70 inducible genes HSPA1A and HSPA1B are regulated in the same manner including the oncogenic transformation. Please, indicate why the preference was given to HSPA1A. The experiments in the manuscript do not discriminate HSPA1A and HSPA1B expression.  The targeted by shRNA sequence TTGATGCTCTTGTTCAGGTCG presents in both genes.
2)    Fig1a definitely indicates that there is no full knockdown of HSP70 with shRNA. The cell clone should be renamed into DLD1 HSP70 shRNA or HSP70-deficient clone
3)    The abbreviation of DLD1scr should be deciphered in the Material and Methods section.
4)    The description of the chaperone assay is missing. Please, describe this assay in the Material and Methods section.

Author Response

First of all, let us thank all three reviewers for their overall positive assessment of our work and for their valuable comments that helped make our work better. Then we answer the reviewers' questions one by one, and we really hope that our answers will be received favorably.

Reviewer #2

Q1.The HSF1 part is quite confusing. Fig3a shows the wrong quantification of HSF1 and HSP70 levels after CL-43 treatment without heat shock. Obviously, there is no difference between treated and untreated cells, especially, taking into consideration significant differences in loading control marker tubulin. CL-43 without any doubts inhibit activation of HSF1 and HSP70 induction during heat shock, but without heat shock, CL-43 had no effect on HSF1, and the following analysis of EMT after CL-43 treatment may be explained by non-specific actions of the chemical drug. It is strongly advised to remove Figure 3 and corresponding text from the manuscript. HSF1 may directly regulate EMT and the levels of EMT markers (e.g. Snail -Liu D. et al. 2016). The data with HSF1 inhibitor in the present form does not help to support the conclusion on HSP70 effects on EMT. 

R1: We are grateful to the referee for this remark, however cannot fully agree with him/her. The number of HSF1 molecules in the cell is usually low, and to detect HSF1 on the blot, it is necessary to use tricky methods, such as the development of staining with the use of FEMTO chemiluminescence. The decrease in the Hsp70 level, according to the modern paradigm, may be due to dysfunction rather than the low content of the HSF1, although some data indicate that the amount of HSF1 can be reduced under the f CL-43 treatment, apparently due to proteolysis in the heat-stressed cells

We have recently demonstrated that CL-43 induces a dose-dependent reduction in the Hsp70 level in a variety of tumor cells, including DLD1 (Nikotina et al., 2018, Oncotarget). We know that, besides heat shock proteins, HSF1 regulates numerous genes, including ones involved in oncogenesis. However, since in the EMT model stimulated by high glucose content, the shRNA-mediated depletion of Hsp70 and the violation of the chaperone function of Hsp70 show similar results, we conclude that the effect of EMT suppression when inactivating HSF1 with CL-43 is still mediated by a decrease in the Hsp70 level. Especially for the reviewer, we ran the gel with cells induced by CL-43 once again, but decided not to insert it into the main text, since similar data had already been published by us earlier (Nikotina et al., 2018, Oncotarget). However we will be happy to show this blot to the reviewer. Here DLD1 cells were treated with CL-43 in concentrations indicated and 12 hrs later cells were collected and cell lysates were subjected to western blot.  We would also ask the reviewer not to insist on deleting fig. 3a from the m/s.

Q2: The primary CRC cell line from the patient HCC-9 should be better characterized, data presented in the manuscript, or experiments with HCC-9 should be removed from the paper. Primary tumor cell cultures are frequently a mixture of epithelial tumor cells and mesenchymal stromal cells. The observed changes of EMT markers may be explained by changes in stroma/tumor cells ratio rather than EMT. The paradox of increased E-cadherin after Glu+CL-43 treatment in comparison with control non-treated cells in Fig4d indicates that CL-43 may affect stromal mesenchymal cells proliferation or cell death in cultivating cells from CRC patient biopsy.

R2: Thank you, this is very important point. We used for our experiments the patient's cells that were at passages 5-7 after they were transferred to cell culture, as described in the Materials and Methods, i.e. say directly, it was not PDX. After the first two passages, even before freezing, all cells with fibroblast-like morphology disappeared, and after thawing, we have a monotypic culture. It is unlikely that cellular contaminants can alter the EMT pattern. Here how the cells look like at the 5th passage.

Q3: The main drawback of the manuscript is the absence of analysis of the main regulators of EMT, ZEB1/2. The addition of this analysis will significantly improve the quality of the manuscript and may help to find the mechanism behind HSP70 regulation of EMT through HSP70 chaperoning of Zeb1.

Thank you, and as you rightly pointed out in the beginning of the review, this is a descriptive article and we didn't get into the mechanisms here. It is very possible that chaperoning of ZEB1 with HSP70 takes place, and this may explain the effect of the chaperone on EMT. We have investigated changes in the ZEB1 expression, but we have never been able to obtain figures statistically different from the control after conditioning with glucose. ZEB1 turned out to be not a very suitable factor for this model. We gave up our attempts, and did not include the ZEB1 data in the manuscript. However, we will show you en example

Minor points.
Q1)    Authors mentioned that only HSPA1A expression increased in tumors. Usually, both HSP70 inducible genes HSPA1A and HSPA1B are regulated in the same manner including the oncogenic transformation. Please, indicate why the preference was given to HSPA1A. The experiments in the manuscript do not discriminate HSPA1A and HSPA1B expression.  The targeted by shRNA sequence TTGATGCTCTTGTTCAGGTCG presents in both genes.

R1: You are right in that both proteins are close homologs but according to what I know particularly HSPA1A constitutes the majority of total cellular (inducible) Hsp70; otherwise, in our results both proteins are measured.

Q2)    Fig1a definitely indicates that there is no full knockdown of HSP70 with shRNA. The cell clone should be renamed into DLD1 HSP70 shRNA or HSP70-deficient clone

R2: Yes, you are absolutely right, the use of shRNA does not completely suppress the expression of the protein of interest and should not decrease it to zero. Usually, other approaches are used to completely suppress gene expression, such as the CRISPR/Cas9 technique. In this case, it is called “knock-out”.

Q3)   The abbreviation of DLD1scr should be deciphered in the Material and Methods section.

R3:The DLD1cells were infected with lentivirus with gfp gene. We inserted this information in Material and Methods 2.1.

Q4)  The description of the chaperone assay is missing. Please, describe this assay in the Material and Methods section

R4:The description of the chaperone assay is presented in material and Methods, 2.7.

Author Response File: Author Response.pdf

Reviewer 3 Report

Manuscript: ijms-1240001

In this manuscript, the authors explore the Hsp70-dependent mechanisms in high-glucose-induced epithelial to mesenchymal transition (EMT) of colorectal cancer cell lines. The paper has merit, in general, the experiments are well planned and the conclusions are well stated. However, several technical shortcomings should be fixed before publication in any journal:

1. Please provide the quantification of all three replicates of western blots depicted in figures: 1a, 3a, 4a, 4b, 5b. This should be accompanied by the proper statistical analyzes of the results.
2. For the panels: 1d, 1g, 3d, and 5d. Next to the absolute distance of the wounds, in different experimental conditions and time points, please provide the data as a percentage of the initial wound size (relation of the wound size at 24h time point to the size in 0h). If this representation of results changes the significance of the results, please discuss it.
3. Please provide the quantification of the immunofluorescence pictures presented in figures 2a and 3e. Please, quantify at least 200 randomly chosen cells from each biological replicate of the given experimental condition and present the data as an average of means fluorescence intensity (within each replicate). Please also include appropriate statistics for these experiments.  

Author Response

First of all, let us thank all three reviewers for their overall positive assessment of our work and for their valuable comments that helped make our work better. Then we answer the reviewers' questions one by one, and we really hope that our answers will be received favorably.

Reviewer #3

Q1: Please provide the quantification of all three replicates of western blots depicted in figures: 1a, 3a, 4a, 4b, 5b. This should be accompanied by the proper statistical analyzes of the results.

R1: We performed the quantification and statistical analysis of all western blots pointed by Reviewer #3.

Q2: For the panels: 1d, 1g, 3d, and 5d. Next to the absolute distance of the wounds, in different experimental conditions and time points, please provide the data as a percentage of the initial wound size (relation of the wound size at 24h time point to the size in 0h). If this representation of results changes the significance of the results, please discuss it.

R2: We recalculated the wound closure as the ratio of the distance of the overgrown area of the scratch at 24 hours to the distance at 0 hours after damage in 21 lines (delta). The distance was measured using the Adobe Photoshop CS6 program. Data presented in % of delta. The graphs on figs 1e,h, 3e, 5d look more didactic now. Thank you for the advice.

Q3: Please provide the quantification of the immunofluorescence pictures presented in figures 2a and 3e. Please, quantify at least 200 randomly chosen cells from each biological replicate of the given experimental condition and present the data as an average of means fluorescence intensity (within each replicate). Please also include appropriate statistics for these experiments.  

R3: We did this. The data of the quantification accompanied by statistical analysis are presented on Figs  2b, and 3g.