Insights into the molecular basis of tick-borne encephalitis from multiplatform metabolomics

Background Tick-borne encephalitis virus (TBEV) is the most prevalent arbovirus, with a tentative estimate of 10,000 to 10,500 infections occurring in Europe and Asia every year. Endemic in Northeast China, tick-borne encephalitis (TBE) is emerging as a major threat to public health, local economies and tourism. The complicated array of host physiological changes has hampered elucidation of the molecular mechanisms underlying the pathogenesis of this disease. Methodology/Principle findings System-level characterization of the serum metabolome and lipidome of adult TBEV patients and a healthy control group was performed using liquid chromatography tandem mass spectrometry. By tracking metabolic and lipid changes during disease progression, crucial physiological changes that coincided with disease stages could be identified. Twenty-eight metabolites were significantly altered in the sera of TBE patients in our metabolomic analysis, and 14 lipids were significantly altered in our lipidomics study. Among these metabolites, alpha-linolenic acid, azelaic acid, D-glutamine, glucose-1-phosphate, L-glutamic acid, and mannose-6-phosphate were altered compared to the control group, and PC(38:7), PC(28:3;1), TAG(52:6), etc. were altered based on lipidomics. Major perturbed metabolic pathways included amino acid metabolism, lipid and oxidative stress metabolism (lipoprotein biosynthesis, arachidonic acid biosynthesis, leukotriene biosynthesis and sphingolipid metabolism), phospholipid metabolism and triglyceride metabolism. These metabolites were significantly perturbed during disease progression, implying their latent utility as prognostic markers. Conclusions/Significance TBEV infection causes distinct temporal changes in the serum metabolome and lipidome, and many metabolites are potentially involved in the acute inflammatory response and immune regulation. Our global analysis revealed anti- and pro-inflammatory processes in the host and changes to the entire metabolic profile. Relationships between metabolites and pathologies were established. This study provides important insight into the pathology of TBE, including its pathology, and lays the foundation for further research into putative markers of TBE disease.


Introduction
Line 77. The sentence is not clear.
Line 96. The hypothesis of the study needs to be stated and how the study design will address this hypothesis.
Lines 97-98. The current methodology and findings do not allow the authors to "provide insight into tick biology and pathogen transmission". The authors need to state the objectives and hypotheses of the study.

Methods.
General comments. Several methodological information regards the metabolomics/lipidomics analysis was missing.
Line 110. They mentioned that the final classification of TBEV was based on clinical test results and symptoms. What were the clinical tests and symptoms considered for the classification? A dot is missing.
Line 111. What were the other criteria used to classify a patient as AP?
Line 135. Were all peaks presented a relative standard deviation (RSD) below 10%? This is unlikely to happen considering that signal intensity of many peaks can drift from day to day and sometimes from run to run. Also, the occurrence of missing values is something that also occurs and contributes to a higher variance in the data. Please, clarify. Were these peaks from the QC injection?
Line 145-146. References for XCMS and Metaboanalyst are required. Which XCMS was used (XCMS online or the XCMS package in R)? Please, mention the version of XCMS and metaboanalyst used. XCMS and Metaboanalyst were used to find the metabolites that differed significantly in TBE patients. Was it XCMS used only to process raw spectra? Was Metaboanalyst used only for statistical analysis? Additional details of data processing as well as of the statistical analysis are required for a better comprehension of the metabolomics/lipidomics analysis. What was the criterion used to filter the features? What was the normalization approach used for metabolomics and lipidomics? Moreover, the authors need to inform which algorithm and parameters were used for peak detection (m/z deviation; chromatographic peak width range; signal-to-noise ratio threshold; etc.). Same for retention-time correction and chromatogram alignment. How were the missing values handled?? All these procedures are crucial to remove bias introduced by LC-MS approach, such as outlier runs or peaks, removal of noise and contaminants, reduction the systematic variation of LC-MS data.
Line 147-148. OPLS-DA is a supervised not an unsupervised method.
Line 151. Please indicate when the student t-test and Mann Whitney were used. Each statistical test works with different assumptions and it is not common to use both tests for the same data. The authors need to establish their questions and assumptions about the data first and then choose the statistical test of interest. Also, it is crucial to correct for multiple comparisons to avoid false positives (e.g., false discovery rate). How the QCs samples were run? How many times was the QC sample injected? Line 158. Which established strategy? Cite the references for this strategy and also explain it. In the text, the authors should mention that the identification obtained was a putative identification since no authentic standards were used.
Line 165. Further details are required to understand how the pathways analysis and the interaction networks were performed. In the pathway analysis, what were the criteria to consider a metabolic pathway as altered?

Results.
General comments. Indicate in the figures and tables which statistical test was used. The amount of features obtained for each LC-MS experiment should also be described in the results as well as the number of features retained after all the filtering process. Be clear in the text that these compounds are putative identified since no authentic standards were used to confirm the structures of the compounds.  Line 183. To determine the flux of a metabolite it is needed the use not only of metabolomics approaches but also computational modeling of pathways. Thus the authors cannot claim that they checked the metabolite flux. Table 2. It seems that authors transformed the abundance values to log2. However this was not mentioned in the methodology. Please clarify. General comments. How the metabolic findings could be potentially linked with immune responses observed in TBEV patients.

Line 242. Metabomic??
Lines 323 -324. The results of the study do not allow such conclusion. The metabolic process that occurs during infectious diseases are too complex to establish a cause/consequence link between the metabolic changes and the disease.