Title page, table of contents and abstract only. The complete thesis in print form is available from the University of Adelaide Library. / Mannoproteins released during the fermentation of Saccharomyces cerevisiae in a chemically defined synthetic grape juice medium were isolated by ethanol precipitation. The proteins from the medium were fractionated by ion exchange chromatography. Fractions containing mannoproteins were identified by their UV absorption spectra and by the presence of polymeric mannose. A mannoprotein designated HPF2 was purified from one of the fractions by gel permeation chromatography, and had a high capacity to reduce heat-induced protein haze formation in white wine. After electrophoretic separation and transfer to nitrocellulose, HPF2 stained positively with an antibody (anti-HPFl) which had been raised against a previously isolated mannoprotein with known haze protective activity designated HPFL. Analysis of the carbohydrate portion of HPF2 using methylation analysis confirmed the presence of (1 -> 2), (1 -> 3) and (1 -> 2,1 -> 6) mannosyl residues and showed that it contained N-linked and possibly O-linked carbohydrate. Digestion of the mannoprotein with PNGase F resulted in a reduction in molecular weight, as measured by SDS-PAOE and confirmed the presence of N-linked carbohydrate. N-linked deglycosylation decreased the ability of HPF2 to protect wine from heatinduced protein haze. Protein sequence analysis of the peptides derived from the HPF2 mannoprotein obtained via digestion with endoproteinase Lys C led to the identification of the HPF2 structural gene in Saccharomyces cerevisiae followed by searching the Yeast Proteome Database for related sequences. Analysis of the deduced amino acid sequence of HPF2 from its structural gene indicated that HPF2 possessed a secretion signal at the NH₂-terminus, a putative OPI anchor site at the COOH-terminus and also contained serine and threonine rich regions at both the NH₂-terminus and COOHterminus. These regions may contain O-linked carbohydrate. There were also 15 potential glycosylation sites, five of which were classified from the peptide mapping data as likely glycosylation sites. These characteristics, combined with the results from the carbohydrate analysis, indicate that HPF2 was a mannoprotein derived from the yeast cell walls. / http://proxy.library.adelaide.edu.au/login?url= http://library.adelaide.edu.au/cgi-bin/Pwebrecon.cgi?BBID=678380 / Thesis (Ph.D.) -- University of Adelaide, Dept. of Plant Science, 2000

Title page, table of contents and abstract only. The complete thesis in print form is available from the University of Adelaide Library. / Mannoproteins released during the fermentation of Saccharomyces cerevisiae in a chemically defined synthetic grape juice medium were isolated by ethanol precipitation. The proteins from the medium were fractionated by ion exchange chromatography. Fractions containing mannoproteins were identified by their UV absorption spectra and by the presence of polymeric mannose. A mannoprotein designated HPF2 was purified from one of the fractions by gel permeation chromatography, and had a high capacity to reduce heat-induced protein haze formation in white wine. After electrophoretic separation and transfer to nitrocellulose, HPF2 stained positively with an antibody (anti-HPFl) which had been raised against a previously isolated mannoprotein with known haze protective activity designated HPFL. Analysis of the carbohydrate portion of HPF2 using methylation analysis confirmed the presence of (1 -> 2), (1 -> 3) and (1 -> 2,1 -> 6) mannosyl residues and showed that it contained N-linked and possibly O-linked carbohydrate. Digestion of the mannoprotein with PNGase F resulted in a reduction in molecular weight, as measured by SDS-PAOE and confirmed the presence of N-linked carbohydrate. N-linked deglycosylation decreased the ability of HPF2 to protect wine from heatinduced protein haze. Protein sequence analysis of the peptides derived from the HPF2 mannoprotein obtained via digestion with endoproteinase Lys C led to the identification of the HPF2 structural gene in Saccharomyces cerevisiae followed by searching the Yeast Proteome Database for related sequences. Analysis of the deduced amino acid sequence of HPF2 from its structural gene indicated that HPF2 possessed a secretion signal at the NH₂-terminus, a putative OPI anchor site at the COOH-terminus and also contained serine and threonine rich regions at both the NH₂-terminus and COOHterminus. These regions may contain O-linked carbohydrate. There were also 15 potential glycosylation sites, five of which were classified from the peptide mapping data as likely glycosylation sites. These characteristics, combined with the results from the carbohydrate analysis, indicate that HPF2 was a mannoprotein derived from the yeast cell walls. / http://proxy.library.adelaide.edu.au/login?url= http://library.adelaide.edu.au/cgi-bin/Pwebrecon.cgi?BBID=678380 / Thesis (Ph.D.) -- University of Adelaide, Dept. of Plant Science, 2000

Title page, table of contents and abstract only. The complete thesis in print form is available from the University of Adelaide Library. / Mannoproteins released during the fermentation of Saccharomyces cerevisiae in a chemically defined synthetic grape juice medium were isolated by ethanol precipitation. The proteins from the medium were fractionated by ion exchange chromatography. Fractions containing mannoproteins were identified by their UV absorption spectra and by the presence of polymeric mannose. A mannoprotein designated HPF2 was purified from one of the fractions by gel permeation chromatography, and had a high capacity to reduce heat-induced protein haze formation in white wine. After electrophoretic separation and transfer to nitrocellulose, HPF2 stained positively with an antibody (anti-HPFl) which had been raised against a previously isolated mannoprotein with known haze protective activity designated HPFL. Analysis of the carbohydrate portion of HPF2 using methylation analysis confirmed the presence of (1 -> 2), (1 -> 3) and (1 -> 2,1 -> 6) mannosyl residues and showed that it contained N-linked and possibly O-linked carbohydrate. Digestion of the mannoprotein with PNGase F resulted in a reduction in molecular weight, as measured by SDS-PAOE and confirmed the presence of N-linked carbohydrate. N-linked deglycosylation decreased the ability of HPF2 to protect wine from heatinduced protein haze. Protein sequence analysis of the peptides derived from the HPF2 mannoprotein obtained via digestion with endoproteinase Lys C led to the identification of the HPF2 structural gene in Saccharomyces cerevisiae followed by searching the Yeast Proteome Database for related sequences. Analysis of the deduced amino acid sequence of HPF2 from its structural gene indicated that HPF2 possessed a secretion signal at the NH₂-terminus, a putative OPI anchor site at the COOH-terminus and also contained serine and threonine rich regions at both the NH₂-terminus and COOHterminus. These regions may contain O-linked carbohydrate. There were also 15 potential glycosylation sites, five of which were classified from the peptide mapping data as likely glycosylation sites. These characteristics, combined with the results from the carbohydrate analysis, indicate that HPF2 was a mannoprotein derived from the yeast cell walls. / http://proxy.library.adelaide.edu.au/login?url= http://library.adelaide.edu.au/cgi-bin/Pwebrecon.cgi?BBID=678380 / Thesis (Ph.D.) -- University of Adelaide, Dept. of Plant Science, 2000

Yeast mannoproteins released during the process of aging wine on the yeast lees have been reported to make important contributions to wine quality. However, few mannoproteins have been identified in wine and their lifespan during aging is unknown. As a first step towards better understanding the contributions of yeast mannoproteins to wine quality, a model system was used to follow changes in yeast protein release, and to identify the released proteins over a 9-month time course following completion of fermentation. Model musts were fermented in duplicate by a number of commercial yeast strains, including BM45 and RC212, and were stored on the yeast lees post-fermentation with monthly stirring at 15°C. Wine samples were taken during and after fermentation, and following removal of suspended solids, total protein and total mannoproteins were measured, and individual proteins were identified--but not quantified--by HPLC-MS of tryptic fragments. The total number of identified proteins in all samples increased from between 3-15 following inoculation, to between 70-80 after one month on the lees, and decreased to about 20 after 6 months on the lees. Over 50% of the identified proteins were shared among all yeast samples. For most strains, protein and mannoprotein concentrations increased during, but decreased by the end of fermentation. Both protein and mannoprotein concentrations were found to increase again post-fermentation, reaching values about 2- and 6-fold higher than values measured at 2 days, respectively. Consistent with the increase in mannoprotein concentration, cell wall mannoproteins were the predominant proteins identified after 6 months of aging on the lees. Most cytosolic proteins found during and soon after fermentation were not found after 6 months of aging on the lees. / Graduation date: 2009

White wine clarity is essential, but it can be marred by the presence of a protein haze. This protein haze is predominantly formed by grape-derived thaumatin-like proteins and chitinases, which can slowly denature and aggregate if left in bottled wines. Currently bentonite fining is used by the wine industry to prevent protein haze. Bentonite consists of fine clay particles that, when added to wine, bind and remove the haze-forming proteins. However this method is inconvenient, time-consuming, and causes significant losses of wine. It is estimated that this process costs the Australian wine industry $50 m annually in wine losses alone. Alternatives are thus being investigated. The principal objective of this thesis was to investigate the sensory effects on wine of an alternative method to bentonite fining: addition of haze protective factor 2, known as Hpf2. Hpf2 is a Saccharomyces cerevisiae mannoprotein that has been shown to reduce protein haze in wines. It is a highly mannosylated 180 kDa protein, of which approximately 75% by weight is mannose. Previous work has shown that the addition of approximately 200 mg L⁻¹ Hpf2 to wines reduces the visible haze in wine by approximately 50%. Hpf2 is naturally present in wines at concentrations of less than 10 ng L⁻¹, much lower than the concentration required for haze protection activity. However, the sensory impacts involved with the addition of such high concentrations of Hpf2 in wine have never been studied. This knowledge is essential for the future commercial prospects of this alternative approach to protein stabilisation of wine. To undertake sensory studies, over 1 g of Hpf2 would be required. Presently, the laboratory-scale process for the production of a 6-histidine tagged version of the protein, 6xHisHpf2, in a laboratory yeast strain of S. cerevisiae, produces only milligram quantities. Consequently, the first challenge of this research was to scale up the existing process to produce sufficient quantities of Hpf2. The first attempt to increase the production level was by over-expression in the bacteria Escherichia coli. Although several approaches were trialled, 6xHisHpf2 was unable to be successfully and consistently expressed in this system. The second method was by improving the original yeast expression system, and the expression level was able to be improved approximately 10-fold. This improved expression method was scaled up to produce and then purify over 1 g of protein. Several quantification methods were assessed to determine the efficiencies of each purification step, with slot blot analysis proving successful. Sensory trials were conducted to establish the effect of 6xHisHpf2 on wines, with duo-trio studies conducted assessing both aroma and palate of the wines. Invertase, another yeast haze protective factor, was also trialled. It was found that the addition of an active level of 6xHisHpf2 or invertase did not cause a significant difference in the aroma or palate of wines. In addition to this main study, the role of the glycosylation was studied. 6xHisHpf2, produced in a different yeast, Pichia pastoris, was found to be 83 kDa, with only 50% mannose. This protein was compared to the S. cerevisiae protein in its ability to reduce protein haze, and it was shown that the P. pastoris protein could reduce haze, but not as effectively as the S. cerevisiae protein. The finding that Hpf2 does not affect the sensory properties of wine is essential if Hpf2 is to be used commercially, as winemakers and wine consumers would most likely reject an additive that alters the wine aroma or palate. This work has brought the wine industry a step closer to a new method for protein haze prevention in white wines. / Thesis (Ph.D.) -- University of Adelaide, School of Chemical Engineering, 2008

White wine clarity is essential, but it can be marred by the presence of a protein haze. This protein haze is predominantly formed by grape-derived thaumatin-like proteins and chitinases, which can slowly denature and aggregate if left in bottled wines. Currently bentonite fining is used by the wine industry to prevent protein haze. Bentonite consists of fine clay particles that, when added to wine, bind and remove the haze-forming proteins. However this method is inconvenient, time-consuming, and causes significant losses of wine. It is estimated that this process costs the Australian wine industry $50 m annually in wine losses alone. Alternatives are thus being investigated. The principal objective of this thesis was to investigate the sensory effects on wine of an alternative method to bentonite fining: addition of haze protective factor 2, known as Hpf2. Hpf2 is a Saccharomyces cerevisiae mannoprotein that has been shown to reduce protein haze in wines. It is a highly mannosylated 180 kDa protein, of which approximately 75% by weight is mannose. Previous work has shown that the addition of approximately 200 mg L⁻¹ Hpf2 to wines reduces the visible haze in wine by approximately 50%. Hpf2 is naturally present in wines at concentrations of less than 10 ng L⁻¹, much lower than the concentration required for haze protection activity. However, the sensory impacts involved with the addition of such high concentrations of Hpf2 in wine have never been studied. This knowledge is essential for the future commercial prospects of this alternative approach to protein stabilisation of wine. To undertake sensory studies, over 1 g of Hpf2 would be required. Presently, the laboratory-scale process for the production of a 6-histidine tagged version of the protein, 6xHisHpf2, in a laboratory yeast strain of S. cerevisiae, produces only milligram quantities. Consequently, the first challenge of this research was to scale up the existing process to produce sufficient quantities of Hpf2. The first attempt to increase the production level was by over-expression in the bacteria Escherichia coli. Although several approaches were trialled, 6xHisHpf2 was unable to be successfully and consistently expressed in this system. The second method was by improving the original yeast expression system, and the expression level was able to be improved approximately 10-fold. This improved expression method was scaled up to produce and then purify over 1 g of protein. Several quantification methods were assessed to determine the efficiencies of each purification step, with slot blot analysis proving successful. Sensory trials were conducted to establish the effect of 6xHisHpf2 on wines, with duo-trio studies conducted assessing both aroma and palate of the wines. Invertase, another yeast haze protective factor, was also trialled. It was found that the addition of an active level of 6xHisHpf2 or invertase did not cause a significant difference in the aroma or palate of wines. In addition to this main study, the role of the glycosylation was studied. 6xHisHpf2, produced in a different yeast, Pichia pastoris, was found to be 83 kDa, with only 50% mannose. This protein was compared to the S. cerevisiae protein in its ability to reduce protein haze, and it was shown that the P. pastoris protein could reduce haze, but not as effectively as the S. cerevisiae protein. The finding that Hpf2 does not affect the sensory properties of wine is essential if Hpf2 is to be used commercially, as winemakers and wine consumers would most likely reject an additive that alters the wine aroma or palate. This work has brought the wine industry a step closer to a new method for protein haze prevention in white wines. / Thesis (Ph.D.) -- University of Adelaide, School of Chemical Engineering, 2008

Thesis (PhD (Science) (Viticulture and Oenology. Wine Biotechnology))--University of Stellenbosch, 2010. / ENGLISH ABSTRACT: The yeast Saccharomyces cerevisiae senses change within its environment and responds through specific adaptive cellular programmes, in particular by modifying gene expression. Many adaptive changes affect the physico-chemical properties of the cell wall, and several mechanisms that specifically affect the expression levels of genes that encode for cell wall components have been described previously. Cell wall modification directly impacts on general cell wall properties and cell-cell and cell-surface interactions. Many of these properties have been directly linked to families of cell wall proteins referred to as adhesins. In particular members of the Flocculation (FLO) gene family have been shown to play a crucial role in adhesion phenotypes. Flo11p functions in a variety of phenotypes including agar invasion, plastic adhesion and the formation of pseudohyphae, “flor” and “mats”, whereas Flo1p appears to control flocculation. The regulation of FLO11 expression is well documented and is mainly controlled by the mitogen activated protein kinase (MAPK) and cyclic AMP protein kinase A (cAMP-PKA) signalling cascades. Genetic analysis shows that Mss11p acts downstream and is central to these pathways, and furthermore interacts with the cAMP-PKA component Flo8p to activate transcription. In this study we further explore additional gene targets of Flo8p and Mss11p, as well as their regulation and their impact on cell wall characteristics and associated adhesion phenotypes. Our analysis shows that Mss11p is also required for FLO1 expression, and functions together with Flo8p to control many Flo-dependent adhesion phenotypes. Genome-wide gene expression analysis further reveals that altered Mss11p levels leads to the change in the expression of various cell membrane and cell wall genes, notably AQY2 and members of the DAN and TIR gene families. Further genetic analysis indicates that adhesion phenotypes display an almost exclusive dependence on FLO gene expression. We also demonstrate that these phenotypes require Flo10p and are thus dependent on the specific balance of Flo proteins in the cell wall. The analysis of signalling deletion mutants show that regulation of FLO10 shares signalling components with FLO11, but that the two genes are differentially regulated. Unlike FLO11, FLO10 transcription also does not display an absolute requirement for Mss11p but rather for the MAPK component Ste12p. Whole genome expression analysis were also performed on strains with altered levels of Flo8p which were compared with the above mentioned transcriptome data set. This analysis shows that Flo8p and Mss11p co-regulate the FLO genes, as well as AQY2 and TIR3, but also have significant unique gene targets. The combination of transcriptome data with current information concerning transcription factor (TF) interaction networks reveals the importance of network interaction between Cin5p, Flo8p, Mga1p and Mss11p. From these data we constructed a TF interaction model in which Flo8p acts as the predominantly activating TF component, whereas Mss11p function as a target hub TF, possibly as a mediator- or polymerase II holo-enzyme component. Finally we provide a first report on “mat” formation by an industrial wine yeast strain, and show that by adjusting FLO11 expression in this strain we are able to significantly change this phenotypic behaviour. / AFRIKAANSE OPSOMMING: Die gis Saccharomyces cerevisiae neem veranderinge in sy omgewing waar en reageer daarop deur middel van spesifieke sellulêre programme, in die besonder deur geenuitdrukking aan te pas. Verskeie aanpasbare veranderinge beïnvloed die fisieke, asook chemiese eienskappe van die selwand, en talle meganismes is al beskryf wat die uitdrukkingsvlakke beïnvloed van gene wat vir selwandkomponente kodeer. Die modifikasie van die selwand het ’n direkte impak op selwand-eienskappe, asook die sel-sel- en sel-oppervlak-interaksies. Verskeie van hierdie eienskappe word direk gekoppel aan die selwandproteïenfamilies, wat ook as adhesie-faktore bekend staan. Veral lede van die Flokkulasie (FLO) -geenfamilie het ’n noodsaaklike funksie in adhesie-fenotipes. Flo11p speel ’n rol in verskeie fenotipes, wat insluit die indringende groei van agar, plastiekaanhegting en die vorming van pseudohifes, “flor“ en “matte“, terwyl Flo1p flokkulasie beheer. Die regulering van FLO11-uitdrukking is deeglik gedokumenteerd en dit word hoofsaaklik gereguleer deur die mitogeen-geaktiveerde proteïenkinase (MAPK) en sikliese AMP-proteïenkinase A (cAMP-PKA) seintransduksiekaskades. Genetiese analises toon dat Mss11p stroom-af en sentraal tot hierdie kaskades funksioneer, en dit aktiveer transkripsie deur interaksie met die cAMP-PKA-komponent, Flop8. In hierdie studie word ’n ondersoek gedoen na addisionele teikengene van Flo8p en Mss11p, en hoe hierdie gene gereguleer word, asook hul impak op selwandeienskappe en geassosieerde adhesie-fenotipes. Ons analises toon dat Mss11p ook benodig word vir die ekspressie van FLO1 en dat dit, tesame met Flo8p, beheer uit oefen oor verskeie Flo-afhanklike fenotipes. Genoomwye geenekspressie-analises wys verder daarop dat veranderde Mss11p-vlakke lei tot die aanpassing van die ekspressie van verskeie selmembraan- en selwandgene, naamlik AQY2 asook lede van die DAN- en TIR-geenfamilies. Verdere genetiese analise dui daarop dat adhesie-fenotipes byna eksklusief afhanklik is van FLO-geenekspressie. Daar is verder getoon dat hierdie fenotipes ook Flo10p benodig en dus afhanklik is van die spesifieke balans van Floproteïene in die selwand. Die analise van seintransduksiemutante demonstreer dat FLO10 en FLO11 seintransduksie-komponente deel, maar dat hierdie gene verskillend gereguleer word. Anders as FLO11, toon FLO10 nie ’n absolute noodsaaklikheid vir Mss11p nie, maar eerder vir die MAPK-komponent, Ste12p. Totale genoomekspressie-analises is ook gedoen op gisrasse met aangepaste vlakke van Flo8p en dis vergelyk met bogenoemde transkripsiedatastel. Hierdie analise wys dat Flo8p and Mss11p die FLO-gene, asook AQY2 en TIR3, koreguleer, maar ook beduidende unieke teikengene het. Die kombinasie van transkripsiedata met huidig beskikbare informasie betreffende transkripsiefaktor (TF) -interaksienetwerke dui op die relevansie van netwerkinteraksie tussen Cin5p, Flo8p, Mga1p en Mss11p. Hiervan is daar ’n model opgestel waarin Flo8p in die meeste gevalle as die aktiverende TF-komponent optree, terwyl Mss11p as TF-teiken dien, moontlik as ’n mediator- of polimerase II holoënsiemkomponent. Laatens word daar vir die eerste keer verslag gedoen van ”mat”-vorming deur ’n industriële wyngisras en toon ons verder dat hierdie fenotipe beduidend verander word deur middel van die aanpassing van FLO11-uitdrukking.

Thesis (PhD)--Stellenbosch University, 2012. / ENGLISH ABSTRACT: Wine protein haze formation is a major challenge for wine makers, and several wine clarifying agents such as bentonite are used in the industry to protect wine from this occurrence. However, clarifying agents may have an undesirable impact on wine quality. Yeast mannoproteins have been shown to possess haze-protective properties, while also positively impacting on the sensorial properties of the product. However, while such mannoproteins are released into the wine during the wine making process, the amounts are low and therefore of limited oenological significance. However, and although commercial wine yeast strains display significant genotypic and phenotypic diversity, no broader assessment of haze protective activity and of mannoproteins release by different wine yeast strains has been undertaken. In this study, several yeast strains were screened for their impact on wine haze formation in Chardonnay must and in a grape juice model system. The data show that strains of the species Saccharomyces paradoxus possess better haze protective properties than the common Saccharomyces cerevisiae wine yeast strains. Differences in the nature of the proteins released by these two species were investigated, and indicated that several mannoproteins were released at significantly higher levels by S. paradoxus, and that some of these proteins might indeed contribute to the haze-protective activity. A further exploration of yeast cell wall properties indicated that the cell walls of haze-protective S. paradoxus strains contained higher levels of chitin than non-haze protective strains. Grape chitinases are likely to be primarily responsible for wine haze formation, and the data clearly demonstrate that these enzymes are able to bind to the yeast cell walls, and that strains with higher amounts of chitin in the cell wall will bind more chitinases. This finding suggests that the haze-protective nature of the strains is at least in part linked to the chitin levels of the strains. Furthermore, the impact of some genetic modifications in two wine strains (namely S. cerevisiae VIN13 and S. paradoxus RO88) suggests that several proteins contribute to wine haze protection. However, none of the mannoprotein-encoding flocculation genes, FLO1, FLO5, and FLO11 showed any impact on this property. Further studies are required to assess the full impact of the S. paradoxus strains on haze protection. In particular, the possible use of such strains as starter cultures or the use of S. paradoxus yeast hulls as clarifying agent needs to be further explored. / AFRIKAANSE OPSOMMING: Wyn proteïen-waas vorming is 'n groot uitdaging vir wynmakers en verskeie wyn verhelderings agente soos bentoniet word in die wynbedryf gebruik om wyn te beskerm teen die vorming van waas. Hierdie verheldering agente het egter 'n ongewenste impak op wynkwaliteit. Gis mannoproteïene is uitgewys as proteïene met moontlike waas-beskermende eienskappe wat ook 'n positiewe uitwerking op die sensoriese eienskappe van die produk het. Al word hierdie mannoproteïene egter vrygestel in die wyn tydens die wynmaak proses, is die hoeveelhede oor die algemeen te laag om van wynkundige belang te wees. Verder, ten spyte van die beduidende genotipiese en fenotipiese diversiteit van kommersiële wyngisrasse is daar nog geen breër assessering van die waas beskermende aktiwiteit van mannoproteïene, vrygestel deur verskillende rasse, tot dusver onderneem nie. In hierdie studie is verskeie gisrasse gekeur vir hul impak op wyn waas-vorming in Chardonnay mos en ook in 'n model druiwesap. Die data wys dat rasse van die spesie Saccharomyces paradoxus besit beter waas beskermende eienskappe as die algemene Saccharomyces cerevisiae wyngisrasse. Verskille in die aard van die proteïene wat vrygestel is deur hierdie twee spesies is ondersoek, en dit is aangedui aangedui dat verskeie mannoproteins vrygestel aan aansienlik hoër vlakke deur S. Paradoxus. Dit is ook aangedui dat sommige van hierdie proteïene wel bydra tot die waas-beskermende aktiwiteit. 'n Verdere verkenning van gis selwand eienskappe het aangedui dat die selwande van waas-beskermende rasse van S. paradoxus hoër vlakke chitien as nie-waas beskermende stamme bevat. Druiwe chitinases is waarskynlik hoofsaaklik verantwoordelik vir wyn waas vorming, en die data toon duidelik dat hierdie ensieme in staat is om te bind aan die gis selwande, en dat die stamme met hoër vlakke chitien in die selwand meer chitinases sal bind. Hierdie bevinding dui daarop dat die waas-beskermende aard van die stamme ten minste gedeeltelik gekoppel is aan die chitien vlakke van die stamme. Die impak van sekere genetiese modifikasies in twee verskillende gisrasse, naamlik die S. cerevisiae ras VIN13 en die S. paradoxus ras RO88, dui verder daarop dat verskeie proteïene dra by tot die beskerming teen wyn waas. Geeneen van die mannoprotein-koderende flokkulasie gene, FLO1, FLO5 en FLO11 het egter 'n impak op hierdie eienskap nie. Verdere studies is nodig om die volle impak van die S. paradoxus rasse op waas beskerming te assesseer. In die besonder, die moontlike gebruik van sulke rasse as 'n inkolasie kultuur of die gebruik van S. paradoxus gis doppe as verheldering agent moet verder ondersoek word.

La clarification des vins par procédés membranaires en particulier la microfiltration tangentielle a toujours été limitée par le colmatage, générant des flux de perméation faibles incompatibles avec la rentabilité économique. La compréhension, la maîtrise, ainsi que l'anticipation des phénomènes de colmatage font l'objet de ce travail. Dans un premier temps, la contribution individuelle puis en mélange des composés du vin (tannins, pectines, mannoprotéines et levures) au colmatage d'une membrane céramique multicanaux a été évaluée. Une approche fondamentale a permis de proposer des mécanismes de colmatage : la présence des pectines induit les flux de perméation les plus faibles en formant un gel à la surface de la membrane tandis que les levures semblent au contraire avoir un effet protecteur dans le cas d'un vin brut. Parce qu'il représente le flux au-delà duquel un colmatage irréversible apparaît à la surface de la membrane, le flux critique pour l'irréversibilité est un paramètre clef pour contrôler le colmatage. Dans le cas de la filtration de vin, aucun flux critique n'a pu être déterminé ce qui a conduit à définir un critère identifiant une zone de travail où le degré de colmatage reste acceptable. La dernière partie de cette thèse est consacrée à l'étude de la filtration dynamique (RVF) pour une éventuelle application dans le secteur vinicole. Cette technique est testée en présence de deux membranes organiques différentes: PES (hydrophile) et PTFE (hydrophobe). Les observations ont permis de mettre en évidence l'efficacité du système contre le colmatage des membranes PES induite par l'action de l'agitateur. Le colmatage des membranes PTFE est énormément affecté par les interactions molécules/membrane rendant la filtration dynamique inefficace pour lutter contre le colmatage de ces membranes. / Wine clarification by membrane processes mainly cross-flow microfiltration has been limited by membrane fouling generating low permeate fluxes with economic efficiency. Understanding, controlling and anticipation of fouling are the main goals of this work. In a first time, the individual contribution of wine compounds (tannins, pectins, mannoproteins and yeasts) to a multichannel ceramic membrane fouling was evaluated. The fouling mechanisms were analyzed using a fundamental approach. The presence of pectins induce the lowest fluxes by a gel-type formation at the membrane surface while yeasts presence tends to reduce fouling in the case of crude wine (case of mixed components). Because it represents the flux beyond which irreversible fouling appears on the membrane surface, the critical flux for irreversibility is a key parameter to control fouling. No critical flux for irreversibility could be measured, hence a criterion that identifies a range of operating conditions where the degree of fouling remains acceptable was proposed. The last part of this work was devoted to the study of dynamic filtration (RVF) for further application in wine sector. This technique was tested with two different membranes: hydrophilic PES and hydrophobic PTFE. Results have allowed to demonstrate the efficiency of the system to reduce fouling in the case of PES membrane. Fouling of PTFE membrane is greatly influenced by molecules / membrane interaction making dynamic filtration ineffective in reducing of membrane fouling.