Evaluation of Antibiotic Tolerance in Pseudomonas aeruginosa for Aminoglycosides and Its Predicted Gene Regulations through In-Silico Transcriptomic Analysis
Round 1
Reviewer 1 Report
The authors corrected the paper following reviewers comments
I suggest to ACCEPT the paper in its present form
Author Response
Response to Reviewer 1 Comments
Point 1: The authors corrected the paper following reviewers comments I suggest to ACCEPT the paper in its present form
Response 1: Thank you for your time to review our work. We are delighted about your positive feedback on our work. We noted that you asked for a minor spell check in the options. We have corrected minor spelling and grammar mistaes. Apart from that, we have made changes as per the comments of Reviewers 2 and 3. Kindly look into the changes we have made and give us your inputs. We are looking forward for your report.
Author Response File: 
Author Response.docx
Reviewer 2 Report
A potentially interesting study- but I am very confused about the study design
In vitro exposure of antibiotics to P. aeruginosa - which strain was used? , if not P. aeruginosa ATCC 27853, how do you know the MIC as the same? and why was ATCC 27853 used for MIC - if indeed it was used- but PA14 used for further studies?
Transcriptomic analysis and protein functional regulations: - this does not have sufficient detail about the methods used. Its difficult to determine - but I think PA14 was used, presumably so that the NCBI datasets could be used.
Why was 5 ug/ml tobramycin used? How does this relate to the MIC data of strain ATCC 27853? Do you know if 5ug/ml is an MIC value - or below the MIC. If below by how much?
how was rna extracted? etc etc - many details have been missed
I think the MIC of PA14 should be analysed - this is not a difficult thing to do and would help resolve the issues I have raised
The terms “antibiotic tolerance” and “resistance” appear to be used interchangeably - but they mean different things- so please re-write so that it is clear whether you are examining tolerance or resistance
Is it correct to compare a bacteria grown planktonically in the presence of antibiotic to a gene set from a biofilm grown culture? This seems to introduce too many variables for a meaningful comparison.
If you think it can be justified - you should write a justification into the discussion
Author Response
Response to Reviewer 2 Comments
Point 1: In vitro exposure of antibiotics to P. aeruginosa - which strain was used? if not P. aeruginosa ATCC 27853, how do you know the MIC as the same? and why was ATCC 27853 used for MIC - if indeed it was used- but PA14 used for further studies?
Response 1: P. aeruginosa reference strain ATCC 27853 was used for in vitro exposure of antibiotics. We determined the MIC of ATCC 27853 for amikacin, gentamicin and tobramycin. Based on the determined MICs, the antibiotic concentrations were used for “In vitro exposure of antibiotics” experimentation.
MIC was particularly used because P. aeruginosa can tolerate antibiotics at MICs and antibiotic treatment failure for clinical infections are common. For further in silico analysis, we used Microarray datasets of PA14 mimicking our experimental condition, as PA14 is a model strain for the study of biofilm formation and virulence.
Our aim of the study is to evaluate the antibiotic tolerance in P. aeruginosa for aminoglycoside groups of antibiotics in absence of the external resistance mechanisms and genetic diversity. Hence, we used the reference strain ATCC 27853 with native characteristics of P. aeruginosa to exclude the external factors. This will answer which antibiotic has superior response in absence of aminoglycoside resistance through in vitro experiment. To uncover the mechanism accounting for the antibiotic tolerance (through in silico analysis) we used PA14 which is an infection model strain with presence of complete genes for biofilm formation and virulence, to analyse the treatment of tobramycin in clinical picture. We used two different strains, each for different analysis (but exposed to same condition). From the early results using ATCC 27853 we found tobramycin has the superior antibiotic effects over others. Then we did a predictive analysis using PA14 by exploring the datasets mimicking our experimental condition to understand the up-regulating and down-regulating mechanisms of P. aeruginosa involved in the decreasing kinetics of antibiotic tolerance and virulence in tobramycin, when exposed at the MIC of Tobramycin.
Point 2: Transcriptomic analysis and protein functional regulations: - this does not have sufficient detail about the methods used. Its difficult to determine - but I think PA14 was used, presumably so that the NCBI datasets could be used.
Response 2: Duly noted. We have revised carefully. Materials and methods were elaborated on the procedures, cut-off values, databases used. The references were included for all the methodology and cited all the databases used for analysis. The procedure for both tools were elaborated and the results were segregated and reported separately.
For transcriptomic analysis, the microarray datasets from GEO, NCBI (available as accessible series no. GSE9991 and GSE9989) were analysed using NetworkAnalyst 3.0 for determining significant genes after log2 transformations and applying limma statistics. Then Gene Ontology and functional enrichment analysis were performed for the DEGs using PANTHER Classification System and DAVID Bioinformatics Resources 6.8. Further annotation of target DEGs, Pseudomonas Genome DB and PseudoCyc were used to determine the protein and molecular interactions of DEGs based on previously published resources. The overall method involves in silico analysis using various computational databases which is completely discussed in the “Methods and Methods”.
The overall analysis was performed to understand the functions of the differentially regulated genes. To elaborate more on how it works: The Gene ontology classifies the DEGs into various categories like 1) Molecular functions, 2) Biological process and 3) Protein classes. Each categories have several subcategories in which the DEGs are enriched. The subcategories cover a broad terms to show preliminary classification of DEGs. Then in the functional enrichment analysis, the DEGs were grouped into functional annotation clusters with enrichment score, FDR, p-value to determine the significance of the enrichments. The individual terms were also presented in graph with enrichment percentage. This displays the distribution of DEGs in the pathways from popular databases like KEGG, InterPro, UniProtKB, and SMART which is used by DAVID Bioinformatics Resources 6.8. for enrichment analysis. Then the target DEGs were functionally annotated using Pseudomonas Genome DB and PseudoCyc which holds the complete data of the gene functions from the published sources. Collectively the regulation of proteins was determined and discussed.
Yes, PA14 was used, and the datasets GSE9991 and GSE9989 perfectly mimicked our experimental condition. Also, PA14 is a model strain for the study of biofilm formation and virulence to explore how the antibiotic exposure regulates gene in a clinical strain (which includes genes necessary for pathogenesis).
Point 3: Why was 5 ug/ml tobramycin used? How does this relate to the MIC data of strain ATCC 27853? Do you know if 5ug/ml is an MIC value - or below the MIC. If below by how much? I think the MIC of PA14 should be analysed - this is not a difficult thing to do and would help resolve the issues I have raised
Response 3: In our in vitro experiments, we exposed MICs of antibiotics to P. aeruginosa ATCC 27853. Hence, we were interested in microarray datasets mimicking our experimental condition (antibiotic exposed to P. aeruginosa at its MIC) and we found the datasets of P. aeruginosa exposed to its MIC of tobramycin to understand the gene regulations at the MIC.
5 µg/ml tobramycin was used in the microarray dataset because it is the MIC of PA14. MIC of PA14 strain ranges from 5 µg/ml to 8 µg/ml, which is well knows as it a well-studied model strain.
Point 4: how was rna extracted? etc etc - many details have been missed
Response 4: Duly noted. In the published datasets, RNA was extracted using RNeasy RNA isolation kit. I have included RNA extraction and other details in the revised version of the manuscript. Thank you for asking.
Point 5: The terms “antibiotic tolerance” and “resistance” appear to be used interchangeably - but they mean different things- so please re-write so that it is clear whether you are examining tolerance or resistance
Response 5: Duly noted. I have revised my paper thoroughly and changed in many places including the title of the paper where I have mentioned “resistance development” instead of “tolerance”. Thank you for the comment.
But please note that the term “adaptive resistance” (which I have still mentioned in few places) denotes antibiotic tolerance mechanism.
Point 6: Is it correct to compare a bacteria grown planktonically in the presence of antibiotic to a gene set from a biofilm grown culture? This seems to introduce too many variables for a meaningful comparison.
Response 6: We explored two datasets, 1) GSE9991 - Tobramycin treatment of planktonic P. aeruginosa (which has 4 samples, out of them 2 was tobramycin treated and the rest 2 was untreated controls). 2) GSE9989 - Tobramycin treatment of P. aeruginosa biofilms (which has 6 samples, out of them 3 was tobramycin treated and the rest 3 was untreated controls). The Test versus Control were only compared within the respective datasets. Not between the datasets which will definitely introduce too many variables as you mentioned. Only for the further downstream analysis, we have proceeded with DEGs from both the datasets. The reason we are interested in both planktonic and biofilm datasets is because, in infections caused by P. aeruginosa, the organism can exist in both states. It depends on the organism’s gene regulations during course of the infection. For instance, P. aeruginosa in cystic fibrosis and nosocomial infections like CAUTI, VAP, CLABSI are predominantly biofilm producers. Hence, we proceeded with DEGs from both the datasets for further downstream analysis to look at the effect of tobramycin in both the forms of P. aeruginosa.
Thanks for your time and your valuable inputs. We have noted that you asked for improvement in methods and results. The whole manuscript was revised thoroughly as per the comments, restructured, and added few more relevant references. Please let us know if further changes are required.
Author Response File: Author Response.docx
Reviewer 3 Report
In the present manuscript, the authors present evidence for the emergence of antibiotic tolerance in Pseudomonas aeruginosa. Here, they have calculated MIC values for 3 aminoglycosides antibiotics, namely amikacin, gentamicin, and tobramycin. The authors further provide putative mechanism for the bacterial adaptation to the antibiotics using analysis of previously published transcriptomics data.
Major:
The paper is poorly written and structured and needs to be edited for English language.
The scattered plot for the differential gene expression is provided for GSE9989 and not for GSE9991.
Materials and method are incomplete and needs to be improved.
No description provided for the MALDI-TOF mediated confirmation of the P. aeruginosa.
The inclusion of two different tools (Panther and David) for go-term analysis while reporting data from only one method is not clear.
Moreover, it is not clear why only top 250 DEGs were selected for the analysis.
Please include references for each methodology.
The authors did not provide relevant references, for example, the microarray datasets were previously published in the "Anderson GG, Moreau-Marquis S, Stanton BA, O'Toole GA. In vitro analysis of tobramycin-treated Pseudomonas aeruginosa biofilms on cystic fibrosis-derived airway epithelial cells. Infect Immun 2008 Apr;76(4):1423-33. PMID: 18212077". Authors did not cite this key paper and did not provide a comparative analysis.
minor:
Title: The authors have used a published microarray dataset for differential gene expression analysis and referred to this data as "In-silico Transcriptomic Analysis"; this is an incorrect use of the terminology.
Author Response
Response to Reviewer 3 Comments
Point 1: The paper is poorly written and structured and needs to be edited for English language.
Response 1: Duly noted. The manuscript was revised, restructured, and edited for English language. The grammar mistakes and casual English usages were reframed throughout the paper. Please use the track changes option in the MS Word file to see the changes we made.
Point 2: The scattered plot for the differential gene expression is provided for GSE9989 and not for GSE9991.
Response 2: Duly noted. The new scattered plot for datasets GSE9991 and GSE9989 were added. The DEGs from both datasets were represented in one consolidated volcano plot to display the degree of variation in gene expression after exposure to tobramycin.
Point 3: Materials and method are incomplete and needs to be improved. Please include references for each methodology. The inclusion of two different tools (Panther and David) for go-term analysis while reporting data from only one method is not clear.
Response 3: Duly noted. We have revised carefully as per the comments. Materials and methods were elaborated on the procedures, cut-off values, databases used. The references were included for all the methodology and cited all the databases used for analysis. The procedure for both tools was elaborated and the results were segregated and reported separately.
Point 4: No description provided for the MALDI-TOF mediated confirmation of the P. aeruginosa.
Response 4: Duly noted. Reference included for the sample preparation in methodology and in the results, we mentioned the determining factor for identification by VITEK® MS. We used MALDI-TOF automated identification system from VITEK® MS, BioMérieux that provides species level matching based on the peptide mass fingerprints. It ranks top 10 species and type strains that matches the exact protein profile in the inbuilt database. As it is an automated system with existing proteome databases for microbial identification, we are unable to provide more description. However, we included the reference for sample preparation, and in the results, we added a statement that the identification is based on the peptide mass fingerprint matching.
Point 5: The authors did not provide relevant references, for example, the microarray datasets were previously published in the "Anderson GG, Moreau-Marquis S, Stanton BA, O'Toole GA. In vitro analysis of tobramycin-treated Pseudomonas aeruginosa biofilms on cystic fibrosis-derived airway epithelial cells. Infect Immun 2008 Apr;76(4):1423-33. PMID: 18212077". Authors did not cite this key paper did not provide a comparative analysis.
Response 5: Duly Noted. As we used the microarray datasets of the study (Anderson GG et al. 2008) deposited in GEO, NCBI, we cited the paper in the methodology. But we are unable to provide a comparative analysis, as the aim of both the studies were different and we explored a different panel of genes, which were not studied in the following publication. Our goal was to do predictive analysis by exploring their published datasets which mimic our experimental condition to understand the up-regulating and down-regulating mechanisms of P. aeruginosa involved in the decreasing kinetics of antibiotic tolerance when exposed at the MICs of tobramycin.
Point 6: Title: The authors have used a published microarray dataset for differential gene expression analysis and referred to this data as "In-silico Transcriptomic Analysis"; this is an incorrect use of the terminology.
Response 6: The term “Transcriptomic analysis” in the title means the study involves omnic data analysis. Further, addition of “In-silico” in the prefix tells the omnic data is analysed from published source. Other suitable terms are “In-silico transcriptome analysis”, “In-silico gene expression analysis” or “In-silico microarray analysis” If this justification is still not convincing, please suggest the suitable term. We will change it accordingly.
Thanks for your time and your valuable inputs. We have noted that you asked for improvement in all the headings. The whole manuscript was revised thoroughly as per the comments, restructured, and added few more relevant references. Please let us know if further changes are required.
Author Response File: Author Response.docx
Round 2
Reviewer 2 Report
I think this statement "MIC of PA14 strain ranges from 5 µg/ml to 8 µg/ml." along with appropriate references should be added into the manuscript
Author Response
Response to Reviewer 2 Comments
Point 1: I think this statement "MIC of PA14 strain ranges from 5 µg/ml to 8 µg/ml." along with appropriate references should be added into the manuscript.
Response 1: Duly noted. Please find statements under the methodology and discussion. In methodology we clearly mentioned that we were interested in the microarray datasets, because it mimicked our experimental condition of P. aeruginosa exposed to MIC of tobramycin. Followed by that, we added “PA14 was exposed to MIC of tobramycin 5 µg/mL” with appropriate reference.
In discussion we added that the published microarray datasets of P. aeruginosa was exposed to MIC of tobramycin and we cited their published study.
Thanks for your time and your valuable inputs. Apart from the changes you suggested, we have made changes as per the comments of Reviewer 3. Kindly look into the changes and give us your inputs. We are looking forward for your report.
Author Response File: Author Response.docx
Reviewer 3 Report
In the methods section, it is not clear what was the rationale for "removing unannotated genes" during the raw data processing.
While using FDR values (adjusted p-values) for the enrichment analysis, the use of p-value cutoff is not informative. Please omit the mention of unadjusted p-values.
It is not clear whether the FDR values were used for determining the significance of the DEGs.
Figure 4. I am not able to locate the volcano plots for two datasets.
Plots should be labeled clearly to represent the respective comparisons.
The results of gene ontology analysis using Panther are generic and does not add any additional information. Authors should consider removing it and keeping the focus on the functional enrichment obtained from David.
Even if the current analysis excludes the genes studied from the previous publication, a rationale must be provided for such exclusions. It is also important that the authors provide discussion about the comparison between microarray results obtained between the current manuscript and Anderson et al., 2008. This should also include the similarities and difference between the two conclusions.
Title:
In silico experiments are generally referred to computational analysis that involves math modeling/simulations. Because this manuscript is about differential gene expression analysis of previously published microarray dataset it doesn’t qualify as an in silico experiment. I suggest the replacing “in silico transciptome analysis” with “microarray or transcriptomic reanalysis”.
“Evaluation of Antibiotic Tolerance in Pseudomonas aeruginosa for Aminoglycosides and its Predicted Gene Regulations Through Transcriptomic Reanalysis”
Also, the use of “Predicted Gene Regulations” should be elaborated in the discussion section.
It is not clear how the target DEGs were selected? Authors should substantiate their criteria for such selections.
minor:
Please fix the remaining typos and formatting errors.
Table 3:
Drop the term “transferases” as it does not clear the FDR cutoff.
Author Response
Response to Reviewer 3 Comments
Point 1: In the methods section, it is not clear what was the rationale for "removing unannotated genes" during the raw data processing.
Response 1: The novel genes (without ID and official gene symbol) and hypothetical proteins were considered as unannotated genes and background corrections were made by removing those genes during raw data processing. Enrichment analysis cannot be performed for novel and hypothetical proteins in the datasets, hence we filtered them during raw data processing.
Point 2: While using FDR values (adjusted p-values) for the enrichment analysis, the use of p-value cutoff is not informative. Please omit the mention of unadjusted p-values. Table 3: Drop the term “transferases” as it does not clear the FDR cutoff.
Response 2: Duly noted. The P-value cutoff is removed from the Table 3 as it is not informative and only FDR cutoff is considered for enrichment analysis. The changes were made in the methodology as well. The “transferase” is removed in the Table 3 as it did not qualify the FDR cutoff.
Point 3: It is not clear whether the FDR values were used for determining the significance of the DEGs.
Response 3: Yes, adjusted p-value cut-off of 0.05 were used to determine the significance of the DEGs. Please find the values added in Table 1 and 2.
Point 4: Figure 4. I am not able to locate the volcano plots for two datasets. Plots should be labeled clearly to represent the respective comparisons.
Response 4: Duly noted. Please find the volcano plots for both the microarray datasets in Figure 4. Plots are labelled as suggested.
Point 5: The results of gene ontology analysis using Panther are generic and does not add any additional information. Authors should consider removing it and keeping the focus on the functional enrichment obtained from David.
Response 5: Duly noted. We considered removing Gene Ontology analysis form Panther. The functional enrichment analysis performed in DAVID was elaborated and we discussed more on our enrichment analysis. We have included representation for our most significant enrichment observed in Bacterial secretion system from KEGG pathways.
Point 6: Also, the use of “Predicted Gene Regulations” should be elaborated in the discussion section.
Response 6: Duly noted. The predicted gene regulations were elaborated in the discussion session as suggested. We further included STRING interaction networks of the regulatory genes mucA and gidB to predict the molecular interactions.
Point 7: Even if the current analysis excludes the genes studied from the previous publication, a rationale must be provided for such exclusions. It is also important that the authors provide discussion about the comparison between microarray results obtained between the current manuscript and Anderson et al., 2008. This should also include the similarities and difference between the two conclusions.
Response 7: Duly noted. Please find the comparisons between Anderson GG et al. study and our study in the discussion. We explored different panel of genes that were suitable for our study to elucidate the mechanism of antibiotic tolerance in P. aeruginosa. Still, we could find similarities in the conclusion, and it is discussed in the discussion.
Point 8: In silico experiments are generally referred to computational analysis that involves math modeling/simulations. Because this manuscript is about differential gene expression analysis of previously published microarray dataset it doesn’t qualify as an in silico experiment. I suggest the replacing “in silico transciptome analysis” with “microarray or transcriptomic reanalysis”.
Response 8: We agree that the term “in silico” was majorly referred to modeling/simulations before the era of omics and other bigdata biology. The term “in silico” refers to computational works in general. Hence, in silico experiments are not limited to only modelling/simulations as per our knowledge. Any computational data analysis or omic analysis still falls under the category of “In silico experiments”. We also referred several dataset analyses studies before framing a suitable title. Please search for terms “in silico transcriptome analysis” and “in silico gene expression analysis” to find published studies.
If this justification is still not convincing, we will change it to “transcriptomic reanalysis”.
Point 9: It is not clear how the target DEGs were selected? Authors should substantiate their criteria for such selections.
Response 9: Duly noted and we added about the criteria for the selection of target DEGs. Our primary goal of exploring datasets is to understand the mechanism that regulates the possible effectiveness of tobramycin treatment in P. aeruginosa. Initially we performed functional enrichment analysis for the DEGs in the datasets. Interestingly, we found that the functional enrichments were suitable for our study to elucidate the mechanism of antibiotic tolerance in P. aeruginosa. Hence, we targeted the DEGs significantly enriched in the functional groups and studied its functional regulations and molecular interactions from the previously published resources.
Point 10: Please fix the remaining typos and formatting errors.
Response 10: Duly noted. We received the whole manuscript carefully and corrected the remaining mistakes. All the gene names were changed to Italic. Use of case letters were fixed. Few conjunctions, vocabulary and grammar were corrected.
Thanks for your time and your valuable inputs. We have noted that you asked for improvement in all the headings. The whole manuscript was revised thoroughly as per the comments, restructured, and added few more relevant information.
Author Response File: Author Response.docx
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.
Round 1
Reviewer 1 Report
The study focus on the kinetics of the adaptive resistance as well as shifts in gene expression for P. aeruginosa when exposed to amikacin, gentamicin, and tobramycin In vitro by optical density of the cells .Yet , the transcriptomic profile of P. aeruginosa exposed antibiotic was evaluated from Gene Expression Omnibus (GEO). It was observed a superior in vitro response of tobramycin comparing to gentamicin and amikacin. Yet, P. aeruginosa exposed to tobramycin by microarray analysis showed low expression of catalytic enzyme 16s rRNA Methyltransferase E, B & L, alginate biosynthesis genes and several proteins of Type 2 Secretory. The paper is well written and contains important information for clinicians and scientists working on antibiotic resistances . My suggestion is to ACCEPT and publish it .
Reviewer 2 Report
The purpose of genetic analysis of the acquisition of resistance to aminoglycoside antibiotics is very interesting.The method of the experiment is reasonable and there seems to be no problem, but unfortunately, there is no novelty in the results obtained in this paper. In addition, the discussion is too general and there are no breakthroughs in the mechanism of resistance acquisition.
Please, describe the apparatus, equipment, and materials used, and clarify the source of supply in material and methods
Figures 1-3, 4-5, and 7-9 would be easier to see if they were combined respectively.
Is there a reason why you dare to use mg/L as the unit of MIC in the head of results?
Please increase the resolution of the figures a little more. Especially fig10, 11, 12.
