A Combined Metabolomic and Metagenomic Approach to Discriminate Raw Milk for the Production of Hard Cheese

Using Targeted Metabolomics to Unravel Phenolic Metabolites of Plant Origin in Animal Milk

Milk holds a high nutritional value and is associated with diverse health benefits. The understanding of its composition of (poly)phenolic metabolites is limited, which necessitates a comprehensive evaluation of the subject. This study aimed at analyzing the (poly)phenolic profile of commercial milk samples from cows and goats and investigating their sterilization treatments, fat content, and lactose content. Fingerprinting of phenolic metabolites was achieved by using ultra-high-performance liquid chromatography coupled with triple-quadrupole mass spectrometry (UHPLC-QqQ-MS/MS). Two hundred and three potential microbial and phase II metabolites of the main dietary (poly)phenols were targeted. Twenty-five metabolites were identified, revealing a diverse array of phenolic metabolites in milk, including isoflavones and their microbial catabolites equol and O-desmethylangolensin, phenyl-γ-valerolactones (flavan-3-ol microbial catabolites), enterolignans, urolithins (ellagitannin microbial catabolites), benzene diols, and hippuric acid derivates. Goat's milk contained higher concentrations of these metabolites than cow's milk, while the sterilization process and milk composition (fat and lactose content) had minimal impact on the metabolite profiles. Thus, the consumption of goat's milk might serve as a potential means to supplement bioactive phenolic metabolites, especially in individuals with limited production capacity. However, further research is needed to elucidate the potential health effects of milk-derived phenolics.

Multi-omics in food safety and authenticity in terms of food components

One of the main goals of food science is to ensure the high quality and safety of food. The inspection technology for known hazards has matured, and the identification of unknown and potential food safety hazards, as well as the identification of their composition and origin, is a challenge faced by food safety. Food safety and authenticity require multi-omics methods to support the implementation of qualitative discrimination to precise quantitative analysis, from targeted screening to non-target detection, and from multi component to full component analysis to address these challenges. The present review aims to provide characterizations, advantages, the latest progress, and prospects of using omics (including genomics, proteomics, and metabonomics) in food safety and authenticity. Multi omics strategies used to detect and verify different standard biomarkers of food will contribute to understanding the basic relationship between raw materials, processing, foods, nutrition, food safety, and human health.

The integrated metabolomics and sensory analyses unravel the peculiarities of mountain grassland-based cheese production: The case of Parmigiano Reggiano PDO

Untargeted metabolomics based on ultra-high-performance liquid chromatography coupled with high-resolution mass spectrometry was combined with sensory analysis to provide new insights into the impact of the feeding system from mountain regions (grassland deriving from permanent meadows) on the chemical fingerprint of Parmigiano Reggiano PDO hard cheese. In the framework of a representative investigation, two different ripening times (12 and 24 months) were also considered. Multivariate statistics allowed discriminating cheese samples from different feeding regimens according to their metabolomics signatures. Interestingly, mountain grassland-based cheese samples were characterized by a more favourable fatty acid profile, recording also feed-derived compounds (such as terpenoids and linoleic acid derivatives) potentially associated with both beneficial effects on human health and sensory properties. According to the sensory analysis, the impact of herbs and grass enhanced the colour and retro-olfactive complexity of Parmigiano Reggiano PDO cheese, with spicy, umami and intense vegetal aromatic notes representing distinctive features.

Interrogating the role of the milk microbiome in mastitis in the multi-omics era

There is growing interest in a functional understanding of milk-associated microbiota as there is ample evidence that host-associated microbial communities play an active role in host health and phenotype. Mastitis, characterized by painful inflammation of the mammary gland, is prevalent among lactating humans and agricultural animals and is associated with significant clinical and economic consequences. The etiology of mastitis is complex and polymicrobial and correlative studies have indicated alterations in milk microbial community composition. Recent evidence is beginning to suggest that a causal relationship may exist between the milk microbiota and host phenotype in mastitis. Multi-omic approaches can be leveraged to gain a mechanistic, molecular level understanding of how the milk microbiome might modulate host physiology, thereby informing strategies to prevent and ameliorate mastitis. In this paper, we review existing studies that have utilized omics approaches to investigate the role of the milk microbiome in mastitis. We also summarize the strengths and challenges associated with the different omics techniques including metagenomics, metatranscriptomics, metaproteomics, metabolomics and lipidomics and provide perspective on the integration of multiple omics technologies for a better functional understanding of the milk microbiome.

Buffalo milk and rumen fluid metabolome are significantly affected by green feed

The use of green feed for livestock breeding is an important strategy to encounter both the increasing demand for animal derived products and the perceptions of the consumers regarding animal welfare and sustainability. The aim of this study was to compare different feeding strategies in lactating water buffaloes by using a metabolomic approach. The study was carried out on 32 milking buffaloes that were randomly divided into two groups for a total period of 90 days (3 sampling times). DD Group (dry diet) received a standard total mixed ratio (TMR) characterized by dry forages and concentrates; ZG Group (zero grazing) fed an isoenergetic and isoproteic diet obtained using 30% of sorghum as green forage. Samples of milk and rumen fluid were analyzed by liquid chromatography-mass spectrometry (LC-MS) techniques. Data analyses revealed the presence of several differentially accumulated metabolites and among these, ten compounds were putatively identified in milk samples (i.e. L-carnitine, acetylcarnitine, propionylcarnitine, butyrylcarnitine, 2-methylbutyroylcarnitine, 2-hexenoylcarnitine, hexanoylcarnitine, glycerophosphocholine, δ-valerobetaine and γ-butyrobetaine) and four in rumen fluid (3-(2-hydroxyphenyl) propanoate, Indole-3-acrylic acid, oleamide (cis-9,10-octadecenoamide) and 20-carboxy-leukotriene B4). The modulation of these molecules in buffalo milk is significantly related to the green/dry based feeding and some the natural compound detected could be considered as health-promoting nutrients.

Impact of Pasture-Based Diets on the Untargeted Metabolomics Profile of Sarda Sheep Milk

In this work, untargeted metabolomics was used to shed light on the impact of different pasture-based diets on the chemical profile of Sarda sheep milk. The study considered 11 dairy sheep farms located in Sardinia, and milk samples were collected in 4 different periods, namely January, March, May, and July 2019, when all sheep had 58, 98, 138, and 178 days in milk, respectively. The animal diet composition was based on the intake of grazed herbage in natural pasture, hay, and concentrate. Overall, the combination of two comprehensive databases on food, namely the Milk Composition Database and Phenol-Explorer, allowed the putative identification of 406 metabolites, with a significant (p < 0.01) enrichment of several metabolite classes, namely amino acids and peptides, monosaccharides, fatty acids, phenylacetic acids, benzoic acids, cinnamic acids, and flavonoids. The multivariate statistical approach based on supervised orthogonal projections to latent structures (OPLS-DA) allowed us to predict the chemical profile of sheep milk samples as a function of the high vs no fresh herbage intake, while the prediction model was not significant when considering both hay and concentrate intake. Among the discriminant markers of the herbage intake, we found five phenolic metabolites (such as hippuric and coumaric acids), together with lutein and cresol (belonging to carotenoids and their metabolites). Additionally, a high discriminant power was outlined for lipid derivatives followed by sugars, amino acids, and peptides. Finally, a pathway analysis revealed that the herbage intake affected mainly five biochemical pathways in milk, namely galactose metabolism, phenylalanine metabolism, alpha-linolenic acid metabolism, linoleic acid metabolism, and aromatic amino acids involved in protein synthesis (namely tyrosine, phenylalanine, and tryptophan).

1H NMR Metabolic Profiling and Meat Quality in Three Beef Cattle Breeds from Northeastern Thailand

The increasing need for effective analytical tools to evaluate beef quality has prompted the development of new procedures to improve the animal sector's performance. In this study, three beef breeds-Thai native (TN), crossbred Brahman × Thai native (BT), and crossbred Charolais × Brahman (CB)-were compared in terms of their physicochemical and metabolic profiles. The findings demonstrated that TN beef was lighter and tougher than other beef. Beef odor was stronger in BT. In addition, CB beef was the most tender and had the highest intramuscular fat content. Twenty-one different metabolites were found overall through NMR and chemometric approaches. The primary factors contributing to the difference in OPLS-DA loading plots were acetylcholine, valine, adenine, leucine, phosphocreatine, β-hydroxypyruvate, ethanol, adenosine diphosphate, creatine, acetylcholine, and lactate. The multivariate analysis indicated that these metabolites in beef cattle breeds could be distinguished using NMR spectroscopy. The results of this study provide valuable information on the quality and meat metabolites of different breeds. This could help in the development of a more accurate assessment of the quality of beef in future research.

Grape Pomace in Ewes Diet Affects Metagenomic Profile, Volatile Compounds and Biogenic Amines Contents of Ripened Cheese

The main objective of this research was to evaluate the development of volatile organic compounds (VOCs) and the accumulation of biogenic amines (BAs) in relation to the dynamic of microbial population composition in fresh and ripened cheese produced from raw milk of ewes fed a diet containing grape pomace (GP+) and fed a standard diet (Ctrl). Genomic DNA was extracted from the cheeses at 2 (T2), 60 (T60), 90 (T90) and 120 (T120) days of ripening and prepared for 16S rRNA-gene sequencing to characterize the cheese microbiota; furthermore, VOCs were determined via solid-phase microextraction combined with gas chromatography-mass spectrometry and biogenic amines by HPLC analyses. Diet did not affect the relative abundance of the main phyla identified, Proteobacteria characterized T2 samples, but the scenario changed during the ripening. At genus level, Pseudomonas, Chryseobacterium and Acinetobacter were the dominant taxa, however, a lower percentage of Pseudomonas was detected in GP+ cheeses. Enterococcus became dominant in ripened cheeses followed in Ctrl cheeses by Lactobacillus and in GP+ cheeses by Lactococcus. The diet affected the development of carboxylic acids and ketones but not of aldehydes. Low levels of esters were identified in all the samples. In total, four biogenic amines were determined in cheeses samples and their levels differed between the two groups and during ripening time. In 60, T90 and T120 GP+ cheeses, a lower amount of 2-phenylethylamine was found compared to Ctrl. Putrescine was detected only in GP+ samples and reached the highest level at 120 days. Conversely, the amount of cadaverine in GP+ samples was invariable during the ripening. The concentration of tyramine in GP+ samples was compared to Ctrl during the ripening. Overall, significant positive correlations between some families of bacteria and the formation of VOCs and BAs were found.

Elderberry (Sambucus nigra L.) Encapsulated Extracts as Meat Extenders against Lipid and Protein Oxidation during the Shelf-Life of Beef Burgers

In this work, we studied the impact of encapsulated elderberry extracts as natural meat extenders to preserve both the quality and the oxidative stability of beef burgers. In particular, the comprehensive chemical changes of beef burgers treated with different antioxidants, namely, (a) a control without antioxidants, (b) 0.5 g/kg sodium erythorbate (ERY), (c) 2.5 g/kg encapsulated elderberry extract (EE 2.5), and (d) 5 g/kg encapsulated elderberry extract (EE 5), each one packaged under modified atmosphere (80% O2 and 20% CO2) for 13 days storage at 2 ± 1 °C, were deeply evaluated. Overall, EEs showed a wide array of antioxidant compounds, namely polyphenols like anthocyanins, flavonols, and phenolic acids. Multivariate statistics provided marked chemical differences between burgers manufactured with EEs and synthetic antioxidants (ERY) during 13-days storage in terms of both metabolomic profiles and typical lipid/protein oxidation markers (such as malondialdehyde and total carbonyls). Most of the differences could be attributed to some discriminant compounds, namely glutathione, 4-hydroxy-2-nonenal, hydroxy/peroxy-derivatives of fatty acids, carbonyl compounds (such as 5-nonen-2-one and 1,5-octadien-3-one), and cholesterol. Interestingly, significant correlations (p < 0.01) were observed between malondialdehyde, total carbonyls, and these discriminant metabolites. The combination of spectrophotometric approaches and a high-throughput untargeted metabolomics analysis outlined a strong modulation of both lipid and protein oxidations, likely promoted by the encapsulated meat extender (elderberry), thus confirming its ability to delay oxidative phenomena during the shelf-life of beef burgers.

Milk metabolome reveals pyrimidine and its degradation products as the discriminant markers of different corn silage-based nutritional strategies

The purpose of this study was to evaluate the effect of 6 different feeding systems (based on corn silage as the main ingredient) on the chemical composition of milk and to highlight the potential of untargeted metabolomics to find discriminant marker compounds of different nutritional strategies. Interestingly, the multivariate statistical analysis discriminated milk samples mainly according to the high-moisture ear corn (HMC) included in the diet formulation. Overall, the most discriminant compounds, identified as a function of the HMC, belonged to AA (10 compounds), peptides (71 compounds), pyrimidines (38 compounds), purines (15 compounds), and pyridines (14 compounds). The discriminant milk metabolites were found to significantly explain the metabolic pathways of pyrimidines and vitamin B₆. Interestingly, pathway analyses revealed that the inclusion of HMC in the diet formulation strongly affected the pyrimidine metabolism in milk, determining a significant up-accumulation of pyrimidine degradation products, such as 3-ureidopropionic acid, 3-ureidoisobutyric acid, and 3-aminoisobutyric acid. Also, some pyrimidine intermediates (such as l-aspartic acid, N-carbamoyl-l-aspartic acid, and orotic acid) were found to possess a high discrimination degree. Additionally, our findings suggested that the inclusion of alfalfa silage in the diet formulation was potentially correlated with the vitamin B₆ metabolism in milk, being 4-pyridoxic acid (a pyridoxal phosphate degradation product) the most significant and up-accumulated compound. Taken together, the accumulation trends of different marker compounds revealed that both pyrimidine intermediates and degradation products are potential marker compounds of HMC-based diets, likely involving a complex metabolism of microbial nitrogen based on total splanchnic fluxes from the rumen to mammary gland in dairy cows. Also, our findings highlight the potential of untargeted metabolomics in both foodomics and foodomics-based studies involving dairy products.

Comprehensive identification of molecular profiles related to sensory and nutritional changes in Mongolian cheese during storage by untargeted metabolomics coupled with quantification of free amino acids

Non-targeted metabolomics was used to study metabolites with low molecular weight which may contribute to quality deterioration of Mongolian cheese during storage. Microbiological analysis, pH, FAAs (free amino acids), volatile compounds, and sensory evaluation of the cheese during storage were also studied. A total of 278 metabolites were identified in Mongolian cheese, of which 51 metabolites were used as differential metabolites, including amino acids, peptides, organic acids, lipids, and carbohydrates. Bitter amino acids, bitter peptide (Phe-Ile), and organic acids (sinapic acid, butyric acid) increased during storage. Metabolic pathway analysis showed that differential metabolites were mainly related to amino acid metabolism, such as β-alanine metabolism and glycine, serine, and threonine metabolism. Moreover, accompanied with the increased contents of short-chain fatty acids, 2-undecanone and ethyl esters, strength of odor and unpleasant smell increased but overall acceptability decreased during Mongolian cheese storage. This research provides suitable strategies for quality control of Mongolian cheese during shelf life.

Characterization of Microbial Shifts during the Production and Ripening of Raw Ewe Milk-Derived Idiazabal Cheese by High-Throughput Sequencing

Simple Summary

Idiazabal is a traditional cheese produced from raw ewe milk in the Basque Country (Southwestern Europe). The sensory properties of raw milk cheeses have been attributed, among other factors, to microbial shifts that occur during the production and ripening processes. In this study, we used high-throughput sequencing technologies to investigate the microbiota of Latxa ewe raw milk and the dynamics during cheese production and ripening processes. The microbiota of raw milk was composed of lactic acid bacteria (LAB), environmental bacteria and non-desirable bacteria. Throughout the cheese making and ripening processes, the growth of LAB was promoted, whereas that of non-desirable and environmental bacteria was inhibited. Moreover, some genera not reported previously in raw ewe milk were detected and clear differences were observed in the bacterial composition of raw milk and cheese among producers, in relation to LAB and environmental or non-desirable bacteria, some of which could be attributed to the production of flavour related compounds.

Abstract

In this study, we used high-throughput sequencing technologies (sequencing of V3–V4 hypervariable regions of 16S rRNA gene) to investigate for the first time the microbiota of Latxa ewe raw milk and the bacterial shifts that occur during the production and ripening of Idiazabal cheese. Results revealed several bacterial genera not reported previously in raw ewe milk and cheese, such as Buttiauxella and Obesumbacterium. Both the cheese making and ripening processes had a significant impact on bacterial communities. Overall, the growth of lactic acid bacteria (LAB) (Lactococcus, Lactobacillus, Leuconostoc, Enterococcus, Streptococcus and Carnobacterium) was promoted, whereas that of non-desirable and environmental bacteria was inhibited (such as Pseudomonas and Clostridium). However, considerable differences were observed among producers. It is noteworthy that the starter LAB (Lactococcus) predominated up to 30 or 60 days of ripening and then, the growth of non-starter LAB (Lactobacillus, Leuconostoc, Enterococcus and Streptococcus) was promoted. Moreover, in some cases, bacteria related to the production of volatile compounds (such as Hafnia, Brevibacterium and Psychrobacter) also showed notable abundance during the first few weeks of ripening. Overall, the results of this study enhance our understanding of microbial shifts that occur during the production and ripening of a raw ewe milk-derived cheese (Idiazabal), and could indicate that the practices adopted by producers have a great impact on the microbiota and final quality of this cheese.

Foodomic-Based Approach for the Control and Quality Improvement of Dairy Products

The food quality assurance before selling is a needed requirement intended for protecting consumer interests. In the same way, it is also indispensable to promote continuous improvement of sensory and nutritional properties. In this regard, food research has recently contributed with studies focused on the use of 'foodomics'. This review focuses on the use of this technology, represented by transcriptomics, proteomics, and metabolomics, for the control and quality improvement of dairy products. The complex matrix of these foods requires sophisticated technology able to extract large amounts of information with which to influence their aptitude for consumption. Thus, throughout the article, different applications of the aforementioned technologies are described and discussed in essential matters related to food quality, such as the detection of fraud and/or adulterations, microbiological safety, and the assessment and improvement of transformation industrial processes (e.g., fermentation and ripening). The magnitude of the reported results may open the door to an in-depth transformation of the most conventional analytical processes, with the introduction of new techniques that allow a greater understanding of the biochemical phenomena occurred in this type of food.

Case Study on the Microbiological Quality, Chemical and Sensorial Profiles of Different Dairy Creams and Ricotta Cheese during Shelf-Life

This work investigated the microbiological quality and chemical profiles of two different dairy creams obtained by centrifugation vs. natural creaming separation systems. To this aim, an untargeted metabolomics approach based on UHPLC-QTOF mass spectrometry was used in combination with multivariate statistical tools to find potential marker compounds of the two different types of two dairy creams. Thereafter, we evaluated the chemical, microbiological and sensorial changes of a ricotta cheese made with a 30% milk cream (i.e., made by combining dairy creams from centrifugation and natural creaming separation) during its shelf-life period (12 days). Overall, microbiological analysis revealed no significant differences between the two types of dairy creams. On the contrary, the trend observed in the growth of degradative bacteria in ricotta during shelf-life was significant. Metabolomics revealed that triacylglycerols and phospholipids showed significant strong down-accumulation trends when comparing samples from the centrifugation and natural creaming separation methods. Additionally, 2,3-Pentanedione was among the best discriminant compounds characterising the shelf-life period of ricotta cheese (VIP score = 1.02), mainly related to sensorial descriptors, such as buttery and cheesy. Multivariate statistics showed a clear impact of the shelf-life period on the ricotta cheese, revealing 139 potential marker compounds (mainly included in amino acids and lipids). Therefore, the approach used showed the potential of a combined metabolomic, microbiological and sensory approach to discriminate ricotta cheese during the shelf-life period.

A combined metabolomics and peptidomics approach to discriminate anomalous rind inclusion levels in Parmigiano Reggiano PDO grated hard cheese from different ripening stages

Parmigiano Reggiano is a hard cheese with a Protected Designation of Origin (PDO) certification that also applies to the grated product. The percentage of rind in grated Parmigiano Reggiano is regulated by the PDO production Specification and must not exceed the limit of 18% (w/w). The present study evaluates the potential of an untargeted foodomics approach to detect anomalous inclusions of rind in grated Parmigiano Reggiano cheese. In particular, a combined metabolomics and peptidomics approach was used to detect potential markers of counterfeits (rind > 18%). In the framework of realistic food integrity purposes, non-Parmigiano Reggiano grated samples and different ripening times were also considered. Untargeted metabolomics allowed detecting 347 compounds, with a prevalence of amino acids and peptide derivatives, followed by fatty acyls and other compounds (such as lactones, ketones, and aldehydes) typically related to proteolysis and lipolysis events. Overall, the unsupervised multivariate statistics showed that the ripening time plays a hierarchically higher impact than rind inclusion in determining the main differences in the chemical profiles detected. Interestingly, supervised statistics highlighted distinctive markers for ripening time and rind inclusion, with only 16 common discriminant compounds being shared between the two conditions. The best markers of rind inclusion > 18% were 2-hydroxyadenine (VIP score = 1.937; AUC value = 0.83) and the amino acid derivatives argininic acid (VIP score = 1.462; AUC value = 0.75) and 5-hydroxyindole acetaldehyde (VIP score = 1.710; AUC value = 0.86). Interestingly, the medium-chain aldehyde 4-hydroperoxy-2-nonenal was a common marker of both ripening time and anomalous rind inclusion (>18%), likely arising from the lipid oxidation processes. Finally, among potential marker peptides of rind inclusion, the alpha-S1 casein proteolytic product (F)FVAPFPEVFGK(E) could be identified.

Screening of Regulated and Emerging Mycotoxins in Bulk Milk Samples by High-Resolution Mass Spectrometry

In this work, a retrospective screening based on ultra-high-performance liquid chromatography (UHPLC) coupled with high-resolution mass spectrometry (HRMS) based on Orbitrap-Q-Exactive Focus™ was used to check the occurrence of regulated and emerging mycotoxins in bulk milk samples. Milk samples were collected from dairy farms in which corn silage was the main ingredient of the feeding system. The 45 bulk milk samples were previously analyzed for a detailed untargeted metabolomic profiling and classified into five clusters according to the corn silage contamination profile, namely: (1) low levels of Aspergillus- and Penicillium-mycotoxins; (2) low levels of fumonisins and other Fusarium-mycotoxins; (3) high levels of Aspergillus-mycotoxins; (4) high levels of non-regulated Fusarium-mycotoxins; (5) high levels of fumonisins and their metabolites. Multivariate statistics based on both unsupervised and supervised analyses were used to evaluate the significant fold-change variations of the main groups of mycotoxins detected when comparing milk samples from clusters 3, 4, and 5 (high contamination levels of the corn silages) with cluster 1 and 2 (low contamination levels of the corn silages). Overall, 14 compounds showed a significant prediction ability, with antibiotic Y (VIP score = 2.579), bikaverin (VIP score = 1.975) and fumonisin B2 (VIP score = 1.846) being the best markers. The k-means clustering combined with supervised statistics showed two discriminant groups of milk samples, thus revealing a hierarchically higher impact of the whole feeding system (rather than the only corn silages) together with other factors of variability on the final mycotoxin contamination profile. Among the discriminant metabolites we found some Fusarium mycotoxins, together with the tetrapeptide tentoxin (an Alternaria toxin), the α-zearalenol (a catabolite of zearalenone), mycophenolic acid and apicidin. These preliminary findings provide new insights into the potential role of UHPLC-HRMS to evaluate the contamination profile and the safety of raw milk to produce hard cheese.

Changes of Milk Metabolomic Profiles Resulting from a Mycotoxins-Contaminated Corn Silage Intake by Dairy Cows

In this study, an untargeted metabolomics approach based on ultra-high-performance liquid chromatography coupled with high-resolution mass spectrometry (UHPLC-HRMS) was used for investigating changes in chemical profiles of cow milk considering diets based on mycotoxins-contaminated corn silages. For this purpose, 45 milk samples were classified into five clusters according to the corn silage contamination profile, namely (1) low levels of Aspergillus- and Penicillium-mycotoxins; (2) low levels of fumonisins and other Fusarium-mycotoxins; (3) high levels of Aspergillus-mycotoxins; (4) high levels of non-regulated Fusarium-mycotoxins; (5) high levels of fumonisins and their metabolites, and subsequently analyzed by UHPLC-HRMS followed by a multivariate statistical analysis (both unsupervised and supervised statistical approaches). Overall, the milk metabolomic profile highlighted potential correlations between the quality of contaminated corn silages (as part of the total mixed ration) and milk composition. Metabolomics allowed to identify 628 significant milk metabolites as affected by the five levels of corn silage contamination considered, with amino acids and peptides showing the highest metabolite set enrichment (134 compounds). Additionally, 78 metabolites were selected as the best discriminant of the prediction model built, possessing a variable importance in projection score >1.2. The average Log Fold-Change variations of the discriminant metabolites provided evidence that sphingolipids, together with purine and pyrimidine-derived metabolites were the most affected chemical classes. Also, metabolomics revealed a significant accumulation of oxidized glutathione in milk samples belonging to the silage cluster contaminated by emerging Aspergillus toxins, likely involved in the oxidative imbalance. These preliminary findings provide new insights into the potential role of milk metabolomics to provide chemical indicators of mycotoxins-contaminated corn silage feeding systems.