Insight into the lncRNA–mRNA Co-Expression Profile and ceRNA Network in Lipopolysaccharide-Induced Acute Lung Injury
Round 1
Reviewer 1 Report
The manuscript by Shen et al., focuses on elucidating the ceRNA network associated with acute lung injury (ALI). Authors claim to have identified lncRNA-mRNA coexpression profile in LPS-treated BEAS-2B cells that may have implications in ALI. The study is primarily based on bioinformatic analyses with BEAS-2B cells and can be potentially significant in understanding the molecular basis of ALI. The study may have potential for diagnosis and treatment of ALI. The manuscript can be improved further following suggestions described below:
1. What is the rationale for using LPS? Does LPS-treated BEAS-2B cells mimic cells with ALI? Authors need to explain that LPS may initiate a signaling cascade in BEAS-2B cells, which may lead to cell activation and expression of immunoregulatory or inflammatory cytokines. If it represents a cell culture model of ALI, an appropriate reference needs to be cited.
2. In results section 3.9., authors suddenly started talking about miRNA in the context of ceRNA network. How did they identify miRNA that facilitated the lncRNA-mRNA interaction?
3. Authors are taking it for granted that the reader knows everything and they have nothing to explain. For example, in section 3.2 of results authors wrote…”The intersection of the CPC[23], 219 PLEK[24], CNCI[25], and Pfam[26] was exploited to predict 1,418 novel lncRNAs (Supplementary Figure S1e).”…No attempt was made to explain what CPC, PLEK, CNCI and Pfam are. For understanding of a general reader of CIMB, please explain each and how they were used in data analysis.
4. Any acronym or abbreviation must be expanded when using the first time in text.
5. Figure legends are sketchy and provide little information about the figure. All figures legends need more elaborate description about the figure so that it is easier for readers to understand results.
Minor English editing is required.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
In this manuscript, the expression of lncRNA and mRNA BEAS-486 2B cells after LPS stimulation was measured and analyzed including the effect of lncRNA on mRNA and building ceRNA networks. This work also suggests it can help find potential biomarkers for Acute Lung Injury (ALI). While this study is interesting, I have many concerns regarding the rigor of the work, which I listed below. I hope my comments will help the authors improve their work.
Major comments:
1) Abstract: the authors suggest their findings might be useful in developing biomarkers for ALI. The study was done mostly in-vitro with only 1 lncRNA tested in patients with ALI/control with a very small sample size. So, this sounds premature to suggest in the abstract that this work is towards biomarkers. Also, ALI diagnosis criteria are well-known, and diagnosis is routinely made in clinical settings. So the need for a biomarker to detect ALI is not clear.
2) ALI has 2 sub-phenotypes. It is not clear if the authors tried to model 1 of these 2 sub-phenotypes (if that is even possible in-vitro) or if the model is for both sub-types. It is not clear if for the human study patients were from sub-phenotype 1 or 2 or both (which can affect the results).
3) Lines 60-61: “In recent years, several lncRNAs have has been shown to participate in the process of acute lung inflammation…”. I could see any later reference to these earlier studies. The authors should compare their sequencing results to the previously reported transcripts, to see if there is any replication between studies.
4) Methods, lines 195-6: “ Data were presented as mean ± standard deviation (SD) or median (interquartile range)” Please indicate in the relevant legends when mean+sd or median was used.
5) Fig1 shows LPS caused less viability and proliferation. Can the authors discuss the options that transcriptomic differences are due to cell death, more cell-free RNA etc .. and are not specific to LPS?
6) Results line 202- dose response for 1-50 mg/Ml is described. In the methods section, only the 10mg dose is described . Please correct and justify while this dose was chosen for the next experiments.
7) RNA-seq data analysis: The was no correction for multiple comparisons and or FDR. Only a cutoff for fold-change and nominal p-value of 0.05. I couldn’t find a table with FDR-corrected values. Table 1 shows the top 10 lncRNA. Even in this shortlist of top transcripts. the p-values of at least some of them would become not significant after FDRF correction.
This is very troubling. Basically, there is no way to know if all the results of this study are just false-positive.
I tried to have a look at the raw data deposited at the NCBI GEO database details in the text: https://www.ncbi.nlm.nih.gov/sra/PRJNA872330, but it is not available.
8) In the qPCR validation ENST00000642173 had the largest effect. But in the human samples, ENST00000627824 was tested and not ENST00000642173. Why ? If there is available RNA from patients one would accept to see the quantification of ENST00000642173 that was the “strongest” effect in-vitro, and then also all other genes as well. Did the authors measure all other transcripts that were tested by qPCR in-vitro ? please show the data for all comparisons even if they are not significant.
9) I’m sorry to write this but Table 2 is very sloppy. There were 12 patients (6 ALI and 6 healthy controls), but clinical data are described for only half of them (6)! And it is not clear if they are ALI or healthy (or mix ?!). Also, there are individuals with high blood pressure and smokers these are two factors that might have a big impact on the results regardless of ALI. So as it stands the human results are not convincing.
Minor comments:
1) “Thus identifying lncRNAs that make a difference in the ALI model…” (lines 70-71). “that make a difference”. Did the authors mean “differentially expressed” ? Please rephrase.
2) Methods: analysis was done with DeSeq that was published in 2012. The improved version DeSeq2, was published in 2014 (https://genomebiology.biomedcentral.com/articles/10.1186/s13059-014-0550-8) and it allows much improved interpretability of the results. I encourage the authors to reanalyze with the newer version.
3) DElncRNA + DEmRNA > please make it explicitly explained in the first time these terms are mentioned.
Differentially expressed lncRNA (DElncRNA). These are not so common abbreviations.
4) Figure 4B- the X-axis is missing.
I mentioned 1-2 minor edits needed in my comments to the authors
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
The authors answered most of my comments and corrected their manuscript accordingly.
Yet, I do not support publishing transcriptomic data without correction for multiple comparisons / false-detection ratio. In this case, also the fold change cutoff was low.
The authors referred to previously published works that analyzed rna-seq without FDR. However, this is a poor analysis practice, and we should not encourage keeping with this approach.
The authors should reanalyze the data accordingly.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 3
Reviewer 2 Report
The authors provided a detailed table of their lncRNA-seq data, and they mentioned that most of the top lncRNA did not reach the cutoff of FDR<0.05.
Notably, in some cases, despite a large log2 fold change, the significance level was not reached, implying the data has large variations between biological replicates.
I suggest that in order to make these data more trustable, the authors include this table in the paper (or as a supplement), clearly that most of the data is insignificant after FDR correction and validate the top 10 by other means (qPCR) on independent biological replicates.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 4
Reviewer 2 Report
The authors answered my questions and comments.
Thanks
