Lactobacillus: the Next Generation of Malolactic Fermentation Starter Cultures an Overview

Food Bioprocess Technol (2011) 4:876 906 DOI 10.1007/s11947-010-0448-8 REVIEW PAPER Lactobacillus: the Next Generation of Malolactic Fermentation Starter Cultures an Overview Maret du Toit & Lynn Engelbrecht & Elda Lerm & Sibylle Krieger-Weber Received: 22 June 2010 / Accepted: 30 September 2010 / Published online: 16 October 2010 # Springer Science+Business Media, LLC 2010 Abstract Malolactic fermentation (MLF) is a secondary wine fermentation conducted by lactic acid bacteria (LAB). This fermentation is important in winemaking as it deacidifies the wine, it contributes to microbial stability and lastly it contributes to wine aroma through the production of metabolites. Oenococcus oeni is the main species used in commercially available starter culture currently, but research has indicated that different Lactobacillus species also partake in MLF and this has shifted the focus in the MLF field to evaluate the potential of lactobacilli as starter cultures for the future. There are 17 different species of Lactobacillus associated with winemaking either being associated with the grapes/beginning of alcoholic fermentation or the MLF and wine. Lactobacillus associated with wine is mainly facultative or obligatory heterofermentative and can withstand the harsh wine conditions such as high ethanol levels, low ph and temperatures and sulphur dioxide. Wine lactobacilli contain the malolactic enzyme encoding gene, but sequence homology shows that it clusters separate from O. oeni. Lactobacillus also possesses more enzyme encoding genes compared to O. oeni, important for the production of wine aroma compounds such as glycosidase, protease, esterase, phenolic acid decarboxylase and citrate lyase. Another characteristic associated with wine lactobacilli is the M. du Toit (*) : L. Engelbrecht : E. Lerm Institute for Wine Biotechnology, Stellenbosch University, Private Bag X1, Matieland, 7602 Stellenbosch, South Africa e-mail: mdt@sun.ac.za S. Krieger-Weber Lallemand, In den Seite 53, 70825 Korntal-Münchingen, Germany production of bacteriocins, especially plantaricins which would enable them to combat spoilage LAB. All these characteristics, together with their ability to conduct MLF just as efficiently as O. oeni, make them suitable for a new generation of MLF starter cultures. Keywords Lactobacillus. Wine. Malolactic fermentation. Starter cultures. Metabolism Introduction Winemaking is a complex process involving the interaction of different microbes. The two main groups of microorganisms involved are yeasts and bacteria. The yeasts and bacteria taking part in the fermentation can originate from grapes, the winery environment or as inoculated starter cultures. Yeasts are mainly responsible for the alcoholic fermentation, especially Saccharomyces cerevisiae, and contribute to the production of major aroma compounds such as esters, higher alcohols, aldehydes and fatty acids. Yeasts, especially genera of the non- Saccharomyces group, can also be detrimental to wine quality by spoiling the wine through the production of haze, film layers or off-flavours. Malolactic fermentation (MLF) is a secondary fermentation that usually takes place during alcoholic fermentation or at the end of alcoholic fermentation and is carried out by one or more species of lactic acid bacteria (LAB). Four genera were identified as the principal organisms involved in winemaking: Lactobacillus, Leuconostoc, Oenococcus and Pediococcus. Of all the species of LAB, Oenococcus oeni is probably the best adapted to overcome the harsh environmental wine conditions and therefore represents the majority of commercial MLF starter cultures. Depending on

Food Bioprocess Technol (2011) 4:876 906 877 the wine style, MLF can be beneficial or detrimental. It contributes to the stabilisation of wine by deacidification and removal of residual nutrients. Moreover, the organoleptic profile and quality of the final product are changed via secondary metabolic reactions (Wibowo et al. 1985; Davis et al. 1988; Drici-Cachon et al. 1996; Lonvaud-Funel 1999). Lactobacillus spp. have shown that they can survive the winemaking conditions and that they possess many favourable characteristics that would make them suitable MLF starter cultures in the future. Secondary metabolic reactions are of great importance for aroma and flavour development in wine which include citrate metabolism, amino acid metabolism, metabolism of polysaccharides, metabolism of polyols, catabolism of aldehydes, hydrolysis of glycosides, synthesis and hydrolysis of esters, degradation of phenolic acids, lipolysis, proteolysis and peptidolysis (Liu 2002; Matthews et al. 2004). It was shown by several authors that many wine-associated lactobacilli have the genes that encode for the enzymes involved in the metabolic activities mentioned above and that some of the enzymes are active under winemaking conditions (Cavin et al. 1997; Vaquero et al. 2004; Grimaldi et al. 2005; Spano et al. 2005; Matthews et al. 2007; Nelson2008; DeLasRivasetal.2009; Mtshali et al. 2010). LAB have the ability to produce antimicrobial peptides which might help prevent the production of undesired compounds by inhibiting the indigenous LAB microflora and allowing the MLF to be conducted by the inoculated bacterial strains (Du Toit and Pretorius 2000). The possibility of controlling bacterial growth during winemaking and preservation by bacteriocins such as leucocin, nisin, pediocin PA-1, pediocin N5p and plantaricin 423 has been investigated in several studies (Radler 1990a, b; Daeschel et al. 1991; Strasser de Saad and Manca de Nadra 1993; Du Toit 2002; Yurdugül and Bozoglu 2002; Bauer et al. 2003; Rojo-Bezares et al. 2007). Several wine-related lactobacilli also have the ability to produce bacteriocins or have the genes encoding forbacteriocinproduction(navarroetal.2000; Holoetal. 2001; Yurdugül and Bozoglu 2002; Rojo-Bezares et al. 2008; Knoll et al. 2008; Sáenz et al. 2009). Lactobacillus plantarum has genetically shown to have the most diverse array of plantaricins (Knoll et al. 2008; Rojo-Bezares et al. 2008; Sáenz et al. 2009). This characteristic can be an important selection criterion for future lactobacilli MLF starter cultures to assist in their dominance of a wine fermentation and also to reduce the level of sulphur dioxide (SO 2 )used. The fast progress of genomics, transcriptomics, proteomics and metabolomics of numerous LAB on genus, species and strain level provides great knowledge about their diversity and evolution. Moreover, this information helps to understand the mechanisms that control and regulate bacterial growth, signalling, survival, stress response and fermentation processes. Several genomes have been sequenced of wine relevant species such as L. plantarum, Lactobacillus casei, Lactobacillus brevis, Lactobacillus fermentum, O. oeni, Leuconostoc mesenteroides and Pediococcus pentosaceus. The information on the genome of O. oeni has led to the development of an array-based comparative genome hybridization technique that has shown the genetic diversity in the strains evaluated and the same technology to study the diversity within the Lactobacillus spp. can be applied in the future (Borneman et al. 2010). Proteomic approaches have been used to study the sugar fermentation patterns in L. plantarum (Plumed-Ferrer et al. 2008) and the effect of tannic acid on Lactobacillus hilgardii (Bossi et al. 2007). Metabolomics have been used to compare the effect of L. plantarum and O. oeni on primary and secondary metabolite production in wine. Results obtained showed that they could be clearly differentiated from each other withregardstobothprimaryandsecondarymetabolites produced (Lee et al. 2009). These insights and knowledge gained on the omics of wine LAB will facilitate the selection of novel strains or breeding of strains that are better adapted as starter cultures for MLF in the future. This review will focus on the potential of Lactobacillus spp. as the next generation of MLF starter cultures, discussing their occurrence during winemaking, major metabolic activities, factors influencing their growth and their performance in conducting MLF. Lactobacillus spp. and the Wine Environment Apart from O. oeni being the dominant LAB isolated from MLF, it is shown by several studies that lactobacilli are found throughout the wine environment, also during MLF and bottled wine, indicating that certain strains have adapted to survive under winemaking conditions. Recently, the focus has started to shift towards species from other LAB genera for attributes that could be beneficial in the winemaking process. One such group includes species from the genus Lactobacillus. Lactobacillus is the largest genus amongst all LAB and consists of approximately 100 species and at least 16 subspecies. The growth temperature ranges from 2 to 53 C, with an optimum of 30 40 C and an optimum ph range of 5.5 6.2 (Dicks and Endo 2009). The genus Lactobacillus is non-mobile, non-sporulating, regular elongated cells that are 0.5 1.2 μm by1.0 10 μm in dimension. They tend to be long rod-like forms that are often assembled in pairs or chains of varying lengths. Lactobacillus spp. possess fermentative metabolisms and

878 Food Bioprocess Technol (2011) 4:876 906 can be divided into three main metabolic groups based on their metabolite production from glucose/pentose: obligatory homofermentative, facultative heterofermentative and obligatory heterofermentative (Dicks and Endo 2009). Carbohydrate metabolism will be discussed in a later section. Dicks and Endo (2009) recently reviewed the taxonomic status of wine LAB and key phenotypic characteristics to differentiate LAB. The LAB that have been isolated from grapes, must and wine are listed in Table 1. In order to understand the prevalence of Lactobacillus spp. in wine, we need to understand the progression of the different LAB genera and species during the winemaking process. The evolution of LAB from the vineyard to the final vinification stages has been documented, but shows considerable variability due to region, cultivar and vinification procedures. It is clear that there is a successional growth of several species of LAB during vinification (Wibowo et al. 1985; Boulton et al. 1996; Fugelsang and Edwards 1997). Several studies have been done regarding the diversity of Lactobacillus spp. associated with the winemaking process. As a result, many novel Lactobacillus spp. have been described based on investigations into the phylogenetic, genotypic and phenotypic characterisation of isolates (Mañes-Lázaro et al. 2009). In the vineyard, the diversity and population density of LAB are very limited, especially in comparison to the indigenous yeast population found on grapes (Fugelsang and Edwards 1997). LAB occur on grapes and leaf surfaces (Wibowo et al. 1985), but population numbers Table 1 The main lactic acid bacteria species often associated with grapes, must and wine (adapted from Dicks and Endo 2009; Pozo-Bayón et al. 2009) Genus Species Reference Pediococcus P. damnosus Peynaud and Domercq (1967), Back (1978), Lonvaud-Funel et al. (1991), Dueñas et al. (1995), Beneduce et al. (2004), Ribéreau-Gayon et al. (2006) P. inopinatus Back (1978), Edwards and Jensen (1992) P. parvulus Nonomura et al. (1967), Edwards and Jensen (1992), Davis et al. (1986a, b), Rodas et al. (2003) P. pentosaceus Lonvaud-Funel et al. (1991), Salado and Strasser De Saad (1995), Eliseeva et al. (2001), Rodas et al. (2003), Ribéreau-Gayon et al. (2006) Leuconostoc Leuc. mesenteroides Garvie (1979, 1983), Lonvaud-Funel and Strasser De Saad (1982), Lonvaud-Funel et al. (1991), Ribéreau-Gayon et al. (2006), Izquierdo et al. (2009), Ruiz et al. (2010) Leuc. paramesenteroides Garvie (1983) Oenococcus O. oeni Garvie (1967b), Lonvaud-Funel et al. (1991), Ribéreau-Gayon et al. (2006), López et al. (2007), Ruiz et al. (2008, 2010) Lactobacillus L. brevis Vaughn (1955), Du Plessis and Van Zyl (1963), Ribéreau-Gayon et al. (2006) L. bobalius Mañes-Lázaro et al. (2008a) L. buchneri Vaughn (1955), Du Plessis and Van Zyl (1963) L. casei Vaughn (1955), Carre (1982), Lonvaud-Funel et al. (1991), Izquierdo et al. (2009), Ruiz et al. (2010) L. collinoides Carr and Davies (1972), Couto and Hogg (1994) L. fermentum Vaughn (1955) L. fructivorans Amerine and Kunkee (1968), Couto and Hogg (1994) L. hilgardii Douglas and Cruess (1936), Vaughn (1955), Carre (1982), Couto and Hogg (1994), Ribéreau- Gayon et al. (2006), Izquierdo et al. (2009), Ruiz et al. (2010) L. kunkeei Edwards et al. (1998), Bae et al. (2006) L. lindneri Bae et al. (2006) L. mali Carr and Davies (1970), Couto and Hogg (1994), Bae et al. (2006) L. nagelii Edwards et al. (2000) L. oeni Mañes-Lázaro et al. (2009) L. paracasei Du Plessis et al. (2004) L. paraplantarum Curk et al. (1996), Krieling (2003) L. plantarum Carre (1982), Wibowo et al. (1985), Lonvaud-Funel et al. (1991), Beneduce et al. (2004), Bae et al. (2006), Ribéreau-Gayon et al. (2006), Izquierdo et al. (2009), Ruiz et al. (2010) L. uvarum Mañes-Lázaro et al. (2008b) L. vini Rodas et al. (2006)

Food Bioprocess Technol (2011) 4:876 906 879 on undamaged grapes and in grape must are rarely higher than 10 3 cfu/g (Lafon-Lafourcade et al. 1983). The population size on grape surfaces depends largely on the maturity and sanitary state of the grapes (Wibowo et al. 1985; Jackson2008), and Pediococcus and Leuconostoc spp. occur on grapes more frequently than O. oeni (Jackson 2008). The LAB species isolated from grapes before harvest by Carre (1982) belonged to the Lactobacillus genus and included L. plantarum, L. hilgardii and in some cases, L. casei. In 2006, Bae et al. investigated LAB associated with Australian wine grapes. Lactobacillus lindneri was the main species isolated from Cabernet Sauvignon, Merlot and Shiraz while L. plantarum and Lactobacillus mali were also present on the Cabernet Sauvignon grapes. L. lindneri and Lactobacillus kunkeei were the main species present on Chardonnay, Semillon and Sauvignon blanc grapes. Besides grape surfaces, bacterial strains can also be isolated from the cellar environment, including barrels and poorly sanitised winery equipment like pipes and valves (Donnelly 1977; Boulton et al. 1996; Jackson 2008). Shortly after crushing and the start of alcoholic fermentation (AF), the LAB population in the grape must generally ranges from 10 3 to 10 4 cfu/ml. During the first days of alcoholic fermentation, when bacterial cell counts are mainly influenced by the original sanitary state of the grapes and the fermentation parameters like temperature, bacterial microflora are present in varying quantities of 10 2 10 4 cfu/ml (Ribéreau-Gayon et al. 2006). The major species of LAB present at this stage include L. plantarum, L. casei, Leuc. mesenteroides and Pediococcus damnosus as well as O. oeni to a lesser extent (Wibowo et al. 1985; Lonvaud-Funel et al. 1991; Boulton et al. 1996; Powell et al. 2006). In comparison to the microflora found in the vineyard, the species found in grape must are more diverse and include L. plantarum, L. hilgardii, L. brevis, P. damnosus, P. pentosaceus, Leuc. mesenteroides and O. oeni (Ribéreau-Gayon et al. 2006). Most of these LAB species generally do not multiply and decline towards the end of alcoholic fermentation, with the exception of O. oeni (Wibowo et al. 1985; Lonvaud-Funel et al. 1991; Van Vuuren and Dicks 1993; Fugelsang and Edwards 1997; Volschenk et al. 2006). There is a natural progression of different LAB species, which is in part dependent on the sensitivity of the individual bacterial strains to yeast interactions. The bacterial population decreases to approximately 10 2 10 3 cfu/ml during the active stages and towards the end of alcoholic fermentation (Ribéreau-Gayon et al. 2006). The decrease could be attributed to increased ethanol concentrations, high SO 2 concentrations, low ph, low temperatures, the nutritional status and competitive interactions with the yeast culture (Fugelsang and Edwards 1997; Volschenk et al. 2006). During the active stages of alcoholic fermentation, some bacteria species like O. oeni, as well as the yeast population multiply, but bacterial growth still remains limited. This increase is dependent on and influenced by the ph and the addition of SO 2. A maximum population of approximately 10 4 10 5 cfu/ml is reached (Ribéreau-Gayon et al. 2006). After the completion of alcoholic fermentation and the LAB lag phase, as a result of ph, ethanol and temperature parameters, the surviving bacterial population will enter a latent phase which is followed by the active growth phase. The active growth phase can last for several days, and during this time, the population increases to 10 6 cfu/ml. At this stage, MLF will commence when the total population exceeds 10 6 cfu/ml and sufficient biomass is achieved. During this stage, temperature is essential in determining the LAB activity (Ribéreau-Gayon et al. 2006). The ph of the wine is imperative in determining which species of LAB are present, with values above ph 3.5 favouring the growth of Lactobacillus and Pediococcus spp., whereas the O. oeni population tend to dominate at lower ph values (Davis et al. 1986b; Henick-Kling 1993). Lonvaud-Funel et al. (1991) isolated LAB from musts and wines of Merlot, Cabernet franc and Cabernet Sauvignon. These isolates were identified as L. plantarum, L. casei, Leuc. mesenteroides, O. oeni, P. damnosus and P. pentosaceus. This study also found that a natural selection process took place during alcoholic fermentation. The homofermentative bacteria, followed by the heterofermentative Lactobacillus disappear, and subsequent to this, the homofermentative cocci and Leuc. mesenteroides disappear, all in favour of the growth of O. oeni. In similar findings, López et al. (2007) identified 98.5% of 204 isolates as being strains of O. oeni, isolated from Spanish red wines from the Rioja region. Ruiz et al. (2008) foundo. oeni to be the dominant LAB species present during spontaneous MLF in Cencibel wines from Spain. Similarly, a genetic study of the diversity of LAB isolated during spontaneous MLF in Tempranillo wines from six different wineries in Spain found O. oeni to be the dominant species. In the same study, the percentage of non-oenococcus species present during MLF varied between 2% and 10% and consisted of L. casei, L. hilgardii, L. plantarum and Leuc. mesenteroides (Izquierdo et al. 2009). Ruiz et al. (2010) had similar findings in an investigation of the LAB populations during spontaneous MLF in Tempranillo wines from Spain. The study was conducted over two vintages at five wineries. Samples were collected at three different stages: the end of alcoholic fermentation and the middle and end of MLF. Using both molecular and phenotypic methods, O. oeni was identified as the predominant species. The other strains were identified as L. casei, L. plantarum, L. hilgardii and Leuc. mesenteroides. These non-oenococcus

880 Food Bioprocess Technol (2011) 4:876 906 strains were isolated from all the wineries and at all the stages previously mentioned, but were more abundant at the beginning of MLF. Beneduce et al. (2004) investigated the prevalence of LAB from table wines produced in the Apulia Region, Italy. Of the 150 strains that were identified, most belonged to L. plantarum. These strains were isolated from wines with high ph, low SO 2 concentrations and without the addition of commercial cultures. Du Plessis et al. (2004) isolated 54 LAB strains from grape juice and at different stages of brandy base wine production and found the predominant species to be O. oeni, L. brevis, L. paracasei and L. plantarum. Couto and Hogg (1994) obtained 82 isolates from Douro fortified wines and winery equipment. Four species were identified, 89% of the isolates as L. hilgardii and one strain each of Lactobacillus fructivorans, Lactobacillus collinoides and L. mali. These findings highlight the ability of Lactobacillus strains to survive in high alcohol environments. When MLF is complete, the remaining LAB are still able to metabolise residual sugar, which could result in spoilage including the production of volatile acid (Fugelsang and Edwards 1997). This is particularly prevalent in high ph wines, where Lactobacillus and Pediococcus may occur and this could contribute to wine spoilage (Wibowo et al. 1985). When the malic acid is completely transformed, SO 2 additions are necessary to prevent the growth of spoilage microflora (Ribéreau- Gayon et al. 2006). The diversity of Lactobacillus species associated with the wine environment shows their potential application as co-inoculant in grape must or as inoculant after alcoholic fermentation, as they can dominate or survive under these conditions. Factors Affecting Their Growth Wine is a complex environment with sugar, ethanol, organic acids, amino acids, fatty acids, phenol contents, ph and SO 2 determining the growth of wine microorganisms. Various investigators reported many factors that influence the occurrence of LAB and MLF in wines. Henick-Kling (1988) listed besides oxygen and CO 2, carbohydrates, amino acids, vitamins and minerals, organic acid content, as well as the alcohol level, ph, SO 2, the method of vinification and interrelationships between LAB and other wine microorganisms to be the most influential factors to affect LAB growth. The wine ph is one of the most important factors which limit LAB growth and MLF in wine (Radler 1966; Dittrich 1977; Wibowo et al. 1985) and determines the type of LAB which will be present. Ideally for table wines, the ph should be between 3.1 and 3.6 (Amerine and Ough 1980), but due to global warming wine ph increased over the last years in almost all wine regions. ph Most LAB are neutrophilic (Guzzo and Desroche 2009), and generally, the optimum ph for the growth of LAB is close to neutrality (Hutkins and Nannen 1993). Some genera of bacteria such as Lactobacillus and Oenococcus show more acidophilic behaviour. At ph values less than 3.0, bacterial growth is difficult or impossible according to the presence or absence of other wine-limiting parameters (Lonvaud-Funel 1995). Wine ph affects the metabolism of sugars (Peynaud and Sapis-Domercq 1970) and also has a selective effect on the species (Mayer and Vetsch 1973). In wines of ph below 3.5, strains of O. oeni will generally dominate, and in wines of ph above 3.5, various strains of Lactobacillus and Pediococcus will dominate. Vaillant et al. (1995) studied the influence of some physico-chemical factors by experimental design assays. Malolactic activity of the tested L. plantarum strain had been mainly affected by a low ph compared to O. oeni. Strains of O. oeni have the lowest ph optimum for growth and it can grow optimally at ph values ranging from 4.5 to 6.5 (Radler 1966; Henick-Kling 1986). The ability of the bacteria to obtain energy from the metabolism of glucose is inhibited at the low ph of wine (Henick-Kling 1988). The optimum ph for catabolism of glucose by O. oeni and L. plantarum is between 4.5 and 6.0 (Henick-Kling 1986). According to Henick-Kling (1986) at low ph, glucose metabolism is most likely inhibited by the lower intracellular ph, but Cox and Henick-Kling (1989) could also show that in cells catabolising L-malate, the internal ph is increased by up to 0.5 units. McDonald et al. (1990) explained the ability of L. plantarum to terminate vegetable fermentations by the comparatively low growth-limiting internal ph and the ability to maintain a ph gradient at high organic acid concentration. Olsen et al. (1991) and Cox (1991) measured a 28 50% increase in cell yield for L. plantarum during fermentation under glucose-limited and low ph conditions. In addition, the preference of L. plantarum for malate as an energy source at low ph, even in the presence of glucose, suggests this species as starter for malate decarboxylation in fermenting musts (Guerzoni et al. 1995). Ethanol Ethanol is frequently used as a disinfection agent because of its bactericidal properties; it is therefore obvious that the ethanol produced during alcoholic fermentation will impact on LAB growth. It is difficult to specify the concentrations which will completely prevent LAB development. Radler (1966), Peynaud and Domercq (1968) and Henick-Kling

Food Bioprocess Technol (2011) 4:876 906 881 (1986) reported an increasing inhibition above 5% (v/v). Wibowo et al. (1985) stated in their review that the ability of LAB to survive and grow in wine decreases as the alcohol concentration increases above 10% (v/v). Optimum growth (i.e. shortest lag phase, fastest growth rate and highest cell yield) in the presence of 10% to 14% (v/v) occurs between 18 and 20 C whereas optimum growth at 30 C is achieved at 0% to 4% (v/v) ethanol (Henick-Kling 1993), indicating a strong impact of higher temperatures on the toxicity of ethanol. The degree of ethanol tolerance is, however, strain dependent and also depends, besides temperature, on the ph and nitrogen status of the environment. Specific details of alcohol sensitivity for the various species of wine LAB are not available and information is contradicting. Davis et al. (1988) reported strains of Lactobacillus and Pediococcus being in general more tolerant to high ethanol concentrations than O. oeni. From the observation of Wibowo et al. (1985), most Lactobacillus spp. can tolerate about 15% (v/v). Lactobacillus trichodes has been isolated from wines of 20% ethanol (Vaughn 1955). Guerzoni et al. (1995) studied the effects of ph, temperature, ethanol and malate concentrations on L. plantarum and O. oeni. The comparison of the results obtained from the two representative strains confirmed that L. plantarum is more resistant to the combined action of various stresses, at least when the ethanol was lower than 6% (v/v). They suggest that L. plantarum is, therefore, more competitive in the early steps of alcoholic fermentation. However, more severe conditions, e.g. ethanol concentrations higher than 6% (v/v), favour O. oeni. G-Alegría et al. (2004) observed similar high LAB populations, around 10 8 cfu/ml for both O. oeni and L. plantarum strains in a synthetic medium with 13% (v/v) ethanol and 18 C. Contrary to this, Henick- Kling (1993) reported O. oeni being only partially inhibited by ethanol concentrations above 5% (v/v) and able to tolerate up to 14% (v/v) alcohol while growth of L. plantarum stops at ethanol concentrations of 5 6% (v/v). Strains of L. casei and L. brevis are described to be more tolerant and have been successful to introduce MLF (Krieger 1989). Temperature Temperature modifies growth speed of all microorganisms (yeast and bacteria). The majority of LAB are described to be mesophilic (van de Guchte et al. 2002), andtheiroptimumgrowthisbetween25and30 Cin laboratory culture. In wine, the optimum temperature of growth is different from what is obtained in a laboratory culture. The optimum range is dependent on other physical and chemical parameters of the wine, notably the ethanol content. A higher ethanol content will lead to a decrease in the optimum growth temperature. According to French literature, bacterial growth is limited by temperature. Lowering the temperature by 5 C is sufficient (together with other unfavourable factors) to decrease the formed biomass and to prolong the MLF by 3 weeks (Lafon- Lafourcade 1983). The optimum temperature for LAB growth in wine is between 20 and 25 C. At 15 C or lower, the chance of bacterial growth is slight. However, Ribéreau-Gayon et al. (1975) proposed that once MLF has started at 15 C it may proceed at lower temperatures but at a slower rate. Guerzoni et al. (1995) studied the effects of several chemico-physical factors (ph, SO 2, ethanol concentration and temperature) on L. plantarum and O. oeni and modelled the malolactic activity. A temperature increase only positively affected the lag phase of O. oeni, but not of L. plantarum. A temperature increase exhibited a negative and positive influence on O. oeni and L. plantarum, respectively. In general, MLF usually occurs at sub-optimal LAB temperatures (below or around 18 C). Spano and Massa (2006) investigated the molecular and physiological bases of the stress response in wine LAB. In contrast to O. oeni, confined only to a few fermented beverages such as wine and cider, L. plantarum is a versatile bacterium that is found in a variety of ecological niches, ranging from vegetable and plant fermentations to the human gastrointestinal tract. Three small heat shock proteins and three highly homologous cold shock proteins were recently characterised in L. plantarum. From the large set of genes involved in stress response, they consider L. plantarum suited to tolerate or escape several environmental fluctuations. As part of the cold adaptive response, microorganisms have developed a transient adaptive response, termed the cold shock response. During this, a number of cold-induced proteins are synthesised in order to maintain both membrane fluidity, by increasing the proportion of shorter and/or unsaturated fatty acids in the lipids, and transcription and translation needed for cellular adaptation to low temperature (Phadtare et al. 2002). The most strongly induced proteins include a family of closely related low molecular weight proteins termed cold shock proteins (Csp). Three csp genes, recently identified in L. plantarum strain NC8 (Derré et al. 1999), may allow L. plantarum to become a predominant population during MLF. Sulphur Dioxide Sulphur dioxide is another factor which plays an essential role in the growth of LAB and development of MLF. This component is found in wine with variable concentrations according to the winemaking conditions and the yeast strain responsible for alcoholic fermentation. Sulphur dioxide is added to wines to inhibit the

882 Food Bioprocess Technol (2011) 4:876 906 growth of undesirable microorganisms and for its antioxidant effect. Sulphur dioxide in its free, as well as in its bound form with aldehydes and ketones, is a potent inhibitor of LAB. Three liberated forms of SO 2 are present in wine: molecular SO 2, bisulphite (HSO 3 ) and sulphite (SO 2 3 ). The predominant form varies according to the wine ph. The molecular SO 2 is responsible for the antimicrobial effect. It can freely diffuse to the inside of LAB and react with the cell constituents. According to Chang et al. (1997), it interferes with protein bridges and associates with co-enzymes and vitamins, which finally causes cell death. The sensitivity of LAB to SO 2 varies; generally, concentrations of 50 to 100 mg/l bound SO 2 and 1 10 mg/l free SO 2 can inhibit growth of LAB and MLF. Henick-Kling (1993) reported the malolactic activity of the cell being equally sensitive to SO 2 ; 20 mg/l bound SO 2 reduces malolactic activity by 13%, 50 mg/l reduces it by 50% and 100 mg/l inhibits malolactic activity completely. At low ph, SO 2 will be more inhibitory due to the higher percentage of active molecular SO 2. Delfini and Morsiani (1992) investigated the SO 2 resistance of malolactic strains of O. oeni and Lactobacillus spp. isolated from wines. The experiments had been performed in a synthetic medium at different ph values and in a buffered solution at ph 3.2. The results showed that at ph 4.0 the strains of O. oeni were more sensitive than the strains of Lactobacillus. AtpH3.5,both species showed similar behaviours. A ph-dependent factor was identified, which mediates the inhibition of molecular SO 2 (H 2 SO 3 )andtheso 2 3 ions in a different way in Lactobacillus and Oenococcus bacteria. The consequence is a higher sensitivity of Oenococcus to molecular SO 2 and a stronger inhibition effect of the sulphite (SO 2 3 )ionon Lactobacillus. To explain the different resistant mechanisms of O. oeni and Lactobacillus spp., the authors postulated that one of the resistance mechanisms may involve the existence of a sulphite transport system through the cell wall, which may be affected by the ph. Since the wine ph is usually under 3.5, the importance of SO 2 3 is of little significance in oenology. Both species could remain inactive for a long time without losing the capacity for growth even in the presence of large amounts of SO 2. Strains grown in the presence of different concentrations of SO 2 in nutrient broth had been reinoculated into a medium containing 10 mg/l SO 2 more than what they generally would be able to tolerate. Thus, the resistance level of LAB to SO 2 couldbeimprovedby adaptation. Yeasts Numerous studies have investigated the relationship between yeast and the MLF process. The relationship can be stimulating, inhibiting or have no effect, dependingontheyeast bacterial pairing. The negative impact of yeast on LAB can be due to the production of certain metabolites such as ethanol, SO 2, medium chain fatty acids and antibacterial proteins. Yeast can positively affect LAB growth and MLF by the production of mannoproteins and the release of essential nutrients via yeast autolysis (for review see Alexandre et al. 2004). Antimicrobial Metabolites Besides ethanol, SO 2 is a dominating factor in the inhibition of MLF. Henick-Kling and Park (1994) demonstrated the inhibition of malolactic starter cultures by active growing yeasts due to the production of high levels of SO 2 during the early stage of alcoholic fermentation. The concentration in wine is a result of adding SO 2 to the must and the production of SO 2 by S. cerevisiae. Although most yeast strains produce less than 30 mg/l of SO 2, some release more than 100 mg/l (Rankine and Pocock 1969; Eschenbruch 1974; Dott et al. 1976). Three groups of yeasts can be distinguished according to their level of SO 2 production: low (<40 mg/l), medium (40 70 mg/l) and high (>70 mg/l). High SO 2 producing yeast strains can delay or not allow the beginning of MLF. King and Beelman (1986) suggested that the growth of O. oeni during alcoholic fermentation may be delayed by the production of toxic compounds by yeasts other than ethanol and SO 2. Fleet et al. (1984) reported that yeast can also produce toxic compounds inhibitory to both the yeast and the malolactic bacteria. These compounds limit the bacterial growth and can considerably reduce the capacity of the bacteria to metabolise malic acid. These toxic compounds were demonstrated to be medium chain fatty acids, primarily decanoic acid. Shortly after Fleet, octanoic acid and decanoic acid were reported to be inhibitory to malolactic bacteria (Lonvaud-Funel et al. 1988). Edwards and Beelman (1987) found decanoic acid to have potential use as a natural wine preservative to prevent growth of malolactic bacteria in wine, although higher concentrations than normally found in wine may be needed for LAB inhibition. Yeast ghosts were observed to facilitate more rapid and predictable MLF in wines as well as reducing bacterial antagonism by growing yeast. This permits the more efficient production of mix cultures of yeast and malolactic bacteria. Dick et al. (1992) and Comitini et al. (2005) showed the existence of antibacterial S. cerevisiae products of a proteinaceous nature. In 2007, Osborne and Charles identified an inhibiting yeast peptide of 5.9 kda, which acts in synergy with SO 2 present in high concentrations (Osborne and Charles 2007). More recently, Nehme et al. (2010) described two MLF-inhibiting yeast peptides between 5 and 10 kda. Certain aromatic

Food Bioprocess Technol (2011) 4:876 906 883 compounds, such as β-phenylethanol or succinic acid have also been described as MLF inhibitors (Lonvaud-Funel et al. 1988; CaridiandCorte1997). Amino Acid Deficiency The utilisation of nutrients by S. cerevisiae can lead to deficiencies that inhibit MLF. Slow growing yeast with high nutrient requirements showed a strong negative effect on the growth of LAB strains. More recently, most commercial active dry yeast strains have been categorised with regard to their nitrogen demand and nitrogen needs, varying from extremely low to extremely high (Julien et al. 2000). Free amino acids and small peptides are the most frequent sources of nitrogen for LAB growth. Terrade and Mira de Orduña (2009) investigated the essential nutrient requirements of wine-related strains from the genera Oenococcus and Lactobacillus using the single omission technique with a suitable chemically defined medium. The essential bacterial nutrient requirements were found to be strain specific. Ten compounds were essential for all wine LAB tested: carbon and phosphate sources, manganese, as well as several amino acids. The two O. oeni strains revealed a larger number of auxotrophies (18 and 21) and had a higher degree of nutritional similarity (87.5%) (defined as the percentage of common requirements per maximum total requirements) compared to the two Lactobacillus species that had a nutritional similarity of 78.6%, but only 11 and 14 auxotrophies. Both Lactobacillus spp. were auxotroph for riboflavin, which was not needed by the O. oeni strains. This shows that O. oeni is more sensitive to amino acid deficiency whereas Lactobacillus spp. depend more on vitamin supply. Stimulation of Malolactic Bacteria by Wine Yeasts The concentrations of various nitrogen compounds in wine vary according to the cultivar, the grape s maturity, nitrogen fertilisation of the vine and the yeast strain. The autolytic activity of yeast at the end of alcoholic fermentation modifies the concentration of amino acids, peptides and proteins in the wine. Glucans and mannoproteins are released. The levels depend on the yeast strain (Rosi et al. 1999; Escot et al. 2001) and winemaking practices. Mannoproteins can stimulate the malolactic bacteria through two mechanisms: adsorption of medium chain fatty acids that detoxifies the medium or enzymatic hydrolysis by LAB can provide a source of nitrogen. Released vitamins, nucleotides and long chain fatty acids could also encourage the growth of LAB. Most studies on yeast and LAB interactions had been done with O. oeni strains. A recent study (Fumi et al. 2010) investigated the compatibility of a L. plantarum strain with various wine yeast strains using co-inoculation (inoculation of the bacteria 24 h after the yeast) compared to sequential inoculation after alcoholic fermentation in high ph conditions. Yeast strains compatible with O. oeni starter cultures were also shown to be compatible with L. plantarum. Phenolics Recent research has shown that certain grape tannins can have a negative influence on malolactic bacteria and consequently on the course of MLF. In fact, some research has indicated that certain red cultivars, such as Merlot, can have great difficulty undergoing successful MLF (Vivas et al. 2000; Lonvaud-Funel 2001a). The latest results indicate that phenolic acids influence the growth of certain LAB strains in laboratory growth media. The effect on growth can be either positive or negative, depending on the LAB species, the specific phenolic acid used and its concentration. Caffeic acid at 50 and 100 mg/l elicited a positive effect on bacterial growth and degradation of malic acid. On the other hand, ferulic acid can affect bacterial growth and malic acid degradation detrimentally, but is strain dependent. The inhibitory effect of ρ-coumaric acid was the greatest and increased with concentration. These results agree, in essence, with those reported previously (Kunkee 1967). Bossi et al. (2007) analysed the effect of tannic acid on L. hilgardii, a wine spoilage bacteria, by a proteomic approach. The growth tests revealed the negative effect of tannic acid on L. hilgardii in culture medium related to the concentrations used. In particular, within oenological tannic acid concentrations (100 and 250 mg/l), the reduced inhibitory effects could be correlated to the survival and growth conditions for L. hilgardii occurring after MLF when wine can be in contact with wood. A significant growth inhibition had been observed at higher concentrations of tannic acid (500 and 1,000 mg/l). The inhibition of the cells growing in the presence of tannic acid was explained by the interference of tannins on cell protein expression. Figueiredo et al. (2008) found O. oeni in general to be more sensitive than L. hilgardii to the phenolic compounds studied. L. hilgardii was only inhibited by sinapaldehyde and coniferaldehyde. Amongst the flavonoids, quercetin and kaempferol exerted an inhibitory effect, especially on O. oeni. Condensed tannins (particularly tetramers and pentamers) were found to strongly affect cell viability, especially in the case of O. oeni. Campos et al. (2009a) showed that important oenological characteristics of wine LAB, such as the malolactic activity and the production of volatile organic acids, were differently affected by the presence of phenolic acids, depending on the bacterial species or strain. Despite the strong inhibitory effect of most tested phenolic acids on the growth of O. oeni, the malolactic activity of this strain had not been considerably affected

884 Food Bioprocess Technol (2011) 4:876 906 by these compounds. While less affected in its growth, the capacity of L. hilgardii to degrade malic acid was clearly diminished. This group also studied the effect of phenolic acids on the cell membrane of wine LAB (Campos et al. 2009b). Generally, phenolic acids increased the cell membrane permeability in LAB from wine. The different effects of phenolic acids on membrane permeability have been related to differences in their structure and lipophilic character. Recently, Curiel et al. (2010) reported a positive effect of quercetin on the fermentation capacity of L. plantarum. Quercetin was not catabolised by L. plantarum, so the antioxidant properties of the flavonol were protected against degradation while the bacterium improved its growth performance. Other researchers studied the contribution of MLF to the changes in the non-anthocyanin polyphenolic composition of red wine (Hernández et al. 2007). Results showed different malolactic behaviours in relation to wine phenolic compositions for O. oeni and L. plantarum. Diversity was found within each group. The hydroxycinnamic acids and their derivatives, the flavonols and their glycosides, the flavanol monomers and oligomers and trans-resveratrol and its glucoside were the main compounds modified by the different LAB. Studies of Landete et al. (2007b) indicated that L. plantarum is able to grow in the presence of high concentrations of some wine phenolic compounds. Of the ten compounds analysed, only the hydroxycinnamic acids, gallic acid and methyl gallate were metabolised by the L. plantarum strains studied. They propose the use of L. plantarum strains to obtain high-added-value antioxidants from the degradation of phenolic compounds found in wine waste. Studying the effect of phenolic compounds on the putrescine production by L. hilgardii, Alberto et al. (2007) demonstrated a decrease in putrescine formation in the presence of phenolic compounds, proposing a natural way of diminishing the formation of this undesirable compound. Metabolism Over the past few years, it was shown that the role LAB play in wine is much more than deacidification. Whilst wine consists of a large amount of metabolisable compounds, LAB have a very complex and diverse metabolism which can induce a range of compositional changes that may affect the quality of the final product positively or negatively. Carbohydrate Metabolism Wine contains various monosaccharides, mainly pentoses and hexoses as well as dissacharides. The major sugars in wine are represented by arabinose, glucose, fructose and trehalose. There are clear species and strain differences in sugar metabolism with glucose and trehalose being the preferred carbon sources (Liu 2002). The sugars left over by the yeast after alcoholic fermentation are available for fermentation by LAB (Lonvaud-Funel 1999). Oxidation of these sugars is the primary energy-producing pathway and this energy is essential for LAB growth. The amount of energy and the final products produced during sugar metabolism is dependent on environmental factors and the type of sugars being fermented. The sugar transport systems in LAB are specific with regards to the sugars being transported and adenosine triphosphate (ATP) dependent and they activate complex enzymatic systems (Ribéreau- Gayon et al. 2006). LAB can be divided into three major metabolic groups based on their carbohydrate metabolism. Hexose sugars are transported into the cell as free sugar molecules where they are phosphorylated by ATP-dependent glucokinase before entering either the homofermentative or heterofermentative pathway. The homofermentative fermentation pathway, the Embden Meyerhof Parnas (EMP) pathway, results in the fermentation of glucose to produce lactic acid and ATP (Fig. 1a) (Hornsey2007). The first phase of the reaction is glycolysis whereby glucose is converted to pyruvate and the second part is the conversion of pyruvate to produce lactic acid (Ribéreau- Gayon et al. 2006). LAB that are obligatory homofermentative include Pediococcus species and some Lactobacillus species (Fugelsang and Edwards 1997). LAB in this group possess aldolase activity while lacking the presence of phophoketolase (Hornsey 2007). The heterofermentative pathway, also known as the pentose phosphate pathway, the phosphoketolase pathway or the 6-phosphogluconate/phosphoketolase pathway, results in the formation of other end products besides lactate, including acetate and ethanol (Fig. 1b). LAB can be either obligatory heterofermentative or facultative heterofermentative. Obligatory heterofermentative LAB species include Leuconostoc species, O. oeni such as L. brevis and L. hilgardii. These species all possess phosphoketolase activity. The facultative heterofermentors are homofermentative for hexose and heterofermentative for pentose and as example L. casei and L. plantarum (Fugelsang and Edwards 1997; Liu 2002; Hornsey 2007; Ribéreau-Gayon et al. 2006). Species in this group possess both the aldolase and phosphoketolase enzymes. Pentose sugars enter the cell via specific permeases. The pentose sugars are phosphorylated and converted to phosphate derivatives, which in turn are metabolised by the bottom half of the pentose phosphate pathway to produce acetate and lactate as the end products (Fig. 1c) (Hornsey 2007). Both the heterofermentative and homo-

Food Bioprocess Technol (2011) 4:876 906 885 Fig. 1 The three possible carbohydrate fermentation pathways in wine LAB. The EMP (homofermentative) pathway for the fermentation of hexose sugars (a) (compiled from Hornsey 2007; Ribéreau-Gayon et al. 2006; Fugelsang and Edwards 1997). The heterofermentative pathway (b) (adapted from Fugelsang and Edwards 1997) and the fermentation pathway for pentose sugars (c) (adapted from Ribéreau-Gayon et al. 2006) A EMP pathway glucose-6-phosphate fructose-6-phosphate phosphorylation glucokinase fructose-1,6-diphosphate glucose fructose diphosphate aldolase glyceraldehyde-3- dihydroxyacetone phosphate phosphate substrate level phosphorylation pyruvate reduction lactate dehydrogenase LACTATE Pentose phosphate pathway glucose glucose-6-phosphate gluconate-6-phosphate ribulose-5-phosphate xylulose-5-phosphate phosphoketolase ACETATE glyceraldehyde-3-phosphate acetyl-phosphate B acetyl-coa pyruvate acetaldehyde LACTATE ETHANOL Pentose fermentation pathway ribose ribose-5-phosphate xylulose-5-phosphate acetyl phosphate glyceraldehyde-3-phosphate ACETATE pyruvate C LACTATE fermentative LAB metabolise pentoses according to this metabolic pathway. A further product of fructose metabolism is mannitol, a six-carbon sugar alcohol described as being viscous and sweet with an irritating finish (Bartowsky 2009). Fructose can act as an electron acceptor, thereby being reduced to mannitol (Liu 2002). Homofermentative bacteria produce very little or no mannitol via the reduction of fructose-6- phosphate or mannitol-6-phosphate by mannitol phosphate dehydrogenase activity. Contrary to this, heterofermentative LAB produce mannitol in large amounts by the reduction of fructose to mannitol via mannitol dehydrogenase activity

886 Food Bioprocess Technol (2011) 4:876 906 (Wisselink et al. 2002). The acetyl, which is then formed during hexose metabolism, is converted to acetate instead of being reduced to ethanol and ATP is generated (Liu 2002). Mannitol is considered a bacterial spoilage product (Liu 2002) and mannitol-tainted wines are described as complex, usually accompanied by high levels of acetic acid, d-lactic acid, n-propanol and 2-butanol. Mannitol wines can also have a slimy texture, vinegar-estery aroma and a slightly sweet taste (Bartowsky 2009). It is important to understand the metabolic processes associated with the different species of LAB and the main products that are formed as a result. This will aid the winemaker in making informed decisions regarding the process of MLF. Organic Acid Metabolism The major organic acids in wine grapes are tartaric acid, malic acid and citric acid. Lactic acid bacteria are able to metabolise some organic acids which can affect the quality of the end product by producing important MLF aroma compounds or negatively by off-flavours. Malic Acid Several wine LAB strains, belonging to the genera Oenococcus, Leuconostoc, Lactobacillus and Pediococcus, have the ability to decarboxylate L-malic acid into L-lactic acid and carbon dioxide. This reaction usually occurs during the secondary MLF in wine and is catalysed by the malolactic enzyme in the presence of co-factors NAD + and Mg 2+ (Wibowo et al. 1985; Lonvaud-Funel 1999). It was shown by Mtshali et al. (2010) that the mlea gene is different between the lactobacilli and oenococci. The transformation can affect the acidity, microbial stability and sensory complexity of wine and thus have the potential to improve or worsen the wine quality. The principal value of the decarboxylation reaction is the biological deacidification of wine by the conversion of a dicarboxylic acid to a monocarboxylic acid. This usually results in an average reduction in total acidity of 1 to 3 g/l and an increase in ph of 0.1 to 0.3 units (Margalit 1997). In cool climate regions, which usually result in high acid or low ph wines, deacidification is recommended for the production of acid-balanced wines. In warmer climate regions, which usually result in low acid or high ph wines, deacidification could have a negative impact on the wine quality. The reduction in wine acidity could result in unwanted flat wines that lack sufficient acidity. In addition, an increase in ph could enhance the risk of survival and growth of spoilage organisms and could also cause a loss of red colour intensity in red wines (Volschenk et al. 2006). Although L-malic acid is not sufficient as the sole energy source for growth of LAB, it stimulates their growth and serves as a relatively good nutritional resource (Pilone and Kunkee 1976). Thus, by removing malic acid from the wine matrix, the risk of bacterial spoilage is minimised. Furthermore, the main product of this reaction, lactic acid, is a registered antibacterial agent and could have the potential to inhibit growth of unwanted organisms. The degradation of malic acid also contributes to wine aroma. As the strong and sharp green taste of malic acid is replaced by the less aggressive and milder taste of lactic acid, an improved and softer mouthfeel can be expected (Lonvaud-Funel 1999). Citric Acid Citric acid in wine can be degraded by all heterofermentative cocco-bacilli (Leuconostoc and Oenococcus) and facultative heterofermentative lactobacilli (L. plantarum and L. casei) (Lonvaud-Funel 1999). The metabolic pathway for citric acid metabolism has been explained in O. oeni (Fig. 2) (Ramos et al. 1995). The metabolites produced during citrate metabolism are acetic acid, lactic acid, acetoin, 2,3-butanediol, diacetyl as well as aspartic acid. Most of these products are of sensorial importance in wine. Following uptake, citric acid is first cleaved to oxaloacetate and acetic acid. Oxaloacetate is subsequently converted to aspartic acid and pyruvic acid. Results from a study by Ramos et al. (1995) showed that aspartic acid, a precursor for other amino acids (e.g. asparagine, methionine and threonine), was a product of citrate metabolism that was not excreted to the extracellular medium and could be used to synthesise amino acids. Pyruvic acid can be metabolised to lactic acid, acetic acid and α-acetolactic acid. The fate of pyruvate depends on conditions such as ph, aeration and carbohydrate availability (Starrenburg and Hugenholtz 1991). α-acetolactic acid can be decarboxylated to acetoin, which accumulates in the medium or could be further reduced to 2,3-butanediol. Diacetyl is produced by a spontaneous decarboxylation of α-acetolactic acid in the presence of oxygen and/or low ph (Ramos et al. 1995). The most important significance of citric acid metabolism is the production of diacetyl. Diacetyl is best known as the compound responsible for the characteristic buttery flavour note, one of the most distinct flavour changes in wine during MLF (Martineau et al. 1995a). At concentrations of 1 to 4 mg/l, diacetyl is considered to contribute to wine aroma, but slightly higher concentrations (between 5 and 7 mg/l or higher) give an undesirable rancid butter-like flavour (Davis et al. 1985). Like other carbonyl compounds, diacetyl is an unstable end product and can be further reduced to acetoin and 2,3-butanediol by yeast and