Eur Food Res Technol (2011) 232:191–204 DOI 10.1007/s00217-010-1412-6 REVIEW PAPER Protein changes during malting and brewing with focus on haze and foam formation: a review Elisabeth Steiner • Martina Gastl • Thomas Becker Received: 17 October 2010 / Revised: December 2010 / Accepted: 13 December 2010 / Published online: January 2011 Ó Springer-Verlag 2010 Abstract Beer is a complex mixture of over 450 constituents and, in addition, it contains macromolecules such as proteins, nucleic acids, polysaccharides, and lipids In beer, several different protein groups, originating from barley, barley malt, and yeast, are known to influence beer quality Some of them play a role in foam formation and mouthfeel, and others are known to form haze and have to be precipitated to guarantee haze stability, since turbidity gives a first visual impression of the quality of beer to the consumer These proteins are derived from the malt used and are influenced, modified, and aggregated throughout the whole malting and brewing process During malting, barley storage proteins are partially degraded by proteinases into amino acids and peptides that are critical for obtaining high-quality malt and therefore high-quality wort and beer During mashing, proteins are solubilized and transferred into the produced wort Throughout wort boiling proteins are glycated and coagulated being possible to separate those coagulated proteins from the wort as hot trub In fermentation and maturation process, proteins aggregate as well, because of low pH, and can be separated The understanding of beer protein also requires knowledge about the barley cultivar characteristics on barley/malt proteins, hordeins, protein Z, and LTP1 This review summarizes the protein composition and functions and the changes of malt proteins in beer during the malting and brewing process Also methods for protein identification are described E Steiner (&) Á M Gastl Á T Becker Lehrstuhl fu¨r Brau- und Getra¨nketechnologie, Technische Universita¨t Mu¨nchen, Weihenstephaner Steig 20, 85354 Freising, Germany e-mail: Elisabeth.Steiner@wzw.tum.de Keywords Proteins Á Barley Á Malt Á Beer Á Haze formation Á Foam formation Proteins in barley and malt Barley (Hordeum vulgare L.) is a major food and animal feed crop It ranks fourth in area of cultivation of cereal crops in the world Barley is commonly used as raw material for malting and subsequently production of beer, where certain specifications have to be fulfilled These specifications are among others: germinative capacity, protein content, sorting (kernel size), water content, kernel abnormalities, and infestation Malting includes the controlled germination of barley in which hydrolytic enzymes are synthesized, and the cell walls, proteins, and starch of the endosperm are largely digested, making the grain more friable [1–3] Proteins in beer are mainly derived from the barley used The mature barley grain contains a spectrum of proteins that differ in function, location, structure, and other physical and chemical characteristics Barley seed tissues have different soluble protein contents and distinct proteomes The three main tissues of the barley seed are the aleurone layer, embryo, and starchy endosperm that account for about 9, 4, and 87%, respectively, of the seed dry weight [4, 5] The level of protein in barley is an important determinant in considering the final product quality of beer, for example for cultivar identification or as an indication of malting quality parameters [4], and it is influenced by soil conditions, crop rotation, fertilization, and weather conditions For malting barley, the balance between carbohydrates and proteins is important, since high protein content reduces primarily the amount of available carbohydrates Proteins present in barley seeds are important quality 123 192 determinants During malting, barley storage proteins are partially degraded by proteinases into amino acids and peptides which are critical for obtaining high-quality malt and therefore high-quality wort and beer [1, 6, 7] Germination provides the necessary hydrolytic enzymes to modify the grain, which are, in the case of proteins, endoproteases, and carboxypeptidases These enzymes degrade storage proteins, especially prolamins (hordeins) and glutelins [8] and produce free amino acids during germination by cleavage of reserve proteins in the endosperm [9] According to Mikola [10], there exist five seine carboxypeptidases in germinating barley, which have complementary specificities and mostly an acidic pH optimum All of these carboxypeptidases consist of identical subunits, each compose of two polypeptide chains, cross-linked by disulphide bridges [9, 11, 12] Barley malt endoproteases (EC.3.4.21) develop multiple isoforms mainly during grain germination and pass through kilning almost intact [8, 13] Jones [13–17] surveyed those enzymes and their behavior during malting and mashing Cysteine proteases (EC 3.4.22) are clearly important players in the hydrolysis of barley proteins during malting and mashing However, it seems likely that they not play as predominant a role as was attributed to them in the past [15, 16, 18–22] It has been found out that metalloproteases (EC 3.4.24) play a very significant role in solubilizing proteins, especially during mashing at pH 5.8–6.0 [23] All current evidence suggests that the serine proteases (EC 3.4.21) play little or no direct role in the solubilization of barley storage proteins [23, 24], even though they comprise one of the most active enzyme forms present in malt [22] While none of the barley aspartic proteases (EC 3.4.23), that have been purified and characterized, seem to be involved in hydrolyzing the seed storage proteins, it is likely that other members of this group participate Jones [17] investigated endoproteases in malt and wort and discovered that they were inactivated at temperatures above 60 °C Jones et al [6] examined the influence of the kilning process toward the endoproteolytic activity These enzymes were affected by heating at 68 and 85 °C, during the final stages of kilning, but these changes did not influence the overall proteolytic activity Other proteins are involved in protein folding, such as protein disulfide isomerase (EC 5.3.4.1), which catalyzes the formation of protein disulfide bridges Due to their heat-sensitivity, proteinases are inactivated when the temperature rises above 72 °C [25–30] They are almost totally inactive within 16 [1, 7, 13] Summarizing the most important factors for the protein composition, as origin in finished beer are barley cultivar and the level of protein modification during malting, which is judged by malt modification which is conventionally measured in the brewing industry as the Kolbach index (soluble nitrogen/total nitrogen*100) [31, 32] 123 Eur Food Res Technol (2011) 232:191–204 To get an overview of the main proteins in malt and beer, the most studied proteins are described in the next paragraphs Proteins can be classified pursuant to their solubility Osborne [33–37] took advantage of this fact and developed a procedure to separate the proteins Proteins are divided into water-soluble (albumins), salt-soluble (globulins), alcohol-soluble (prolamins), and alkali-soluble (glutelins) fractions [34–36, 38, 39] Osborne fractionation is a relatively simple, fast, and sensitive extraction–analysis procedure for the routine quantitation of all protein types in cereals in relative and absolute quantities, including the optimization of protein extraction and of quantitative analysis by RP-HPLC High-performance liquid chromatography (or high-pressure liquid chromatography, HPLC) is a chromatographic technique that can separate a mixture of compounds and is used in biochemistry and analytical chemistry to identify, quantify, and purify the individual components of the mixture Not only Osborne fractionation and HPLC but also several other methods exist to separate and identify proteins in barley, malt, wort, and beer To get an overview over the applications of the described methods in the review, a description follows in the next paragraphs Several authors [5, 39–60] characterized barley and barley malt proteins with help of 2D-PAGE Other authors [25, 26, 29, 30, 32, 41, 61–65] used 2D-PAGE and mass spectrometry to fingerprint the protein composition in beer and to evaluate protein composition with regard to foam stability and haze formation Klose [39] followed protein changes during malting with the help of a Lab-on-a-Chip technique and validated the results with 2D-PAGE Iimure et al [64] invented a protein map for the use in beer quality control This beer proteome map provides a strong detection platform for the behaviors of beer quality–related proteins, like foam stability and haze formation The nucleotide and amino acid sequences defined by the protein identification in the beer proteome map may have advantages for barley breeding and process control for beer brewing The nucleotide sequences also give access to DNA markers in barley breeding by detecting sequence polymorphisms Hejgaard et al [66–73] worked with immunoelectrophoresis and could identify several malt and beer proteins Shewry et al [54, 74–78] determined several methods for investigation of proteins in barley, malt, and beer mainly with different electrophoresis methods Asano et al [62, 63] worked with size-exclusion chromatography, immunoelectrophoresis and SDS–PAGE Mills et al [79] made immunological studies of hydrophobic proteins in beer with main focus and foam proteins He discovered that the most hydrophilic protein group contained the majority of the proteinaceous material but it also comprised polypeptides with the least amount of tertiary structure Eur Food Res Technol (2011) 232:191–204 193 Fig Shematic longitudinal section of a barley grain [81] Vaag et al [28] established a quantitative ELISA method to identify a 17 kDa Protein and Ishibashi et al [80] used an ELISA technique to quantify the range of foam-active protein found in malts produced in different geographic regions, and using different barley cultivars Van Nierop et al [30] used an ELISA technique to follow LTP1 content during the brewing process Osman et al [18–20] investigated the activity of endoproteases in barley, malt, and mash Hence, protein degradation during malting and brewing is very important for the later beer quality (mouthfeel, foam, and haze stability) It was suggested that estimation of the levels of degraded hordein (the estimation of the levels of hordein degraded during malting truly reflects the changes in proteins during malting and can measure the difference in barley varieties related to proteins and their degrading enzymes) during malting is a sensitive indicator of the total proteolytic action of proteinases as well as the degradability of the reserve proteins And therefore, it is possible to predict several beer quality parameters according the total activity of all proteinases and the protein modification during malting To obtain good results, those separation and identification methods can be combined Van Nierop et al [30], for example, used ELISA, 2D-PAGE, RP-HPLC, electrospray mass spectrometry (ESMS), and circular dichroism (CD) spectrophotometry to follow the changes of LTP1 before and after boiling Since there exist various methods to separate and identify proteins in this review, an overview over existent proteins in barley, malt, wort, and beer is provided according to only one method, which is Osborne fractionation These fractions are described more closely in the next sections Barley glutelin About 30% of barley protein is glutelin that dissolves only in diluted alkali [54] Glutelin is localized almost entirely in the starchy endosperm (Fig 1), is not broken down later on, and passes unchanged into the spent grains [81, 82] Glutelin is the least well-understood grain protein fraction This is partly because the poor solubility of the components has necessitated the use of extreme extraction conditions and powerful solvents which often cause denaturation and even degradation (e.g., by the use of alkali) of the proteins, rendering electrophoretic analysis difficult Also, because glutelin is the last fraction to be extracted, it is frequently affected by previous treatments and contaminated with residual proteins from other fractions, notably prolamins, which are incompletely extracted by classical Osborne procedures [83] It has not been possible to prepare an undenatured glutelin fraction totally free of contaminating hordein [3] Barley prolamin The prolamin in barley is called hordein and it constitutes about 37% of the barley protein It dissolves in 80% alcohol and part of it passes into spent grains Hordein is a lowlysine, high-proline, and high-glutamine alcohol-soluble protein family found in barley endosperm (Fig 1) It is the major nitrogenous fraction of barley endosperm composing 35–55% of the total nitrogen in the mature grain [1, 84–86] Hordeins are accumulated relatively late in grain development, first being observed about 22 days after anthesis (when the grain weighs about 33% of its final dry weight) and increasing in amount until maximum dry weight is reached [87] The major storage proteins in most cereal grains are alcohol-soluble prolamins These are not single components, but form a polymorphic series of polypeptides of considerable complexity [88] Hordein is synthesized on the rough endoplasmic reticulum during later stages of grain filling and deposited within vacuoles in protein bodies [89, 90] Silva et al [91] ascertained that the exposure of hordeins to a proteolytic process during germination reduces its content and originates in less hydrophobic peptides 123 194 Some malt water–soluble proteins result from the hordein proteolysis Hordeins are the most abundant proteins in barley endosperm characterized by their solubility in alcohol These storage proteins form a matrix around the starch granules, and it is suggested that their degradation during malting directly affects the availability of starch to amylolytic attack during mashing [92] Shewry [75, 77] divided the hordeins according to their size and amino acid composition in four different fractions (A-D), dependent on their size and amino acid composition A-hordeins (15–25 kDa) seem to be no genuine storage proteins as they contain protease inhibitors and a-amylases B-hordeins (32–45 kDa) are rich in sulfur content and are, with 80%, the biggest hordein fraction B-hordeins have a general structure, with an assumed signal peptide of 19 aminoacid residues, a central repetitive domain rich in proline and glutamine residues, and a C-terminal domain containing most of the cysteine residues are encoded by a single structural locus, Hor2, located on the short arm of chromosome 1H(5), 7–8 cM distal to the Hor1 locus which codes for the C-hordeins C-hordeins (49–72 kDa) are low in sulfur content, and D-hordeins ([100 kDa) are the largest storage proteins and are encoded by the Hor3 locus located on the long arm of chromosome 1H(5) [85, 87, 93, 94] Cereal prolamins are not single proteins but complex polymorphic mixtures of polypeptides [54] During malting, disulfide bonds are reduced and B- and D-hordeins are broken down by proteolysis Well-modified malt contains less than half the amount of hordeins present in the original barley D-hordeins are degraded more rapidly than their B-type counterparts, and the latter are more rapidly degraded than C-hordeins [3, 95] Barley albumins and globulins Many researchers extract a combined salt-soluble protein fraction, because water extracts contain globulins as well as albumins The two classes of proteins may be separated by dialysis, but there is considerable overlap between the two [83] Albumins and globulins consist mainly of metabolic proteins, at least in the cereal grains [96] and are found in the embryo and the aleurone layer, respectively [81, 82] Whereas prolamins are degraded during germination, albumins and other soluble proteins increased during the germination process [92] Globulins The globulin fraction of barley is called edestin It dissolves in dilute salt solutions and hence also in the mash It forms about 15% of the barley protein Edestin forms components (a, b, c, and d) of which the sulfur-containing b-globulin does 123 Eur Food Res Technol (2011) 232:191–204 not completely precipitate even on prolonged boiling and can give rise to haze in beer Enzymes and enzyme-related proteins are mainly albumins and globulins [42] Albumins The albumin of barley is called leucosin It dissolves in pure water and constitutes about 11% of the protein in barley During boiling, it is completely precipitated a-Amylase, protein Z, and lipid transfer proteins are barley albumins and are important for the beer quality attributes: foam stability and haze formation [97] Albumins can be further divided into protein Z and lipid transfer proteins as functional proteins Protein Z Protein Z belongs to a family of barley serpins and consists of at least four antigenically identical molecular forms with isoelectric points in the range 5.55—5.80 (in beer: 5.1–5.4), but same molecular mass near 40 kDa [1, 55, 67, 68, 98] Protein Z is hydrophobe and exists in free and bound forms in barley, like a-amylase, and there also exist heterodimers Protein Z contains cysteine and 20 lysine residues per monomer molecule and is relatively rich in leucine and other hydrophobic residues Protein Z accounts for 5% of the albumin fraction and more than 7% in some high-lysine barleys [67, 99] The content of protein Z in barley grains depends on the level of nitrogen fertilization [67, 100] Protein Z makes up to 20–170 mg/L of beer protein [79] In mature seeds, protein Z is present in thiol bound forms, which are released during germination [101] The function of the protein is at present unknown but it is known that it is deposited specifically in the endosperm responding to nitrogen fertilizer, similar to the hordein storage proteins The synthesis is regulated during grain development at the transcriptional level in dependence of the supply of nitrogen [98, 100, 102, 103] It is stated that upregulation of transcript levels could be effectuated within hours, if ammonium nitrate was supplied through the peduncle, and equally rapid reduced when the supply was stopped [103] Finnie et al [49] investigated the proteome of grain filling and seed maturation in barley They identified a group of proteins that increased gradually both in intensity and abundance, during the entire examination period of development and were identified as serpins Also Sorensen [55] and Giese [98] could detect the expression of protein Z4 (a subform of protein Z) only during germination Protein Z4 has an expression profile similar to b-amylase and seed storage proteins (hordeins) Three distinct serpin sequences from barley could be found in the databases SWISSPROT and TREMBL: protein Z4, protein Z7, and protein Zx These different protein Eur Food Res Technol (2011) 232:191–204 Z forms are thought to have a role as storage proteins in plants, due to their high ‘‘Lys’’ content and the fact that serpin gene expression is regulated by the ‘‘high-Lys’’ alleles lys1 and lys3a [49, 104] Hejgaard et al [68] suggest that the precursors of protein Z originate from chromosomes and 7, and thus, they are named protein Z4 and protein Z7 Rasmussen and co-workers [105] were able to estimate the size of protein Z mRNA at 1.800 b This is sufficient to code for the 46.000 or 44.000 MW precursor peptides found in vitro translations plus leave 400–500 b for the 50 and 30 non-coding regions Doll [106] and Rasmussen [107] suggest that protein Z could be a candidate for modulation of the barley seed protein composition to balance the nutritional quality of the grain Giese and Hejgaard [98] found out that during germination, protein Z becomes the dominant protein in the salt-soluble fraction in developing barley The proteins in barley malt are known to be glycated by D-glucose, which is a product of starch degradation during malting [108] Bobalova et al [109] investigated in their research the glycation of protein Z and found out that protein Z glycation is detectable from the second day of malting The role of protein Z in beer is described more detailed in the sections foam and haze formation Lipid transfer protein Lipid transfer proteins (LTPs) are ubiquitous plant lipidbinding proteins that were originally identified by their ability to catalyze the transfer of lipids between membranes LTPs are abundant soluble proteins of the aleurone layers from barley endosperm The compact structure of the barley LTP1 comprises four helices stabilized by four disulfide bonds and a well-defined C-terminal arm with no regular secondary structure [110] which is shown in Fig 2, where a 3D and surface protein of barley LTP native protein (here called 1LIP, red) is shown [111] In comparison with other plant lipid transfer proteins, the barley protein has a small hydrophobic cavity but is capable of binding different lipids such as fatty acids and acyl-CoA [25, 112, 113] According to molecular mass, this multigene family is subdivided into two subfamilies, ns-LTP1 (9 kDa) and ns-LTP2 (7 kDa); both located in the aleurone layer of the cereal grain endosperm [56, 114] LTP1 and LTP2 are expressed in barley grain but only LTP1 has been able to be detected in beer LTP1 is claimed to be an inhibitor of malt cysteine endoproteases [14, 115] The role of LTP1 in beer is described more detailed in the sections foam and haze formation Protein Z and LTP1 Evans [116, 117] investigated the influence of the malting process on the different protein Z types and LTP1 He 195 Fig 3D and surface protein of barley LTP native protein (1LIP, red) is shown [111] discovered that the amount of LTP1 did not change during germination but a significant proportion of the bound/latent protein Z was converted into the free fraction He claims that during germination, proteolytic cleavage in the reactive site loop converts protein Z to a heat and protease stable forms, and hence, they can survive the brewing process He ascertained also that kilning reduced the amount of protein Z and LTP1 [66, 118] Evans [116] analyzed feed and malting barley varieties and could not find any differences in the level of protein Z and LTP1 He also ascertained malt-derived factors that influence beer foam stability, such as protein Z4, b-glucan, viscosity, and Kohlbach index Beer components (protein Z4, free amino nitrogen, b-glucan, arabinoxylan, and viscosity) were correlated with foam stability [117] Protein Z4, protein Z7, and LTP1 have been shown to act as protease inhibitors [116, 119, 120] Proteins in wort and beer Proteins influence the whole brewing process not only in the form of enzymes but also in combination with other substances such as polyphenols As enzymes, they degrade starch, b-glucans, and proteins In protein–protein linkages, they stabilize foams and are responsible for mouthfeel and flavor stability, and in combination with polyphenols, they are thought to form haze As amino acids, peptides, and sal ammoniac, they are important nitrogen sources for yeast [121] Only about 20% of the total grain proteins are water soluble Barley water-soluble proteins are believed to be resistant to proteolysis and heat coagulation and hence pass through the processing steps, intact or somewhat modified, to beer [116, 122, 123] Several aspects of the brewing process are affected by soluble proteins, peptides, and/or 123 196 Eur Food Res Technol (2011) 232:191–204 Table Enzymes in barley and barley malt [1, 7, 166, 167] Cytolysis Enzyme Substrate Product b-glucan-solubilase Matrix linked b-glucan Soluble, high molecular weight b-glucan Endo-b-(1-3) glucanase Soluble, high molecular weight b-glucan Low molecular weight b-glucan, cellobiose, laminaribiose Endo-b-(1-4) glucanase Exo-b-glucanase Soluble, high molecular weight b-glucan Cellobiose, laminaribiose Low molecular weight b-glucan, cellobiose, laminaribiose Glucose Xylanase Hemicellulose b-D-Xylose Proteins Peptides, free amino acids Carboxypeptidase Proteins, peptides Free amino acids Aminopeptidase Proteins, peptides Free amino acids Dipeptidase Dipeptides Free amino acids Proteolysis Endopeptidase Amylolysis a-amylase Other High and low molecular weight a-glucans Melagosaccharides, oligosaccharides b-amylase a-glucans Maltose Maltase Maltose Glucose Limit dextrinase Limit dextrins Dextrins Pullulanase Linear amylose fractions Lipase a-1,6-D-glucans in amylopectin, glykogen, pullulan Lipids, lipidhydroperoxide Lipoxygenase Free fatty acids Fatty acid hydroperoxide amino acids that are released No more than one-third of the total protein content passes into the finished beer which is obtained throughout mainly two processes; mashing and the wort boiling Mashing is the first biochemical process step of brewing and completes the enzymatic degradation started during malting Enzymes synthesized during malting are absolutely essential for the degradation of large molecules during mashing These enzymes are displayed in Table [1, 7] The three biochemical basic processes taking place during malting are cytolysis, proteolysis, and amylolysis, which are indicated by b-glucan, FAN, and extract concentration, respectively In order to get good brews, part of the insoluble native protein must be converted into ‘‘soluble protein’’ during malting and mashing This fraction comprises a mixture of amino acids, peptides, and dissolved proteins, and a major portion of it arises by proteolysis of barley proteins [23] During the brewing process, there are three possibilities to discard the (unwanted) proteinic particles The first opportunity is given during wort boiling, where proteins coagulate and can be removed in the ‘‘whirlpool’’ The second, during fermentation, where the pH decreases and proteinic particles can be separated by sedimentation The third step is during maturation of beer During maturation, proteins adhere on the yeast and can be discarded [124] It has also been demonstrated that yeast proteins are present in beer, but only as minor constituents [73] Beer contains *500 mg/L of proteinaceous material including a variety of polypeptides with molecular masses ranging from \5 to [100 kDa These polypeptides, which mainly 123 Glycerine, free fatty acids, fatty acid hydroperoxide originate from barley proteins, are the product of the enzymatic (proteolytic) and chemical modifications (hydrogen bonds, Maillard reaction) that occur during brewing, especially during mashing, where proteolytic enzymes are liable for those modifications [125] A beer protein may be defined as a more or less heterogeneous mixture of molecules containing the same core of peptide structure, originating from only one distinct protein present in the brewing materials [126] Jones [13–17] surveyed proteinases and their behavior during malting and mashing Proteinases are not active in beer anymore; hence, they are inactivated when the temperature rises above 72 °C, which happens already during mashing [1, 7, 13, 25–30] Proteins influence two main quality aspects in the final beer: 1st haze formation and 2nd foam stability In the following lines, these quality attributes are described in a more detailed way Haze formation Proteins play a major role in beer stability; hence, they are, beside polyphenols, part of colloidal haze There exist two forms of haze; cold break (chill haze) and age-related haze [127] Cold break haze forms at °C and dissolves at higher temperatures If cold break haze does not dissolve, age-related haze develops, which is non-reversible Chill haze is formed when polypeptides and polyphenols are bound non-covalently Permanent haze forms in the same manner initially, but covalent bonds soon form and Eur Food Res Technol (2011) 232:191–204 197 Table Distribution of hordeins in barley according to their size [75] Type MW (kDa) % of total hordeins A 10–16 [5 B 30–46 80–90 C 48–72 10–20 D [100 [5 insoluble complexes are created which will not dissolve when heated [128] Proanthocyanidins (condensed tannins) from the testa tissue (seed coat) of the barley grain are carried from the malt into the wort and are also found after fermentation of the wort in the beer There they cause precipitation of proteins and haze formation especially after refrigeration of the beer, even if it previously had been filtered to be brilliantly clear [129] Proteins, as the main cause of haze formation in beer, can be divided into two main groups: 1st proteins and 2nd their breakdown products Protein breakdown products are characterized by always being soluble in water and not precipitate during boiling Finished beer contains almost only protein breakdown products [126] The content of only mg/L protein is enough to form haze [118] Beer contains a number of barley proteins that are modified chemically (hydrogen bond formation, Maillard reaction) and enzymatically (proteolysis) during the malting and brewing processes, which can influence final beer haze stability Leiper et al [130, 131] found out that the mashing stage of brewing affects the amount of haze-active protein in beer If a beer has been brewed with a protein rest (48–52 °C), it may contain less total protein but more haze-active proteins because the extra proteolysis caused release of more haze causing polypeptides Asano et al [62] investigated different protein fractions and split them in categories: 1st high, 2nd middle and 3rd low molecular weight fractions being high molecular weight fractions: [40 kDa, middle molecular weight fraction: 15–40 kDa and low molecular weight fraction: \15 kDa Nummi et al [132] even suggested that acidic proteins derived from albumins and globulins of barley are responsible for chill haze formation (Table 2) Researchers proofed that proline-rich proteins are involved in haze formation [63, 65, 124, 127, 128, 130, 131, 133–137] Outtrup et al [138] say that haze-active proteins are known to be dependent on the distribution of proline within the protein Nadzeyka et al [127] suggested that proteins in the size range between 15–35 kDa comprised the highest amount of proline It was also investigated that proline and glutamic acid-rich hordeins, in the size range between 10–30 kDa, are the main initiators causing haze development [63, 74] b-Amylase, protein Z, and two chymotrypsin inhibitors have relatively high-lysine contents [100] Barley storage proteins that are available for hydrolysis are all proline-rich proteins [15] Dadic and Belleau [139, 140] on contrary say that there is no specific amino acid composition for haze-active proteins Leiper [130, 131] even says that not only the mainly consistence of proline and glutamic acid of the glycoproteins is responsible for causing haze but also that the carbohydrate component consists largely of hexose It was found out that the most important glycoproteins for haze formation are 16.5 and 30.7 kDa in size Glycation is a common form of non-enzymatic modification that influences the properties of proteins [109] Nonenzymatic glycation of lysine or arginine residues is due to the chemical reactions in proteins, which happen during the Maillard reaction [109] It is one of the most widely spread side-chain-specific modifications formed by the reaction of a-oxoaldehydes, reducing carbohydrates or their derivatives with free amine groups in peptides and proteins, such as e-amino groups in lysine and guanidine groups in arginine [141, 142] The proteins in barley malt are known to be glycated by D-glucose, which is a product of starch degradation during malting [108] D-glucose reacts with a free amine group yielding a Schiff base, which undergoes a rapid rearrangement forming more stable Amadori compounds Haze-sensitive proteins Polypeptides that are involved in haze formation are also known as sensitive proteins They will precipitate with tannic acid, which provides a mean to determine their levels in beer Proline sites of these polypeptides bind to silica gel hydroxyl groups so that haze-forming proteins are selectively adsorbed, since foam proteins contain little proline and are thus not affected by silica treatment [143] Removal of haze forming tannoids can be effected using PVPP [143] To assure colloidal stability, it is not necessary to remove all of the sensitive proteins or tannoids Identification of a tolerable level of these proteins can be used to define a beer composition at bottling that delivers satisfactory haze stability [94, 99] To prolong stability of beer, stabilization aids are used Haze-forming particles are removed with: (a) silica, which is used to remove prolinerich proteins that have the ability to interact with polyphenols to form haze in bright beer, or (b) PVPP, which is used to remove haze-active polyphenols Evans et al [144] investigated the composition of the fractions which were absorbed by silica This analysis revealed that the mole percentage of proline ranged between 33.2 and 38.0%, and of glutamate/glutamine between 32.7 and 33.0%, consistent with the proline/glutamine–rich composition of the hordeins [144] Iimure et al [65] stated in their studies that proteins adsorbed onto silica gel (PAS) are protein Z4, protein Z7, and trypsin/ amylase inhibitor pUP13 (TAI), rather than BDAI-1 123 198 (a-amylase inhibitor), CMb, and CMe La´zaro et al [145] investigated the CM proteins CMa, CMb, and CMe The CM proteins are a group of major salt-soluble endosperm proteins encoded by a disperse multigene family and act as serine proteinase inhibitors Genes CMa, CMb, and CMe are located in chromosomes 1, 4, and 3, respectively Protein CMe has been found to be identical with a previously described trypsin inhibitor Furthermore, Iimure et al [64] analyzed proline compositions in beer proteins, PAS, and haze proteins It was proofed that the proline compositions of PAS were higher (ca 20 mol%) than those in the beer proteins (ca 10 mol%), although those of the hazeactive proteins such as BDAI-1, CMb, and CMe were 6.6–8.7 mol% These results suggest that BDAI-1, CMb, and CMe are not predominant haze-active proteins, but growth factors of beer colloidal haze Serine proteinase inhibitors have also been called trypsin/a-amylase inhibitors, and it has been proposed that some of them might inhibit the activities of barley serine proteinases However, none have been shown to affect barley enzymes [16] Robinson et al [146] identified a polymorphism for beer haze-active proteins and surveyed by immunoblot analysis throughout the brewing process In this polymorphism, some barley varieties contained a molecular weight band at 12 kDa, while in other varieties, this band was absent Pilot brewing trials have shown that the absence of this 12 kDa protein conferred improved beer haze stability on the resulting beer This band was detected by a polyclonal antibody raised against a haze-active, proline/glutamine– rich protein fraction; it was initially assumed that the band was a member of the hordein protein family [144, 147] Foam formation Beer foam is an important quality parameter for customers Good foam formation and stability gives an impression of a freshly brewed and well-tasting beer Therefore, it is necessary to investigate mechanisms that are behind foam formation Beer foam is characterized by its stability, adherence to glass, and texture [148] Foam occurs on dispensing the beer as a result of the formation of CO2 bubbles released by the reduction in pressure The CO2 bubbles collect surface-active materials as they rise These surface-active substances have a low surface tension, this means that within limits they can increase their surface area and also, after the bubbles have risen, they form an elastic skin around the gas bubble The greater the amount of dissolved CO2 the more foam is formed But foam formation is not the same as foam stability Foam is only stable in the presence of these surface-active substances [1] Beer foam is stabilized by the interaction between certain beer proteins, for example LTP1, and isomerized 123 Eur Food Res Technol (2011) 232:191–204 hop a-acids, but destabilized by lipids [30, 148] The intention is to find a good compromise of balancing foampositive and foam-negative components Foam-positive components such as hop acids, proteins, metal ions, gas composition (ratio of nitrogen to carbon dioxide), and gas level, generally improve foam, when increased Whereas foam negatives, such as lipids, basic amino acids, ethanol, yeast protease activity, and excessive malt modification, decrease foam formation and stability Free fatty acids, which are extracted during mashing, have a negative effect on foam stability [64, 65, 80, 85, 88, 128–131, 166] Foam-positive proteins can be divided into high molecular weight proteins (35–50 kDa) and low molecular weight proteins (5–15 kDa) which primary originate from malt but in small amount can also originate from yeast [62, 73, 148] It is thought that during foam formation, amphiphile proteins surround foam cells and stabilize them by forming a layer They arrange themselves into bilayers, by positioning their polar groups toward the surrounding aqueous medium and their lipophilic chains toward the inside of the bilayer, defining a non-polar region between two polar ones [149] There are two main opinions concerning the nature of foaming polypeptides in beer The first position claims the existence of specific proteins which basically influence foam stability Those proteins are known as protein Z and LTP1 [150, 151] The second argument claims the existence of a diversity of polypeptides which stabilize foam; the more hydrophobic their nature, the more foam active they are [122, 152], like hordeins that are rich in proline and glutamine content and exhibit a hydrophobic b-turn-rich structure [74] KAPP [153] investigated the influence of albumin and hordein fractions from barley on foam stability, because both are able to increase the foam stability The ability to form more stabile foams seems to be higher by albumins than by hordeins Denaturation of these proteins causes an increase in their hydrophobic character and also in their foam stability This confirms the already known opinion that the more hydrophobic the protein, the better is the foam stability [122, 152] The foams from albumins are more stable than those from hordeins This may also be the reason for the increased ability of albumin fractions to withstand the presence of ethanol The foam stability of both albumins and hordeins is increased by bitter acids derived from hops Whereas the barley LTP1 does not display any foaming properties, the corresponding beer protein is surface active Such an improvement is related to glycation by Maillard reactions on malting, acylation on mashing, and structural unfolding on brewing which was ascertained by Perrocheau et al [25] During the malting and brewing processes, LTP1 becomes a surface-active protein that concentrates in beer foam [55] LTP1 is modified during boiling and this modified form influences foam stability [28, 150] The two Eur Food Res Technol (2011) 232:191–204 forms have been recovered in beer with marked chemical modifications including disulfide bond reduction and rearrangement and especially glycation by Maillard reaction The glycation is heterogeneous with variable amounts of hexose units bound to LTPs [112] The four lysine residues of LTP1 are the potential sites of glycation [112] Altogether, glycation, lipid adduction, and unfolding should increase the amphiphilic character of LTP1 polypeptides and contribute to a better adsorption at air–water interfaces and thus promote foam stability Van Nierop et al [30] established that LTP1 denaturation reduces its ability to act as a binding protein for foam damaging free fatty acids and therefore boiling and boiling temperature are important factors in determining the level and conformation of LTP1 and so enhance foam stability Perrocheau et al [25] showed that unfolding of LTP1 occurred on wort boiling before fermentation and that the reducing conditions are provided by malt extract Van Nierop et al [30] showed that the wort boiling temperature during the brewing process was critical in determining the final beer LTP1 content and conformation It was discovered that higher wort boiling temperatures (102 °C) resulted in lower LTP1 levels than lower wort boiling temperatures (96 °C) Combination of low levels of LTP1 and increased levels of free fatty acids resulted in low foam stability, whereas beer produced with low levels of LTP1 and free fatty acids had satisfactory foam stability LTP1 has been demonstrated to be foam promoting only in its heat denatured form [55, 150, 154] Perrocheau et al [26] investigated heat-stable, watersoluble proteins that influence foam stability Most of the heat stable proteins were disulfide-rich proteins, implicated in the defense of plants against their bio-aggressors, e.g., serpin-like chymotrypsin inhibitors (protein Z), amylase and amylase-protease inhibitors, and lipid transfer proteins (LTP1 and LTP2) Leisegang et al [95–97] identified LTP1 as a substrate for proteinase A, which degrades LTP1, but does not influence protein Z and may have a negative influence on beer foam stability Iimure et al [32] invented a prediction method of beer foam stability using protein Z, barley dimeric a-amylase inhibitor-1 (BDAI-1) and yeast thioredoxin and confirmed BDAI-1 and protein Z as foam-positive factors and identified yeast thioredoxin as a possible novel foam-negative factor Jin et al [155, 156] found out in their research that structural changes of proteins during the wort boiling process are independent of the malt variety It was discovered that barley trypsin inhibitor CMe and protein Z were resistant to proteolysis and heat denaturation during the brewing process and might be important contributors to beer haze formation Vaag et al [28] found a new protein of 17 kDa which seemed to influence foam stability even more than protein Z and barley like LTP1 She could support this theory by the 199 correlation of the content of this so called 17 kDa protein and the foam half-life of lager beers LTP1 and the 17 kDa protein exhibit some similarities; their tertiary structures are characterized by disulfide bridges, both are rich in cysteine and are modified during heating to a more foam promoting form Ishibashi et al [80] agrees that both malting and mashing conditions influence the foam-active protein levels in experimental mashes Proteinaceous materials in beer have as well been implicated in the stabilization of beer foam Molecular weight has been reported to be important for foaming potential, while the hydrophobicity of polypeptides has been cited as a controlling factor [62] Kordialik-Bogacka et al [157, 158] investigated also foam-active polypeptides in beer In contrary to Osman et al [123] in this investigation, it was confirmed that fermentation influences the protein composition of beer and particularly in beer foam Yeast polypeptides were also found in beer foam It was noted that, especially during the fermentation of high gravity wort, excessive foaming may occur, and this may be one of the reasons why beer brewed at higher gravities has a poor head It was detected that polypeptides of molecular weight about 40 kDa present in fermented wort and foam originated not only from malt but also from yeast cells Okada et al [159] studied on the influence of protein modification on foam stability during malting They found that the foam stability of beer samples brewed from barley malts of cultivars decreased as the level of malt modification increased, but the foam stability of another cultivar did not change In this research, they defined BDAI-I as an important contributor to beer foam stability Conclusion Proteins not only influence haze formation; furthermore, they play an important role for mouthfeel and foam stability These aspects are important for brewers, since consumer judge beer also according to these quality attributes As it is known, most foam-positive proteins are also haze active, Evans et al [144] made an investigation to immunologically differentiate between those two protein forms (foam and haze-active proteins) and concluded that no barley variety or growing condition have any significant influence on beer stability It was also demonstrated in a regression analysis that a prediction of foam stability is not possible, which underlines the complexity of these problems It is suggested that both foam-active and foam-negative components should be measured and that the amount of hordeins and protein Z4 are somehow related It was also ascertained that foam and haze-active proteins share some epitopes and that oxygen during the brewing process influence haze stability of beer [147] 123 123 a Papua New Guinea Western Highlands -5.23/144.33 -8.75/125.40 0/102 18.2/-76.83 19 19 57 58 118 19 56 60 19 55 118 19 54 59 19 53 231 17 51 52 17 49 50 294 47 48 752 294 – 429 188 – 317 899 156 93 93 71 99 33 33 – 486 Distance to the sea (Km) 45 46 44 43 42 41 40 39 38 37 36 35 34 33 32 30 31 Sample Negative latitude values correspond to South latitudes and negative longitude values correspond to West longitudes Melanesia Mandeling Ermera East Timor Ecuador Indonesia San Cristobal Island Peru South-East Asia -20.47/-45.97 -10.57/-75.40 Fazenda Nossa Yanesha Sierra del Centro – Unknown Jamaica -20.25/-42.33 Zona da Mata -0.38/-89.72 – -22.37/-46.95 -22.12/-45.12 Fazenda S Benedito Mogiana -20.52/-47.38 Fazenda da Terra Fazenda Muzambo Brazil 14.28/-85.18 14.20/-89.21 Santa Rita 14/-89 San Miguel 14.55/-90.93 Acatenango San Antonio 15.32/-91.97 9.7/-84.1 – 16.75/-93.12 Coordinatesa (latitude/longitude) Huehuetenango Caribbean South America Nicaragua El Salvador Guatemala San Marcos Costa Rica de Tarrazu Santa Rita Tuxtla (Chiapas) Mexico Central America Location Country Region Table continued 2,000 2,000 1,195 1,195 1,195 1,195 1,195 1,195 30 750 750 1,492 1,492 4,500 857 4,500 – 877 607 – 900 700 200 1,100 1,100 1,586 1,908 1,450 1,450 – 750 Altitude (m) 2,970 2,970 1,987 1,987 1,987 1,987 1,987 1,987 2,304 2,439 2,439 – – 594 1,580 594 – 1,580 1,580 – 1,580 1,580 – 2,278 2,278 1,218 1,021 1,790 1,790 – 690 Average annual precipitation (mm) 364 Eur Food Res Technol (2011) 232:361–373 Eur Food Res Technol (2011) 232:361–373 between samples to control stability and to allow drift correction when necessary Precision was 0.14 % Strontium isotope ratio measurement by multicollector inductively plasma mass spectrometry (MC-ICP-MS) 365 intensities of 88Sr from to volts Fractionation effect of the column extraction procedure on the Sr isotopic ratio was checked according to Schultheis [24] and proved to be insignificant Instrumentation Reagents Pro analysi (p.a.) grade 65% HNO3 (Merck, Darmstadt, Germany) was subboiled doubly in an ultrapure quartz apparatus (MLS DuoPur, MLS, Leutkirch im Allga¨u, Germany) Deionised water (18 MX cm; REWA HQ5 Austria Wasseraufbereitung, Guntramsdorf, Austria) was subboiled prior to usage as well Subboiled HNO3 and 31% H2O2 (p.a., Merck) were used for digestion Polyethylene flasks and cartridges as well as polypropylene tubes and lids were cleaned sequentially with HNO3 (10% w/w) and HNO3 (1% w/w) and rinsed with deionised water before use Dilution of standards and samples was performed gravimetrically with HNO3 (1% w/w), prepared from subboiled water and doubly subboiled HNO3 Mass bias correction was performed by analysing a 20 ng g-1 solution of SRM 987 SrCO3 (NIST, Gaithersburg, MD, USA) The certified 87Sr/86Sr ratio is 0.71034 ± 0.00026, whereas a generally ‘accepted value’ of the 87Sr/86Sr ratio for this reference material is reported in the literature as 0.710263 ± 0.000016 (the error represents a standard deviation of 2r from the external reproducibility) [23] Sample preparation Four to six beans (amounting to about 1.0 g) were grinded in a Retsch mill Type MM2, three times for to obtain particle sizes of less than mm Approximately 0.5 g of the grinded material was directly weighed into Teflon bombs for subsequent microwave-assisted digestion (MLS 1200mega, MLS, Leutkirch im Allga¨u, Germany) Concentrated double-subboiled HNO3 (6 mL) and H2O2 (1 mL) were used as digestion reagents Details are presented elsewhere [14] The samples were finally transferred into 20-mL flasks, filled with HNO3 (1% w/w) to 20 g and stored at °C for further analysis A digestion blank was prepared with each digestion batch Screening of the solutions for Rb and Sr prior and after separation was performed by using a quadrupole-based inductively coupled plasma mass spectrometer ICP-MS (Elan DRCe, Perkin Elmer, Waltham, MC, USA) Sr isotope ratio measurements of the final solutions were accomplished using a double-focusing multicollector inductively coupled plasma mass spectrometer (MC-ICPMS) (Nu Plasma, Nu Instruments, Wrexham, UK) coupled to a membrane desolvating system (DSN-100, Nu Instruments) The DSN-100 instrument was equipped with a PFA nebuliser (MicroFlow nebuliser, Elemental Scientific, Omaha, NE, USA) and a spray chamber with additional hot gas flow to eliminate condensation and droplet formation The multicollector inductively coupled plasma mass spectrometer is equipped with a collector configuration consisting of 12 Faraday cups and three ion counters All isotopes in this work (82Kr, 83Kr, 84Sr, 85Rb, 86Sr, 87Sr, 88Sr) were measured simultaneously using Faraday cups Experimental parameters of the MC-ICP-MS including nebuliser gas, rf power and ion transfer lens potentials were optimised to achieve the maximum ion intensity for 88Sr, using NIST SRM 987 solution with a concentration of 20 ng-1 g The operation parameters are given in Table Blank correction and mass bias correction were performed according to previous measurements [14] Data analysis Statistical analysis of data (principal component analysis and Spearman’s correlation) was performed with Statistica 8.0 software (Statsoft Inc., USA) and MatLab software (version R2007b) (The MathWorks, Inc., USA) Spearman’s rank correlation coefficient was calculated between analytical measurements and available climatic and geographical data A level of significance of p \ 0.05 for Spearman’s correlation was the criteria for selecting climatic and geographical factors that were included as variables in principal component analysis Strontium/matrix separation The obtained digestion solutions of green coffee bean samples were separated according to Swoboda et al [14], using Eichrom Sr resin (Eichrom Industries, Darien, IL, USA) The solutions were diluted after separation to a final Sr concentration of about 20 ng g-1 to obtain optimum signal Results and discussion Strontium isotope abundance ratios of green coffee 87 Sr/86Sr isotope abundance ratios along with Sr concentration determined for 60 green coffee bean samples are 123 366 Eur Food Res Technol (2011) 232:361–373 Table Operating parameters and scheme of the monitored isotopes for the Sr measurements Nu Plasma settings Rf power 1,300 W Auxiliary gas flow rate/cooling gas flow rate 0.75 mL min-1/13.0 mL min-1 Sample uptake rate Sample/skimmer cone 100 lL min-1 Ni Nebuliser Perfluoroalkoxy nebuliser Sampling mode blocks of 10 measurements Measurement time 10 per sample Mass analyser pressure \10-8 mbar Background/baseline determination HNO3 (1% w/w) Washout time Axial mass/mass separation 86.05/0.5 Detection system 12 Faraday collectors Cups L5 L4 L3 L2 L1 Isotope 82 Kr 83 Kr 84 Sr Ax H1 H2 H3 H4 H5 H6 85 Rb 86 Sr 87 Sr 88 Sr DSN-100 nebuliser settings Nebuliser pressure bar (30 psi) Hot gas flow 0.7–0.9 L min-1 Membrane gas flow Spray chamber temperature L min-1 112 °C Membrane temperature 122 °C shown in Table Coffees from Rwanda, Malawi, Zambia, Peru, East Timor and a part of the samples from Brazil showed 87 Sr/86Sr values higher than 0.710 (Table 3) All other coffee samples had lower 87Sr/86Sr isotopes abundance ratios American coffees had variable 87Sr/86Sr values in the range of 0.7041–0.7155 Asian coffee from Papua New Guinea and Indonesia revealed significantly different 87 Sr/86Sr values from East Timor coffee African green coffee beans had 87Sr/86Sr range of 0.7047–0.7148 The latter allowed a clear discrimination between the group of coffees from Rwanda, Malawi and Zambia and the other three geographical origins, UR Tanzania, Kenya and Ethiopia Samples from Kenya showed 87Sr/86Sr range from 0.7061 to 0.7075 (Table 3) Kenyan samples (except samples 21 and 22; Table 3) originated from Mount Kenya and had been cultivated in soils with moderate to high fertility Kenyan coffee from Kirimiri (samples 21 and 22; Table 3) came from an area of deep, red, strongly weathered acid soil with low fertility common in wet (sub) tropical climates (Exploratory Soil Map of Kenya, [21]) Both regions are rich in mugearites (basalts) which have an expected 87Sr/86Sr range from 0.702 to 0.707 87Sr/86Sr values determined in Kenyan coffees (Table 3) overlap with this range 87Sr/86Sr values found for all Kenyan coffees are in the same range of values reported for parent rock at the same locations (Table 3) 123 Green coffee from UR Tanzania had 87Sr/86Sr values from 0.7047 to 0.7072 (Table 3) UR Tanzania sample came from a region classified as Eutric Cambisol (Provisional Soils Map of Tanzania, [21]) and had a higher 87 Sr/86Sr ratio (0.7072) compared with the other Tanzanian coffees that stem from Mount Kilimanjaro, where Eutric Nitisols prevail (Provisional Soils Map of Tanzania, [21]) In this region, EarthChem [21] reports the existence of basalts (absarokite; GEOROC 118466, Table 3) with higher 87Sr/86Sr values (0.7063) comparing with Mount Kilimanjaro Basalts in locations at Mount Kilimajaro show lower values of 87Sr/86Sr This is in agreement with the values determined for coffee from Kilimanjaro and with a87Sr/86Sr range from 0.702–0.707 which is reported for basaltic rocks, specifically ocean basalts [25] Coffee from Ethiopia had an 87Sr/86Sr ratio ranging from 0.7073 to 0.7077 which is higher than the 87Sr/86Sr ratio of parent rock at that location (Table 3) The difference may be explained by the fact that the Sr in the mobile phase can differ from the composition of the total rock [26] This hypothesis is taken into account in the case of coffee plants and beans from certain locations (e.g coffee from Ethiopia and Zambia and some of the coffees from Brazil, Table 3) Green coffees from Zambia, Malawi and Rwanda show the highest 87Sr/86Sr values among African coffees (Table 3) Malawi samples 23 and 24 came from a location with soil characterised as Lithosol (Soil Map of Malawi, [21]) with clay minerals formed from silicate bearing rocks This may explain 87Sr/86Sr ratios found in coffee that are similar to values for silicate rocks (C0.710) Rwanda samples and came from a location rich in silicate rocks (87Sr/86Sr C 0.710) like paragneiss and orthogneiss (Carte Lithologique du Rwanda, [21]) 87Sr/86Sr ratios (0.7140 and 0.7144, Table 3) of Rwandese coffees match the range of values reported for this type of parent rock 87 Sr/86Sr values allow the discrimination between the different American coffees (Ecuador, Jamaica, Nicaragua, El Salvador, Brazil and Hawaii), as well, with the exception of samples from Mexico which show overlapping isotope ratios with coffees from Hawaii, part of the Brazilian coffees Coffee from Hawaii was produced in a region of soil over pahoehoe lava bedrock at the Island of Hawaii (General Soil Map of Island of Hawaii, [21]) Values of 87 Sr/86Sr found for Hawaiian coffee (Table 3) are in accordance with the values of 87Sr/86Sr reported for ocean basalts (0.702 to 0.707) We could not obtain further information about parent rock nature at the Mexican locations included in this study, neither by the analysis of soil or geological maps or by consulting database GEOROC For this reason, Sr isotopic composition of Mexican Pacific 30.3 31.2 29.5 25.6 28.2 10 Lunji Mount Kilimanjaro UR Tanzania (Hawaii island) Hawaii Kona Greenwell estate Mubuyu Estate Peruzu Zambia Ludwing-Mzuzu Zimbabwe Malawi Kirimiri Mount Kenya 30.8 Kenya 34.8 31.7 33.6 34.1 34.1 28.1 16 17 18 19 20 29 28 27 26 25 29.7 29.2 26.6 30.3 33.4 29.9 26.8 23 24 31.8 30.9 15 22 33.8 14 29.6 29.6 13 21 31.7 31.6 11 12 27.2 31.5 30.9 Gatare´ d18Obean(%) Yirga Che´fe Sample Rwanda Eastern Africa Location Ethiopia Country -3.1 -3.1 -3.1 – -5.3 -5.4 -5.4 -3.9 -3.9 -10.2 -10.2 -10.2 -10.2 -10.2 -10.2 -10.2 -10.2 -10.2 -10.2 -2.7 -2.7 -2.7 -2.7 -4.6 -1.8 -1.8 -1.8 -4.0 -4.0 Mean annual d18Oprecipitation (%)a 0.7063 0.7067 0.7059 0.7169 0.7121 0.7148 0.7131 0.7072 0.7073 0.7063 0.7075 0.7066 0.7072 0.7067 0.7065 0.7062 0.7074 0.7074 0.7061 0.7047 0.7058 0.7047 0.7047 0.7072 0.7073 0.7074 0.7077 0.7140 0.7144 Sr/86Srbean (mean)b 87 2.8 4.8 3.1 3.4 6.0 6.2 5.4 6.7 7.6 7.1 8.1 5.2 8.1 6.3 6.1 7.8 6.0 6.4 7.6 26.5 2.6 25.0 21.8 9.2 4.2 3.1 3.1 7.7 7.6 Srbean (lg g-1)c 0.7041 (0.7038–0.7047) – – – 0.7068 (0.7061–0.7073) 0.7068 (0.7061–0.7073) 0.7035 0.7063 0.7035 (0.7034–0.7035) 0.7067 (0.7054–0.7080) Mean 87Sr/86Sr in parent rock (range)d 25 – – – 793 793 470 1,034 625 941 Mean Sr in parent rock (ppm)d GEOROC 28810/12/13/14 – – – GEOROC 89749, 118458, 118400 GEOROC 89749, 118458, 118400 GEOROC 113112 GEOROC 118466 GEOROC 73614, 73622, 73621 GEOROC 85572, 85576/7/ 9, 85592/7, 85607, 85629, 85636/7, 85641 EarthChem referenced Sr/86Sr of the 60 green coffees included in this study and strontium concentration and isotope ratio of parent rock reported at the geographical coordinates (whenever 87 Region Table d18O and available) Eur Food Res Technol (2011) 232:361–373 367 123 123 Western Highlands 25.8 58 20.0 25.4 57 60 23.7 56 18.7 21.3 55 59 26.2 54 – -11.1 -11.1 -7.5 -7.5 -7.5 -7.5 -7.5 -7.5 -7.0 -5.2 -5.2 -6.4 -6.4 -13.5 -13.5 -5.4 0.7044 0.7042 0.7259 0.7296 0.7285 0.7227 0.7159 0.7270 0.7062 0.7053 0.7057 0.7052 0.7049 0.7112 0.7127 0.7154 0.7077 0.7126 0.7077 0.7139 0.7155 0.7068 0.7047 0.7041 0.7041 0.7045 0.7067 0.7051 0.7064 0.7064 0.7076 Sr/86Srbean (mean)b 87 5.0 5.9 4.0 5.1 4.9 3.7 3.2 5.2 8.9 6.7 – 2.2 2.5 3.6 3.7 5.6 4.4 2.9 4.2 6.6 9.0 4.2 4.7 5.8 5.4 3.8 4.4 9.7 6.8 4.0 7.7 Srbean (lg g-1)c 0.7044 (0.7036–0.7054) 0.7109 0.7047 (0.7045–0.7047) 0.7055 (0.7035–0.7088) 0.7030 (0.7024–0.7038) – 0.7125 (0.7057–0.7288) – – 0.7055 (0.7055–0.7056) – 0.7060 (0.7059–0.7061) 0.7057 0.7037 0.7032 0.7055 (0.7039–0.7066) – 0.7039 (0.7035–0.7049) – Mean 87Sr/86Sr in parent rock (range)d 759 268 375 523 334 – 363 – – 377 – 357 452 – 482 169 – 362 – Mean Sr in parent rock (ppm)d GEOROC 82070/1/2/3/4/5/ 7/8/9 82081/2/3/4/5 GEOROC 45581 15143/4/ 5/6/8/9 15150/1/2 GEOROC 146992/3/4 GEOROC 100921/2/4 GEOROC 78680/8 111071/ 2/3/4/5/6/7/8/9 112983 – GEOROC 114909/14 – – GEOROC 43567 – GEOROC 43474, 43541 GEOROC 145808/9/27 GEOROC 16708 GEOROC 135763 GEOROC 6277, 6278 – GEOROC 8838,10339,10342,38652 – EarthChem referenced In digested solution Relative standard deviation of results from independent prepared samples RSD values varied from to 0.0006 with the exception of samples 32 and 54 with RSD of 0.0012 and 0.0147, respectively From OIPC [20] Papua New Guinea 26.7 53 24.9 26.8 51 52 25.0 26.6 49 50 24.2 21.1 48 47 25.6 27.6 -5.3 -5.2 – -5.8 -5.1 -5.2 -7.3 -7.3 -8.1 -8.4 -7.6 -7.6 -5.8 -6.4 Mean annual d18Oprecipitation (%)a From EarthChem—Advanced Data Management in Solid Earth Geochemistry The values in this table are means of values of 87Sr/86Sr of parent rock/s found in the same or very approximate geographical coordinates from where green coffee samples were obtained (the range of values is also indicated, whenever there was more than one parent rock sample per location) d c b a Melanesia Mandeling Ermera Indonesia East Timor South-East Asia Sierra del Centro Jamaica San Cristobal Island Ecuador 45 Fazenda Nossa 21.6 44 – 27.9 26.7 27.0 28.2 30.1 22.7 24.0 25.7 23.8 24.3 21.1 19.2 24.7 25.2 d18Obean(%) 46 43 Zona da Mata Yanesha 41 42 Mogiana 40 Fazenda S Benedito Fazenda Muzambo 39 Fazenda da Terra 38 37 San Miguel Santa Rita 36 San Antonio 34 35 Huehuetenango Acatenango Peru Brazil Nicaragua El Salvador Guatemala 32 33 San Marcos de Tarrazu 30 31 Santa Rita Sample Tuxtla (Chiapas) Location Caribbean South America Mexico Central America Costa Rica Country Region Table continued 368 Eur Food Res Technol (2011) 232:361–373 Eur Food Res Technol (2011) 232:361–373 coffees could not be included in the comparison with 87 Sr/86Sr of local parent rock Moreover, we could not access soil maps or geological maps of Jamaica, Costa Rica, Guatemala and Nicaragua However, GEOROC samples of these locations (Table 3) reported in EarthChem [21] allowed us to compare 87Sr/86Sr values of rock samples and green coffee beans Coffees from Jamaica and Guatemala showed 87Sr/86Sr values equivalent to those reported for parent rock, whereas coffees from Costa Rica and Nicaragua had values of 87Sr/86Sr above those reported to corresponding rock samples (Table 3) Brazilian coffee beans revealed a wide variation in Sr isotopes abundance ratios although these samples were distinguishable from coffees obtained from Ecuador, Jamaica, Nicaragua and El Salvador (Table 3) We were unable to find information on Sr isotopic composition of parent rock for each Brazilian geographical location Nevertheless, it was possible to compare 87Sr/86Sr of green coffee to 87Sr/86Sr of the parent rock for samples 39, 40, 42 and 45 (Tables 3) With the exception of sample 45, these Brazilian green coffees had higher 87Sr/86Sr ratios than values reported for parent rock We were not able to obtain information on soil and parent rock for Peruvian coffee, making it difficult to relate 87 Sr/86Sr of the coffee bean to the geology Coffee from Ecuador came from a region rich in clays, metamorphic rocks derived from volcanic rocks and also granites (Mapa General de Suelos del Ecuador, [21]) Values of 87Sr/86Sr for the Ecuadorian coffees are close to the range of values referenced for basalts which is the type of rock reported for GEOROC samples [21] used to compare 87Sr/86Sr values in our work (Table 3) East Timor green coffee could be distinguished from all other origins included in this study solely based on the 87 Sr/86Sr isotopes abundance ratio In spite of this, East Timor coffees showed high scattering of 87Sr/86Sr ranging from 0.7159 to 0.7296 (Table 3) Coffees from Papua New Guinea had different 87Sr/86Sr ratio compared with coffee from Indonesia and lower 87Sr/86Sr ratio compared with East Timor coffee, where parent rock had higher 87Sr/86Sr values compared with the other geographical origins included in this study (Table 3) A correlation was found between the 87Sr/86Sr ratios of African green coffee and the distance from coast of each geographical location (with a Spearman’s correlation coefficient (r) of 0.74; significant at p \ 0.05) Positive correlations between 87Sr/86Sr of the African green coffees and the mean annual precipitation (r = 0.76) and between precipitation and distance from coast (r = 0.87) were obtained, as well This indicates an influence from ocean and precipitation Sr inputs in Africa and consequently in the coffee plants and their seeds developed in these regions We cannot fully exclude that Sr isotopic 369 composition is influenced by dry deposition Sr atmospheric signature can be estimated by direct measurement on bulk precipitation Unfortunately, to our knowledge, there are no available databases with this information Precipitation samples from each location and harvest time were not available to our study The effect of dry deposition in combination with wet deposition has to be further evaluated depending on the geographical region (e.g geology, climate) Sea-salt aerosols show constant marine 87 Sr/86Sr, but continental dust sources may impart spatial and temporal variations on this value on certain locations [27] Whereas the isotopic signature of the atmospheric inputs and the plant materials are relatively easy to determine, the estimation of the weathering end-member remains more problematic Nonetheless, natural weathering can be simulated by extraction of soil samples with acidic solutions The soil exchangeable fraction, which corresponds to the plant-available pool, can be leached from soil by salt solution [14] Future steps of our study will involve strontium isotopes analysis not only of the bulk coffee bean but also of corresponding precipitation water and soil These observations cannot be verified in the case of Latin American, Caribbean coffees and Asian coffees due to the limited number of samples Oxygen isotope abundance ratios of green coffees Oxygen isotopic ratios (d18O values) of green coffee beans varied globally from 18.7 to 34.8% (Table 3) In Africa, we found values ranging from 25.6 to 34.8% Pacific (Hawaii), Central American and Caribbean (Jamaica) origins as well as South American coffee displayed a lower range of values for d18O (19.2 to 30.1%) The same was observed in South Asia and Melanesia (Papua New Guinea) with values for d18O of the coffee bean between 18.7 and 26.7% A solid interpretation of the 18O abundance in the green coffee bean is complicated by the combination of environmental, climatic and physiological processes Moreover, oxygen isotopes are known to fractionate in plant leaves [9] but less is known concerning fractionation in seeds (e.g the coffee bean) However, most of seed organic matter should derive from leaf photosynthesis, probably with minor contribution of seed photosynthesis d18O of rainwater directly translates to the d18O of soil water [28] After soil water has been uptake through plant roots, plant stem water remains virtually with the same d18O value as for soil water and for plant water lost by transpiration However, oxygen isotopes fractionation occurs in leafs during photosynthesis [28] Newly synthesised plant organic compounds d18O will be dependent not only on fractionation occurring during this process but also on the specific biosynthetic pathways involved in their synthesis (their enzymes and regulation) Also, the leaf 123 370 water oxygen isotope signal is dampened in organic material formed from exported sucrose in other plant parts As a consequence, organic molecules may reflect or not the water in which they are formed In spite of this, several authors report that oxygen isotope composition of plant organic material is known to reflect that of source water and leaf evaporative conditions at the time the material was formed [29] On this basis, we have tried to correlate green coffee bean d18O with environmental factors known for the correspondent origins included in this study In fact, Serra et al [18] already indicated the importance of relating isotope analysis with known data on environmental variables in order to build the best model for origin discrimination We calculated Spearman’s rank correlation coefficient (r) between isotopic composition of oxygen and climate and geographical information of each known geographical location A weak correlation was found between values of d18O of green coffees and the distance from coast (r = 0.27; significant at p \ 0.05) Several authors reported a continental effect on isotopic fractionation of oxygen in precipitation in which a trend of lighter isotopic composition is seen as a function of the distance from coast Most water vapour in the atmosphere is derived from evaporation of low-latitude oceans Precipitation derived from this vapour is always enriched in 18O relative to the vapour The fractionation between rain and vapour is a function of condensation temperature Therefore, as clouds move across the continent, progressive rainout leads to increasingly isotopically lighter rain [9, 30] However, our results not show this continentality effect on d18O of green coffee beans This suggests that a variety of processes are influencing the d18O values of the bulk bean, e.g variations in d18O of water taken up by plants and leaf water enrichment in 18O during plant transpiration, depending on the atmospheric conditions (relative humidity and 18O of water vapour in the atmosphere) As seen for the distance to the sea, weak correlations between d18O of the coffee bean and mean annual precipitation and latitude were observed (r = -0.34 for both correlations; significant at p \ 0.05) Isotopically lighter (depleted; with less 18O) rain is observed with increasing latitude [8] This effect is known as latitude effect on isotopic fractionation of 18O in precipitation and is well described in literature [30] The correlations found between d18O of the coffee bean and values of latitude and mean annual precipitation could be an indication that variations in the isotopic composition of oxygen of the green coffee bean were related to these factors However, this was not found on the analysis at global scale Additionally, we have not found any correlation between oxygen isotopic composition of the coffee bean and values of altitude The statistical analysis was repeated for smaller regions Oxygen isotopic composition of green coffees from islands correlated with values of 123 Eur Food Res Technol (2011) 232:361–373 altitude (r = -0.91; significant at p \ 0.05) Lower values of altitude correspond to enriched values of d18O of green coffees In this system, the altitude effect on oxygen isotopic composition seems clear (this pattern is often not observed in interior mountains or on the leeward side of mountains) Due to Rayleigh distillation [30], rainfall values become more depleted in 18O as a storm moves across the landscape Rain at higher altitudes (where lower temperatures are also verified) becomes more depleted in 18O This leaves a plausible explanation of the variation on the d18O of the green coffees originating from islands The same approach was applied to the case of South American coffee beans (Peru, Ecuador and Brazil), and the oxygen composition of the mean annual precipitation as well as the corresponding values for latitude and altitude seemed to be relevant to the d18O values of green coffee (rd18O bean versus d18O mean annual precipitation = 0.84, rd18O bean versus mean annual precipitation = 0.76, rd18O bean versus latitude = 0.68 and rd18O bean versus altitude = -0.89) An altitude effect on the isotopic composition of oxygen was observed as well as a strong correlation with precipitation As verified in plant leaf organic material [9], we assume that the oxygen isotope composition of a seed (i.e the green coffee bean) reflects that of source water and coffee plant leaf evaporative conditions at the time the seed was formed Lack of studies and experimental data still limit the understanding of the fractionation of oxygen isotopes of the green coffee bean during its formation Seeds are very complex matrices, rich in a large number of compounds originating from primary and secondary metabolism leading to possible divergence to models that have been described for other plants and/or plant tissues Multivariate analysis The variable reduction method of principal component analysis (PCA) was used for exploratory data visualisation to determine to what extent we could discern differentiation of the samples according to geographic origin Data on Sr amount and isotope ratio of coffees and correspondent parent rock (whenever available) (Table 3) were used to determine principal components and green coffee samples scores on principal components (figure not shown) The first two principal components (PCs) selected to plot samples scores had eigenvalues of 1.9792 and 0.9877 and together explained [74% of the total variance of the data PC1 explains 49.48% of the total variance and correlates negatively with 87Sr/86Sr of green coffees and 87Sr/86Sr of parent rock (the two variables have loading values approximate or higher than |0.65|) PC2 explains 24.69% of the remaining variance with Sr (ppm) of parent rock having the major loading (0.7900) In PC3, explaining 21.39% of Eur Food Res Technol (2011) 232:361–373 variance, the loading of Sr (ppm) in green coffees was higher than |0.65| (0.8793) Figure shows that coffees from UR Tanzania, East Timor, Hawaii, Rwanda and Kenya appear in groups and discriminate from other origins Spearman’s correlation r (significant at p \ 0.05) between 87Sr/86Sr of green coffees and 87Sr/86Sr of rock was 0.698907 and between Sr (lg g-1) in green coffees and Sr (lg g-1) in rock was 0.522956, showing that these parameters are related to each other There was evidence that most 87Sr/86Sr ratios of the green coffee bean are related to Sr isotope abundance ratio of the parent rock and/ or soil Reserves of bioavailable cations in the soil are mainly in exchangeable form, adsorbed on the mineral and organic matter surface (exchange complex) The advantage of the use of the Sr isotope technique is that soil 87Sr/86Sr remains a robust signature, although concentrations in major elements released by weathering can be modified by the formation of secondary minerals or exchange processes Fig Principal component analysis of oxygen and strontium isotopic composition of African green coffees (a) and with values of annual mean precipitation, distance from coast and altitude (b) (Legend: ? Ethiopia; h Kenya; È Malawi; ' Rwanda; H UR Tanzania; Zambia; Zimbabwe) 371 on the adsorbing complex Lacking agreement of isotope ratios of total rock and soil with bioavailable Sr can be interpreted that Sr fractions in soil have different solubility [14] Thus, the isotopic composition of all sources of Sr such as soil, soil extracts, wet and dry precipitates, surface and groundwater as well as water used for watering has to be subject to broader investigations for further interpretation The final combination of the isotopic composition of both oxygen and strontium did not allow total discrimination between all 20 countries in one single analysis However, it is possible to achieve a separation between selected origins and groups of provenances East Timor discriminated significantly from all other origins The samples from Peru separated from the group of Costa Rica and some coffees from Brazil as well as from Papua New Guinea and from the group enclosing coffee from Malawi, Rwanda, Zambia and finally from a major group of origins including samples from UR Tanzania, Kenya, Ethiopia, Hawaii, Jamaica, Nicaragua, El Salvador, Guatemala, Ecuador, Mexico and Indonesia (figure not shown) We have furthermore applied the same multivariate analysis to smaller regions (Africa, Islands and South America) A strong competitive market is observed within these regions PCA of the isotopic composition of oxygen and strontium of the African green coffees is shown in Fig 1a The first two PCs selected to plot samples scores had eigenvalues of 1.038 and 0.962 Coffees from Ethiopia, UR Tanzania and Kenya discriminate significantly from coffees from Rwanda, Malawi and Zambia 87Sr/86Sr of the samples originating from Rwanda, Malawi and Zambia show 87 Sr/86Sr values higher than 0.710, while Ethiopian, Tanzanian and Kenyan coffees have significantly lower values (Table 3) A positive correlation between d18O values of African green bean samples and higher altitudes (r = 0.56 (significant at p \ 0.05)) as well as between mean annual precipitation and distance from coast with strontium isotopic composition of the African green coffees (r = 0.76 and 0.74, respectively; significant at p \ 0.05) was observed When PCA was repeated including these factors as variables, the result was the distinction between the African origins (Fig 1b) The two-first PCs had eigenvalues of 2.378 and 1.666, respectively, and explained 80% of the variability observed 87Sr/86Sr and the distance to the sea were positive in the two PCs, whereas mean annual precipitation, altitude and d18O of the green coffee beans were negative on the first PC but positive on the second The same approach was applied to coffees originating from islands (Hawaii, Jamaica, East Timor, Indonesia and Papua New Guinea) (Fig 2) East Timor coffees discriminated from coffees from Papua New Guinea and from a third group enclosing Indonesian, Hawaiian and Jamaican coffees In this analysis, the eigenvalues of the two PCs were 1.021 and 0.979, respectively Figure shows the 123 372 Fig Principal component analysis of oxygen and strontium isotopic composition of green coffees from islands (Legend: h Hawaii; È Indonesia; ? Jamaica; H Papua New Guinea; ' East Timor) Fig Principal component analysis of oxygen and strontium isotopic composition of green coffees from South America (Legend: h Brazil; ? Ecuador; ' Peru) result of PCA of isotopic composition of oxygen and strontium of green coffees from South America The two PCs had eigenvalues of 1.028 and 0.972, respectively Green coffee samples from Ecuador discriminate from coffees from Peru and from Brazilian coffees Conclusion We proved evident that a combination of the isotopic systems of oxygen and strontium is a good approach to discriminate coffees according to their geographical origin in smaller competitive regions (such as South America or Africa) The results indicate that the isotopic composition 123 Eur Food Res Technol (2011) 232:361–373 of oxygen and strontium of the green coffee bean is related to the isotopic composition of oxygen of wet precipitation and to the isotopic composition of bioavailable strontium in soil We still have to consider additional influences such as CO2 sources, variations in temperature, the time of the year when rain and plant irrigation occur as well as coffee plant metabolism (primary and secondary) Further research on Sr isotopic composition variation during coffee bean phenology leaves room for investigations (similar to considerations for the interpretation of the variation of the oxygen isotopic composition) even though Sr fractionation caused by metabolic processes is expected to be far less pronounced (if even measurable) The use of isotope ratios for origin determination can be further improved by establishing a sampling network allowing the use of additional climatic and geological data as variables Additional information on the stability of environmental parameters that influence isotopic signature of foods is fundamental This will bring an innovative approach in which it may be possible to relate environment with food isotopes and so explain ‘why’ it is possible to get a discrimination based on isotope analysis of the food A conclusive Coffee Geographical Origin Discrimination Model requires an increased number of samples besides fundamental studies of influencing parameters and on how these are stable with time The work is a key for using both strontium isotopes abundance ratios and d18O for origin discrimination of coffee as well as it proves that the combination with additional parameter enhances the validity of a discrimination model Acknowledgment The authors wish to thank Novadelta, Come´rcio e Indu´stria de Cafe´s, S.A., Campo Maior, Portugal for supplying green bean coffee samples and geographic location information Rodrigo Maia, Marion Brunner, Johanna Irrgeher and Stefanie Schweigkofler are acknowledged for technical support of the sample preparation and measurement Carla Rodrigues wishes to thank Fundac¸a˜o para a Cieˆncia e a Tecnologia for a grant (SFRH/BD/ 28354/2006) Financial support by the Austrian Science Foundation (FWF START grant 267 N11) is highly acknowledged This work was financed by the project PTDC/AGR-AAM/104357/2008 (IsoGeoCoffee) References Costa Freitas AM, Parreira C, Vilas-Boas L (2001) The use of an electronic aroma-sensing device to assess coffee differentiationcomparison with spme gas chromatography-mass spectrometry aroma patterns J Food Compost Anal 14:513–522 Bertrand B, Etienne H, Lashermes P, Guyot B, Davrieux F (2005) Can near-infrared reflectance of green coffee be used to detect introgression in coffea arabica cultivars? J Sci Food Agric 85:955–962 Martı´n MJ, Pablos F, Gonza´lez AG, Valdenebro MS, Leo´nCamacho M (2001) Fatty acid profiles as discriminant parameters for coffee varieties differentiation Talanta 54:291–297 Eur Food Res Technol (2011) 232:361–373 Gonza´lez AG, Pablos F, Matı´n MJ, Le´on-Camacho M, Valdenebro MS (2001) Hplc analysis of tocopherols and triglycerides in coffee and then use as authentication parameters Food Chem 73:93–101 Weckerle B, Richling E, Heirinch S, Schreier P (2002) Origin assessment of green coffee (coffea arabica) by multi-element stable isotope analysis of caffeine Anal Bioanal Chem 374:886–890 Suzuki Y, Chikaraishi Y, Ogawa NO, Ohkoushi N, Korenaga T (2008) Geographical origin of polished rice based on multiple element and stable isotope analysis Food Chem 109:470–475 Camin F, Larcher R, Perini M, Bontempo L, Bertoli D, Gagliano G, Nicolini G, Versini G (2010) Characterisation of authentic It alian extra-virgin olive oils by stable isotope ratios of c, o and h and mineral composition Food Chem 118:901–909 Bowen GJ, Wilkinson B (2002) Spatial distribution of d18O in meteoric precipitation Geology 30:315–318 Barbour MM (2007) Stable oxygen isotope composition of plant tissue: A review Funct Plant Biol 34:83–94 ´ ggarwal J, Habich-Mauche J, Juarez C (2008) Application of 10 A heavy stable isotopes in forensic isotope geochemistry: a review Appl Geochem 23:2658–2666 11 Critteren RG, Andrew AS, LeFournour M, Young MD, Middleton H, Stockmann R (2007) Determining the geographic origin of milk in australasia using multi-element stable isotope ratio analysis Int Dairy J 17:421–428 12 Rummel S, Hoelzl S, Horn P, Rossman A, Schlicht C (2008) The combination of stable isotope abundance ratios of h, c, n and s with 87sr/86sr for the origin assignment of orange juices Food Chem 118:890–900 13 Almeida CMR, Vasconcelos MTSD (2004) Does the winemaking process influence the wine 87sr/86sr? A case study Food Chem 85:7–12 14 Swoboda S, Brunner M, Boulyga SF, Galler P, Horacek M, Prohaska T (2008) Identification of marchfeld asparagus using sr isotope ratio measurements by mc-icp-ms Anal Bioanal Chem 390:487–494 15 Krivan V, Barth P, Morales AF (1993) Multielement analysis of green coffee and its possible use for the discrimination of origin Mikrochimica Acta 110:217–236 16 Anderson KA, Smith BW (2002) Chemical profile to differentiate geographic growing origins of coffee 50:2068–2075 17 Wieser EM, Iyer SS, Krouse HR, Cantagallo MI (2001) Variations in the boron isotope composition of coffea arabica beans Appl Geochem 16:317–322 373 18 Serra F, Guillou CG, Reniero F, Ballarin L, Cantagallo MI, Wieser M, Iyer SS, He´berger K, Vanhaecke F (2005) Determination of the geographical origin of green coffee by principal component analysis of carbon, nitrogen and boron stable isotope ratios Rapid Commun Mass Spectrom 19:2111–2115 19 Global gazetteer (2006) Falling Rain Genomics, Inc http://www fallingrain.com/world/ Accessed December 2009 20 Bowen GJ (2010) The Online Isotopes in Precipitation Calculator, version 2.2 http://www.waterisotopes.org 21 European digital archive of soil maps (2005) Maps are holdings of ISRIC http://eusoils.jrc.it/esdb_archive/EuDASM/EUDASM htm 22 Advanced data management in solid earth geochemistry (2003) http://earthchem.org/earthchemWeb/index.jsp Accessed December 2009 23 Stein M, Starinsky A, Katz A, Goldstein SL, Machlus M, Schramm A (1997) Strontium isotopic, chemical, and sedimentological evidence for the evolution of lake lisan and the dead sea Geochim Cosmochim Acta 61:3975–3992 24 Schultheis G (2003) Analysis of isotope ratios in anthropological and archaeological samples by high resolution inductively coupled plasma mass spectrometry (hr-icp-ms) Universita¨t fu¨r Bodenkultur am Institut fu¨r Chemie, Vienna 25 Capo RC, Stewart BW, Chadwock OA (1998) Strontium isotopes as tracers of ecosystem processes: theory and methods Geoderma 82:197–225 26 Prohaska T, Wenzel W, Stingeder G (2005) Icp-ms-based tracing of metal sources and mobility in a soil depth profile via the isotopic variation of sr and pb Int J Mass Spectrom 242:243–250 27 Drouet T, Herbauts J, Demaiffe D (2007) Change of the origin of calcium in forrest ecosystems in the twentieth century highlighted by natural sr isotopes In: Dawson TE, Siegwolf RTW (eds) Stable isotopes as indicators of ecological change Elsevier, London, pp 333–343 28 Yakir D, Sternberg LDSL (2000) The use of stable isotopes to study ecosystem gas exchange Oecologia 123:297–311 29 Barbour MM, Schurr U, Henry BK, Wong SC, Farquhar D (2000) Variation in the oxygen isotope ratio of phloem sap sucrose from castor bean Evidence in support of the pe´clet effect Plant Physiol 123:671–679 30 Augusti A, Schleucher J (2007) The ins and outs of stable isotopes in plants New Phytol 174:473–475 123 Eur Food Res Technol (2011) 232:375–379 DOI 10.1007/s00217-010-1383-7 SHORT COMMUNICATION Pyroglutamic acid development during grape must cooking Giuseppe Montevecchi • Francesca Masino Andrea Antonelli • Received: 23 June 2010 / Revised: October 2010 / Accepted: 15 October 2010 / Published online: 10 November 2010 Ó Springer-Verlag 2010 Abstract A study of the pyroglutamic acid development during must cooking as glutamine and glutamic acid degradation product was carried out Pyroglutamic acid was detected in two white musts (Trebbiano toscano and Spergola) and in a red one (Lambrusco) previously cooked by means of a laboratory-scale equipment emulating the real process Results showed that pyroglutamic acid sharply increased and glutamine quickly disappeared during the first stages of the cooking process, while glutamic acid was almost linearly reduced throughout all the process A study carried out by heating the model solutions of the single amino acids showed that both of them underwent degradation following the same trend observed in the musts Keywords Grape must Á Heat concentration Á Pyroglutamic acid Á Glutamine thermal degradation Á Glutamic acid thermal degradation Introduction Cooked must is a widespread product in many wine-producing countries For centuries, cooked must has been represented the only alternative to honey as a sweetener in the regions where the viticulture was practiced The cooking process kills bacteria and other microorganisms, and the high concentration of sugars prevents their growth allowing a long storage G Montevecchi (&) Á F Masino Á A Antonelli Dipartimento di Scienze Agrarie e degli Alimenti, Universita` degli Studi di Modena e Reggio Emilia, Via Amendola 2, 42100 Reggio Emilia, Italy e-mail: giuseppe.montevecchi@unimore.it The traditional use is still perpetuated in many regions for several uses [1] as garnish in pastry making, jam, and pudding, and for production of fermented beverages and fermented seasonings [2–5] Fermented seasonings are typical products of the Emilia-Romagna region, in the North of Italy, known as Balsamic Vinegars Among them, traditional balsamic vinegars (TBV) [6] are produced by slow ageing of cooked must from local grapes Grape must cooking is performed in open boilers or pans at ambient pressure and over a direct flame to give a cooked must up to 2–4 times more concentrated than the raw grape juice [7, 8] This process is quite long (up to 30 h), and the temperature is set at about 85–90 °C In a previous work [3], different grape musts were cooked by means of a laboratory equipment able to emulate the real process The concentrations of 17 amino acids, ammonium ion, c-aminobutyric acid, and ethanolamine were followed during the whole process Except for ammonium ion that was almost at the same concentration during the experiment, the other amino acids showed a common pattern of degradation, but at different rates [3] Glutamine was the most sensitive amino acid to temperature In fact in all the musts, it completely disappeared only after h However, no amino acid degradation product was identified Nevertheless, under the experimental conditions, glutamine was likely converted into pyroglutamic acid (5-oxoproline, 5-pyrrolidine-2-carboxylic acid) [9, 10] When heated, in fact, glutamine releases ammonia on a molar basis mainly due to the deamination of the amide group (Fig 1, route 1) [11], contrary to more stable lower homologue, the asparagine [12] The same conditions could promote glutamic acid conversion into pyroglutamic acid [13], although more slowly than glutamine (Fig 1, route 2) 123 376 Eur Food Res Technol (2011) 232:375–379 Model solutions Fig Pyroglutamic acid production from glutamine (route 1), from glutamic acid (route 2), and conversion of glutamine into glutamic acid (route 3) This work explores the fate of two amino acids—glutamine and glutamic acid—that have shown the most remarkable rate of degradation Four model solutions were set Each solution was prepared by dissolving two different amounts of glutamine or glutamic acid into 20 mL of water The first two solutions were adjusted to pH 3.0 with diluted H2SO4 A set of solutions (glutamine 0.104 g, 0.71 mmol; glutamic acid 0.110 g, 0.75 mmol) was heated in closed test tubes into a water bath at 90 °C for 30 h, and the other two solutions (glutamine 0.109 g, 0.74 mmol; glutamic acid 0.099 g, 0.68 mmol) were heated under the same conditions, but for h only Samples (1.5 mL each) were collected at h, 18 h, and 30 h for the first two solutions; and 0, 1, 2, and h, for the other two solutions A pyroglutamic acid solution (pH = 3) was heated into a water bath (90 °C for 10 h) and a sand bath (270 °C for min) Determinations of ammonium ion, glutamic acid, glutamine, and pyroglutamic acid in real samples and model solutions by HPLC Materials and methods Grape musts and their concentration Grape musts of Lambrusco, Trebbiano toscano (from here on Trebbiano), and Spergola were collected from local wineries They were stored at -20 °C and defrosted immediately before use Musts were divided into aliquots of L each and concentrated to half of the initial volume in partially capped 1-L conical flasks [3] Every h, 1.5 mL of must were withdrawn and immediately stored at -20 °C for analyses Analyses were carried out in duplicate Reagents Pure reference compounds (ammonium chloride, L-glutamic acid, L-glutamine, and DL-pyroglutamic acid), hydrochloric acid, sulphuric acid, and diethylethoxymethylenemalonate (DEEMM) were purchased from Fluka Sigma–AldrichÒ (Milan, Italy), while high-purity solvents (acetonitrile and methanol) were supplied by different companies Deionised water was obtained by a Milli-Q purification system (Millipore, Milan, Italy) Standard stock solutions Standard 2,000-ppm stock solutions for ammonium ion (as ammonium chloride), glutamic acid, glutamine, and pyroglutamic acid in HCl 0.01 M were prepared 123 The determination of amino acids and ammonium ion was performed by a pre-column derivatisation method, which utilised DEEMM, a derivatising agent able to convert amino compounds into aminoenones The method [14] was used on diluted samples, as already reported [3] As DEEMM is unable to react with amidic nitrogen, pyroglutamic acid was separated by anion-exclusion HPLC with a method used for organic acid determination [15], with some minor modifications [3] The column—Bio-Rad Aminex HPX-87H hydrogen-form cation exchange resinbased column (300 7.8-mm i.d.), protected by a precolumn (30 4.6 mm)—operated at 50 °C, with a flow rate of 0.5 mL/min using a solvent system (H2SO4 0.045 N adjusted to pH 1.35; CH3CN 10%) A diode-array detector was used to quantify the pyroglutamic acid at 200 nm [16] Chromatograms were acquired and processed with TotalChrom Workstation version 6.2.1 chromatography system software (PerkinElmer, Inc.) Peaks were identified by comparing retention times of pure standards Quantification was performed through an external standard calibration method Results and discussion Calibration gave good results Pyroglutamic acid equation for calibration was calculated by peak areas of standard solutions Linearity in the concentration range was satisfactory as shown by determination coefficient (R2) of 0.996, and the limit of detection was 0.55 ppm Eur Food Res Technol (2011) 232:375–379 A study of retention time of pyroglutamic acid was carried out by modifying the acetonitrile percentage and the pH of the mobile phase Acetonitrile reduced the retention time of pyroglutamic acid by min, passing from 26 (no CH3CN) to 22 with 10% CH3CN Working close to the highest operative limit of the column (pH = 3), a further 2-min reduction of retention time was observed In previous works [3, 4] wherein real must samples were heat-concentrated in open vessels, the presence of pyroglutamic acid was ignored, as studies on amino acid behaviour were at a pioneering stage However, the quick decrease in glutamine [3] suggested investigating more in detail about its evolution As a matter of fact, the presence of pyroglutamic acid was easily detected reanalysing the samples used for the research on amino acid evolution The concentration of pyroglutamic acid sharply increased during cooking time, while glutamine quickly disappeared After only h, glutamine concentration was below its LOD, while glutamic acid, another pyroglutamic acid precursor decreased more regularly (Fig 2a–c) Ammonia is a common degradation substance of almost all amino acids [12] Its content was almost at the same concentration during the whole process (Fig 3) [3] The cooking process was carried out in an open vessel and ammonia could freely volatilise On the contrary, the acid media contrasted this tendency by salifying part of the ammonia evolved As a consequence, at the end of the process, the level of ammonia was comparable to the concentration of this substance at the beginning of the process However, within the 30-hour cooking, some variations were observed In order to eliminate all these uncertainties and to gain a deeper comprehension of this phenomenon, similar experiments were carried out with model solutions Temperature and pH were perfectly comparable to must cooking conditions However, musts in real concerns are cooked in open vessels to allow their concentration, while these experiments were carried out in closed tubes to prevent volatile losses and solute concentration Moreover, time is comparable in the first set of samples (30 h) only Finally in model solutions, the amino acid concentrations were deliberately higher than those ones detected in the musts to have a more marked vision of the phenomenon Considered on a molar basis, results were more congruent Glutamine (Table 1) was almost quantitatively transformed into pyroglutamic acid and ammonia (Fig 1, route 1) The same behaviour was confirmed in 3-hour experiment The glutamine loss was quick and regular Finally, a slight tendency to glutamine hydration to yield glutamic acid (Fig 1, route 3) was also evident in both cases [9] Compared to glutamine, the same experiment applied to glutamic acid (Fig 1, route 2) showed a less pronounced 377 Fig Glutamic acid, glutamine, pyroglutamic acid mmol/Brix versus time of cooking in the three different musts: Lambrusco (a), Trebbiano (b), Spergola (c) Data were divided by Brix to eliminate the effect of concentration tendency to degradation with a negligible amount of ammonia production (Table 1) The ratio of glutamine and glutamic acid degradation in real samples vs model solutions indicated that acid media was able to reduce the amount of both amino acids although it seemed likely an additional contribution to the degradation due to other pathways, more relevant for glutamic acid than for glutamine 123 378 Eur Food Res Technol (2011) 232:375–379 of N-acetyl-pyroglutamic acid (1-acetyl-5-oxopyrrolidine2-carboxylic acid) could be likely At the moment, GC–MS analysis performed on a single TBV sample did not allow any confirmation of this hypothesis Conclusions Fig Ammonia ppm/Brix versus time of cooking in the three different musts: Lambrusco, Trebbiano, and Spergola Data were divided by Brix to eliminate the effect of concentration Pyroglutamic acid has showed the characteristic of a final product In fact, a further thermal stress of this substance had negligible consequences Pyroglutamic acid heated for 10 h at 90 °C gave no loss, while a more severe treatment (270 °C for min) caused a 16% loss However, a reaction with other constituents may not be excluded Webb, Kepner, and Maggiora [17] reported the presence of pyroglutamic ethyl ester (ethyl 5-oxopyrrolidine-2-carboxylate) in Sherry wines, while in vinegars the presence The presence of pyroglutamic acid in must and its increase during cooking process was clearly explained by the degradation of glutamine and, in a lesser extent, of glutamic acid, as well Ammonia was another important product of degradation Its importance cannot be ignored because it could play a key role in many characteristics of the cooked must as well as of the other products, which are elaborated by its ageing (i.e TBV) Its high reactivity could yield important odorants, and its role for microorganism metabolism is very important, as well Because of the susceptibility to heat of glutamine, the study of the ratio glutamine/pyroglutamic acid could be a useful tool to detect wine pasteurisation The influence of cooking process on glutamine enhancement is clearly demonstrated It could be very interesting to evaluate the effect of TBV ageing on the Table Ammonium ion, glutamic acid, glutamine, and pyroglutamic acid content (mmol) in the model solutions Hours 18 30 Glutamine 30 h Ammonium ion 0.011 (±0.001) 0.539 (±0.005) 0.596 (±0.031) Glutamic acid 0.000 (±0.000) 0.002 (±0.000) 0.003 (±0.000) Glutamine 0.709 (±0.007) 0.006 (±0.002) 0.001 (±0.000) Pyroglutamic acid Nd 0.637 (±0.002) 0.742 (±0.008) Glutamic acid 30 h Ammonium ion 0.020 (±0.000) 0.026 (±0.004) 0.023 (±0.002) Glutamic acid 0.751 (±0.013) 0.631 (±0.017) 0.479 (±0.003) Pyroglutamic acid Nd 0.335 (±0.006) 0.524 (±0.011) Hours Glutamine h Ammonium ion 0.009 (±0.001) 0.211 (±0.003) 0.352 (±0.007) 0.517 (±0.011) Glutamic acid 0.000 (±0.000) 0.006 (±0.001) 0.008 (±0.001) 0.059 (±0.006) Glutamine 0.744 (±0.017) 0.430 (±0.008) 0.317 (±0.024) 0.199 (±0.006) Pyroglutamic acid Nd 0.329 (±0.008) 0.464 (±0.006) 0.679 (±0.007) Ammonium ion 0.033 (±0.002) 0.032 (±0.002) 0.031 (±0.008) 0.020 (±0.000) Glutamic acid 0.676 (±0.026) 0.643 (±0.009) 0.644 (±0.026) 0.603 (±0.023) Pyroglutamic acid Nd 0.014 (±0.001) 0.039 (±0.004) 0.105 (±0.028) Glutamic acid h Data (±STD) are means of two replications Nd not detected 123 Eur Food Res Technol (2011) 232:375–379 glutamine/pyroglutamic acid ratio It is very likely that the accumulation of pyroglutamic acid has a positive trend as a consequence of time and of summer heat References Repubblica Italiana (2007) Decreto Direttoriale 19/06/2007 Settima revisione dell’elenco nazionale dei prodotti agroalimentari tradizionali Gazzetta Ufficiale della Repubblica Italiana, 147(27 June) (S.O n 146) Bozkurt H, Go¨gu¨s F, Eren S (1999) Food Chem 64:89–93 Montevecchi G, Masino F, Chinnici F, Antonelli A (2010) Food Chem 121:69–77 Piva A, Di Mattia C, Neri L, Dimitri G, Chiarini M, Sacchetti G (2008) Food Chem 106:1057–1065 Rivero-Pe´rez M, Pe´rez-Magarin˜o S, Gonza´lez-San Jose´ M (2002) Anal Chim Acta 458:169–175 379 Repubblica Italiana (2000) Gazzetta Ufficiale della Repubblica Italiana, 124, 30 May 2000, and following modifications Antonelli A, Chinnici F, Masino F (2004) Food Chem 88:63–68 Cocchi M, Lambertini P, Manzini D, Marchetti A, Ulrici A (2002) J Agric Food Chem 50:5255–5261 Archibald RM (1945) Chem Rev 37:161–208 10 Vickery HB, Pucher GW, Clark HE, Chibnall AC, Westall RG (1935) Biochem J 29:2710–2720 11 Airaudo CB, Gayte-Sorbier A, Armand P (1987) J Food Sci 52:1750–1752 12 Sohn M, Ho CT (1995) J Agric Food Chem 43:3001–3003 13 Gayte-Sorbier A, Airaudo CB, Armand P (1985) J Food Sci 50:350–352 14 Go´mez-Alonso S, Hermosı´n-Gutie´rrez I, Garcı´a-Romero E (2007) J Agric Food Chem 55:608–613 15 Castellari M, Versari A, Spinabelli U, Galassi S, Amati A (2000) J Liq Chromatogr Relat Technol 23:2047–2056 16 Shih FF (1985) J Chromatogr A 322:248–256 17 Webb AD, Kepner RE, Maggiora L (1967) Am J Enol Vitic 18:190–199 123 [...]... obtain a final concentration of 0.5 mM aldehyde Control was aldehyde-free and was prepared in the same manner, except an equivalent volume of ethanol was used instead The treated samples were collected at 0, 1, 2, 4, 6, and 8 days of incubation at 4 °C for analysis To determine the crosslinking effect of aldehydes, the mixture containing hexanal and hexenal at a final concentration of 12.5 mM was also... peptidases and are promising candidates for the detoxification of gluten-containing foods [8] and possibly for oral therapy of CD patients They are composed of endopeptidases and exopeptidases and have unique specificities optimized by nature for the fragmentation of gluten proteins and peptides [7] and are derived from a naturally safe and cheap raw material Their production is part of well-established... power but a lower antioxidant capacity measured by the DPPH technique [45] This biplot clearly shows that the samples were separated in terms of extracts and not in terms of thermal treatments Extracts A and C are darker, and they have a higher amount of TC Extracts D and E have a higher amount of individual phenolic content Finally, extract B clearly has the highest amount of GA and the lowest AA The... For example, the epimerisation and degradation of flavan-3-ols have been evaluated and modelled for a green tea extract at high temperatures ranging from 100 to 165 °C with various durations of up to 120 min These tests showed that the epimerisation and degradation of the tea’s catechins followed first-order reactions, and the rate constants of the reaction kinetics followed the Arrhenius equation [15]... different batches of samples Data were subjected to analysis of variance (ANOVA) Comparison of means was carried out by Duncan’s multiple range test For pair comparison, t-test was used [21] Statistical analysis was performed using the Statistical Package for Social Science (SPSS 11.0 for Windows, SPSS Inc., Chicago, IL, USA) Results and discussion Effect of Fenton’s reactants on the changes of Eastern... reagent, a mixture of bathophenanthroline disulfonic acid, double-deionized water, and saturated sodium acetate solution at a ratio of 1:20:20 (w/v/v), prepared freshly After vortexing and standing for 10 min, the absorbance was measured at 540 nm The non-heme iron content was calculated from iron standard curve The iron standard solutions (Fe(NO3) in HNO3), with the concentrations ranging from 0 to... grown because polyphenols can be utilised as antioxidants in the food industry, and they benefit human health in various ways The beneficial effects of natural antioxidants on human health come mainly from the capability of polyphenols to scavenge free radicals and therefore protect cells from the damage caused by free radicals [1] Polyphenols can also act as anti-inflammatory agents [2, 3], they can inhibit... tendency of GA to increase may b a 70,0 b ba,b a, b a, b a, b a b a, b aa,b a a 6,00 d mgCyE/g 50,0 c,d b b b,c d c,d ab ba mMolesTrolox/g 60,0 a a aa a a a 40,0 30,0 20,0 a a,b a, b a a 5,00 c c b,c a, b a, b a b b d b b b b a a a a a a a a a a a 4,00 3,00 2,00 1,00 10,0 0,0 A B C D 0,00 E A Extracts c 7,00 B C D E Extracts 3,00 e f d Absobance at 420 nm 2,50 a b f e c d 2,00 f d e 1,50 a b f c a b c c d... balansae) by matrix-assisted laser desorption/ionization time -of- flight mass spectrometry and thioacidolysis/liquid chromatography/electrospray ionization mass spectrometry Anal Chim Acta 513(1):247–256 31 De Freitas VAP, Glories Y (1999) Concentration and compositional changes of procyanidins in grape seeds and skin of white vitis vinifera varieties J Sci Food Agric 79(12):1601–1606 32 Zhu QY, Zhang... 113(4):1226–1233 46 Hagerman AE, Riedl KM, Jones GA, Sovik KN, Ritchard NT, Hartzfeld PW, Riechel TL (1998) High molecular weight plant polyphenolics (tannins) as biological antioxidants J Agric Food Chem 46(5):1887–1892 47 Jayabalan R, Marimuthu S, Thangaraj P, Sathishkumar M, Binupriya AR, Swaminathan K, Sei EY (2008) Preservation of kombucha tea—effect of temperature on tea components and free radical scavenging