Background

Aroma is one of the most appreciated traits in assessing wine quality, and among the hundreds of volatile compounds characterized only a small number influence its sensory perception (see [1, 2] and references therein). These aroma active compounds (e.g. terpenes, esters, alcohols) have their origins in grapes, the metabolism of microorganisms (especially the wine-making yeast Saccharomyces cerevisiae), and wine aging and storage conditions.

Monoterpenes (a C10 class of terpenes mainly derived from grapes) are key odorants associated with the varietal (or primary) aromas of certain white wines. Linalool, geraniol, nerol, citronellol and α-terpineol are the major constituents of aromatic grape varieties (e.g. Muscat d’Alexandrie, Gewürztraminer, Riesling), imparting floral and fruity attributes (reviewed in [3, 4]), and certain dietary monoterpenes are of nutraceutical importance because of their antimicrobial, antiviral, anti-proliferative, antioxidative, anxiolytic, hypotensive or anti-inflammatory properties, among other activities (see [5–8] and references therein). Apart from the natural properties of a grape variety, monoterpene content is also influenced by uncontrollable factors such as climate and soil. A large proportion of these monoterpenes is present in grape musts as non-volatile odorless sugar glycoconjugates that can be released enzymatically using industrial glycosidase cocktails or recombinant wine yeast strains expressing such activities (for reviews see [9–11]). Nevertheless, a number of grape varieties are aromatically ‘neutral’ and almost completely lack free monoterpenes and their precursors [4]. Thus there is considerable variability in monoterpene contents in grapes.

Monoterpene biosynthesis in plants is effected by monoterpene synthases (MTPSs). Many of their corresponding genes have been characterized [12, 13] and considerable expansion of these has been observed in grapevine (Vitis vinifera) [14, 15]. S. cerevisiae wine strains themselves produce only tiny quantities of monoterpenes (e.g. up to 1.2 or 4 μg/L of geraniol and linalool, respectively) [16] because they lack MTPSs and cannot therefore contribute to ameliorating monoterpene deficiency in grape must. Notwithstanding the non-acceptability of GMOs, especially by European wine consumers and industries, vinification by engineered monoterpene-producing wine yeast strains could thus constitute a means to enhance varietal wine aroma. In this regard, successful expression of the Clarkia breweri S-linalool synthase (LIS) gene in a S. cerevisiae wine yeast strain has provided proof of concept by virtue of de novo production of linalool in wines to about 19 μg/L [17]. This metabolic manipulation was possible because plant MTPSs catalyze the synthesis of monoterpenes from geranyl pyrophosphate (GPP) in a single step, and S. cerevisiae has enough free GPP (an intermediate in ergosterol biosynthesis) under vinification conditions to be used as a substrate by these plant enzymes. In addition S. cerevisiae has the ability to metabolize supplemented monoterpenes, the bioconversions of (i) geraniol into citronellol, linalool, nerol and geranyl acetate, (ii) nerol into geraniol, linalool and α-terpineol, (iii) linalool into α-terpineol and (iv) citronellol into citronellyl acetate having been reported (see [18–20] and references therein). Thus an engineered monoterpene-producing yeast could also play a valuable additional role in the development of wine aroma by producing a wider spectrum of monoterpenes.

Previous work has shown that wine yeast strain T73 has a greater inherent capacity for recombinant monoterpene production compared to other laboratory and industrial wine strains [21]. Here we report the substantial modification of the terpene profile of a wine produced from a neutral grape variety using the T73 strain expressing the geraniol synthase (GES) gene from Ocimum basilicum (sweet basil) [22].

Results and discussion

Production of geraniol by a wine yeast strain expressing the GES gene of O. basilicum and its metabolic fate in synthetic defined (YPD) media

The truncated O. basilicum GES cDNA [22] (GenBank Accession No. AY362553) coding for a geraniol synthase lacking the first 34 codons—which encode the plastid transit peptide—was cloned under the control of the S. cerevisiae ACT1 (encoding actin) promoter (ACT1 p) and the HIS3 (encoding imidazole glycerol-phosphate dehydratase) terminator (HIS3 t) into the binary vector YEplac195 [23]. The resulting plasmid (YEp195Ges) was used to transform the S. cerevisiae T73-4 [24] wine strain and the uracil prototrophic (ura+) transformants YR377 and YR378 (T73Ges) were isolated. The growth rates of YR377, YR378 and the control strain YR70 (T73-4 transformed with the empty plasmid) on liquid YPD media were almost identical, albeit slightly slower than that of the industrial strain T73 (Fig. 1a) as observed previously for other recombinant yeasts [17]. This indicates that neither the amount of geraniol nor the putative reduction of precursors from the isoprenoid pathway apparently produce deleterious effects on yeast growth under such conditions. In addition, GC and GC–MS analyses of these culture media showed similar extraordinarily high geraniol yields (8,017.85 ± 1,245.81 and 7,859.12 ± 1,614.62 μg/L after 32 h) (Fig. 1b). These levels are about 16-fold higher than those produced by recombinant S. cerevisiae laboratory strains expressing the same GES gene, about 1.6-fold the amount produced by laboratory yeasts co-expressing GES and an optimized farnesyl diphosphate synthase [25, 26], and about 120-fold the amount of linalool excreted by engineered T73-4 wine strains expressing LIS [17, 21]. These results clearly reinforce the previously shown importance of the genetic background of this industrial yeast for monoterpene production [21] but also that of the monoterpene synthase being expressed. In contrast to the T73Lis strains that produced linalool as the only end product, and in agreement with the reported ability of S. cerevisiae T73 to metabolise supplemented geraniol and its reaction products [20], the T73Ges strains produced geraniol (84.83%) and geraniol derivatives i.e. citronellol (10.92%), nerol (3.90%), linalyl acetate (0.13%), geranyl acetate (0.12%) and linalool (0.1%). As expected, monoterpene production by the control strains lacking GES (YR70 and T73) was practically negligible (7.13 ± 1.12 μg/L; >1,300-fold lower than YR377 and YR378) (Fig. 1b). YR377 was chosen for the microvinification experiments.

Fig. 1
figure 1

Growth and monoterpene production in YPD of recombinant wine yeast T73-4 expressing the O. basilicum GES gene. a Growth curves of T73Ges (YR377 and YR378) and control strains YR70 (T73-4 transformed with the empty plasmid) and T73. b Monoterpene production at 24 and 32 h by YR377, YR378 and controls. Numbers above the bars corresponding to 32 h indicate μg/L. Terpene concentrations are represented on a logarithmic scale. Results are presented as the mean and standard deviations of two independent assays with three replicates each.

Full size image

Aromatic wines from neutral grapes using the self-aromatizing wine yeast YR377

Microvinification experiments were carried out in parallel on sterile Parellada white grape must using the wine yeast strain T73-4 that carries the GES expression cassette (YR377) and a control strain lacking GES (YR70). Both alcoholic fermentations progressed similarly (Fig. 2b) and reached completion in about 19 days leaving approximately 2 g/L residual sugar (i.e. dry wine). Given the persistence of the ura+ phenotype (around 85%) in YR377 and hence high maintenance of the GES expression cassette throughout the process it is evident that neither the expression of the GES gene nor its consequences affected the growth or fermentative capacity of the engineered wine strain.

Fig. 2
figure 2

Analyses of microvinifications. Microvinifications were carried out with YR377 (T73Ges) and the YR70 control strain transformed with the empty vector. a Schematic representation of the engineered mevalonate pathway in the T73Ges strains. IPP, isopentenyl pyrophosphate; GPP, geranyl pyrophosphate; FPP, farnesyl pyrophosphate; FPPS, FPP synthase. b Growth curves and kinetics of sugar consumption by YR377 and YR70 during the course of fermentations. Results are presented as the mean and standard deviation.

Full size image

To evaluate the influence of GES expression on wine aroma, volatile profiles were determined by GC and GC–MS (Fig. 3a). As expected given the aromatic neutrality of the Parellada grape, free geraniol was undetectable in wines produced by YR70. In contrast, geraniol concentrations (~750 μg/L) well in excess of its olfactory perception threshold (40–75 μg/L) and exceeding those present in wines obtained from the highly aromatic Muscat grapes (Additional file 1: Table S1) were found in wines fermented with the ‘self-aromatizing’ wine yeast YR377 (Fig. 3b; Table 1). Remarkably, GC analysis (Fig. 3) showed that apart from the geraniol peak there were also notable quantities (~810 μg/L) of additional monoterpenes and esters associated with strain YR377: citronellol, linalool, nerol, citronellyl acetate and geranyl acetate, resulting in a total terpene concentration >220-fold greater than the control wine. With the exception of nerol and citronellyl acetate, the other compounds are present above their perception thresholds (Table 1). The presence of geraniol and its derivatives will enrich these wines with flowery and fruity notes.

Fig. 3
figure 3

Presence of geraniol, citronellol, linalool, nerol, citronellyl acetate and geranyl acetate in wines produced by YR377. a Comparison of the chromatograms of wines produced by YR377 (T73Ges) and YR70 (control). Arrows indicate peaks of geraniol and its derivatives. Peak numbers refer to the aromatic compounds listed in Table 1. Asterisks indicate retention time of the internal control 2-octanol. The insert amplifies the region of the chromatogram corresponding to the monoterpenes. b Geraniol, and geraniol derivative structures and their contents in wines.

Full size image
Table 1 Concentrations (μg/L), odor quality and thresholds of a selected subset of aromatic compounds found in Parellada wines fermented with the T73Ges strain
Full size table

GES and E. coli-expressed recombinant GES both exclusively catalyze the synthesis of geraniol from GPP [22]. Our comparative GC–MS data (Figs. 2, 3) revealed that the same terpenes found in wine were also excreted by YR377 when grown in synthetic (YPD) medium. Thus during vinification the enzymatic activities intrinsic to this wine yeast strain are also able to metabolize geraniol and its derivatives resulting in their conversion to other monoterpenes and aromatic esters, a situation resembling the metabolic diversion occurring in tomato expressing the GES gene [27]. The reduction of geraniol to citronellol and the acetylation of geraniol and citronellol are probably catalyzed by the oxidoreductase Oye2 and the alcohol acetyl transferase Atf1 [28], respectively. An obvious strategy to further expand our ability to modulate wine aroma would therefore be to promote or suppress the formation of these geraniol derivatives by modification of these enzyme activities.

To investigate whether GES expression could lead to additional changes in a wine’s volatile profile, determinations of other volatile compounds of oenological relevance were carried out on both recombinant yeast derived and control wines. The compositions and concentrations of higher alcohols (e.g. 2-phenylethyl alcohol) and acetate esters (e.g. isoamyl acetate), the presence of which is considered favorable for the aromatic properties of wines, were seen to be statistically similar in wines fermented with YR377 and YR70 strains (Table 1).

Introduction of the C. breweri LIS gene into wine yeast strain T73-4 (T73Lis) under the control of the TDH3 yeast promoter was our first attempt to construct a self-aromatizing wine yeast [17]. This resulted in de novo accumulation in wine of linalool alone to levels exceeding its odor perception threshold. Remarkably, the amount of geraniol-derived linalool produced by YR377 (T73Ges) was about 7.5 times greater than that obtained with T73Lis (~141 versus ~19 μg/L) and the total de novo terpene concentration is more than 80-times greater, illustrating the importance of the MTPS employed in engineering strain T73. These results justify the strategy of engineering the wine yeast isoprenoid pathway as a means of achieving efficient plant-derived aromatic monoterpene production during alcoholic fermentation.

Mixed fermentation with T73Ges and S. cerevisiae strains not producing monoterpenes serves to modulate levels of terpenes

In order to assess whether it would be feasible to produce wines with a predetermined monoterpene content, vinifications were conducted using mixed starters (1:1) of yeast strains YR377 and YR70 and were compared to those obtained using pure cultures of YR377. The monoterpene profiles of wines derived from mixed fermentations were directly related to the composition of the initial inocula. Thus the amounts of geraniol (~388 μg/L) and its derivatives (~311 μg/L) detected were about half of those obtained using inocula of YR377 alone (Table 2).

Table 2 Concentrations (μg/L) of geraniol and derivatives found in Parellada wines co-fermented with GES strains
Full size table

Terpenes are also important flavor compounds in other fermented drinks. Geraniol, linalool and citronellol have all been shown to be important contributors to the floral, fruity and citrus flavors of beer [29], and biotransformations of these monoterpenes by ale and lager yeasts have been reported [19]. Engineered brewing-yeasts designed as vehicles for the de novo production of these monoterpenes thus have potential for use in the brewing industry. Moreover, certain monoterpenes have been shown to display a plethora of potential health benefits (see [5–8] and references therein).

Conclusions

These results demonstrate the considerable potential for geraniol-engineered yeasts in the development of wines with aromas ‘à la carte’. Fermentation of grape musts with these and/or other yeast strains expressing novel plant MTPS genes and thus the possibility of producing monoterpenes absent from grapes will provide variety and novelty to the wine industry. Approaches including the manipulation of enzyme activities responsible for monoterpene bioconversions [28], the engineering of rate-limiting reactions in the mevalonate pathway [21] and/or the possibility of using diverse mixed starters to pre-determine monoterpene content could contribute to the enhancement of complexity in wine aroma (Fig. 4).

Fig. 4
figure 4

Schematic representation of the isoprenoid pathway in S. cerevisiae including the branch point to monoterpenes. Gene names of S. cerevisiae appear in blue. Red and green arrows indicate engineered steps to increase monoterpene content in wines (this work and [17], respectively) catalysed by plant linalool (LIS) and O. basilicum geraniol (GES) synthases. Monoterpene bioconversions appear with red letters [18–20, 28]. HMG-CoA, 3-hydroxy-3-methylglutaryl-coenzyme A; IPP, isopentenyl pyrophosphate; DMAPP, dimethylallyl pyrophosphate; GPP, geranyl pyrophosphate; FPP, farnesyl pyrophosphate; GGPP, geranyl geranylpyrophosphate.

Full size image

The work reported brings to the fore once again the question of whether modern genetic technologies, in this instance for improving wine yeasts, may become acceptable to industry and consumers given the continued resistance to transgenic foods mainly in Europe. The advance reported in our study illustrates the biotechnological improvement of a food beyond the use of this type of technology for generating resistance to herbicides and pests via the genetic manipulation of a plant, and instead offers a clear alternative to transgenic grapes engineered to enhance free monoterpene content.

Methods

Strains and culture conditions

Escherichia coli DH5α [endA1, hsdR17, gyrA96, thi-1, relA1, supE44, recA1, ΔlacU169 (Φ80 lacZΔM15)] was used for cloning experiments and plasmid propagation. S. cerevisiae wine strain T73-4 [ura3::470/ura3::470] [24] (derived from T73, Lallemand) was used for GES expression. E. coli was maintained in LB medium (1% tryptone, 0.5% yeast extract, 1% NaCl) with or without 100 μg/mL ampicillin. S. cerevisiae strains were maintained in YPD-rich medium (1% yeast extract, 2% bacteriological peptone, 2% glucose) or SD-minimal medium (0.17% yeast nitrogen base without amino acids—Difco Laboratories, Detroit, USA—2% glucose, 0.5% ammonium sulphate) with or without 20 mg/L uracil. For solid media, 1.5% agar was added. To determine terpene yields of recombinant yeasts, aliquots from overnight cultures of selected transformants grown in SD medium lacking uracil were transferred to 250 mL flasks containing 50 mL of YPD medium at an initial OD600 of 0.05. Yeast cultures were grown with continuous shaking (200 rpm) at 30°C and aliquots of the cultures were taken at different times.

Construction of yeast plasmids carrying the GES gene of C. breweri and yeast transformation

The GES cDNA was obtained from pCRT7/CT-TOPO/GES [22] via PCR as a 1.6-kb EcoRI [T4 DNA polymerase treated for blunt-ending]-BspLU11I fragment using the oligonucleotide pair GES-L35-Bs (5′-CCCACGCTA C A T G T CTGCTTGCACGCCTTTGG-3′; BspLU11I is in italics and the artificial translation start site codon ATG and the GES-S35 TCT codon appear in bold) and GES-STOP-RI (5′-CCCCCGAATTCTATTTATTGAGTGAAGAAGAGG-3′). The HIS3 t was isolated as a 0.66-kb HincII-SphI fragment obtained by PCR using genomic DNA of the S. cerevisiae strain FY1679 (MAT a/MATα ura3-52/ura3-52) and the oligonucleotide pair His3_SalI (5′-AGGT CGACTAGTGACACCGATTATTTAAAGCTG-3′) and His3_SphI (5′-AGG C ATG CGAATTCGGATCCTCGGGGACACCAAATATGG-3′). These two fragments were subcloned downstream of ACT1 p in the plasmid YEpACT4 [30] previously digested with NcoI and SphI, thereby generating plasmid YEp181Ges (2 μ; LEU2). The expression cassette ACT1 p::GES::HIS3 t was isolated from this plasmid as a 2.8-kb EcoRI fragment and subcloned into the same site of YEplac195 (2 μ; URA3). The resulting plasmid (YEp195Ges) was used to transform the S. cerevisiae T73-4 [24] wine strain and uracil (ura+) prototrophic transformants (T73Ges) were thus isolated. To obtain the control strain YR70, T73-4 was transformed with YEplac195.

DNA manipulations were performed following standard protocols [31]. PCR fragments were individually cloned into the pGEM-T Easy vector (Promega) and the absence of mutations was confirmed by sequencing. Transformation of the T73 derived strain was done using lithium acetate to permeabilize the cells as previously described [24, 32]. Transformants were selected and maintained on SD plates without uracil. For plasmid stability analyses, transformants were grown under both selective (SD) and non-selective (YPD) conditions and the colonies growing under each condition were counted.

Microvinifications

Two temporally independent microvinifications were done in triplicate at 20°C using 250 mL glass bottles containing 200 mL of Parellada white grape must (Villafranca del Penedés, Spain). The must (ºBrix ~15) was centrifuged and sterilized with 0.2% (v/v) dimethyl dicarbonate (Velcorin; Bayer, Levercusen, Germany) and inoculated with 9 × 105 cells/ml from overnight cultures of YR70 (uracil nutritional control) and YR377 (T73Ges). Samples were collected periodically to measure yeast growth and sugar consumption and thus monitor the progress of the fermentations. Sugar concentrations were initially measured as Brix grades using a Euromex RD. 5645 digital refractometer. After 15 days ºBrix stabilized to about 5, and reducing sugar concentrations were measured using the Nelson–Somogyi method [33, 34] to determine the end of the fermentations (‘dry wine’; sugar concentration below 2 g/L). At this point (day 19), the persistence of the plasmids was measured (% colonies grown on selective SD compared to those grown on complete YPD medium), the wines were centrifuged to remove yeast cells and then transferred to new bottles that were kept at −20°C until their analysis.

GC–MS analysis of volatiles

Geraniol, geraniol derivatives and other volatiles were extracted and analyzed by headspace solid-phase microextraction (HS-SPME) using poly(dimethylsiloxane) (PDMS) fibers (Supelco, USA) coupled to gas chromatography (GC) and GC–mass spectrometry (MS) as previously reported [17]. 2-Octanol (0.2 μg) was used as internal control. Identification of compounds was determined by comparing retention times and mass spectra with those of standards using a Thermo-Scientific model Focus-GC equipped with a HP-Innowax column (length 30 m; inside diameter 0.25 mm; film thickness 0.25 μm) and a Thermo Trace GC Ultra gas chromatograph coupled to a Thermo DSQ mass spectrometer (Thermo-Scientific), under the same chromatographic conditions. Ion spectra of the peaks of interest were identified by comparison with computerized libraries (e.g. Wiley6, NIST). The oven temperature was programmed as follows: 60°C for 5 min, raised to 190°C at 5°C/min, then raised to 250°C at 20°C/min and held 2 min at 250°C. The injector temperature was 220°C. Helium was the carrier gas at 1 mL/min in the splitless mode. Compounds were quantified by integrating the peak areas of GC chromatograms.