Yeast Yeast 2015; 32: 123–143. Published online 16 December 2014 in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/yea.3058
Special Issue Article
Process engineering for bioflavour production with metabolically active yeasts – a mini-review Magnus Carlquist1, Brian Gibson2, Yonca Karagul Yuceer3, Adamantini Paraskevopoulou4, Mari Sandell5, Angel I. Angelov6, Velitchka Gotcheva6, Angel D. Angelov6, Marlene Etschmann7, Gustavo M. de Billerbeck8,9,10,11 and Gunnar Lidén12* 1
Division of Applied Microbiology, Department of Chemistry, Lund University, Sweden VTT Technical Research Centre of Finland, Espoo, Finland 3 Department of Food Engineering, Faculty of Engineering – Architecture, Canakkale Onsekiz Mart University, Terzioglu Campus, Canakkale, Turkey 4 Laboratory of Food Chemistry and Technology, Department of Chemistry, Aristotle University of Thessaloniki, Greece 5 University of Turku, Functional Foods Forum, Turku, Finland 6 Department of Biotechnology, University of Food Technologies, Plovdiv, Bulgaria 7 DECHEMA Research Institute, Frankfurt am Main, Germany 8 Université de Toulouse, INSA, UPS, INP, LISBP, Toulouse, France 9 INRA, UMR792 Ingénierie des Systèmes Biologiques et des Procédés, Toulouse, France 10 CNRS, UMR5504, Toulouse, France 11 INP-ENSAT, Castanet-Tolosan, France 12 Department of Chemical Engineering, Lund University, Sweden 2
*Correspondence to: G. Lidén, Department of Chemical Engineering, Lund University, PO Box 124, SE-22100 Lund, Sweden. E-mail:
[email protected]
Received: 30 June 2014 Accepted: 1 September 2014
Abstract Flavours are biologically active molecules of large commercial interest in the food, cosmetics, detergent and pharmaceutical industries. The production of flavours can take place by either extraction from plant materials, chemical synthesis, biological conversion of precursor molecules or de novo biosynthesis. The latter alternatives are gaining importance through the rapidly growing fields of systems biology and metabolic engineering, giving efficient production hosts for the so-called ’bioflavours’, which are natural flavour and/or fragrance compounds obtained with cell factories or enzymatic systems. Yeasts are potential production hosts for bioflavours. In this mini-review, we give an overview of bioflavour production in yeasts from the process-engineering perspective. Two specific examples, production of 2-phenylethanol and vanillin, are used to illustrate the process challenges and strategies used. Copyright © 2014 John Wiley & Sons, Ltd.
Purpose and scope of this review Flavours and fragrances are products of widespread use in food, detergents, cosmetics and pharmaceuticals. The world market was estimated to be close to $24 billion in 2013 (www.leffingwell. com), so the economic importance of these compounds is quite significant. The concept of flavour is complex and involves most of our senses (Barham et al., 2010). However, the central components most often discussed are taste, which is sensed by receptors on the Copyright © 2014 John Wiley & Sons, Ltd.
tongue distinguishing salt, sweet, bitter, sour and umami, and smell, which is detected by sometimes amazingly sensitive receptors in the olfactory system in the nose. The chemical diversity in flavour composition is quite large, but in order to generate a smell, the compound must be sufficiently volatile that it can reach the sensory system in the upper part of the nose (Buck and Axel, 1991; Lundström et al., 2011). Chemical synthesis and extraction from plant cells are the most common procedures for producing flavour compounds. Extractiondependent production has disadvantages, such as
M. Carlquist et al.
124
seasonal variation, risk of plant diseases, stability of the compound and trade restrictions. Chemical synthesis, on the other hand, will give compounds that, according to EU regulation (EC 1334/2008), will be termed ’flavouring substances’. The term ’nature-identical’, which was used in EC Directive 88/388 as a distinction from ’artificial’, no longer applies. Since consumers typically favour ’natural’ compounds, the price levels are substantially higher for these (Schrader, 2007). The European COST Action Bioflavour (Yeast Flavour Production – New Biocatalysts and Novel Molecular Mechanisms) was initiated to promote the development of biotechnological and eco-efficient production of natural flavour compounds. This mini-review aims to provide a background to bioflavour production in yeasts from an applied perspective, including flavour analysis and sensory evaluation aspects. Production of two specific flavour compounds, 2-phenylethanol (2-PE) and vanillin, will be used to exemplify process challenges and possible solutions from a processengineering perspective.
The flavour chemistry of yeasts Yeasts are microbes with large synthetic capacity that are able to convert simple carbohydrates and nitrogen sources into many complex molecules, including many flavour compounds, via enzymecatalysed reactions. Control of yeast-derived flavour compounds as part of fermented beverages has been of interest to producers of beer, wine, sake and other fermented beverages, as long as these processes have existed. Tailoring of process conditions to develop specific organoleptic characteristics occurred over time through artisanal observation and trial-and-error approaches. Latterly, more accurate analytical methods and an improved understanding of the biochemical mechanisms that determine flavour production have enabled more rapid development of process conditions, via hypothesis-driven research, to achieve desired flavour characteristics for particular products. A flavour compound can thus be an integrated part of fermented liquors or food, but it can also be a dedicated product in itself. Flavour and fragrance compounds can be produced de novo, i.e. simple sugars and nutrients can be metabolized into Copyright © 2014 John Wiley & Sons, Ltd.
flavour and fragrance compounds via biochemical pathways. Alternatively, bioconversion of a precursor in a single-step (or a few steps) enzymecatalysed process may also occur. Flavour and fragrance compounds that can be formed by yeasts include ketones, aldehydes, alcohols, carboxylic acids, esters, lactones and terpenoids (Figure 1).
Flavours as an integrated part of a fermented product The biological functions of flavours in food are manifold. They may, for example, attract animals to improve seed dispersal and propagation, or warn that the food is spoilt. Flavour compounds play an important role in consumer preference and acceptance of food products, and consumers typically prefer natural over synthetic flavour compounds. Odour- and taste-stimulating components can be classified in the following groups (Reineccius, 1999): • Volatiles and non-volatiles formed during normal plant metabolism and remaining in the plant after harvest, e.g. essential oils, fruits and vegetable flavours. • Flavour compounds produced by enzymatic reactions. • Flavours developed by microorganisms and fermentation (wines and dairy products). • Flavour compounds resulting from processing (heat treatment, cooking, etc.). Fermentation processes provide wide varieties of flavour compounds in, for example, cheese, yogurt, kefir, beers, wines, soy sauce, sausages, sauerkraut, kimchi and fermented fish products. Flavour may be generated by biochemical reactions of microbial metabolism or by the activities of residual enzymes after microorganisms have lysed (Reineccius, 1999). The use of metabolically active yeasts for integrated flavour production is exemplified below by the production of alcoholic beverages, e.g. beer and wine.
Beverages Yeast is historically linked to the production of fermented beverages. One aspect is, of course, that Yeast 2015; 32: 123–143. DOI: 10.1002/yea
Bioflavour production with yeasts – a mini-review
125
Figure 1. Flavour and fragrance compounds produced by yeast de novo from sugars, or via biotransformation of precursors (source: Wang et al., 2011b: 404–407)
the fermentations give rise to ethanol in the beverage – thereby preserving it. However, equally important for the enjoyment of the beverage are the flavour compounds produced during the fermentation. In fermented beverages, fusel alcohols formed through the Ehrlich pathway, schematically shown in Figure 2 (see e.g. Hazelwood et al., 2008), are central flavour compounds. While all of the amino acids necessary for the formation of fusel alcohols can be found in brewery worts and grape musts, isoleucine, leucine, phenylalanine and valine may be considered to be the most important, due to the fact that their Ehrlich pathway-derived products are often found at concentrations close to flavour thresholds, and small changes may therefore have a significant impact on flavour perception (Christoph and Bauer-Christoph, 2007; Meilgaard, 1975). An intimate relationship exists between amino acid availability and yeast higher alcohol production during fermentation (Schulthess and Ettlinger, 1978; Äyräpää, 1971). Yeast production of phenyl
ethanol and other fusel alcohols is determined not only by the substrate concentration but also by the various factors that influence uptake and assimilation of the substrate. Utilization may, for example, be affected by the assimilable nitrogen matrix, the physical conditions, such as temperature, that may affect uptake and the affinity of transporters for the particular substrate. All of these factors may influence flavour differentially and distinct differences are observed, depending on the strain or species involved. When nitrogen supplementation is considered as an option for control of the concentration of yeast volatiles, the impact of the nitrogen matrix must be taken into account. Hernàndez-Orte et al. (2005) noted that an increase in available nitrogen, regardless of whether this was due to higher amino acid or ammonium concentrations, resulted in lower levels of phenylethanol (as well as methionol and isoamylalcohol). The results suggest that a higher level of assimilable nitrogen may reduce the uptake
Figure 2. Ehrlich pathway overview. The R group is different, depending on which amino acid is converted (source: Adler et al., 2011: 285–292) Copyright © 2014 John Wiley & Sons, Ltd.
Yeast 2015; 32: 123–143. DOI: 10.1002/yea
126
of certain individual amino acids, presumably due to competition for transmembrane transport. A similar phenomenon is known to affect the levels of the vicinal diketone diacetyl during brewery fermentation. The precursor of diacetyl is produced by yeasts as a by-product of valine metabolism, and any increase in amino acid concentration will reduce uptake of valine and have a direct influence on the diacetyl concentration of beer (and a corresponding influence on process efficiency) (Krogerus and Gibson, 2013). While specific amino acids may be added to fermentation media to tailor the final aroma and flavour profile of the medium, in practice this may not be feasible, due to the expense of purified amino acids. Other approaches that alter the amino acid profile may be equally effective without a major effect on process cost. Process changes such as extended must maceration times can be used to raise the concentration and alter the relative proportions of amino acids available to yeasts during wine fermentation (Guttart et al., 1997), a result which may explain the improved aroma quality of wines produced after extended maceration (Defranoux and Joseph, 1992). The amino acid profile of brewery worts can be influenced through selective hydrolysis of the available barley proteins. Wort contains, in particular, high levels of the glutelin and prolamin protein classes (Steiner et al., 2011). The different protein types are known to contain different concentrations of individual amino acids with, for example, hordein, a prolamin protein, characterized by high levels of proline and glutamic acid and relatively low levels of lysine (Steiner et al., 2011). Different proteases and peptidases are known to differentially hydrolyse barley proteins, with cysteine proteases particularly effective against the hordein proteins (Jones and Budde, 2005). A commercial enzyme preparation has been shown to be particularly effective at liberating proline and glutamic acid from spent barley grain, suggesting that the enzyme was specifically targeting the hordein proteins present (Treimo et al., 2008). In this case, if the amino acids, glutamic acid and proline were made available, it would be expected that the former, being a preferred source of nitrogen, would promote yeast growth but possibly reduce the catabolism of other amino acids via the Ehrlich pathway. By contrast, another commercial protease was found to specifically liberate branched-chain amino acids Copyright © 2014 John Wiley & Sons, Ltd.
M. Carlquist et al.
(Piddocke et al., 2011), with potential for increased synthesis of higher alcohols during fermentation. Different protein-solubilizing enzymes operate maximally under different conditions (Swanston et al., 2014). Where multiple enzymes act in concert, such as in the case of brewery malting or mashing, the amino acid profile will be influenced by process conditions, such as mashing temperature or pH. Igyor et al. (2001), for example, found that changing the malting and mashing temperature significantly influenced wort nitrogen availability and resulted in beers with altered concentrations of important volatile compounds, including ethyl acetate, isobutanol and isoamyl alcohol (Igyor et al., 2001), thus providing further evidence that proteins in brewing may be manipulated for the generation of free amino nitrogen fractions with specific amino acid profiles and potentially different functional properties. This feature may be exploited to alter the amino acid profile and hence the flavour of the beer. While substrate availability has an important influence on concentrations of higher alcohols in fermented beverages, ultimately it is the uptake and assimilation of these substrates that determines higher alcohol concentration. This fact has been exploited to manipulate the levels of 2-phenylethanol and other yeast-derived com pound during fermentation. Evolutionary engineering of yeasts has been used to alter phenylalanine metabolism and hence product level. Fukuda and colleagues have successfully applied this method to increase the production of 2-phenylethanol and 2-phenylethylacetate by yeast during sake production (Akita et al., 1990; Fukuda et al., 1990a, 1990b). Yeasts resistant to toxic analogues of phenylalanine, such as o-fluoro-D,L-phenylalanine or p-fluoro-D,Lphenylalanine, showed changes in the action of phenylalanine-dependent DAHP synthase, the first step in production of aromatic amino acids. Concentrations of 2-phenylethanol produced could be increased by as much as 40-fold in adapted mutants. A similar process has been applied to brewing yeast strains for increasing 2-phenylethanol concentrations in lager beer (Lee et al., 1995). In addition to exploiting the potential genetic diversity of Saccharomyces yeast strains, one may also consider the feasibility of utilizing non-Saccharomyces yeasts for manipulation of Yeast 2015; 32: 123–143. DOI: 10.1002/yea
Bioflavour production with yeasts – a mini-review
the concentration of 2-phenylethanol and other important yeast-derived volatiles. Such an approach may be more acceptable in the wine industry, where traditional spontaneous fermentations would be influenced to some extent by the local non-Saccharomyces microflora. Many yeasts, such as Kluyveromyces spp., produce high levels of 2-phenylethanol (Fabre et al., 1998) but are not suitable for use in anaerobic fermentations (Garavaglia et al., 2007). Mixed-culture wine fermentations involving Hanseniaspora guilliermondii or Pichia anomala have been proposed by Rojas et al. (2003) as being suitable for increasing the levels of certain volatiles in wine. In particular, H. guilliermondii was shown to increase the levels of 2-phenylethyl acetate (although not 2-phenylethanol) in wines. Concentrations of individual volatiles in wine may be controlled by careful selection of non-Saccharomyces yeasts in co-culture fermentations (Moreira et al., 2005, 2008, 2011; Viana et al., 2009). The brewing industry is, however, more conservative than the wine industry in terms of the yeast employed and considerable effort is put into eliminating any possible source of wild yeast contamination. Lager beer is also characterized by a finely-balanced volatile composition which would be detrimentally affected by a disproportionate increase in any one volatile compound. Adaptive evolution or careful selection of brewing yeast strains may therefore be more acceptable than utilization of wild yeasts for flavour control. Any flavour changes that can be made through simple adjustments to process conditions are preferable to major changes in raw material use or production yeast. Disproportionate uptake of phenylalanine has been observed at higher temperatures (Beltran et al., 2007) and this phenomenon may, to some extent, explain the higher concentrations of 2-phenylethanol in wine and beer fermented at higher temperatures (Molina et al., 2007; Saerens et al., 2008).
Bioflavour compounds from substrates in the agricultural industry The agricultural industry generates large volumes of plant material that typically contain important precursors for flavours, or flavour compounds themselves, and are thus an interesting source of substrates for eco-efficient microbial flavour Copyright © 2014 John Wiley & Sons, Ltd.
127
production. Substrates from the agro-industry often have low costs, and the closeness to the food industry in general leads to more easily obtained consumer acceptance as compared to synthetic flavours of petrochemical origin. A number of studies that focus on characterizing product spectra from the bioprocessing of agro-industrial biomaterial by various yeast species have been reported previously. These flavour-mapping studies have often been based on the use of gas chromatography (GC) analysis for the identification and quantification of volatile compounds in the headspace during fermentation of the substrates in litre-scale bioreactors. Christen et al. (2000) characterized volatile compounds produced via Rhizopus-catalysed bioprocessing of agro-industrial solid wastes, such as cassava bagasse, apple pomace, soya bean, amaranth grain and soybean oil. As can be expected, large differences were found in the profiles of volatile compounds, depending on the specific substrate. Acetaldehyde, ethanol, 1-propanol, ethyl acetate, ethyl propionate and 3-methyl butanol were some of the volatiles determined in the samples. The effect of the substrate type and use of precursors on fruit flavour production in solid-state fermentation by Ceratocystis fimbriata was also demonstrated (Christen et al., 1997). Cassava bagasse has also been studied as a substrate for production of flavour compounds by Kluyveromyces marxianus in an aerobic packed-bed column bioreactor (Medeiros et al., 2001). The main volatiles determined by headspace GC analysis were ethyl acetate, ethanol and acetaldehyde. Coffee pulp and husk are generated in large amounts in the food industry. Coffee husk was reported to be a good substrate for aroma production by C. fimbriata (Medeiros et al., 2003); ethyl acetate, ethanol and acetaldehyde were identified as the major volatiles. In a similar study on solid-state fermentation using coffee husk (Soares et al., 2000), acetaldehyde, ethanol, isopropanol, ethyl acetate, ethyl isobutyrate, isobutyl acetate, isoamyl acetate and ethyl-3-hexanoate were identified. Orange and citric pulps are other agro-industrial substrates with high potential for yeast-based flavour production. According to the Statistical Database of the Food and Agriculture Organization of the United Nations (FAOSTAT), world orange production in 2012 was estimated to be over 68 million tons (FAOSTAT, 2012). About 50–60% Yeast 2015; 32: 123–143. DOI: 10.1002/yea
M. Carlquist et al.
128
of the processed fruit turns into citrus peel waste, consisting of peels, seeds and membranes left over after juice extraction (Wilkins et al., 2007). Volatile bio-ester production from orange pulp-containing medium using Saccharomyces cerevisiae was investigated (Mantzouridou and Paraskevopoulou, 2013). The results of this study showed that orange pulp stimulated the de novo biosynthesis of isoamyl acetate, 2-phenylethyl acetate and ethyl esters, including hexanoate, octanoate, decanoate and dodecanoate, by S. cerevisiae. Similar results were also found by using immobilization technology (Lalou et al., 2013). Immobilized cells of S. cerevisiae displayed better growth performance and produced a higher amount of volatile esters of ’fruity’ aroma (2-phenylethyl acetate, ethyl hexanoate, octanoate, decanoate and dodecanoate) than cells in suspension. Rossi et al. (2009) studied fruit aroma production by C. fimbriata, using citric pulp including the addition of other carbon (sugar cane molasses, soya molasses) and nitrogen (soya bran or urea) sources. The highest overall concentration of volatile compounds was found when the citric pulp was supplemented with 50% soya bran, 25%
sugar cane molasses and mineral saline solution. The compounds identified from this investigation included acetaldehyde, ethanol, ethyl acetate, propyl acetate, ethyl isobutyrate, 2-hexanone, 2hexanol and isoamyl acetate.
Flavours as primary products We have previously seen that flavours are naturally produced in many fermented foods, and that plant material can also serve as a substrate for flavour production. Several specific flavour and fragrance compounds have economic importance and are used to increase the sensory qualities of various products, not only food but also cosmetics or perfumes. Yeasts are potential catalysts for the production of many of these compounds, either via de novo biosynthesis from sugars or by biotransformation via one- or multistep whole-cell biocatalysis from specific precursors (Serra et al., 2005). Table 1 summarizes different flavour products derived from wild-type and recombinant yeast fermentation or whole-cell biotransformation. A
Table 1. Flavour compounds from wild-type or recombinant yeasts Product Alcohols 2-Phenylethanol
Substrate
Sensorial description
Etschmann et al., 2003
Fruit
De novo production from glucose with Candida utilis
Schrader, 2007
Ricinoleic acid (cator oil) 10-Hydroxystearic acid
Fruit, peach
Bioconversion with Yarrowia lipolytica
Pagot et al., 1997
Fruit, peach
An et al., 2013
Plant flavours – synthetic yeast Valencene
Batch bioconversion with Waltomyces lipofer
Glucose, galactose
Citrus, orange
Asadollahi et al., 2008
Vanillin
Glucose
Vanilla
p-Hydroxybenzalacetone (intermediate for raspberry ketone)
p-Coumaric acid
Raspberry
Aerabic batch cultivation, defined mineral medium with 20 g/l galactose or glucose Vanillin glucoside was produced to avoid product toxicity Batch bioconversion
Lactones γ-Decalactone γ-Dodecalactone
Rose
Leucine Isoleucine Valine
Fruit Fruit Fruit
Sugar
Reference
Fed-batch cultivation, phenylalanine as sole nitrogen source, ISPR Isolated from fusel oil Isolated from fusel oil Isolated from fusel oil
2-Methylbutanol 3-Methylbutanol 2-Methylpropanol Esters Ethyl acetate
Phenylalanine
Comments
Copyright © 2014 John Wiley & Sons, Ltd.
Schrader, 2007 Schrader, 2007 Schrader, 2007
Brochado et al., 2010 Beekwilder et al., 2007
Yeast 2015; 32: 123–143. DOI: 10.1002/yea
Bioflavour production with yeasts – a mini-review
central pathway for the production of flavour compounds is the Ehrlich pathway (Hazelwood et al., 2008), which, in addition to the improvement of organoleptic properties of alcoholic beverages, as mentioned above, is also exploited for the production of specific flavour compounds. An example of industrial interest is the conversion of L-phenylalanine to 2-phenylethanol, the main aroma compound of rose flowers. This is covered more in detail below (see Process example – 2-phenylethanol). Other important fusel alcohols are 2-methylbutanol, 3-methylbutanol and 2-methylpropanol, which are used as fruit flavour ingredients and can be obtained from fusel oil, a by-product from S. cerevisiae fermentation processes. Oleaginous yeast can be used to convert long-chain fatty acids into lactones that are characterized by their fruity peach flavour (Schrader et al., 2004); for example, γ-decalactone is produced by Yarrowia lipolytica through conversion of ricinoleic acid, which is the main constituent of castor oil. Another example is the production of γ-dodecalactone from 10-hydroxystearic acid by Waltomyces lipofer (An et al., 2013). Production of both γ-deca- and dodecalactones functions via repeated β-oxidation into 4-hydroxydecanoic acid or 4-hydroxydodecanoic acid, followed by subsequent spontaneous lactone formation facilitated in acidic media. S. cerevisiae has also been used as a whole-cell biocatalyst for asymmetrical synthesis of various chiral flavour products (Serra et al., 2005). Most often the high capacity of baker’s yeast to catalyse stereoselective reductions of prochiral carbon–carbon double bonds or carbonyl groups has been exploited. For example, baker’s yeast was efficiently used for the preparation of chiral aerangis lactones via whole-cell bioreduction, followed by spontaneous lactonization (Brenna et al., 2001). The emerging field of synthetic biology allows for the production of a completely new set of yeast-derived flavours. Efficient activity-based and in silico-based enzyme-mining strategies, together with combinatorial and semi-rational cloning approaches to assemble recombinant biochemical pathways, have led to the rapid development of engineered yeast strains that are able to produce a variety of different flavours. The development of an increasingly advanced synthetic biology toolbox will allow for the construction of yeast strains with unprecedented ability to produce both natural and novel flavour molecules. In addition, with the development of new procedures for fast Copyright © 2014 John Wiley & Sons, Ltd.
129
and efficient high-throughput screening for optimal process conditions and for engineered yeast strains that are able to produce metabolites, the time to go from idea to assembling a functional biosynthetic route is decreasing yearly. There are a number of examples where yeast has been genetically engineered to express recombinant enzymes for the biosynthesis of different industrially important flavour compounds. For example, de novo production of p-hydroxybenzaldehyde in recombinant S. cerevisiae overexpressing 4-coumaratecoenzyme A ligase and mutated chalcone synthase from raspberry plant Rubus idaeus has been reported (Beekwilder et al., 2007). Other examples of recombinantly produced flavour compounds in S. cerevisiae are vanillin (Brochado et al., 2010) and valencene (Asadollahi et al., 2008), which are both on the verge of entering the sales market (Hayden, 2014). De novo production of vanillin with engineered yeast and related process challenges are covered in more detail below (see Process example – Vanillin).
Process design General features The overall key numbers in any biotechnological process are: product yield (i.e. moles of product formed/mole of substrate converted); degree of substrate conversion (mole of substrate converted/mole of substrate added); productivity (mole of product formed/mole of biomass/h, or mole of product formed/reactor volume/h) and product titre (g/l). The relative importance of these factors will vary, depending on the value of the product and the cost of the substrate. For bioflavours, the product value can be quite high, meaning that the product yield is less critical (provided that expensive precursors are not used). Therefore, more important are the productivity (translating into capital cost for equipment) and product titres, which translate into operational costs (separation costs). Some specific features that characterize bioflavour production by yeasts are: (a) the products formed are likely to be secondary metabolites; (b) the flavour compound is likely to be inhibitory, or even toxic, to the yeast – sometimes this is true also for precursors in the pathway or even the substrate used in the production; and Yeast 2015; 32: 123–143. DOI: 10.1002/yea
M. Carlquist et al.
130
Table 2. Process engineering targets for improved bioflavour production Problem Low productivity, yield and product titre Substrate toxicity
Product toxicity By-product formation (e.g. ethanol) giving increased toxicity due to synergy Slow yeast growth
Solution Yeast strain engineering Immobilization Cell recirculation Feeding strategies Strain engineering Immobilization ISPR/strain engineering Strain adaptation Use Crabtree-negative yeast Keep glucose levels low
Media composition
(c) titres obtained are generally low, with final concentrations typically
