The Aminoacyl-tRNA Synthetase and tRNA Expression Levels Are Deregulated in Cancer and Correlate Independently with Patient Survival

Round 1

Reviewer 1 Report

Sangha and Kantidakis perform an analysis of genomic alterations, mRNA expression, and survival correlations for the aminoacyl-tRNA synthetases in multiple cancer types from the TCGA. Additionally, tRNA expression and correlation with survival in these same cancers is examined. While similar analysis has been performed before for just the tRNA synthetases, the work is especially impactful for its comparison of the synthetases and tRNAs together. Analysis of specific tRNA isoacceptors and their relationship to survival is also useful. The analyses are thoughtful, and with some minor improvements to clarity and additional considerations for some statistical analyses, this work will be a helpful contribution for the field.

General Comments:

1.     For analysis of significance with multiple comparisons, you are calculating an FDR-adjusted p-value, which is a q-value. Your FDR should remain at 0.05 (the value used in this work). As it is currently used in the manuscript, FDR should be replaced with q-value.

2.     How is tRNA expression evaluated for the TCGA dataset? tRNA is typically difficult to quantify due to its structure and modifications. Could changes in tRNA levels in cancer be artifacts caused by changes in modifications (affecting the tRNA sequencing?)

3.     It would be interesting to test the correlation between ARS expression and expression of certain tRNA isoacceptors instead of all isoacceptors together for a single amino acid. It appears changes in specific isoacceptors are much larger and you may be able to get more interesting correlations that way.

4.     Is there a conflict between the results found in this work looking at the effects of ARS expression on survival (such as in KIRC) with the results of Ref. 35 (Wang et al.)? Wang et al. showed the opposite trend for survival in KIRC where lower expression of some ARS led to worse survival.

Specific Comments:

1.     Line 28: There are 36 ARSs in humans, but the distribution is 17 in the cytoplasm, 17 in the mitochondria, and 2 in both (mitochondrial GlnRS has not been identified).

2.     Figure 1B: CNA frequency is represented, but as defined in the legend, this includes both amplifications and deletions, which are very different events. There is no indication of whether a particular ARS is usually amplified or deleted in this figure, although there is reference to this data in the discussion (Line 311). It may be helpful to define whether a box includes mostly amplifications or deletions, such as outlining or coloring the box.

3.     For Figure 2A/4A (and Supplementary Figures): Please include the expression units (such as FPKM) for the mRNA expression. This also helps to show that the data is normalized and can be compared across samples.

4.     Figure 2B/4B: Coloring and scale can be improved for clarity. Red and green should not be used together for colorblind people. The Fold-change (FC) scale can be unintuitive to interpret because underexpression values (FC between 0-1) are more condensed than overexpression (1-3). It is difficult to determine differences in overexpression due to the three different colors (Green, Black, and Red). To solve these problems, I suggest using a log2(FC) scale so that underexpression values are negative (and can be one color) while overexpression values are positive (can be another color). Also on the log scale, differences in expression are symmetric around 0 while for the current FC scale, they are asymmetric around 1. Having the log2 scale would also translate easily from the Supplementary Figures where expression is quantified on a log2 scale.

5.     Supplementary Figure S1: Minor point, the figure legend can be reworded to better match was is written in the main text. “mRNA expression of samples with or without CNAs”. **** P-value can be defined.

6.     Supplementary Figures S2, S3: Scale coloring can be changed here as well. Maybe a single color gradient to make comparing differences easier.

7.     Paragraph starting with Line 160: It may be helpful to insert a sentence saying there are no ARSs where lower expression is associated with worse survival.

8.     Figures 3/5: 3A/5A It may not be necessary to have both p-values and q-values plotted together.  If you go by p-value<0.05 only, there is a chance these could be false positives, although it is understandable to show broader trends. q-value < 0.05 should always mean p-value < 0.05. For these Figures and Figures 3B/5B, it may help to compute a Hazard Ratio with Cox’s regression to determine how big the effect on survival is. You could combine significance (q-value) with the Hazard Ratio to give a fuller picture, especially since you do not have Kaplan-Meier curves for all cases. For Figure 3A figure legend, be sure to mention what the q-value represents (q<0.05 means patients with higher ARS expression have worse survival; otherwise this is not clear from the figure).

9.     Figure 6A (Line 241): This analysis appears to be done with the p-values of the previous Figures 3A and 5A. It would be helpful to mention p<0.05 was used for this analysis. Keep in mind for some of the p-values < 0.05, the q-values are > 0.05, meaning they have a higher probability of being false positives. You can consider using q-values only, which would reduce the number of overlapping tRNA/ARS connections. It is interesting that tRNA and ARS expression are mostly uncoupled (Figure 6B), and this change would further highlight that tRNA and ARS are mostly uncoupled with regards to survival.

10.  Figure 7A: Can label the Axis more clearly (# of overlapping overexpressed tRNA isoacceptors), but not strictly necessary.

11.  Figure 7B: How was the p-value calculated and is it meaningful for the analysis? It is not completely clear from the text what it is for.

12.  For Figures where you only show Fold Change (FC) (Figures 2B, 4B), are all of these differences in expression significant? If so, you can consider writing somewhere that they are.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

 

Sangha and Kantidakis presented a manuscript on the expression levels of the aminoacyl-tRNA syntetases (ARSs) and tRNA in cancer. Through in silico approach, the authors investigated the relationship between the ARSs and tRNA expression levels and patients’ overall survival in 10 TCGA cancers. Specifically, they found that ARSs were upregulated in most tumor types and often correlated with worse outcome, while tRNAs were either up- or down-regulated in tumors and their expression levels sometimes correlated with worse patient survival. The manuscript is well written and well supported by consistent references. Overall, the obtained results encourage further studies to validate ARSs and tRNA as potential cancer biomarkers. However, I have some major comments as follow:

1) The extended full name of AIMP1, AIMP2 and AIMP3/EEF1E1 scaffold proteins should be reported when they are cited for the first time along the text (Line 32).

2) The authors analyzed 20 cytoplasmic ARSs genes (Table S2). However, in the introduction section (lines 28-29) they reported that “There are 36 ARSs in humans, 16 of which function exclusively in the cytoplasm, 17 in the mitochondria and three in both”. Please, better clarify the function of the selected ARSs.

3) The raw data used in the study should be provided as supplementary tables in Excel format to better evaluate the obtained results.

4) Lines 171-172, the authors indicated that “MARS and TARS correlated with three different types of tumors (BRCA, KIRC, LIHC)”. According to data reported in Figure 3A, TARS does not correlate with KIRC but with HNSC. Please, verify and correct.

5) The authors reported only a few indicative Kaplan-Meier survival plots (Figures 3B, 5B, and 7D). To provide a complete view, the remaining Kaplan-Meier survival plots should be presented or discuss in the main text if not significant. Log-rank test statistics should be added to each Kaplan-Meier survival plot (Figures 3B, 5B, and 7D).

6) Figure S4 may be integrated to Figure 6 to complete the analysis and provide a better visualization of the obtained results.

7) The data reported in Figures 7C and 7D should be better discussed in the main text.

8) Please, separate the Conclusions.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Dear authors, thank you for your responce.  All questions were addressed.

Best regards.

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.