Skin Mucus as a Relevant Low-Invasive Biological Matrix for the Measurement of an Acute Stress Response in Rainbow Trout (Oncorhynchus mykiss)

Round 1

Reviewer 1 Report

Dear Authors,

your manuscript is clear and well written and deals with a topic of great interest. I was a pleasure for me read it. I think it can be accepted after some minor changes:

Line 21: put the scientific name in italics.

Line 54: after welfare, quote Fuochi et al., 2017 please. DOI: 10.3390/md15110342 in Marine Drugs

Author Response

Answers to the Comments of reviewer 1 on MS Water-1655158

Title: Skin mucus as a relevant low-invasive biological matrix for the measurement of an acute stress response in rainbow trout (Oncorhynchus mykiss)

REVIEWER 1

Comments and Suggestions for Authors

Dear Authors,

your manuscript is clear and well written and deals with a topic of great interest. I was a pleasure for me read it. I think it can be accepted after some minor changes.

Comments reviewer 1

Comment 1: Line 21: put the scientific name in italics.

Answer to comment 1: Done.

Comment 2: Line 54: after welfare, quote Fuochi et al., 2017 please. DOI: 10.3390/md15110342 in Marine Drugs

Answer to comment 2: Done.

Author Response File: Author Response.docx

Reviewer 2 Report

See attached file

Comments for author File: Comments.docx

Author Response

REVIEWER 2

Comment 1:

Comment 1.1. The authors try to demonstrate that skin mucus can be used as an alternative for blood plasma for the analysis of stress in fish, not only for cortisol and plasma glucose, but also for many other blood plasma parameters (biomarkers). They conclude that most biomarkers showed similar behaviour in mucus and plasma after air exposure as stressor. Their results do not support this conclusion: 12 of the 15 biomarkers measured in plasma as well as mucus do not show such a correlation. A more than low correlation is only demonstrated for cortisol, glucose and protein (for cortisol and glucose this is already known for several species). The potential key biomarkers mentioned in the main conclusion should be specified. The statistical treatment of the data is raising questions.

Answer to comment 1.1: Taking into account the comments of the reviewer 2 the conclusions were rewriten. Also following the reviewer’s comment that the correlation analysis was not useful and was not providing interesting information for the results, we decided to eliminate the correlation analysis from the article. However, we do consider that methods and statistical analyses were correctly performed. It is important to point out, that if each statistical analysis that was performed in this study had been performed independently, it would have a limited value. However, all together they contribute to obtaining additional information and lead to the conclusions made. Nevertheless, as we state above, we have eliminated some of the statistical analysis performed, following the reviewer recommendations.

Detailed major comments:

Comment 2: For demonstrating that mucus can be used as an alternative for plasma the authors apparently assume that it is sufficient to demonstrate a correlation with cortisol in the values of the biomarkers. This is not correct. Such a correlation is a prerequisite only. A second prerequisite is that biomarkers that show a significant respons to a stressor in the blood plasma do also show a significant response to the stressor in the mucus. Only cortisol, glucose and cholesterol comply with both criteria. One cannot recommend the use of mucus as an alternative for plasma when most parameters measured in mucus do not show significant differences after an acute stressor. The results of Spearman’s test and principal component analysis are too general and are not linked to the main question of the study.

Answer to comment 2: Could it be that there was some misunderstanding? The aim of the study was to evaluate the possible utility of skin mucus as an alternative matrix to plasma for stress studies. Therefore, several approaches were adopted: 1) fish were subjected to a standard stressor and data were obtained at different time points; 2) correlation with cortisol was performed to evaluate if there was a relation with this widely recognized stress biomarker, being widely known the role cortisol plays as a stress biomarker in welfare and stress studies (please see more detailed explanaition regarding correlations “Answer to comment 8”); 3) PCA was performed precisely with the idea to identify biomarkers other than cortisol that contribute to the stress response in fish and evaluate the role of skin mucus biomarkers, but not in an individual manner. We consider that all methods and statistical analyses were correctly performed. It is important to point out, that if each statistical analysis performed in this study had been performed independently, it would have a limited value. However, by analyzing them all together, they contribute to obtaining additional information and lead to the presented conclusions. Nevertheless, following the reviewer’s recommendations we decided to eliminate the correlation analysis. Furthermore, if the editor and reviewer considers that the PCA analyses are also unnecessary, we will take this information out of the manuscript.

Comment 3: The primary results should have been analysed more critically, for instance:

Cholesterol is one of the biomarkers that is considered to be positively correlated with cortisol levels in plasma and mucus (line 267). However, the cholesterol data of the plasma show a minimum after 24 h. In the mucus a minimum is observed already after 6h and a recovery after 24 h. The authors do not show a recovery in the plasma, since their observations stopped after 24 h. Thus, it cannot be concluded that the time curve of the cholesterol concentrations in the mucus followed that of cholesterol in the plasma. The comment (line 267) that cholesterol measured in both matrices correlates positively with the cortisol levels has no biological significance because of lack of data. Moreover, if the mucus value would be a reflection of the plasma values, one would expect first a reduction in the plasma, followed by a reduction in the mucus.

Answer to comment 3: We agree with the reviewer that, since our observations stopped after 24 h (which is a common time-point in the literature), we cannot conclude if the curves are similar but with a delay between plasma and skin mucus. However, in this study, when we refer that an analyte is correlated between plasma and skin mucus (or with cortisol in both matrices) we refer only that we observe a correlation with statistical significance (p<0.05), which is an objective fact. However, to avoid possible confusion, we have included the following sentence in the manuscript: “As commonly performed in literature, our observations stopped 24 h after the stressor episode. Thus, we cannot evaluate if there is a delay in some biomarkers recovery in skin mucus with respect to plasma, and focused on the possible correlations presented between both matrices at the same time points. Therefore, further studies are desirable to discern the possible biological significance of our findings.” Lines 271-277.

Comment 4: The reverse is observed. Similar criticism applies to the conclusions with respect to many other biomarkers, for instance creatinine (line 259), triglycerides (line 267), AST (line 254; this parameter shows an increase in plasma and a reduction in mucus.

Answer to comment 4: Please, see response to the previous comment.

Comment 5: For EA the authors conclude that it “appeared unaltered for all conditions, although in skin mucus EA was positively correlated with cortisol”. This reasoning is difficult to follow in the absence of convincing data.

Answer to comment 5: As stated in comment #3, it is true that data from EA was positively correlated with cortisol in skin mucus (r=0.563, p<0.001). However, it is not the objective of this study to find if this and other correlations are of biological significance. Instead, we aimed to discern if a battery of biomarkers could be employed to detect fish subjected to an air-exposure stressor (another stressor could run into different results) in both plasma and skin mucus as biological matrices.

Comment 6: For any study, and in particular for a study on methodology, it is important to present data on the specificity, sensitivity and reproducibility of the assays involved. No information is presented. The concentrations of many of the biomarkers show marked differences in the time curves between plasma and mucus. In the absence of information on specificity, reliability and sensitivity of the assays such information is difficult to interprete.

Answer to comment 6: Summarized validation data was included in the new version of the mansucript: “All biomarkers were previously assessed for linearity and intra-assay precision in fish plasma and skin mucus as described elsewhere (Franco-Martínez et al. 2019). For all biomarkers, intra-assay coefficients of variation were below the 15%, and values from ½ dilution with ultrapure (MilliPore) water showed less than 15% variation in comparison to the expected results. We also added information on Page 4, line 152, indicating that all the samples were measured in one batch to avoid possible inter-day variations. All the measurements were performed within a month after sample obtaining (Page 3, lines 141-148).”

Comment 7: The standard deviations of most of the means of the concentrations are very high and the causes (biological, methodological or both) are unknown. These comments in particular apply to mucus: a highly complicated fluid.

Answer to comment 7: Yes, agree that the reasons of the high standard deviations in skin mucus are still unknown, and deserve further research. However, this is also observed in other biofluids such as saliva, the use of which is increasing everyday, regardless of these complications.

Comment 8: Table 1: The Spearman correlation test is a test for two variables. The data on biomarker levels in mucus and plasma contain a third variable: time (0, 1, 6 and 24 h). Thus, the application of a simple Spearman seems not correct. Since the authors state that all data have been included, the value of the correlation data can be disputed, the more so since most of the data vary in time in a non-linear fashion. What exactly can be concluded from these correlation coefficients? The use of r normally indicates Pearson’s r; should it read Spearman’s  rho (or rs)? Or has Pearson’s test been applied? How have outliers been treated? In general the range of the values is wide, as indicated by the confidence intervals. A wide range normally tends to lead to a higher correlation than a small range. In this respect it is worth mentioning that the P values presented in Table 2 do not refer to the statistical significance of the correlations. A critical discussion of the results is missing. Any correlation determined requires a specific interpretation, including its biological significance. In general the correlations are low to very low, with only few that can be characterised as moderate (the ones printed in bold). Therefore, general conclusions such as “mucus is a low-invasive alternative matrix for stress assessment pointing to potential key biomarkers” are premature and inadequate.

Answer to comment 8: Yes, we agree with the reviewer that Spearman is a test to compare two variables. But this is what we use it for. Spearman’s correlation test was performed in order to compare data obtained in mucus with data obtained in plasma of the same individuals, with the aim to see if higher values of the biomarker in one biofluid corresponded to higher values of the same biomarker in the other biofluid. With regard to the different time points, this approach was evaluated using a one-way analysis of variance (one-way ANOVA). These data and the outcome of the statistical analysis are presented in Figure 1.

Regarding the relatively low correlations in some biomarkers, it is important to understand that, although there is no correlation between plasma and mucus, the analyte can still be a biomarker of stress. The low correlations could be due to the delayed transfer of some analytes from organism to mucus, the analyte could be subjected to a different transport out of the organism, and even to local production in peripheral tissues. Furthermore, the lack of correlation between target biomarker and cortisol could also be due to their involvement in different pathways. However, by doing correlations we obtain more information about the behaviour of the biomarkers in different tissues and in different situations. Finally, regarding the general conclusion, as mentioned before, having or not a perfect correlation, some target analytes can be acknowledged as a good biomarker of some conditions, and in our case – stress. And this is not just because of the correlation, this is because of the shreds of evidence obtained using three major statistical approaches – anova, correlation, PCA. However, as we already state in the responses to the other comments, following the suggestions of the reviewer 2, that considers that the correlation analysis was not useful in this particular study, we decided to eliminate all that was related to correlations from the manuscript.

Comment 9: The application of principal component analysis (PCA) also raises questions. The information presented is rather poor, also with respect to the data set. That all data were included does not point to a critical data analysis. Was Bartlett’s test applied and what was the result?  No conclusions are drawn on the basis of the results of this analysis. The biological significance of the observed separation is not discussed and its relevance in relation to the main hypothesis of this study remains unclear.

Answer to comment 9: As described in the Material and Method section, the PCA was performed with the online open-access programme Metaboanalyst 5.0 (MetaboAnalyst),  which is highly accepted by the scientific community. However, as far as we know, this programme does not permit to perform the Bartletts test. Nevertheless, based on PCA analysis in our study, we highlighted that when all data (obtained from mucus and plasma) were assessed the groups of stressed and non-stressed fishes were totally separated and when data from plasma alone were used, the plots slightly overlapped. For this reason, on Page 11, line 342, we indicated that “These data (…) spot-out the drawbacks of just determining plasma analytes. In addition, it highlights the usefulness of skin mucus as a complementary matrix.” Furthermore, biomarkers contributing the most to the differentiation of groups were also stated (Page 10, line 333). Based on this the conclusion that stress has a generalised negative effect in fish affecting multiple systems was drawn (Page 11, line 342).

Comment 10: The stress terminology is frequently used incorrectly. In particular the use of the term post-stress should be avoided for the period in which the fishes were clearly stressed (as indicated by the cortisol values), and be replaced by post stressor. Exposure to acute stress should read: to an acute stressor.

Answer to comment 10: These alterations were done as suggested by the reviewer.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

see attached file

Comments for author File: Comments.pdf

Author Response

Dear Editor and Reviewer 2,

First of all, we want to thank you for giving us the opportunity to resubmit our revised article. Please find attached answers to the comments of reviewer 2. We also want to resubmit later on the revised and reworked version of the manuscript. We followed the changes recommended by reviewer 2. However, we do not agree with some of the comments suggested by the reviewer, mainly with those comments related with the statistical analysis. Therefore, we decided to keep the structure of the article and the analysis performed, and we hope that the editor will consider our opinion. We also did the changes that we considered necessary in the manuscript taking into account the similarity index report.

Please see the answers to the reviewer comments in the attached file.

Thanking you.

Yours sincerely,

Mariana Teles

Author Response File: Author Response.docx