Volatile organic compounds in cheese production chain (VOCHEESE)

In recent years, consumers have become increasingly interested in the quality aspects of food. Food quality, in turn, is strongly related to the sensory characteristics such as the flavor. Several scientific studies have shown that the Volatile Organic Compounds (VOCs) released by the food are related to the flavor and can be considered as assistive markers in the production chain. Today, the analysis of VOCs requires fast, non-invasive, and solvent free devices. It has been shown that the VOCs can be extracted, identified, and measured with a Gas ChromatographyMass Spectrometry (GC-MS) without any pre-concentration or pre-treatment of the food. The main objective of this PhD thesis was to investigate the presence of volatile compounds in dairy products. More precisely, this study aimed in i) qualifying and quantifying VOCs in dairy products, ii) examining their formation and iii) integrating knowledge on VOCs by tracking their release during the whole production process from the raw materials till the final dairy product. In addition, statistical analysis was applied to link VOCs with the genetic characterization of animals, dairy system and individual cow-factors (e.g. stage of lactation, order of parity and milk yield). The identification and quantification of VOCs were performed using fast and non-invasive analytical approaches (Solid Phase Micro Extraction/Gas Chromatography-Mass Spectrometry SPME/GC-MS and Proton Transfer Reaction-Time of Flight-Mass Spectrometry PTR-ToF-MS) that can monitor the evolution of VOCs. To achieve the overall goal, the research was partitioned in four interrelated subparts as described below. The aim of the first chapter was to study the VOCs presence in the headspace of cheese. To this purpose, 150 cheeses ripened for two months were used. The cheeses were obtained through an individual model cheese-making approach using milk from individual Brown Swiss cows. Animals reared in 30 herds belonging to different dairy systems, from traditional (typical of the mountainous area) to modern ones. The study identified 55 VOCs classified in the chemical families of free fatty acids, esters, alcohols, aldehydes, ketones, lactones, terpenes, and pyrazines. We found that dairy system and individual cow characteristics (lactation stage, order of parity and daily milk yield) influenced the volatile compounds. In order, to test the instrument reproducibility and the model cheese-making procedure; data of GC analysis, order of injection of the sample into instrument, and vat were included in the statistical model. In many cases, these analytical factors did not affect the amount of VOCs released by cheese. In the second chapter, the potential of a new spectrometric technique (PTR-ToF-MS) was investigated to study cheese quality traits on a large scale. The PTR-ToF-MS allows direct injection of the sample headspace without extraction or pre-concentration steps, has a shorter analysis time (only a few seconds per sample) and greater sensitivity that permit to monitor on-line the evolution of volatile compounds. The resulting spectral information can provide a very detailed description of samples, which is useful for characterizing food quality and typicality. In particular, we analyzed the volatile fingerprint of 1,075 model cheeses produced using individual milk of Brown Swiss cows reared in 72 herds of different dairy systems. The output of PTR (spectrum) was characterized by more than 600 spectrometric peaks (variables). After removing interfering ions and background noise a set of 240 peaks was selected. Further, based on the results of the first contribution and literature, 61 peaks were identified. These peaks represent the major part of the cheese flavor. To summarize the amount of information, a multivariate analysis (PCA) was applied associating principal components (PC) with the 240 spectrometric peaks. Following, we tried to characterize the PCs through the correlations between PCs and the spectrometric peaks. The effects of dairy system, herd within dairy system, individual cows characteristics (lactation stage, order of parity and milk yield), and vat used for the cheese-making on the PCs and on the 240 peaks were analyzed. Dairy system was correlated with PC and 57 spectrometric peaks, especially when the herds were using Total Mixer Ration (TMR) as feeding technique, including or not maize silage in the diets. Regarding the individual animal characteristics, the most significant effect was the stage of lactation (139 peaks), followed by milk yield and parity, with 31 and 21 peaks, respectively. Finally, the vat used for the cheese-making was not found to be significant, confirming the good reproducibility of the model cheese-making procedure used to study cheese quality aspects. In the third chapter, the effect of cows’ genetics to the VOCs of ripened cheeses was assessed. Principal components and the 240 spectrometric peaks (as described above in the second contribution) were used fitting an animal model in a Bayesian framework. On average, heritability (h2) of 7% for PCs was found, which is similar to h2 of somatic cell count and much lower than the h2 of milk fat content and daily milk yield. It is interesting to note that only a small proportion of peaks showed very low h2 (<7%). The major part of them showed values similar to those found for PCs, while forty peaks presented heritability similar to that of milk yield and other milk quality traits. The variability attributed to the herd was different for the various PC. Results suggest a potential of improvement for several cheese VOCs through genetic selection in dairy cow breeding programs. The aim of the fourth chapter was to study the effect of summer transhumance on the quality traits of dairy products. Due to the extended work, this contribution was further splitted into two parts. In the first part, the evolution of milk and cheese quality characteristics were studied, while in the second part the evolution of VOC content of dairy products was analyzed. For the first part, chemical characteristics and technological properties of 11 dairy products obtained during summer transhumance of cows to Alpine pastures (Malga) were analyzed. Dairy products obtained throughout this period are known to give origin to high-value, healthier products, and extra tasty,. Bulk milk from 148 dairy cows reared day and night on Alpine pasture (1,860 m a.s.l.) was used. We performed 7 experimental cheese-making according to traditional mountain techniques, one every two weeks, using milk produced during the summer transhumance (from June to September). For each cheese-making we collected: milk from the evening milking (day before the cheese-making), the same milk the following morning (after natural creaming), the cream separated, the whole milk from the morning milking, the milk in vat obtained mixing the creamed evening milk with the whole morning milk, the fresh curd, the whey, the ricotta obtained from whey, and the residual scotta. Moreover, the curd was used to produce typical “Malga” cheese that was ripened for 6 and 12 months. The chemical characteristics were measured with infrared technology. Results highlighted variation in milk yield, milk chemical composition, cheese yield and curd recoveries and/or loss of nutrients in the traditional cheese-making. In particular, a reduction of milk yield, fat, protein and lactose contents of milk during summer transhumance was observed. Nevertheless, the return to lowland farming systems of the cows at the end of grazing season, positively affected milk yield and milk chemical composition. The average of cheese yield was 14.2%, while recoveries of fat, protein, total solids and energy were 85.1%, 77.8%, 49.4% and 58.1%, respectively. These results were in accordance to those found in the literature. For the second part of this chapter, the VOCs content of sample headspace was measured through SPME/GC-MS. Forth nine VOCs belonging to the chemical families of alcohols, aldehydes, free fatty acids, ketones, esters, lactones, terpenes, phenolic, and sulphur compounds were detected. In addition, the evolution of VOCs and their chemical family across the cheese- and ricotta-making processing as well as during the cheese ripening period was tracked. The comparison between VOCs concentration of 4 types of milk (whole evening, creaming milk, whole morning, milk in vat) showed that the creaming process significantly affected about half of all the volatile organic compounds analyzed, followed by the effects of milking (evening milking vs. morning milking) and the mixing (creamed milk mixed with whole morning milk). In general, the cream, in contrast to curd and ricotta, showed higher content of free fatty acids, sulphurs and terpenes compounds. Moreover, in ricotta a higher VOC concentration was observed compared to the curd, probably due to the high temperature required during the ricotta process. The effect of the progressive nutrient depletion of milk was investigated by contrasts between VOC concentration of milk in the vat, whey, and scotta. Although milk contains a greater amount of nutrients, whey and scotta have shown a higher concentration of VOCs with the exceptions of esters, sulphurs, terpenes and phenolic compounds. Finally, the effect of ripening was tested by comparing the quantity of VOCs of curd and of aged cheeses (6 and 12 months). The release of volatile compounds increased with increasing ripening period in relation with the enzymatic and microbiological activity of cheese. In summary, the spectrometric techniques (SPME/GC-MS and PTR-ToF-MS) used in this work demonstrated to be very efficient to characterize the volatile organic compounds of dairy products. The dairy system, and cow related factors affected the volatile fingerprint of ripened cheeses. Particularly, concerning the individual animal source of variation, lactation stage was the most important effect followed by the cow’s parity and the milk yield. On the basis of phenotypes used in this work, the traits collected offered the potential for a genetic analysis to be carried out. The genetic analysis demonstrated the existence of an exploitable genetic variability of the volatile profile of cheese that might be useful for an (in)direct selection of dairy cows for cheese quality traits in breeding programs. Nevertheless, further research is needed in this area. In the era of genomics for e.g., it might be interesting to associate genomic regions to specific VOCs. This information might be useful for genomic breeding programs. The evolution of volatile compounds across the production chain depends on specific technological aspects, such as the process of natural creaming, the temperature of coagulation, and the ripening period. The monitoring of volatile fingerprint permits to obtain dairy products with specific organoleptic characteristics useful to differentiate them on the market and to improve the supply chain efficiency on the basis of quality aspects.