Compounds with Antiviral, Anti-Inflammatory and Anticancer Activity Identified in Wine from Hungary’s Tokaj Region via High Resolution Mass Spectrometry and Bioinformatics Analyses

(1) Background: Wine contains a variety of molecules with potential beneficial effects on human health. Our aim was to examine the wine components with high-resolution mass spectrometry including high-resolution tandem mass spectrometry in two wine types made from grapes with or without the fungus Botrytis cinerea, or “noble rot”. (2) For LC-MS/MS analysis, 12 wine samples (7 without and 5 with noble rotting) from 4 different wineries were used and wine components were identified and quantified. (3) Results: 288 molecules were identified in the wines and the amount of 169 molecules was statistically significantly different between the two wine types. A database search was carried out to find the molecules, which were examined in functional studies so far, with high emphasis on molecules with antiviral, anti-inflammatory and anticancer activities. (4) Conclusions: A comprehensive functional dataset related to identified wine components is also provided highlighting the importance of components with potential health benefits.


Introduction
In recent years, products of grapes have received great interest due to the discovery that several of their components have beneficial health effects on the human metabolism [1][2][3]. While grapes are carbohydrate-rich fruits, their glycemic index is quite low [4]. Furthermore, the polyphenol levels in grapes are relatively high and studies suggest the benefits of the polyphenol content of grapes and Table 1. Antiviral, anticancer and anti-inflammatory roles of representative compounds found in wine. The full list of the compounds with detailed biological role and references are listed in Table S4.

Compound Name
PubChem ID Role, Biological Activity Anti-inflammatory activity, inhibition of NO and proinflammatory cytokine production

Compound Name PubChem ID Role, Biological Activity
Coumarin 323 Coumarin derivatives exert anti-coagulant, anti-tumor, anti-viral, anti-inflammatory and antioxidant effects, as well as anti-microbial and enzyme inhibition properties

Comparative Analysis of "aszú" and "furmint" Wines
Data acquired in positive and negative polarity modes were subjected to principal component analysis ( Figure 1). Regarding the results, the differentiation between the "aszú" and "furmint" samples was successful using the data acquired both in positive ( Figure 1A) and negative ( Figure 1B) polarity modes. Besides the differentiation of the "aszú" and "furmint" wine types, we could also differentiate between the wineries using the data from positive polarity mode experiments (57.3%). In case of the data acquired in negative polarity mode, the differentiation of the wineries was not accurate enough (53%). Data acquired in positive and negative polarity modes were subjected to principal component analysis ( Figure 1). Regarding the results, the differentiation between the "aszú" and "furmint" samples was successful using the data acquired both in positive ( Figure 1A) and negative ( Figure 1B) polarity modes. Besides the differentiation of the "aszú" and "furmint" wine types, we could also differentiate between the wineries using the data from positive polarity mode experiments (57.3%). In case of the data acquired in negative polarity mode, the differentiation of the wineries was not accurate enough (53%).
The acquired data were also subjected to hierarchical cluster analysis and heat maps were generated ( Figure 2). The differentiation between the "aszú" and "furmint" wines was successful using the data registered in both positive ( Figure 2A) and negative ( Figure 2B) polarity mode. The "furmint" samples were clustered by the wineries in case of positive mode, while in negative polarity mode, the clustering was not as accurate as in positive mode. In case of the "aszú" samples, we observed that the clustering by wineries was successful in negative polarity mode, but not in positive mode. Based on the heat maps, several clusters of differentially expressed molecules between the two studied wine types could be identified. The "x" axis shows PC1 while the "y" axis shows PC2. The orange dots represent the "furmint" samples while the blue dots represent the "aszú" samples. The wineries are also indicated with numbering.  The "x" axis shows PC1 while the "y" axis shows PC2. The orange dots represent the "furmint" samples while the blue dots represent the "aszú" samples. The wineries are also indicated with numbering.
The acquired data were also subjected to hierarchical cluster analysis and heat maps were generated ( Figure 2). The differentiation between the "aszú" and "furmint" wines was successful using the data registered in both positive ( Figure 2A) and negative ( Figure 2B) polarity mode. The "furmint" samples were clustered by the wineries in case of positive mode, while in negative polarity mode, the clustering was not as accurate as in positive mode. In case of the "aszú" samples, we observed that the clustering by wineries was successful in negative polarity mode, but not in positive mode. Based on the heat maps, several clusters of differentially expressed molecules between the two studied wine types could be identified. Data acquired in positive and negative polarity modes were subjected to principal component analysis ( Figure 1). Regarding the results, the differentiation between the "aszú" and "furmint" samples was successful using the data acquired both in positive ( Figure 1A) and negative ( Figure 1B) polarity modes. Besides the differentiation of the "aszú" and "furmint" wine types, we could also differentiate between the wineries using the data from positive polarity mode experiments (57.3%). In case of the data acquired in negative polarity mode, the differentiation of the wineries was not accurate enough (53%).
The acquired data were also subjected to hierarchical cluster analysis and heat maps were generated ( Figure 2). The differentiation between the "aszú" and "furmint" wines was successful using the data registered in both positive ( Figure 2A) and negative ( Figure 2B) polarity mode. The "furmint" samples were clustered by the wineries in case of positive mode, while in negative polarity mode, the clustering was not as accurate as in positive mode. In case of the "aszú" samples, we observed that the clustering by wineries was successful in negative polarity mode, but not in positive mode. Based on the heat maps, several clusters of differentially expressed molecules between the two studied wine types could be identified. The "x" axis shows PC1 while the "y" axis shows PC2. The orange dots represent the "furmint" samples while the blue dots represent the "aszú" samples. The wineries are also indicated with numbering.  After fold change analysis, 169 molecules with ±1 log2 fold change and with p < 0.05 between the "aszú" and "furmint" groups were identified ( Figure 3, Table S5).
compounds with lower amount, while the red color shows compounds with higher amount in the comparison of "aszú" and "furmint" samples.
The top 10 molecules with the highest changes in "furmint" and top 10 molecules with the highest changes in "aszú" were further analyzed, and the biological roles of these molecules [20][21][22][23][24][25][26][27][28][29][30][31] are shown in Table 2. Figure 3. Comparative analysis of the identified compounds in "aszú" and "furmint" wines. The "x" axis represents the Log2 fold-change values, while the "y" axis indicates the -Log10 p values. The green box sows the compounds having statistically significantly higher level in "furmint" compared to "aszú", while the red box indicates compounds with statistically significantly higher level in "aszú" compared to "furmint". The top 10 components that showed the highest change in "furmint" and in "aszú", repsepectively, are denoted with numbers (F1-F10 and A1-A10, respectively).  Table 2. Biological roles of the top 10 differentially expressed molecules in "aszú" and the top 10 differentially expressed molecules in "furmint" wines. The name, log2 fold change, adjusted p-value and the biological role is indicated in case of each compound. Negative fold change represents significantly higher level in "furmint", while positive fold change shows significantly higher level in "aszú".  Figure 3. Comparative analysis of the identified compounds in "aszú" and "furmint" wines.
The "x" axis represents the Log2 fold-change values, while the "y" axis indicates the -Log10 p values. The green box sows the compounds having statistically significantly higher level in "furmint" compared to "aszú", while the red box indicates compounds with statistically significantly higher level in "aszú" compared to "furmint". The top 10 components that showed the highest change in "furmint" and in "aszú", repsepectively, are denoted with numbers (F1-F10 and A1-A10, respectively). The top 10 molecules with the highest changes in "furmint" and top 10 molecules with the highest changes in "aszú" were further analyzed, and the biological roles of these molecules [20][21][22][23][24][25][26][27][28][29][30][31] are shown in Table 2. Table 2. Biological roles of the top 10 differentially expressed molecules in "aszú" and the top 10 differentially expressed molecules in "furmint" wines. The name, log2 fold change, adjusted p-value and the biological role is indicated in case of each compound. Negative fold change represents significantly higher level in "furmint", while positive fold change shows significantly higher level in "aszú".  Membrane metalloendopeptidase inhibitor in mouse model [20] Plant metabolite DL-tyrosyl-DL-prolyl-DL-isoleucine −6.

Discussion
Appearing in ancient books, wine has been considered for a long time to be a preparation with health beneficial effects [12]. These effects are mainly attributed to the different phytochemicals present in wine, but there were also studies which demonstrated the beneficial effect of low doses of alcohol [32,33]. However, according to current studies, there are contradictory data regarding the beneficial or harmful effect of alcohol consumption [33]. Considering these data, we do not intend either to promote or discourage alcohol consumption; with this study, we aimed to carry out, with scientific rigorousness, a metabolomics examination of wine components followed by bioinformatics analysis.
The metabolomics analysis of "aszú" and "furmint" wine types identified altogether 288 different components. Although many common compounds were found in the two different wine types, we have identified 169 molecules characteristic either to "aszú" or "furmint". The production of "furmint" and "aszú" wines are different, therefore, even though "aszú" is based on further processing and modification of "furmint" their composition is expected to be different, as well. In this pilot study, we have successfully differentiated the "aszú" wines involving berries having undergone noble rotting from the "furmint" wines based on the identified molecules and their relative quantities.
Phenolic compounds of the wine, including phenolic acids, flavonols, stilbenoids, dihydroflavonols, anthocyanins and flavanol monomers and polymers, influence the color and the taste of the wines. This large group of phenols can be separated into two broad groups, flavonoids and non-flavonoids [34,35]. Our analyses identified many phenolic compounds in the "aszú" and "furmint" wines such as 3-feruloylquinic acid, caffeic acid, caffeic acid 3-sulfate, chlorogenic acid, ethyl caffeate, dimethylcaffeic acid, ethyl gallate, fertaric acid, gentisic acid, rhododendrine and quercetin, and the compound annotation revealed that many identified phenolic compounds have antiviral, anticancer and anti-inflammatory activities. The statistical analysis has not shown statistically significant differences in the levels of caffeic acid, ethyl caffeate, ethyl gallate, fertaric acid, gentisic acid and quercetin between the "aszú" and "furmint" wines. The level of caffeic acid 3-sulfate and rhododendrine was significantly higher in "aszú" wines while the level of 3-feruloylquinic acid and chlorogenic acid was significantly higher in "furmint" wines. The reason behind the different level of several polyphenols in the different wines can be the additional fermentation process of the "aszú" wines and/or the infection of the grape berries with Botrytis cinerea [10].
Further studies are needed to gain more insights into the composition of wines, but our study has shown the power of LC-MS/MS-based metabolomics in wine examination. By applying the mass-spectrometry-based methods able to generate both qualitative and quantitative information, molecular fingerprints of different wines based on their compounds can possibly be created. By the bioinformatics analysis of the biological function of the wine components, we could generate comprehensive lists of the wine components highlighting their antiviral, anticancer and anti-inflammatory properties.
Viruses activate the immune system of the host that can further lead to inflammation [36]. The antiviral activity of the compounds described above mainly consists of the modulation of the immune response and initiating the inflammatory pathways. However, the proper balance between the pro-inflammatory and anti-inflammatory processes is required for the cells and organs to maintain their physiological functions. Thus, strictly regulated pro-and anti-inflammatory pathways are necessary for the homeostasis of the cells and tissues [77]. By database search and literature mining, we could identify 20 molecules in the "aszú" and "furmint" wine samples with anti-inflammatory activity (Table S4). From the 20 identified anti-inflammatory molecules, eight phenolic compounds, rhododendrin, caffeic acid, ethyl caffeate, chlorogenic acid, ethyl gallate, fertaric acid, quercetin and taxifolin were identified. Their anti-inflammatory effects involve the inhibition of toll-like receptor 4 and toll-like receptor 7 mediated signal transduction pathways [78,79], suppression of the NF-κB pathway [80][81][82][83][84][85][86], downregulation of COX-2 expression [87], a decrease in the level of inflammatory cytokines such as IL-1β, IL-6, IL-8, TNF-α and INF-γ [88][89][90], and reduced NO production [91]. In addition to phenolic components, tetrahydroharman-3-carboxylic acid, achalensolide, (E)-p-coumaric acid, zedoarondiol, asperlin, 9S, 13R-12-oxophytodienoic acid, dehydrocostus lactone, eicosapentaenoic acid, indole-3-carbinol, kynurenic acid, melatonin, and umbelliferone were also identified as molecules with anti-inflammatory activity. It is interesting to note that many of the compounds, such as umbelliferone, indole-3-carbinol, and melatonin, have both antiviral and anti-inflammatory activities at the same time, reflecting a complex mechanism of action [92][93][94][95].
The comprehensive collection of the biological functions of the identified wine components can provide a rich dataset to design in vitro and in vivo studies in order to test the beneficial effects of the different compounds. The data generated in this study can be used to design targeted examinations.
Validation of the wine components with health beneficial effects can provide high quality wines as functional food in the future. This can be especially important in case of many polyphenols such as quercetin, chlorogenic acid, caffeic acid, etc., which are insoluble in water but are soluble in alcoholic solutions. The alcohol content of wine helps the solvation of water insoluble polyphenols making wine a complex mixture of both water soluble and insoluble compounds such as phenolic compounds, acids, lipids, amino acids and other biologically active molecules [7]. The rich composition and the identified molecules with beneficial health effects highlight the potential of wine as a functional food [12].

LC-MS Analysis
Prior to mass spectrometry analysis the components of the wines were separated using a Transcend II TLX-1 UPLC system (Thermo Scientific, Palo Alto, CA, USA) in LX mode. Chromatographic separations were performed on a TFS Hypersil gold reverse phase analytical column (50 × 2.1 mm, 1.9 µm particle size, 175 Å pore size, Thermo Scientific) using a 5 min water/acetonitrile gradient. The first step of the separation was a 25 s equilibration with 100% buffer A followed by the increase in solvent B to 30% during 5 s. Solvent B was further increased to 50% in 70 s and then increased to 95% in 60 s followed by a 60 s hold-on. In the last steps, the solvent condition was changed to 100% A in 30 s followed by the equilibration of the system with 100% solvent A. The flow rate was set to 0.8 mL/min. Solvent A was 0.1% formic acid in LC grade water (VWR Ltd., Radnor, PA, USA) and solvent B was 0.1% formic acid in LC grade acetonitrile (VWR Ltd.). The 100 µL samples were injected in duplicates.
Mass spectrometry analyses were performed on an Orbitrap Fusion tribrid mass spectrometer (Thermo Scientific) using data-dependent acquisition. Survey scans were taken in the Orbitrap mass analyzer with 120,000 mass resolution scanning a 100-1000 m/z range in profile mode. The AGC target was set to 4.0 × 10 5 . MS/MS spectra also were acquired in the Orbitrap mass analyzer by the fragmentation of the selected parent ions using HCD dissociation with 40% collision energy. However, MS/MS spectra were recorded in centroid mode at resolution and AGC target set to 50,000 and 5.0 × 10 4 , respectively. The cycle time of the analyses was 0.6 s. Spectra were acquired in both positive and negative polarity modes. The mass spectrometry data are available at the MassIVE database (ftp://MSV000085599@massive.ucsd.edu).

Data Analysis and Compound Identification
The acquired data were subjected to metabolite identification using the Compound Discoverer 3.1 software (Thermo Scientific). Both positive and negative polarity mode data were loaded for the analysis. For compound detection, the mass tolerance was set to 5 ppm, the intensity tolerance was 30%, the signal/noise ratio threshold was set to 3 and the minimum peak intensity was set to 100,000 cps. For the detection of compounds, {M+H] + , [M+Na] + , [M+K] + and [M-H]ions were used. The grouping of the compounds was done with 5 ppm mass tolerance and 0.2 min retention time tolerance. The acquired MS2 spectra were searched against three different databases implemented into the Compound Discoverer for compound identification. The first database was the m/z cloud [130], the autoprocessed and reference libraries were used for the identification of small molecules. Besides the m/z cloud, the Chemspider database [131] was also used for identification. In the Chemspider, we used the Carotenoids database [132], FooDB [133], KEGG [134], LipidMAPS [135], Peptides [136], Phenol explorer [137], Plantcyc database [138] and the Yeast metabolome database [139]. The mass list library of the Compound Discoverer software was also used for compound identification. From the library, the Flavonoid structure database and the Endogenous metabolites database was selected for the search. The spectra of the identified molecules were analyzed manually and the hits were curated using the FISh scoring algorithm. For FISh scoring, the high mass accuracy tolerance was set to 2.5 Da while the low accuracy mass tolerance was 0.5 Da and the signal/noise threshold ratio was 3. The best hits were selected as compound annotations. Besides the identification, the comparative analysis of "aszú" and "furmint" wine types was also performed with the help of the software.
The peak areas were normalized by the software using constant mean normalization. For relative quantification the mean normalized peak areas were compared between the two studied groups. Principal component analysis implemented into the Compound Discoverer software was used considering the normalized peak areas. For heat map construction, normalized peak areas were used and the scaling was done before the clustering. The distance function was set to Euclidean and the linkage method was set to complete. For quantitative comparison of the "aszú" and "furmint" samples molecules with ±1 log2 fold change were accepted, and the significance threshold was set to p < 0.05.

Annotation of the Identified Compounds
In order to obtain information about the biological and pharmacological properties of the identified compounds, an in silico approach was implemented utilizing several in-house developed bash scripts to access ChemSpider, PubChem [140,141], and PubMed [142] databases via their programmatic web services. Based on the chemical names, first the ChemSpiderIDs, and SMILES identifiers were retrieved. SMILES strings were then included in PUG-REST web service [143] requests to access PubChem's BioAssays [144] records and PubChem compound identifiers (CID) of each identified molecules. The retrieved CIDs were used to collect annotation data for each compound in the PubChem PUG-View interface [145]. CIDs were also used to collect literature metadata from PubMed by searching for PubMed IDs linked to the retrieved CIDs. The listed article titles, abstracts and BioAssays records were screened for specific keywords related to anti-inflammatory, antiviral and anticancer effects.
Supplementary Materials: The following are available online at http://www.mdpi.com/1422-0067/21/24/9547/s1, Table S1: List of the compounds identified in "aszú" and "furmint" wines. The novel and previously annotated components identified in wine are highlighted in bold and italics, while the novel compounds without annotation are highlighted in bold, Table S2: Compound annotation based on the PubChem BioAssay database, Table S3: Compound annotation based on the PUG-View database, Table S4: Identified wine compounds with antiviral, anti-inflammatory and anticancer activity based on the BioAssay database and literature mining, Table S5: Compounds with significantly different levels in "aszú" and "furmint" wines, Table S6: Routine chemical analysis of the examined wines according to Winsscan FTIR analysis (Foss Analytical A/S-HillerØd, Denmark).

Acknowledgments:
The technical help of Renáta Kovács and Andrea Guba is greatly acknowledged. We thank Károly Vékey for reading and critically reviewing the manuscript.

Conflicts of Interest:
The authors declare no conflict of interest.