Untargeted Metabolomics Discriminates Grapes and Wines from Two Syrah Vineyards Located in the Same Wine Region

Round 1

Reviewer 1 Report

This paper investigates from a metabolomic point of view the differences in grapes and wines from two Syrah vineyards located in the same wine Demarcated Region of Lisbon. That with the aim to understand how terroir influences grape and wine metabolite profiles as well as sensory parameters. Ultra-High resolution Liquid Chromatography coupled with tandem Mass Spectrometry (UHPLC-Q-ToF-MS/MS) was applied to identify grape and wine metabolites. With the use of untargeted metabolomics the anthocyanin, chrysanthemin, was identified as a putative biomarker for terroir 1, the wine of which was better scored from the overall sensorial analysis.

The paper is interesting and falls within the scope of the journal. The paper is well-structured and the main aspects to characterize wines were addressed. However, some methodological mistakes makes the Ms. weak in some parts. In particular, metabolite quantification was not performed. All samples were referred to Intensities (arbitrary units?). In this way, comparisons between berry and wine was unreliable. In fact, metabolites from berries should be expressed on weight basis (calculated considering volume obtained from 1 g of berries) whereas metabolites from other samples should be better expressed per L. For this reason, data reported in Figs. 2-4 should be revised and, in particular, data shown in Fig. 4 recalculated making thus reliable the comparisons.

Minor:

-        Table 1: is “Initial 2 (S2.1)” correct?

Author Response

Thanks for the comment. In fact, what is shown in figure 2 and 3 is the evolution of the compounds during the fermentation process. We think that there is no need of making calculations of concentrations in the must because the analysis was performed using always the same injected volume. Therefore, what is done is a relative  comparison among all of them. The samples were withdrawn exactly from the same amount of juice or must  and The dilution done to all of them was the same and the volume injected in the apparatus also the same. The comparison of wine and berries was carried out using the same volumes and the same dilution, as grape samples were first crushed and the homogenized juice used for UHPLC/Q-ToF-MS/MS, injecting the same amount of each sample. Once again, the comparisons are relative  to each other.

A sentence reinforcing this situation was added to “Material and Methods” section: «All samples, including the grape juice, the fermentation must and the wine  were first diluted in MilliQ H₂O, in a proportion of 1:2. A volume of 5 μL of all samples was injected (auto injector)…». A new sentence was also added reinforcing this point: Data directly obtained from UHPLC/Q-ToF-MS/MS was used for comparison between samples.

 Minor:

-        Table 1: is “Initial 2 (S2.1)” correct?

It is now corrected to Initial 2 (S2.2).

Author Response File: Author Response.pdf

Reviewer 2 Report

This manuscript relies upon specific grape microbiome and metabolome as a means to differentiate two vineyard sites based on terroir. The primary concern is that only one year’s worth of data were collected and sample sizes (e.g., grapes for fermentation) were not only small but may or may not represent diversity with a given vineyard site. 

Due to these concerns, it is recommended that this manuscript should not be published in its current form. With a large amount of modification, it is possible that portions could be acceptable for the Journal. Would a different journal be more applicable? Specific comments are as follows:

Line            Comment

25-26         If only one year’s worth of data were collected, can these inferences be made? It seems like a shift in climate/weather would also greatly impact scores and such. And, would independent judges all score these wines the same way; wines from site 1 were better scored than those from site 2? Another example is 2-keto-D-glucuronic acid which is thought to be a microbial by-product. What if there was an microbial infection on one site but not the other? Wouldn’t that skew the results?

72-77         Sample size appears small. What about sample variation within each vineyard site? Not knowing the size of the vineyards made establishment of far reaching conclusions difficult. Were these vines the same age, grown under the same conditions, same clone, etc.? If the grape vines were different in any way, those differences could also account for "terror" differences. Another source of error would be fermentation volume where 50 to 60 liters is very different from commercial winemaking involving thousands of liters. Variation due to the small sample and fermentation size could result in very different inferences being made.

121            What is an expert panel, winemakers? Were these judges initially trained?

177            Unclear how a replicate was actually defined? If only two replicates (refer to line 78), it is unclear whether enough data could be collected to make inferences regarding the differences between site 1 and site 2.

224            What about clonal differences between cultivars?

270            It is unclear how strong this science was to distinguish the two terroirs given small sample sizes, unclear vineyard diversity, and some marker molecules originating from microbial sources.

Author Response

We understand the concern given by the referee, but we would like to point out that our biological replicates were collected per terroir in order to represent diversity within the vineyard. This diversity can be easily distinguished in the PCA analyses. In this work, we preferred to document sequential stages of wine production.

Taking into consideration the comments of all reviewers, we have considerably improved the manuscript, namely, clarifying some issues, better explaining the data and improving the results discussion.

New text highlighted in yellow in the manuscript.

  Answer to specific comments are as follows:   Line 25-26: 

This is an exploratory study and instead of more than one year of analysis we have preferred to present the metabolites evolution in different stages of the wine production, starting with the grapes. This work allowed us to obtain indicators that may be used in future studies on wine quality in this region. As will be clarified ahead (answer to line 121) the panel was composed by independent expert tasters.

No visual symptoms of infections were noticed in the grapes from both sites. Additionally, gluconic acid, an infection marker, was not detected. The presence of 2-keto-D-glucuronic acid can be due to functional microbiome, it does not need to mean infection.

Line 72-77: 

For sample collection the all vineyard was considered in both sites. Grapes were randomly collected and used either for grape analysis as well as for microvinifications. Although in literature there are studies using laboratory scale volumes, in our case assays were carried out at an experimental station and the volumes used are the most common in experimental assays, and replicates were performed in order to produce stronger results. From each site, four biological replicate samples for grape analysis were prepared by collecting 50 berries each from different bunches. The text was clarified in the manuscript.

Line 121:

For the sensory analysis the expert panel is constituted by trained tasters which belong to the INIAV-Dois Portos staff, namely researchers, technicians, project fellows. This panel is highly trained with different, experimental and commercial, red and white wines as well as with different type of sensory evaluation methodologies.

The text was changed in order to clarify that point: The sensory analysis was carried out by INIAV-Dois Portos expert panel, using nine highly trained and independent wine tasters, in a standardized tasting room with individual white boots.

Lines 177:

We understand the text was a little bit confusing, therefore the manuscript was modified in order to answer the question (lines 75-85) Two biological replicates of microvinifications were performed from the grapes randomly collected from all the vineyards of each site. For grape samples four biological replicates were used in order to get higher diversity, due to a lower sample size.

Line 224:

The vineyards were not of a unique clone. The vineyards on site 1 and 2 have similar characteristics using polyclonal material, which is considered standard material.

Line 270:

These concerns were already clarified above. Most of all, chrysanthemin, pointed out as a putative marker for site, is not a microbial originating molecule. 

Author Response File: Author Response.pdf

Reviewer 3 Report

The manuscript in reference describes the metabolic profiling-based comparative study of grapes, must, and wine from the Syrah grape variety cultivated in two sites from the same region using (UHPLC-Q-ToF-MS/MS) and statistical analysis. The manuscript is interesting and includes relevant information for readers. However, some points should be addressed before being considered further.

1.       Detailed scrutiny should be performed throughout the manuscript to revise some grammar, stylistic, and even typos issues.

2.       Lines 60-69: These two paragraphs can be merged since they share similar information.

3.       Lines 75-76: Improve the exact meaning of “biological replicate” since it is something confusing in the manuscript. I mean, the authors mentioned in the manuscript that 50 berries from different grape bunches per location were collected, but it is challenging to understand if these 50 berries constituted the biological replicate or if these 50 berries were subdivided into four replicates. In addition, other collection information is missing. For instance, how many bunches were used to collect the 50 berries? How many berries per bunch were collected? How many grape plants were employed for such a collection?

4.       Line 81 and Table 1: The authors should specify in the manuscript why only two replicates were used for initial fermentation. This seems to have no sense since four biological replicates were collected. Availability for microvinifications? Additionally, it is unclear how four replicates were used in the next mid and end fermentation. Do you mean technical replicates? I suggest that the authors revise this paragraph since it is highly confusing.

5.       Line 114: The authors should report the confidence level of compound identification by HRMS. It would be best if you used the MSI recommendations to communicate the confidence level for the metabolite identification. This level should be added in table 3 as a column (if the confidence level is different per metabolite) or in the caption if the same for all identified compounds.

6.       Lines 148-154 and Figure 1: The cumulative R2 and Q2 values per PCA-based comparison should be informed to provide the explanation level of the variance.

7.       Figures 2-4 and related information in the text: The comparisons based on MS-based signal intensity can be valid if a quality control to correct MSD response variations is used. However, no quality control is informed. In this sense, such a comparison is not totally reliable since it depends so much on the sample preparation, chromatographic analysis, ionization, MSD response, etc. In this regard, the authors should add an adequate clarification/explanation, supported by references, of the actual scope of such a comparison to avoid confusing interpretation by readers.

8.       Table 4: Why was ANOVA used if the authors did not mention verifying the normal distribution of data? If yes, the test to verify normality should be informed. If not, the authors should verify that and, depending on the results, use the proper parametric or non-parametric test. In addition, the authors should inform the test used for the multiple comparisons to get the different groups defined by the lowercase letters. The table must be self-explanatory, and the caption information is crucial for that.

9.       R&D section: This section is highly descriptive, but a proper discussion is missing. The authors should compare the results with theory and literature.

 

10.   Conclusions must be improved since they are related to general information or summarize results. The authors should define conclusions from conceptual perspectives and mechanistic point-of-view.

Author Response

A thorough revision of the manuscript was performed. Taking into consideration the comments of all reviewers, we have considerably improved the manuscript, namely, clarifying some issues, better explaining the data and improving the results discussion. New text is highlighted in yellow in the manuscript.

2. Lines 60-69: These two paragraphs can be merged since they share similar information.

Answer: These two paragraphs were merged as suggested.

3. Lines 75-76:

Answer: The text was a little bit confusing and the manuscript was modified in order to answer the question (lines 75-80). Grapes were randomly collected in all vineyard of each site. Grape sample biological replicates were obtained by collecting 50 berries each from different grape bunches from each site. Four biological replicates were used in order to get higher diversity, due to lower sample size.

4. Line 81 and Table 1:

Answer: The text was clarified (lines 80-85) and the table corrected. Two microvinification biological replicates were performed and initial, mid and end fermentation samples were collected in duplicate from each replicate.

5. Line 114:

Answer: Although rules for confidence level in compounds identification started to be proposed [for instance, J. Am. Soc. Mass Spectrom. (2017) 28:709Y723. DOI: 10.1007/s13361-016-1556-0] in the present work they were not used because all the identifications were based on the chemical formula proposed by the DataAnalysis program, the chemical structure searched for in Pubchem, MetLin, and HMDB, and checked using MS2 fragmentation using the program MassFrag from Bruker. To certify the chrysanthemin proposed as an hypothetical biomarker, standard was used and again the MS2 of the standard was compared with the fragmentation obtained for the proposed compound in the chromatogram of the sample. The name given to each compound was checked with fragmentation, drawn with the program MassFrag and the obtained fragments with the equipment were possible to be drawn.  A sentence was added saying precisely how this identification was carried out: For this identification, the procedure indicted in Material and Method section was followed that is, molecular mass was used by the DataAnalysis program to propose several chemical formula, each with an error of estimation and with two or three proposals, according to the lowest error, the databases were used to have a chemical structure. These structures were drawn employing MassFrag, and the MS²indicated in the chromatograms were used to check each chemical structure proposed. Salicylic acid was applied as a standard to control the methodology in use and chrysanthemin was also used as a standard to verify the hypothetical biomarker that is proposed in this work.

6. Lines 148-154 and Figure 1:

Answer: the statistic was added to Fig 1 legend

7. Figures 2-4 and related information in the text:

Answer: Yes of course no concentration values can be inferred from a detector used in this work, nevertheless the comparisons are done for the same compound, that is x and y in the begging of the fermentation and the same compound at the end, taking into consideration that all had the same dilution and injected the same amount. So, we think that these comparisons can be accomplished giving an idea of the metabolites´evolution during the process.

8. Table 4:

Answer: In fact the ANOVA description relating sensory analysis is lacking some important information.Bartlett's, Cochran's, and Hartley's tests were used in order to test the homogeneity of the variances, the corresponding bibliographic reference was added. The least significant difference (LSD) was applied for the comparison of the different averages, using the Fisher LSD test. Table 4 was improved.

The text was modified in order to better explain what was done (lines 140-146): One-way ANOVA with alpha=0.05 (significance level) and alpha=0.95 (confidence limits) was performed to the sensory results of each replicate wine in order to evaluate the effect of the site factor. Bartlett's, Cochran's, and Hartley's Tests were used in order to test the homogeneity of the variances [16]. When the effect of site factor was detected the calculation of least significant difference (LSD), using Fisher LSD test, with alpha=0.05, was applied for the comparison of the different averages [17].

9. R&D section:

Answer: The R&D section text was improved with a deeper discussion of the data based on literature. New text highlighted in yellow in the manuscript.

10. Conclusions must be improved since they are related to general information or summarize results. The authors should define conclusions from conceptual perspectives and mechanistic point-of-view.

Answer: The conclusions were improved accordingly. New text highlighted in yellow in the manuscript.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

With the present version of the Ms. Authors clarified my doubts and improved the paper.

Reviewer 2 Report

This study has a difficult experimental design which does not allow for preparation of strong conclusions. In their response to the initial review, the Authors stated, "We understand the concern given by the referee, but we would like to point out that our biological replicates were collected per terroir in order to represent diversity within the vineyard. This diversity can be easily distinguished in the PCA analyses." What does the first sentence actually mean? With only four samples taken from a vineyard, how were the authors assured that that sample size was, in fact, representative of the vineyard? Diversity can be distinguished through PCA analysis but with such a low number of samples from a vineyard, how are the authors sure of the actual diversity and therefore their conclusions? What were the variations within each vineyard including soils, irrigation, elevation, slopes, etc.? How was a sampling scheme of 50 berries per four vines per vineyard determined? Does this imply that a sample size of just 200 berries represents an entire vineyard of unknown acreage? Fermenting only duplicate samples and not providing any information about the composition of the grapes at harvest again hampers formulation of conclusions. Why choose those concentrations of total SO2 which is normally based on the chemical composition of the grapes (i.e., pH)? 

Furthermore, it was unclear what what the Authors meant by, "This is an exploratory study and instead of more than one year of analysis ...". An "exploratory" research study usually sets up a much larger, more exhaustive study because certain parameters needed to be evaluated first. For instance, a far stronger experimental design which would be "exploratory" would have focused on multiple years for a single vineyard to determine that impacts of terroir (climate, in particular) towards year to year variation. Using that data, the Authors could then apply the same experimental design to different vineyards which would allow preparation of stronger conclusions as part of a more intensive study.

This study would have been strengthened by consultation with a viticultural expert working with the target vineyards to develop a stronger experimental design. Without a strong design, the remaining portions of the research would be questionable.