gene technology in winemaking: new approaches to an

27REVIEW PAPER

Agriculturae Conspectus Scientificus, Vol. 66, No. 1, 2001 (27-47)

Gene Technology in Winemaking:New Approaches to an Ancient Art

Institute for Wine Biotechnology, University of StellenboschStellenbosch 7600, South Africa

Received: December 20, 2000

ACKNOWLEDGEMENTS

The research conducted in the Institute for Wine Biotechnology is supported bygrants from the National Research Foundation (NRF) and the South African WineIndustry (Winetech).

Isak S. PRETORIUS

SUMMARY

For the last century, the availability of pure culture yeast has improvedreproducibility in wine fermentations and product quality. However, there isnot a single wine yeast strain that possesses an ideal combination ofoenological characteristics that are optimised for the task set by today’s leadingwinemakers. With new developments in modern winemaking there has arisenan urgent need to modify wine yeast strains in order to take full advantage oftechnology and to satisfy the demands of the sophisticated wine consumers.The combined use of mutagenesis, hybridisation and recombinant DNAmethods have significantly increased the genetic diversity that can beintroduced into Saccharomyces cerevisiae strains. The overall aim of the straindevelopment programmes extends far beyond the primary role of wine yeastto catalyse the rapid and complete conversion of grape sugars into alcoholand carbon dioxide without distorting the flavour of the final product. Startercultures of S. cerevisiae must now possess a range of other properties thatdiffer with the type and style of wine to be made and the technicalrequirements of the winery. Our strain development programme focuses ona number of targets that are amenable to a genetic approach, including strainsecurity and quality control, the increase of fermentation and processingefficiencies, and the enhancement of the sensorial quality and healthproperties of wine and other grape-based beverages. However, successfulcommercialisation of transgenic wine yeasts will depend on a multitude ofscientific, technical, economic, marketing, safety, regulatory, legal and ethicalissues. Therefore, it would be foolish to entertain unrealistic expectations overrapid commercialisation and short-term benefits. However, it will be equallyunwise to deny the potential advantages of genetically improved wine yeaststo both the winemaker and consumer in the third millennium.

KEY WORDS

Saccharomyces cerevisiae, wine yeast, genetic improvement

Agric. conspec. sci. Vol. 66, No. 1, 2001

Isaak S. PRETORIUS28

INTRODUCTION

Yeasts have been used in the production of wine andother fermented beverages and foods for several thou-sand years. Originally, yeast species and strainspresent on the grapes and cellar equipment wereresponsible for the “spontaneous” and unpredictablefermentation which took place. But, over the lastcentury, Saccharomyces cerevisiae strains have beenselected on the basis of their fermentation behaviourand this has led to significant improvements in bothfermentation control and wine quality.

No controlled breeding of wine yeasts has been at-tempted until relatively recently. This is largely be-cause the yeasts used in most research laboratoriesdiffer significantly from those used in winemakingand, as a consequence, the techniques developedwith laboratory-bred strains cannot be applied di-rectly to wine yeasts. For the same reason studies ofthe genetics of wine yeasts have lagged behind inves-tigations of laboratory strains of S. cerevisiae. In ad-dition to these technical difficulties that restrict theprogress with strain development, it is not surpris-ing that with traditional fermentation methods andproducts there was initially little need to change theyeast strain; in fact, with local wines with their ownunique character there was every reason for preserv-ing the strain or strains then in use. However, today’sfierce competition for market share in a world wherethere is an over supply of wine at several price points,fundamental innovations in viticultural andwinemaking practices will undoubtedly continue torevolutionise the wine industry during the 21st cen-tury. These forces of market-pull and technology-push

are therefore expected to challenge the tension be-tween the tradition and innovation. Clearly, “tradi-tional” wine yeasts are far from optimised for thedemanding task which they are set by modern day’sinnovative winemakers and sophisticated consumers.

There is an ever-growing demand for new and im-proved wine yeast strains. In addition to the primaryrole of wine yeast to catalyse the efficient and com-plete conversion of grape sugars to alcohol withoutthe development of off-flavours, starter culturestrains of S. cerevisiae must now possess a range ofother properties. The importance of these additionalyeast characteristics differs with the type and styleof wine to be made and the technical requirementsof the winery. The need is for S. cerevisiae strains thatare better adapted to the different wine-producingregions of the world with their respective grapevarietals, viticultural practices and winemaking tech-niques.

GENETIC TECHNIQUES FOR THE ANALYSISAND DEVELOPMENT OF WINE YEASTS

S. cerevisiae can be modified genetically in manyways. Some techniques alter limited regions of the ge-nome, whereas other techniques are used to recom-bine or rearrange the entire genome. Techniques

having the greatest potential in genetic programmingof wine yeast strains are: clonal selection of variants,mutation and selection, hybridisation, rare-mating,spheroplast fusion and gene cloning and transforma-tion. The combined use of mutagenesis, hybridisationand recombinant DNA methods have dramaticallyincreased the genetic diversity that can be introducedinto yeast cells.

The classical genetic techniques all have value instrain development programmes, but these methodslack the specificity required to modify wine yeasts ina well-controlled fashion. It may not be possible todefine precisely the change required using thesegenetic techniques, and a new strain may bring animprovement in some aspects, while compromisingother desired characteristics. Gene cloning and trans-formation offer the possibility to alter the character-istics of wine yeasts with surgical precision: themodification of an existing property, the introductionof a new characteristic without adversely affectingother desirable properties, or the elimination of anunwanted trait. By using such procedures it is possi-ble to construct new wine yeast strains that differfrom the original only in single specific characteris-tics.

TARGETS FOR STRAIN DEVELOPMENT

It is well established that wine yeasts vary markedlyin their winemaking abilities. Some of the propertiesrequired by many leading winemakers are complexand difficult to define genetically without a betterunderstanding of the biochemistry and physiologyinvolved. To date, no wine yeast in commercial usepossesses an ideal combination of oenological char-acteristics. While some degree of variation can beachieved by altering the fermentation conditions, amajor source of variation is the genetic constitutionof the wine yeasts. Fortunately, recombinant DNAprosedures are now available which allow this prob-lem to be addressed. Many possible targets for geneticmodification have been identified and reviewed ex-tensively by Pretorius (2000). The following sectionsare an excerpt from this recent review.

Improved quality control and strainhandling

Strain maintenance

One of the main objectives for using pure culturesin winemaking is to ensure reproducible fermenta-tion performance and product quality. It is thereforeimportant to maintain the genetic identity of wineyeasts and to slow down the rate of strain evolutioncaused by sporulation and mating, mutations, geneconversions and genetic transpositions. Total preven-tion of heterogeneity in pure cultures is impossible,since homothallism, inability to sporulate and mate,and polyploidy (multiple gene structure) only protectagainst genetic drift caused by sexual reproductionand mutation, but not against that caused by gene


GENE TECHNOLOGY IN WINEMAKING: NEW APPROACHES TO AN ANCIENT ART 29

conversion and transposition. Even stringently-con-trolled conditions for maintenance of culture collec-tions (i.e., freeze-dried cultures, cultures preserved inliquid nitrogen or in silica gel) will not render fullprotection against genetic drift in pure yeast cultures.In fact, freeze-drying (lyophilisation) for long-termmaintenance of yeast stock cultures causes phasetransitions in membrane lipids and cell death duringfreeze-thawing and may also induce respiratory defi-cient variants. In an attempt to improve resistance tocryo-damage, anti-freeze peptides from polar fishwere successfully expressed in S. cerevisiae.Cryoprotectants and osmoprotectants such as cellu-lar trehalose and glycerol also alleviate freeze-thawand water stress. Trehalose appears to stabilise cellmembranes of lyophilised or cryopreserved stockcultures as well as their cellular proteins by replac-ing water and forming a hydration shell around pro-teins. Glycerol accumulation seems to controlintracellular solute potential relative to that of theculture medium thereby counteracting the deleteri-ous effects of dehydration on lyophilised yeast cells.

A better understanding of how yeast cells preciselyacquire cryotolerance and osmotolerance may lead togenetic modification of starter culture strains withgreater robustness for industrial fermentations. How-ever, at present, fermentation trials, continuous strainevaluation and early detection of genetic changesusing comparative molecular techniques are the onlypractical ways to limit possible economic loss.

Molecular marking

The potential for genetic markers in wine yeast iden-tification has been recognised and deliberatelymarked oenological strains were developed as an aidto monitor the kinetics of yeast populations duringwine fermentations. Genetic labeling could also beregarded as a quality control tool in general yeastculture management as well as in trouble-shooting,particularly for wineries using more than one yeaststrain, as the genomes of commercial wine yeasts canbe tagged. This would also discourage illegitimate useof (patented) commercial wine yeast strains by ‘pi-rate’ yeast and wine producers.

The marking of wine yeast strains usually entails theintegration of specific genetic markers into theirgenomes. This could take the form of syntheticoligonucleotides or foreign genes of known nucle-otide sequences. These DNA sequences can then beused as ‘diagnostic probes’ to identify specific wineyeast strains. In one instance a wine yeast was dou-ble-marked with diuron and erythromycin resistancegenes. A more sophisticated manner of marking wasthe expression of the Escherichia coli β-glucuronidase(GUS) gene (uidA) under control of the yeast alco-hol dehydrogenase I promoter and terminator se-quences. The GUS construct was integrated into theILV2 gene of S. cerevisiae and a simple assay proce-dure was devised to detect GUS activity in yeast cellsor colonies. In a GMO (‘genetically modified organ-

ism’) risk assessment experiment, this yeast is cur-rently being used to monitor the dissemination oftransgenic yeast strains on vines cultivated in a bio-logically contained glass house. This will undoubtedlyprovide an insight into the kinetics of transgenic andnative yeast populations on vines.

Improvement of fermentation performance

The primary selection criteria applied to most straindevelopment programmes relate to the overall objec-tive of achieving a better than 98% conversion ofgrape sugar to alcohol and carbon dioxide, at a con-trolled rate and without the development of off-fla-vours. The growth and fermentation properties ofwine yeasts have, however, yet to be genetically de-fined. What makes the genetic definition of theseattributes even more complex is the fact that lagphase, rate and efficiency of sugar conversion, resist-ance to inhibitory substances and total time of fer-mentation are strongly affected by the physiologicalcondition of the yeast as well as by the physicochemi-cal and nutrient properties of grape must.

Generally, sugar catabolism and fermentation pro-ceed at a rate greater than desired, and are usuallycontrolled by lowering the fermentation temperature.Occasionally, wine fermentation ceases prematurelyor proceeds too slowly. Measures to rescue such ‘slug-gish’ or ‘stuck’ fermentations include the increase offermentation temperature, addition of vitamin sup-plements, limited aeration by pumping over and re-inoculation. The commercial implications of‘runaway’ wine fermentations arise from the fact thatfermentor space is reduced because of foaming andvolatile aroma compounds are lost by entrainmentwith the evolving carbon dioxide. Conversely, finan-cial losses through sluggish or incomplete winefermentations are usually attributed to inefficientutilisation of fermentor space and wine spoilage re-sulting from the low rate of protective carbon diox-ide evolution and high residual sugar content.Optimal performance of wine yeasts in white winefermentations, conducted at cooler temperatures (10-150C) so as to minimise the loss of aromatic volatiles,and red wine fermentations, performed at highertemperatures (18-300C) to enhance extraction of an-thocyanin pigments, is therefore of critical impor-tance to wine quality and cost-effectiveness.

Fermentation predictability and wine quality is di-rectly dependent on wine yeast attributes that assistin the rapid establishment of numerical and meta-bolic dominance in the early phase of wine fermen-tation, and that determine the ability to conduct aneven and efficient fermentation with a desirable re-sidual sugar level. A wide range of factors affect thefermentation performance of wine yeasts. Apart froma successful inoculation with the appropriate starterculture strain, the physiological condition of such anactive dried wine yeast culture, and its ability to adaptto and cope with nutritional deficiency and the pres-



ence of inhibitory substances, is of vital importanceto fermentation performance.

Successful yeast cellular adaptation to changes in ex-tracellular parameters during wine fermentation re-quires the timely perception (sensing) of chemical orphysical environmental parameters, followed by ac-curate transmission of the information to the relevantcompartments of the cell. Chemical signals emanat-ing during wine fermentations include the availabil-ity/concentration of certain nutrients (e.g.,fermentable sugars, assimilable nitrogen, oxygen, vi-tamins, minerals, ergosterol and unsaturated fattyacids) and the presence of inhibitory substances (e.g.,ethanol, acetic acid, fatty acids, sulphite,agrochemical residues and killer toxins). Signals of aphysical nature include factors such as temperature,pH, agitation and osmotic pressure. As an example,physiological and morphological modifications inresponse to a limited supply of essential nutrientssuch as carbon and nitrogen sources include a shiftin transcription patterns, the modification of the cellcycle, a change in budding pattern and strongly po-larised growth. It is becoming clear that a complexnetwork of interconnected and cross-talking signaltransduction pathways, relying on a limited numberof signaling modules, governs the required adaptiveresponses to changes that occur as the fermentationprogresses.

This complexity explains why it is so difficult to de-fine all the key genetic determinants of a yeast’s fer-mentation performance that may be candidates forgenetic engineering. However, general targets in-clude increased tolerance to desiccation and viabil-ity of active dried yeast; improved grape sugar uptakeand assimilation; increased ethanol tolerance; im-proved nitrogen assimilation; enhanced resistance tomicrobial metabolites and toxins; resistance to heavymetals and agrochemical residues; tolerance to sul-phite; and reduced foam formation.

Improved viability and vitality of active dried

wine yeast starter cultures

Both the genetic and physiological stability of stockcultures of seed yeast and wine yeast starter culturesare essential to optimal fermentation performance.Physiological stability and ‘fitness’ of active driedwine yeast cultures relate to the maintenance of cellviability and vitality during the process of yeast manu-facturing, including desiccation and storage. The dif-ferentiation between yeast ‘viability’ and ‘vitality’ isbased upon the fact that cells which irreversibly losetheir ability to reproduce may still be capable of ac-tive metabolism. Therefore ‘viability’ is defined as therelative proportion of living cells within an activedried starter culture, whereas ‘vitality’ refers to themeasure of metabolic activity and relates to fitnessor vigour of a starter culture. Yeast viability can beassessed directly by determining loss of cell reproduc-tion/division (e.g., plate and slide counts) and indi-rectly by assessing cellular damage (e.g., vital staining

with bright-field or fluorochrome stains) or loss ofmetabolic activity (e.g., ATP bioluminescence andNADH fluorescence). Yeast vitality can be indirectlyassessed by measuring metabolic/fermentative activ-ity (e.g., CO2 evolution in mini-scale fermentations),storage molecules (e.g., glycogen), intracellular/extra-cellular pH (acidification power) and gaseous ex-change coefficients (e.g., respiratory quotients orRQ). Automatic in-line monitoring of yeast cell viabil-ity in fermentation plants can be achieved withelectrosensors such as capacitance probes or withfluorescent probes coupled with flow cytometrywhich can rapidly determine cell viability and otheraspects of yeast physiology (e.g., stress responses).These techniques generally show varying degrees ofcorrelation with fermentative performance and noneof them alone can accurately predict the physiologi-cal activity of a active dried wine yeast starter culture.

The manufacturers of active dried wine yeast startercultures can positively influence the degree of viabil-ity and vitality as well as the subsequent fermenta-tion performance of their cultures by the way theycultivate their yeasts. Industrial cultivation of wineyeasts can have a profound effect on the microbio-logical quality, fermentation rate, production of hy-drogen sulphide, ethanol yield and tolerance,resistance to sulphur dioxide as well as tolerance todrying and rehydration. For example, if a protein tophosphate ratio (P2O5:N) of 1:3 in a yeast cell is ex-ceeded it would result in an excess of water linkedto the protein which would, in turn, negatively affectthe drying procedure, viability and final activity ofthe dry yeast. Due to the roles that trehalose and gly-cogen play in a yeast cell’s response to variations inenvironmental conditions, it is generally recom-mended that the manufacturers of active dried wineyeast starter cultures cultivate their yeast in such away that the maximum amount of these storage car-bohydrates is accumulated in the yeast cells.

In S. cerevisiae, trehalose (α-D-glucopyranosyl α-D-glucopyranoside) is synthesised from glucose-6-phos-phate and UDP-glucose by the TPS1-encodedtrehalose-6-phosphate synthetase and converted totrehalose by the TPS2-encoded trehalose-6-phosphatephosphatase. The regulation of trehalose synthesisand degradation (by trehalase) is mediated by cAMP-dependent phosphorylation mechanisms. Trehaloseis associated with nutrient-induced control of cellcycle progression; control of glucose sensing, trans-port and initial stages of glucose metabolism; as wellas stress protection against dehydration, freezing,heating and osmo-stress, and toxic chemicals such asethanol, oxygen radicals and heavy metals. This stor-age carbohydrate plays an important role duringsporulation, nutrient starvation, growth resumptionand growth rate. Trehalose content in the yeast cellis probably one of the most important factors affect-ing the resistance of yeasts to drying and subsequentrehydration. The accumulation of this disaccharideon both sides of the plasma membrane is thought to



confer stress protection by stabilising the yeast’smembrane structure.

Glycogen, another carbohydrate reserve whose accu-mulation by yeast propagated for drying has also beenlinked to enhanced viability and vitality upon reacti-vation, provides a readily mobilisable carbon andenergy source during the adaptation phase. The bio-synthesis of glycogen (α -1,4-glucan with α -1,6branches) is effected by glycogen synthase, whichcatalyses the sequential addition of glucose fromUDP-glucose to a polysaccharide acceptor in a linearα-1,4 linkage, while branching enzymes are respon-sible for the formation of α-1,6 branches. There aretwo forms of glycogen synthase in S. cerevisiae,Gsy1p and Gsy2p. The GSY1 gene is expressed con-stitutively at a low level along with growth on glucose,while the level of the GSY2-encoded glycogen syn-thase increases at the end of the exponential phaseof growth when glycogen accumulates. This indicatesthat GSY2 encodes the major glycogen synthase. Gly-cogen breakdown, catalysed by glycogen phosphory-lase quickly following depletion of nutrients at theend of fermentation, is accompanied by sterol forma-tion. Since sterol is essential for yeast vitality, lowlevels of accumulated glycogen in active dried wineyeast starter cultures may result in insufficient yeaststerols, which, in turn, may impair yeast performanceupon inoculation into grape juice. In this regard, itis important to note that the overexpression of theSUT1 and SUT2 genes has been shown to promote theuptake of sterol from the medium under fermentativeconditions.

Owing to its multiple roles in increasing survival ofS. cerevisiae cells exposed to several physical andchemical stresses, trehalose and glycogen have impor-tant implications for the viability, vitality and physi-ological activity of active dried wine yeast startercultures upon reactivation. Therefore, there is astrong incentive to develop wine yeast strains with asuperior trehalose and glycogen accumulation abil-ity. However, due to the complexity of yeast viability,vitality and physiological activity, it is unclear at thisstage whether the modification of the expression lev-els of the TPS1, TPS2, GSY1, GSY2, SUT1 and/or SUT2

genes, would contribute to yeast fitness and fermen-tation performance of starter culture strains.

Efficient sugar utilisation

In S. cerevisiae, glucose and fructose, the main sug-ars present in grape must, is metabolised to pyruvatevia the glycolytic pathway. Pyruvate is decarboxylatedto acetaldehyde, which is then reduced to ethanol.The rate of fermentation and the amount of alcoholproduced per unit of sugar during the transformationof grape must into wine is of considerable commer-cial importance. During wine yeast glycolysis, onemolecule of glucose or fructose yield two moleculeseach of ethanol and carbon dioxide. However, thetheoretical conversion of 180 g sugar into 92 g etha-nol (51.1%) and 88 g carbon dioxide (48.9%) could

only be expected in the absence of any yeast growth,production of other metabolites and loss of ethanolas vapor. In a model fermentation, about 95% of thesugar is converted into ethanol and carbon dioxide,1% into cellular material and 4% into other productssuch as glycerol.

The first step to ensure efficient utilisation of grapesugar by wine yeasts is to replace any mutant allelesof genes encoding the key glycolytic enzymes,namely hexokinase (HXK), glucokinase (GLK),phosphoglucose isomerase (PGI), phosphofructoki-nase (PFK), aldolase (FBA), triosephosphate isomer-ase (TPI), glyceraldehyde-3-phosphate dehydrogenase(TDH), phosphoglycerate kinase (PGK), phos-phoglycerate mutase (PGM), enolase (ENO), pyruvatekinase (PYK), pyruvate decarboxylase (PDC) and al-cohol dehydrogenase (ADH). The genes encodingPGI, TPI, PGM and PYK appear to be present in sin-gle copy in a haploid genome, while multiple formsexist for TDH (three isozymes), ENO (two isozymes)and ENO/GLK (three isozymes).

The assumption that an increase in the dosage ofgenes encoding these glycolytic enzymes would re-sult in an increase in the efficiency of conversion ofgrape sugar to alcohol has been disproved; it has beendemonstrated that overproduction of the enzymeshas no effect on the rate of ethanol formation. Thisindicates that the step of sugar uptake represents themajor control site for the rate of glycolytic flux un-der anaerobic conditions, whereas the remainingenzymatic steps do not appear to be rate limiting. Inother words, the rate of alcohol production by wineyeast is primarily limited by the rate of glucose andfructose uptake. Therefore, in winemaking, the lossof hexose transport toward the end of fermentationmay result in reduced alcohol yields.

Sugars enter yeast cells in one of three ways: simplenet diffusion, facilitated (carrier-mediated) diffusionand active (energy-dependent) transport. In grapemust fermentations where sugar concentrationsabove 1 M are common, free diffusion may accountfor a very small proportion of sugar uptake into yeastcells. However, since the plasma membranes of yeastcells are not freely permeable to highly polar sugarmolecules, various complex mechanisms are requiredfor efficient translocation of glucose, fructose andother minor grape sugars into the cell. The hexosetransporter family of S. cerevisiae consists of morethan 20 proteins comprising high, intermediate andlow affinity transporters and at least two glucosesensors. Many factors affect both the abundance andintrinsic affinities for hexoses of these transporterspresent in the plasma membrane of wine yeast cells,among them glucose concentration, stage of growth,presence or absence of molecular oxygen, growthrate, rate of flux through the glycolytic pathway andnutrient availability (particularly of nitrogen).

Although the precise mechanisms and regulation ofgrape sugar transport of wine yeast are still unclear,some aspects about glucose and fructose uptake can



be noted. Glucose uptake is rapid down a concentra-tion gradient, reaching an equilibrium and is there-fore not accumulative. Several specific,energy-dependent glucose carriers mediate the proc-ess of facilitated diffusion of glucose and protonsymport is not involved. Phosphorylation by the theHXK1- and HXK2-encoded hexokinases and theGLK1-encoded glucokinase is linked to high-affinityglucose uptake. Glucose transporters, encoded byHXT1-HXT18 and SNF3 are stereospecific for certainhexoses and will translocate glucose, fructose andmannose. Some members of this multigene permeasefamily affect glucose, galactose, glucose and mannose,or glucose, fructose and galactose uptake, but thusfarnone has been described as specifically affecting fruc-tose uptake. It appears that in S. cerevisiae fructoseis transported via facilitated diffusion rather thanactive transport, whereas related species (S. bayanus

and S. pastorianus) within the Saccharomyces sensu

stricto group do possess fructose-proton symporters.

Based on the spectacular increase in the amount ofinformation on sugar sensing and their entry intoyeast cells that has come to the fore over the last fewyears, several laboratories have identified this mainpoint of control of glycolytic flux as one of the keytargets for the improvement of wine yeasts. For ex-ample, in some instances, certain members of theHXT permease gene family are being overexpressedin an effort to enhance sugar uptake, thereby improv-ing the fermentative performance of wine yeaststrains. However, more in-depth details are requiredabout the complex regulation of glucose and fructoseuptake as well as glycolysis as it occurs in grape juice(especially in the presence of high sugar levels dur-ing the early phase of fermentation and during thefinal stages of sugar depletion coupled to nutrientlimitation) before it will be possible to devise novelstrategies to improve wine yeast’s fermentation per-formance and to prevent sluggish or stuckfermentations.

Improved nitrogen assimilation

Of all nutrients assimilated by yeast during winefermentations, nitrogen is quantitatively second onlyto carbon. Carbon-nitrogen imbalances and more spe-cifically, deficiencies in the supply of assimilable ni-trogenous compounds, remain the most commoncauses of poor fermentative performance and slug-gish or stuck fermentations. Such problematic andincomplete fermentations occur because nitrogendepletion irreversibly arrests hexose transport. Otherproblems that are related to the nitrogen compositionof grape must include the formation of reduced-sul-phur compounds, in particular hydrogen sulphide,and the potential formation of ethyl carbamate frommetabolically produced urea.

Unlike grape sugars that are usually present in largeexcess (often exceeding 20% w/v) to that needed formaximal yeast growth, the total nitrogen content ofgrape juices ranges 40-fold from 60 to 2400 mg/l and

can therefore be growth limiting. The assimilablecontent of grape must is dependent upon grapecultivar and root stock, as well as several aspects ofvineyard management, including nitrogen fertilisa-tion, berry maturation, vine water status, soil type andfungal infection. Grape juices with nitrogen levelsbelow 150 mg/l have a high probability of becomingproblem ferments due to inadequate yeast growthand poor fermentative activity. There are two basicstrategies to circumvent problems linked to nitrogendeficiency: prevention of nitrogen deficiency ingrape juice by optimising vineyard fertility, and morecommonly, supplementation with ammonium saltssuch as diammonium phosphate (DAP). However, theinjudicious use of DAP supplements often contra-venes the wine industry’s desire to minimise its useof additives while producing wines of high quality.Moreover, excessive addition of inorganic nitrogenoften results in excessive levels of residual nitrogen,leading to microbial instability and ethyl carbamate(and phosphate in the case of DAP) accumulation inwine. The degree of supplementation of inorganicnitrogen in grape juice is therefore often regulated.This implies that knowledge of the nitrogen contentof grape juice and the requirement for nitrogen byyeast are important considerations for optimal fer-mentation performance and the production of winesthat comply with the demands of regulatory authori-ties and consumers.

The major nitrogenous compounds in the averagegrape must are proline, arginine, alanine, glutamate,glutamine, serine and threonine, while the ammo-nium ion levels may also be high, depending on grapevariety and vineyard management. Proline and ar-ginine accounts for 30 to 65% of the total amino acidcontent of grape juices. High proline accumulationin grape must is associated with grapevine stress, inparticular with low moisture, whereas high levels ofγ-aminobutyrate, another nitrogen compound, may beformed in the grape berries most probablypostharvest and prior to processing of the grapes.

S. cerevisiae is incapable of adequately hydrolysinggrape proteins to supplement nitrogen-deficientmusts, and relies therefore on the ammonium andamino acids present in the juice. Wine yeasts can dis-tinguish between readily and poorly used nitrogensources. Ammonium is the preferred nitrogen sourceand as it is consumed, the amino acids are taken upin a pattern determined by their concentration rela-tive to yeast’s requirements for biosynthesis and tototal nitrogen availability. When a readily used nitro-gen such as ammonium, glutamine and asparagine ispresent, genes involved in the uptake and catabolismof poorly utilised nitrogen sources (including pro-line) are repressed. This nitrogen catabolite repres-sion exerted upon nonpreferred nitrogenouscompounds rigorously impairs the assimilation ofproline as well as arginine since both amino acidsdepend on the proline utilisation pathway. Since theproline content of wine is generally not less than



grape juice, it appears that proline is not taken up bywine yeast under anaerobic fermentative conditions.Proline is transported into S. cerevisiae by the gen-eral amino acid permease and the PUT4-encodedproline-specific permease. Once inside the yeast cell,proline is converted to glutamate in mitochondria bythe PUT1-encoded proline oxidase and PUT2-en-coded pyrroline-5-carboxylate dehydrogenase. Theexpression of both PUT1 and PUT2 is regulated bythe PUT3-encoded activator and the URE2-encodedrepressor. Ure2p represses transcription of PUT1 andPUT2 under nitrogen-limiting conditions, while theGLN3-encoded regulator has no effect on these genes.

Since wine yeast strains vary widely in their nitrogenrequirement, an obvious target for strain improve-ment is to select or develop starter strains that aremore nitrogen efficient for use in low nitrogen musts.To achieve this a thorough understanding of the regu-lation of nitrogen assimilation by yeast underfermentative conditions is required. In an effort todevelop wine yeast strains that are relieved from ni-trogen catabolite repression and that are capable ofutilising proline more efficiently under winemakingconditions, a mutant containing a ure2 recessive al-lele was constructed. It was demonstrated that thismutation strongly deregulates the proline utilisationpathway, thereby improving the overall fermentationperformance of the ure2-carrying yeast. This may bethe first step towards the development of wine yeaststhat are able to efficiently assimilate the abundantsupply of proline in grape juice under fermentativeconditions.

Improved ethanol tolerance

The winemaker is confronted by the dilemma that,while ethyl alcohol is the major desired metabolicproduct of grape juice fermentation, it is also a po-tent chemical stress factor that is often the underly-ing cause of sluggish or stuck fermentations. Apartfrom the inhibitory effect of excessive sugar contenton yeast growth and vinification fermentation, theproduction of excessive amounts of ethanol, comingfrom harvest of over-ripe grapes, is known to inhibitthe uptake of solutes (e.g., sugars and amino acids)and to inhibit yeast growth rate, viability and fermen-tation capacity.

The physiological basis of ethanol toxicity is complexand not well understood, but it appears that ethanolmainly impacts upon membrane structural integrityand membrane permeability. The chief sites of etha-nol action include the yeast cell’s plasma membrane,hydrophobic proteins of mitochondrial membranes,nuclear membrane, vacuolar membrane, endoplasmicreticulum and cytosolic hydrophilic proteins. In-creased membrane fluidity and permeability due toethanol challenge seem to result in futile cycling ofprotons and dissipation of ATP energy. However, thedissipation of the proton gradient across the mem-brane and ATP is not only affected by increased per-meability to protons, but ethanol may also directly

affect the expression of the ATPase-encoding genes(PMA1 and PMA2) and membrane ATPase activity.This explains the interference of ethanol with energy-coupled solute transport in yeast cells.

Several intrinsic and environmental factors areknown to synergistically enhance the inhibitory ef-fects of ethanol. These factors include high fermen-tation temperatures, nutrient limitation (especiallyoxygen, nitrogen, lipids and magnesium ions) andmetabolic by-products such as other alcohols, alde-hydes, esters, organic acids (especially octanoic anddecanoic acids), certain fatty acids and carbonyl andphenolic compounds. By manipulating the physico-chemical environment during the cultivation andmanufacturing of active dried wine yeast starter cul-tures and during the actual vinification process, theyeast cells’ self-protective adaptations can be pro-moted. Prior exposure of yeast cells to ethanol (physi-ological pre-conditioning) elicits adaptive stressresponses that confer a degree of resistance to sub-sequent exposure to high levels of ethanol. Further-more, osmotic pressure, media composition, modesof substrate feeding and by-product formation playimportant roles in dictating how yeast cells tolerateethanol during vinification. Most of the so-called sur-vival factors (e.g., certain unsaturated long chain fattyacids and sterols) are formed only in the presence ofmolecular oxygen which in part explains the greatsuccess in the use of commercial starter cultures thatare cultivated under highly aerobic conditions and inlow glucose concentrations. These starter yeast cellscontain high levels of the survival factors that can bepassed onto the progeny cells during the six or sevengenerations of growth in a typical wine fermentation.

Wine yeast strains usually contain higher levels ofsurvival factors than nonwine Saccharomyces strains.The physiological response of wine yeast to ethanolchallenge is also greater than is the case withnonwine strains. These defensive adaptations of wineyeasts, conferring enhanced ethanol tolerance, rangefrom alterations in membrane fluidity to synthesis ofdetoxification enzymes. Responses include a decreasein membrane saturated fatty acids (e.g., palmitic acid);an increase in membrane unsaturated long chain fattyacids (e.g., oleic acid); phosphatidylinositol biosynthe-sis (thereby increasing the phospholipid: protein ra-tio in the membrane); elevated levels of cellulartrehalose that neutralise the membrane-damagingeffects of ethanol; stimulation of stress protein bio-synthesis; enhanced mitochondrial superoxidedismutase activity that countereffects ethanol-in-duced free radical synthesis; increased synthesis ofcytochrome P450, alcohol dehydrogenase activityand ethanol metabolism.

From this it is clear that the genetics of ethanol toler-ance are polyvalent and very complex. It is speculatedthat more than 250 genes are involved in the controlof ethanol tolerance in yeast. Nevertheless, some re-ports claim that continuous culture of yeasts in a feed-back system in which the ethanol was controlled by



the rate of carbon dioxide evolution, enabled theselection of viable mutants with improved ethanoltolerance and fermentation capabilities. Dramaticincreases in ethanol tolerance, however, seem toelude researchers. It therefore appears that for thetime being, ethanol tolerance in wine yeasts will beaddressed by ‘cell engineering’ rather than ‘geneticengineering’.

Increased tolerance to antimicrobial compounds

Besides the various yeast metabolites such as alcohols,acetic acid and medium chain length fatty acids (e.g.,decanoic acid) that can interfere with efficient grapemust fermentations, there are several other antimicro-bial compounds that can impede the fermentationperformance of wine yeasts. These compounds in-clude killer toxins, chemical preservatives (especiallysulphite) and agrochemicals containing heavy metals(e.g., copper). Since S. cerevisiae strains vary widelyin their ability to resist or tolerate these compounds,the differences may lend themselves as targets forstrain development.

Killer toxins are proteins produced by some yeaststhat are lethal to sensitive wine yeast strains. The kill-ers themselves, however, are immune to thesemycovirus-associated toxins. Whether the growth andzymocidal activity of some wild killer yeasts have thepotential to delay the onset of fermentation, causesluggish or stuck fermentations and produce wineswith increased levels of acetaldehyde, lactic acid,acetic acid and other undesirable sensory qualities isstill a matter of controversy. However, it appears thatunder certain conditions (e.g., inefficient inoculationwith highly sensitive starter cultures in low-nitrogenmusts) that favour the development of killer yeastcontamination of grape juice, potent zymocidal yeastsmay indeed contribute to incomplete fermentations.While zymocidal toxins produced by killer strains (K1,K2, K3, K28) of S. cerevisiae are lethal only to sensi-tive strains of the same species, those produced bynon-Saccharomyces killer species (K4 to K11) may betoxic to a wider range of wine yeast strains and otherwild yeasts. The killing of sensitive wine yeasts by thetwo S. cerevisiae killer toxins that function at winepH, K2 and K28, occur via two different mechanisms:the K2 toxin acts as an ionophore affecting membranepermeability and leakage of protons, potassium cati-ons, ATP and amino acids, whereas the K28 toxin in-hibits DNA synthesis.

An unfortunate consequence of ignorance regardingthe role of killer yeasts in wine fermentations is thatsome winemakers use co-cultures to inoculatefermentations; one strain being a killer and the othera sensitive strain. The advantage of using killer orneutral wine yeasts should therefore not be underes-timated. For this reason the aim of many strain de-velopment programmmes is to incorporate themycoviruses from killer yeasts into commercial winestrains. Mycoviruses are readily transmitted by cyto-plasmic fusion and have been used to transfer the

killer character into commercial yeasts. In most cases,however, the mixing of the genomes of commercialstrains and donor strains containing the killer char-acter would prove undesirable even though repeatedback-crossing could be used to minimise the un-wanted effects.

One way to circumvent this problem is cytoductionbetween a donor killer strain deficient in nuclearfusion and a haploid derived from a commercial winestrain. Another means is to cross a haploid derivedfrom a killer wine yeast with haploid cells orascospores from a sensitive wine yeast. An alternativeto the use of cytoduction and hybridisation to de-velop broad spectrum zymocidal resistance into wineyeasts would be to clone and introduce the toxin-immunity genes from non-Saccharomyces killeryeasts into wine yeasts.

Sulphur dioxide is widely used in wineries to sup-press the growth of unwanted microbes, and toler-ance to sulphite forms the basis of selectiveimplantation of active dried wine yeast starter cul-tures into grape must. Membrane transport of sul-phite in wine yeasts is by simple diffusion of liberatedsulphur dioxide rather than being carrier-mediated.SO2 dissociates within the cell to SO3

2- and HSO3- and

the resulting decline in intracellular pH forms thebasis of the inhibitory action. Although S. cerevisiae

tolerates much higher levels of sulphite than most un-wanted yeasts and bacteria, excessive SO2 dosagesmay cause sluggish or stuck fermentations.

Wine yeasts vary widely in their tolerance of sulphite,and the underlying mechanism of tolerance as wellas the genetic basis for resistance are still unclear.Once these have been better defined, it may be ad-vantageous to engineer wine yeast starter strains withelevated SO2 tolerance. This, however, should notreplace efforts to lower the the levels of chemicalpreservatives in wine.

Improper application of copper-containing fungalpesticides (copper oxychloride) to control downymildew (Plasmopara viticola) and, to a lesser extent,dead arm (Phomopsis viticola) and anthracnose(Gloeosporium ampelophagum) could lead to cop-per residues in musts that may cause lagging fermen-tation and affect wine quality detrimentally. Coppertoxicity towards wine yeast cells involves the disrup-tion of plasma membrane integrity and perhaps alsointracellular interaction between copper and nucleicacids and enzymes. Several copper uptake, efflux andchelation strategies have been developed by yeasts tocontrol copper ion homeostasis. One such protectivemechanism relates sequestration of copper by theCUP1-encoded copper-binding protein, copper-chelatin. Such metallothein proteins are generallysynthesised when S. cerevisiae cells are exposed topotentially lethal levels of toxic metals. The copperresistance level of a given yeast strain correlates di-rectly with the CUP1 copy number. One way to engi-neer wine yeasts resistant to copper would be toclone and integrate the CUP1 gene at multiple sites



into their genomes. This would enable the wine yeastto tolerate higher concentrations of copper residuesin musts. Copper-resistant wine yeasts should, how-ever, not be used to encourage disrespect for recom-mended fungicide withholding periods.

Reduced foam formation

Excessive foaming, caused by certain wine yeaststrains during the early stages of wine fermentation,can result in the loss of grape juice. Moreover, forma-tion of a froth-head can reduce fermenter capacity,as part of the fermentation vessel may have to bereserved to prevent the froth from spilling over. Insome cases foaming may also reduce the suspendedyeast cell density in the fermenting must.

Froth generation varies widely among S. cerevisiae

strains. Genetic analysis of the foaming characteris-tic suggests that this trait is under the control of atleast two dominant genes, FRO1 and FRO2. Appar-ently these genes, located on chromosome VII, codefor proteins that interact with the grape juice therebycausing foaming. Several researchers have success-fully used intragenomic hybridisation to cross out thegenes that are responsible for foaming. However, theFRO1 and FRO2 genes have yet to be cloned and theirencoded proteins characterised. Once this is done,the regulation of FRO1 and FRO2 can be unravelled.Gene disruption through targeted homologous re-combination would then also become possible whichwould eliminate the foaming characteristic of wineyeast strains without changing the remainder of theirgenetic backgrounds.

Improvement of processing efficiency

Improved protein and polysaccharide

clarification

The main objectives of fining and clarification dur-ing wine processing include the removal of excesslevels of certain components to achieve clarity andensure the physicochemical stability of the end prod-uct. The need for fining and clarification depends onthe composition of the must and the winemakingpractices that have been employed. Fining of wineentails the deliberate addition of an adsorptive com-pound, followed by the settling or precipitation ofpartially soluble components from the wine. Furtherclarification is usually achieved by sedimentation andracking, centrifugation and filtration. Wine filtrationinvolves a wide range of objectives, from the partialremoval of large suspended solids by various gradesof diatomaceous earth or filter sheets to the completeretention of microbes by perpendicular flow poly-meric membranes. While clarification of wine is gen-erally thought to produce insignificant compositionalchanges, fining is intended to bring about changesthat will prevent further precipitation. Fining cantherefore be used to modify the sensory attributes ofwine even though existing clarity may not be a prob-lem.

Fining reactions include the the removal of colloidssuch as partially soluble, haze-forming proteins, fil-ter-clogging polysaccharides as well as complexesbetween proteins and phenols, and between proteinsand polysaccharides. The removal of tannic or brownpolymeric phenols is usually achieved byproteinaceous fining agents (e.g., casein, isinglas, al-bumin and gelatin), whereas the depletion ofmonomeric and small polymeric phenols is reachedby treatment with polyamide materials (e.g.,polyvinylpolypyrrolidone or PVPP). Haze-formingproteins are removed by exchanging clays such asbentonites, while the removal of fine colloidal parti-cles and incipient precipitates is achieved by thesieving effect of other gelatinous materials.

The slow development of protein hazes in white wineis considered to be the next most common physicalinstability after the precipitation of potassiumbitartrate. Protein instability, occurring after bottlingand shelf storage, is induced by high ethanol and lowpH. Protein haze is not dependent upon total proteincontent but rather upon specific grape-derived pro-teins whose size or isoelectric properties make themparticularly susceptible to solubility limitations. Pro-tein instability is presumably associated withpathogenesis-related (PR) proteins produced in grapeberries when challenged by fungal attack. Althoughthe removal of these haze-forming proteins by ben-tonite treatment is effective, the non-specific natureof this diatomaceous clay can result in the loss ofimportant aroma and flavour compounds, therebyaltering the sensory characteristics of the wine. Fur-thermore, bentonite fining is an expensive and labo-rious practice that generates large volumes of lees fordisposal and causes a 5 to 20% loss of wine.

To omit the bentonite treatment, an application of anappropriate acid protease to hydrolyse the grape PR-proteins has been suggested. However, the search forfungal enzymes that could degrade these haze-form-ing proteins has so far remained unsuccessful.

We have investigated the feasibility of engineering aproteolytic wine yeast that could facilitate proteinhaze reduction. Proteolytic activities of S. cerevisiae

include the acid endoprotease, protease ysc A; theendo serine-sulphydryl protease, protease ysc B; theserine exopeptidase, carboxypeptidase ysc Y; and thefour metallo exopeptidases, namely carboxypeptidaseysc S, aminopeptidase ysc I, aminopeptidase ysc IIand aminopeptidase ysc Co. However, the vacuolarprotease A, encoded by the PEP4 gene, is the only onethat is active at the low pH of wine. Furthermore, ithas been reported that the prolonged storage of wineon the lees after the completion of the alcoholic fer-mentation renders a wine more protein stable. Thisphenomenon was attributed to the action ofproteinase A during autolysis.

This acid endoprotease is synthesised as a preproteinin S. cerevisiae. The prepeptide is cleaved early in thesecretory pathway and the propeptide is cleavedupon entrance of proteinase A into the vacuole. The



propeptide contains the vacuolar targeting informa-tion and serves as an inhibitor to keep protease Ainactive during transport through the secretory path-way. The PEP4 gene was cloned and expressed in awine yeast by using different combinations of severalpromotor, leader and termination sequences. North-ern blot analysis indicated the presence of these PEP4

transcripts in the various transformants. Upon replac-ing the PEP4-encoded prepro-region (vacuolar locali-sation signal) with the yeast mating pheromonea-factor (MFa1-encoded) prepro-region (secretion sig-nal), no extracellular protease activity was detected.However, western blot analysis revealed the presenceof extracellular protease A when the PEP4 gene wasoverexpressed under control of the constitutive yeastalcohol dehydrogenase I promoter (ADH1p) and ter-minator (ADH1T) signals. Casein agar test plates con-firmed that these transformants secreted biologicallyactive protease A. Overexpression of PEP4 inS. cerevisiae seems to have saturated the vacuolartargeting machinery and resulted in secretion of bio-logically active protease.

Later, it became known, however, that bentonite fin-ing is unlikely to be replaced by the addition of pro-teolytic enzymes to wine or by engineering aproteolytic wine yeast. This is not because theseproteases are inactive in must and wine, but becausethe haze-forming proteins in wine are inherently re-sistant to proteolysis. Their resistance is not due toprotection by other wine components in wine nor isit due to covalently bound sugars (glycosylation) orassociated phenolic compounds. It appears that pro-tein conformation bestows stability to these PR-pro-teins and that appropriate viticultural practices,rather than postharvest processing, may hold the keyto controlling the concentrations of protein in wine.

Like grape proteins, polysaccharides also influencethe clarification and stabilisation of must and wine.Polysaccharides, found in wines at levels between 300and 1000 mg/l, originate in the grape itself, the fungion the grape and the microorganisms present duringwinemaking. The main polysaccharides responsiblefor turbidity, viscosity and filter stoppages arepectins, glucans (a component of cellulose) and, to alesser extent, hemicellulose (mainly xylans). Grapepectic substances are heteroplosaccharides consist-ing of partially methylated α -1,4-D-galacturonanchains linked to L-rhamnopyranose units carryingneutral side chains. Glucans such as β-1,3-1,6-glucanproduced by the grey mould Botrytis cinerea inbotrytised grape juice, comprise β-D-glucopyranoseunits with a high degree of polymerisation. Xylans arecomplex polymers consisting of a β-D-1,4-linkedxylopyranoside backbone substituted with acetyl,arabinosyl and glucuronosyl side chains. Enzymaticbreakdown of pectic polymers occurs by the de-es-terifying action of pectinesterase, releasing the metylester groups of the pectin molecule, and by the hy-drolase or lyase action of the depolymerases (pectinlyase, pectate lyase and polygalacturonase), splitting

the α-1,4-glycosidic linkages in the polygalacturonatechain. Glucans are hydrolysed by endoglucanases(β-1,4-D-glucan glucanohydrolase), exoglucanases(β-1,4-D-glucan cellobiohydrolase), cellodextrinasesand cellobiases (β-1,4-D-glucoside glucohydrolase, amember of the β-glucosidase family). Enzymatic deg-radation of xylans is catalysed by the synergistic ac-tions of endo-β-1,4-D-xylanases, β-D-xylosidases andα-L-arabinofuranosidases.

The endogenous pectinase, glucanase, xylanase andarabinofuranosidase activities of grapes and yeastsare often neither efficient nor sufficient underwinemaking conditions to prevent polysaccharidehazes and filterstoppages. Industrial enzyme prepa-rations are widely used to supplement these polysac-charide-degrading activities. Most commercialpectinase and glucanase preparations are derivedfrom Aspergillus and Trichoderma, respectively.

Since the addition of these commercial enzymepreparations can be quite expensive, some research-ers are looking at the native pectinases andglucanases of S. cerevisiae. Certain strains ofS. cerevisiae were reported to produce pectineste-rase, polygalacturonase and pectin lyase, while allstrains of S. cerevisiae show some form of glucanaseactivity. All of these glucanase genes have beencloned and characterised. The EXG1 (BGL1) geneencodes a protein whose differential glycosylationaccounts for the two main extracellular exo-β-1,3-glucanases (EXGI and EXGII), while EXG2 encodesa minor exo-β-1,3-glucanase (EXGIII). BGL2 encodesa cell wall associated endo-β-1,3-glucanase, while SSG1

(SPR1) codes for a sporulation-specific exo-β-1,3-glucanase.

Since these endogenous pectinolytic and glucanolyticactivities of S. cerevisiae are not sufficient to avoidclarification and filtration problems, we have intro-duced a wide variety of heterologous pectinase,glucanase, xylanase and arabinofuranosidase genesinto S. cerevisiae. A pectinolytic wine yeast was de-veloped by co-expressing the Erwinia chrysanthemi

pectate lyase gene (pelE) and the Erwinia carotovora

polygalacturonase gene (peh1) in S. cerevisiae. Boththese bacterial genes were inserted in the ADH1P-

MFα1S-ΑDH1T yeast expression-secretion cassette.The pectinase gene cassette, consisting of ADH1P-

MFα1S-pelE-ADH1T (designated PEL5) and ADH1P-

MFα1S-peh1-ADH1T (designated PEH1) enabled wineyeast strains of S. cerevisiae to degrade polypectateefficiently. Likewise, our laboratory has also con-structed a glucanase gene cassette comprising theButyrivibrio fibrisolvens endo-β-1,4-glucanase gene(END1), the Bacillus subtilis endo-β-1,3-1,4-glucanase(BEG1), the Ruminococcus flavefaciens

cellodextrinase gene (CEL1), the Phanerochaeta

chrysosporium cellobiohydrolase gene (CBH1) andthe Saccharomycopsis fibuliger cellobiase gene(BGL1). Upon introduction of this glucanase genecassette, S. cerevisiae transformants were able to de-grade glucans efficiently. We have also successfully



expressed in S. cerevisiae the endo-β-xylanase genesfrom Aspergillus kawachii (XYN1), Aspergillus niger

(XYN4 and XYN5) and Trichoderma reesei (XYN2)as well as the Bacillus pumilus xylosidase (XLO1) andthe A. niger α-L-arabinofuranosidase gene (ABF2).The xlnA and xlnB genes from Aspergillus nidulans

were also reported to be expressed in S. cerevisiae.

It is hoped that these efforts will lay the foundationfor developing pectolytic, glucanolytic andxylanolytic wine yeasts that would contribute to theclarification of wine and replace or reduce the levelsof commercial enzyme preparations needed. Further-more, polysaccharide-degrading enzymes secreted bywine yeasts may also improve liquefaction of thegrapes, thereby increasing the juice yield. Since muchof the flavour compounds are trapped in the grapeskins, pectolysis, glucanolysis and xylanolysis mayrelease more of these aromatic compounds duringskin contact in red wine fermentations and make apositive contribution to the wine bouquet.

Controlled cell sedimentation and flocculation

S. cerevisiae adapts its growth pattern in response toa wide range of physical and chemical signals sensedby the cells. These changes include yeastfilamentation, agglomeration, flocculation and flota-tion, influenced by a variety of genetic, physiologi-cal and biochemical factors which are not alwaysclearly understood.

Filamentous growth and the formation ofpseudohyphae and hyphal-like structures often resultin dimorphism, known to be affected by nutrient limi-tation and the availability of oxygen. The phenom-enon of agglomeration involves an extensive,non-reversible cell aggregation process; flocculationrefers to an asexual cellular aggregation when yeastcells adhere, reversibly, to one another to form mi-croscopic flocs which sediment out of suspension.Yeast cell flotation, the converse of flocculation, de-fines the ability of non-aggregated yeast cells to trapCO2 bubbles in a fermenting liquid and form a filmor vellum at the top of fermentation vessels. All thesephenomena are highly relevant to the production ofseveral yeast-fermented products. Grittiness, causedby agglomerated baker’s yeast strains and the con-comitant appearance of granular material, is detri-mental since it results in inadequate mixing intobread dough leading to limited leavening ability. Yeastflocculation, on the other hand, is often exploited inthe production of lager beer and wine (especiallybottle-fermented sparkling wine). The flocs that set-tle to the bottom of the fermentor by the end of theprimary fermentation can easily be removed from thefermentation product, thereby allowing for rapid andefficient clarification and reduced handling of wine.Yeast flotation is important for the production of tra-ditional ale beer by top-fermenting strains, and florsherry by vellum-forming strains.

Flocculation in S. cerevisiae is thought to be mediatedby specific calcium-activated lectins, the FLO-gene en-

coded flocculins which are surface glycoproteinscapable of directly binding mannoproteins of adja-cent cells. Proteinaceous ‘hairy’ protrutions called‘fimbrae’ often emanate from the cell surface offlocculent S. cerevisiae cells. Cell surface charge andhydrophobicity have also been implicated in a pri-mary or complementary role with lectins to facilitatethe onset of flocculation. Environmental factors thatmay influence the level of flocculent S. cerevisiae

strains include temperature, pH, calcium and zincions, certain inhibitors, oxygen content, sugar and in-ositol depletion, growth phase and cell density.

Several dominant, semi-dominant and recessive genesare known to be involved in flocculation, and distinctflocculation phenotypes have been identified basedon their sensitivities to sugar inhibition andproteolylitic enzymes. These phenotypes, designatedFlo and NewFlo, also display different sensitivities toyeast growth conditions, most notably temperature,acidity of the culture medium and glucose availablity.The flocculation genes include FLO1, FLO2, flo3,

FLO4, FLO5, flo6, flo7, FLO9, FLO10 and FLO11/MUC1. The FLO1/MUC1 gene was also shown to beinvolved in pseudohyphal development and invasivegrowth, while FLO8 was reported to encode a tran-scriptional activator of FLO1 and FLO11/MUC1. Ap-parently, Flo8p inactivates the TUP1 and CYC8/SSN6encoded cascade which represses flocculation andpseudohyphal differentiation in certain strains. How-ever, when the expression of FLO1 and FLO11/MUC1

was investigated in 25 commercial wine yeast strains,it was found that they are not co-regulated. Further-more, it is unclear what the advantage would be tothe yeast cell by co-regulating the expression ofFLO11/MUC1 and three glucoamylase-encoding genes(STA1, STA2 and STA3) involved in starch metabolism.In fact, the unusually long (> 3 kb) promoter se-quences of FLO11/MUC1 and STA1-3 are almost iden-tical, and we have shown that several transcriptionalactivators (e.g., Flo8p, Msn1p and Mss11p) co-regu-late FLO11/MUC1-mediated filamentous growth andSTA1-STA3-facilitated starch assimilation.

The overall structure of the FLO11/MUC1-encodedcell wall associated protein is similar to those of theFlo1p, Flo5p and Flo10p. All these flocculins com-prise an amino-terminal domain containing a hydro-phobic signal sequence and a carboxyl-terminaldomain with homology to the glycosyl-phosphatidyl-inositol-anchor-containing proteins, separated by acentral domain of highly repeated sequences rich inserine and threonine residues. Of all the flocculationgenes, FLO1 is considered to be the best studied andperhaps most important, capable of conferringflocculation when transformed into non-flocculent S.

cerevisiae strains.

Regulated expression of the flocculation genes is im-portant in wine production, because yeast must per-form conflicting roles; during fermentation of grapemust, a high suspended yeast count ensures a rapidfermentation rate, while at completion of sugar con-



version, efficient settling is needed to minimise prob-lems with wine clarification. Moreover, flocculationhas also been linked to enhanced ester production.For these reasons we have linked the FLO1 gene tothe HSP30 gene promoter. It is known that the HSP30

promoter induces high gene expression during latestationary phase. We have shown that the expressionof FLO1, linked to the late-fermentation HSP30 pro-moter, can be induced by a heat-shock treatment, con-firming that controlled flocculation is indeedpossible during fermentation.

Controlled cell flotation and flor formation

Flor sherry is produced using certain strains of S.

cerevisiae (formerly known as S. beticus andS. capensis) capable of forming a yeast film on thesurface (flor) of a base wine exposed to air. Thesestrains are known for their high ethanol tolerance,superior film-forming ability and desirable oxidativemetabolism. Flor sherry is characterised by a highethanol (> 15%), low sugar and high aldehyde content.The typical nutty character of flor sherry can be as-cribed to the partial oxidation of ethanol to acetal-dehyde and to the specific contribution made by theflor strains of S. cerevisiae.

Although initial reports indicated that the vellum-forming trait segregated according to Mendelianrules in asci of sherry yeasts, it now seems unlikelythat the flor trait is controlled by a single dominantgene. Several genes encoding cell wall associated,hydrophobic proteins have been implicated in vel-lum-formation. Since few yeasts capable of growth onwine are suitable for flor sherry production, thegenotype of sherry yeasts is likely to be more com-plex than originally expected. However, once themost important genes responsible for film formationand the characteristic nutty bouquet have been iden-tified, the relevant genetic and metabolic mechanismsthat would allow for controlled vellum formation inflor sherry production may be brought to light.

Improvement of wine flavour and othersensory qualities

The single most important factor in winemaking isthe organoleptic quality of the final product. A wine’sbouquet is determined by the presence of desirableflavour compounds and metabolites in a well-bal-anced ratio, and the absence of undesirable ones.

Many variables contribute to the distinctive flavourof wine, brandy and other grape-derived alcoholicbeverages. Grape variety, viticultural practices, andterroir affect vine development and berry composi-tion, and exert major influences on the distinctive-ness of wine and brandy flavour. Oenologicalpractices, including the yeast and fermentation con-ditions, have a prominent effect on the primary fla-vours of V. vinifera wines. The volatile profile ofwines is dominated by those components that areformed during fermentation, since these compounds

are present in the highest concentrations. In brandy,the character is further changed as distillation altersthe absolute and relative amounts of volatiles.

The flavour of wine and brandy immediately after fer-mentation or distillation only approximates that ofthe finished product. After the sudden and dramaticchanges in composition during fermentation anddistillation, chemical constituents generally reactslowly during aging to move to their equilibria, result-ing in gradual changes in flavour. The harmoniouscomplexity of wine and brandy can subsequently befurther increased by volatile extraction during oakbarrel aging.

Despite the extensive information published on fla-vour chemistry, odour thresholds and aroma descrip-tions, the flavour of complex products such as wineand brandy cannot be predicted. With a few excep-tions (e.g., terpenes in the aromatic varieties andalkoxypyrazines in the vegetative or herbaceouscultivars), perceived flavour is the result of specificratios of many compounds rather than being attrib-utable to a single ‘impact’ compound. In wines andbrandies, the major products of yeast fermentation,esters and alcohols, contribute to a generic back-ground flavour, whereas subtle combinations of tracecomponents derived from the grapes usually elicit thecharacteristic aroma notes of these complex bever-ages.

Enhanced liberation of grape terpenoids

Varietal flavour of grapes is mainly determined by theaccumulation and profile of volatile secondarymetabolites in V. vinifera. However, a high percent-age of these metabolites occur as their respective,non-volatile O-glycosides. Several studies have shownthat increased enzymatic hydrolysis of aroma precur-sors present in grape juice can liberate the aglyconto intensify the varietal character of wines. For in-stance, terpenols such as geraniol and nerol can bereleased from terpenyl-glycosides by the grape-de-rived β-D-glycosidase activity present in muscat grapejuice. However, grape glycosidases are unable to hy-drolyse sugar conjugates of tertiary alcohols such aslinalool. Moreover, these grape enzyme activities areinhibited by glucose and exhibit poor stability at thelow pH and high ethanol levels of wine. Thanks tothese limiting characteristics of grape-derivedglycosidases and the fact that certain processing stepsduring the clarification of must and wine profoundlyreduce their activity, these endogenous enzymes ofgrapes have a minimal effect in enhancing varietalaroma during winemaking.

As an alternative to the inefficient grape glycosidases,aroma-liberating β-glucosidases from Aspergillus andother fungal species have been developed as compo-nents of commercial enzyme preparations to beadded to fermented juice (as soon as the glucose hasbeen consumed by the yeast) or to young wine. Theaddition of exogenous enzyme preparations to wine,however, is an expensive practice, and is viewed by



many purists as an ‘artificial’ or ‘unnatural’ interven-tion by the winemaker. This has led to renewed in-terest in the more active β-glucosidases produced bycertain strains of S. cerevisiae and other wine associ-ated yeasts such as Candida, Hanseniaspora andPichia (formerly Hansenula) species.

Unlike the grape β-glycosidases, yeast β-glucosidasesare not inhibited by glucose, and the liberation ofterpenols during fermentation can be ascribed totheir action on the terpenyl-glycoside precursors.Since these b-glucosidases are absent in mostS. cerevisiae starter culture strains, we have function-ally expressed the b-glucosidase gene (BGL1) of theyeast Saccharomycopsis fibuliger in S. cerevisiae.When the β-1,4-glucanase gene from Trichoderma

longibrachiatum was expressed in wine yeast thearoma intensity of wine increased, presumably dueto the hydrolysis of glycosylated flavour precursors.Likewise, we have overexpressed the S. cerevisiae

exo-β-1,3-glucanase gene (EXG1) and introduced theendo-β-1,4-glucanase gene (END1) from Butyrivibrio

fibrisolvens, the endo-β-1,3-1,4-glucanase (BEG1)from Bacillus subtilis and the α-arabinofuranosidase(ABF2) in S. cerevisiae. Further trials are underwayto determine the effect of these transgenic yeasts onthe varietal character of muscat wines.

Another intriguing discovery gives yeast the poten-tial to modify the ‘impact’ compound profile of low-flavoured wines. Certain mutants of the yeast sterolbiosynthetic pathway are able to producemonoterpenes (geraniol, citronelol and linalool) simi-lar to those of the muscat grape cultivars.

Enhanced production of desirable volatile esters

During the primary or alcoholic fermentation ofgrape sugars, wine yeast produces ethanol, carbondioxide and a number of by-products including es-ters, of which alcohol acetates and C4-C10 fatty acidethyl esters are found in the highest concentrationin wine and brandy. Although these compounds areubiquitous to all wines and brandies, the level of es-ters formed varies significantly. Apart from factorssuch as grape cultivar, rootstock and grape maturity,the ester concentration produced during fermenta-tion is dependent on the yeast strain, fermentationtemperature, insoluble material in the grape must,vinification methods, skin contact, must pH, theamount of sulphur dioxide, amino acids present inthe must and malolactic fermentation. Furthermore,the ester content of distilled beverages is greatly de-pendent on whether the yeast lees is present duringdistillation.

The characteristic fruity odours of wine are prima-rily due to a mixture of hexyl acetate, ethyl caproateand caprylate (apple-like aroma), iso-amyl acetate(banana-like aroma), and 2-phenylethyl acetate (fruity,flowery flavour with a honey note). The synthesis ofacetate esters such as iso-amyl acetate and ethyl ac-etate in S. cerevisiae is ascribed to at least threeacetyltransferase activities: alcohol acetyltransferase

(AAT), ethanol acetyltransferase (EAT) and iso-amylalcohol acetyltransferase (IAT). Theseacetyltransferases are sulfhydryl (SH) enzymes whichreact with acetyl coenzyme A (acetyl-CoA) and, de-pending on the degree of affinity, with various higheralcohols to produce esters. It has also been shownthat these enzymatic activities are strongly repressedunder aerobic conditions and by the addition of un-saturated fatty acids to a culture.

The ATF1-encoded alcohol acetyltransferase activity(AAT) is the best-studied acetyltransferase inS. cerevisiae. It has been reported that the 61-kDaATF1 gene product (Atf1p) is located within theyeast’s cellular vacuomes, and that it plays a majorrole in the production of iso-amyl acetate and to alesser extent ethyl acetate during beer fermentation.To investigate the role of AAT in wine and brandycomposition we have cloned the ATF1 gene from awidely used commercial wine yeast strain (VIN13)and placed it under control of the constitutive yeastphosphoglycerate kinase gene (PGK1) promoter andterminator. Integration of this modified copy of ATF1

into the genomes of three commercial wine yeaststrains (VIN7, VIN13 and WE228) resulted in the over-expression of AAT activity and increased levels ofethyl acetate, iso-amyl acetate and 2-phenylethyl ac-etate in wine and distillates. The concentration ofethyl caprate, ethyl caprylate and hexyl acetateshowed only minor changes, whereas the acetic acidconcentration decreased by more than half. Thesechanges in the wine and distillate composition had apronounced effect on the solvent/chemical (associ-ated with ethyl acetate and iso-amyl acetate), herba-ceous and heads-associated aroma of the finaldistillate and the solvent/chemical and fruity/flowerycharacter of Chenin blanc wines. This study estab-lished the concept that the over-expression ofacetyltransferase genes such as ATF1 could pro-foundly affect the flavour profiles of wines and dis-tillates deficient in aroma, thereby paving the way forthe production of products maintaining a fruitiercharacter for longer periods after bottling.

Optimised fusel oil production

Alcohols with carbon numbers greater than that ofethanol, such as isobutyl, isoamyl and active amyl al-cohol, are termed fusel oil. These higher alcohols areproduced by wine yeasts during alcoholic fermenta-tion from intermediates in the branched chain aminoacids pathway leading to production of isoleucine,leucine and valine by decarboxylation, transamina-tion and reduction. At high concentrations thesehigher alcohols have undesirable flavour and odourcharacteristics. Higher alcohols in wines, however,are usually present at concentration levels belowtheir threshold values and do not affect the taste ofwine unfavourably. In some cases, they may evencontribute to wine quality. However, since higheralcohols are concentrated by the distilling process,their reduction in wines that are to be distilled forbrandy production is of great importance.



Initial attempts to use Ile-, Leu- and Val- auxotrophicmutants succeeded in lowering the levels ofisobutanol, active amyl alcohol and isoamyl alcoholproduction in fermentations, but these mutants wereof no commercial use as their growth and fermenta-tion rates were compromised. A Leu- mutant derivedfrom the widely used Montrachet wine yeast wasreported to produce more than 50% less isoamyl al-cohol during fermentation than the prototrophicparent. It will be of great interest to see whether in-tegrative disruption of specific ILE, LEU and VAL

genes of wine yeasts will result in lower levels of fuseloil in wine for distillation.

Enhanced glycerol production

Due to its non-volatile nature, glycerol has no directimpact on the aromatic characteristics of wine. How-ever, this triol imparts certain other sensory qualities;it has a slightly sweet taste, and owing to its viscousnature, also contributes to the smoothness, consist-ency and overall body of wine.

The amount of glycerol in wines depends on manyfactors: grape variety, nitrogen composition, degreesof ripeness (sugar levels) and mould infection (dur-ing which glycerol is produced), sulfite levels and pHof grape must, fermentation temperature, aeration,and choice of starter culture strain and inoculationlevel. Typically, under controlled conditions glycerolconcentrations are higher in red wines than in whitewines, varying from 1 to 15 g/l. The threshold tastelevel of glycerol is observed at 5.2 g/l in wine,whereas a change in the viscosity is only perceivedat a level of 25 g/l. Wine yeast strains overproducingglycerol would therefore be of considerable value inimproving the organoleptic quality of wine.

In addition, the overproduction of glycerol at the ex-pense of ethanol could fulfil a growing need for ta-ble wine with lower levels of ethanol. About 4 to 10%of the carbon source is usually converted to glycerol,resulting in glycerol levels of 7 to 10% of that of etha-nol. Redirecting more of the grape sugars to glycerolwould provide a desirable alternative to the currentphysical ethanol-removing processes that non-specifi-cally alter the sensorial properties of the final prod-uct. Conversely, wine yeasts in which the glycerolpathway has been minimised would yield more alco-hol, which would be of great value for the produc-tion of brandy and other distilled products.

The physiological functions of glycerol synthesis arerelated to redox balancing, resistance tohyperosmotic and oxidative stress, recycling ofcytosolic inorganic phosphate and nitrogen metabo-lism. Futhermore, glycerol-3-phosphate, the precursorof glycerol, is an essential intermediate in the biosyn-thesis of membrane lipid. It is also noteworthy thatglycerol is not only produced by yeasts, but can alsoserve as carbon source in aerobically grown cultures.

During wine fermentations, the main role of glycerolsynthesis is to supply the yeast cell with an osmotic

stress responsive solute and to equilibrate the intra-cellular redox balance by converting the excessNADH generated during biomass formation to NAD+.Glycerol formation entails the reduction ofdihydroxyacetone phosphate to glycerol-3-phosphate,a reaction catalysed by glycerol-3-phosphate dehydro-genase and followed by the dephosphorylation ofglycerol-3-phosphate to glycerol by glycerol-3-phos-phatase. Two cytosolic glycerol-3-phosphatedehydrogenases, considered the key limiting enzymesfor glycerol formation in wine, are encoded by GPD1

and GPD2. The expression of GPD1 is usually in-creased by hyperosmotic stress, whereas GPD2 ex-pression is increased by anaerobic conditions. Thelevel of glycerol in S. cerevisiae and the expressionof both these genes are partially controlled by theHOG (High Osmotic Glycerol) signal transductionpathway when cells are exposed to hyperosmoticstress.

Conversely, the utilisation of glycerol is coupled torespiration via a glycerol kinase. This GUT1-encodedglycerol kinase converts glycerol to glycerol-3-phos-phate, which is then oxidised to dihydroxyacetonephosphate by the GUT2-encoded, flavin-dependentand membrane-bound mitochondrial glycerol-3-phos-phate dehydrogenase. GUT2 is strongly repressed inthe presence of glucose. FPS1 that encodes a chan-nel protein belonging to the MIP family, was shownto act as a glycerol transport facilitator controllingboth glycerol influx and efflux.

Slight increases in glycerol production in wine canbe achieved by using yeast strains selected or bredfor high glycerol production, and by optimising fer-mentation conditions. More recently it was reportedthat the overexpression of GPD1, together with con-stitutive expression of FPS1, successfully redirectedthe carbon flux towards glycerol and the extracellu-lar accumulation of glycerol. Depending on the ge-netic background in these engineered strains, 1.5 to4-fold increases in glycerol levels were obtained. Asa result of redox imbalances resulting from glyceroloverproduction, ethanol formation was decreasedand the metabolite pattern of these recombinantwine yeasts was considerably changed. A lowerbiomass concentration was attained in the GPD1-overexpressing strains, probably due to high acetal-dehyde production during the growth phase.Interestingly, the fermentation rate during the sta-tionary phase of wine fermentation was stimulatedin these strains, suggesting that the availability ofNAD may be a factor controlling the rate of glycolyticflux. Other side-effects of these glycerol-overproduc-ing yeasts included the accumulation of by-productssuch as pyruvate, acetate, acetoin, 2,3-butanediol andsuccinate.

A method was recently devised to overcome the mostnegative side-effect of glycerol overproduction,namely a marked increase in acetate formation. Sinceacetaldehyde dehydrogenases were shown to play aprominent role in acetate formation, the ALD6 and



ALD7 genes encoding a cytosolic Mg2+-activated,NADP-dependent and a mitochondrial K+-activated,NAD(P)-dependent acetaldehyde dehydrogenase, re-spectively, were disrupted. A wine yeast strain inwhich GPD1 was overexpressed in conjunction withthe deletion of ALD6 produced 2- to 3-fold more glyc-erol and a similar amount of acetate compared to theuntransformed strain. The redox balance was main-tained in these recombinant wine yeasts by increas-ing the formation of succinate and 2,3-butanediol toconcentrations remaining in the range of that foundin wine. These yeasts offer new prospects to improvethe quality of wine lacking in smoothness and body,and to production of low-alcohol wines.

Bio-adjustment of wine acidity

The acidity of grape juice and wine plays an impor-tant role in many aspects of winemaking and winequality, including the sensory quality of the wine andits physical, biochemical and microbial stability. Thejuice and wine acidity, in particular the pH, has aprofound influence on the survival and growth of allmicroorganisms; the effectiveness of anti-oxidants,antimicrobial compounds and enzyme additions, thesolubility of proteins and tartrate salts, the effective-ness of bentonite treatment, the polymerisation of thecolour pigments, and the oxidative and browningreactions. Wine contains a large number of organicand inorganic acids. The predominant organic acidsare tartaric and malic acid, accounting for 90% of thetitratable acidity of grapes. The main features of wineacidity include the acids themselves, the extent oftheir dissociation, the titratable acidity and pH. Fac-tors affecting the pH and titratable acidity of grapesinclude soil potassium and soil moisture; the natureof the rootstock and characteristics of the root sys-tem; viticultural practices such as canopy manage-ment and irrigation; climatic conditions andprevailing temperature during ripening; the cultivarand final berry volume at harvest. Of all these factorsthe climatic conditions and ambient temperaturehave a critical effect on grape maturation and result-ing acidity of the fruit. Under certain climatic condi-tions, the development of acidic compounds in thegrape during maturation and the subsequent physi-cal and microbial modification of these compoundsduring the process of winemaking can cause imbal-ances in the acidity of wines. Unless the acidity ofsuch wines with suboptimal pH values is adjusted, thewines will be considered as unbalanced or spoilt. Incooler climates (northern Europe, Canada, north-east-ern USA) chemical adjustment generally means a re-duction in titratable acidity by physicochemicalpractices such as blending, chemical neutralisationby double salting (addition of calcium carbonate) andprecipitation. These procedures often reduce winequality and require extensive labour and capital in-put.

In the warmer viticultural regions of southern Eu-rope, California, South Africa and Australia, blessedwith adequate sunshine during the growing season

and grape ripening period, malic acid is catabolisedat a faster rate. Here, adjustment of wine acidity gen-erally entails increasing the titratable acidity, or morecritically, lowering the pH by the addition of tartaricacid, and sometimes malic and citric acids, depend-ing on the laws of the country. Since the addition ofcalcium carbonate and acids are highly contentiouspractices that sometimes affect free trade in wine,several laboratories explored biological alternativesin order to minimise such chemical intervention.

At present, biodeacidification of wine is mediated bylactic acid bacteria, in particular Oenococcus oeni

(formerly Leuconostoc oenos). During malolactic fer-mentation, L-malic acid is decarboxylated to L(+)-lac-tic acid and carbon dioxide. Malolactic fermentationnot only reduces the total acidity of wine, it also en-hances microbiological stability and presumably im-proves the organoleptic quality of wine. However,owing to nutrient limitation, low temperature, acidicpH, and high alcohol and sulphur dioxide levels, themalolactic bacteria often grow poorly in wine,thereby complicating the management of this proc-ess. Stuck or sluggish malolactic fermentation oftenleads to spoilage of wine. Several alternatives wereexplored, including the possible use of malate-degrad-ing yeasts. During malo-ethanolic fermentations con-ducted by the fission yeast Schizosaccharomyces

pombe, malate is effectively converted to ethanol butoff-flavours were produced. Attempts to fuse wineyeasts with malate-assimilating yeast also failed. Theapplication of high density cell suspensions of sev-eral yeasts including S. cerevisiae, in an effort to in-crease the rate at which L-malate was degradedduring fermentation, was unsuccessful.

Their lack of success forced investigators back to thewine yeast itself. The ability of S. cerevisiae strains toassimilate L-malate acid varies widely. Unlike S.

pombe, S. cerevisiae lacks an active malate transportsystem and L-malate enters wine yeast by simple dif-fusion. Once inside the cell, S. cerevisiae’s own con-stitutive NAD-dependent malic enzyme convertsL-malate to pyruvate, which, under anaerobic condi-tions, will be converted to ethanol and carbon diox-ide. Aerobically, malic acid is decarboxylated intowater and carbon dioxide. Although the biochemicalmechanism for malate degradation in S. cerevisiae isthe same as in S. pombe, the substrate specificity ofthe S. cerevisiae malic enzyme is about 15-fold lowerthan that of the S. pombe malic enzyme. This lowsubstrate specificity together with the absence of anactive malate transport system is responsible forS. cerevisiae’s inefficient metabolism of malate.

Genetic engineering of wine yeast to conduct alco-holic fermentation and malate degradation simulta-neously has been explored by several groups. In orderto engineer a malolactic pathway in S. cerevisiae themalolactic genes (mleS) from Lactococcus lactis,Lactobacillus delbrueckii and the mleA gene fromO. oeni were cloned and expressed in S. cerevisiae.The mleS gene encodes a NAD-dependent malolactic



enzyme that converts L-malate to L-lactate and carbondioxide. However, due to the absence of an activemalate transport system in S. cerevisiae, these engi-neered strains could still not metabolise malate effi-ciently. Efficient malolactic fermentation wasachieved only when the L. lactis mleS gene was co-expressed with the S. pombe mae1 gene encodingmalate permease.

Similarly, an efficient malo-ethanolic S. cerevisiae wasconstructed by co-expressing mae1 permease geneand the mae2 malic enzyme gene from S. pombe inS. cerevisiae. A functional malolactic wine yeast couldreplace the unreliable bacterial malolactic fermenta-tion, whereas a malo-ethanolic strain of S. cerevisiae

would be more useful for the production of fruityfloral wines in the cooler wine-producing regions ofthe world.

Conversely, acidification of high-pH wines producedin the warmer regions with a wine yeast would bean inexpensive and convenient biological alternative.The formation of high levels of L(+) lactic acid byS. cerevisiae during alcoholic fermentation would beuseful for reducing the pH. In addition to its acidifi-cation properties, L(+) lactic acid, the main productof the metabolism of lactic acid bacteria, is stable. Dueto its pleasant acidic flavour and its properties as apreservative, lactic acid is widely used as a foodacidulant. Moreover, it is naturally present in mostfermented products, including wine, where it may bepresent in amounts of up to 6 g/l after malolacticfermentation.

Due to the inefficiency of the mitochondriallacticodehydrogenases under fermentation condi-tions, natural S. cerevisiae strains produce only tracesof lactic acid during alcoholic fermentation. In anattempt to redirect glucose carbon to lactic acid inS. cerevisiae, the lacticodehydrogenase-encodinggenes from Lactobacillus casei and bovine were ex-pressed in laboratory yeast strains. Encouraged by thefact that the L. casei lacticodehydrogenase gene, ex-pressed under control of the yeast alcohol dehydro-genase gene, converted 20% of the glucose intoL(+) lactic acid, this construct was also introducedinto eight wine yeast strains. Wines obtained withthese engineered lactic acid-alcoholic fermentationyeasts were shown to be effectively acidified, but thefermentation rate was slower.

Elimination of phenolic off-flavour

Excessive amounts of volatile phenols such as 4-vinylphenol, 4-vinylguaiacol, 4-ethylphenol and 4-ethylguaiacol often confer undesirable organolepticattributes on wine. These phenolic off-flavours canbe described as smoky, woody, clove-like, spicy andmedicinal. The POF1 gene in some strains ofS. cerevisiae encodes a substituted cinnamic acid car-boxylase that is able to decarboxylate grapehydroxycinnamic acids in a nonoxidative fashion tovinylphenols. Perhaps the disruption of POF1 could

provide a way to reduce the content of volatilephenols in, at least, white wines.

Reduced sulphite and sulphide production

Owing to their high volatility, reactivity and potencyat very low threshold levels, sulphur-containing com-pounds have a profound effect on the flavour of wine.These substances are formed in grapes during ripen-ing; dusting of vines with fungicides containing el-emental sulphur provide another source. During thewinemaking process sulphite is deliberately added tomost wines as an antioxidant and antimicrobial agent.Health concerns and an unfavourable public percep-tion of sulphite have led to demands for restrictionof its use and reassessment of all aspects of sulphiteaccumulation in wine. Consequently, the productionof sulphur-containing compounds by wine yeast itselfhas become a focal point of research.

Sulphur is essential for yeast growth and S. cerevisiae

can use sulphate, sulphite and elemental sulphur assole sources. The formation of sulphite and sulphideby wine yeasts greatly affects the quality of wine.Unlike sulphur dioxide (SO2), which when properlyused, has some beneficial effects, hydrogen sulphide(H2S) is one of the most undesirable of yeastmetabolites, since it causes, above threshold levels of50-80 g/l, an off-flavour reminiscent of rotten eggs.Sulphite is only formed from sulphate, while sulphideis formed from sulphate, sulphite, from elementalsulphur applied as a fungicide, and from cysteine. Theformation of sulphite and sulphide is affected bymany factors, including the composition of the fer-mentation medium.

Apart from strain effect, the nutrient composition ofgrape juice, the concentration of sulphate, must clari-fication, the initial pH and temperature all affect sul-phite formation by wine yeasts. Defects in sulphateuptake and reduction, which is normally regulated bymethionine via its metabolites methionyl-tRNA and S-adenosylmethionine, can result in excessive sulphiteproduction. During investigations into the regulationof sulphur metabolism in high and low sulphite-pro-ducing wine yeast strains, considerable differences inthe levels of activity of sulphate permease, ATP-sulphurylase and sulphite reductase were reported.Sulphate permease, mediating the uptake of sulphateby the yeasts, was shown not to be repressed by me-thionine in high sulphite-producing strains. ATP-sulphurylase and ADP-sulphurylase are not regulatedby sulphur intermediates in high or low sulphite-pro-ducing strains. Unlike the high sulphite-producingstrains, the low sulphite-producing strains showed anincreased biosynthesis of NADPH-dependent sulphitereductase, O-acetylserine sulphydrylase and O-acetylhomoserine sulphydrylase during theexponentional growth phase in the presence of sul-phate, sulphite and djencolic-acid. Methionine andcysteine are known to prevent an increase in the lev-els of sulphite reductase, O-acetylserinesulphydrylase and O-acetylhomoserine sulphydrylase.



Since sulphite production is very energy dependent,the cellular metabolism of high SO2-forming yeaststrains is reduced, explaining the decreased produc-tion of biomass and slow fermentation rate.

The formation of H2S by yeast during fermentationis largely in response to nutrient depletion, especiallyassimilable nitrogen and possibly certain vitaminssuch as pantothenate or pyridoxine. In the absenceof the H2S sequestering molecules O-acetylserine andO-acetylhomoserine, as caused by nitrogen starvation,free H2S accumulates and diffuses from the cell. De-pending on soil type and vintage conditions, somegrape varieties (e.g., Riesling, Chardonnay and Syrah),tend to have a low nitrogen content. This problemcan usually be suppressed by the addition of nitro-gen (typically in the form of diammonium phos-phate) during active fermentation. However, it hasbeen reported that impaired membrane transportfunction and intracellular deficiency of certain vita-mins can also cause H2S accumulatation.

The amount of H2S produced can also be affected bythe addition of a high level of SO2 to the must shortlybefore inoculating with yeast, and by the strain ofyeast involved. Certain yeasts more readily reducesulphate and SO2 to H2S when deprived of nitrogen,in a futile effort to synthesise and supply sulphur-containing amino acids to the growing yeast cell. Theaddition of ammonium salts prevents H2S accumula-tion in wine, not by stopping its formation but byenabling the yeast to synthesise amino acid precur-sor compounds which react with H2S to form sul-phur-containing acids. Due to higher fermentationtemperatures in hot climate red wine production,yeast cells use more nitrogen during rapidfermentations and tend to develop sulphidic smells.Fortunately, H2S is highly volatile and can usually beremoved by the stripping action of CO2 producedduring these rapid high-temperature fermentations.However, H2S formed towards the end of, or after,fermentation can react with other wine componentsto form mercaptans, thiols and disulphides, whichhave pungent garlic, onion and rubber aromas.

Yeast strains differ widely in their ability to producesulphite and sulphide. One way to take advantage ofthis fact is to select or develop a wine yeast strain thatwill either produce less H2S or that will retain mostof the H2S produced intracellularly. It was amply dem-onstrated in several laboratories that yeast strainswith low H2S production and improved winemakingproperties can be bred by hybridisation. In additionto exploiting the genetic heterogeneity in sulphiteand sulphide formation, the deliberate introductionof mutations in certain enzymes of the sulphur, sul-phur amino acids, pantothenate and pyridoxine path-ways might well enable a stepwise elimination ofthese characteristics in wine yeasts. The MET3 geneencoding ATP sulphurylase (the first enzyme in theconversion of intracellular sulphate to sulphite) hasbeen cloned and shown to be regulated at the tran-scriptional level. This may lead to the elucidation of

sulphite and sulphide formation by wine yeasts. H2Sproduction also appears to be closely related to theactivity of sulphite reductase and this could also pro-vide a target for down regulation of H2S formationin wine yeasts.

Improvement of wine wholesomeness andhealth properties of wine

Until the 18th century wine played a pivotal role inmedical practice, not least because it was a safer drinkthan most available water. Thanks to its alcohol andacid content, wine inhibits the growth of many spoil-age and pathogenic microorganisms. By the secondhalf of the 20th century, though, alcohol consumption,including wine drinking, had become a target ofsome health campaigners, who, with some success,demanded warning labels on wine bottles. By the 90’smedical science was reporting that moderate con-sumption of especially red wine, can reduce the inci-dence of heart disease. Today, it is generally acceptedthat moderate wine drinking can be socially benefi-cial, and that it can be effective in the managementof stress and reducing the risk of coronary heart dis-ease. The prudent wine drinker, however, continuesto keep a close eye on what and how he or she drinksto ensure that the benefits exceed the risks. Theworldwide decrease of alcohol consumption testifiesto this effect.

In developing wine yeast strains, it is therefore of theutmost importance to focus on these health aspectsand to develop yeasts that may reduce the risks andenhance the benefits. It is therefore no surprise that,since glycerol and ethanol are inversely related, partof the objective in developing glycerol-overproducingS. cerevisiae strains is to reduce alcohol content in theend product. Likewise, research in several laborato-ries around the world is directed towards the elimi-nation of suspected carcinogenic compounds in winesuch as ethyl carbamate, and asthmatic chemical pre-servatives such as sulphites.

It might even be possible to develop wine yeasts thatcould increase the levels of phenolic and anti-oxidative substances (e.g., resveratrol) associatedwith the so-called ‘French paradox’, in which, despitethe high dietary fat intake of the cheese-loving popu-lation of southern France, the death rate from coro-nary heart disease is significantly lower than incomparable industrialised countries. Several possibleexplanations have been offered, but the best case forresolving this paradox has been made for red winephenolics that chemically modify blood lipoproteinsin cholesterol-furred arteries.

Resveratrol production

Phytoalexins, including stilbenes such as resveratrol,have been shown to reduce the risk of coronary heartdisease. By acting as an antioxidant and as anantimutagen, resveratrol shows cancerchemopreventive activity, as well as the ability to in-



duce specific enzymes which metabolise carcino-genic substances.

Stilbenes are secondary plant products producedthrough the phenylalanine/ polymalonate pathway.Resveratrol is a stress metabolite produced byV. vinifera during fungal infection, wounding or ul-tra-violet radiation. Resveratrol is synthesised particu-larly in the skin cells of grape berries and only tracesare found in the fruit flesh. Red wine therefore con-tains a much higher resveratrol concentration thanwhite wine due to skin contact during the first phaseof fermentation.

One way to increase the levels of resveratrol in bothred and white wine is to develop wine yeasts able toproduce resveratrol during fermentation. To achievethis goal, the phenylpropanoid pathway inS. cerevisiae will have to be modifed to produce p-coumaroyl-coenzyme A, one of the substances forresveratrol synthesis. This can be done by introduc-ing the phenylalanine ammonia-lyase gene (PAL),cinnamate 4-hydroxylase gene (D4H) and thecoenzyme A ligase gene (4CL216) in S. cerevisiae. Theintroduction of the grape silbene synthase gene(Vst1) may then catalyse the addition of three acetateunits from malonyl-coenzyme A, already found inyeast, to p-coumaryl-coenzyme A, resulting in the for-mation of resveratrol. At this stage, however, there islittle indication of the chances for success in devel-oping resveratrol-producing wine yeast strains.

Reduced formation of ethyl carbamate

Ethyl carbamate (also known as urethane) is a sus-pected carcinogen that occurs in most fermentedfoods and beverages. Given the potential health haz-ard, there is a growing demand from consumers andliquor control authorities to reduce the allowable lim-its of ethyl carbamate in wines and related products.Although young wines do not contain measurablelevels (< 10 µg/l) of ethyl carbamate, the requiredprecursors are present which can generate a consid-erable amount of this mutagenic compound whenwine is aged or stored at elevated temperatures. High-alcohol beverages such as sherries, dessert wines anddistilled products also tend to contain much higherlevels of ethyl carbamate than table wine. It is be-lieved that ethyl carbamate forms in aging wines, for-tified wines and brandies by reaction between ureaand ethanol. For this reason excessive application ofurea-containing fertilisers to vines and spraying ofurea shortly before harvest to remove leaves are notrecommended. Furthermore, the use of urea-contain-ing nutrient supplements for yeast during winefermentations to avoid stuck or sluggishfermentations is also prohibited. Apart from thesefactors that could lead to high urea levels and con-comitant transgression of ethyl carbamate limits,S. cerevisiae strains also vary widely with regard totheir urea-forming ability.

In S. cerevisiae urea is formed during the breakdownof arginine, one of the main amino acids in grape

juice, by the CAR1-encoded arginase. During this re-action, arginine is converted to ornithine, ammoniaand carbon dioxide, while urea is formed as an inter-mediate product. Certain yeast strains secrete ureainto wine and, depending on fermentation condi-tions, may be unable to further metabolise the exter-nal urea. Although all S. cerevisiae strains secreteurea, the extent to which they re-absorb the ureadiffers. S. cerevisiae secretes more urea at higher fer-mentation temperatures, whereas high ammonia con-centrations suppress the re-absorption of urea by theyeast. It is therefore important to inoculate grapemust with a low-urea producing wine yeast strainwhen the juice has a high arginine content.

Strain selection is only one way of reducing the ac-cumulation of urea in wine. As an alternative meansof curbing ethyl carbamate formation in the endproduct, successive disruption of the CAR1 arginasegene in an industrial saké yeast proved to be success-ful in eliminating urea accumulation in rice wine.This arginase deletion mutation resulted in a yeaststrain that could not metabolise arginine but it alsoimpeded growth, thereby limiting the commercial useof such a strain.

Another possibility is adding commercial prepara-tions of acidic urease, enabling the hydrolysis of ureain wine. This practice has recently been approved bythe OIV and is used in some wine-producing coun-tries to lower ethyl carbamate levels in their winesand related products. A less expensive route to lowerlevels of ethyl carbamate would be to develop a wineyeast that produces an extracellular, acidic urease. Inone such attempt a novel urease gene was con-structed by fusing the α, β and γ subunits of the Lacto-

bacillus fermentum urease operon. In addition, jackbean urease linker sequences were inserted betweenthe α and β, as well as the β and γ subunits. Both geneconstructs were successfully expressed under thecontrol of the S. cerevisiae PGK1 promoter and ter-minator signals in the yeasts S. cerevisiae andS. pombe. Although the level of transcription inS. cerevisiae was much higher than in S. pombe, thesecretion of urease peptides was extremely low. Un-like the S. pombe urease, the S. cerevisiae-derivedurease was unable to convert urea into ammonia andcarbon dioxide. The absence of recombinant ureaseactivity in transformed S. cerevisiae cells is probablydue to the lack of the essential auxiliary proteinspresent only in urease-producing species such asS. pombe. Without these proteins, S. cerevisiae isunable to assemble the various subunits into an ac-tive urease. It seems, therefore, that accessory genesof L. fermentum will also have to be cloned and ex-pressed in addition to the structural urease genes toenable S. cerevisiae to express an active urease.

Improved biological control of wine spoilagemicroorganisms

Uncontrolled microbial growth before, during or af-ter wine fermentation can alter the chemical compo-



sition of the end product, detracting from its sensoryproperties of appearance, aroma and flavour. In se-vere cases of microbial spoilage the wine becomesunpalatable. Owing to the high initial sugar content,low pH, anaerobic fermentation conditions and highalcohol levels at the end of fermentation, only a fewspoilage yeasts and bacteria can survive the strongselective pressures present in fermenting grape mustand in wine.

Moulds usually spoil wine by infecting the grapes orspoiling cork slabs. These include species of Penicil-

lium, Anahanocladium, Mucor, Monilia, Trichode-

rma, Oidiodendron, Botrytis, Rhizopus,

Cladosporium and Paecilomyces. Penicillium

glabrum is considered the major mould on corkslabs, while some strains of Botrytis cinerea are as-sociated with grey rot (‘pourriture grise’) of grapes.They confer mouldiness and cork taints to wine. Thisearthy, musty, sometimes mushroom-like aroma isassociated with the presence of 2,4,6-trichloroanisolein bottled wine.

Spoilage yeasts include species from Brettanomyces,the osmotolerant yeast Zygosaccharomyces and thefilm-forming yeast species Pichia and Candida.Brettanomyces intermedius is known to producehaze, turbidity, volatile acidity and a mousy taint;Zygosaccharomyces baillii causes turbidity after re-fermentation during storage of wine or after bottling,resulting in sediment formation and reduction inacidity. Wines spoiled by Pichia membranaefaciens

and Candida krusei taste oxidised and less acid.

Without underestimating the degree of wine spoilagethat can be caused by moulds and yeasts, it is widelyaccepted that bacteria are the primary culprits, espe-cially acetic acid and lactic acid bacteria. A vinegarytaint in wine is often associated with the activity ofacetic acid bacteria such as Acetobacter aceti,Acetobacter pasteurianus and Gluconobacter

oxydans. Although some lactic acid bacteria play akey role in the malolactic fermentation of wine, oth-ers may cause serious faults. Excessive volatile acid-ity, mannitol taint, ropiness, mousiness, acroleinformation and bitterness, tartaric acid degradation,diacetyl overproduction and rancidness, as well as thevery unpleasant geranium off-flavour are often theconsequence of uncontrolled growth of some speciesof Lactabacillus (e.g., L. brevis, L. hilgardii,L. plantarum), Leuconostoc (e.g., L. mesenteroides),Streptococcus (S. mucilaginosus) and Pediococcus

(e.g., P. cerevisiae).

Healthy grapes, cellar hygiene and sound oenologicalpractices (e.g., appropriate pH, fermentation tem-perature, filtration, application of fining agents, etc.)will remain the corner stones of the winemaker’sstrategy against uncontrolled proliferation of spoil-age microbes. But the use of efficient S. cerevisiae

and O. oeni starter cultures at appropriate inocula-tion levels will usually outcompete undesirable con-taminants, thereby limiting the risk of poor qualitywine and concomitant financial loss. For additional

safety, chemical preservatives such as sulphur diox-ide and dimethyl dicarbonate are commonly addedto control the growth of unwanted microbial con-taminants. However, the excessive use of these chemi-cal preservatives is deleterious to the quality of wineand related fortified and distilled products, and isconfronted by mounting consumer resistance.

Consumer concerns have spurred a worldwide searchfor safe, food-grade preservatives of biological origin.A major focus of these investigations into novelbiopreservatives includes the identification and ap-plication of effective antimicrobial enzymes (e.g., lys-ozyme) and peptides (e.g., zymocins andbacteriocins). These efforts have been encouraged bythe successful application of lysozyme and nisin toprotect beer, wine and fruit brandies from spoilagelactic acid bacteria. But wine is a market-sensitivecommodity, and large scale industrial application ofpurified antibacterial enzymes and bacteriocins isexpensive, resulting in an increase in retail costs, asobserved in the case of beer production. This may beovercome by developing wine yeast starter culturestrains producing appropriate levels of efficient an-timicrobial enzymes and peptides.

Wine yeasts producing antimicrobial enzymes

Antimicrobial enzymes are ubiquitous in nature, play-ing a pivitol role in the defense mechanisms of hostorganisms against infection by fungi and bacteria.Hydrolytic antimicrobial enzymes such as chitinases,β-glucanases and lysozyme function by degrading keystructural components of the cell walls of moulds andbacteria. Chitinases and β-glucanases synergisticallyattack the main components of fungal cell walls,chitin and β-1,3-glucan. Lysozyme, an N-acetylhexosaminidase, lyses the cell walls of certainGram-positive species of bacteria lacking an outermembrane by hydrolysing the β-1,4-glucosidic link-ages of peptidoglycan in the cell wall. Its alkalinenature contributes to the antibacterial activity of lys-ozyme. Furthermore, Gram-negative bacteria contain-ing an outer membrane are more sensitive tolysozyme in combination with a chelating agent suchas EDTA or when lysozyme is modified byperillaldehyde. Conjugation to galactomanan alsoincreases the potency of lysozyme towards Gram-negative bacteria by enabling diffusion of the enzymeacross the outer membrane of the target cell.

The OIV has recently approved the use of commer-cial lysozyme preparations to control malolactic fer-mentation and to stabilise wine afterwards. However,the general use of lysozyme in winemaking is limitedbecause of its low cost-efficiency. This has encour-aged efforts to develop lysozyme-producingS. cerevisiae strains. The lysozyme-encoding genefrom chicken egg white was successfully expressedin E. coli and S. cerevisiae. In E. coli, the bactericidalaction of the recombinant lysozyme against Gram-negative bacteria was enhanced when a pentapeptidewas inserted into C-terminus. Research is underway



to express a modified lysozyme gene in wine yeastthat would avoid hyperglycosylation and broaden itsactivity to effectively eliminate spoilage by lactic andacetic acid bacteria.

Wine yeasts producing antimicrobial peptides

The killer phenomenon is wide spread among grape,must and wine related yeast genera, including Can-

dida, Cryptococcus, Debaryomyces, Hanseniaspora,

Kloeckera, Kluyveromyces Pichia, and Rhodutorula.Most zymocidal strains of S. cerevisiae associatedwith wine fermentation produce the K2 or K28zymocins which are functional at the low pH of grapemust and wine. Zymocidal yeast contaminants areimplicated as one of the causes of sluggish or stuckfermentations, but they are also promoted for inhib-iting the proliferation of unwanted yeast contami-nants. However, their efficacy under winemakingconditions has yet to be demonstrated. Furthermore,zymocins produced by S. cerevisiae are lethal only tosensitive strains of S. cerevisiae, whereas those pro-duced by non-Saccharomyces species may be toxic toS. cerevisiae as well as non-Saccharomyces species.

Several attempts have been made over the years toexpand the zymocidal activity of S. cerevisiae so thatit could also eliminate other yeast contaminants. Insome instances different killer types of S. cerevisiae

were hybridised by mating, cytoduction and sphero-plast fusion, while in another case a DNA copy of theK1 dsRNA was introduced in a K2 strain ofS. cerevisiae. However, even a K1/K2 double killerS. cerevisiae is very limited as to the variety of yeastcontaminants that can be eliminated. Rather attentionis now focused on the identification of genes encod-ing more effective zymocins in other yeasts such asPichia and Hanseniaspora and their possible intro-duction into S. cerevisiae.

We have investigated the feasibility of controllingspoilage bacteria during wine fermentations by en-gineering bactericidal strains of S. cerevisiae. To testthis novel concept, we have succesfully expressed twobacteriocin genes in yeast, the one encoding apediocin and the other a leucocin. The pediocin geneoriginates from Pediococcus acidilactici PAC1·0 andthe leucocin gene from Leuconostoc carnosum B-Ta11a.

The pediocin operon of P. acidilactici consists of fourclustered genes, namely pedA (encoding a 62 aminoacid precursor of the PA-1 pediocin), pedB (encod-ing an immunity factor), pedC (encoding a PA-1 trans-port protein) and pedD (encoding a protein involvedin the transport and processing of PA-1). The leucocinoperon of L. carnosum comprises two genes: lcaB

(encoding a 61 amino acid precursor of the B-Ta11aleucocin) and lcaB1 (encoding a 113 amino acid im-munity factor. Both the P. acidilactici pedA andL. carnosum lcaB genes were inserted into a yeastexpression-secretion cassette and introduced asmulticopy episomal plasmids into laboratory strainsof S. cerevisiae. Northern blot analysis confirmed that

the pedA and lca1 structural genes in these con-structs (ADH1P-MFa1S-pedA-ADH1T, designated PED1

and ADH1P-MFa1S-lcaB-ADH1T, designated LCA1),were efficiently expressed under the control of theyeast alcohol dehydrogenase I gene promoter(ADH1P) and terminator (ADH1T). Secretion of thePED1-encoded pediocin and LCA1-encoded leucocinwas directed by the yeast mating pheromone a-fac-tor’s secretion signal (MFa1S). The presence of bio-logically active antimicrobial peptides produced bythe S. cerevisiae transformants was indicated by agardiffusion assays against sensitive indicator bacteria(e.g., Listeria monocytogenes B73). The heterologouspeptides were present at relatively low levels in theyeast supernatant but pediocin and leucocin activi-ties were readily detected when intact yeast colonieswere used in sensitive strain overlays. These prelimi-nary results indicate that it is indeed possible to de-velop bactericidal wine yeast strains that could beuseful in the production of wine with reduced levelsof potentially harmful chemical preservatives.

CONCLUDING REMARKS

Yeast genetics provides both the means to constructimproved strains for use in winemaking and a pow-erful analytical approach to a better understandingof yeast behaviour in industrial fermentations. Thereis still much to learn of the genetic make-up of indus-trial yeasts and much to do to demonstrate the advan-tages of strain development in practice. Nevertheless,over the last few years considerable progress hasbeen made in developing new wine yeast strains.However, for a significant expansion of this effortthere is a need for further research into yeast physi-ology. In particular, it is vital that research continuesin the areas of flavour formation and the factors af-fecting yeast growth under winemaking conditions.Legislation is anothor important factor affecting thedevelopment of new wine yeasts. As outlined above,recombinant DNA techniques have been extensivelyused for the construction of new wine yeast strains.However, the need for approval is a major barrier tocommercialise these genetically modified yeaststrains. These strict requirements of genetically modi-fied food and beverage regulations are a response toperceived consumer concerns about the risk of ge-netically modified organisms (GMOs). From severalrecent surveys in many countries it is apparent thatthe average consumer is poorly informed about thebenefits and risks of biotechnology. It is therefore ofkey importance that this problem is addressed andthat the public is familiarised with the relevant sci-entific facts and factors involved in the used of GMOsfor food and drink production. It is essential to con-vince public opinion that the risks are small, beforewine made from genetically modified grapevinecultivars or fermented with genetically modifiedyeasts will become generally acceptable. All winebiotechnologists must work hard towards ensuringthat any possible risks are seen in the correct perspec-



tive. Otherwise the application of innovative biotech-nology in the wine industry will be restricted andmany of the benefits will be lost.

REFERENCES

This article is condensed from a comprehensive re-view by Pretorius (2000), and all references are notlisted here. The reader is referred to the followingreview articles and references therein:

Bauer F.F., and Pretorius, I.S., 2000. Yeast stressresponse and fermentation efficiency: How tosurvive the making of wine – A review.S. Afr. J. Enol. Vitic. 21: 5 - 26.

Du Toit M., and Pretorius I.S., 2000. Microbial spoilageand preservation of wine: Using weapons fromnature’s own arsenal – A review. S. Afr. J. Enol.Vitic. 21: 74 – 96.

Lambrechts M.G., and Pretorius I.S., 2000. Yeast and itsimportance to wine aroma – A review. S. Afr. J.Enol. Vitic. 21: 97 – 129.

Pretorius I.S., 1997. Utilization of polysaccharides bySaccharomyces cervisiae. In Zimmermann, F. K.and Entian, K.-D. (eds), Yeast Sugar Metabolism.Technomic Publishing, Pennsylvania, USA,pp. 459 - 501.

acs66_04

Pretorius I. S., 1999. Engineering designer genes forwine yeasts. Aust. N.Z. Wine Indust. J. 14: 42 - 47.

Pretorrius I.S., 2000. Tailoring wine yeast for the newmillennium: novel approaches to the ancient art ofwinemaking. Yeast 16: 675 - 729.

Pretorius I.S., and Van der Westhuizen T.J., 1991. Theimpact of yeast genetics and recombinant DNAtechnology on the wine industry – A review. S. Afr.J. Enol. Vitic. 12: 3 - 31.

Pretorius I.S., Van der Westhuizen T.J., and AugustynO.P.H., 1999. Yeast biodiversity in vineyards andwineries and its importance to the South Africanwine industry. S. Afr. J. Enol. Vitic. – A review. 20:61 - 74.

Rainieri S., and Pretorius I.S., 2000. Selection andinprovement of wine yeast. Ann. Microbiol. 50:15 - 31.

Van Rensburg P., and Pretorius I.S., 2000. Enzymes inwinemaking: Harnessing natural catalysts forefficient biotransformations – A review. S. Afr. J.Enol. Vitic. 21: 52 – 73.

Vivier M.A., and Pretorius I.S., 2000. Genetic improve-ment of grapevine: Tailoring grape varieties for thethird millennium – A review. S. Afr. J. Enol. Vitic.21: 5 – 26.

gene technology in winemaking: new approaches to an

Documents