LOXL3 Silencing Affected Cell Adhesion and Invasion in U87MG Glioma Cells

Round 1

Reviewer 1 Report

The manuscript “LOXL3 Silencing Affected Mitochondria Localization and Protein Recycling in U87MG Glioma Cells” by Laurentino et al, analysed the expression of lysyl oxidase-like 3 (LOXL3) in tumours and cancer cell lines. The authors found that glioblastoma (GBM) is among the tumour entities with highest LOXL3 expression and proceeded to study some of the effects of LOXL3 silencing in U87 glioblastoma cells. The authors performed transcriptomic analysis and showed that LOXL3 silencing alters cell morphology and reduces cells numbers, while increasing apoptosis and adhesion.

The study made good use of public databases and analysis tools to characterise the expression of LOXL3 in different types of tumours and cancer cell lines. It also provides some interesting insights into the function of LOXL3 in glioblastoma cells and the effects of its silencing on gene expression. Most parts of the manuscript are well presented and can be of interest to researchers working on LOX family proteins or GBM.

There are however certain deficiencies in the study.

1) The data on LOXL3 localisation in mitochondria and its effect on mitochondrial distribution are unconvincing. The staining pattern for LOXL3 is not really suggestive of mitochondrial localisation and given the rather wide area covered by the LOXL3 signal, much of the “colocalisation” with MitoTracker is likely to be spurious. A decrease in the supposed “colocalisation” (Mander’s coefficient) in the LOXL3 knockdown may well be due to the reduced overall LOXL3 levels and the observed changes in cell morphology. Furthermore, since the LOXL3 KD cells become larger, the fact that one detects mitochondria further away from the centre or a greater total MitoTracker signal may not have any relation to a specific effect of LOXL3 on mitochondria.

Thus, much of the data in Fig. 3 (except for panels F and G) do not seem biologically meaningful and are best left out, together with any associated claims about effects on mitochondria throughout the text. The alternative is that the authors provide a substantial amount of additional analyses and controls, e.g. comparison of the “effects” on mitochondria to those on the ER and/or endosomes; demonstration of enrichment of LOXL3 in purified mitochondria or immuno-electron microscopy to confirm specific localisation in/on mitochondria; normalisation of distance from centre and overall signal to cell size (Fig. 3B-D).

2) The authors should perform invasion assays using the siLOXL3 cells. Given the role of LOX proteins in ECM remodelling, as well as the interaction of LOXL3 with Snail, one of the most relevant effects to characterise would be that on invasion, but such data are lacking.

3) The transcriptomic data are vastly overinterpreted.

The RNA-seq data in themselves are well presented and Fig. 6 provides a nice illustration of some of the differentially expressed genes and processes in which they might potentially be involved. However, the authors proceed to make a huge leap from the expression profiling to postulating functional changes in the siLOXL3 cells, without providing any experimental evidence for them. Some parts of the text are formulated as if the authors have actually shown these hypothesised functional effects, although they haven’t examined them at all beyond the weak correlative links suggested by the RNA-seq data. The most egregious example is the legend to Fig. 6, which reads as if the authors have firmly demonstrated a plethora of functional effects upon LOXL3 depletion, whereas in fact they can at best tentatively speculate on the theoretical plausibility of such effects based on changes in the expression of a handful of genes annotated to compatible functional groups. Notably, even the expression changes detected by RNA-seq (let alone any functional consequences that they might have) are not validated by any other method for any of the genes – yet, the authors build such complex models on them.

In general, Fig. 6 is very well composed and has presumably required a considerable amount of work and creative thinking. However, since the large majority of the functional effects and underlying molecular mechanisms proposed in this figure have not been experimentally demonstrated by the manuscript at all, I feel that this figure is misleading and should best be left out. Alternatively, it might fit as an add-on in Fig. 4 or as a supplement, but only if the legend is completely rewritten to make it clear that any claims of functional effects are pure speculation at this point, not based on any experimental data.

Some of the Conclusions section is also overstated in that respect, in particular the sentence: “This finding suggested that dysregulation of LOXL3 interrupted the energy source to maintain cellular focal adhesion in sprawled tumor cell.”

The title of the manuscript also falls in this category and must be changed: as discussed under point 1, the data on mitochondrial localisation are unconvincing and the authors have performed no analysis of protein recycling.

4) The references cited to demonstrate that LOXL3 is processed by BMP1 are not appropriate, e.g. PMID 11386757 would fit better.

 

Technical points:

a) In Fig. 2, “##” should be defined. In panel B the names of the antigens should be indicated next to the blots. In panel C, was the signal measured per cell, per field or something else?

b) Some of the font sizes used are difficult to read, e.g. in Fig. 1, Fig. 2B, Fig. 4, Fig. 5B, C.

c) It is stated that the glioma lines exhibited the highest LOXL3 expression levels, which does not match the data shown in Fig. 1B.

d) The term DEG(s) is frequently used in the results, but is first defined in the Materials and methods.

e) Lines 252-254: “An increase of approximately 10% was observed with the combined treatment versus TMZ monotherapy (Figure 5B).” If by combined treatment the authors mean TMZ + siLOXL3, the graph appears to show a greater than 10% difference.

f) The paper is generally well written, but there are occasional typos, grammatical mistakes or confusing formulations, e.g. in the following lines:

- 72

- 94

- 122 (Figure 2.2B)

- 133

- 190 (Enrichment of the ECM of DEGs)

- 205-206 (caveolae plasma membrane caveolins)

- 222 (four binding proteins)

- 570: the last sentence in the legend of Fig. 6 is unfinished, although, as discussed above, this is by far the least of the problems with this legend – in fact, the last two sentences of this legend are the only ones that have any foundation in the manuscript’s findings.

Author Response

Response to Reviewer 1 Comments

 

Dear Reviewer,

 

We appreciated the careful revision of our manuscript and the suggestions to improve it. We took into consideration your comments and we made the modifications as suggested. We marked up all the alterations in the manuscript using Track Changes.

 

1) The data on LOXL3 localisation in mitochondria and its effect on mitochondrial distribution are unconvincing. The staining pattern for LOXL3 is not really suggestive of mitochondrial localisation and given the rather wide area covered by the LOXL3 signal, much of the “colocalisation” with MitoTracker is likely to be spurious. A decrease in the supposed “colocalisation” (Mander’s coefficient) in the LOXL3 knockdown may well be due to the reduced overall LOXL3 levels and the observed changes in cell morphology. Furthermore, since the LOXL3 KD cells become larger, the fact that one detects mitochondria further away from the centre or a greater total MitoTracker signal may not have any relation to a specific effect of LOXL3 on mitochondria.

Thus, much of the data in Fig. 3 (except for panels F and G) do not seem biologically meaningful and are best left out, together with any associated claims about effects on mitochondria throughout the text. The alternative is that the authors provide a substantial amount of additional analyses and controls, e.g. comparison of the “effects” on mitochondria to those on the ER and/or endosomes; demonstration of enrichment of LOXL3 in purified mitochondria or immuno-electron microscopy to confirm specific localisation in/on mitochondria; normalisation of distance from centre and overall signal to cell size (Fig. 3B-D).

Response: We agree with the reviewer that we do not presented enough data to support the LOXL3 localisation in mitochondria, and we took out the results and discussion concerning this aspect. We will perform further experiments to explore the intracellular localization of LOXL3 in the near future. We modified the figures excluding the data concerning mitochondria localization.

 

2) The authors should perform invasion assays using the siLOXL3 cells. Given the role of LOX proteins in ECM remodelling, as well as the interaction of LOXL3 with Snail, one of the most relevant effects to characterise would be that on invasion, but such data are lacking.

Response: We performed invasion assay as suggested, and LOXL3 silencing led indeed a decrease in U87MG tumor cells invasion. These new data were added into the figure 4D, and the description of this results in RESULTS. We also highlighted this behavior in the DISCUSSION in the context of ECM remodeling.

 

3) The transcriptomic data are vastly overinterpreted.

The RNA-seq data in themselves are well presented and Fig. 6 provides a nice illustration of some of the differentially expressed genes and processes in which they might potentially be involved. However, the authors proceed to make a huge leap from the expression profiling to postulating functional changes in the siLOXL3 cells, without providing any experimental evidence for them. Some parts of the text are formulated as if the authors have actually shown these hypothesised functional effects, although they haven’t examined them at all beyond the weak correlative links suggested by the RNA-seq data. The most egregious example is the legend to Fig. 6, which reads as if the authors have firmly demonstrated a plethora of functional effects upon LOXL3 depletion, whereas in fact they can at best tentatively speculate on the theoretical plausibility of such effects based on changes in the expression of a handful of genes annotated to compatible functional groups. Notably, even the expression changes detected by RNA-seq (let alone any functional consequences that they might have) are not validated by any other method for any of the genes – yet, the authors build such complex models on them.

In general, Fig. 6 is very well composed and has presumably required a considerable amount of work and creative thinking. However, since the large majority of the functional effects and underlying molecular mechanisms proposed in this figure have not been experimentally demonstrated by the manuscript at all, I feel that this figure is misleading and should best be left out. Alternatively, it might fit as an add-on in Fig. 4 or as a supplement, but only if the legend is completely rewritten to make it clear that any claims of functional effects are pure speculation at this point, not based on any experimental data.

Some of the Conclusions section is also overstated in that respect, in particular the sentence: “This finding suggested that dysregulation of LOXL3 interrupted the energy source to maintain cellular focal adhesion in sprawled tumor cell.”

The title of the manuscript also falls in this category and must be changed: as discussed under point 1, the data on mitochondrial localisation are unconvincing and the authors have performed no analysis of protein recycling.

Response: We agree with the reviewer that the pathways depicted in the Figure 6 are speculative based on the RNASeq results, but the experimental evidence are missing at this point. Therefore, we excluded the Figure 6. We rewrite the DISCUSSION, CONCLUSION, and modified the TITLE accordingly.

 

4) The references cited to demonstrate that LOXL3 is processed by BMP1 are not appropriate, e.g. PMID 11386757 would fit better.

Response: The reference was changed as suggested.

 

Technical points:

1. a) In Fig. 2, “##” should be defined. In panel B the names of the antigens should be indicated next to the blots. In panel C, was the signal measured per cell, per field or something else?

Response: “##” was defined. The antigens names were indicated in the Figure 2B. In Figure 2C the signal was measured per cell, and this aspect was clarified in the figure legend.

 

1. b) Some of the font sizes used are difficult to read, e.g. in Fig. 1, Fig. 2B, Fig. 4, Fig. 5B, C.

Response: The mentioned font sizes were increased.

 

1. c) It is stated that the glioma lines exhibited the highest LOXL3 expression levels, which does not match the data shown in Fig. 1B.

Response:

We agree and corrected the statement to the following:

“The cell lines derived from gliomas (U138MG and U87MG) were in the top five cell lines presenting the highest LOXL3 expression levels among the 64 human cell lines of the HPA study.”

 

 

1. d) The term DEG(s) is frequently used in the results, but is first defined in the Materials and methods.

Response: The term DEG was defined in the Results (item 2.4).

 

1. e) Lines 252-254: “An increase of approximately 10% was observed with the combined treatment versus TMZ monotherapy (Figure 5B).” If by combined treatment the authors mean TMZ + siLOXL3, the graph appears to show a greater than 10% difference.

Response: We corrected the result description. Combined treatment (siLOXL3 + TMZ) presented an increase of approximately 20% when compared to TMZ treatment

 

1. f) The paper is generally well written, but there are occasional typos, grammatical mistakes or confusing formulations, e.g. in the following lines:

- 72

“the expression of LOXL3 (expression) in GBM”: the second expression was excluded.

- 94

“patients survived in the high- and low-expression groups (survived), respectively”: the second survived was excluded.

- 122 (Figure 2.2B)

We corrected to Figure 2B

- 133

“Quantification of LOXL3 fluorescence (quantification)”: the second quantification was excluded.

- 190 (Enrichment of the ECM of DEGs)

We corrected to: Enrichment of DEGs related to ECM

- 205-206 (caveolae plasma membrane caveolins)

We corrected to: genes coding for the caveolae plasma membrane caveolins

- 222 (four binding proteins)

We corrected to: which included three genes coding for  binding proteins

- 570: the last sentence in the legend of Fig. 6 is unfinished, although, as discussed above, this is by far the least of the problems with this legend – in fact, the last two sentences of this legend are the only ones that have any foundation in the manuscript’s findings.

This figure was excluded.

 

Reviewer 2 Report

In this manuscript the authors investigated in the glioblastoma cell line U87MG the effects of LOXL3 silencing, through well executed transcriptomic analysis and immunofluorescence proving the colocalization of LOXL3 with mitochondria and after its downregulation the authors suggested a movement towards focal adhesion sites.

The work is well-executed where experiments are technically sounding and the aim of the study is noteworthy, but I think that the research design is not robust for the level/impact of this Journal. 

Glioblastoma multiforme is characterized by a high degree of cellular and genetic heterogeneity with multiple pathway-dependencies both within and across each particular subtype (as mesenchymal, classical and proneural). So it would be expected that, as for this type of heterogeneous tumor, the analysis was also carried out in at least one other line of a different molecular subtype.

Author Response

Response to Reviewer 2 Comments

 

Dear Reviewer,

 

We appreciated the careful revision of our manuscript and the suggestions to improve it. We took into consideration your comments and we made the modifications as suggested. We marked up all the modifications in the manuscript using Track Changes.

 

In this manuscript the authors investigated in the glioblastoma cell line U87MG the effects of LOXL3 silencing, through well executed transcriptomic analysis and immunofluorescence proving the colocalization of LOXL3 with mitochondria and after its downregulation the authors suggested a movement towards focal adhesion sites.

The work is well-executed where experiments are technically sounding and the aim of the study is noteworthy, but I think that the research design is not robust for the level/impact of this Journal.

Glioblastoma multiforme is characterized by a high degree of cellular and genetic heterogeneity with multiple pathway-dependencies both within and across each particular subtype (as mesenchymal, classical and proneural). So it would be expected that, as for this type of heterogeneous tumor, the analysis was also carried out in at least one other line of a different molecular subtype.

Response: We agree with the reviewer that glioblastoma is a very heterogeneous tumor. We carried out the LOXL3 silencing in another GBM cell line of a molecular subtype, T98G, presenting TP53 mutation and less aggressive in vitro and in vivo characteristic than U87MG. We obtained LOXL3 silencing in T98G cells at day 4 demonstrated both at gene and protein expressions demonstrated by RT-qPCR and Western blot, respectively. We also demonstrated a morphological change in T98G cells after LOXL3 silencing, similar to that observed in U87MG cells, with increasing of the cell surface area. These results were presented in Figure 6 A and B. Therefore, a similar effect of LOXL3 silencing was observed in these different molecular subtypes of GBM cells. We also performed in silico analysis in the TCGA RNASeq dataset of different molecular subtypes of GBMs. The heatmap presented in Figure 6D shows the differential expression of gene associated to LOXL3, mainly genes conding for tubulin and genes related to focal adhesion in G-CIMP, proneural, classical and mesenchymal subtypes. The correlation of the gene expressions presented in Figure 6E shows a significant correlation between LOXL3 expression and tubulin (TUBA1C, TUBB6), actin related protein (ACTR2), vimentin (VIM) and gelsolin (GSN), which are genes coding for protein related to cytoskeleton microfilament and cell adhesion. These results further corroborate the LOXL3 association with these genes in different GBM molecular subtypes.

 

Round 2

Reviewer 1 Report

The authors have taken my major concerns into account, mostly by removing unsubstantiated claims and the corresponding data from the manuscript, which has made the latter more robust and credible – an example of less being more. The added invasion assays have also increased the relevance of the findings.

On the other hand, some of the minor points have not been properly addressed. For example, the fonts in the figures I had mentioned in my previous report, if anything, have become less readable in the revised version. As a minimum, the authors should take care to use higher resolution images and lower compression when generating the PDFs, so that text is legible at least upon magnification.

Panel 2B still does not show the names of the proteins for which the WBs were performed.

In the revised abstract, a full stop is missing in line 14, and a space is missing in line 21.

Author Response

Dear Reviewer,

Thank you very much again for your careful review of our manuscript, which helped to increase the relevance of our findings. We took into consideration your comments and we made the modifications as requested.

We improved the resolution of all images to make text more legible at higher magnification. Additionally we added the names of the protein in the Figure 2B, and made the modification in the Abstract.

 

 

Reviewer 2 Report

In the revised manuscript, LOXL3 Silencing Affected Mitochondria Localization and Protein Recycling in U87MG Glioma Cells, now LOXL3 Silencing Affected Cell Adhesion and Invasion in U87MG Glioma Cells the authors have addressed most of the concerns and they have answered many of our initial queries, so the main scientific message is stronger and clearer in this version.

In particular the authors have included the analysis of LOXL3 silencing in T98G glioma cell line and argued these new data in the revised paper.

I think that this new version of the manuscript it will be acceptable for the publication of IJMS, but several minor points need to be corrected

• Line 14: add a point (.) after LOXL3 expression level;
• Line 21: divide observeddecrease;
• Line 149: differentiaLLy with 2 L;
• Line 260: the efficacy of loxl3 silencing is confirmed not by gene but by RNA expression analysis;
• Line 366: point 6 not 8;
• Line 408: were not was;
• Line 550: Statistical analyses paragraph is now 4.14 no more 4.13 and in a newline;
• Line 582: Reference in a newline

Author Response

Dear Reviewer,

We thank you for all your previous suggestions, which helped us to improve our manuscript. We took into consideration your minor points and we made all corrections as requested.