Salinomycin Treatment Specifically Inhibits Cell Proliferation of Cancer Stem Cells Revealed by Longitudinal Single Cell Tracking in Combination with Fluorescence Microscopy

Round 1

Reviewer 1 Report

The manuscript by Kamlund et al is interesting, as it shows an original perspective allowing to see how individual cells in a population can be affected by a drug. Something, that would be difficult to observe with classical methods looking at averages in bulk populations. The manuscript is well written, though data can be presented in a more clear way. First of all, statistical significances are not given on the graphs, which makes it difficult to compare the groups. Since salinomycin-treated cells are compared to control cells why not to show it on one graph with scaling the values to the control?

Figure 2 A – it is not easy to statistically interpret these graphs. It would be very helpful to show another graph indication the % of cells dividing in the CD24- and CD24+ groups (in control and salinomycin-treated groups).

On Figure 2 B proportion of CD44+/CD24- cells (stem cells phenotype) in control cell culture seems quite high, especially after 48h (at around 70%). One would expect that the proportion of stem cells in the cell culture would be much lower. What are the reported percentages of CD44+/CD24- cells in JIMT-1 cell line in the literature? Also, it is visible, that the proportion of CD44+/CD24- cells increases in the control within 24 h from around 40% to 70%. How can this be explained? The population of cells should be in equilibrium, 30% increase in a control cell line is quite significant. Could this be related to experimental conditions?

Could salinomycin-treatment be related to the inhibition of EMT transition, since it changes E-cadherin levels (the main indicator of EMT)? Are there any literature data about this?

Author Response

We appreciate the comments by the reviewer and have done our best to answer the questions and make appropriate changes in the manuscript. We believe the changes made by the comments have improved the manuscript. We hope the reviwer reads the comments to reviewer number 2 as well. Our comments are found italic style.

The manuscript by Kamlund et al is interesting, as it shows an original perspective allowing to see how individual cells in a population can be affected by a drug. Something, that would be difficult to observe with classical methods looking at averages in bulk populations. The manuscript is well written, though data can be presented in a more clear way. First of all, statistical significances are not given on the graphs, which makes it difficult to compare the groups.

Statistics is found for Figure 3 in the figure legend as well as in the main text describing Figure 3. We tried to put statistics marks in the figure from the beginning but it proved to be to complicated with too many lines. Therefore, we put it in the text. We have rewritten the main text explaining Figure 3 to make comparison easier and pointing to the most interesting findings (lines 316-355). We have now also included statistics comparing control with salinomycin.

Since salinomycin-treated cells are compared to control cells why not to show it on one graph with scaling the values to the control?

Either way of showing the results are commonly accepted, but we find it more informative to show data from both control and salinomycin-treated cultures.

Figure 2 A – it is not easy to statistically interpret these graphs. It would be very helpful to show another graph indication the % of cells dividing in the CD24- and CD24+ groups (in control and salinomycin-treated groups).

Figure 2A only shows one representative experiment for each panel. All data are shown in Figure S2. We have now added a Table S1 where the cell divisions and identities are shown for the individual tracked cells. We have added text at line 257: Table S1 shows that no cells were found to have completed 3 cell division in salinomycin-treated cultures in contrast to control where a large proportion of the cells had gone through 3 divisions during 48 hours. In the legend of Figure 2 we have added: The data are compiled from all data shown in Figure S2. They are derived from 3 independent experiments with 2 independent time-lapses in each.   

On Figure 2 B proportion of CD44+/CD24- cells (stem cells phenotype) in control cell culture seems quite high, especially after 48h (at around 70%). One would expect that the proportion of stem cells in the cell culture would be much lower. What are the reported percentages of CD44+/CD24- cells in JIMT-1 cell line in the literature? Also, it is visible, that the proportion of CD44+/CD24- cells increases in the control within 24 h from around 40% to 70%. How can this be explained? The population of cells should be in equilibrium, 30% increase in a control cell line is quite significant. Could this be related to experimental conditions?

The proportion of cancer stem cells is high in the JIMT-1 cell line (Huang et al., ACS Chem. Biol., 2014, 9, 1587-1594, doi: 10.1021/cb5002153) and can vary between 50 and 80 % at 96 hours after seeding (late exponential phase) when we have performed most of our observations earlier. In the present study, the experiments ended at 72 hours after seeding i.e. during the exponential phase. Salinomycin treatment (or treatment with salinomycin analogues) always lowers the proportion of cancer stem cells as evaluated with different cancer stem cell specific assays (flow cytometric evaluation of the CD44⁺CD24^- population and ALDH⁺ population as well as a functional assay determining colony forming efficiency in soft agar). We have actually never looked at the dynamics in the change of the cancer stem cell population. There is ongoing mathematical modeling to understand the dynamics in the cancer stem cell population (https://doi.org/10.1371/journal.pone.0224787). If a cell population always was growing at a constant rate, the cancer stem cell population should be at equilibrium as the reviewer writes. However, in cell culturing the growth dynamics vary and specifically after seeding of cells and thus, changes in the composition of sub-populations can be expected. The question posed by the reviewer is interesting and warrants further studies in the connection of understanding cancer stem cell dynamics. We prefer to not comment on this in the present manuscript as it will deviate from the main message, which is the combined used of DHM and fluorescence microscopy.

Could salinomycin-treatment be related to the inhibition of EMT transition, since it changes E-cadherin levels (the main indicator of EMT)? Are there any literature data about this?

Salinomycin treatment induces MET ([44]    X. Huang et al., “Breast cancer stem cell selectivity of synthetic nanomolar-active salinomycin analogs,” BMC Cancer, vol. 16, pp. 1–13, 2016). Here we show that salinomycin treatment shifts the distribution of E-cadherin (Fig. 2B) and increases the expression which is in line with our observation of salinomycin inducing MET.

Reviewer 2 Report

The method and results reported are potentially interesting, however, the current manuscript appeared immature to be published. If the authors are going to report the new method, which combines digital holography and immuno-fluorescent staining to investigate the reactions of each individual cells in the heterogenous cell population, they should evaluate the obtained data quantitatively to demonstrate the advantages of their method.

1. The authors are not taking advantages of digital holography. In spite of the title that “Salinomycin Treatment Specifically Inhibits Cell Proliferation of Cancer Stem Cells Revealed by Longitudinal Single Cell Tracking”, just traces of longitudinal tracking are presented (Fig.2A). The authors did not evaluate the relationships between cell proliferation and salinomycin treatment, CD24 and E-cadherin expression in the individual cells to reveal the reactions of each individual cells in the heterogenous cell population.

2. The picture of Holomonitor M4 in Figure 1A is not necessary. The reviewer cannot understand what the authors want to show by Fig. 1B and 1C.

3. The number of the cell tracked should be presented for Fig.2 and Fig.3. Since the authors conducted the experiment three times with two time-lapses (lane 133), they should show the variability between measurements. It would be helpful to demonstrate the effectiveness of their method. The reviewer cannot understand what the authors want to show by Fig. 2A.

4. Circular statics might be employed to evaluate the migration directness of the cells.

5. For convenience of the readers, the authors might mention the reasons why it is appropriate to use JIMT-1 cell line as a model of Cancer Stem Cells in Introduction or Discussion section.

6. The manuscript lacks the discussion about the relationship between salinomycin treatment, cell motility and expression of CD24 or E-cadherin.

7. The reviewer strongly recommend to analyze quantitatively and statistically the data presented in Supplementary figure S2 and add to the manuscript.

7. Histograms of the fluorescence intensities of the cells might be presented as supplementary data to confirm the classification of the cells by the authors.

Author Response

We appreciate the comments by the reviewer and have done our best to answer the questions and make appropriate changes in the manuscript. We believe the changes made by the comments have improved the manuscript.We hope the reviewer will read the comments to reviewer 1 as well. Our answers are in italic style.

The method and results reported are potentially interesting, however, the current manuscript appeared immature to be published. If the authors are going to report the new method, which combines digital holography and immuno-fluorescent staining to investigate the reactions of each individual cells in the heterogenous cell population, they should evaluate the obtained data quantitatively to demonstrate the advantages of their method.

1. The authors are not taking advantages of digital holography. In spite of the title that “Salinomycin Treatment Specifically Inhibits Cell Proliferation of Cancer Stem Cells Revealed by Longitudinal Single Cell Tracking”, just traces of longitudinal tracking are presented (Fig.2A). The authors did not evaluate the relationships between cell proliferation and salinomycin treatment, CD24 and E-cadherin expression in the individual cells to reveal the reactions of each individual cells in the heterogenous cell population.

This study can only be undertaken with DHM since we follow the cells longitudinally and investigate how they proliferate and move. In addition, besides DHM, this study requires that DHM is combined with fluorescence microscopy. Fluorescence microscopy is only used at the end for cell identification. We have changed the title to include also fluorescence microscopy. It now reads: Salinomycin Treatment Specifically Inhibits Cell Proliferation of Cancer Stem Cells Revealed by Longitudinal Single Cell Tracking in Combination with Fluorescence Microscopy.

To clarify the effects on cell proliferation in different cell populations, which may be difficult to discern in Figure 2A and supplementary Figure S2, to we have added a Supplementary Table 1. As the CSCs were only identified in the last frame of the time-lapse it is not possible to create growth curves for the different subgroups of cells. Cell proliferation data consisting of number of cell divisions are shown in Figure S2 and Table S1. Please also see comments above to reviewer number 1.

2. The picture of Holomonitor M4 in Figure 1A is not necessary. The reviewer cannot understand what the authors want to show by Fig. 1B and 1C.

Figure 1B and 1C are there to explain the experimental set up clearly for the reader. They also show that we use DHM, that DHM is the basis for the entire experimental set up.

3. The number of cells tracked should be presented for Fig.2 and Fig.3. Since the authors conducted the experiment three times with two time-lapses (lane 133), they should show the variability between measurements. It would be helpful to demonstrate the effectiveness of their method. The reviewer cannot understand what the authors want to show by Fig. 2A.

The number of analysed cells can now be deduced from Supplementary Table S1. At the start of the time-lapses there were approximately 10-15 cells per frame. This is a recommended cell number to be able to follow cells longitudinally for several days. Figure 2A shows one representative experiment while Figure S2 shows all experiments. We wish to show Figure 2A to demonstrate clearly the principle of constructing family trees for the different treatments and sampling times.

4. Circular statics might be employed to evaluate the migration directness of the cells.

Circular statistics may be employed but is not the most commonly used form of statistics for cell migration.

5. For convenience of the readers, the authors might mention the reasons why it is appropriate to use JIMT-1 cell line as a model of Cancer Stem Cells in Introduction or Discussion section.

We have used the JIMT-1 breast cancer cell line in studies of the effect of different compounds including salinomycin and salinomycin analogues on CSCs as well in studies using digital holographic microscopy [doi: 10.1021/cb5002153, 37, doi: 10.1186/s12885-016-2142-3, doi: 10.1371/journal.pone.0184304, 41]. The cell line contains a high proportion of CSCs that are sensitive to different treatments. Here we deepen our insight into dynamics of how the salinomycin treatment decreases the CSC sub-population of JIMT-1 cells.

6. The manuscript lacks the discussion about the relationship between salinomycin treatment, cell motility and expression of CD24 or E-cadherin.

In two paragraphs in the discussion we comment on salinomycin, cell motility, E-cadherin, and CD24 (lines 540-543). We have now added: “We have previously shown that 72 hours of treatment of JIMT-1 cells with salinomycin or salinomycin analogues at their respective IC₅₀ concentrations increases the number of cells expressing E-cadherin and b-catenin [44]. Here we further analyse the dynamics in the change of E-cadherin expression”.

Lines 554-557 have been added further confirming the data presented here as they show the same results as we found in [41] with another experimental set up.

7. The reviewer strongly recommend to analyze quantitatively and statistically the data presented in Supplementary figure S2 and add to the manuscript.

We have added Table S1 containing all information about cell divisions in relation to cell identity. Please see comments above as well.

7. Histograms of the fluorescence intensities of the cells might be presented as supplementary data to confirm the classification of the cells by the authors.

The images were evaluated visually and to clarify the text written in Materials and Methods (line 200): “For E-cadherin, the cells were visually quantified as E-cadherin low (E-cad^low) with a weak expression over the surface, E-cadherin mid (E-cad^mid) with a stronger expression over the surface but weak edges, or E-cadherin high (E-cad^high) with a strong expression on the cell edges as well as expression over the entire surface.”, we have added Figure S3 to the supplement.

Round 2

Reviewer 2 Report

Information has been added to the introduction and discussion sections to make it easier for readers to understand the background and the achievements of the study. I appreciate that the description of the results is more detailed and the desired corrections have been made.

If possible, consider how to display Fig2A before published. The author conducted CD24 immunostaining after the each experiments was over. In the time sequence, CD24 positives and CD24 negatives (and unkowns) were the last to be judged, and the cell lineage was traced back from the last. The CD24- and CD24+ displays might be placed on the right side of the figure instead of the left side for the reader's convenience, since the reviewer was confused at the first look.