Hydrolysis of terpenyl glycosides in grape juice and other fruit juices: a review

Transcription

1 Appl Microbiol Biotechnol (2005) 67: DOI /s MINI-REVIEW Sergi Maicas. José Juan Mateo Hydrolysis of terpenyl glycosides in grape juice and other fruit juices: a review Received: 28 July 2004 / Revised: 30 September 2004 / Accepted: 19 October 2004 / Published online: 6 January 2005 # Springer-Verlag 2005 Abstract The importance of monoterpenes on varietal flavour of must and other fruit juices has been reviewed. These compounds were mainly found linked to sugar moieties in grape juice and wines, showing no olfactory characteristics. In this way, analytical techniques developed to study these compounds, in both free or glycosidically forms, are discussed. Mechanisms to liberate terpenes were studied, making a comparative study between acidic and enzymic hydrolysis of terpene glycosides; as enzymic hydrolysis seems to be the most natural way to liberate terpenes, the ability to use glycosidases from grapes, yeasts, bacterial or exogenous, i.e. fungal commercial preparations, were reviewed. Re-arrangements of terpenes after acidic hydrolysis of glycoconjugated are discussed as well as potential adverse effects of enzyme preparations. Introduction Research over the last decades has revealed that a great number of plant-tissue flavour compounds are glycosilated and accumulate as non-volatile and flavourless glycoconjugates (Stahl-Bishop et al. 1993; Winterhalter and Skouroumounis 1997; Mateo and Jimenez 2000). Although results in literature had long suggested the occurrence of glycosidically bound flavour compounds in plants, the first clear evidence was found in 1969 by Francis and Allock in S. Maicas Department of Food Science and Technology, Universidad Cardenal Herrera-CEU, Seminari s/n, Montcada, Spain S. Maicas. J. Mateo J. Department of Microbiology and Ecology, Universitat de València, Dr. Moliner 50, 46100, Burjassot, Spain Tel.: Fax: Jose.J.Mateo@uv.es rose (Francis and Allcock 1969). The work of Cordonnier and Bayonove (1974), suggesting the occurrence in grapes of monoterpenes (important flavour compounds) as glycoconjugates on the basis of enzymatic works, was later confirmed by identification of glycosides (Williams et al. 1982). These findings opened a new field of intensive research on the chemistry of glycoconjugated flavour compounds to exploit this important flavour source present in both plants and fruit tissues. Glycoconjugates of flavour compounds are present in several fruits such as grapes (Williams et al. 1982; Gunata et al. 1985), apricot (Krammer et al. 1991; Salles et al. 1991), peach (Krammer et al. 1991), yellow plum (Krammer et al. 1991), quince (Lutz and Winterhalter 1992), sour cherry (Schwab et al. 1990), passion fruit (Chassagne et al. 1996; Winterhalter 1990), kiwi (Young and Paterson 1995), papaya (Heidlas et al. 1984; Schwab et al. 1989), pineapple (Wu et al. 1991), mango (Sakho et al. 1997), lulo (Suarez et al. 1991), raspberry (Pabst et al. 1991) and strawberry (Roscher et al. 1997). The occurrence of glycosidically bound volatiles is typically two to eight times greater than that of their free counterparts (Gunata et al. 1985; Krammer et al. 1991). Moreover, most norisoprenoids in fruit, some of which are precursors of very potent flavour compounds, have been detected mainly in glycosidic forms. This, together with the low aroma threshold and sensory properties of aglycones, makes the glycosidic compounds an important potential source of flavour volatiles during fruit juice processing. Some aglycones are already odourous when released from glycosides. They can therefore contribute to the floral aroma of some wines (Mateo and Jimenez, 2000), grapes (Gunata et al. 1993), apricots (Chairote et al. 1981), peaches (Engel et al. 1988) and tea (Ogawa et al. 1997). This is the case of monoterpenes such as geraniol, nerol and linalool, which possess mainly floral attributes and low odour thresholds ( ppb, Rapp and Mandery 1986). Terpene compounds belong to the secondary plant constituents, the biosynthesis of which begins with acetyl- CoA (Manitto 1980; Fig. 1). Microorganisms are also able to synthesize terpene compounds (Hock et al. 1984), but the

2 Fig. 1 Mechanism of biosynthesis of monoterpenols in plants 323

3 324 formation of terpenes by Saccharomyces cerevisiae has not yet been observed (Rapp and Mandery 1986). Several authors have shown that terpenes play a significant role in the varietal flavour of wines by means of their transformation to other compounds (Cordonnier and Bayonove 1974; Gunata et al. 1985; Wilson et al. 1986). Classification of monoterpenes Three types of categories of monoterpenes exist in plant tissues, with some interrelationships among the categories. On the top of the complex are the free aroma compounds, commonly dominated by linalool, geraniol and nerol, together with the pyran and furan forms of the linalool oxides. However, depending on how the juice has been treated and on certain factors (which may include climate), many additional monoterpenes can be found in this group, i.e. citronellol, α-terpineol, hotrienol, nerol oxide, myrcenol, the ocimenols, plus several other oxides, aldehydes and hydrocarbons. In wines, several monoterpene ethyl ethers and acetate esters have also been found among the free aroma compounds. Second, there are the polyhydroxylated forms of the monoterpenes, or free odourless polyols. A most significant feature of the polyols is that, although these compounds make no direct contribution to the aroma, some of them are reactive and can break down with great ease to give pleasant and potent volatiles, e.g. diendiol (3,7-dimethylocta- 1,5-diene-3,7-diol) can give hotrienol and nerol oxide (Williams et al. 1980). Third, there are the glycosidically conjugated forms of the monoterpenes which also make no direct contribution to the aroma of the grape. Glycosides are, in most cases, more abundant than the unglycosilated forms of individual monoterpenes and polyols (Mateo and Jimenez 2000). Structure of glycosides Glycosidically bound volatiles identified in fruits and plants are highly complex and diverse, especially the aglycone moiety. The sugar parts consist of β-d-glucopyranosides and different diglycosides: 6-O-α-L-arabinofuranosyl-β- D-glucopyranosides, 6-O-α-L-arabinopyranosyl-β-D-glucopyranosides (vicianosides), 6-O-α-L-rhamnopyranosyl-β- D-glucopyranoside (rutinosides), 6-O-β-D-apiofuranosylβ-D-glucopyranosides, 6-O-β-D-glucopyranosyl-β-D-glucopyranosides and 6-O-β-D-xilopyranosyl-β-D-glucopyranosides (primeverosides) (Gunata et al. 1985; Vasserot et al. 1995). Figure 2 shows main glycosides found in Muscat grapes. In rare cases, trisaccaharide glycoconjugates have been isolated (Winterhalter and Skouroumounis 1997). The aglycon part is often formed with terpenols, but linalool oxides, terpene diols and triols can also been found. However, other flavour precursors can occur such as linear or cyclic alcohols, e.g. hexanol, phenylethanol, benzyl alcohol, C-13 norisoprenoids, phenolic acids and probably volatile phenols such as vanillin (Fig. 3; Allen et al. 1989; Schwab et al. 1990; Sarry and Gunata 2004). If we consider in grape juice, only glycosides with the most flavourant aglycons (i.e. with the same state of oxidation than linalool) are the most abundant are apiosylglycosides (up to 50% according to grape variety), followed by rutinosides (6 13%) and finally, glucosides (4 9%). A more accurate analysis indicates that all glycosides are not present in all cultivars, and that their proportions also differ according to grapes (Bayonove et al. 1993). The glycoside flavour potential from grapes remains, unfortunately, quite stable during winemaking and in young wines as well. Isolation of glycosides Isolation of glycosides by selective retention of the compounds on a solid-phase adsorbent is a commonly used technique. By washing the adsorbent with water following the adsorption step, free sugars and other polar constituents can be removed, while the less polar glycosides are retained. Elution with an organic solvent then gives the glycosidic fraction. Different methodologies have been proposed to extract glycoside precursors from grape juices and wines. Williams et al. (1982) used glass column chromatography containing C-18 reversed-phase adsorbent to extract glycosides from juice or de-alcoholised wine. After washing with water and eluting free compounds with 20% aqueous acetic acid, precursors were eluted in two fractions with 30% aqueous acetic acid and methanol. A modification of the methodology has been proposed by using 1 g solid phase extraction C-18 cartridges: hydrophylic compounds were eluted with water, free terpenes with dichloromethane and glycosides with methanol (Di Stefano 1991). This method has been improved in the last years, but it has a disadvantage, because separation is different depending on the commercial origin of the cartridges (R. Di Stefano, personal communication). A second approach to the problem has been proposed by Montpellier researchers by using Amberlite XAD-2 resin, because it possesses an excellent capacity for adsorption of free terpenols from grape juice (Gunata et al. 1985). This resin had been previously used to isolate naringin and limonin from grape juices. Once wine has been eluted through resin, free compounds were eluted with pentane and glycosides with ethyl acetate. Amberlite XAD- 2 resin displays extraction capacities similar to those of coated octadecyl silica. Furthermore, it has the advantage of being sold in large particle sizes; it is thus possible to use it in a wide preparative column at atmospheric pressure. A modification of this method has been suggested, and free compounds were eluted with pentane:dichloromethane to improve extraction (Di Stefano 1989). Nevertheless, even with extensive washing of the adsorbent beds, both Amberlite XAD-2 and Amberlite XAD-16 (the Amberlite XAD-16 resin has a surface area and capacity greater than that of Amberlite XAD-2) retained free glucose. Both of these adsorbents had the disadvantage of retaining free glucose in addition to the adsorbed glycosides (Williams et al. 1995).

4 325 Fig. 2 Main monoterpene compounds in grape juice and wines. Numbers like Table 1 Reversed-phase silica gel has been found to be a particularly suitable adsorbent for the isolation of glycosidic terpenes (Williams et al. 1995). The commercial availability of this adsorbent in uniform, pre-packed cartridges was an additional advantage in development of the isolation step. A method has been proposed to extract glycosilated flavour precursors from grape juices by microwaves (Bureau et al. 1996). This method was compared to the method using a column of Amberlite XAD-2 resin. Although a further purification of the extracts was required, this method provided the advantages of rapidity, ease of extraction and possibility of extract berries without deseeding or crushing. Glycosides fractionation Williams et al. (1982) laid stress on the existence, in the aromatic varieties of grapes, of two classes of precursors of C-10 and C-13 components with different polarity, separable by chromatography on C-18 cartridges. The method from these authors described of the separation of two glycosides classes, the absorption of de-alcoholised wine on C-18 cartridges of adequate size to the wine volume used, the elimination of hydrophylic components with water, the elution of more polar precursors with 30% acetic acid and the elution of monohydroxilated terpenic alcohols with methanol. The components in both fractions were recuperated by evaporation of solvent and subjected to another purification or hydrolysis.

5 326 Fig. 3 Structure of some glycoconjugated aroma compounds The critical points of this method are: (1) the incomplete elimination of volatile components during the evaporation of wine alcohol under vacuum and (2) the partial recuperation of the precursors with an acidic solvent, with the opportunity of hydrolytic adulterations and chemical transformations during the solvent evaporation. In addition, Strauss et al. (1988) fractionated by doplet countercurrent chromatography the C-10, C-13 and aromatic ring precursors in must previous isolation on C-18 RP column, and Bitteur et al. (1989) described an analytical HPLC method with C-18 column to fractionate aroma precursors, using acetonitrile and water as solvents. Following developments of chromatographic techniques in totally liquid phase (Skouroumounis and Winterhalter 1994) led to the fractionation of the precursors in leaves of Renan Riesling by using multilayer coil countercurrent chromatography. In this case, the precursors were previously isolated on Amberlite XAD-2 resin column. This technique has been used to isolate two novel terpenoid glucose esters from Riesling wines (Bonnlander et al. 1998). Thinking it helpful to also bring the fractionation techniques into laboratories without instruments for chromatography in totally liquid phase, an attempt to create a separation system for volatile precursors, using commercial C-18 RP columns and inert solvents, has been performed. The glycosides of mono-, di- and trihydroxilated terpene and norisoprenoid alcohols, as well the related shikimate pathway ones, have been isolated on C-18 RP cartridges and fractionated in classes having different polarity, with eluents at increasing percentages of methanol. The benzyl alcohol glycosides appear the most polar, while those of terpene monohydroxilated alcohols and of geranic acid the least polar. The terpene diols, linalool furanoid and pyranoid oxides, as well norisoprenoid precursors, show intermediate polarity and place themselves in well-fixed fractions according their polarity (Mateo et al. 1997). Acidic hydrolysis of terpene glycosides Acidic hydrolysis of terpene glycosides can induce a molecular rearrangement of the monoterpenols, which are transformed in other compounds. Nevertheless, these ways to liberate terpenes simulate the reactions taking place during ageing of wines, and the different terpenic alcohols were produced in similar quantitative rations. Experiments on both whole juice and monoterpene glycosides isolated from juice have demonstrated that significantly different patterns of volatile monoterpenes are produced when each is hydrolysed at different ph values. Furthermore, there appears to be a ph-dependent inter-

6 327 relationship among several of the grape monoterpenes. Thus, e.g., the isomerics ocimenols appear to be formed hydrolytically in juice at ph 1 at the expense of linalool, nerol and geraniol, the last three compounds being ph 3 products (Williams et al. 1982). For acidic hydrolysis, samples were dissolved in tartrate buffer at ph 3.2 or water to which perchloric acid was added to give ph of 1.0. After removing any residual volatile compound remaining, solutions were heated on a steam bath for 15 min, cooled and resulting free volatile compounds extracted and analysed by GC. Mateo et al. (1997) used tartaric acid 5 g l 1, NaCl 10% with a ph of 3.0 (adjusted with 1M NaOH) to dissolve samples and then heated them for 60 min in a boiling-water bath. R. Di Stefano (personal communication) used H 2 SO N to adjust the ph to 1.0 and heated te solutions for 30 min in a boiling-water bath. Significant differences can be seen in the patterns of volatile monoterpenoids produced from material by hydrolysis at ph 1.0 and ph 3.0. It is apparent that the more acidic conditions bring out extensive rearrangements of monoterpenoids (Williams et al. 1982). On hydrolysis at ph 3.0, linalyl, geranyl and neryl β-d-glucosides each gave the same major products, linalool and α-terpineol. Additionally, several compounds were observed in the hydrolysis of linalyl and geranyl glucosides that were not given by the neryl derivative, as geraniol, (E)- and (Z)-ocimene, α-terpinene and myrcene. α-terpenyl β-d-glucoside gave no acyclic monoterpenes, and α-terpineol was its predominant hydrolysis product. In contrast with the reaction at ph 3.0, the precursor material at low ph gave little linalool, and no geraniol or nerol was detected. While α-terpineol was still a significant product at ph 1.0, it was now accompanied by many other monoterpenoids not given by the precursors at ph 3.0 (Williams et al. 1982). Acid hydrolysis products are not diagnostic of the monoterpene aglycon composition of the grape precursors. Thus, while the grape glycosides are made up predominantly of geranyl, linalyl and neryl derivatives and only trace quantities of α-terpenyl glycosides, the hydrolysis products at ph 3.0 are dominated by linalool and α-terpineol, with geraniol relatively less abundant. Most of the compounds given under hydrolytic conditions at ph 3.0 are the free terpenes of the juice (Cordonnier and Bayonove 1974; Sarry and Gunata 2004). However, many of those products developed at ph 1.0 were seen in the headspace of Muscat juice heated for 15 min at 70 C and ph 3.2 (Williams et al. 1982). It has been observed that prolonged heating of juice at ph 3.0 ultimately altered the sensory character by imparting a eucalyptus-like aroma, Fig. 4 Acid catalysed rearrangement of monoterpenes

7 328 attributable to the presence of excessive quantities of 1,8- cineole in the headspace composition of the juice. The occurrence of ph 1.0 products 1-terpineol, 4-terpineol, β- terpineol and myrcenol in cognac can also be accounted for by hydrolytic degradation of grape monoterpene glycosides during wine distillation. The content of various monoterpene components change during bottle storage or rather during the maturation of wine by means of acid-catalysed reactions (Rapp and Mandery 1986; Di Stefano 1989). In the case of several components (e.g. linalool, geraniol, hotrienol and isomers of linalool oxide), an obvious decrease in concentration can be ascertained during the course of ageing. In addition to this, compounds are formed which are not present in young wines: among others the formation of cis- and trans-1,8 terpine. During the maturation of wine, monoterpene alcohols can be formed from linalool. According to these results, linalool is transformed in an aqueous acid medium to α-terpineol by cyclation, to hydroxy-linalool through hydration in the seventh position and to geraniol and nerol by a nucleophilic 1,3-transition (allyl) (Fig. 4) (Rapp and Mandery 1986). Enzymic hydrolysis of terpene glycosides Terpene glycosides can also be hydrolysed by an enzymic way, a more interesting way because it produces a more natural flavour in the wine (Cordonnier and Bayonove 1974; Gunata et al. 1985; Mateo and Jimenez 2000; Gunata 2002). The glycosidase flavour potential from grape remains, unfortunately, quite stable during winemaking and in young wines as well. So, to enrich wine flavour by release of free aromatic compounds from natural glycoside precursors, particularly pathways are required. Mainly, enzymic hydrolysis of glycosides is carried out with various enzymes which act sequentially according to two steps: firstly, α-l-rhamnosidase, α-l-arabinosidase or β-d-apiosidase make the cleavage of the terminal sugar and rhamnose, arabinose or apiose and the corresponding β-d-glucosides are released; subsequently liberation of monoterpenol takes place after action of a β-d-glucosidase (Fig. 5; Gunata et al. 1988). Nevertheless, one-step hydrolysis of disaccharide glycosides has also been described; enzymes catalysing this reaction have been isolated from tea leaves (Ogawa et al. 1997) and grapes (Gunata et al. 1998). This one step reaction occurs through the cleavage of the aglycone linkage which yields a disaccharide and aglycone, the identity of which have been confirmed by HPLC and GC/MS (Ogawa et al. 1997; Gunata et al. 1998) (Fig. 6). Studies regarding the mechanism of the enzymes which participate in the enzymic hydrolysis of terpene glycosides, individual or sequentially, have been made on glycoside fraction of the grapes or on different synthetic glycosides. Enzymic hydrolysis of glycoside extracts from Muscat, Riesling, Semillon, Chardonnay, Sauvignon and Sirah Fig. 5 Sequential enzymatic hydrolysis of disaccharidic flavour precursors

8 329 Fig. 6 GC-MS chromatogram of terpenes obtained by enzymic hydrolysis of glycosides. Numbers like Table 1 varieties have provoked the liberation not only of terpenes, but also C-13 norisoprenoids such as 3-oxo-α-ionol and 3- hydroxy-β-damascenona (Gunata et al. 1990a). These compounds are totally glycosilated in the grape and, as opposed to terpenes, they are found in the same quantities in all the grape varieties, aromatics or neutral, and they are capable of awarding certain typicity to the wine flavour, because they have lower threshold values than terpenes, and they contribute characteristic aromatic features (Razungles et al. 1987). Plant glycosidases Most work has been devoted to grape glycosidases, because the first evidence of multiple forms of glycosidic flavour precursors was found in this fruit. The potential use of enzymic systems of grapes to liberate terpenes from grape juices or wines has been always the subject of different works regarding the enzymic hydrolysis of terpene glycosides. Grapes have an enzyme with β-glucosidase activity (Table 1; Aryan et al. 1987; Gunata et al. 1990a,b; Di Stefano and García 1995) but only low α-rhamnosidase, α- arabinosidase or β-xylosidase activities have been detected (Gunata et al. 1990a); the presence of β-apiosidase in grapes has not yet been confirmed (Gunata 2002). On the other hand, grape β-glucosidase was not quite stable and showed low activity at grape juice or wines ph values (Lecas et al. 1991). β-glucosidase was isolated from mature grapes (cv. Muscat of Alexandria) and partially purified by chromatography on Ultrogel AcA 44 and DEAE-Sepharose CL- 6B. Two distinct proteins with β-glucosidase activity were isolated but, except for their mass, no differences occurred in other biochemical characteristics; the optimum ph of activity was 5.0 and the optimum temperature 45 C. For natural substrates, the two enzymes had a high affinity for geranyl β-d-glucoside and were inhibited by glucose, gluconolactone, Ca 2+,Cu 2+ and p-chloromercuribenzoate (Lecas et al. 1991). Similarly, certain β-glucosidase activity has been observed in grape leaves (Biron et al 1988; Gunata et al. 1990a; Di Stefano and García 1995). In general, β-glucosidases with a vegetal origin show a low activity on monoglucosides of terpenes with a tertiary alcohol group (linalool, α-terpineol), and they are only capable to hydrolyse monoglucosides of terpenes with a primary alcohol group (geraniol, nerol, citronellol) (Aryan et al. 1987; Gunata et al. 1990a). Endoglycosidase, active on disaccharidic flavour precursors in fruits, has been isolated and partiatilly characterized from grape skins (Gunata et al. 1998). Structure of this enzyme differs from that obtained from tea and could be a membrane bound protein. Moreover, grape endoglycosi-

9 330 Table 1 Results obtained by hydrolysis of terpene glycosides treated with different enzymatic preparations. Data are normalized to 100 in untreated wines No. Terpenes Grape skin Saccharomyces cerevisiae glucosidase Klerzyme200 Exogenous glycosidases Pectinol C Rohapect C Hemicellulase Sweet wine Dry wine 1 trans-furan linalool oxide 2 cis-furan linalool oxide 3 Linalool Hotrienol Neral 6 α-terpineol Geranial 8 trans-pyran linalool oxide 9 cis-pyran linalool > oxide 10 Citronellol Nerol , Geraniol Diol I Endiol Diol II Hydroxy-cityronellol Hydroxy-dihydrolinalool Hydroxy-nerol trans-8-hydroxylinalool , Hydroxy-geraniol cis-8-hydroxylinalool , Geranic acid Triol dase also seems to possess β-glucosidase activity (Gunata et al. 1998). Microbial glycosidases Glycosidases with oenological implications have been widely reported in distant taxomical species such as yeasts, bacteria and fungi (Grossmann et al. 1987; Gunata et al. 1990b; Boido et al. 2002). Yeast glycosidases Yeasts of the Hansenula species isolated from fermenting must were reported to have an inducible β-glucosidase activity, but this enzyme was inhibited by glucose (Grossmann et al. 1987). Other yeast strains such as Candida molischiana (Gonde et al. 1985) and C. wickerhamii (Leclerc et al. 1984) also possess activities towards various β-glucosides, and they were little influenced by the nature of aglycon (Gunata et al. 1990b). β-glucosidase from C. molischiana was immobilized to Duolite A-568 resin, showing similar physicochemical properties to those of free enzyme. The immobilized enzyme was found to be very stable under wine conditions and could be used repeatedly for several hydrolyses of bound aroma (Gueguen et al. 1997). Endomyces fibuliger also produces extracellular β-glucosidase when grown in malt extract broth (Brimer et al. 1998). Screening 370 strains belonging to 20 species of yeasts, all of the strains of the species Debaryomyces castelli, D. hansenii, D. polymorphus, Kloeckera apiculata and Hansenula anomala showed β-glucosidase activity (Rosi et al. 1994). A strain of D. hansenii exhibited the highest exocellular activity, and some wall-bound and intracellular activity and its synthesis, occurring during exponential growth, was enhanced by aerobic conditions and repressed by high glucose concentration. The optimum condition for this enzyme was ph and 40 C. This enzyme was immobilized using a one-step procedure on hydroxyapatite. The immobilized enzyme exhibited a lower activity

10 331 than the purified free enzyme, but was much more stable than the enzyme in cell-free supernatant (Riccio et al. 1999). These studies have shown the ability of several wine yeasts to hydrolyse terpenoids, norisoprenoids and benzenoids glycosides; among wine yeasts Hanseniaspora uvarum was able to hydrolyse both glycoconjugated forms of pyranic and furanic oxides of linalool (Fernandez et al. 2003) Other authors have also shown the important role of non-saccharomyces species in releasing glycosidid bound fraction of grape aroma components (Esteve et al. 1998; Mendes et al. 2001). Finally, the situation regarding S. cerevisiae is more complex, because this yeast is capable of modifying the terpenic profile of the wine; thus, it can produce citronellol from geraniol and nerol; the intensity of this transformation depends on the yeast strain used (Dugelay et al. 1992; Hernandez et al. 2003). Other authors propose a more complex scheme: geraniol was transformed by these yeasts into geranyl acetate, citronellyl acetate and citronellol, while nerol was transformed into neryl acetate; in addition, geraniol was transformed into linalool and nerol was cyclized to α-terpineol at must ph (Di Stefano et al. 1992). Few data are available regarding glycosidase activities of oenological yeast strains and the technological properties of the enzymes. Low α-rhamnosidase, α-arabinosidase or β-apiosidase activities were detected in S. cerevisiae (Delcroix et al. 1994). Nevertheless, data on β-glucosidase activity on Saccharomyces are contradictory. First results showed that these yeasts had a very low activity (Gunata et al. 1990c), but Delcroix et al. (1994) found three enological strains showing high β-glucosidase activity. On the other hand, Darriet et al. (1988) have shown that oxidases located in the periplasmic space of a strain of S. cerevisiae were able to hydrolyse monoterpene glucosides of Muscat grapes; they found also that the activity of this β-glucosidase was glucose independent. Mateo and Di Stefano (1998) have detected β-glucosidase activity in different Saccharomyces strains on the basis of its hydrolytic activity on para-nitrophenyl-β-d-glucoside (pnpg) and terpene glucosides of Muscat juice (Table 1). This enzymic activity is induced by the presence of bound β- glucose as a carbon source in the medium and seems to be a characteristic of the yeast strain. This β-glucosidase is associated with the yeast cell wall is quite glucose independent but is inhibited by ethanol. These results could open new pathways regarding other glycosidase activities in S. cerevisiae; α-rhamnosidase, α-arabinosidase or β- apiosidase activities could be induced in wine yeast by changing the composition of the medium including inductive compounds, as well as in filamentous fungi (Shoseyov et al. 1990; Dupin et al. 1992). Bacterial glycosidases Although little information is available regarding the β- glucosidase activity of lactic acid bacteria (LAB) involved in wine-making, Oenococcus oeni is the species generally recognized as beneficial for final aroma compound (Henick-Klling 1995; Maicas et al. 1999). This bacterium is involved in the malolactic fermentation (MLF), the conversion of the malic acid in wine to lactic acid and carbon dioxide. This reaction, generally associated to a finally equilibrated good wine (Fleet 1993), can also be performed by other LAB (Lactobacillus and Pediococcus), although it is usually performed by O. oeni (Davis et al. 1985). It has been demonstrated that O. oeni was able to cleave the glucose moiety from the major red wine anthocyanin, malvidin-3-glucoside, to use it as a carbon source. Release of glycosylated volatile precursors in Tannat wine was also observed (Boido et al. 2002), and only minimal O. oeni glycosidase activity was noted in Viognier glycosidic extracts (McMahon et al. 1999). The effective β-glucosidase activity of some strains of O. oeni has been recently reported (Grimaldi et al. 2000; Boido et al. 2002; Barbagallo et al. 2004; D Incecco et al. 2004). Mansfield et al. (2002) have also detected the production of β-glucosidase enzymes in several strains of O. oeni, although cultures of the same strains failed to hydrolyse native grape glycosides. The study of the hydrolysis of wine aroma precursors (linalool, α-terpineol, nerol and geraniol) during MLF has been carried out with some O. oeni starter cultures in model wine solutions by Ugliano et al. (2003). The liberation of glycosidically bound aroma compounds was assessed for various strains, which also performed the MLF. Although the quantity of released precursors was strain dependent, the large release of glycosylated aroma compounds observed during their experiments suggests that O. oeni can actively contribute to the changes of sensory characteristics of wine after MLF through the hydrolysis of aroma precursors. Fungal glycosidases Taking into account that enzymic systems of grapes are not suitable to hydrolyse terpene glycosides in grape juice or wine and that more studies are needed regarding the ability of S. cerevisiae and lactic acid bacteria to produce all the enzymes which take part in this process, several exogenous enzymes, mainly with a fungal origin, have been developed to liberate terpenes in wines. They are the so-called exogenous (fungal) glycosidases. Comparing 34 different enzymic preparations, the most suitable ones to be used during winemaking process are those which posses all β-dglucopyranosidase, α-l-arabinofuranosidase, α-l-rhamnopyranosidase and β-d-apiofuranosidase activities (Table 1; Cordonnier et al. 1989). With regard to the oenological uses of exogenous enzymes, the non-selectivity of the glycosidases in Rohapect C, together with its relative tolerance to ethanol, suggest a possible application in releasing any glycosidically bound terpenes present in wines. But even for this application, the high ph values required for the enzymes in this preparation ( ) to function effectively limits their practical utility. The great sensitivity to glucose of β-glucosidase of Rohapect C and several other commercially non-plant gly-

11 332 cosidases precludes any application of these enzymes in juice processing (Aryan et al. 1987). Aspergillus niger β-glucosidases from two different commercial preparations do not show the same hydrolysis rates towards linalyl β-d-glucosides. After 24 h of incubation, β-glucosidase from the Hemicellulase preparation (Gist Brocades, France) hydrolysed the substrate completely, whereas β-glucosidase from the Pectinol VP preparation (Röhm, Germany) hydrolysed only one third of it. α-terpenyl β-d-glucoside was not a substrate for A. niger β-glucosidase from Pectinol preparation; however, this substrate was hydrolysed by A. niger β-glucosidase from Hemicellulase preparation. Such variations in both commercial enzymic preparations are probably due to the use of different strains of A. niger (Gunata et al. 1990b). The ability of purified enzymes, synthetic substrates and suitable analytical methods are needed for the rigorous progress in the field of enzymic hydrolysis of terpenyl glycosides (Williams et al. 1982; Gunata et al. 1988; Bitteur et al. 1989; Voirin et al. 1990, 1992; Skouroumounis et al. 1995). Enzymes have been isolated from fungal enzymic preparations, vegetal extracts or synthetic culture media inoculated with fungal cultures and have been purified by different chromatographic techniques (gel filtration, ionexchange chromatography, affinity chromatography, chromatofocusing). β-apiosidase has been partially purified by filtration chromatography on Ultrogel AcA 44 and ion exchange chromatography on DEAE-Sepharose CL-6B from A. niger cultures. The sugar linkage (1 6) of apiosylglucosides of nerol, geraniol and pyran linalool oxide was cleaved by the enzyme with liberation of the corresponding monoglucosides. Furthermore, apiosylglucosides of furan linalool oxide did not appear to be a good substrate for β-apiosidase, unlike other apiosylglucosides. The production of the enzyme by A. niger is inducible, as it is only produced when apiin (2-O-β-D-apiofuranosyl-β-D-glucoside of apigenin) was present in the culture medium as the carbon source. In the same way, peptone (2% w/v) and Tween 80 (0.15% v/v) activate β-apiosidase synthesis. With regard to the activity, the optimum ph and temperature were, respectively, 5.6 and 50 C, and different ions have inhibitory effects on β- apiosidase activity. Conversely, the enzyme was inhibited neither by glucose nor by ethanol (Dupin et al. 1992). An α-l-arabinofuranosidase from A. niger was purified from a commercial crude preparation of Hemicellulase REG2 by gel filtration on Ultrogel AcA 44, ion exchange chromatography on DEAE-Sepharose CL-6B and affinity chromatography on Concanavalin A-Ultrogel AcA 22. The optimum ph of the enzyme was 3.9, and the temperature of maximal activity 60 C. This arabinosidase was active against monoterpenyl α-l-arabinofuranosylglucosides from grapes by liberating monoterpenyl β-dglucosides and arabinose regardless the structure of the aglycon moiety (Gunata et al. 1990b). This enzyme has been immobilised on chitosan by conjugation; this method proved to be the better one as it ensures good biocatalyst activity and better stability than the reported for the free enzyme (Spagna et al. 1998). An α-l-rhamnopyranosidase from a naringinase commercial preparation from Penicillium sp. and freed from contaminating β-d-glucosidase activity by chromatofocusing on PBE 94 using a ph gradient. Rhamnosidase isoenzymes were eluted at ph 6.2 and 5.7, respectively (Gunata et al. 1988). A strain of A. niger excreted into the medium a β-glucosidase which was partially purified by affinity chromatography. The enzyme was found to hydrolyse natural glycosides and was competitively inhibited by glucose. The ph optimum was 3.4, and ethanol enhanced the activity. Immobilization of fungus using Ca 2+ alginate beads enabled an increase in enzyme production in a continuous fermentation (Shoseyov et al. 1988). Later, an endo-βglucosidase fom A. niger was immobilized to acrylic beads (Shoseyov et al. 1990) or to γ-alumina activated with dodecamethylendiamine and glutaraldehyde in sequence (Fu-Mian et al. 1994). GC-MS analysis of the wines indicated that the enzyme treatment increased concentrations of free monoterpene alcohols. Aspergillus oryzae was found to secrete two distinct β-glucosidases when it was grown in liquid culture. The major form was highly inhibited by glucose, but the minor form, which was induced on quercitin, exhibited high tolerance to glucose or gluconolactone inhibition. The enzyme was optimally active at 50 C and ph 5.0. It exhibits exoglucanase activity and was able to release flavour compounds (Riou et al. 1998). Similar results were obtained in A. niger (Gunata and Vallier 1999). These enzymes are present only in low quantities in the majority of commercial fungal enzymic preparations, mainly regarding β-d-apiosidase activity (Dupin et al 1992). Determination of free volatile compond terpenols, norisoprenoids and volatile phenols indicates that concentration in enzyme-treated wines highly increase, not only in aromatic varieties, but also in neutral ones. Enhancements widely varied according to the original grapes, from 265% to 2,000% (Gunata et al. 1990c). Nevertheless, multiple forms of glycosidases could be present in an enzymic preparation. So, chromatofocusing on PBE 94 of a crude enzyme preparation from A. niger showed the presence of multiple forms of β-apiosidase, β- glucosidase, α-rhamnosidase and α-arabinosidase (Gunata et al. 1997). Enzymes are effective during some days from their addition to the wine, while glycosidic precursors remain almost unchanged along fermentation. The method used actually to enrich wine in volatile compounds is only effective when dry wines are used, because β-glucosidase in fungal enzymic preparations is hardly inhibited by glucose. The hydrolysis of terpene glycosides is not completed in sweet wines obtained from Muscat de Frontignan grapes, breaking in the monoglucoside stay due to inhibition of β-glucosidase by the presence of glucose (Gunata et al. 1990b).

12 Potential adverse effects of enzyme preparations When enzymic preparations have been used to improve the aromatic characteristics of the wines, undesirable odours have been sometimes detected, even if liberation of terpenes has been produced. These wines show high concentrations in vinyl-phenols (4-vinyl-phenol, 4-vinyl-guayacol), reaching concentrations up to 1 mg l 1. The presence of these compounds is due to the action of two different enzymic systems. Firstly, a cinnamyl esterase from the enzymic preparation hydrolyses cinnamic esters, the main phenolic derivatives in grape juices; so, the medium enriches in p- coumaric and ferulic acids, which are subsequently decarboxylated by yeast cinnamate decarboxylase producing vinyl-phenols. It had been impossible to explain the formation of these undesirable compounds in wines when enzymic preparations were used in winemaking (mainly for clarification processes) until this conjugate action of enzymes was discovered. Based on these data, industries have developed new fungal enzymic preparations with no cinnamyl esterase activity (Gunata 1993; Sefton and Williams 1994). Hydrolytic activity of fungal enzyme preparations towards anthocyanins was reported as early as 1955 (Huang 1955). Enzymic activities from commercial preparations may potentially degrade anthocianins, composed of anthocyanidin glycosilates (Yang and Steele 1958; Blom 1983). The de-coloration effect of these enzymes has been reported for raspberry (Jiang et al. 1990), strawberry (Rwahahizi and Wrolstad 1988), blackberry (Yang and Steele 1958) and grape anthocyanins (Fu-mian et al. 1994; Wightman et al. 1997). So, the choice of enzyme preparations for flavour enhancement in fruit juices and derived beverages should be made carefully. Glycosidases that are highly active on anthocyanins may have specific applications for the decoloration of some orange juice or wines containing low levels of anthocyanins pigments. On the other hand, fungal enzyme preparations generate oxidation artefacts during the hydrolysis of glycosides (Sefton and Williams 1994); this fact implies that, perhaps, the review of all the information regarding the use of these enzymic preparations will be necessary. Trends and prospects The presence of terpenes, in their different forms, in grape juices and wines represents an enormous potential in a way to increase the varietal characteristics of the wines, contributing the final product with higher fruit-like characteristics. Actually, researchers have sufficient information and tools to study the presence of terpenes and their evolution in grape juices and wines, but it is not yet possible to translate all of the acquired knowledge to the wineries, because an efficient methodology to improve the terpene content of all the wines present in the market has not been found. Several approaches through genetic engineering techniques will doubtless be used in the near future to overcome, by using recently discovered glucosidases, the limitations on the exploitation of the glycosylated aroma source in the processing of fruit juices and derived beverages. On the other hand, construction of fungal strains expressing these proteins involved in flavour liberation may enable the production of tailor made glycosidases without undesirable activities. So, great work remains to be made in a way to obtain a methodology, probably of enzymic nature, which allows that wine-consumer sense of smell can, at last, appreciate the whole organoleptic richness of the product that he has on his hand or in his mouth. Acknowledgements This work was supported by Conselleria de Cultura, Educació i Esport, Generalitat Valenciana (Spain) project GV04B-175. References 333 Allen MS, Lacey MJ, Brown WV, Harris RLN (1989) Sauvignon blanc varietal aroma. Aust Grapegrower Winemaker 292:51 56 Aryan AP, Wilson B, Strauss CR, Williams PJ (1987) The properties of glycosidases of Vitis vinifera and comparison of their β-glucosidase activity with that of exogenous enzymes. An assessment of possible applications in enology. Am J Enol Viticult 38: Barbagallo RN, Spagna G, Palmeri R, Restuccia C, Giudici P (2004) Selection, characterization and comparison of β-glucosidase from mould and yeasts employables for enological applications. Enzyme Microbiol Technol 35:58 66 Bayonove C, Gunata Y, Sapis JC, Baumes RL, Dugelay I, Grassin C (1993) L aumento degli aromi nel vino mediante l uso degli enzimi. Vignevini 9:33 36 Biron C, Cordonnier C, Glory O, Gunata Z, Sapis JC (1988) Etude dans le raisin de l activité β-glucosidase. Conn Vigne Vin 22: Bitteur S, Gunata Z, Brillouet JM, Bayonove C, Cordonnier R (1989) GC and HPLC of grape monoterpenyl glycosides. J Sci Food Agric 47: Blom H (1983) Partial characterization of a thermostable anthocyanin-β-glycosidase from Aspergillus niger. Food Chem 12: Boido E, Lloret A, Medina K, Carrau F, Dellacasa E (2002) Effect of β-glucosidase activity of Oenococcus oeni on the glycosylated flavor precursors of Tannat wine during malolactic fermentation. J Agric Food Chem 50: Bonnlander B, Baderschnider B, Messerer M, Winterhalter P (1998) Isolation of two novel terpenoid glucose esters from Riesling wine. J Agric Food Chem 46: Brimer L, Nout MJR, Tuncel G (1998) Glycosidase (amygdalase and linamarase) from Endomyces fibuliger (LU677): formation and crude enzyme properties. Appl Microbiol Biotechnol 49: Bureau S, Razungles A, Baumes R, Bayonove C (1996) Glycosylated flavor precursor extraction by microwaves from grape juice and grapes. J Food Sci 61: Chairote G, Rodriguez F, Crouzet J (1981) Characterization of additional volatile flavor components of apricot. J Agric Food Chem 46: Chassagne D, Crouzet JC, Bayonnove CL, Brillouet JM, Baumes RL (1996) 6-O-α-D-Arabinopyranoyl-β-D-glucopyranosides as aroma precursors from passion fruit. Phytochemistry 41: Cordonnier R, Bayonove C (1974) Mise en evidence dans la baie de raisin, variété Muscat d Alexandrie, de monotèrpenes liés, révélables par une ou plusieurs encimes du fruti. CR Acad Sci Paris 278:

13 334 Cordonnier RE, Gunata Z, Baumes RL, Bayonove CL (1989) Recherche d un materiel enzymatique adapté à l hydrolyse des précurseurs d arômes de nature glucosidique du raisin. Conn Vigne Vin 23:7 23 Darriet P, Boidron JN, Dubourdieu D (1988) L hydrolyse des hétérosides terpéniques du Muscat a Petit Grains par les enzymes périplasmiques de Saccharomyces cerevisiae. Conn Vigne Vin 22: Davis CR, Wibowo D, Eschenbruch R, Lee TH, Fleet GH (1985) Practical implications of malolactic fermentation: a review. Am J Enol Viticult 36: Delcroix A, Gunata Z, Sapis JC, Salmon JM, Bayonove C (1994) Glycosidase activities of three enological yeast strains during wine making. Effect on the terpenol content of Muscat wine. Am J Enol Viticult 45: Di Stefano R (1989) Evoluzione dei composti terpenici liberi e glucosidici e degli actinidioli durante la conservazione dei mosti e dei vini in funzione del ph. Riv Vitic Enol 2:11 23 Di Stefano R(1991) Proposition d une methode de preparation de l echantillon pour la determination des terpenes libres et glycosides des raisins et des vins. Bull OIV 4: Di Stefano R, Magiorotto G, Gianotti S (1992) Transformazioni di nerolo e geraniolo indotte dai lieviti. Riv Vitic Enol 42:43 49 Di Stefano R, Garcia E (1995) Attività idrolitica di preparati enzimatici da foglie e da acini di Moscato sui glicosidi terpenici. Riv Vitic Enol 48:53 58 D Incecco N, Bartowsky E, Kassara S, Lante A, Spettoli P, Henschke P (2004) Release of glycosidically bound flavour compounds of Chardonnay by Oenococcus oeni during malolactic fermentation. Food Microbiol 21: Dugelay I, Gunata Z, Sapis JC, Baumes R, Bayonove C (1992) Etude de l origine du citronellol dans les vins. J Int Sci Vigne Vin 26: Dupin I, Gunata Z, Sapis JC, Bayonove C, M Bairaroua O, Tapiero C (1992) Production of a β-apiosidase by Aspergillus niger. Partial purification, properties and effect on terpenyl apiosylglucosides from grape. J Agric Food Chem 40: Engel KH, Flath RA, Butter RG, Mon TR, Ramming DW, Teranishi R (1988) Investigation of volatile constituents in nectarines. 1. Analytical and sensory characterization of aroma components in some nectarine cultivars. J Agric Food Chem 36: Esteve B, Manzanares P, Ramon D, Querol A (1998) The role of non-saccharomyces yeasts in industrial wine making. Int Microbiol 1: Fernandez M, Di Stefano R, Briones A (2003) Hydrolysis and transformation of terpene glycosides from Muscat must by different yeast species. Food Microbiol 20:35 41 Fleet GH (1993) Wine microbiology and biotechnology. Harwood, Switzerland Francis MJO, Allcock C (1969) Geraniol β-d-glucoside: occurrence and synthesis in rose flowers. Phytochemistry 8: Fu-Mian C, Pifferi PG, Setti L, Spagna G, Martino A (1994) Immobilizzazione di un antocianasi (β-glucosidasi) da Aspergillus niger. Ital J Food Sci 1:31 42 Gonde P, Ratomahenina R, Arnaud A, Galzy P (1985) Purification and properties of the exocellular β-glucosidase of Candida wickerhamii (Zikes) Meyer and Yarrow capable of hydrolysing soluble cellodextrines. Can J Biochem Cell Biol 63: Grimaldi A, McLean H, Jiranek V (2000) Identification and partial characterization of glycosidic activities of commercial strains of the lactic acid bacterium, Oenococcus oeni. Am J Enol Viticult 51: Grossmann M, Rapp A, Rieth W (1987) Enzymatische freisetzung gebundener aromastoffe in wein. Dtsch Lebensm Rdsch 83:7 12 Gueguen Y, Chemardin P, Pien S, Arnaud A, Galzy P (1997) Enhancement of aromatic quality of Muscat wine by the use of immobilized β-glucosidase. J Biotechnol 55: Gunata YZ, Bayonove CL, Baumes RL, Cordonnier RE (1985) The aroma of grapes. I. Extraction and determination of free and glycosidically bound fraction of some grape aroma components. J Chromatogr 331:83 90 Gunata Z, Bitteur S, Brillouet JM, Bayonove C, Cordonnier R (1988) Sequential enzymic hydrolysis of potentially aromatic glycosides from grape. Carbohydr Res 184: Gunata Z, Bayonove C, Tapiero C, Cordonnier R (1990a) Hydrolysis of grape monoterpenyl β-d-glucosides by various β- glucosidases. J Agric Food Chem 38: Gunata Z, Brillouet JM, Voirin S, Baumes B, Cordonnier R (1990b) Purification and some properties of an α-l-arabinofuranosidase from Aspergillus niger. Action on grape monoterpenyl arabinofuranosyl glucosidases. J Agric Food Chem 38: Gunata Z, Dugelay I, Sapis JC, Baumes R, Bayonove C (1990c) Action des glycosidases exogènes au cours de la vinification: Liberation de l arôme à partir des précurseurs glycosidiques. J Int Sci Vigne Vin 24: Gunata Z (1993) Etude et exploitation par voie enzymatique des precurseurs d arômes du raisin de nature glycosidique. Rev Oenol 74:22 27 Gunata Z, Dugelay I, Sapis JC, Baumes R, Bayonove C (1993) Role of the enzymes in the use of the flavour potential from grape glycosides in winemaking. In: Schreier P, Winterhalter P (eds) Progress in flavour precursor studies. Alured, Carol Stream, pp Gunata Z, Dugelay I, Vallier MJ, Sapis JC, Bayonove C (1997) Multiple forms of glycosidases in an enzyme preparation from Aspergillus niger: partial characterization of an apiosidase. Enzyme Microbiol Technol 21:39 44 Gunata Z, Blondeel C, Vallier MJ, Lepoutre JP, Sapis JC, Watanabe N (1998) An endoglycosidase from grape berry skin of cv. M. Alexandrie, hydrolysing potentially aromatic disscharide glycosides. J Agric Food Chem 46: Gunata Z, Vallier MJ (1999) Production of highly glucose-tolerant extracellular β-glucosidase by three Aspergillus strains. Biotechnol Lett 21: Gunata Z (2002) Flavor enhancement in fruit juices and derived beverages by exogenous glycosidases and consequences of the use of enzyme preparations. In: Whitaker JR (ed) Handbook of food enzymology. Marcel Dekker, New York, pp Heidlas H, Lehr M, Idstein H, Schreier P (1984) Free and bound terpene compounds in papaya (Carica papaya L.) fruit pulp. J Agric Food Chem 32: Henick-Klling T (1995) Control of malo-lactic fermentation in wine: energetics, flavour modification and methods of starter culture preparation. J Appl Bacteriol 79(Suppl):29 37 Hernandez LF, Espinosa JC, Fernandez M, Briones A (2003) β-glucosidase activity in a Saccharomyces cerevisiae wine strain. Int J Food Microbiol 80: Hock R, Benda J, Schreier P (1984) Formation of terpenes by yeasts during alcoholic fermentation. Z Lebensm Unters Forsch 179: Huang HT (1955) Fruit colour destruction: decolorization of anthocianins by fungal enzymes. J Agric Food Chem 3: Jiang J, Paterson A, Piggott JR (1990) Effects of pectolytic enzyme treatments on anthocyanins in raspberry juice. Int J Food Sci Technol 25: Krammer G, Winterhalter P, Schwab M, Schreier P (1991) Glycosidically aroma bound compounds in the fruits os Prunus species: apricot (P. armeniaca L.), peach (P. persica L.), yellow plum (P. domestica L. ssp syriaca). J Agric Food Chem 39: Lecas M, Gunata Z, Sapis JC, Bayonove C (1991) Purification and partial characterization of β-glucosidase from grape. Phytochemistry 30: Leclerc M, Arnaud A, Ratomahenina R, Galzy P, Nicolas M (1984) The enzyme system in a strain of Candida wickerhamii Meyer and Yarrow participating in the hydrolysis of cellodextrins. J Gen Appl Microbiol 30: Lutz A, Winterhalter P (1992) Isolation of additional carotenoid metabolites from quince fruit (Cydonia oblonga Mill). J Agric Food Chem 40: