STRATEGIA ANALITICA PER LA CARATTERIZZAZIONE DEL PROFILO VITAMINICO LIPOSOLUBILE E CAROTENOIDEO DEL LATTE DI DIFFERENTI SPECIE ANIMALI MEDIANTE IFENAZIONE HPLC-DAD-TANDEM MS

Vitamin determination is a complex and challenging task, especially when aimed at the analysis of biological and food samples. Milk is a unique matrix, being at the same time biological fluid and food with the chemical characteristics of three phases: emulsion, suspension and colloidal solution. Furthermore, it is an almost complete food and an excellent source of vitamins, in particular vitamins A, B1, B2, B5 and B12. The water-soluble vitamins occur in the serum, while the fat-soluble vitamins and carotenoids are associated to the lipid fraction. Bovine milk has been one of the most investigated food matrices, but also milk from other animal species deserve to be investigated since it is essential for human diet in several parts of the world. As a matter of fact, India and Pakistan are the main producers of buffalo milk, the second most produced in the world after bovine milk. Europe, China and some zones of Africa are excellent producers of sheep milk, while goat milk is produced in some parts of Asia and in the countries of the Mediterranean area. Donkey milk has a niche market, but it is considered a valuable alternative for people who are allergic to cow milk. In the scientific literature, a large number of papers have focused on the macronutrient composition of bovine milk, while its vitamin and caroteinoid profile still have to be completed. On the other hand, this information is almost completely lacking for buffalo, sheep, goat and donkey milk. Official procedures are based on liquid chromatography with UV-visible or fluorescence detection for the individual determination of vitamins A, E, D and K. The scientific literature describes various methods for multivitamin determination [1-8], but most of them are addressed to the analysis of fortified foods. Only two are aimed at the determination of endogenous forms in breast milk [6,7] and in vegetable foods [8]. This lack of literature is likely due to the several problems coming up during the development of a multivitamin method for the determination of endogenous forms [9]. A first problem concerns the commercial unavailability of authentic standards of some vitamin forms, carotenoids and their geometrical isomers. The subtle structural difference between homologues belonging to the same group hampers their chromatographic separation, whereas the chemical heterogeneity among vitamin groups makes difficult to find common conditions of extraction and detection. However, the development of a simultaneous extraction procedure is the most critical point. In the case of fatty foods such as milk, the high lipid content compromises the extraction efficiency. In the literature, hot saponification is the most adopted solution to free vitamins and carotenoids from saponifiable fraction; nevertheless, this reaction, typically performed at 70-80°C for 30 minutes, is responsible for the rapid decomposition of vitamers K, a severe loss of xanthophylls, the thermal isomerization of all-trans--carotene and vitamin D. Thence, this work was aimed at developing a method, based on the hyphenation HPLC-DAD-tandem MS, to rapidly and completely characterize fat-soluble micronutrients in milk of different species of herbivores (cow, buffalo, sheep, goat, donkey). Overnight cold saponification was optimized as simultaneous extraction procedure. Bovine milk, more easily available, was used to develop the method, which was then optimized for the other types of milk. We have chosen to analyze raw milk, produced by pasture-fed animals, so to avoid the occurrence of vitamins due to the intake of fortified feedstuffs or losses due to the processing. The analyte, were separated by non-aqueous reversed-phase (NARP) chromatography: carotenoids on a C30 column, while the fat-soluble vitamins on a tandem system of C18 columns. The atmospheric pressure chemical ionization (APCI) in positive ion mode was the most suitable technique for the mass-spectrometric detection of -tocopherol, -tocopherol, -tocopherol, ergocalciferol, cholecalciferol, phylloquinone, menaquinone-4, all-trans-retinol, all-trans-lutein, all-trans-zeaxanthin, all-trans--cryptoxanthin, all-trans--carotene. In addition to the 12 target analytes, the combined DAD-MS detection system allowed the screening of other carotenoids, whose standards are not available on the markets, basing their identification on the expected retention time, the absorbance spectra, acquired between 200 and 700 nm, and the mass-spectrometric data. For each target analyte, the LC-tandem MS method was validated in terms of both quality (identification, selectivity) and quantitative parameters (recoveries, precision, limits of detection and quantitation, sensitivity, linear dynamic range). Regarding the two selected transitions for each analyte, the most intense one was used to perform quantitative analysis, whereas the least intense one for identification purposes. The presence of each compound in matrix was confirmed by matching its retention time and relative abundance of the two SRM transitions with the values of the corresponding standard in solvent. The recoveries, assessed on 6 replicates, were above 80% for all analytes, with the exception of vitamers K (54%-67%). The relative standard deviations (RSD) associated to recoveries were representative of the intra-day precision, whereas the inter-day precision was estimated as the RSD of 12 replicates performed within 2 weeks. The quantitative analysis was carried out using the standard additions method. The linear dynamic range was investigated up to 200 ng injected for all analytes, while for -retinol and γ-tocopherol up to 2000 ng injected. A linear correlation coefficient greater than 0.99 was achieved for all the analytes in the different types of milk. After the LC-MS method validation, the whole analytical approach, based on the HPLC-DAD-tandem MS hyphenation, was applied for the characterization of cow, buffalo, goat, sheep and donkey milk. The survey aimed to trace a species-dependent profile, without taking into account the dependence on parameters such as season, stage of lactation and intra- and inter-individual variability. It is known that milk is a good source of vitamin A and E, but the tested milk samples were particularly rich in these vitamins, probably because obtained from pasture-fed animals and analyzed immediately after sampling; on the other hand the milk of donkey was poor of these micronutrients, probably because of its low fat content. As regards vitamin E, α-tocopherol was the most abundant vitamin form (more than vitamin A) found in all kinds of milk analyzed in this work, with the exception of the buffalo one. γ-Tocopherol occurred in lower amount, whereas the δ-homologue was absent in bovine, buffalo, and donkey milk. The vitamers D were detected in buffalo milk, and, in trace amounts, in goat and donkey milk. Although the literature reports for bovine milk the presence of phylloquinone and menaquinones, from MK-4 to MK-9 [10,11], the high selectivity of the developed method allowed excluding the occurrence of MK-7 in the analyzed milk samples; for this confirmation, the standard of MK-7 was obtained in our laboratory, purifying a dietary supplement purchased in a drugstore on a semi-preparative column. Among all kinds of milk, cow milk was that had the lowest content of vitamin A but a significant amount of β-carotene and a variety of carotenoids lacking in the other types of milk, with the exception of lutein and zeaxanthin. Missing the authentic standards, the screening of carotenoids in cow milk was achieved by combining LC-DAD-MS data. In formulating a hypothesis to identify a pigment, the UV-Vis spectrum is fundamental; indeed, most of carotenoids show a characteristic three-peak spectrum: the identification is based on the position of the maxima and on the fine structure. For the carotenoids selected in this study, the wavelength of the central peak (MAX) was calculated applying Fieser-Kuhn rules. A cis isomer was identified comparing its spectrum to that of the corresponding all-trans isomer and evaluating: i) the extent of the ipsocromic shift of the λMAX; ii) the hypochromic effect and the reduction of the fine structure of the entire spectrum; iii) the appearance of a "cis"-peak in the near-UV region (330-350 nm), iv) the Qratio, i.e. the ratio of the intensity of the cis band to the central band. In this way, on the LC-DAD-MS/MS chromatograms of the bovine milk samples, were identified: zeinoxanthin, all-trans--cryptoxanthin, a cis-isomer of -cryptoxanthin, 3-hydroxy-β-zeacarotene and β-zeacarotene. It was also detected a group of structural and geometric isomers of all-trans--carotene: a cis-isomer of -carotene, 13-cis--carotene, all-trans--carotene and -carotene. In all the analyzed samples, it was also found a compound with an absorption maximum at 422-424 nm, characteristic of the Soret band; it was probably ascribable to a degradation product of chlorophyll a, generated in rumen of animals. On the other hand, it was not possible to identify two unknown compounds because of the low signal intensity of both detectors. In the case of cow milk, besides raw milk, other categories were analyzed: fresh pasteurized whole milk, biological fresh pasteurized whole milk, high quality fresh whole milk, semi-skimmed fresh milk, UHT, whole yogurt. High levels of vitamins and carotenoids were found in the biological commercial milk while, unexpectedly, low levels occurred in the high quality milk samples. The low concentrations found in the semi-skimmed milk are due to the skimming process which eliminates not only fat but also part of vitamins and carotenoids. Yogurt showed the same micronutrient levels of fresh pasteurized milk. UHT milk was particularly abundant in 13-cis--carotene, probably produced during the high temperature sterilization because of the thermal isomerization of the all-trans isomer. After this exhaustive characterization, another significant part of this thesis work was further addressed to define the detailed composition of vitamin A vitamers of the same varieties of milk, with the only exception of donkey milk. It has known that the compounds with vitamin A activity are present in milk mainly as retinoids [12]. Among retinoids, the most abundant forms are esters of retinol with saturated and unsaturated fatty acids, while only a small fraction is constituted by free retinol. This information has been provided from a single work, based on HPLC-UV and published in 1989 [13]; in that paper the identification of the various forms was exclusively based on the chromatographic retention time, which is a parameter necessary but not sufficient to ensure a certain analyte identification in a complex matrix such as milk. In the present study, cold saponification allowed us to determine the total content of vitamin A as retinol. In order to define the detailed distribution of 17 vitamers A (retinol, retinoic acid, retinal and esters: retinyl caprylate, caprate, palmitoleate, laurate, myristate, pentadecanoate, arachidonate, palmitate, eptadecanoate, linoleate, oleate, stearate, linolenate, eicosanoate) in milk, direct extraction with solvent and a tandem system of reversed phase columns (C18/C18 and C18/C30) coupled to a tandem mass spectrometer were used. Taking into consideration that deuterated structural analogues of these retinoids are not available on the markets, the internal standards were chosen on the basis of what has been reported in literature and results obtained by preliminary tests; accordingly, retinyl propionate was selected as internal standard for the quantitative analysis of retinyl esters with medium-chain fatty acids (8-12 carbon atoms), while retinyl arachidonate for esters with longer chain fatty acids. For all analytes, the recoveries, evaluated from the average of six replicates, were  68%; analytical limits were similar for all four analyzed types of milk, indicating the presence of a similar matrix effect for the different extracts. The linear correlation coefficients of the calibration curves, valued by applying the standard-addition method, were between 0.9941 and 0.9999. The validated method was then applied to the analysis of samples of cow, goat, sheep and buffalo milk. In the survey, both qualitative and quantitative differences concerning the composition of retinyl esters were observed in the milk samples from the different animal species. From a qualitative point of view, the results showed the presence of: - 6 retinoids common to the four types of milk: retinol, retinyl linolenate, retinyl oleate, retinyl palmitate, retinyl stearate and retinyl eptadecanoate; - 1 retinoid common to sheep and goat milk, retinyl eicosanoate; - 1 retinoid common to cow and buffalo milk, retinyl linoleate; - 1 retinoid characteristic of goat milk, retinyl caprate; - 1 retinoid characteristic of cow milk, retinyl myristate. From a quantitative point of view, buffalo milk differed from the others for the highest concentration of free retinol and retinyl linolenate; this latter form was about seven times higher than bovine milk. Furthermore, buffalo milk showed the highest content of retinyl palmitate, even if a considerable concentration was also found in sheep milk. Milk of small ruminants was characterized by a content of retinyl eptadecanoate from 4 to 6 times higher than cow and buffalo milk. In this work, the distribution of retinyl caprate, laurate, pentadecanoate, palmitoleate and myristate in bovine milk resulted different from that presented in the only pre-existing work [13]. In particular, the authors found retinyl caprate in cow milk but not in goat milk; this result is anomalous since capric acid is particularly abundant in goat milk. In addition, retinyl palmitate followed by retinyl oleate were always found to be the most abundant vitamers A, while our results agree with these observations only partially. In conclusion, the interest in this work is motivated by several reasons: firstly, a LC-DAD-MS based approach was proposed for a complete characterization of fat-soluble micronutrients in milk, providing a more certain identification than methods reported in the literature. Secondly, detailed data on the composition of fat-soluble micronutrients were achieved for five kinds of milk, filling the information gap of literature. Eventually, it could provide a tool to detect adulteration: - this work has definitively established that β-carotene (and other provitamin A carotenoids) occurs only in cow milk; so, traces of β-carotene in buffalo mozzarella could indicate use of bovine milk during its production. - another finding of this work is the very high concentration of retinyl linolenate in buffalo milk and retinyl eicosanoate in sheep milk; these esters could be simple biomarkers to detect other kinds of adulteration. (1) Salo-Väänänen, P.; Ollilainen, V.; Mattila, P.; Lehikoinen, K.; Salmela-Mölsä, E.; Piironen, V. Simultaneous HPLC analysis of fat-soluble vitamins in selected animal products after small-scale extraction. Food Chem., 2000, 71, 535-543. (2) Herrero-Barbudo, M. C.; Granado-Lorencio, F.; Blanco-Navarro, I.; Olmedilla-Alonso, B. Retinol, α- and γ-tocopherol and carotenoids in natural and vitamin A and E fortified dairy products commercialized in Spain. Int. Dairy J. 2005, 15, 521-526. (3) Gomis, D. B.; Fernández, M. P.; Gutièrrez Alvarez, M. D. Simultaneous determination of fat-soluble vitamins and provitamins in milk by microcolumn liquid chromatography. J. Chromatogr. A, 2000, 891, 109-114. (4) Blanco, D.; Fernandez, M. P.; Gutierrez, M. D. Simultaneous determination of fat-soluble vitamins and provitamins in dairy products by liquid chromatography with a narrow-bore column. Analyst, 2000, 125, 427-431. (5) Chauveau-Duriot, B.; Doreau, M.; Noziere, P.; Grailet, B. Simultaneous quantification of carotenoids, retinol, and tocopherols in forages, bovine plasma, and milk: validation of a novel UPLC method. Anal. Bioanal. Chem. 2010, 397, 777-790. (6) Heudi, O.; Trisconi, M. J.; Blake, C. J. Simultaneous quantification of Vitamins A, D3 and E in fortified infant formulae by liquid chromatography–mass spectrometry. J. Chromatogr. A, 2004, 1022, 115-123. (7) Kamao, M.; Tsugawa, N.; Suhara, Y.; Wada, A.; Mori, T.; Murata, K.; Nishino, R.; Ukita, T.; Uenishi, K.; Tanaka, K.; Okano, T. Quantification of fat-soluble vitamins in human breast milk by liquid chromatography-tandem mass spectrometry. J. Chromatogr. B, 2007, 859, 192-200. (8) Gentili, A.; Caretti, F. Evaluation of a method based on liquid chromatography–diode array detector–tandem mass spectrometry for a rapid and comprehensive characterization of the fat-soluble vitamin and carotenoid profile of selected plant foods. J. Chromatogr. A, 2011, 1218, 684-697. (9) Gentili A.; Caretti F., Multimethod for water-soluble vitamins in foods by using LC-MS In Fortified Foods with Vitamins– Analytical Concepts to Assure Better and Safer Products. Editor M. Rychlik, publisher ‘Wiley –VCH VerlagGmbH & Co. KGaA’, 2001. Print ISBN: 9783527330782. Online ISBN: 9783527634156. DOI: 10.1002/9783527634156. (10) Indyk, H. E.; Wollard, D. C. Vitamin K in milk and infant formulas: determination and distribution of phylloquinone and menaquinone-4. Analyst, 1997, 122, 465-469. (11) Koivu-Tikkanen, T. J.; Ollilainen, V.; Piironen, V. I. Determination of phylloquinone and menaquinones in animal products with fluorescence detection after postcolumn reduction with metallic zinc. J. Agric. Food Chem. 2000, 48, 6325-6331. (12) Gentili A., The Chemistry of Vitamin A (Chapter 5) In Food and Nutritional Components in Focus No. 1, Vitamin A and Carotenoids: Chemistry, Analysis, Function and Effects. Edited by Victor R Preedy, RCS Publishing, 2012, 73-89. ISBN: 978-1-84973-550-6. DOI:10.1039/9781849735506-00073. (13) Wollard, D. C.; Indyk, H. The distribution of retinyl esters in milks and milk products. J. Micronutr. Anal. 1989, 5, 35-52