Improving fructose utilization in

wine yeast using adaptive evolution

Tommaso Liccioli

A thesis submitted for the degree of Doctor of

Philosophy in the School of Agriculture, Food and Wine

Faculty of Sciences

The University of Adelaide

August 2010

i

SUMMARY

Saccharomyces cerevisiae is the most important micro organism involved in the

production of fermented alcoholic beverages such as wine. Despite its fermentative

capacity and production of desirable metabolites, grape juice represents a hostile

environment for yeasts. Sometimes, adverse conditions reduce yeast biomass

formation or catabolic capacity, which may lead to stuck or sluggish fermentation.

These phenomena represent one of the most common problems during the wine

production process and mean that winery throughput is reduced and residual sugar

adds unwanted sweetness in dry wine styles while offering substrates for microbial

spoilage.

The scientific community has always been alert to the problems linked with

fermentation, considering the vital role of this organism during the production

process. For this reason research has focussed on developing a range of techniques for

strain improvement. With the emergence of modern molecular genetics, the new

methodologies of hybridization and genetic engineering have been used to isolate and

create new yeast strains. However, their application in wine microbiology is not

without complications, as genetically modified yeasts are not universally favoured for

commercial use in the food industry.

A recent development is the notion of using the natural capacity of a population of

single celled organisms to adapt themselves to an environment imposing a specific

stress. The technique is termed “adaptive evolution” or “directed evolution”. In

principle the process is simple: when a species is constricted to live and replicate

under stressful conditions for many generations, some cells will present adaptive

characteristics: i.e. “adaptive mutations” and outgrow the starting population. A key

benefit of this technique is that it does not rely on direct manipulation at the level of

DNA, and can be used to reproduce the stress conditions found in nature or in

fermentation tanks. However, adaptive evolution is a technology that needs to be

more fully explored and developed for its possible use in improving wine yeast

strains.

ii

A possible improvement for wine yeasts targets their sugar catabolic capacity. The

different affinity of S. cerevisiae for glucose and fructose is thought to be a cause of

stuck or sluggish fermentations in the winemaking process. The possibility of

obtaining a strain with improved fructose utilization using adaptive evolution is

therefore the topic of this investigation.

This thesis describes work that can be divided into four sections. The first part is the

identification of a candidate strain from a selection of commercially available wine

yeasts. The second part is aimed at evolving the candidate strain under a selective

pressure. The third validates new methods for assessing the populations of candidate

evolved yeast in order to isolate clones that can metabolize fructose more efficiently

compared to the parental strain. The last part is focussed on a deeper investigation and

comparison of a number of potentially evolved candidates with the parent.

To identify a candidate strain for use in the adaptive evolution process, it was

necessary to compare fermentative performances of commercially available strains.

Fermentations for 20 strains were conducted in synthetic media, containing fructose

as sole sugar or else an equivalent concentration of glucose and fructose. Particular

attention was focussed on the rate of fructose consumption relative to glucose, and

thus it was necessary to identify a methodology that was independent of sugar

concentration, overall fermentation rate or duration. As such the value of the area

under the fermentation curves determined by the composite trapezoid rule was utilised

to compare glucose and fructose utilisation and hence define the fructophilicity of

each strain screened. This approach allowed the most suitable candidate strain to be

chosen for the application of adaptive evolution. Accordingly, strain AWRI 796 was

cultured under fermentative conditions that elicited an appropriate selective pressure

over some 350 generations. Samples of the population were collected every 50

generations for characterization of individual clones. The next stage of the project

focussed on the identification of clones which showed improved fructose utilization

compared to the parental strain. To define fermentative performance of a high number

of isolates from the adaptive evolution experiment, it was necessary to develop

screening methodologies. For this purpose fermentations in microtiter plates and

automated colorimetric assays for determination of residual sugar were adopted. From

378 clones examined, four were identified to be faster consumers of fructose relative

to the parent. Patterns of glucose utilisation in these clones were unchanged. The last

iii

stage of the study validated the improved fermentation ability of these novel

phenotypes under winemaking related conditions in fermentations of 20 kg of red

grapes. In these experiments two isolates again showed a significant reduction in the

time required for completion of the fermentation. The results validate the approach

used and the selective pressures applied as a means introducing specific

improvements into wine yeast strains.

iv

TABLE OF CONTENTS

SUMMARY _____________________________________________________ i STATEMENT OF AUTHORSHIP __________________________________ v ACKNOWLEDGEMENTS ________________________________________ vi CHAPTER 1: INTRODUCTION AND LITERATURE REVIEW ___________________ 1

1.1 – Introduction ______________________________________________ 2 1.2 – Yeast and the winemaking process ____________________________ 3 1.3 – Stuck and sluggish fermentations: one of the biggest problems

during the winemaking process ______________________________ 15 1.4 – Improving wine yeast strains ________________________________ 18 1.5 – Aim of the project _________________________________________ 25

CHAPTER 2: CANDIDATE STRAIN SELECTION ___________________________ 27 2.1 – A novel methodology independent of fermentation rate for assessment

of the fructophilic character of wine yeast strains _______________ 30 2.2 – Basis of candidate strain selection ____________________________ 48

CHAPTER 3: ESTABLISHING AN ADAPTIVE EVOLUTION STRATEGY USING CONTINUOUS CULTURE TO GENERATE FRUCTOPHILIC GENOTYPES _ 51

3.1 – Isolation of a representative single clone of AWRI 796 ___________ 52 3.2 – Induced mutagenesis of AWRI 796 ___________________________ 55 3.3 – Defining experimental conditions for adaptive evolution __________ 55

CHAPTER 4: SCREENING OF ISOLATES FROM CONTINUOUS CULTURE TO IDENTIFY FRUCTOPHILIC GENOTYPES _______________________ 63

4.1 – Screening for isolates showing improved fructophilic ability _______ 68 4.2 – Fermentative performances of 19 isolates identified as candidates

for further characterization _________________________________ 75 4.3 – Conclusions _____________________________________________ 75

CHAPTER 5: PHYSIOLOGICAL AND GENETIC CHARACTERIZATION OF THE IDENTIFIED ISOLATES ___________________________________ 78

5.1 – Fermentation performance of 4 isolates in a higher sugar concentration medium ____________________________________ 79

5.2 – Evaluation of the fermentation performance of isolates 9 and 11, the parental strain and two commercially available strains ________ 82

5.3 – Metabolite production during fermentation ____________________ 84 5.4 – DNA fingerprinting of evolved strains ________________________ 86 5.5 – Phenotype stability ________________________________________ 88 5.6 – Grape juice fermentations __________________________________ 90

CHAPTER 6: CONCLUSIONS, DISCUSSION AND FUTURE DIRECTIONS __________ 97 6.1 – Conclusions ______________________________________________ 98 6.2 – Discussion and future directions ______________________________ 98

APPENDIX 1 ____________________________________________________ 101 APPENDIX 2 ____________________________________________________ 108 REFERENCES ___________________________________________________ 110

v

STATEMENT OF AUTHORSHIP

This work contains no material which has been accepted for the award of any other

degree or diploma in any university or other tertiary institution to Tommaso Liccioli

and, to the best of my knowledge and belief, contains no material previously

published or written by another person, except where due reference has been made in

the text.

I give consent to this copy of my thesis, when deposited in the University Library,

being made available for loan and photocopying, subject to the provisions of the

Copyright Act 1968.

I also give permission for the digital version of my thesis to be made available on the

web, via the University’s digital research repository, the Library catalogue, the

Australasian Digital Theses Program (ADTP) and also through web search engines,

unless permission has been granted by the University to restrict access for a period of

time.

Tommaso Liccioli

vi

ACKNOWLEDGMENTS

I would like to firstly thank my supervisors Assoc. Professor Vladimir Jiranek and Dr. Paul J.

Chambers for their guidance, constant encouragement and patience. They were my reference point

during this study and there is no way to repay what I have learnt from them.

I am also very thankful to Frank Schmid (especially when I did not “listen” to him). Every minute

spent together was an occasion to increase my knowledge, personal experience, scientific

approach and, above all, life. I cannot thank him enough for all the good times I had with him.

Simon Schmidt, with his patience, experience and critical view in this field of study, was always

ready with a solution for solving problems during my experiments and suggesting future

directions. It was great talking with him about life and playing music together.

Thank you to Michelle Walker, Jennie Gardner and Paul Grbin, for their experience, knowledge

and support, indispensable people for my academic life. Cristian Varela for his help with the

continuous culture experiments. Colin McBryde, for the work previously done with the adaptive

evolution, from which this study was a continuation.

A special thought goes to Alana Capaldo and Krista Sumby, who started their PhD in the same

period as me and were a great support during this time: it has been great being in the lab with

them.

Thanks go to Angus Forgan, Jenny Bellon, Darek Kutyna, Maurizio Ugliano, Radka Kolouchova,

Caroline Abrahamse, Mariola Kwiatkowski and all of the other members of AWRI and WIC

building, for their support and making my time enjoyable.

I am also very grateful to Michael Brinkley and his family for their help and lovely moments spent

together. Richard (old mate) Freebairn and Peter (uncle P) Start for their patience of sharing

houses together and “introducing” me to Aussie lifestyle. Medyan Ghareeb, for all of the

incredible experiences shared, reciprocal respect and help: thanks dude, and it is not finished here!

Words can never thank my family enough, that fully understood my problems and supported all

my decisions. I could have not been here writing these words without the knowledge of having my

family. Living so far away from each other was only marginally compensated with the satisfaction

of having lived this experience.

I am grateful to Australia’s grape growers and winemakers through their investment body, the

Grape and Wine Research and Development Corporation, with matching funds from the

Australian Government for the financial support.

1

CHAPTER 1

INTRODUCTION AND LITERATURE REVIEW

2

1.1 INTRODUCTION Mankind has used yeasts for many millennia in the modification of food products

(Samuel, 1996). In particular, Saccharomyces cerevisiae is the most well know of

these yeast as it is used in the baking, brewing and winemaking processes (Spencer

and Spencer, 1983). It was not until 1863 that its catalytic capacity in wine

fermentation was described by Louis Pasteur, basing his work on previous

observations of other pioneers (Leeuwenhoek, 1680, Cagniard-Latour, Kützing and

Schwann in the late 1830s – for review see Rose and Harrison, 1969). With such

revolutionary knowledge winemakers could start to control the process of

transformation of grape sugars into alcohol and carbon dioxide, and the influence that

yeasts have on the final product. In fact, in this context, in 1890 Müller-Thurgau

introduced the concept of “inoculating fermentation” with pure cultures of superior,

previously selected yeasts (Kunkee and Amerine, 1970). This innovative practice

changed the wine industry, resulting in increased quality and quantity of wine. Pure

culture wine yeasts gave more consistent, better quality and more satisfactory results

than uncontrolled natural fermentations (Mestre and Mestre, 1946) and for this reason

yeast selection has become a tool appreciated among winemakers.

S. cerevisiae strongly influences the composition and structure of the wine, thereby

making yeast one of the principal factors determining wine sensorial profile. In

addition, the yeast in use and the manner in which it interacts with the grape juice

determines the progress of fermentation and whether stuck or sluggish fermentations

are encountered. Such problem fermentations can be linked to issues such as nutritive

limitations or adverse media composition, but also to the capacity of the yeast to take

up sugars per se or specifically fructose compared to glucose (Berthels et al., 2004;

Berthels et al., 2008; Bisson, 1999; Guillaume et al., 2007), the two most important

sugars in grape juice. Possible solutions to these problems come from two directions:

correcting must composition before and during the winemaking process or improving

the resilience and fermentative capacity of the yeast. To change the chemical

composition of the grape juice presents some difficulties and may adversely influence

the quality of the wine. Therefore, improving the fermentative capacity of the yeast in

the presence of adverse physico-chemical conditions remains a primary target for

wine microbiological research.

3

Natural selection and recombinant techniques are the two approaches used presently

to isolate, improve or create new yeast strains for the wine industry. In isolating yeast

from nature it is possible to identify representatives of a particularly interesting wine

area and to capture the peculiarities of that “terroir”. But it is difficult to improve

determinate characteristics beyond what is found in the originating environment.

Recombinant techniques can overcome these limitations, but some difficulties are

linked to DNA manipulation (i.e. the need for a high level of knowledge, ethical

debates on GMOs, etc), and thus more research is required.

A technique that seeks to capture the cell’s spontaneous modification of its genome

during exposure to or in response to environmental stress is called adaptive (or

directed) evolution. This technique does not involve DNA manipulation and yet

shows some of the great potential of recombinant techniques, thereby offering much

promise as a means to strain optimisation.

The following literature review provides background to yeast sugar metabolism

during winemaking and reviews recent studies which indicate the potential of adaptive

evolution as a new strategy specifically for improving wine yeast. Particular emphasis

is given to the different ability of strains of S. cerevisiae to ferment glucose and

fructose and this introduction also debates the possibility of the use of adaptive

evolution as strategy for improving fructose metabolism and therefore limiting

problems of stuck or sluggish fermentations.

1.2 YEAST AND THE WINEMAKING PROCESS

During winemaking, the most evident phenomena linked with wine yeast is the

metabolism of the sugar present in grape juice. This process, called alcoholic

fermentation, is complex and involves a high number of chemical and enzymatic

reactions (Boulton et al., 1998). The two principal products of the alcoholic

fermentation are ethanol and carbon dioxide (Brock et al., 1994). However, other

substances present in grape juice are able to penetrate the yeast cell membrane and

participate in biochemical reactions producing numerous volatile end-products. These

include long-chain fatty acids, organic nitrogen-containing compounds and sulphur-

containing compounds and others (Boulton et al., 1998). As such, during the different

4

phases of winemaking, the grape matrix is markedly changed and important

influences on the sensory characteristics are detectable in the final product.

This section will be briefly described the main features of the evolution of sensorial

characteristics during winemaking, with particular attention to the influence of yeast.

Then, the metabolic pathways of sugar catabolism by yeast will be discussed.

Fermentation and wine flavours

Wine flavour can be described as the overall sensory impression of wine (Robinson,

1994). However the largest contribution to what is described as flavour is in fact wine

aroma (“associated with odorous, volatile compounds”) or bouquet (”more complex

flavour compounds which evolve as result of fermentation, elevage and ageing” – for

a more extensive review see Lambrechts and Pretorius, 2000). Thus these three terms

are often used interchangeably (Lambrechts and Pretorius, 2000).

It is possible to classify the compounds contributing to wine aroma into four different

classes: varietal, pre-fermentative, fermentative and post-fermentative (Boulton et al.,

1998; Rapp, 1998; Schreier and Jennings, 1979). The first class is based on

compounds which originate from grapes and provides the basis of varietal character.

Those in the second class are formed during extraction and conditioning of must

preceding the fermentation. During the alcoholic and malolactic fermentations,

flavours that belong to the fermentative class are produced by yeast and bacteria. The

last class of compounds appears as a result of enzymatic or physico-chemical

reactions during the process of aging in wood or in the bottle. As already stated, the

factors that most influence the flavour of wine are yeast and fermentation conditions

(Lambrechts and Pretorius, 2000).

During a typical wine fermentation, about 95% of the sugars present in the must are

converted into ethanol and carbon dioxide, 1% into cellular material and 4% into

other end products (Boulton et al., 1998). Ethanol, glycerol and carbon dioxide play

fundamental roles in final aroma perception, but the greatest influence on wine

bouquet comes from organic acids, higher alcohols, esters and, to a lesser extent,

aldehydes. These classes of compounds may also have undesirable effects,

particularly if present at higher concentrations. Other products of microbial activity

5

that negatively influence final wine aroma are reduced sulphur compounds, such as

hydrogen sulphide, and at certain concentrations, organic sulphides and thiols. Wine

bouquet characteristics are also influenced by the yeast population. Every yeast genus

and species has a specific metabolic activity, with some strains of the same species

exhibiting large differences in terms of the production of secondary metabolites

(Romano et al., 2003).

The alcoholic fermentation is a process that is carried out by different yeast genera

and species where S. cerevisiae is the most important species, in both inoculated

fermentations and spontaneous fermentations (Fleet et al., 1984; Heard and Fleet,

1985; Lema et al., 1996). The early stages can nonetheless be dominated by the

growth of non-Saccharomyces yeasts, until the ethanol concentration reaches

inhibitory levels (around 5% v/v; Romano et al., 2003). The contribution that the

indigenous micro-flora can make to aroma depends on several factors, related to the

species (taxonomic identity, kinetics of growth, biochemical properties), the condition

and treatment of the grapes in the early stages of the winemaking process and the

operations before and during the alcoholic fermentation. In particular, sensory

differences were shown to exist between spontaneous and inoculated fermentations

(Lambrechts and Pretorius, 2000; Romano et al., 2003), where the influence of wild

yeast on wine quality was minimized in the inoculated wines.

An influence on the composition of wines obtained through the use of different yeast

species and strains of the same species was investigated by Romano and co-workers

(2003) and provided analytical profiles of the wines. The metabolites produced across

the various yeast species were usually the same (with rare exceptions), although they

differed in their concentrations. Thus is seems that only differences in the absolute

amount of specific metabolites will be enough to significantly change the final aroma

of a wine. Therefore an important consideration relates to the differences between

spontaneous or inoculated fermentation. Although the use of commercial starter yeasts

induces reliable and rapid fermentations (favouring good quality wines), it is also

suggested that the product obtained in this way is often nondescript. To carry out a

spontaneous fermentation may result in enhanced aroma complexity and perhaps

terroir identity (Lambrechts and Pretorius, 2000; Romano et al., 2003), but control of

the fermentation may be reduced as a trade off (Rainieri and Pretorius, 2000).

6

C6H12O6 2C2H5OH + 2CO2

This last consideration has to be kept in mind during the development and isolation of

new commercial strains. If reliable fermentative capacities and innovative yeast

contributions to the wine are fundamental characteristics of an improved commercial

yeast strain (Fleet, 2008; Rainieri and Pretorius, 2000), it is not possible to create

strains without respecting the main oenological traits of the parental strain (Fleet,

2008). For this reason, techniques that permit the yeast to adapt itself to the

environment where it has to grow and live (i.e. fermentation conditions) may also

allow preservation of the original desirable characteristics of that specific strain.

Adaptive evolution therefore may fit this purpose perfectly (see Section 1.4).

Sugar catabolism by yeast

As previously stated, the two most important products of alcoholic fermentation are

ethanol and carbon dioxide:

For every molecule of hexose (mostly glucose and fructose in the grape juice) two

molecules of ethanol and two of carbon dioxide are produced (Brock et al., 1994).

The molecular weight of the C6 sugars are 180 g/mol, whilst it is 46 g/mol for ethanol

and 44 g/mol for CO2. Thus the production of ethanol corresponds to 51.1% by

weight of the sugars fermented (Boulton et al., 1998). In winemaking, ethanol is not

measured by weigh but by volume (specific gravity at 20°C = 0.78 g/l), thus its

theoretical volumetric yield would be:

However this value is only theoretical and some of the sugar is utilised by the yeasts

as constituents of cell material and other products such as glycerol and succinic acid.

It is assumed that in a wine fermentation the actual percentage of conversion from

sugar to ethanol is around 60% (Boulton et al., 1998).

51.11 / 0.78 = 65.5% v/v

7

While the chemical reaction of the alcoholic fermentation can be described by the

equation above, the biochemical mechanisms by which yeasts ferment sugars to

ethanol are complex processes and include many steps, catalyzed by many enzymes

(Boulton et al., 1998). Essentially it is possible to divide this process into three main

parts: the first is the uptake of sugar into the cell through the plasma membrane; the

second is the breakdown of the sugars to pyruvate through glycolysis, which in turn

can have two remain fates: respiration or fermentation. Respiration leads to a more

efficient energy utilization of the sugar, but under winemaking conditions yeast

usually utilise the fermentative pathway.

Another important detail of the alcoholic fermentation is the higher preference of S.

cerevisiae for glucose than for fructose. During the progress of fermentation the

discrepancy between glucose and fructose utilization increases, due to associated

stresses such as increasing ethanol concentration and nutrient limitation (Berthels et

al., 2004; Guillaume et al., 2007). Although more work is required to determine the

precise basis for differences in the rate of glucose and fructose fermentation, two steps

in the utilization of hexoses are implicated: plasma membrane sugar transport systems

and hexose phosphorylation during the first step of glycolysis.

The principal steps of sugar catabolism and the points possibly involved in the

different preference for glucose and fructose are discussed below.

First step of sugar utilization: hexose uptake through the plasma membrane

Hexose transport by S. cerevisiae has been shown by many authors to be a critical

point for the different preference shown by this organism for glucose and fructose. In

this yeast, sugar uptake is mediate by facilitated diffusion, where 34 proteins

constitute the permease family (Figure 1.1; Wieczorke et al., 1999). Of these proteins,

twenty form the subfamily of hexose transporters and glucose sensors (Hxt1-17, Gal2,

Snf3 and Rgt2 – Table 1.1; Reifenberger et al., 1995). However, under fermentation

conditions, Hxt1 to Hxt7 are the most important for glucose and fructose consumption

(Guillaume et al., 2007; Karpel et al., 2008; Perez et al., 2005; Verwaal et al., 2002).

While these transporters have the ability to transport both glucose and fructose with

either low or high affinity, all typically have a higher affinity for glucose than for

fructose.

8

Figure 1.1 – The yeast sugar transporter homologues. From Wieczorke et al. (1999)

Table 1.1 – Twenty proteins forming the hexose transporter and glucose sensor family

Protein Affinity/Comment

Hxt1 Low

Hxt2 High (controversial)

Hxt3 Low

Hxt4 Moderately low

Hxt5 Moderate for glucose, low for fructose

Hxt6 High

Hxt7 High

Hxt8-17, Gal2 (Less important during wine fermentation)

Snf3, Rgt2 (Sensing of the available glucose)

9

The expression of each HXT gene is regulated by environmental factors, especially the

extracellular hexose concentration. At high hexose concentration in the media, the low

affinity system is principally responsible for sugar uptake into the cell, while the high

affinity is under control of catabolite repression (Bisson, 1988). With the progressive

depletion of hexose from the medium the cell shifts from the low to the high affinity

system. This mechanism, although complex, has in part been explained (Ramos et al.,

1988). The low affinity system is a constitutive, kinase-independent, facilitated

diffusion process. With the de-repression of the high affinity transporter system, the

low affinity system is not exhibited and vice-versa. However this is a dynamic and

progressive process and some of the low and high affinity membrane transporters are

simultaneously expressed during certain stages of fermentation (Perez et al., 2005).

Expression of HXT1-7 is also strain dependent. Various authors have studied the

expression of hexose transporters under oenological conditions, showing some

controversies result depending on which strain is studied. For example, expression of

HXT5 was not found in a study of a wine yeast strain (Perez et al., 2005), while the

same membrane transporter was expressed during mid-fermentation in a different

strain (Karpel et al., 2008). The comparison of these and similar studies (Varela et al.,

2005; Verwaal et al., 2002), also highlights that in different strains HXTs are

expressed at different levels during different phases of fermentation (i.e. different

substrate concentrations).

The high-affinity transporters are induced when the glucose concentration is low (≈ 1

– 4 mM or 0.18 – 0.72 g/l) and are repressed when the concentration of glucose is

higher. Conversely low-affinity carriers are induced by high glucose concentration (≈

50 – 100 mM or 9 – 18 g/l) or in fact are thought to be constitutively express (Ciriacy

and Reifenberger, 1997). Hxt1 and Hxt3 are considered low-affinity transporters, Hxt4

moderately low-affinity and Hxt2, Hxt6 and Hxt7 high-affinity. However, according

to Perez et al. (2005) and Luyten et al. (2002) the Hxt2 transporter is only transiently

expressed during the first hours of the lag phase, when the hexose concentration is

high. This suggests that Hxt2 may be able to bypass glucose repression in the initial

stages of the lag phase, playing an important role in the growth phase. However HXT2

was not expressed at the latter stages of the fermentation, when the concentration of

hexoses drops below repressive levels. Hxt5 has moderate affinity for glucose (Km =

100 mM) and low affinity for fructose (Km = 40 mM; Diderich et al., 2001). During

10

fermentation, HXT3 is the only gene of the hexose transport family expressed

throughout the entire process (Luyten et al., 2002; Perez et al., 2005; Varela et al.,

2005). For this reason HXT3 may be the dominant hexose transporter and as such play

a fundamental role in determining glucose and fructose uptake rates (Guillaume et al.,

2007).

More work is required to determine the reasons for the higher affinity of these

transporters for glucose than for fructose. Ethanol is known to have protein denaturing

properties and to disrupt the plasma membrane (Bisson and Block, 2002; Stanley et

al., 2010) and intracellular enzymes and structures (due to the increase in membrane

permeability and passive proton flux). Berthels et al. (2004) observed that a high

ethanol level inhibits sugar utilization, but with a different effect on glucose compared

fructose. This led them to hypothesise that the glucose utilization capability may be

more robust than fructose utilization. In a study of HXT1-7 deletion strains, Karpel et

al. (2008) found that only the strain lacking HXT3 was unable to complete

fermentation in media containing of 5% (v/v) exogenous ethanol. These authors

suggested that HXT3 may have an important role in ethanol tolerance. Ethanol also

has a differing ability to inhibit different Hxt transporters and influence their affinity

for glucose or fructose (Santos et al., 2008). Moreover ethanol is able to shift the

tautomeric equilibrium of fructose from the readily transported pyranose form to the

furanose form. Thus unlike the aldose glucose which is entirely in the transportable

(pyranose) form, the situation for fructose, of which typically only 70% is in the

pyranose form, is made even worse by ethanol. This has been suggested to be one

reason for the generally lower utilisation of fructose by this yeast (Berthels et al.,

2004).

It is important to note that the discrepancy between glucose and fructose is not

constant during the progress of the fermentation. Rather it has been shown to be time

of fermentation and strain dependent (Berthels et al., 2004). Although the

mechanisms are unknown, a possible explanation stems from the different ethanol

tolerance shown by different strains. Another possibility is linked with nitrogen

availability and utilization. Different strains utilise nitrogen at different rates and with

different efficiencies (Jiranek et al., 1995). Moreover, the depletion of nitrogen in

combination with the rapid turnover of sugar transporters in the stationary phase, is

11

thought to be responsible for inactivation of sugar transport systems with a resulting

reduction in fermentative capacity. Berthels et al. (2004) showed that in every case of

nitrogen supplementation, fructose consumption was enhanced with respect to

glucose. Another possibility related to glucose and fructose discrepancy could be the

different affinity of the sensor-proteins that are located on the plasma membrane

(Ozcan et al., 1996). Some have been identified as having affinity for glucose and at

least one has a differing affinity for glucose compared to fructose (Rolland et al.,

2001a; 2001b). The presence or location of specific fructose sensors has not be

reported (Berthels et al., 2004).

Second step of sugar utilization: glycolysis

Once sugar has entered the cell, yeast start to break it down liberating energy (as

ATP) and in turn re-oxidising NAD. This first part of the process is called glycolysis.

Two molecules of ATP are required for the production of glyceraldehyde-3-

phosphate, while 4 molecules of ATP are formed with the production of pyruvate

(Figure 1.2). The formation of pyruvate is common for the metabolism of sugar

whether concluding with fermentation or respiration.

Once glucose and fructose have entered into the cell, they are phosphorylated to

fructose-6-phosphate. For fructose this is a direct step, while glucose has to be

phosphorylated to glucose-6-phosphate and then converted in fructose-6-phosphate by

the enzyme phosphoglucose isomerase (PGI). From this point there is no more

differentiation between glucose and fructose. Glucose and fructose are both

phosphorylated by the enzymes hexokinase Hxk1 and Hxk2 (Figure 1.2), albeit at

different rates (glucose faster than fructose). Additionally glucokinase Glk1

phosphorylates glucose (Berthels et al., 2004; Guillaume et al., 2007; Serrano and

Delafuente, 1974). According to these authors, the phosphorylation capacity of these

enzymes exceeds the amount of sugar transported into the cell. This would indicate

that these enzymatic processes do not contribute to the ability of yeast to utilise

glucose and fructose at different rates. However, a recent study (Berthels et al., 2008)

demonstrates that over-expression of hexokinases could in fact alter the rate of

fermentation. Thus, more work is necessary for a better understanding of the topic. A

final consideration relates to the catabolic repression capacity: Hxk2 is required to

maintain catabolic repression by glucose, while fructose catabolic repression

12

Figure 1.2 – Principal steps and enzymes of glycolysis (green background) and alcoholic fermentation (blue background). HXT (hexose transporter), HXK (hexokinase), GLK (glucokinase), PGI (phosphoglucose isomerase), PFK (phosphofructokinase), FBA (aldolase), TPI (triosephosphate isomerase), TDH (glyceraldehyde-3-phosphate dehydrogenase), PGK (phosphoglycerate kinase), PGM (phosphoglycerate mutase), ENO (enolase), PYK (pyruvate kinase), PDC (pyruvate decarboxylase), ADH (alcohol dehydrogenase). Glycolysis ends with the formation of pyruvate, while alcoholic fermentation is the transformation of pyruvate in ethanol and CO2. (Adapted from Boulton et al., 1998).

Glycolysis

Alcoholic fermentation

13

requires either Hxk1 or Hxk2. Also in this case the mechanism involving glucose and

fructose may be different (Berthels et al., 2004).

Pyruvate is the final product of glycolysis. From this point, pyruvate is further

modified through the Krebs cycle if the respirative route is followed. Alternatively, in

alcoholic fermentation the process ends with the decarboxylation of pyruvate to

acetaldehyde followed by reduction to ethanol and re-oxidation of NAD (Figure 1.2).

Third step of sugar utilization: respiration vs. fermentation.

Wine yeast have the ability to grow under aerobic or anaerobic conditions, as they can

both respire and ferment sugars. The consequences of this dual activity are important

and the production of ATP and biomass differs markedly depending on which

pathway is followed.

In Saccharomyces respiration occurs when the concentration of sugar is below

repressive concentrations (between 1 – 4 g/l and 30 g/l; De Deken, 1966; Hornsey,

2007) and oxygen is available. Respiration allows complete oxidation of one molecule

of sugar to produces 36 molecules of ATP, thus no alcohol is produced and the full

energetic potential of the sugar is achieved. As a result, the yeast has a high growth

rate and produces a large biomass. With respiration, every gram of sugar respired

produces 0.25 g of dry cell weight (i.e. at a ratio of 1:4). This information is important

when the target is the production of biomass, for example in the production of

commercial yeast. However, under winemaking conditions, yeast usually do not

respire. The ideal maximum yield of cells is readily decreased by a poor supply of

oxygen to the cells. Moreover, as stated previously, another phenomena that drives

yeast metabolism through the fermentation pathway, is linked to the sugar

concentration: if sugar exceeds 30 g/l (De Deken, 1966; Postma et al., 1989), even if

in presence of oxygen, a complete suppression of the respiration capacity is observed,

and metabolism switches to an aerobic alcoholic fermentation. This phenomena is

known as the “Crabtree effect”.

Alcoholic fermentation is a process that utilises sugar in the absence of oxygen. In this

case, only part of the energetic potential of sugar is released. For every molecule of

sugar fermented 2 molecules of ATP are produced along with 56 kcal of heat, which

is released to the environment. The remaining energetic potential of the sugar

14

is bound in ethanol. In this case, 176 g of sugar are required to obtain 1 g of dry cells

(a ratio of 1:176). In an industrial setting, the fermentative condition is easier to

achieve than respiration, because the formation of CO2 in tanks and the high level of

sugars inhibit the respirative capacity of yeast from the first hours of contact with

grape juice (conditions typical of winemaking).

Other considerations in sugar utilization

As previously stated, during the alcoholic fermentation yeast produces compounds

apart from CO2 and ethanol. Glycerol, succinic acid, acetic acid, aldehydes, pyruvate

and acetoin are the most common and are produced as a result of secondary

metabolism. The production of glycerol is due to the inability of the cell to re-oxide

NADH2 to NAD via acetaldehyde to ethanol. If a block occurs at this level, the NAD

is formed via reduction of dihydroxyacetone-phosphate to glycerol-phosphate (and

then glycerol). External factors can influence or deviate the alcoholic fermentation

pathway. This is the case for SO2 added to the grape juice, which combines with

acetaldehyde, preventing its reduction and redirecting the pathway to glycerol. If the

addition of SO2 is high, a total arrest of ethanol production is possible, with glycerol,

acetaldehyde and CO2 being the only metabolites formed. It is important to note that

different concentrations of side-products can be strain dependent. For example, some

strains naturally produce different levels of SO2 that can influence the final

concentration of acetaldehyde. Other reasons for differences in ethanol yield relate to

the fact that the fermentation is mediated by variable enzymatic actions and thereby

also influenced by nutrients (especially nitrogen) availability in the media. For a more

extensive review see Alexandre and Charpentier (1998), Bisson (1999) and Boulton et

al. (1998).

To summarize the significance of the discussions so far, sugar catabolism by yeast is a

process that involves several steps. Sometimes a block can occur at the level of one of

these steps and the result is an arrest or a slowdown of sugar catabolism. In the

winemaking process, catabolism of sugar is carried out under fermentative conditions.

Thus, if any problem affects sugar catabolism, it is possible to observe a stuck or

sluggish fermentation. Once a fermentation sticks it can be difficult to restart. This

can have an effect on the logistics of the winery (holding precious tank volume in the

15

busy vintage period) and leave residual sugars in the wine, which can affect sensory

qualities and microbial stability.

Due to the different rates of consumption of glucose and fructose, it is possible to

observe higher concentrations of fructose in the latter and more difficult stages of

fermentation. If the fermentation sticks or becomes sluggish, higher concentrations of

the less preferred hexose (fructose) increase the difficulty of restarting sugar

catabolism. Under this light, generating a wine yeast strain with improved fructophilic

characteristics, might lead to strains showing more robust fermentation capabilities

and a more balanced ratio between glucose and fructose during the progress of

fermentation. Despite hexose uptake and consumption being a mechanism that

depends on numerous environmental and biological parameters, and more work is

needed to clarify the reasons for the glucose and fructose discrepancy, it is clear that

one or more causes will relate to the kinetics of the transport systems and initial

enzymatic processes of the fermentation pathway.

Understanding the causes of slowdown or arrest of fermentation during winemaking is

vital and a necessary step in an improvement program for wine yeast. The main

factors influencing yeast growth and metabolic activity are numerous and a more

extensive discussion of these is given in the following section.

1.3 STUCK AND SLUGGISH FERMENTATIONS: ONE OF THE BIGGEST PROBLEMS DURING THE WINEMAKING PROCESS

Despite yeast being a somewhat robust microorganism, well adapted to winemaking

conditions, fermenting grape juice represents a harsh environment for S. cerevisiae. If

yeast loses its metabolic activity or viability, the most evident outcome is a decrease

of fermentation rate, which sometimes results in a complete arrest of sugar

catabolism. There are many reasons for stuck or sluggish fermentations and thus it can

be difficult to identify the precise cause in a given situation (Bisson and Butzke, 2000;

Malherbe et al., 2007). In fact, the system ‘yeast-must-wine’ presents many

interactions and complicated equilibria which are involved throughout the progression

of fermentation. Total consumption of sugar by yeast can fail even if only one

16

parameter is out of balance. However, most problem fermentations occur as a result of

multiple parameters. For this reason, reduction of the risks of sluggish or stuck

fermentations requires a sufficient understanding of the potential causes of these

problems. Nutrient limitation, physical factors, toxic substances, microbial

incompatibly are arguably the most important (Bisson, 1999). In Table 1.2 are

detailed the main causes implied in problem fermentations. For a more extensive

review on this topic see Alexandre and Charpentier (1998), Bisson (1999), Boulton et

al. (1998) and Fleet (1992).

Many of the factors affecting wine yeast during fermentation can be eliminated pre- or

during winemaking. Modern wineries are usually well equipped to analyse juice

composition and control several parameters of the winemaking. Thus, it is possible to

supply the juice with nutrients necessary for the yeast to complete the fermentation. In

the same way, it is possible to correct and control physical factors, such as

temperature, pH and lack of oxygen, to better suit yeast requirements. Resistance to

toxic substances or competition with other microorganisms can be limited by

inoculating the juice with a strong yeast culture or with more careful vineyard

management. Oenological practices in the winery can be applied to preserve the

viability and the metabolic ability of S. cerevisiae. However at the present it is not

possible to control the differing ability of yeast to utilise fructose compared to

glucose.

As described extensively above, S. cerevisiae shows a higher fermentation rate for

glucose compared to fructose. Thus, in the latter stage of alcoholic fermentation,

fructose is the main sugar remaining. The lower fermentability of fructose, combined

with nutrient depletion and the high alcohol content, contribute to possible

fermentative problems (Berthels et al., 2004; Guillaume et al., 2007). Another

important consideration is linked with the ratio between glucose and fructose

concentrations. In grape juice this ratio is approximately 1.0, progressively decreasing

during fermentation. Schutz and Gafner (1993) in fact linked stuck fermentations with

the value of the glucose and fructose ratio falling below 0.1. These authors also

suggested that it can be stimulatory to restore this ratio to a value above 0.1. Thus a

stuck fermentation can be restarted by artificially adding glucose to the medium. This

finding may suggest that the predominance of fructose in the latter stages of

fermentation is not only a consequence of the higher fermentability of

17

Table 1.2 – Possible factors influencing the catabolic activity and growth of yeast during alcoholic fermentation in wine.

Factors affecting yeast during fermentation

Details Implications and comments

Nutrient limitation

� Macro-nutrients (carbon, nitrogen, phosphate)

� Micro-nutrients (vitamins, minerals)

� Source of energy (carbon) � Proteins synthesis (macro-

nutrients) � Survival factors (micro-

nutrients)

Physical factors � Temperature � Low pH (< 2.8) � Lack of oxygen

Most notable effects of temperature are on the plasma membrane, reducing or increasing its fluidity. Low pH reduces ethanol and fatty acid tolerance. Oxygen is important for biosynthesis of cell membrane components.

Toxic substances

� Ethanol � Medium-chain fatty acids � Fungicides and pesticides

(vineyard residues) � Antifungal agents (plant

response to fungal infections)

Inhibition (to complete arrest) of cell metabolism. Manifold interactions at different levels.

Microbial incompatibly

� Non Saccharomyces yeast � Bacteria

� Nutrient competition � Toxic substances (i.e. yeast

killer factors and mycotoxins)

Other factors Oenological practices Can generate adverse conditions for the growth and metabolism of yeast

Different ability to utilise glucose and fructose

Less efficiency for fructose transport and/or catabolism

Fructose predominates in the latter and more difficult stages of fermentation

18

glucose by wine yeast, but it could also be a cause of the arrest of fermentation when

residual sugar is almost completely composed of fructose.

Many of the possible factors influencing the metabolic activity and growth of yeast

during alcoholic fermentation presented above, can be controlled. In wineries it is

possible to add nutrients, correct pH, regulate temperature, supply oxygen, manage

the micro-flora and control many other parameters. However it is not possible to

change the sugar composition of the juice and thus it is not possible to regulate the

different consumption rate for glucose and fructose, as is characteristic of

Saccharomyces cerevisiae. For this reason, improving wine yeast appears to be the

only solution to achieving a more uniform consumption of glucose and fructose.

1.4 IMPROVING WINE YEAST STRAINS

The importance of improved strains for the modern wine industry

A suitable yeast for winemaking has to satisfy several requirements and show

determinate characteristics. These aspects of yeast can be divided into technological

and qualitative traits (Zambonelli, 1998). Technological traits influence the efficiency

of the fermentation process. Those most often targeted for improvement are

fermentative vigour, ethanol yield and tolerance, growth temperature range, inhibition

by other microbial species, resistance to SO2, type of growth in liquid media

(dispersed or aggregated cells, flocculence, foam and film formation and

sedimentation speed), tolerance of extreme temperatures and resistance to killer factor

(Rainieri and Pretorius, 2000). Qualitative characteristics influence the chemical

composition and sensorial profile of wines: fermentation by-products (glycerol,

succinic acid, acetic acid, acetaldehyde, n-propanol, iso-butanol, isoamyl alcohol, �-

phenylethanol, etc.), production of sulphur compounds (H2S, SO2, etc.) and the action

of enzymatic activities (�-glucosidase, esterase, proteolytic enzymes, autolysis, etc).

An ideal improved strain is obtained with consideration and satisfaction of both

groups of attributes (Rainieri and Pretorius, 2000).

Yeast improvement can be carried out using a broad spectrum of techniques (Table

1.3). It is highly unlikely that one can identify strains possessing an ideal combination

19

Table 1.3 – Possible approaches for the improvement of specific traits in wine yeasts.

Approach Technique Advantages Disadvantages

Non recombinant

Natural selection

Preserves the natural characteristics of a strain

Likely to be limited in the extent of the improvement

� Mutagenesis � Hybridization � Rare-mating � Spheroplast fusion

Increases genetic diversity

Can introduce significant additional non-specific changes to the strains phenotype

Adaptive (directed) evolution

� High and specific potential of improvement

� No direct DNA manipulation

� Difficulties in defining the appropriate selective condition for desired the outcomes

� Potential lost of desirable attributes

Recombinant � Gene cloning and transformation

High and specific potential of improvement

� Requires high knowledge of the relevant biological and genetic mechanisms

� Must consider views on the use of genetically modified microorganisms in the food industry

20

of technological and qualitative traits by only relying on the natural availability of

different phenotypes (Rainieri and Pretorius, 2000). However, selection of clones is

usually carried out from must or wine, where the yeasts are well adapted to the

oenological environment. Typically a high number of yeast strains are considered,

which are then submitted for analysis of their oenological properties. Thus, “clonal

selection” provides a collection of diverse backgrounds that are very useful for

successive genetic improvement programmes (Giudici et al., 2005).

Progress in the field of microbiology has made it possible to obtain strains that are

quite different from the parental strains (Pretorius, 2000): induced mutation and

selection, hybridisation, rare-mating, spheroplast fusion, and gene cloning and

transformation are among the most widely used techniques. For a more extensive

review see Barre et al. (1993), Pretorius (2000), Pretorius and van der Westhuizen

(1991) and Zambonelli (1998). However, generating a strain with many differences

from its parent it is not always desirable. The challenge between tradition and

innovation is the key to describing this phenomenon and the maintenance of the

original identity of the wine has to be considered during strain improvement.

Conversely, the possibility of obtaining genetically engineered strains expressing

novel genes can open doors to some new and exciting winemaking approaches, and

more specifically combine desired oenological traits (Rainieri and Pretorius, 2000).

Recombinant DNA technologies are the most modern approach for yeast

improvement programs. However, in order to manipulate DNA, the complexity of

yeast biology requires a high level of knowledge and a deep understanding of the

intricate interactions between different biological and gene expression and regulation

mechanisms. Moreover, comment needs to be made about the ethical debates existing

around DNA manipulation in the food industry. To some extent, genetic improvement

of plants, animals and micro-organisms is necessary and required by modern

production processes of foods and beverages, but on the other hand a large segment of

the market rejects GM organisms (Pretorius, 2000). Some consumers consider the GM

approach undesirable and thus avoid the consumption of products derived from DNA-

manipulated organisms. In the absence of extensive results about the safety of food

and beverages derived from GM organisms, this ethical aspect has to be considered

and the genetic improvements required by the industry have to be balanced against

consumer expectations and needs.

21

A technique that needs exploration in improving microorganism is adaptive evolution.

Adaptive evolution can selectively retain phenotypes presenting mutations which

permit a better adaptation to a specific stress. Mutation can arise spontaneously from

exposure to or in response to environmental stress. Alternatively mutants can be

artificially induced. This makes adaptive evolution a technique that can be readily

combined with other improvement strategies, using a pre-selected strain from a

natural winemaking environment as a genetic background to isolate phenotypes with

specific improved characteristics. Adaptive evolution doesn’t require a deep

knowledge of the genetics of the process targeted for the improvement and does not

produce GM organisms. Moreover, this technique can preserve the original

oenological traits of the parent, resulting in a perfect tool with which to maintain

oenological traditions and terroir, whilst improving and bringing innovation to the

winemaking process.

The ability of adaptive evolution to yield isolates better suited for a selective

environment has been demonstrated, and several studies have extensively investigated

the genetic mechanisms of the occurrence of adaptive mutations as well as the

dynamics of the evolving populations (see below). What has been less studied, is the

possibility of the practical use of adaptive evolution as an improvement strategy. The

identification of a particular area of interest for improvement appears to be the next

step. Thus in winemaking, the possibility of reducing differences in the consumption

rate of fructose compared to glucose seems to be a perfect case study for this purpose.

The adaptive evolution technique

Historical background

Since the beginning of the last century there have been open debates on evolution

theories. Moreover a better knowledge on unicellular organisms opened new fields of

investigations and two different schools of thought began. The first supported the

“mutation theory”: a random genetic mutation pre-existing in the population prior the

beginning of a stress, giving a specific adaptation to the descending population. The

other notion is the “adaptation theory”: the adaptation to an induced stress is too

22

precise to be explained as a result of a random mutation and thus the adaptation is a

physiological response to a particular stress, i.e. the mutations are not present before

the occurrence of the stress. The first important study that tried to explain these

controversial notions came in 1943 (Luria and Delbruck, 1943). Observing the

distribution of the accumulation of adapted bacteria in culture, these authors

disproved the adaption theory. However they did not completely support the mutation

theory as they believed proof was required.

It was only at the end of the century, armed with a more detailed knowledge of cell

biology, that Cairns and co-workers demonstrated that cells may have mechanisms to

choose which mutations will occur to allow adaptation to a particular environment:

“Bacteria, in stationary phase, have some way of producing (or selectively retaining)

only the most appropriate mutations” (Cairns et al., 1988). Thus, these authors

showed that mutants not only pre-exist in a population as result of random and not-

specific mutation, but they can arise in response to a specific selective pressure.

Several other authors confirmed the Cairnsian theory: Foster (1992), Foster and

Cairns (1992), Hall (1988, 1989, 1992, 1997) and Steele and Jinks-Robertson (1992).

However, the exact mechanisms of how the cell is able to induce a specific mutation

to relieve a stress are not completely understood and more work is necessary.

Pioneering work was carried out on prokaryotic cells. Bacteria have been used to

prove the existence of selection-induced mutations since the beginning of the

evolutionary debate at the end of the nineteenth century. However more recent

experiments proved that selection-induced mutations also occur in yeast (Brown et al,

1998; Hall, 1992; Paquin and Adams, 1983). Thus, adaptive evolution theories are not

limited to prokaryotes and eukaryote cells also have some system to adapt to a

specific stress (Steele and Jinks-Robertson, 1992).

Another field of investigation related with the mechanism of adaptive mutation has

been the frequency of the occurrence of a mutation. This knowledge is fundamental to

understanding the evolution in a population exposed to a selective pressure. One of

the early debates was solved in 1983 (Paquin and Adams, 1983): mutations occur

more frequently in diploid cells than in haploid. These authors support the idea that

diploid cells should generate twice as many adaptive mutations than haploids because

23

the genome is double in size and so too the number of duplications. However they

demonstrated a rate of adaptive mutation only 1.6 times higher in diploids compared

to haploids. This divergence from the theoretical value can be attributed to deleterious

mutations in the recessive heterozygote state. Moreover the size of a haploid

population is usually higher that the isogenic diploid. Population size also affects the

frequency of mutations (Paquin and Adams, 1983; Wahl and Krakauer, 2000; Wick et

al., 2002). Thus, this phenomenon can contribute to a compensation for the lower

frequency of mutation produced by haploid cells. Conversely, for an induced

adaptation to become a permanent and deep-seated characteristic of a population it

requires a mutation at the allele level, its fixation and the replacement of the new

allele in the entire population (Zeyl, 2004). Despite the high frequency of mutation

occurring in diploids (Paquin and Adams, 1983), in larger population experiments the

fixing time of a mutation can become the determinant factor of adaptive evolution

(Chambers et al., 2007). This can be explained in part by the non-dominant mutations,

requiring longer to bring adaptive advantage to a heterozygous population (Zeyl et al.,

2003). According to this point of view, haploids should evolve faster than diploids

(Orr and Otto, 1994). As confirmation, this phenomena has been observed in a 1.0

litre adaptive evolution fermentation experiment (McBryde et al., 2006), where a

diploid population took a longer span of generations to adapt compared to the derivate

haploid.

Another key consideration in adaptive evolution is the relation between the

occurrence of a mutation and the genetic and physiological characteristic of the

studied organism. Different species and environmental conditions drastically affect

the frequency of adaptive mutations. Hall (1988) observed a rate of advantageous

mutations of around 2 x 10-12 per cell divisions. Paquin and Adams (1983) obtained

rates between 5.68 x 10-12 and 3.55 x 10-12 per cell division, while Zeyl (2004)

reported a value of one adaptive mutation every 1011 cell divisions. McBryde and co-

authors (2006) isolated adapt-evolved mutants after 250 and 350 generations, while

Brown and colleagues (1998) observed evolution phenomena after 450 generations.

Since work on adaptive evolution began, new studies have yielded a large amount of

data to demonstrate the validity of this theory and to explain the mechanisms of

adaptive mutation. On the other hand, the exploitation of adaptive mutations as a tool

to induce cell modifications in order to improve specific industrial characteristics of

24

strains is an underdeveloped area of investigation. There are few studies which have

evaluated the adaptive evolution technique in this context. Moreover, only a small

fraction of these investigations have utilised yeast and most work has been performed

on bacteria.

Adaptive evolution

In adaptive evolution, a population of micro-organisms is grown for hundreds or

thousands of generations, while preserving a sample of the population at chosen

intervals so that they can be compared directly with each other and with their

ancestors (Zeyl, 2004). Thus, the critical steps in adaptive evolution are: to create a

specific stress condition; to maintain the population under stress for many

generations; to keep population samples for further comparison with the original

parental strain.

Studies on adaptive evolution have been carried out in either immobilized cells or

populations maintained in liquid media. In the first approach, cells are grown on

plates containing solid media, which impose a specific stress condition. After a period

of time, it is possible to observe growth of mutants, adapted to the specific stress.

Mutants can be easily isolated for characterization or be re-plated for further

adaptation.

For a population maintained in liquid media, two approaches exist (Zeyl, 2004):

chemostats and serial transfers (or batch cultures). In the chemostat, a population is

maintained in a constant environment, adding fresh media to the culture, while

partially exhausted media is removed at the same rate. For serial transfer a small

percentage of the population is transferred into a new vessel containing fresh media.

With this approach it is possible to create a cyclic condition of induced stress,

alternating with periods of non-stress. With both of these methods it is possible to

choose a specific stress (e.g. by changing the chemical composition of the media) to

“force” the population to adapt to that particular environment, by inducing adaptive

mutations. These two approaches of adaptive evolution may seem similar, but a

difference exists that can influence the choice between one approach over the other.

Continuous culture tends to maintain a population in a largely stable environment for

an indefinite time. This is somewhat artificial because in nature (and in almost all

25

industrial food processes) a population is usually subject to a definite growth phase,

depending on the chemical and physical parameters (substrate, temperature, stresses,

etc.) that defines the kinetics of the growth of that population. Continuous culture

represents only a fraction of the life of a culture in the real world. For example, if the

target is to improve a yeast strain under the stressful conditions of the wine

fermentation, continuous culture only reproduces a particular phase of that

fermentation. This can be advantageous because it can drastically reduce the run time

of the experiment since, for example, the population can be maintained in the

exponential growth phase permanently (McBryde et al., 2006). However the

continuous culture is not representative of the dynamic fermentation process and for

this reason it may not be suitable for improving a strain for industrial/commercial use.

This problem can be solved using a sequential batch cultivation system. This approach

ensures the representation of conditions similar to an industrial environment. Thus in

terms of the previous example of an oenological fermentation, cells are inoculated in

a defined medium representing a grape juice (e.g. CDGJM; Henschke and Jiranek,

1993) and allowed to ferment under oenological conditions until the sugar

concentration reaches that of a finished wine (< 2.5 g/l). At this stage sufficient

numbers of cells are re-inoculated into a fresh batch of media and a new fermentation

is initiated. The sequential batch fermentation system has the disadvantage of

requiring an extended culture time, as the cells are not always at the maximum

division rate of the exponential phase. However this system is more representative of

the industrial process and would conceivably eliminate the selection of alterations that

are disadvantageous to general fermentation performance (McBryde et al., 2006).

1.5 AIM OF THE PROJECT As fermentation is one of the most important steps in the winemaking process,

research must focus on all of its resources to make this process much more reliable.

To avoid the risks of stuck and sluggish fermentations, one of the best approaches is

represented by the improvement of yeast strains. Without forgetting the aroma

contribution and other fundamental influences that yeasts have on the final product,

sugar metabolic activity is a key consideration when creating or isolating a new strain

26

for industrial use. In particular, the different affinity of yeasts for glucose and fructose

(present in equal concentration in grape juice) is one of the greatest difficulties for

completion of fermentation under oenological conditions by glucophilic yeast.

To improve wine yeast, research has the possibility of following different approaches

ranging from natural selection of clones to genetic manipulation techniques. However,

some problems linked to inadequate efficiency, application difficulties,

inappropriateness for commercial application and ethical aspects, require alternate

approaches to be used, particularly those that recognise consumer demands. For this

reason, the application of adaptive evolution to improve wine yeast may be an

important methodology to develop. It presents clear evidence of adequacy: adaptive

mutations have been shown to occur; induced mutations arise either in prokaryotes or

eukaryotes; occurrence of mutations has been studied in haploid and diploid states;

frequency of mutation has been reported in several studies; application principles of

the technique have been described along with the advantages and disadvantages of

different approaches. Moreover, the ability of adaptive evolution to generate new

phenotypes in a wine related study has been demonstrated (McBryde et al., 2006).

Thus, the possibility of testing the capacity of adaptive evolution to improve

commercial wine yeast strains for specific characteristics warrants attention. For this

purpose, fructose fermentation efficiency seems to be a good topic for such an

investigation. The use of fructose by yeast is an easy parameter to monitor

analytically, it is regulated by genetic inheritance and it has strong commercial

interest.

The aim of the research proposed here is to choose a S. cerevisiae strain normally

used in the commercial winemaking process and demonstrate the validity of the

adaptive evolution technique to generate improvements specifically in its fructose

utilization.

27

CHAPTER 2

CANDIDATE STRAIN SELECTION

28

The work presented in this chapter was aimed at identifying candidate strains suitable

for improvement of their fructose utilization capabilities. The genetic background to

be used for the adaptive evolution experiment was to be that of a commercially

available strain. To identify the candidate strain, the phenotype of 20 commercially

available strains was characterized in terms of their fermentative properties,

particularly their ability to ferment fructose, either in the presence or absence of

glucose. A description of this screening exercise is given in the following manuscript

(accepted for publication in Journal of Industrial Microbiology and Biotechnology on

17th of August 2010 – In press. Ms. No. JIMB-D-10-00444R1).

29

STATEMENT OF AUTHORSHIP

A NOVEL METHODOLOGY INDEPENDENT OF FERMENTATION RATE FOR

ASSESSMENT OF THE FRUCTOPHILIC CHARACTER OF WINE YEAST

STRAINS

NOTE: Statements of authorship appear in the print copy of the thesis held in the University of Adelaide Library.

30

2.1 A NOVEL METHODOLOGY INDEPENDENT OF

FERMENTATION RATE FOR ASSESSMENT OF

THE FRUCTOPHILIC CHARACTER OF WINE

YEAST STRAINS

T. Liccioli1, P.J. Chambers2, V. Jiranek1

1The University of Adelaide, School of Agriculture, Food and Wine,

PMB 1 Glen Osmond, SA 5064, Australia.

2The Australian Wine Research Institute, PO Box 197, Glen Osmond, SA

5064, Australia.

Corresponding author:

Vladimir Jiranek

The University of Adelaide, School of Agriculture, Food and Wine,

PMB 1 Glen Osmond, SA5064.

Phone: +61 08 8303 6651

Fax: +61 08 8303 7415

e-mail: vladimir.jiranek@adelaide.edu.au

31

Abstract

The yeast Saccharomyces cerevisiae has a fundamental role in fermenting grape juice

to wine. During alcoholic fermentation its catabolic activity converts sugars (which in

grape juice are a near equal ratio of glucose and fructose) and other grape compounds

into ethanol, carbon dioxide and sensorily important metabolites. However, S.

cerevisiae typically utilises glucose and fructose with different efficiency: glucose is

preferred and is consumed at a higher rate than fructose. This results in an increasing

difference between the concentrations of glucose and fructose during fermentation. In

this study 20 commercially available strains were investigated to determine their

relative abilities to utilise glucose and fructose. Parameters measured included

fermentation duration and the kinetics of utilisation of fructose when supplied as sole

carbon source or in an equimolar mix with glucose. The data were then analysed using

mathematical calculations in an effort to identify fermentation attributes which were

indicative of overall fructose utilisation and fermentation performance. Fermentation

durations ranged from 74.6 to over 150 hours with clear differences in the degree to

which glucose utilisation was preferential. Given this variability we sought to gain a

more holistic indication of strain performance that was independent of fermentation

rate and therefore utilise the Area Under the Curve (AUC) of fermentation of

individual or combined sugars. In this way it was possible to rank the 20 strains for

their ability to consume fructose relatively to glucose. Moreover, it was shown that

fermentations performed in media containing fructose as sole carbon source did not

predict the fructophilicity of strains in wine like conditions (equimolar mixture of

glucose and fructose). This work provides important information for programs which

seek to generate strains that are faster or more reliable fermenters.

Keywords: glucose, fructose, fermentation progress, strain comparison, composite

trapezoid rule. Abbreviation: Area Under the Curve, AUC.

Introduction

When exposed to mixtures of glucose and fructose, as occurs during the fermentation

of grape juice into wine, Saccharomyces cerevisiae utilises these sugars at different

rates (Berthels et al., 2004; Júnior et al., 2008). As a result the ratio between the

32

concentration of glucose and fructose changes, so that late in fermentation fructose

becomes the predominant sugar. This is reported to be one of the causes of arrested or

so-called stuck fermentation (Gafner and Schütz, 1996). Attempts to restart the

fermentation through re-inoculation with fresh but nonetheless glucophilic cultures

are challenging (Cavazza et al., 2004). Regardless of the cause of stuck fermentation,

excess residual sugar is undesirable as it puts the palate of the wine out of balance.

Furthermore, the fact that fructose predominates compounds the problem since

fructose is sweeter than glucose (Lee, 1987).

In seeking the precise basis for differences in the rate of glucose vs fructose

utilisation, steps in the metabolism of hexoses prior to the formation of fructose 1,6

bisphosphate are implicated. In particular this includes the systems for sensing of

extracellular sugars, their transport across the plasma membrane and phosphorylation

as part of the first steps of glycolysis. Plasma membrane sensor-proteins have been

identified which show affinity for glucose, with at least one exhibiting a different

affinity for glucose and fructose (Rolland et al., 2001a; 2001b). The presence of

sensors specific for fructose has not been reported (Berthels et al., 2004).

Phosphorylation of internalised glucose and fructose occurs via the hexokinases Hxk1

and Hxk2, albeit at different efficiencies, with glucose additionally being acted upon

by glucokinase Glk1 (Berthels et al., 2004; Guillaume et al., 2007; Serrano and

Delafuente, 1974). Since the phosphorylation capacity of these enzymes exceeds the

amount of sugar transported into the cell, it would appear that this enzymatic process

is not the basis for differences in the utilisation of glucose compared to fructose.

However, a recent study (Berthels et al., 2008) demonstrate that over-expression of

hexokinases could in fact alter the rate of fermentation. On the other hand, the hexose

transport system has been clearly shown by many authors to be a critical point in

determining fermentation rate.

In S. cerevisiae, hexose uptake is largely mediated by facilitated diffusion (Bisson and

Fraenkel, 1983), where 20 genes (HXT1 to HXT17, GAL2, SFN3 and RGT2) encode

the related proteins. Despite such plurality, only those transporters encoded by HXT1

to HXT7 appear to be of importance for glucose and fructose utilisation under

fermentation conditions (Diderich et al., 1999; 2001; Elbing et al., 2004; Ozcan and

Johnston, 1995, 1999; Reifenberger et al., 1995). Some of these latter HXT

transporters have relatively low-affinity (high Km) for hexose, whilst other are

33

considered high-affinity (low Km), and all have the ability to transport both glucose

and fructose. Importantly, whether they are high- or low-affinity systems, all have a

greater affinity for glucose than for fructose. The expression of each HXT gene is

regulated by environmental factors, especially the extracellular hexose concentration

(Ozcan and Johnston, 1995). The high-affinity transporters are induced when the

glucose concentration is low (~1 – 4 mM or 0.18 – 0.72 g/l) and are repressed when

the concentration of glucose is higher (Ciriacy and Reifenberger, 1997). Conversely

low-affinity carriers are induced by high glucose concentrations (~50 – 100 mM or 9

– 18 g/l) if not constitutively expressed. HXT3, a low-affinity transporter, is

considered key to determining glucose transport and, thereby, could play a

fundamental role in the different rates of glucose and fructose utilization (Guillaume

et al., 2007). Further work is required to define this role.

Beyond affinity differences, other factors such as nitrogen availability and response to

ethanol may be of importance. Ethanol is known to have protein denaturing properties

and disrupts plasma membrane components (Bisson and Block, 2002; Stanley et al.,

2010). Due to increases in membrane permeability and passive proton flux upon

ethanol exposure, damage to intracellular enzymes and structures might also occur.

Berthels and co-authors (2004) observed that a high ethanol concentration inhibited

sugar utilization, but to different extents for glucose and fructose. This led them to

hypothesise that the glucose utilization capability was more robust than fructose

utilization. Moreover ethanol is able to shift the tautomeric equilibrium of fructose

from the readily transported pyranose form to the furanose form. Thus unlike glucose

which is entirely in the transportable pyranose form, typically only 70% of fructose

takes the pyranose form, and less in the presence of ethanol. Thus not only is there a

discrepancy between glucose and fructose transport, it also changes during the

progress of fermentation (Berthels et al., 2004) and may be part cause and part effect

of different ethanol tolerance of various strains.

A further possibility is linked with nitrogen utilisation and availability. Sugar

transporters have been shown to be turned over quickly (t½ ~5 hours) compared to

other proteins (Salmon et al., 1993) thereby creating a demand for active protein

synthesis and assimilable nitrogen. In every case of nitrogen supplementation to

nitrogen starved cultures, fructose consumption was enhanced to a greater extent than

glucose (Berthels et al., 2004). Different strains have been shown to utilise nitrogen to

34

different extents (Jiranek et al., 1995). Perhaps therefore as fermentation progresses

strains with higher nitrogen demands experience greater or earlier restrictions on

assimilable nitrogen availability and thus their ability to maintain fructose transport,

thereby resulting in a higher rate of glucose transport.

Available evidence affirms that hexose utilisation depends on numerous

environmental and biological parameters. Thus determining the precise basis for the

glucose/fructose discrepancy in order to target efforts to reduce such differences and

presumably improve strain performance and wine composition are difficult. In this

study 20 commercially available wine strains were chosen for characterization of their

fructose utilisation proprieties. Such data was sought so as to begin to determine the

relationship between abilities around utilisation of individual sugars and relative

fermentative performance. Fermentations were conducted in a chemically defined

grape juice medium (CDGJM; Henschke and Jiranek, 1993) with two different sugar

compositions. Either an equimolar mix of glucose and fructose or else a medium

containing fructose as the only sugar were used. Various parameters related to total

fermentation duration and kinetics of fructose and/or glucose utilisation were

measured and the resulting values compared though mathematical calculations to

identify relationships.

Materials and methods

Strains and maintenance

A total of 20 commercial strains of S. cerevisiae were selected as representative of

commonly used wine yeast and, where possible, chosen with consideration of

published information about their ability to consume fructose compared to glucose.

(Berthels et al., 2004; Guillaume et al., 2007). The strains used where: B, UCD522,

Cru-Blanc, Primeur, AWRI 350, AWRI 796, AWRI 1503, Elegance (AB Mauri,

Sydney, Australia); EC1118, V1116, D254, W27, BM45, Syrah, Bordeaux Red, S6U,

Uvaferm 43 (Lallemand, Montreal, Canada); VIN13, NT202 (Anchor Yeast, Cape

Town, South Africa), Fermichamp (DSM, Netherland). Strains were collected

aseptically from active dried commercial preparations, re-hydrated in sterile water (20

min) and inoculated into YEPD medium

35

(20 g/l D-glucose, 10 g/l yeast extract, 20 g/l Bacto peptone) in a flask (air/liquid ratio

> 66%) before overnight incubation at 28°C with shaking at 180 rpm. Cultures were

then streaked onto YEPD agar plates and grown overnight at 28°C to check for purity.

Multiple representative colonies were inoculated into 25 ml of YEPD broth and

grown as above. These served as starter cultures for the fermentation experiments

detailed below or, with the combination of 1 ml of culture with 0.5 ml of sterile 80%

(v/v) glycerol, enabled long term storage at -80°C.

Fermentation experiments

Starter cultures were used to generate pre-cultures, which in turn were used to

inoculate fermentation experiments as detailed elsewhere (McBryde et al., 2006). For

each of the 20 strains, fermentations were performed in order to define their sugar

utilisation and fermentation kinetics. To do this, two different formulations of a

CDGJM were used. The first was representative of a typical grape juice in that

glucose and fructose were supplied in equimolar amounts to a combined total of 230

g/L. For the second condition an equivalent amount of sugar was supplied but as

fructose only. In each case 600 mg N/l (as amino acids) were used and the triplicate

fermentations were incubated at 28°C with shaking at 160 rpm.

Fermentation progress was estimated from the Brix value of clarified (14,000 rpm, 2

min) culture samples and fermentation completion (< 2.5 g/l) determined using

Clinitest® tablets (Bayer). Supernatants were stored at -20°C for subsequent

determination of residual sugar content by an enzymatic method (Boehringer-

Mannheim, 1989) adapted for 48 well plates.

Analyses

For every strain, the values obtained from the determination of individual sugars in

each medium where considered separately. Four datasets with regard to sugar

concentration during the progress of fermentation were obtained, i.e. for glucose,

fructose and glucose + fructose in the mixed sugar condition as well as fructose from

the fructose-only media. These datasets were used to plot the respective sugar

utilization curves for each strain. In this way, 20 charts were obtained each containing

four areas delimited by the fermentation curves for the four sugars or combinations

(Figure 2.1).

36

Figure 2.1 – Composite plots showing sugar utilisation profiles of 20 commercial wine strains from two growth media. Yeast were grown in CDGJM containing either 230 g/l of fructose only (top curve in each plot) or 115 g/l each of glucose and fructose (1:1; bottom 3 curves in each plot). Curves are derived from the mean of triplicate fermentations with error bars indicating standard deviation. Moving from top to bottom, the curves correspond to 1) fructose fermentation in a fructose-only medium, 2) combined glucose and fructose fermentation from the mixed sugar medium, 3) fructose fermentation from the mixed sugar medium and 4) glucose fermentation from the mixed sugar medium. The hatched area between the top curve (fructose fermentation in fructose-only medium) and the curve below (combined glucose and fructose fermentation from the mixed sugar medium) highlights the difference in total fermentation profile in the two media. The area corresponding to glucose fermentation from the mixed sugar medium is highlighted in dark grey, while the extent to which fructose utilisation from the mixed sugar condition was delayed compared to glucose is highlighted in light grey.

B

0 50 100 1500

50

100

150

200

250

PRIMEUR

0 50 100 1500

50

100

150

200

250

UCD 522

0 50 100 1500

50

100

150

200

250

AWRI 1503

0 50 100 1500

50

100

150

200

250

AWRI 796

0 50 100 1500

50

100

150

200

250

CRU-BLANC

0 50 100 1500

50

100

150

200

250

EC 1118

0 50 100 1500

50

100

150

200

250

AWRI 350

0 50 100 1500

50

100

150

200

250

V1116

0 50 100 1500

50

100

150

200

250

D 254

0 50 100 1500

50

100

150

200

250

W 27

0 50 100 1500

50

100

150

200

250

BM 45

0 50 100 1500

50

100

150

200

250

SYRAH

0 50 100 1500

50

100

150

200

250

BORDEAUX RED

0 50 100 1500

50

100

150

200

250

S6U

0 50 100 1500

50

100

150

200

250

FERMICHAMP

0 50 100 1500

50

100

150

200

250

UVAFERM 43

0 50 100 1500

50

100

150

200

250

QA23

0 50 100 1500

50

100

150

200

250

VIN 13

0 50 100 1500

50

100

150

200

250

NT 202

0 50 100 1500

50

100

150

200

250

Suga

r con

cent

ratio

n (g

/l)

Time elapsed (hours)

37

The value of these areas was calculated using the composite trapezoid rule (GraphPad

Prism 5 - GraphPad Software Inc., La Jolla, C.A., U.S.A.), a numerical integration

method for approximating an integral or AUC where the function of the curve is

unknown. The value of the AUC is returned in units of the X axis times the units of

the Y axis. Statistical analysis of the data (one-way ANOVA with Dunnet and

Tukey’s multiple comparison post-test analysis) was performed with the same

software.

Results

Fermentation duration

The times required for complete utilisation of glucose and/or fructose were defined

for 20 commercial strains of yeast and are summarised in Table 2.1. For most strains,

fermentation of each medium was completed (i.e. < 2.5 g/l residual sugar) in less than

150 h. The only exception was Maurivin strain B, for which the mixed sugar

fermentation required 151 hrs and the fructose-only condition failed to complete

within this preset maximum duration. In every case the utilisation of fructose lagged

behind that of glucose. Similarly fermentation of fructose took longer than the

fermentation of an equivalent amount of mixed sugars. Glucose depletion from the

mixed sugar condition occurred in between 60 h (UCD522) and 115 h (B) whereas

fructose depletion occurred within 75 h (UCD522) and 151 h (B). Since fructose

depletion always took longer than that of glucose, the time for the former to occur

also equated to the total duration of the mixed sugar fermentations. By comparison,

where complete utilisation of 230 g/l of fructose was seen, a total of between 94 h

(UCD522) and 134 h (BM45) was required for this eventuality. Thus the complete

fermentation of the fructose-only medium took longer than the equivalent mixed sugar

fermentation. These observations confirm the apparent glucophilicity of

Saccharomyces yeast, and suggest that the utilisation of fructose is the rate-limiting

step and that the time taken for this defines the total duration of fermentation.

The inclusion of a medium containing fructose as the only sugar provided an

opportunity to determine whether this condition was predictive of fructose utilisation

capabilities in the mixed sugar medium. The duration of a fructose-only fermentation

38

Table 2.1. Duration of sugar catabolism and fermentation for 20 commercial yeast strains during growth in media containing glucose and fructose or only fructose to a total concentration of 230 g/L. Values are the average of triplicate fermentations ± standard deviation (SD).

Strain

Duration (h) Glucose

(mixed sugars) Fructose or Total sugars

(mixed sugars) Fructose

(fructose only)

Average SD Average SD Average SD B 115.3 12.5 151.3 20.2 DNC1 DNC BORDEAUX RED 73.0 0.0 114.5 5.2 120.3 4.9 AWRI 350 77.2 4.9 101.0 0.0 115.3 3.7 BM45 75.5 4.3 99.2 2.3 134.3 7.2 UVAFERM 43 80.0 0.0 97.0 2.6 106.3 2.3 W27 73.0 0.0 96.5 0.0 104.5 0.0 D254 71.5 0.0 95.0 0.0 109.5 2.6 PRIMEUR 70.0 0.0 94.3 4.6 101.0 0.0 FERMICHAMP 80.0 0.0 94.0 0.0 99.8 1.1 QA23 80.0 0.0 94.0 0.0 102.8 3.7 VIN13 72.0 0.0 94.0 0.0 105.0 0.0 NT202 77.3 4.6 94.0 0.0 112.3 2.9 S6U 73.0 0.0 93.5 5.2 110.0 2.6 SYRAH 64.5 0.0 87.5 0.0 104.5 0.0 V1116 71.5 0.0 86.5 0.0 104.7 3.2 AWRI 796 67.7 4.0 86.0 1.7 102.3 1.1 CRU-BLANC 66.2 4.6 85.2 2.3 95.0 0.0 AWRI 1503 67.7 4.0 84.0 5.2 99.7 2.3 EC1118 71.5 0.0 82.5 0.0 95.0 0.0 UCD522 60.3 4.6 74.6 4.0 94.3 4.6

1 DNC, Fermentations did not complete.

39

was not strongly correlated (R2 = 0.524) with the time required for fructose depletion

from the mixed sugar medium. This indicates that the duration of utilisation of

fructose as sole carbon source is not necessarily a good predictor of the duration of

fructose utilisation in the presence of glucose.

Since fermentation duration alone did not fully describe the performance of individual

strains, other features of the pattern of sugar utilisation were sought by plotting the

utilisation data (Figure 2.1). Several points are evident when the data were presented

graphically. Firstly, all fermentations showed some degree of an initial lag, followed

by an extended period of rapid fermentation before a progressive slowing toward the

complete catabolism of sugars. Thus these periods would collectively and to different

degrees contribute to the overall times required for catabolism of individual sugars.

Secondly, differences existed between strains in terms of the extent to which the

pattern of utilisation of a given sugar(s) reflected that of another sugar(s). Such

differences are highlighted by shading of the area below each curve which

distinguishes it from the adjacent curves (Figure 2.1). Using this approach and

ignoring total fermentation duration, it is evident that strains differ in the extent to

which they preferentially utilised glucose compared to fructose in the mixed sugar

condition. Consequently, Fermichamp has the smallest discrepancy between the two

curves (Figure 2.1, light gray area), whereas strains such as Primeur, AWRI 796 and

Bordeaux Red appear to have the largest discrepancy. The other key observation is

that the difference between the profile for the combined catabolism of glucose and

fructose compared with the pattern of fructose utilisation from the single sugar

medium, also highlights differences between strains (Figure 2.1, hatched area).

Accordingly, strains such as AWRI 796, Syrah and BM45 have a large discrepancy,

while Fermichamp, Cru-Blanc and Uvaferm 43 do not (Figure 2.1).

Application of the composite trapezoid rule

Casual observation suggests that there are no examples of particularly anomalous

behaviour amongst the strains. That is, there are no instances where a given

fermentation duration is in fact not the result of a relatively consistent progression of

fermentation, but instead the result of a protracted commencement followed by a very

rapid completion.

40

However, the validity of this notion was best assessed through a closer, more

quantitative, examination of the data. For this reason much of the subsequent analysis

and strain comparison considers the AUC of utilisation of each sugar as determined

using the composite trapezoid rule (Table 2.2).

If, in fact, the rate of fermentation was consistent, then there should be a relationship

between the AUC of utilisation for that particular sugar and the time taken for the

sugar to be depleted. This was not always the case (N.B. Maurivin strain B was

excluded from this comparison as it was a clear outlier). Thus a comparison of the

AUC of glucose (Table 2.2) was somewhat correlated to the time to glucose depletion

(R2 = 0.715). However, the correlation for the analogous values for fructose in the

mixed sugar condition yielded an R2 value of 0.645. These findings are explained by

the fact that in the case of the latter, there are examples of strains (e.g. Bordeaux Red

and NT202) with similar AUC (i.e. 5833 and 5918) but markedly different times for

fructose catabolism (i.e. 115 and 94 h). The reverse was also evident, that is similar

durations of fructose utilisation (i.e. 101 and 99 h) but different areas under the

corresponding fructose utilisation curve (6016 and 5404) for AWRI 350 and MB45,

respectively. These examples highlight the complexity of fermentation phenotypes

and the difficulty of comparing strains. When considering the combined catabolism of

glucose and fructose from the mixed sugar fermentation, a poor correlation (R2 =

0.461) between the value for the AUC and the total duration of catabolism was again

observed, as was the case for the fructose only condition (R2 = 0.482). Given these

obvious complexities, we therefore suggest that compared to duration alone, the area

under the sugar utilisation curve provides a more complete account of sugar

utilisation, as it incorporates trends occurring during fermentation.

Utilisation of fructose compared with glucose

Of key interest in this study was the manner in which the kinetics of fructose

utilisation influenced the overall fermentation performance of individual strains. For

this reason we sought to identify a basis for comparison of strains which was

independent of overall fermentation duration. As such we compared each strain in

terms of the ratio between the areas under each of the glucose and fructose utilisation

curves (Table 2.2). All strains exhibited a glucose:fructose ratio which was markedly

41

Table 2.2. Area under the glucose and fructose utilisation curves for fermentations by 20 commercial yeast strains of media containing mixed sugars (glucose and fructose) or only fructose to a total concentration of 230 g/L. Values are the average of triplicate fermentations ± standard deviation (SD).

Area under the fermentation curve (trapezoid rule)

Strains Glucose

(mixed sugars) Fructose

(mixed sugars) Glucose area: Fructose area (mixed sugars)

Total sugars (mixed sugars)

Fructose (fructose only)

Total sugars area (mixed sugars): Fructose area (fructose only) Average SD Average SD Average SD Average SD

B 4677 384 8186 828 0.57 12864 1211 DNC1 - - AWRI 350 3861 77 6015 250 0.64 9876 324 11123 252 0.89 NT 202 3743 180 5918 303 0.63 9662 481 11123 572 0.87 UVAFERM 43 3732 77 5731 88 0.65 9463 159 10201 468 0.93 QA 23 3844 111 5548 261 0.69 9392 371 10550 565 0.89 FERMICHAMP 4047 116 5317 91 0.76 9364 184 10108 184 0.93 BORDEAUX RED 3336 65 5833 64 0.57 9169 129 10026 122 0.91 D 254 3404 52 5535 64 0.61 8939 109 9937 352 0.90 VIN 13 3560 77 5363 68 0.66 8923 140 10311 180 0.87 PRIMEUR 3426 45 5476 157 0.63 8902 159 10082 136 0.88 EC 1118 3646 73 5201 216 0.70 8847 146 9456 111 0.94 S6U 3650 16 5180 123 0.70 8830 110 9857 53 0.90 W27 3400 55 5403 84 0.63 8804 139 9709 105 0.91 BM 45 3385 113 5405 76 0.63 8790 127 11065 330 0.79 V1116 3410 95 5201 46 0.66 8611 98 9789 289 0.88 CRU-BLANC 3516 92 5052 107 0.70 8568 178 8946 119 0.96 AWRI 796 3297 7 5135 267 0.64 8432 272 10227 495 0.82 AWRI 1503 3394 8 5017 252 0.68 8411 244 9944 224 0.85 SYRAH 3216 25 4832 20 0.67 8048 25 9915 27 0.81 UCD 522 3001 70 4650 191 0.65 7651 258 9067 179 0.84 1DNC, Fermentations did not complete

42

less than 1, clearly demonstrating the ability of these strains to preferentially utilise

glucose over fructose. The strains ranged in terms of this ratio, such that Bordeaux

Red exhibited a value of 0.57 compared to 0.76 for Fermichamp. The ratio of AUC of

the utilisation of glucose compared with fructose in the mixed sugar condition, is

proposed to give an indication of the relative glucophilicity (increased fructose or

poor glucose consumption) of individual strains. Based on this concept, we averaged

the values of the ratio between glucose and fructose areas for the 19 strains (B

excluded) and used the resulting average as the reference ratio (0.66). Thus, it was

possible to rank all the strains based on their individual glucose:fructose ratio

deviation from the mean of all ratios (Figure 2.2). Two groups of strains can be

identified, the first represents the more fructose efficient strains (negative values), and

the second represents the less fructose efficient strains (positive values). In this way

Fermichamp appears to be either more fructophilic or less efficient at utilising

glucose. Additionally, all strains were compared against each other to determine any

significant differences (Table 2.3). Most strikingly Fermichamp and Bordeaux Red

were the two strains with most significant differences to the other strains.

As was shown for the duration of fermentation, further evidence for the poor

relationship between performances in one condition (medium) compared to another is

provided by comparison of the ratio of the AUC of utilisation of both glucose and

fructose compared to fructose alone (Table 2.2). Such a comparison yielded a

correlation of R2 = 0.449. Further consideration of strain performance was therefore

limited to a comparison of the relative utilisation of glucose and fructose in the mixed

sugar condition.

Discussion

Developing a valid method to assess glucose and fructose utilization during alcoholic

fermentation is not a straightforward matter. Several authors have proposed criteria

for this, but what is clear, and supported by our data, is that the method must be

independent of the overall rate or duration of the fermentation. Also, to simplify its

application, especially for comparison of many strains, the approach must minimise

the need for real-time analysis and, given the dynamic nature of fermentation by

different strains, must incorporate trends from as many stages of the fermentation as

43

Figure 2.2 – Differences from the mean (baseline) of ratios between glucose and fructose areas in the mixed media condition (as reported in Table 2.3). The values were calculated with one-way ANOVA Dunnett’s multiple comparison post-test at 95% confidence interval. FERMICHAMP and BORDEAUX RED (dark bars) were the only two strains showing a significant difference from the average. Table 2.3. One-way ANOVA with Tukey’s multiple comparison post-test analysis of 19 commercial yeast strains (and average of all strains) in terms of their ratio of the area under the curve of glucose utilisation and fructose utilisation from media containing each sugar at 115 g/L (as reported in Table 2.2).

Strains

S6U

EC

1118

C

RU

-BLA

NC

Q

A23

A

WR

I 150

3 SY

RA

H

VIN

13

V11

16

UV

AFE

RM

43

UC

D 5

22

AW

RI 7

96

AW

RI 3

50

NT

202

W 2

7 B

M 4

5 PR

IMEU

R

D 2

54

BO

RD

EUX

RED

A

LL

FERMICHAMP - - • • •• ••• ••• ••• ••• ••• ••• ••• ••• ••• ••• ••• ••• ••• ••• S6U - - - - - - - - - • • •• •• •• •• ••• ••• -

EC1118 - - - - - - - - - • •• •• •• •• ••• ••• - CRU-BLANC - - - - - - - - - • • •• •• •• ••• -

QA 23 - - - - - - - - • • • • •• ••• - AWRI 1503 - - - - - - - - - - - • ••• -

SYRAH - - - - - - - - - - - ••• - VIN 13 - - - - - - - - - - ••• - V1116 - - - - - - - - - ••• -

UVAFERM 43 - - - - - - - - •• - UCD 522 - - - - - - - •• -

AWRI 796 - - - - - - •• - AWRI 350 - - - - - •• -

NT 202 - - - - • - W 27 - - - - -

BM 45 - - - - PRIMEUR - - -

D 254 - - BORDEAUX RED •••

Significant differences were calculated at p value of < 0.05 (•), 0.01 (••) or 0.001 (•••). Non-significant differences also indicated (-).

-0.1

-0.06

-0.02

0.02

0.06

0.1

FER

MIC

HA

MP

S6U

EC11

18

CR

U-B

LAN

C

QA

23

AW

RI 1

503

SYR

AH

VIN

13

V11

16

UV

AFE

RM

43

UC

D 5

22

AW

RI 7

96

AW

RI 3

50

NT

202

W27

BM

45

PRIM

EUR

D 2

54

BO

RD

EAU

X R

ED

Diff

ernc

e fr

om m

ean

44

possible. Thus, even if the second half of the fermentation is the most critical, it is not

appropriate to ignore overall fermentation performance up to this point. The method

suggested from findings of the present study takes into account all these

considerations.

The analysis of the fermentation performance of the 20 strains revealed differences

between some strains to be as much as 2-fold, which equates to about 3 days under

our conditions. Such a difference is likely to have dramatic impacts on winery

throughput and juice processing, particularly at the height of vintage, and no doubt

forms an important criterion in the selection of strains by winemakers. In terms of the

profiles of sugar utilisation, some common trends were seen. Thus, after an initial lag,

sugar utilisation increased markedly before this phase was followed by one of a

slowing of utilisation. In addition an ordered utilisation was evident such that glucose

was removed more rapidly than fructose. Similar reports of glucophilicity in industrial

strains have been made by several authors (Júnior et al., 2008; Meneses et al., 2002;

Meneses and Jiranek, 2002; Tronchoni et al., 2009; Wang et al., 2004). Whilst the

fermentation of a fructose medium was slower compared to an equivalent amount of

mixed sugars, as reported recently (Júnior et al., 2008), it is interesting that the

relative strain performance in the former was not an effective predictor of

performance in the mixed sugar condition.

A recent study of the ability of yeast to grow in a fructose-only medium (Arroyo-

Lopez et al., 2009), compared the area under growth curves to estimate preference or

tolerance of different Saccharomyces yeast for fructose. This method was suggested

as a possible tool for initial screening of yeast. However, with the caveat that our

study examined sugar utilisation rather than growth, the mismatch we observed

between a glucose and fructose mixture and that of fructose alone (R2 = 0.083)

suggests it unlikely that a fructose-only medium will be useful to screen for

performance in an extended mixed sugar fermentation. The presence of glucose is

highly influential on fructose metabolism (Gonçalves et al., 2000; Júnior et al., 2008;

Karpel et al., 2008; Luyten et al., 2002; Perez et al., 2005; Ramos et al., 1988;

Salmon, 1989; Varela et al., 2005; Verwaal et al., 2002). At this point it is not

possible to state how glucose influenced fructose consumption. The effect may be

elicited at the level of the membrane (transport and sensing) or phosphorylation

during the first steps of glycolysis, through a higher Km for fructose compared to

45

glucose. Thus, further work is necessary to better explain the complex interaction

between yeast and the simultaneous presence of two or more sugars. For the moment,

however, it is possible to affirm that this phenomenon is highly strain dependent and

that fermentation performance in presence or absence of glucose can vary

considerably between strains.

In attempting to define fermentation curves of different yeast strains in a mixed sugar

medium, others have developed equations to fit fermentation profiles (Wang et al.,

2004). We chose not to do this, given the high frequency of sampling and good

agreement between replicates in our study. Instead, we used the composite trapezoid

rule to determine the AUC describing the fermentation of each sugar in either

medium. The result is a bi-dimensional measurement of fermentation, which relates

residual sugar concentration and duration of fermentation. This would appear to be the

most inclusive approach to defining fermentation performance used to date. The

subsequent comparison of areas under each of the fructose and glucose curves

therefore provides an overall indication of the sugar affinity of each strain, taking into

account all stages of fermentation. Importantly, the ratio of values derived using this

approach also provides a convenient means for normalising data for strains that

require different times to complete fermentations.

Other approaches to describe different utilisation rates of glucose and fructose whilst

normalising data for different fermentation durations have been described. Berthels

and colleagues (2004) calculated the ratio between glucose and fructose at four points

in a fermentation when 20, 30, 40 and 50% of the glucose had been consumed. Over

this part of the fermentation they observed a linear increase in the glucose:fructose

discrepancy. Moreover they were able to sort strains according to the slope of the

increase in glucose:fructose discrepancy. The reason for selecting four specific

residual glucose concentrations as the points for comparing all strains is not made

clear, but presumably other points within this window would suffice. If not, following

such a prescriptive approach would require a possibly unmanageable high degree of

frequent and real-time glucose quantitation.

As an alternative to such real-time analysis, Guillaume et al. (2007) plotted

fermentation according to g/l of residual sugar normalised against g/l of CO2 released,

the latter being easily determine by weight loss measurements. With this

46

approach they were able to graphically rank two strains based on their different

pattern of fructose and glucose utilization and thereby differentiate the strains. A

shortcoming of the work arises from the fact that the curves were mathematically

fitted and therefore clearly approximations. In a similar approach Dumont and co-

workers (2009) introduced a fructophilic index as a criterion by which to describe the

abilities of yeast to consume fructose compared to glucose. These workers focussed

on the area between the fermentation curves for glucose vs fructose, in the latter ⅔rd of

the fermentation curve (also expressed as g/l of residual sugars vs g/l of CO2

released). Accordingly, strains showing the lowest value, due to smallest area of

differences between the glucose and fructose curves were said to be fructophilic and

were presumed to perform better in situations with high fructose concentrations.

Finally, Tronchoni and colleagues (2009) fitted sugar consumption curves with

various mathematical equations and achieved R2 values of 0.95 and higher. This

enabled them to confidentially estimate the time necessary for different strains to

consume 50% and 100% of glucose and fructose, and, in turn, the residual fructose

concentration at these points. Thus ultimately the comparison between strains was at

one or two time points and did not consider the characteristics of the fermentation

beyond these.

From our results, it was encouraging to observe that Fermichamp showed the highest

ratio between glucose and fructose AUC (0.76). This result agrees with the finding

described by Guillaume and collaborators (2007) about the exceptional ability of this

strain to consume fructose. Similarly, Berthels and colleagues (2004) described the

discrepancy in glucose and fructose for several strains, including EC1118, VIN13 and

Bordeaux Red. Similar results for these three strains have been found in our study

where Bordeaux Red was the slowest fructose fermenter, VIN13 medium and EC1118

one of the fastest. These similarities with other studies increase the confidence of

considering our glucose/fructose comparative approach as a valid method in

describing relative sugar consumption profile during fermentation. Moreover it was

possible to rank 19 commercial wine strains according to their ability to consume

fructose in relation to glucose and identify which strain was significantly different

from the others in terms of fructophilicity (Table 2.3, Figure 2.2).

47

Therefore Fermichamp and Bordeaux red are located at the opposite ends of the chart

and they are the only two strains significantly different from the mean. Other strains

such as S6U and EC1118 are significantly different from most of the strains belonging

to the less fructophilic group, while strains such as D254 and Primeur were

significantly different from most of the strains grouped as the most fructophilic.

In this study, the sampling frequency produced comprehensive fermentation curves

and obviated the need for curve fitting and limitations arising out of this means of

approximation. The post fermentation determination of residual sugars reduced the

analytical burden during the fermentation and the calculation of the AUC is the most

comprehensive encapsulation of all aspects of the fermentation performance of each

strain. Finally, the latter benefit can be said to apply to the ratio of the areas of the

glucose and fructose curves, whilst such ratios also have the advantage of normalising

performance of strains with regard to fermentation duration.

Conclusion

The study proposes a novel approach (AUC) to determine fermentation performances

in wine yeast strains, with particular attention in identifying a methodology to

describe the fructophilicity of the strains. Although one of the goals of the study was

to use a medium containing fructose as only carbon source to predict relative

preference for fructose by strains and to relate this to fermentation performance, the

complexity of glucose and fructose utilization for individual strains remains too great.

Further work is warranted to ascertain the significance of fructose utilising capability

to overall fermentation behaviour and wine quality, and thereby to optimize strain

development programs.

Acknowledgements

This project was supported by Australia’s grape growers and winemakers through

their investment body, the Grape and Wine Research and Development Corporation,

with matching funds from the Australian Government. We would also like to thank

Frank Schmid and Simon Schmidt for their comments during the preparation of this

manuscript. The AWRI and UofA are part of the Wine Innovation Cluster, Adelaide,

South Australia.

48

2.2 BASIS OF CANDIDATE STRAIN SELECTION

In the investigation of 20 commercially available strains (described in the previous

manuscript), use of the area under the fermentation curve (AUC) rather than the

overall fermentation duration was intended to provide a better characterization of the

relative ability of the strains to utilise fructose and glucose. Thus, the most robust

(rapid) fermenter was not necessarily selected for the adaptive evolution experiment,

but rather a strain with a clear difference in its ability to consume fructose in relation

to glucose (lower ratio of the areas under the glucose vs. fructose fermentation

curves). Such a strain should provide a more suitable target for improving fructose

utilization, where the possible margin of improvement is greater and more appreciable

than when using a strain with a smaller difference between glucose and fructose

fermentation. To aid identification of a strain with these attributes, the ratios of AUC

presented in Table 2.2 in the manuscript, were sorted from the smallest to the largest

in Figures 2.3 and 2.4. From Figure 2.3 it appears clear that a strain such as Bordeaux

Red could be a perfect candidate: it shows a very low ratio between the glucose and

fructose AUC in the mixed sugar medium (0.57), thereby making it a very inefficient

fructose fermenter relatively to glucose.

However, other factors have to be considered. A number of approaches could be used

when imposing a fructose related stress during the adaptive evolution experiment. For

example, an improvement in fructose utilization might be induced by forcing a

population to ferment fructose as the only and limited carbon source in the medium. If

this were the intended stress factor to be use, it appears immediately clear that

Bordeaux Red is not longer a suitable candidate. In fact, as shown in Figure 2.4 it

paradoxically ferments sugar (in a fructose only medium) quite fast in relation to

fermentation in presence of glucose (ratio of total sugar AUC in mixed

medium:fructose AUC in fructose only medium = 0.91). Thus a strain that shows a

small difference in fermentation rates between a medium containing glucose

compared to one without, clearly ferments fructose easily in the absence of glucose. If

this is the case, adaptive evolution will further improve the ability of such a strain to

ferment fructose in absence of glucose. Once the strain is returned to a wine-like

medium, its hexose uptake system would again be submitted to glucose regulation.

49

Figure 2.3 – Ratio between areas under fermentation curves for 19 strains. Glucose area : Fructose area (mixed sugars). N.B. Strain Maurivin B was excluded as it was a clear outlier.

Figure 2.4 - Ratio between areas under fermentation curves for 19 strains. Total sugar area (mixed sugars) : Fructose area (fructose only). N.B. Strain Maurivin B was excluded as it was a clear outlier.

0.5

0.6

0.7

0.8B

OR

DEA

UX

RED

D 2

54

BM

45

PRIM

EUR

W27

NT

202

AW

RI 3

50

AW

RI 7

96

UC

D 5

22

UV

AFE

RM

43

V11

16

VIN

13

SYR

AH

AW

RI 1

503

QA

23

CR

U-B

LAN

C

EC 1

118

S6U

FER

MIC

HA

MP

Rat

io o

f are

as

Strains

Glucose area : Fructose area (mixed sugars)

0.7

0.8

0.9

1

BM

45

SYR

AH

AW

RI 7

96

UC

D 5

22

AW

RI 1

503

VIN

13

NT

202

V11

16

PRIM

EUR

AW

RI 3

50

QA

23

S6U

D 2

54

W27

BO

RD

EAU

X R

ED

FER

MIC

HA

MP

UV

AFE

RM

43

EC 1

118

CR

U-B

LAN

C

Rat

io o

f are

as

Strains

Total sugar area (mixed sugars) : Fructose area (fructose only)

50

The strain might again show slow fructose catabolism compared to glucose, as was

the case before the adaptation. On the other hand, in a strain showing a slow fructose

fermentative rate either with or without the presence of glucose, the slow uptake of

fructose can be attributed to factors other than the presence of glucose. Thus, an

improvement arising from the imposition of a fructose-only condition, could

potentially arise in a different step of the hexose uptake system, independent from

glucose regulation. The potential improvement in fructose catabolism arising in such a

strain, may be more easily retained in a fermentation in the presence of glucose.

From these considerations, a suitable candidate has to demonstrate poor fructose

fermentative abilities in media either with or with glucose. BM45 shows a small AUC

ratio in each of such media (0.63 for the mixed sugar condition and 0.79 for the

fructose only medium compared to the mixed sugar medium). However, this strain

required 134.3 hours to complete fermentation in the fructose only media, the slowest

out of all the 19 strains (Table 2.1). Because adaptive evolution requires an extended

running time, a suitable candidate strain would preferably show a reasonably fast

fermentation rate. Otherwise, the total duration of the evolution experiment could be

excessively extended. Thus, BM45 does not satisfy the above criteria and it was not

used further in this study. For the same reason strain B was previously excluded from

the list of candidate strain (see the manuscript), where it was not able to complete

sugar catabolism after 163 hours. Instead, AWRI 796 appears to be the most suitable

strain. It has a medium-low ratio (0.64) between glucose and fructose AUC in the

mixed sugar medium, and a low ratio (0.82) between the total sugar AUC (mixed

sugar media) and the fructose AUC (fructose only media). Moreover, this strain

shows a medium to fast fermentation rate in the fructose only medium, requiring

102.3 hours to catabolise all the sugar.

51

CHAPTER 3

ESTABLISHING AN ADAPTIVE EVOLUTION STRATEGY USING CONTINUOUS CULTURE TO

GENERATE FRUCTOPHILIC GENOTYPES

52

This chapter describes the development of a strategy to generate phenotypes with

improved fructose fermentation rates, applying a selective pressure during continues

culture. The previous chapter dealt with identifying a suitable candidate wine yeast

strain for adaptive evolution to generate fructophilic phenotypes. AWRI 796 was

chosen because it had a large difference between glucose and fructose utilization

rates, but an overall medium-fast fermentation rate, either in media containing an

equimolar mixture of glucose and fructose or fructose alone. This chapter describes

the generation of a variant wine yeast with increased fructose consumption rates,

beginning with the preparation of a starting population for an adaptive evolution

strategy using continuous fermentation.

Despite claims of suppliers, commercial packages of yeast are not pure cultures. Thus,

for the purpose of this study it was important to isolate a single clone of AWRI 796,

physiologically representative of the average of the population. This isolate provided

the genetic background for selective improvement and was a reference for

characterization of adaptively evolved isolates.

Genetic variation is the raw material for adaptive evolution (Chambers et al., 2007).

Thus, the more genetic diversity that exists in a population, the more rapidly a

selection pressure will isolate adaptively evolved novel phenotypes. Starting an

adaptive evolution experiment with a clonal population will therefore greatly reduce

the chance of isolating evolved strains with desirable phenotypes in a reasonable time.

Thus, chemical mutagenesis was used to generate genetic variation in the starting

population, to be used in the adaptive evolution experiment described later in this

chapter.

3.1 ISOLATION OF A REPRESENTATIVE SINGLE CLONE OF AWRI 796

To isolate a representative clone of AWRI 796, the strain was plated onto YEPD agar

(Appendix 1 – Method 2a) from the same glycerol stock as used in Chapter 2 (see also

Appendix 1 – Method 1). From these plates, six individual colonies and a mixed

population derived from a patch from a densely growing region of the plate, were

53

transferred into separate 250 ml conical flasks containing 50 ml YEPD broth. The

seven flasks were incubated overnight at 28°C, shaking at 160 rpm. Each was then

inoculated at a rate of 1 x 106 cells/ml into 50 ml CDGJM starter (50 g/l glucose, 50

g/l fructose – Appendix 1, Method 3) in 250 ml conical flasks (air/liquid ratio > 66%)

and incubated overnight at 28°C, shaking at 180 rpm. Each of these cultures was then

used to inoculate 3 replicates of 100 ml CDGJM (115 g/l glucose, 115 g/l fructose –

Appendix 1, Method 3) at a rate of 5 x 106 cells/ml in 250 ml conical flasks fitted with

water filled air locks and with sampling ports. Fermentation progress was monitored

by CO2 loss, weighing flasks every hour using the Multi-Scale Fermentation Facility

(The University of Adelaide – http://www.sciences.adelaide.edu.au/wine/msff/),

which controls the fermentation temperature (30°C), agitates the culture (with

magnetic stir bars) and monitoring weight loss of the flasks, robotically, on high

sensitivity balances.

CO2 losses (Figure 3.1A) were expressed in g of CO2 lost per 100 g of medium

fermented and fitted with a non-linear, third order polynomial fit using GraphPad

Prism (GraphPad Software Inc., La Jolla, C.A., U.S.A.). For the last third of the

fermentation °Brix of clarified (20,800 rcf, 2 min) culture samples was also

monitored. Fermentation completion (< 2.5 g/l of sugar) was determined using

Clinitest® tablets (Bayer). Supernatants were stored at -20°C for subsequent

determination of residual sugar content (Figure 3.1B and C) as described in

(Boehringer-Mannheim, 1989) with final volumes adjusted to 300 µl for analysis in

48 well microtiter plates.

Fermentation progress of six clonal isolates and the mixed population was monitored.

All fermentation curves show a similar initial profile, but some differences become

evident, especially in the last third of the fermentation (Figure 3.1A). However, clone

5 shows a CO2 weight loss profile almost identical to mixed population. This data was

confirmed in the sugar utilization profile, where clone 5 was able to catabolise all of

the sugar in 113.5 hours, which was the same time as required by the mixed

population (Figure 3.1 B and C).

Due to its similarity in fermentation profile to the mixed population sample of AWRI

796, clone 5 was chosen as the parent strain for subsequent adaptive evolution

experiments.

54

0 25 50 75 100 1250.0

2.5

5.0

7.5

10.0

12.5

0

10

20

30

40

50

60

70

80

Mixed populationClone 5Other clones A

B

C

Time (h)

CO

2 - W

eigh

loss

(g/1

00 g

ram

s of m

ediu

m)

Suga

r con

c (g

/l) Figure 3.1 – A - Non linear (third order polynomial – confidence interval 95%, R2 > 0.98 for the fit of all curves) fit for a mixed population and six clones of AWRI 796 (left axis). B – Glucose concentration (plotted on right axis). C – Fructose concentration (right axis). Each curve is plotted from data from the average of three replicate fermentations. Standard deviations are shown as error bars.

55

3.2 INDUCED MUTAGENESIS OF AWRI 796.

Mutagenesis was induced artificially on clone 5 of AWRI 796 using ethyl

methanesulfonate (EMS) using a method adapted from Fink (1970) and described in

Appendix 1 – Method 4. Thirteen samples were collected at 10 minute intervals

following exposure to 45 µl/ml of EMS and the mutagenesis process arrested with

sodium phosphate buffer. Samples were appropriately diluted and plated onto YEPD

plates, while the remaining of the samples was stored in glycerol at -80°C. From

YEPD plates viable cell counts was determined (see Appendix 1 – Method 4). As can

be seen residual viability decreased progressively with time of exposure to EMS, such

that no viable cells were individuated after 120 minutes (Figure 3.2).

The population treated with EMS for 50 minutes (EMS5), which generated ≈ 40%

death (60% survival), was chosen as the starting population for an adaptive evolution

experiment. Based on past experiences in the laboratory, this population was

anticipated to have a substantial amount of genetic variation.

3.3 DEFINING EXPERIMENTAL CONDITIONS FOR ADAPTIVE EVOLUTION.

As previously mentioned (Chapter 1), there are essentially two different approaches

for adaptive evolution of microorganisms in the laboratory: sequential batch or

continuous culture. Continuous fermentation was the strategy chosen for this work as

it considerably reduced the running time of the experiment. In fact, maintaining a

population in the exponential phase eliminates the time necessary for the complete

depletion of sugar from the media and the sequential inoculation into a fresh batch

ferment. In previous adaptive evolution experiments, McBryde and co-workers (2006)

using similar conditions to those required for the experiment described here,

sequential batch proliferation over 250 generations of a population of wine strain of S.

cerevisiae required 198 days. To achieve the same number of generations in a

chemostat (with a dilution rate of 0.2 h-1) should take approximately 35 days. While it

is recognised that the possibility of affecting the oenological properties of the evolved

56

0 40 80 1200

20

40

60

80

100

Exposure time (min)

Surv

ival

(%)

Figure 3.2 – Survival in a population of clone 5 after varying time lengths of exposure to EMS.

57

strains may exist, a continuous fermentation allows for shorter experiments. This

therefore decrease in running time allows completion of several adaptive evolution

experiments in the same amount of time required for a single sequential batch

approach. This allows trialling of several strategies to generate strains with improved

phenotypes.

Before commencing a long term adaptive evolution experiment it is essential to

establish the optimal conditions for selection of the desired phonotype. This includes:

determining the ‘strength’ of selective pressure, optimal cell density, dilution rate and

composition of feed medium.

Determination of the optimal level of stress for adaptive evolution is critical to drive

selection of the desired phenotype (Butler et al., 1996). It was reasoned that having

fructose as the only sugar in the medium, and maintaining it at a limiting

concentration would be an appropriate selective pressure to generate adaptations to

improve the efficiency of fructose utilization. Using a chemostat which maintains a

continuous low level of fructose provided an effective means of achieving this. An

absence of fructose in the waste medium was taken as an indication that the

concentration of fructose supplied to the bioreactor was limiting. In this environment,

the most efficient fructose utilisers should be fittest.

Maintaining an appropriate cell density is crucial to this approach (Paquin and Adams,

1983; Wahl and Krakauer, 2000; Wick et al., 2002) and this can be achieved by

varying the amount of carbon supplied to the chemostat (Lane et al., 1999). In this

context, an appropriate level of selective pressure (i.e. limiting sugar concentration)

will impose sufficient stress to reduce growth rate without compromising cell density.

For this reason it was important to understand the yeast’s physiological response and

tolerance to the intended stress conditions. The data obtained from previous

experiments (Chapter 2) was not sufficient for this. Thus, information on the

maximum fructose consumption rate and associated growth rate for EMS5 was

sought. Preliminary experiments were performed to estimate the conditions for a

continuous culture experiment in which fructose was limiting (data not shown). From

these initial experiments the following conditions were thought to provide a

reasonable starting point: dilution rate = 0.2 h-1, volume of the fermenting population

58

= 500 ml, cell density = 5 x 107 cells/ml, fructose to be supplied to the population per

hour = 2.13 g/l/h.

In the first experiment using conditions determinate from the above preliminary work,

2.13 g/l/h of fructose was supplied initially at a dilution rate of 0.1 h-1. For this

dilution rate the flow rate was equal to 50 ml/h, and the fructose concentration in the

feed was 20 g/l (Figure 3.3). Under these experimental conditions some fructose

remained unfermented in the exhaust medium (data not shown) indicating that

fructose was not limiting. Additionally, the cell density was around 1.5x108 cells/ml.

Over the first 420 hours of the experiment, the fructose concentration in the feed was

progressively decreased to 15, 10 and then 4 g/l. At the same time the dilution rate

was gradually increased to 0.2 h-1 (corresponding to a flow rate of 100 ml/hour).

Decreasing the amount of fructose to 15 g/l in the feed resulted in a complete

depletion of the fructose from the exhaust (i.e. fructose concentration below

detectable level) indicating that conditions were limiting (data not shown). The

fructose concentration was further decreased and flow rate increased, until the target

value (according to previous experimental experience) of 5 x 107 cell/ml was reached.

To maintain this cell density at a dilution rate of 0.2 h-1 it was necessary to maintain a

fructose concentration in the feed medium of 4 g/l. This preliminary continuous

fermentation defined experimental conditions thought to be appropriate to generate

adaptively evolved phenotypes showing improved fructophilic characteristics.

Adaptive evolution to generate fructophilic wine yeast

An aliquot of frozen cells of EMS5 from a -80°C glycerol stock, was streaked onto

YEPF (see Appendix 1 Method 2b) and grown overnight at 30°C. A patch from a

densely growing region of the plate was then inoculated into liquid YEPF and

incubated overnight at 30°C with agitation on a shaker at 160 rpm. A continuous

culture was then set up by inoculating from the overnight culture 105 cells/ml into

CDGJM (500 ml with 4 g/l fructose – Appendix 1, Method 3). All continuous culture

experiments were conducted in a BIOSTAT® A plus (Sartorius BBI System GmbH)

equipped with a 1.0 litre fermentation vessel and controlled using the MFCS/DA A

plus 2.1 software (Sartorius BBI System GmbH).The medium was supplemented with

59

0 500 1000 15000

5.0�10 7

1.0�10 8

1.5�10 8

0

50

100

50 100 150 200 250 300 350 38520

15 10 420

Time (hours)Number of generations

Cel

l den

sity

(cel

ls/m

l) -�

Flow rate (m

l/hour) -�

Figure 3.3 – Establishing suitable conditions for adaptive evolution to generate fructophilic phenotypes. Flow rate (•) is on the right axis; cell density (▪) is on the left axis, with duration of the experiment (hours) and the relative number of generations on the ‘x’ axis. There is a progressive decrease in fructose concentration in the feedstock (20, 15, 10 and 4 g/l – values reported on the top left corner of the chart). At the same time there was a progressive increment in the flow rate. The combination of these two phenomena resulted in a reduction of total cell number (until it reached a target number at about 5 x 107 cells/ml). Fructose concentration in the exhaust medium was not detectable (except when the feedstock had 20 g/l of fructose).

60

ergosterol and Tween80® (Appendix 1 – Method 3) to supply sterols and fatty acids,

essentially for cell membrane synthesis under anaerobic conditions (Ribereau-Gayon

et al., 2005). CDGJM with the same composition was also used as the feed during the

continuous culture experiments. Fermentation temperature was kept at 30°C and cells

were suspended with agitation at 200 rpm. Head space in the fermentation vessel was

saturated with the CO2 produced during the fermentation and pressure released using

a water filled air lock.

Fermentation was started as a batch culture and after 24 hours (cell density ≈ 5 x 107

cells/ml) the feed pump was started, with an initial dilution rate of 0.08 h-1. At the

same time the waste pump was started at the same rate as the feed pump, ensuring a

constant culture volume at ≈ 500 ml. Over the following 220 hours of the experiment,

the dilution rate was increased progressively from 0.08 h-1 to 0.12, 0.14, 0.15, 0.18 h-1,

until reaching a final value of 0.2 h-1. This dilution rate was kept constant until the end

of the experiment. The duration of the experiment was 1278 hours (≈ 53 days), the

time necessary to allow the multiplication of approximately 350 generations. The

bioreactor was sampled every 48 hours. For each sampling point cell density was

measured (cell count on haemocytometer) and fructose concentration in the exhaust

media was monitored using Clinitest®. Figure 3.4 shows the parameters measured

during the progress of the experiment. It was possible to observe the variation in the

flow rate, maintained constant after 220 hours and the moderate fluctuation of the cell

density (between 4.05 x 107 and 6.75 x 107 cells/ml). Fructose was not detectable in

the fermentation vessel (data not shown). For characterization of mutants, 1 ml

samples of the population were collected every 50 generations and centrifuged 20,800

rcf for 1 min. The supernatant was removed and cells resuspended in 1 ml of fresh

YEPF medium and 0.5 ml of sterile 80% (v/v) glycerol to enabled long term storage

at -80°C.

Conclusions

This chapter has described the development of an adaptive evolution strategy using

continuous culture to generate fructophilic phenotypes. A mutagenised population of

AWRI 796, EMS5, was maintained in continuous culture for 350 generations with

61

0 500 1000 15000

2.0�10 7

4.0�10 7

6.0�10 7

8.0�10 7

0

50

100

50 100 150 200 250 300 350Time (hours)

Number of generations

Cel

l den

sity

(cel

ls/m

l) -�

Flow rate (m

l/hour) -�

Figure 3.4 – Parameters measured during continuous fermentation with limiting sugar (fructose). Flow rate (•) is on the right axis; cell density (▪) is on the left axis.

62

limiting fructose as the only source of sugar. Based on the information reported by

other authors (Paquin and Adams, 1983; Zeyl, 2004) the observed frequency of the

occurrence of a adaptive mutation is somewhere between 1011 and 1012 cell divisions

for non-mutagenised starting populations. Considering that, in the experimental

conditions used in this study the initial population was artificially mutagenised (with

the effect of an increase of the genetic diversity), the population size (around 2.5 x

1010 cells) and the time necessary for a mutation to become fixed, it was decided to

sample the culture every 50 generations. These samples were screened for fructophilic

phenotypes as described in the following chapter.

63

CHAPTER 4

SCREENING OF ISOLATES FROM CONTINUOUS CULTURE TO IDENTIFY FRUCTOPHILIC

GENOTYPES

64

A population of microorganisms forced to live in submerged culture under a specific

stress will eventually acquire isolates that show characteristics of adaptation to that

particular environment. However, the resulting population is not a monoclonal

population derived from the one preceding it, as stipulated in the classical model of

adaptive evolution of an asexual population (Muller, 1932). Other studies have shown

that an evolving population of asexual yeast follow more complex dynamics, resulting

in a mixture of different genotypes, ultimately contributing to different extents to the

total population during the progress of fermentation (Kao and Sherlock, 2008). Thus,

it appears clear that the selection of an adapted clone is not a simple process. As such,

a heterogeneous mixture of different sub-populations will require a reliable

methodology capable of screening several clones to identify those that have adapted

to the selective pressure. The previous chapter described the establishment of a

selective pressure under a continuous fermentation regime, in order to generate

genotypes with improved fructophilicity. In this chapter, the evaluation of isolates

from samples of the populations collected at 50 generation intervals is detailed.

Clearly, the screening of an elevated number of clones would increase the possibility

of identifying superior adapted isolates. In addition, an investigation of many isolates,

would allow broad definition of the dynamics of the evolution of the population.

Particularly, it would help identify when the isolates began to appear and the

percentage of mutants over the total population. Thus, it was fundamental to identify a

methodology suited to the screening of several hundred isolates. Moreover, it was

important to identify a valid parameter on which base the comparison of the

fructophilic ability of the isolates against the parental strain. To better explain these

two concepts, a brief background on the available comparison techniques is given

below.

The methods used traditionally to compare different strains in the laboratory can be

divided into two categories: plates containing solid media and liquid broth culture.

Usually, the use of plates to measure differences between isolates is suitable in

situations where it is possible to observe differences in the morphology of the colonies

(for example differences in colour or shape of the colonies) or other situations where

it is possible to induce limitation, specifically changing the composition of the growth

media. This last situation will determine differences in growth rate of the colonies,

65

leading to difference in colony size, colonies number, days of appearance or even

inhibition of growth of the not-adapted mutants. On the other side, the comparison of

isolates using liquid media is usually adopted when it is important to measure

differences in rate of growth or substrate depletion. The comparison of isolates in

liquid media is usually conduced in small flasks. The maximum number of flasks that

can be managed for each screening exercise limits the number of clones screened. In

addition, a screening experiment usually has to be conducted with at least some

replication for each isolate to increase the statistical significance of the data, and this

represents a further limitation in the number of isolates that can be tested. Moreover,

monitoring parameters in flasks, usually requires more work (according to the

frequency of sampling) than observing events in immobilized colonies: whether

monitoring the evolution of the population or the depletion of substrates, more time

consuming measurements are required.

For the purpose of this study, a screening method based on liquid media appeared to

be the most suitable approach. The intent was to identify isolates showing improved

fructose consumption in wine-like conditions. Thus, the depletion rate of this substrate

from the medium was the parameter thought to be the most pertinent. The use of solid

media was excluded in the early stages of the planning of the screening experiment:

the concern being that parental and mutants would grow well on fructose media and

differences in consumption kinetics of this sugar would not be evident from

observations of the colony growth. Also, because fructose is not an inhibitor or a toxic

substance for S. cerevisiae, it is not possible to create a specific environment that

limits the growth of the not-adapted mutants.

The decision to use laboratory scale fermentations to compare isolates, sought to

resolve other problems also. First, to screen hundreds of isolates in replicate

necessitates more cultures than would be logistically possible. The solution was to

reduce the volume of the population to 200 µl and take advantage of 96 well

microtiter plates as the fermentation vessel. These plates also permit the use of

automated robotic systems to simplify handling and thereby increase the number of

isolates that could be examined.

The use of such small volumes required validation. The relationship between the

performance of strains in 96 microtiter plates compared to large cultures (100 ml) was

66

assessed as part of a collaborative project (manuscript submitted to the Australian

Journal of Grape and Wine Research – see Appendix 2). The key findings were:

1. No difference was observed in the maximum biomass yield and rate of growth

between wine yeast cultures propagated in self-induced anaerobic 100 ml

conical flasks and 96 well microtiter plates (Figure 4.1);

2. Comparable fermentation rates were seen between wine yeast cultures

performed in 100 ml self-induced anaerobic conical flasks and 96 well

microtiter plates. Even if in every case fermentations in microtiter plates were

faster, the relative sugar utilization profiles of the strains examined were

unchanged compared to fermentations in flasks (Figure 4.2).

Thus, the use of microtiter plates became a valid tool to screen large numbers of

isolates during this study. However it was important to identify a parameter that

significantly distinguished between isolates and the parental strain for the ability to

ferment fructose. As described previously in this chapter the principal parameters by

which liquid cultures were compared are the rate of growth of a microorganism and

the rate of substrate consumption. Evaluation of the efficiency of growth on fructose

could conceivably be assessed based on comparisons of rates or extents of biomass

formation or else patterns of fructose removal from the growth medium. Given the

ease by which culture growth can be estimated (i.e. via measurement of optical

density), the validity of this approach was assessed.

Preliminary results showed that no correlation existed between the growth rate and

biomass formation and fermentation rate in a CDGJM containing fructose as sole

carbon source (data not shown). These validation experiments, were conducted

comparing growth rates, yield of biomass and rate of fructose depletion either in 96

well microtiter plates or else 250 ml flasks. A range of sugar content was used

between 4 g/l and 230 g/l of fructose. Moreover, because the final target of the project

was to improve the ability to consume fructose in wine-like conditions (i.e. in the

presence of glucose), the use of a fructose only medium was not thought to be

appropriate. For this reason, other pilot experiments aimed to identify any possible

correlation between the rate of growth and sugar depletion from a medium containing

100 g/l of an equimolar mixture of glucose and fructose were conducted (data not

67

0 50 100 150 2000

2

4

6

SAFMTP

A

time (hours)

dry

cell

weig

ht (g

L-1

)

SAF MTP0.00

0.05

0.10

0.15

0.20

0.25 B

grow

th ra

te (h

-1)

Figure 4.1 – Impact of culture vessel type on culture physiology: Chardonnay juice fermentations using strain AWRI 1493 were performed in self-induced anaerobic flasks (SAF) or microtiter plates (MTP). Growth curves showing biomass accumulation (A) and maximum specific growth rates (B) were estimated from optical density measurements. Error bars show ± standard deviation. Columns show means from two independent experiments. Each treatment was performed in triplicate within each experiment. (From Liccioli et al. – submitted. See Appendix 2 for more details).

Figure 4.2 – Screening of yeast strain fermentation performance and validation using selected strains at a larger scale. The fermentation performances of 15 industrial strains were evaluated in Chardonnay juice using microtiter sacrificial plates (A). One plate was prepared for each time point, each strain was fermented in four wells on each plate. Each time point in (A) is the average sugar concentration from four wells on one plate. Four strains, representing a cross-section of performance profiles, were selected for further characterization using 200 ml air-locked self-induced fermenters (SAF) in the same Chardonnay juice (B). Each time point in B is the average total sugar concentration from 3 fermentations. Error bars indicate ± standard deviation. (From Liccioli et al. – submitted. See Appendix 2 for more details).

0 100 200 3000

50

100

150

200

250

B

time (hours)

tota

l sug

ar (g

/l)

0 50 100 1500

50

100

150

200

250

Strain 1

Strain 4Strain 3Strain 2

Other strains in screen

A

time (hours)

tota

l sug

ar (g

/)

68

shown). Again, the growth curves of the yeast did not reflect the rate of sugar

depletion during these fermentations.

As a result it was concluded that there is no correlation between the growth rate of the

strains used and their ability to ferment sugars. This fact was also confirmed from

Simon Schmidt and collaborators (personal communication), that in different series of

experiments they never founded evidence of this kind of correlation. Thus, despite its

appeal due to the ease with which culture growth can be measured (i.e. as optical

density) the only valid method for comparing the fructophilic nature of isolates from

the adaptive evolution experiment and the parental strain was the direct measurement

of glucose and fructose utilization. The screening of isolates from the sample of the

population collected every 50 generations during the continuous culture experiment is

the topic of the next section.

4.1 SCREENING FOR ISOLATES SHOWING IMPROVED FRUCTOPHILIC ABILITY

As introduced in Chapter 3, clones isolated every 50 generations during the

continuous culture experiment, were investigated to identify the presence of

phenotypes showing improved fructophilic characteristics. Thus, isolates from 7

sequential sampling times were tested. For each point, the fermentative performance

of 54 isolates was compared against the parental strain and a representation of the

mixed population at the same 50 generation sampling point. The screening method

used 96 well microtiter plates with a sacrificial sampling approach (see manuscript –

Appendix 2 and below for a more extensive explanation). Essentially, several copies

of the same plate were prepared such that a fresh plate was used at each time point to

allowing monitoring of sugar depletion profiles, without influencing fermentation

performances in the small volume of the microtiter plates cultures.

Specifically, from -80°C glycerol stocks, cells were streaked onto YEPD plates

(Appendix 1 – Method 2a) and incubated overnight at 30°C. The cells used had been

collected previously from the continuous fermentation experiment at 50 generation

intervals. From the YEPD plates, 54 single colonies were picked and inoculated into

69

500 µl of YEPD broth, incubated without shaking overnight at 30°C. The parental

strain and a mixed population derived from a patch of a densely growing region of the

culture sample streak plate were grown similarily. For each culture, 1 ml of CDGJM

starter medium (25 g/l glucose, 25 g/l fructose) was inoculated with 20 µl of YEDP

culture and incubated at 30°C in 10 ml Falcon tubes without agitation. Once the cells

reached saturation (overnight incubation), each culture was supplemented with 4 ml

of CDGJM containing 50 g/l glucose, 50 g/l fructose (i.e. a 1 in 5 dilution). Ten micro

litres of each suspension were used to inoculate 96 well microtiter plates containing

190 µl of CDGJM (50 g/l glucose, 50 g/l fructose) in each well. The 54 isolates were

distributed over 3 plates and each plate also contained the parental strain and the

mixed population. Each plate was then replicated six times, operation performed with

an automated plate and liquid handling machine (Tecan EVO 150). The dilutions,

gave a dilution factor of 100 for the starter culture. Thus, the average density of the

starter culture was therefore diluted from ≈ 108 cells/ml to around 106 cells/ml. For

each culture, four replicates of the same isolate were inoculated on each 96 well

microtiter plate. Some unutilised wells in the plate contained uninoculated CDGJM as

a control for sterility, while others were used to hold standards when sub-sequentially

performing enzymatic sugar assays described in Boehringer-Mannheim (1989) with

final volumes adjusted to 100 µl for analysis in 96 well microtiter plates. Typical plate

layout is shown in Figure 4.3.

Each of the 6 replicate plates, were sealed with breathable membrane (Breathe-easy,

Diversified Biotech) and placed in an incubator at a controlled temperature (20°C),

humidity (75%) and atmosphere saturated with nitrogen gas (O2 concentration < 1%).

The elevate humidity was intended to limit evaporation from the fermentations, while

the nitrogen atmosphere reduced the influence on fermentation due to oxygen. One of

each of the 6 sacrificial plates was frozen at -20°C (for samples time see Figure 4.4)

every 24 hours, for sub-sequent enzymatic determination of glucose and fructose.

Sugar determination was performed semi-automatically, preparing the enzymatic

reactions using the same liquid handling robot (Tecan EVO 150). An integrated

spectrophotometric plate reader (Tecan M200 Infinite) was used for colorimetric

measurements.

70

Figure 4.3 – Example of plate layout used for screening isolates from continuous culture every 50 generations. Each plate contained 18 isolates in four replicates, as well of the parental strain (PAR), mixed population of the corresponding 50-generational point (MIX), uninoculated wells for checking for contaminations (BLK) and empty wells used subsequently for sugar determinations (boxed area). Each plate was replicated six times for analysis of fermentation performance using a sacrificial sampling approach.

71

Examples of the high throughput analysis of fermentation performance are shown in

Figure 4.4, where the sugars utilization of 18 isolates, the parental strain and the

mixed population are represented. In addition to the curve of total sugar depletion

(Figure 4.4a), utilization of glucose (Figure 4.4b) and fructose (Figure 4.4c) are

shown individually.

A clear differentiation between many of the isolates, the parental strain and the mixed

population is evident (Figure 4.4). Several of the isolates show an improvement in

overall fermentation performance (Figure 4.4a), depleting all sugars in a shorter

period of time compared to the parent. Comparing Figure 4.4a with 4.4b and 4.4c it is

possible to see that the greater contribution to such improvement can principally be

attributed to an improvement of the utilization of fructose. Such an observation

suggested a possible improvement in the fructophilicity of at least some of the

isolates, as per the objective of this study.

The spread in fermentation curves seen across the isolates was typical for each batch

investigated during screening. Thus, in each case, the fermentation kinetics of a few

isolates were faster than the parental strain, while others were slower. However, such

simplistic graphical evaluation of fermentation curves did not provide a detailed

enough comparison of the relative fructophilicity of the isolates. For this reason

differences in fermentation rate were quantified paying particular attention to ability

of isolates to utilize fructose compared to glucose. Once again, determination of the

value of the area under the fermentation curve (described in the manuscript included

in Chapter 2) appeared the most appropriate for this purpose. To eliminate possible

external influence on the fermentative performance due to differences in experimental

or environmental conditions between microtiter plates, the fermentation rate of every

isolate was standardized relative to the parental strain included in each plate. Thus, the

value of the area under the fermentation curve was expressed as a percentage of the

parental strain within every plate. With this approach and in turn the calculation of the

ratio between areas, it was possible to identify and quantify isolates showing

improved overall fermentative performance and to rank them according to

fructophilicity.

The ratio between the area under the fermentation curve of glucose and fructose was

calculated (as described in Chapter 2) for each of 378 isolates and for each of the

72

0 20 40 60 80 1000

50

100

150

A

Isolates

ParentalMixed population

Time elapsed (h)

Tota

l sug

ar c

once

ntra

tion

- GLU

+ F

RU

(g/l)

0 20 40 60 80 1000

20

40

60

80

B

Time elapsed (h)

Glu

cose

con

cent

ratio

n (g

/l)

0 20 40 60 80 1000

20

40

60

80

C

Time elapsed (h)

Glu

cose

con

cent

ratio

n (g

/l)

Figure 4.4 – Example of high throughput analysis of fermentation performance obtained using the sacrificial plate approach. Curves of depletion of total sugar (A), glucose (B) and fructose (C) are shown for 18 isolates, the parental strain and the mixed population. Results are the average of four replicates (standard deviation shown as error bars).

73

parental strain (one for each set of plates = 21 fermentations). The distribution of

these values was examined along the seven groups of isolates of each 50 generation

samples (Figure 4.5). An apparent trend was observed: a progressive increase of the

value of the ratio between glucose and fructose area under curve ratios with the

increasing number of generations of exposure of selective pressure. Moreover, a

comparison with the distribution of the parental strains, which, even if with large

approximation, were mainly distributed within the low values of the ratio, showed a

generally improved fructophilicity of the isolates starting from 200 generations

onwards. However, this parameter alone was not sufficient to assure the identification

of isolates showing improved characteristics. Clones showing improved fructophilic

characteristics but slower overall fermentation performances compared to the parental

strain, were considered not suitable for further studies. To identify candidate isolates

suitable for further characterization, the value of the ratio of the areas under the

fermentation curves of glucose and fructose was combined with the overall

fermentation performance. Thus, it was possible to identify 19 isolates showing

improved fructophilic characteristics and faster overall fermentation performance. For

these 19 clones specifically the ratio of the area under the fermentation curves of

glucose and fructose was seen to have increased to between 0.79 and 0.84, while the

average for the parental strains was 0.72. In every case these clones also showed

values of the area under the overall fermentation curve between 85% and 95% of the

parental value, thereby indicating faster fermentations. These 19 isolates were chosen

for further comparison of fermentation performances (Section 4.2).

A final point to be made in regard to the fermentation rate of the parental strain

related to the proportion of improved isolates. For each of the 50-generational samples

screened, the number of isolates showing a faster fermentation rate compared to the

parental strain was shown to change with the progress of the adaptive evolution

experiment (Figure 4.6). As can be seen, while not statistically significant, there

appeared to be a trend to an increasing proportion of improved isolates toward 250-

300 generations. Beyond this number of generations, the proportion of isolates

improved relative to the parent appeared to decline.

74

0.62

5

0.67

5

0.72

5

0.77

5

0.82

5

0.87

5

0.92

50

5

10

15

20

25

30

50 gen100 gen150 gen200 gen250 gen300 gen350 genParent

Area under the curve of glucose / Area under the curve fructose

Num

ber o

f iso

late

s

Figure 4.5 – Absolute value of the area under the fermentation curve of glucose divided by the value of the area under the curve of fructose as an indication of the fructophilicity of individual isolates. Isolates are grouped (refer to shading) according to the 50-generational sample from which were obtained. The distribution in the performance of replicates of the parental strain is indicated by the red bars.

50 100

150

200

250

300

350

0

20

40

60

80

100TOTGLUFRU

Number of generations elapsed during the AE

Perc

enta

ge o

f iso

late

s sho

win

g a

fast

er ra

te o

fsu

gar u

tiliz

atio

n co

mpa

red

to th

e pa

rent

al st

rain

Figure 4.6 – Number of isolates (expressed as a percentage of the 54 isolates screened from each 50-generational sample) showing faster fermentation rate (smallest area under the fermentation curve) compared to the parental strain for total sugar or glucose and fructose individually. The seven generational points are shown separately. Standard deviation is shown as error bars.

75

4.2 FERMENTATIVE PERFORMANCE OF 19 ISOLATES IDENTIFIED AS CANDIDATES FOR FURTHER CHARACTERISATION

As described in the previous section, 19 isolates were identified as having improved

fructophilicity and overall fermentation rate. A further round of screening was applied

to these strains in the hope of reducing the number of candidates to be asses in the

subsequent large scale evaluation. The experimental approached used was the same as

that used for the initial screen, with the exception that the mixed population was

omitted from the layout of the plate. The data were analysed following the same

procedure described above, that is by using the ratio between the areas under the

fermentation curves of glucose and fructose combined with the overall fermentative

performance.

Although all of the 19 isolates showed a faster fermentation rate compared to the

parental strain in the previous screening, a few of them (6, 10, 15, 16, and 19)

displayed a larger area compared to the parental strain in this experiment (Table 4.1).

For this reason they were excluded from further investigation. On the other side,

isolates 3, 9 and 11 were chosen because they showed an improvement of around 5%

in fermentation performances and a higher ratio between the area under the glucose

and fructose curves. In addition, even if its fermentation performance was almost

equal to the parental strain, isolate 7 was retained for further consideration largely

because of its higher fructophilicity (ratio = 0.682).

4.3 CONCLUSIONS

The identification of a methodology suited to the screening of a large number of

isolates and able to describe their fermentation kinetics, especially the ability to utilise

fructose compared to glucose, enabled identification of four candidates for further

investigation. These candidates, proliferated during continuous fermentation under

selective pressure, showed improved fructophilic characteristic and a faster

fermentation rate compared to the parental strain during the screening experiments.

76

Table 4.1 – Nineteen isolates showing higher fructophilicity compared to the parental strain (ratio between areas under the fermentation curves of glucose and fructose). Strains 6, 10, 15, 16, and 19 (highlighted in gray) showed a slower fermentation rate compared to parent (expressed as a percentage of the area under the curve of total sugar fermentation of the parental strain). Isolates 3, 7, 9 and 11 (highlighted in yellow) were chosen for further characterization.

PAR 16 5 9 13 11 17 1 6 2 3 4 7 8 10 12 14 15 18 19Generations 50 200 200 200 250 250 300 300 350 350 350 350 350 350 350 350 350 350 350Area TOT % 100 126.8 95.2 94.3 95.1 90.0 95.3 96.8 102.2 95.7 94.3 97.4 99.4 98.3 105.6 96.6 98.4 103.5 96.9 101.5Ratio Areas (GLU/FRU) 0.631 0.706 0.650 0.671 0.660 0.661 0.639 0.640 0.658 0.665 0.658 0.662 0.682 0.664 0.638 0.665 0.667 0.631 0.642 0.637

77

However, to confirm and better characterize the improved ability of these clones to

consume fructose, especially in larger scale fermentations and in media including

grape juice, the appropriate strain evaluation was undertaken and is discussed in the

next chapter.

Final comment has to be made about the distributions of isolates showing improved

fructophilicity and the improved fermentation abilities compared to the parental strain.

Even though the data analysed are only derived from a screening experiment and

more work is required before it could be considered a detailed study of evolution in an

evolving population, the trends expressed in Figure 4.5 and 4.6 could reflect a

progressive adaptation of the mutants to the imposed stress.

� The trend shown in Figure 4.5 highlighted the higher frequency of improved

fructophilicity with the progress of the adaptive evolution experiment. The

literature reports that adaptation continues with exposure to selective pressure

over time (Helling et al., 1987; Zeyl, 2005). Even if speculative in this study,

it have been the case that a longer exposure to stress would have allowed

isolation of progressively better adapted strains. Accordingly, 11 of the 19

isolates identified as having improved fructophilicity compared to the parental

strain, have been isolated after 350 generation (longest exposure to selective

pressure in this study).

� Previous studies, although on a much larger scale (Helling et al., 1987; Kao

and Sherlock, 2008), showed that improvement is not a linear trend, but that

adaptation of a population to selective pressure follows oscillatory patterns.

The decline in the number of isolates with improved fermentative ability

compared to the parental around 250-300 generation (Figure 4.6) may

represent such an oscillatory effect.

78

CHAPTER 5

PHYSIOLOGICAL AND GENETIC CHARACTERIZATION OF THE IDENTIFIED

ISOLATES

79

Four isolates showing improved fructophilic and overall fermentation performances

were used in further experiments aimed to more fully evaluate the fermentation of

these. Approaches and techniques with increased sensitivity were therefore used and

represented the initial part of the transition between the laboratory scale to more wine-

like environments. In particular, the screening of the isolates to this point had been

performed in CDGJM containing 50 g/l each of glucose and fructose. In grape juice

the sugar concentration is at least double this and so too the concentration of alcohol

produced.

In addition to fermentation performances it was also important to understand if

isolates differed from the parental strain in terms for other oenological traits and

genetic attributes. The concentrations of organic acids, alcohols and hydrogen

sulphide after the complete catabolism of sugars were measured. Molecular typing of

the strains permitted determination of the extent of the genetic relatedness of isolates

to the parental strain.

The last investigations were aimed at evaluating the fermentative performance of the

mutants in grape juices, to understand if the mutation(s) isolated in artificial

(laboratory) conditions conferred a faster rate of fructose fermentation in a more

wine-like environment.

5.1 FERMENTATION PERFORMANCES OF 4 ISOLATES IN A HIGHER SUGAR CONCENTRATION MEDIUM

Isolates 3, 7, 9 and 11 highlighted from the previous screening experiment (Section

4.2) were streaked on YEPD plates (Appendix 1 – Method 2a) from -80°C glycerol

stock, incubated overnight at 30°C and checked for purity. The parental strain was

prepared similarly. A colony representative of each isolated and multiple colonies of

the parental strain were then inoculated into YEPD broth and grown overnight at 30°C

with agitation at 160 rpm on shaker bench. Cells of each culture were inoculated into

50 ml aliquots of CDGJM starter medium (50 g/l glucose, 50 g/l fructose, Appendix 1

– Method 3) to an inoculation density of 2.5 x 106 cells/ml. The starter cultures were

grown as before and triplicates 250 ml aliquots of CDGJM (115 g/l of glucose, 115 g/l

of fructose) inoculated to 5 x 106 cells/ml. Flasks were sealed with lids provided with

80

water filled air locks, placed into an automatic fermenter (Medicel Explorer) with

continuous flushing of the head space with filtered nitrogen (0.45 µm, 5 ml/min) and

kept at 30°C (in a water bath) with constant agitation (magnetic stir bars, 200 rpm).

Three ml samples were collected automatically and held at -5°C until transferred to

-20°C prior to enzymatic determination of glucose and fructose (Boehringer-

Mannheim, 1989) with final assay volumes adjusted to 100 µl for analysis in 96 well

microtiter plates. The frequency of sampling was generally 6-8 hours at the beginning

and final stages of fermentation, but decreased to 12 hours in the middle of the

fermentation.

The strains showed a clear differentiation in the second half of the fermentation

(Figure 5.1). The parental strain completed the fermentation in 153 hours. By

comparison, isolate 9 showed a marked reduction in fermentation duration depleting

all sugars in 117 hours. Isolate 11 showed an intermediate fermentation duration of

135 hours. Isolates 3 and 7 failed to complete the fermentation and became stuck at

about 30 g/l, which incidentally was composed mainly of fructose. The improvement

in the fermentation kinetic of isolates 9 and 11 was reflected in the area under the

fermentation curves (Table 5.1). These isolates showed a reduction in this value of

≈19% and ≈12% respectively for the total area.

By considering the utilization of the individual sugars, it is possible to see that the

improvement in fermentation was mainly due to the improved fructophilicity of the

two isolates. The increased ability to consume fructose compared to glucose was

confirmed by the calculation of the ratio between the area under the curve of glucose

and fructose, which increased from 0.538 of the parental strain to 0.588 and 0.566 for

isolates 9 and 11 respectively.

As stated, isolates 3 and 7 failed to complete fermentation, leaving mainly residual

fructose. Interestingly the utilisation of glucose was essentially the same as seen for

the parental strain and isolates 9 and 11. The basis of the selective loss of fructose

utilising ability is unknown, but may be linked to the higher ethanol yield of these

high sugar fermentations (230 g/l total) compared to the adaptive experiment (ethanol

concentration ≈ 0 g/l). The experimental conditions of the continuous culture have

lead to an increase in fructophilicity, but they may have contributed to a decrease in

the ethanol tolerance in certain isolates.

81

PAR 9 11

Area TOT % 100 81.37 87.99Area GLU % 100 86.40 92.31Area FRU % 100 79.09 87.77

Ratio Areas (GLU/FRU) 0.538 0.588 0.566

50 100 150

100

200

A

Parent37911

Time elapsed (h)

Tota

l sug

ar c

once

ntra

tion

- GLU

+ F

RU

(g/l)

50 100 150

50

100

B

Time elapsed (h)

Glu

cose

con

cent

ratio

n (g

/l)

Figure 5.1 – Curves of depletion of total sugar (A), glucose (B) and fructose (C) are shown for four isolates and the parental strain. Results are the average of triplicate 250 ml CDGJM fermentations, containing 230 g/l of total sugar (standard deviations are shown as error bars).

Table 5.1 – Percentage of the area under the fermentation curves for isolates 9 and 11 compared to the parental strain for total sugar or glucose and fructose individually. The fructophilicity of the strains is expressed as a ratio between area under the fermentation curves of glucose and fructose. Isolates 3 and 7 were not considered further as they were unable to complete fermentation.

50 100 150

50

100

C

Time elapsed (h)

Fruc

tose

con

cent

ratio

n (g

/l)

82

The increased fermentative performance, mainly attributable to the increased

fructophilicity of mutants 9 and 11 compared to the parental strain, represents the first

confirmation of the ability of adaptive evolution to improve fructose utilization in

wine yeast. Further characterization in more wine-like conditions was required, as was

a comparison of the performance of these isolates with commercial strains. The latter

comparison would allow the magnitude of the improvement of the isolates to be put in

a broader oenological context.

5.2 EVALUATION OF THE FERMENTATION KINETICS OF ISOLATES 9 AND 11, THE PARENTAL STRAIN AND TWO COMMERCIALLY AVAILABLE STRAINS.

EC1118 and Fermichamp, two commercially available strains, already used in this

study (Chapter 2), were included as a reference for the performance of isolates 9 and

11. EC1118 is one of the most widely used strains in winemaking and wine research.

Fermichamp has been described as a strain showing increased fructophilicity, due to a

mutation in the hexose transported encoded by HXT3 (Guillaume et al., 2007). The

preparation and execution of the experiment was the same as that described in Section

5.1. The only difference was the replacement of mutants 3 and 7 with the two

reference strains, EC1118 and Fermichamp.

EC1118 and Fermichamp performed similarly to the parental strain, but they did not

quite complete fermentation in the time frame of the experiment, leaving

approximately 10 g/l of sugar (mainly fructose) in the medium (Figure 5.2). By

comparison, isolate 9 and 11 performed strongly and completed the fermentation in

160 and 168 hours, respectively, while the parental strain required 186 hours. Superior

performance of the parental strain compared to EC1118 and Fermichamp was

attributable to more rapid utilization of glucose. Isolates 9 and 11 exhibit more rapid

utilisation of both glucose and fructose compared to the other strains. Unlike the

previous experiment were improvement of these isolates was largely specific to

fructose utilization, in this case a similar proportional improvement was seen in both

sugars. As results there was no change in the fructophilicity of the isolates as

demonstrated by the fact that the index between the areas under the fermentation

83

PAR 9 11 EC1118 Fermichamp

Area TOT % 100 84.84 80.49 112.82 108.50Area GLU % 100 84.13 78.52 122.77 125.59Area FRU % 100 85.24 81.58 107.34 99.12

Ratio Areas (GLU/FRU) 0.549 0.541 0.528 0.627 0.695

50 100 1500

50

100

Time elapsed (h)

Fruc

tose

con

cent

ratio

n (g

/l)

C

50 100 1500

100

200

Parent911EC1118Fermichamp

A

Time elapsed (h)

Tota

l sug

ar c

once

ntra

tion

- GLU

+ F

RU

(g/l)

50 100 1500

50

100

B

Time elapsed (h)

Glu

cose

con

cent

ratio

n (g

/l)

Figure 5.2 – Curves of depletion of total sugar (A), glucose (B) and fructose (C) are shown for isolates 9 and 11, the parental strain and references EC1118 and Fermichamp. Results are the average of triplicate 250 ml CDGJM fermentations, containing 230 g/l of total sugar (standard deviations are shown as error bars).

Table 5.2 – Percentage of the area under the fermentation curves for isolates 9 and 11 compared to the parental strain, EC1118 and Fermichamp for total sugar or glucose and fructose individually. The fructophilicity of the strains is expressed as a ratio between the areas under the fermentation curves for glucose and fructose.

84

curves of glucose and fructose was unchanged compared to parent (Table 5.2).

Despite this the superior fermentation capability of isolates 9 and 11 warranted further

characterization of their physiological and genetic properties.

5.3 METABOLITE PRODUCTION DURING FERMENTATION

The concentration of the organoleptically relevant organic acids (citric, tartaric, malic,

succinic, lactic, acetic) and alcohols (ethanol and glycerol) was measured by HPLC

(Walker et al., 2003) for isolates 9 and 11 and the parental strain in samples from the

fermentations described in Section 5.2. Frozen ferment samples were thawed and

centrifuged for 5 minutes at 20,800 rcf. Samples were diluted 1:10 and resolved on

either a Benson Carbohydrate SS H+ column (300 mm × 7.8 mm – Alltech, Deerfield,

IL, USA) or an Aminex HPX-87H column (300 mm × 7.8 mm – Bio-Rad, Hercules,

CA, USA). Elution was performed at 60°C with 2.5 mM H2SO4 at a flow rate of 0.5

ml/min. Detection was achieved using a RID-10A refractive index detector

(Shimadzu, Kyoto, Japan). Quantification was achieved by comparison with prepared

standards in CDGJM using Delta integration software (Deltaware, Charlottetown,

Canada – Figure 5.3).

Isolates 9 and 11 showed no large variation in the amount of metabolites produced

during fermentation. The only significative differences observed were the lower

concentration of citric acid and the higher production of acetic acid for the parental

strain compared to the isolates. The amounts of acetic acid produced were of the order

of the legal limit for this acid in commercial wines (Hornsey, 2007) and while

somewhat high, were typical for fermentations of CDGJM (e.g. McBryde et al.,

2006). Determination of the sensory significance of these differences in acetic acids

yield was held over for fermentation of genuine grape juice.

Hydrogen sulphide production of the isolates 9 and 11 compared to the parental strain,

EC1118 and Fermichamp was determined by a different set of fermentations. H2S

production was quantified using lead acetate detector tubes (Katagawa Precision Gas

Detector Tubes, model 120SB-Hydrogen Sulphide, range 0.75-300 ppm, Kawasaki-

City, Japan) as described elsewhere (Ugliano and Henschke, 2010). Starter culture

were prepared as detailed in Section 5.1 and fermentations were conducted in

triplicate in 90 ml of CDGJM (115 g/l glucose, 115 g/l fructose – Appendix 1,

85

Citric A

cid

Tartari

c Acid

Malic A

cid

Succin

ic Acid

Lactic

Acid

Acetic

Acid

Glycero

l

Ethano

l0

2

4

6

8110

120

130

140

Parental911

Metabolites

Con

cent

ratio

n (g

/l)

Figure 5.3 – Metabolite concentration in end of fermentation (< 2.5 g/l sugar) samples for isolates 9 and 11 and the parental strain. The values are the average of single determinations from triplicate 250 ml CDGJM fermentations, containing 230 g/l of total sugar (standard deviations indicated as error bars).

86

Method 3) in 250 ml conical flasks with shaking (160 rpm). Fermentations were

complete (total sugar < 2.5 g/l) after 282 hours at 30°C. The quantity of H2S produced

for each flask (read on the graduate scale of the tubes) was converted into micrograms

per litre using a previously determined calibration curve (data not shown).

Although some strain differences were evident, the quantities of H2S produced during

these fermentations (Figure 5.4) were lower than those found in another study and

deemed sensorially perceptible. Ugliano and co-workers (2009) reported values

between 9 and several hundred micrograms per litre of grape must. The higher values

were mainly associated with fermentations deficient in assimilable nitrogen. In the

present study the CDGJM was prepared with a high nitrogen content (600 µg/litre as

mixed amino acids), thereby likely explaining the limited production of H2S.

5.4 DNA FINGERPRINTING OF EVOLVED STRAINS Any change in the physiological characteristics of a strain, most likely reflects

changes at the DNA level (Giudici et al., 2005). It would ultimately be desirable to

determine the genome sequence of the isolates and the parental strain in order to

define the precise nature of any mutation(s) in the former. However, before

committing to such an undertaking, the use of less detailed genetic comparative

techniques was though appropriate to confirm the close relatedness of isolates and

parental strain and confirm that the isolates were not in fact contaminants. The DNA

fingerprinting method of Ness and co-worker (1993) was chosen since this is a

commonly used protocol for industrial strains of S. cerevisiae.

In brief, the method involves amplification of the repeated elements that flank the Ty1

retrotransposon using the primers listed in Appendix 1 (Method 6). Genomic DNA

was purified from duplicate clonal cell cultures (Appendix 1 – Method 5; Adams et

al., 1997) and its integrity checked by resolution on 2% agarose gel in 1 x TAE buffer

(Tris-acetate-EDTA) and electrophoresis. The Polymerase Chain Reaction (PCR) for

the Ty1 amplification is described in Appendix 1 (Method 6).

87

Figure 5.4 – Quantity of H2S produced per litre of fermentation by mutants 9 and 11, the parental strain and references EC1118 and Fermichamp. Values are the mean of triplicate 90 ml fermentations in CDGJM with 230 g/l of total sugar (standard deviations are indicated as error bars).

0

5

10

15

20

25

30

35

40

Pare

ntal 9 11

EC 1

118

Ferm

icha

mp

H2S

(µg/

l of f

erm

enta

tion)

Strains

88

A distinct banding pattern was observed for each strain set (Figure 5.5). In particular

the genotypic profile of the parent strain (lanes 2 and 3) was discernibly different

from EC1118 (lanes 8 and 9) and Fermichamp (lanes 10 and 11). Importantly, the

genotypic profile of the 2 clonal cultures of isolates 9 (lanes 4 and 5) and 11 (lanes 6

and 7) was identical to the parent, but different from EC1118 and Fermichamp.

These results suggest that it is unlikely that isolates 9 and 11 are contaminants and

instead appear to be mutants derived from the parental strain.

5.5 PHENOTYPE STABILITY In order to observe the long term stability of the phenotype, mutants were propagated

under low stress condition for over 60 generations. This number was considered

sufficient considering that 50 generations was the threshold interval at which it was

possible to observe occurrence of new mutants (Figure 4.5 and 4.6). In this way any

mutation that was not fixed in the population would be lost and the efficient

fermentation phenotype lost with it. Similarly such a passaging exercise would

eliminate the possibility that the observed phenotype was mainly due to adaptation or

pre-conditioning (transient expression of e.g. stress response genes).

Mutants (isolates) 9 and 11 were streaked onto YEPD agar (Appendix 1 – Method 2a)

from a -80°C glycerol stock and grown overnight at 30°C. A single colony from each

plate was transferred into 10 ml of liquid YEPD in a 50 ml Falcon tube and grown

overnight at 30°C. Ten µl of each of these cultures were transferred into 10 ml of

fresh YEPD and growth repeated. This operation was repeated 5 more times. It was

estimated that around 10 generations would elapse for each passage. An aliquot of the

resulting populations was streaked onto YEPD plates and grown as before. Similarly

fresh YEPD plates of the parental strain, a reference strain (Fermichamp) and mutants

9 and 11 prior to passaging (9 original and 11 original), were prepared. Sixteen

colonies from each passaged population as well as a colony from the parental strain,

the reference strain and mutants 9 and 11 (original populations) were transferred into

1 ml of YEPD and grown overnight at 30°C. From this point the experiment was the

same as that described in Section 4.1, utilising duplicate sacrificial plates (200 µl per

well). See Figure 5.6 for the plate layout.

89

Figure 5.5 – Genomic fingerprinting obtained with the Polymerase Chain Reaction (PCR) targeting the Ty1 elements in the genome of the parental strain (lanes 2, 3), mutants 9 (lanes 4, 5) and 11 (lanes 6, 7) and references strains EC1118 (lanes 8, 9) and Fermichamp (lanes 10, 11). Lanes 1, 13 include molecular weight markers (100 base pair ladder) and a PCR product without DNA (blank) is shown in lane 12.

Figure 5.6 – Layout used for the evaluation of sixteen isolates from the passaging of mutants 9 or 11 as well as the mixed population of the passaged cultures (MIX), the parental strain (PAR), a culture of the appropriate original (pre-passaging) mutants 9 and 11 (9 origin or 11 origin) and a reference strain (Fermichamp). Wells in column 11 of the plate were filled with uninoculated media as a sterility check. The wells in column 12 were used during subsequent quantification of sugar concentration.

90

There were no differences between fermentation kinetics of the passaged mutants (16

isolates) 9 and 11 compared to the corresponding mixed population and mutants pre-

passaging (Figures 5.7 panels A to F). Moreover all of the mutants showed faster

fermentation compared to the parent and Fermichamp. In particular and as has been

shown previously for these experimental conditions (Section 4.1), the improvements

in fermentation kinetics were largely attributable to the improvement in fructose

utilization. These data indicate that the changes that have occurred in the mutants and

conferred superior fermentation properties are stable to passaging under low stress

conditions for at least 60 generations.

5.6 GRAPE JUICE FERMENTATIONS The conditions used in this study to isolate improved mutants are ultimately artificial

and do not completely match a grape juice matrix. In addition the continuous

fermentation strategy used does not capture the dynamic compositional phases of

fermentation. In order to begin to evaluate these strains in more industrially relevant

conditions, fermentations were conducted in grape juice and must on a 250 ml and a

20 kg scale, respectively.

Evaluation of fermentation attributes in white grape juice

Fermentations were performed in sterile (0.2 µm) juice obtained from 2007

Chardonnay grapes grown in the Eden Valley (South Australia). Basic compositional

features are shown in Table 5.3. The preparation and execution of the experiment was

largely as described in Section 5.1. Key differences were the incubation temperature

(20°C) and the fact that the starter culture was prepared in white grape juice diluted

1:1 with water. The fermentation performance of mutants 9 and 11 was compared

against the parental strain and reference strains EC1118 and Fermichamp. The

experiment was repeated with similar results, thus only one set of result is shown.

Fermentations of the five strains fell into three different profiles (Figure 5.8).

Fermichamp required 144 hours to deplete all of the sugar from the juice, EC1118 and

mutants 9 and 11 required 235 hours, while the parental strain had not finished

fermentation even after 259 hours (9 g/l residual fructose). Rates of glucose

91

PARAMETER VALUE Glucose + fructose 213.5 g/l Malic acid 2.92 g/l Sulphur dioxide (total) 21 mg/l Sulphur dioxide (free) <5 mg/l pH 3.34 Titratable acid (pH 7.0) 5.4 g/l Titratable acid (pH 8.2) 5.7 g/l Yeast assimilable nitrogen 381 mg/l Ammonia 109 mg/l Alpha amino nitrogen 291 mg/l

20 40 600

50

100

9 or 11 OriginalFermichampMix populationParentalIsolates

A

Time elapsed (h)

Suga

r con

cent

ratio

n (g

luco

se +

fruc

tose

) [g/

l]

20 40 600

20

40

60

B

Time elapsed (h)

Suga

r con

cent

ratio

n (g

luco

se) [

g/l]

20 40 600

20

40

60

Time elapsed (h)

Suga

r con

cent

ratio

n (fr

ucto

se) [

g/l]

C

20 40 600

50

100

D

Time elapsed (h)

Suga

r con

cent

ratio

n (g

luco

se +

fruc

tose

) [g/

l]

20 40 600

20

40

60

E

Time elapsed (h)

Suga

r con

cent

ratio

n (g

luco

se) [

g/l]

20 40 600

20

40

60

F

Time elapsed (h)

Suga

r con

cent

ratio

n (fr

ucto

se) [

g/l]

Figure 5.7 – Curves of depletion of total sugar (A, D), glucose (B, E) and fructose (C, F) are shown for 16 isolates from passaged cultures of mutants 9 (A, B and C) and 11 (D, E and F) along with thecorresponding mixed and pre-passaged (original) population of each and the parental strain. Fermichamp was included as reference. Results are the average of four replicates performed in 200 µl in microtiter plates (standard deviations are shown as error bars).

Table 5.3 – Key compositional features of 2007 Eden Valley Chardonnay juice used to evaluate mutant fermentation performance.

92

consumption were similar across all strains, whereas much of the observed

differentiation was attributable to differences in fructose fermentation rates. As result,

the ratio of the area under the glucose and fructose fermentation curves was highest

for Fermichamp (0.75) followed by EC1118 (0.63) and mutants 9 and 11 (0.56 and

0.53; Table 5.4). Once again this experiment demonstrated the extraordinary ability of

Fermichamp to consume fructose (Guillaume et al., 2007).

Evaluation of fermentation attributes in red grape must

As for the previous experiment, the complete set of fermentations was repeated,

however in this case using two different batches of Merlot grapes (Table 5.5). Given

the similarity of the results across the two must, only data from the Lobethal Merlot

fermentations are reported here.

The grapes were handpicked into 20 kg crates and kept at 4°C until crushed. For each

strain, three replicate fermentations of 20 kg of grape were prepared using fruit

randomly taken from different crates.

Fermentations were conducted in 30 litre plastic containers, provided with a 7 mm

hole for release of CO2. The preparation of the inocula followed the same procedure

as described for the Chardonnay fermentations. In this experiment, due to the

presence of solids, the must was not sterilised. Fermentations were conducted at room

temperature (≈ 25°C) and plunged and sampled at 12 hours intervals. Samples of 1 ml

were stored at -20°C for future enzymatic determination of glucose and fructose

concentration (Boehringer-Mannheim, 1989) with final volumes adjusted to 100 µl

for analysis in 96 well microtiter plates. Other parameters measured at the end of the

fermentations were relevant organic acids (malic, succinic, lactic, acetic) and alcohols

(ethanol and glycerol; HPLC method adopted is described in section 5.3). Colour

measurement (CIELab) of cell free supernatants (20,800 rcf, 1 min) was determined at

the end of fermentation. Calculation of CIELab was obtained measuring the visible

transmission spectrum from 380 nm to 780 nm in 5 nm increments (75 �L in a 96

well microtiter plate, µQuant Microplate Spectrophotometer, Bio-Tek Instruments).

The spectral data was then converted to L* (lightness, 0 – black to 100 – white), a* (-

265, green to 265, magenta) and b* (-265, blue to 265, yellow) using the conversion

equations and tables developed by the Commission Internationale d’Eclairage.

93

100 2000

50

100

150

C

Time elapsed (h)

Fruc

tose

con

cent

ratio

n (g

/l)

VALUES Merlot "Waite"

Merlot "Lobethal" PARAMETER

Glucose + fructose (g/l) 250 265 pH 4.02 3.87

Total acidity (g/l) 3.3 5.0 Alpha Amino Nitrogen (mg/l) 96 93

Ammonia (mg/l) 34 22 Yeast Assimilable Nitrogen (mg/l) 124 111

ADDITIONS SO2 (mg/l) 25 25

Tartaric acid (g/l) 2.3 1 Diammonium phosphate (mg/l) 200 40

PAR 9 11 EC1118 Fermichamp

Area TOT % 100.00 93.06 89.03 98.81 71.43Area GLU % 100.00 100.93 93.31 114.91 92.19Area FRU % 100.00 89.37 87.14 90.90 61.12

Ratio Areas (GLU/FRU) 0.49 0.56 0.53 0.63 0.75

100 2000

50

100

150

B

Time elapsed (h)

Glu

cose

con

cent

ratio

n (g

/l)

100 2000

100

200

300

Parental911EC1118Fermichamp

Time elapsed (h)

Tota

l sug

ar c

once

ntra

tion

- Glu

cose

+ fr

ucto

se (g

/l)

A

Figure 5.8 – Curves of depletion of total sugar (A), glucose (B) and fructose (C) are shown for mutants 9 and 11 compared with the parental strain and commercial strains EC1118 and Fermichamp. Results are the average of triplicate 250 ml Chardonnay juice fermentations (standard deviations are shown as error bars).

Table 5.4 – Percentage of the area under the fermentation curves for isolates 9 and 11 compared to the parental strain, EC1118 and Fermichamp for total sugar or glucose and fructose individually. The fructophilicity of the strains is expressed as a ratio between the areas under the fermentation curves for glucose and fructose. Results are derived from triplicate 250 ml Chardonnay juice fermentations.

Table 5.5 – Key compositional features of 2010 Merlot grapes used to evaluate mutant fermentation performance. Grapes were harvested from the Waite Institute experimental vineyards (Adelaide, South Australia) or at Lobethal (South Australia).

94

Results are expressed using a D65 illuminant at 10° standard observer. Hue (H, 0° -

360°) and chroma (C, 0 - 100) values were calculated using the following equations:

°

As for the Chardonnay juice, fermentation in red must the five strains fell into three

profiles (Figure 5.9). EC1118 and Fermichamp depleted the sugars in 240 hours,

mutants 9 and 11 required 287 hours, while the parent had not finished fermentation

even after 293 hours (≈ 10 g/l residual fructose). Fructose fermentation rate mainly

contributed to differences observed in the overall fermentation performance.

However, in this case, the ratio of the area under the glucose and fructose

fermentation curves did not changed for mutants 9 and 11 compared to the parental

strain (≈ 0.65), while EC1118 and Fermichamp showed higher values (0.77 and 0.84

respectively; Table 5.6). Once again, these references strains showed higher

fructophilicity.

From the analysis of the principal metabolites (Figure 5.10), Fermichamp produced a

larger amount of succinic acid (≈ 5 g/l vs ≈ 3 g/l). Mutants 9 and 11 and Fermichamp

showed a greater concentration of glycerol (≈ 13 g/l) than EC1118 and the parental

strain (≈ 11 g/l).

When expressing red wine colour as a CIELab measurement, differences of 3.0 or

more of the calculated DeltaE (ΔE = (L1 – L2)2 + (a1 – a2)2 + (b1 – b2)2), are considered

to exceed the threshold for detectability by the human eye (Martinez et al., 2001). The

values of ΔE calculated in this experiment, were generally higher than 3.0 CIELab

units (Figure 5.11) suggesting that the differences produced by strains in colour would

be detectable. Unfortunately no sensory panel was available to confirm this.

In conclusions, these last experiments confirmed the improved fitness of two mutants

isolated using an adaptive evolution strategy. Thus mutants 9 and 11 showed faster

rate of utilization of both glucose and fructose compared to the parental strain (AWRI

796). These was observed in CDGJM (250 ml, 230 g/l of sugars), Chardonnay juice

(250 ml, 213 g/l of sugar) and Merlot must (20 kg, 265 g/l of sugar). In most cases

fructose was utilised at faster rate than the parental strain, thereby confirming the

effectiveness of the targeted adaptive evolution strategy adopted in this study.

95

100 200 3000

100

200

300

Parental911EC1118Fermichamp

Time elapsed (h)

Tota

l sug

ar c

once

ntra

tion

- GLU

+ F

RU

(g/l)

100 200 3000

50

100

150

Time elapsed (h)

Glu

cose

con

cent

ratio

n (g

/l)

100 200 3000

50

100

150

Time elapsed (h)

Fruc

tose

con

cent

ratio

n (g

/l)

PAR 9 11 EC 1118 FERM

Area TOT % 100.00 88.01 87.38 86.93 87.93Area GLU % 100.00 88.17 86.27 95.59 101.50Area FRU % 100.00 87.90 88.11 81.26 79.06

Ratio Areas (GLU/FRU) 0.65 0.66 0.64 0.77 0.84

Figure 5.9 – Curves of depletion of total sugar (A), glucose (B) and fructose (C) are shown for mutants 9 and 11 compared with the parental strain and commercial strains EC1118 and Fermichamp. Results are the average of triplicate 20 kg Merlot grape fermentations (standard deviations are shown as error bars).

Table 5.6 – Percentage of the area under the fermentation curves for isolates 9 and 11 compared to the parental strain, EC1118 and Fermichamp for total sugar or glucose and fructose individually. The fructophilicity of the strains is expressed as a ratio between the areas under the fermentation curves for glucose and fructose. Results are derived from triplicate 20 kg Merlot grape fermentations.

96

Malic A

cid

Succin

ic Acid

Lactic

Acid

Acetic

Acid

Glycero

l

Ethano

l0

5

10

15140150160170180190

Parental911EC1118Fermichamp

Metabolites

Con

cent

ratio

n (g

/l)

Figure 5.10 – Metabolite concentration in end of fermentation (< 2.5 g/l sugar) samples for isolates 9 and 11, the parental strain and references EC1118 and Fermichamp. The values are the average of single determinations from triplicate 20 kg Merlot grape fermentations (standard deviations indicated as error bars).

L a b

Chroma

Hue0

20

40

60

80Parental911EC1118Fermichamp

CIELab parameters

Val

ues

Figure 5.11 – CIELab parameters measured at end of fermentation (< 2.5 g/l sugar) samples for isolates 9 and 11, the parental strain and references EC1118 and Fermichamp. The values are the average of single determinations from triplicate 20 kg Merlot grape fermentations (standard deviations indicated as error bars).

97

CHAPTER 6

CONCLUSION, DISCUSSION AND FUTURE DIRECTIONS

98

6.1 CONCLUSIONS This project aimed to generate wine yeast strains with improved fructose utilization,

and the strategy employed was adaptive evolution. The objectives of the study were:

1. to test adaptive evolution as a strategy to generate improved phenotypes in

wine yeast for the wine industry;

2. specifically, to improve the fructophilicity of a commercially available wine

yeast strain.

Both of these objectives were achieved; fructophilic variants of AWRI 796 were

generated using adaptive evolution. Fermentation rates of fructose were greater in the

mutants compared to the parental strain and the fructophilic index* for the mutants

also increased when they were grown in CDGJM. Interestingly, however, whilst

overall fermentation rate was also higher for the mutants than its parent in grape juice,

the fructophilic index was unchanged.

6.2 DISCUSSION AND FUTURE DIRECTIONS

Experimental adaptive evolution has been used several times in the past to generate

yeast with improved phenotypes (see for example Brown and Oliver, 1982; Hall,

1992; McBryde et al., 2006). The work described in this thesis built on this past work

and successfully generated a novel yeast strain with improved fructose utilization.

However the results highlighted an important limitation of adaptive evolution

strategies: evolution is not driven solely by the selective pressure, but it is also shaped

by the matrix of the medium. This may be the reason that fructophilic mutants showed

clear increases in fructophilic index for the medium used in the adaptive evolution

experiment (CDGJM), but for grape juice the index was unchanged from the parent.

Clearly, from a wine industry perspective, it will important in future studies to use a

medium that reflects as closely as possible the matrix of the grape juice where the

mutants will be used in industry.

___________________________________________________________________________________*Defined in this study as the ratio between areas under the fermentation curves of glucose and fructose

99

There are numerous ways in which this project could be developed to improve on the

adaptive evolution strategy that was used. For example:

� EMS mutagenesis increases the genetic variation, however the probability of

generating specific advantageous mutations is limited when done on a single

strain. Other techniques for generating genetic variation that could be used

include mass mating of many wine yeast from culture collections. This would

lead to the generation of many novel gene combinations in a range of wine-

relevant genetic backgrounds and should increase the probability of generating

novel genotypes and phenotypes for adaptive evolution to work on.

� The use of other selective pressures to select for fructophilic mutants should

be trialled. For example, 2-deoxy-D-glucose, a glucose homologue, is toxic to

S. cerevisiae and has been used for a study on hexose uptake (Novak et al.,

1990). If used in a medium with fructose as the principal source of sugar it

should select for mutants with superior fructose utilization at the expense of

glucose uptake.

� Other approaches to adaptive evolution have been described and used in the

past. One strategy is BOICS, (Brown and Oliver Interactive Continuous

Selection) derived from the name of the inventors (Brown and Oliver, 1982).

Briefly, it consists of an interactive chemostat selection: the intensity of the

selective pressure can be automatically adjusted via a feedback control circuit.

Thus, by monitoring parameters such as physiological changes in the

bioreactor (e.g. variation in cell density) or fermentation products (e.g. CO2

release), it is possible to increase the selective pressure without the risk of

completely washing out the population. As the authors stated, “the technique

of continuous selection with feedback should be generally applicable to the

isolation of mutants of any microorganism to improved tolerance to any

inhibitory condition of either its physical or chemical environment” (Brown

and Oliver, 1982). Thus the use of an inhibitor such as 2-deoxy-D-glucose

could be used in this type of approach. Moreover, the BOICS strategy could

also be used in a situation where limited fructose concentration determines the

selective pressure. Fructose could be added to the bioreactor, in ever

decreasing concentrations, in the absence of other carbon sources. When the

cell density drops below a certain level, before wash out, the fructose level

100

would be increased sufficiently to rescue survivors, then decreased again. This

could be done over many cycles to select for mutants that have greater

capacity to utilise limiting fructose.

Further characterization of the physiology of the mutants generated in this study will

be essential. It has been shown that they have a faster fermentation rate compared to

the parent, and, from preliminary experiments (metabolite concentration using HCPL,

colour measurements determined with CIELab), there were no large variations to the

final composition of the wine. However, more detailed investigations (larger scale

winery fermentations, wine evaluation through a sensory panel) are required to

confirm the ability of adaptive evolution to generate specifically improved strains

which maintain other oenological traits of parent.

Time did not permit characterization of the mutation(s) that conferred fructophilicity

in the mutants generated from the work described in this thesis. A deeper

investigation of the genome of the mutants and a sub-sequential comparison with the

parental strain would reveal the identity of the ‘fructophilic genes’ in the mutants.

This would facilitate the use of genetic engineering techniques to produce more

efficient fructophilic strains. Genetic manipulations represent an important approach

in the modern generation of mutants for the wine and food industry. Despite the

ethical debates existing at present on the use of genetic modified organisms, it is

likely that in future they will be accepted for the production of food and beverages.

101

APPENDIX 1

102

METHODS Method 1 – Collection of yeast strains from commercial package for long term storage Strains were collected aseptically from active dried commercial preparations, re-

hydrated in sterile water (20 min) and inoculated into YEPD medium (see Method 2a)

in a flask (air/liquid ratio > 66%) before overnight incubation at 28°C with shaking at

180 rpm. Cultures were then streaked onto YEPD agar plates and grown overnight at

28°C to check for purity. Multiple representative colonies were inoculated into 25 ml

of YEPD broth and grown as above. The combination of 1 ml of culture with 0.5 ml

of sterile 80% (v/v) glycerol (final concentration = 15% v/v), enabled long term

storage at -80°C.

Method 2 – a. YEPD medium b. YEPF medium

10 g/l yeast extract 10 g/l yeast extract

20 g/l bacto-peptone 20 g/l bacto-peptone

20 g/l D-glucose 20 g/l D-fructose

For preparation of plates, 20 g/l agar was added. All media were sterilize by autoclaving at 121°C for 20 min.

103

Method 3 – Chemically Defined Grape Juice Media (CDGJM) (Henschke and Jiranek, 1993) CDGJM is a water based medium. All the chemical components listed below were

dissolved in water. and sterilized by filtration (0.2 µm).

SALTS

L-Malic acid 3 g/l Citric acid 0.2 g/l Potassium phosphate dibasic (K2HPO4) 1.14 g/l Magnesium sulphate heptahydrate (MgSO4.7H2O) 1.23 g/l Potassium sodium tartrate tetrahydrate (KNaC4H4O6.4H2O) 3.12 g/l Calcium cloride dihydrate (CaCl2.2H2O) 0.44 g/l

TRACE MINERALS

Manganese chloride tetrahydrate (MnCl2.4H2O) 198.2 μg/l Zinc chloride (ZnCl2) 135.5 μg/l Iron(II) chloride (FeCl2) 32 μg/l Copper(II) chloride (CuCl2) 13.6 μg/l Boric acid (H3BO3) 5.7 μg/l Cobalt(II) nitrate hexaydrate (Co(NO3)2.6H2O) 29.1 μg/l Sodium molybdate dihydrate (Na2MoO4.2H2O) 24.2 μg/l Potassium iodide (KI) 10.8 μg/l

VITAMINS Myo-insitol 100 mg/l Pyridoxide-HCL 2 mg/l Nicotinic acid 2 mg/l Calcium Pantothenate 1 mg/l Thiamine-HCL 0.5 mg/l ρ-Amino benzoic acid 0.2 mg/l Biotin 0.125 mg/l Folic acid 0.2 mg/l Riboflavin 0.2 mg/l

SUGARS Glucose *Variable Frutose *Variable

#NITROGEN amino acids ^Variable

*Glucose and fructose concentration can vary.

^Nitrogen concentration can vary; in all of the experiments presented in this thesis the

nitrogen concentration was 600 mg/l, added as a 25x amino acid solution.

104

#25x Amino acid solution Alanine 100 mg/l Arginine 750 mg/l Asparagine 150 mg/l Aspartic acid 350 mg/l Glutamic acid 500 mg/l Glutamine 200 mg/l Glycine 50 mg/l Histidine 150 mg/l Isoleucine 200 mg/l Leucine 300 mg/l Lysine 250 mg/l Methionine 150 mg/l Phenylalanine 150 mg/l Proline 500 mg/l Serine 400 mg/l Threonine 350 mg/l Tryptophan 100 mg/l Tyrosine 20 mg/l Valine 200 mg/l Ammonium chloride 100 mg/l

1 ml of 25x amino acid solution to supply 20.27 mg of N For CDGJM starter, was used half of the total sugar concentration plus Tween 80® and ergosterol.

STARTER Tween 80® 0.5 ml/l Erosterol 10 mg/l

105

Method 4 – Ethyl methanesulfonate (EMS) mutagenesis (Fink, 1970)

Material Yeast culture – YEPD (2 x 108 Cells/ml) 15 ml Sodium phosphate 0.1 M buffer pH 7

(2.16 g NaH2PO4 – dibasic 6.54 g NaH2PO4 – monobasic Make up to 200 ml with H2O)

Ethyl methanesulfonate (EMS) 45 µl/ml of YEPD Na thiosulphate 5% sterile water solution YEPD agar plates YEPD Glycerol 80% sterile water solution

The culture was grown over night in 25ml of YEPD at 30°C. Cells were collected

(5000 rpm for 5 min) in a 50 ml sterile plastic tube, washed twice in 15 ml of 0.1 M

sodium phosphate buffer pH 7 and resuspended in 2.5 ml of the same buffer. Cells

were counted and 3 x 109 cells were transferred to a 50 ml sterile plastic tube, brought

to a volume of 15 ml with 0.1 M sodium phosphate buffer pH 7 (final cell

concentration = 2 x 108 cells/ml). One ml of suspension was removed and kept (for

plate inoculation and freeze storage). 630 µl of EMS were added and the culture

incubated at 30°C. 1 ml samples were removed every 10 minutes and wash twice with

one volume of 5% of sodium thiosulphate. Cells were resuspended in 1 ml of YEPD.

For each time point, the appropriate amount of cells was transferred in duplicate onto

agar plates for colony counts. The remaining cells were mixed with 0.5 ml of 80%

(v/v) glycerol for storage at -80°C. The average viable cell count was used to

calculate the relative survival curve.

106

Method 5 – Isolation of genomic DNA (Adams et al., 1997)

Materials:

- YEPD (see Method 2a) - MQ (High purity water – DNAse free) - Phenol : chloroform 5:1 with aqueous phase on top – OR – Phenol:

chloroform: isoamyl alcohol (25:24:1) - Acid washed glass beads - TE: 10 mM Tris-Cl pH 8.0, 1 mM EDTA (in 100 mL 0.12 g tris, 0.037 g

EDTA) - RNAse cocktail - 4 M ammonium acetate - Lysis buffer: 2% Triton X-100, 1% SDS, 100 mM NaCl, 10 mM Tris-Cl pH

8.0, 1 mM Na2EDTA. Autoclaved or filter sterilized.

Method:

1 Grow 10 mL yeast cultures to saturation in YEPD at 30°C overnight from

a streak plate.

2 Collect the cells by centrifugation for 5 minutes. Remove supernatant and

re-suspend in 0.5 ml of water. Transfer to a 1.5 ml microfuge tube. Collect

by centrifugation for 5 seconds.

3 Decant the supernatant

4 Add 0.2 ml lysis buffer and re-suspend. Add 0.2 ml phenol : chloroform

5:1, leaving top aqueous layer. Add 0.3 g acid washed glass beads.

5 Vortex for 3-4 minutes, placing on ice every 30 seconds. Add 0.2 ml TE

6 Centrifuge for 5 minutes. Transfer the aqueous layer to a fresh tube. Add

1.0 ml 100% ethanol. Mix by inversion

7 Centrifuge for 2 minutes. Discard supernatant – pipette. Re-suspend in 0.4

ml TE. Add 1 �l RNAse cocktail. Incubate at 37° for 15 minutes. Add 10

�l 4 M ammonium acetate, plus 1 ml 100% ethanol. Mix by inversion

(denature proteins). Keep in cold room for 15 minutes. Centrifuge for 5

minutes.

8 Discard supernatant. Air-dry pellet. Re-suspend in 50 �l MQ, leaving for

15 minutes, while flicking occasionally.

107

Method 6 – Ty1 transposon (Ness et al., 1993)

Ty1 transposon primers: MLD1 (forward primer) 5’-CAAAATTCACCTATA/TTCTCA-3’ MLD2 (reverse primer) 5’-GTGGATTTTTATTCCAACA-3’ Concentration of the components used for Ty1 transposon PCR reaction

Master MIX Components

Amount used for single PCR

reaction

10x reaction buffer 2.5 µl

MgCl2 (25 mM) 2.5 µl

dNTP’s (10 mM) 1.0 µl

Forward primer (10 µM) 1.0 µl

Reverse primer (10 µM) 1.0 µl

H2O 15.8 µl

Taq polymerase (5 U/µl) 0.2 µl

DNA 1.0 µl

Total volume 25 µl

PCR program 5.0 min 95ºC 0.5 min 95ºC 3 X 0.5 min 42ºC 2.0 min 72ºC 0.5 min 95ºC 29 X 0.5 min 45ºC 2.0 min 72ºC 10 min 72ºC hold at 20ºC

108

APPENDIX 2

109

STATEMENT OF AUTHORSHIP

NOTE: Statements of authorship appear in the print copy of the thesis held in the University of Adelaide Library.

A

Microvinification – How small can we go? 1

2

Tommaso Liccioli1, Tina M.T. Tran2,3, Daniel Cozzolino2, Vladimir Jiranek1, 3

Paul J. Chambers2 and Simon A. Schmidt2* 4

1School of Agriculture, Food and Wine, The University of Adelaide, Glen Osmond, 5

SA 5064, Australia. 6

2The Australian Wine Research Institute, PO Box 197, Glen Osmond, SA 5064, 7

Australia 8

3School of Molecular Sciences, Victoria University of Technology, Werribee Campus 9

(W008), P.O. Box 14428, Melbourne City, MC, Victoria, Australia, 8001 10

* Corresponding author. Tel.: +61 08 830366600; fax.: +61 08 83036601; e-mail: 11

simon.schmidt@awri.com.au 12

13

14

B

Abstract 15

Background 16

High throughput methodologies to screen large numbers of micro-organisms 17

necessitate the use of small-scale culture vessels. In this context, an increasing 18

number of researchers are turning to microtiter plate (MTP) formats to conduct 19

experiments. MTPs are now widely used as a culturing vessel for phenotypic 20

screening of aerobic laboratory cultures and their suitability has been assessed for a 21

range of applications. The work presented here extends these previous studies by 22

assessing the metabolic foot-print of MTP fermentation. 23

Results 24

A comparison of Chardonnay grape-juice fermentation in MTPs with fermentations 25

performed in air-locked (self-induced anaerobic) and cotton-plugged (aerobic) flasks 26

was made. Maximum growth rates and biomass accumulation of yeast cultures grown 27

in MTPs were indistinguishable from self-induced anaerobic flask cultures. 28

Metabolic profiles measured using both targeted and non-targeted methods differed 29

depending on the metabolite. While glycerol and acetate accumulation mirrored that 30

of self-induced anaerobic cultures, ethanol accumulation in MTP ferments was limited 31

by the increased propensity of this volatile metabolite for evaporation in microlitre-32

scale culture format. 33

C

Conclusions 34

The data illustrates that microplate cultures can be used as a replacement for self-35

induced anaerobic flasks in some instances and provide a useful and economical 36

platform for the screening of industrial strains and culture media. 37

Keywords 38

Yeast; microplates; grape-juice; fermentation; high-throughput 39

40

D

Background 41

The use of sub-millilitre culture volumes in microtiter plate (MTP) format for the 42

characterization of microbial cultures has become routine for many applications. 43

Extensive evaluation of MTP configuration (2000; Duetz and Witholt 2004; 44

Warringer and Blomberg 2003), sealing membrane performance (Zimmermann et al. 45

2003) and the impact of fluid mixing regimes on oxygen transfer rates (Duetz et al. 46

2004; Hermann et al. 2003; Kensy et al. 2005) has resulted in the use of MTP format 47

in varied applications including phenotypic profiling (Maresova and Sychrova 2007; 48

Stitt et al. 2002; Warringer et al. 2003), optimization of product yield from E. coli and 49

hybridoma cell lines (Ferreira-Torres et al. 2005; Micheletti et al. 2006a), high-50

throughput metabolite (Börner et al. 2007) or metabolic flux analysis (Blank et al. 51

2005a; Blank et al. 2005b), culminating in full automation of MTP fermentation 52

(Zimmermann and Rieth 2007). The breadth and depth of development of MTP-53

based methods was recently reviewed (Duetz 2007; Micheletti and Lye 2006b) with 54

Duetz declaring that MTPs can now be considered a mature alternative to Erlenmeyer 55

shake flasks. 56

57

One area of investigation lacking in this otherwise comprehensively researched field 58

is self-induced anaerobic fermentation. Self-induced anaerobic refers to 59

fermentations in which oxygen is initially present but which quickly tend toward 60

anaerobisis. Self-induced anaerobic fermentation is fundamental to much of industrial 61

microbiology involving, for example, wine, beer and biofuel production. In these 62

environments oxygen is not only limiting but often undesirable. The question 63

associated with culture miniaturization is, therefore, not how to maximize oxygen 64

E

transfer rates, as is the case for aerobic culture, but how to minimize them. As 65

previously noted by Duetz (2007), one of the intrinsic properties associated with 66

culture miniaturization is the high ratio of gas-liquid exchange area relative to the 67

bulk liquid compared with shaken flask cultures. Differences in this ratio inevitably 68

lead to altered oxygen transfer rates (OTRs) and liquid loss due to evaporation. 69

Recent literature addresses the problem of differential OTRs by optimizing fluid 70

mixing regimes (Duetz et al. 2004) however, in the case of self-induced anaerobic 71

Erlenmeyer fermentations, agitation is used to maintain culture homogeneity rather 72

than enhance OTRs. In self-induced anaerobic cultures agitation rates are typically 73

minimized to a level sufficient for the maintenance of biomass suspension, and no 74

more. The purpose of maintaining a suspension is two fold: to enable representative 75

sampling for the estimation of cell biomass and, to ensure that culture reproducibility 76

is not diminished by stochastic effects resulting from the lack of mixing. 77

78

Running self-induced anaerobic cultures in MTP format releases us from the first of 79

these two requirements since culture density is determined non-destructively by 80

measuring optical density in situ, vertically across the whole well; whether or not the 81

cells are in suspension is of little consequence. This leaves us only with the second 82

requirement, culture reproducibility and this raises several questions. Is culture 83

homogeneity a prerequisite for reproducible fermentations when working in small 84

scale culture vessels? Will, for example, settling of biomass to the bottom of a MTP 85

well, limit its access to nutrients and thus diminish or alter the result? What 86

constitutes a set of conditions whereby fermentations conducted in MTP format yield 87

results indistinguishable from self-induced anaerobic Erlenmeyer flasks? We 88

approached these questions by comparing the growth profile and metabolic footprint 89

F

of an industrial yeast during fermentation of grape juice in MTPs, aerobic flasks 90

(AeFs) and self-induced anaerobic flasks (SAFs). In addition, the feasibility of MTP 91

format use in a typical industrial strain development workflow was explored by 92

screening 15 wine yeast strains using fermentation performance as a criterion for 93

selection. The performance profiles of four strains, exhibiting performance diversity 94

in MTP fermentations, were validated using larger scale self-anaerobic fermentation. 95

Using this approach the utility of MTP format for screening of industrial yeast strain 96

properties other than growth and biomass formation is demonstrated. 97

98

Materials and Methods 99

Strain and culture conditions 100

All yeast strains were maintained on 1% w/v yeast extract, 2% w/v bacto peptone and 101

2% w/v glycerol (YPG) agar plates. Overnight cultures were grown in 1% w/v yeast 102

extract, 2% w/v bacto peptone and 2% w/v D-glucose (YPD) liquid broth on a 103

rotating wheel at 28°C. YPD overnight cultures were used to inoculate 50:50 Eden 104

Valley Chardonnay juice : water, which were grown overnight in cotton-plugged 105

Erlenmeyer flasks at 20°C, and used as starter cultures for 100% Chardonnay juice 106

ferments. Inoculation density was 1 � 106 cells ml-1. The Chardonnay juice contained 107

107 gL-1 each of glucose and fructose, 362 mgL-1 yeast assimilable nitrogen and had a 108

pH of 3.32. All juice was 0.2 μm filtered prior to inoculation. Small volume flask 109

fermentations were performed in triplicate in 200 ml Erlenmeyer flasks with side arm 110

sampling ports, stoppered with either water filled airlocks (SAF) or cotton plugs 111

(AeF), incubated at 20°C with shaking at 150 rpm. Immediately following 112

G

inoculation, a sample was drawn from each of the flasks and added to microtiter plates 113

(MTPs) to a final volume of 200 μl per well. Culture from each flask was used to 114

inoculate eight wells on replicate plates, with one plate prepared for each required 115

time point. The MTPs were sealed with Breathe Easy gas permeable membrane 116

(Diversified Biotech, Boston, M.A., U.S.A.) and kept in a humidified box without 117

shaking at 20°C. 118

Analysis of growth data 119

Yeast growth in Erlenmeyer flasks was followed by measuring optical density at 600 120

nm at 1 cm path length corrected for non-linearity at high optical densities (OD600 121

1cmcorr), using a DU530 spectrophotometer (Beckman Coulter Inc., Fullerton, C.A., 122

U.S.A.). Correction for non-linearity was achieved by dilution prior to measurement 123

such that OD600 was always < 0.5. Yeast growth in the MTPs was followed by 124

measuring the optical density at 630 nm using a Multiskan Ascent spectrophotometer 125

(Thermo Fisher Scientific Inc., Waltham, M.A., U.S.A.). A correction function was 126

used to convert observed MTP optical density (OD630MTPobs) to 1 cm path length 127

OD600 corrected for non-linearity (OD600 1 cmcorr). The correction function was 128

applied to MTP absorbance data so that dry cell weight estimates and comparisons 129

with shake flask cultures could be made. This function was derived by measuring the 130

optical density of a yeast culture dilution series in both spectrophotometers, plotting 131

OD600 1 cmcorr (optical density measure at 1 cm pathlength) against OD630 MTPobs 132

(optical density measure from microplate reader) and performing non-linear, least 133

squares regression using GraphPad Prism (GraphPad Software Inc., La Jolla, C.A., 134

U.S.A.). The best fitting equation was selected based on linear regression of 135

OD600 1 cmcorr vs OD600 1 cm predicted. The following equation predicted 136

H

OD600 1 cmcorr from OD630 MTPobs values between 0.007 and 1.46 with RMSE = 137

0.0676; OD600 1 cmcorr= 0.008160 + [2.743 � OD630 MTPobs] + [-0.3495 � OD630 138

MTPobs2] + [0.6282 × OD630 MTPobs

3]. 139

140

OD600 1 cmcorr was used to estimate biomass accumulation as dry cell weight (DCW). 141

This was achieved by transforming OD600 1 cmcorr data to DCW using the formula 142

DCW = OD600 1 cmcorr � 0.64246. The conversion factor was derived from linear 143

regression of OD600 1 cmcorr vs DCW using strain AWRI 1493 generated following 144

direct measurement of DCW using a moisture balance (AMB50, Inscale Measurement 145

Technology Ltd. Sussex, U.K.). 146

147

Maximum growth rates were estimated from natural log transformed DCW data using 148

the Baranyi-model (Baranyi and Roberts 1994) and custom equation curve-fitting in 149

Prism (GraphPad). Final biomass yields were estimated by taking the exponential of 150

the Ymax term (maximum value obtained from the model) following Baranyi-model 151

fitting. Statistical comparisons were made using one-way ANOVA with significance 152

between treatments evaluated using Tukey’s multiple comparison post test. Results 153

are expressed as means ± 95% standard deviation. 154

Ferment analysis 155

Concentrations of glucose, fructose, glycerol, acetate and ethanol were determined by 156

HPLC on an HP 1100 series using the method of Frayne (1986) except that 5 mM 157

H2SO4 was used as the mobile phase. Calibration curves relating concentration to 158

optical density or refractive index measurements were fitted by least squares 159

regression using Chemstation software (Agilent, Santa Clara, C.A., U.S.A.). For high 160

I

throughput microplate fermentations glucose and fructose concentrations were 161

determined using a Randox kit (Randox Laboratories Ltd., Crumlin, Antrim, United 162

Kingdom). 163

Results 164

This work compared fermentation in unshaken MTPs with fermentations typical of 165

those used at laboratory scale in wine research; shaken 200 ml Erlenmeyer flasks 166

fitted with air-locks. In the absence of agitation yeast cells accumulated as a layer on 167

the base of the MTP. Uneven distributions of this layer, as noted by Warringer and 168

Blomberg (2003) when plates were shaken were not evident in the absorbance profiles 169

of these experiments. Furthermore, the observed optical density of settled cells was 170

comparable to that of suspended cells (see additional file 1). The even growth 171

characteristics and similarity of optical density for suspended and settled cell cultures 172

simplified the experimental protocol and permitted the use of plate readers and 173

incubators without built in shakers for the conduct of screening experiments using 174

MTP fermentations. 175

Effect of culture vessel size on maximum growth rate and biomass yield 176

Maximum growth rate and efficiency of growth (biomass yield) have been described 177

as “principally independent variables reflecting strictly different aspects of cell 178

physiology” (Warringer et al. 2003) and are the two parameters most commonly used 179

to characterize performance of a microbial culture. Figure 1A and Figure 1B show a 180

comparison of final biomass yields and maximum growth rates respectively from 181

cultures in non-shaken MTPs, SAFs and AeFs. 182

183

J

Maximum growth rates were indistinguishable, irrespective of culture vessel type 184

(ANOVA and Tukey’s multiple comparison test, alpha = 0.05). In contrast final 185

biomass levels varied depending on the culture vessel used. Aerobic (AeF) cultures 186

produced 1.6 fold (2.4 gL-1 ± 0.2, n = 6) more biomass than those in either SAFs or 187

the MTPs. 188

189

Evaporative volume loss during fermentation in MTPs was minimized by conducting 190

fermentations in humidified chambers. The total volume loss during fermentation 191

was estimated by weighing MTPs. Average weight loss in MTPs was 14.5% of initial 192

weight. A weight loss of 14.5% translates to a volume loss of approximately 12 193

μl/well (6% of initial volume), once weights are adjusted for change in density from 194

juice to wine. No weight loss was observed in un-inoculated MTPs indicating weight 195

loss was not due to evaporation of water during extended incubation. Therefore, 196

while some weight loss can be accounted for by liberation of CO2 some excess 197

volume loss occurs in MTPs that is not evident in larger scale ferments. While the 198

volume loss does not affect the optical density measurement due to reduced path-199

length (see additional file 2) it does lead to altered cell concentration that must be 200

taken into account when calculations of cell concentration are made. 201

Kinetics of sugar utilization 202

The major substrates in grape-juice fermentation are glucose and fructose, present in 203

equimolar concentrations. The kinetics of utilization of these two sugars differed 204

depending on the culture vessel (Figure 2A, B and Figure 3). Sugar utilization in 205

AeFs was both faster and more efficient than in MTPs or SAFs. With access to 206

oxygen provided by the cotton plug, fermentations in AeF flasks completed both 207

K

glucose and fructose utilization within 70 hours and therefore differences between 208

glucose and fructose utilization were minimal in comparison to the other two vessel 209

types. Glucose was exhausted (< 1 gL-1) in within 85 hours in MTPs and 100 hours in 210

SAFs. MTPs required 105 hours to complete fermentation of fructose, a full day 211

longer than was required for completion of glucose fermentation. SAFs still 212

contained 2 gL-1 fructose at 190 hours. These ‘time-to-dryness’ profiles reflected 213

maximum glucose and fructose utilization rates achieved in different vessel types 214

(Figure 3A and B). 215

Ethanol, glycerol and acetic acid accumulation 216

Ethanol, glycerol and acetic acid are metabolites of interest in many industrial 217

fermentation types, particularly in wine and beer production. The maximum ethanol 218

concentrations achieved in AeFs and SAFs did not differ significantly and were 108.7 219

± 1.3 gL-1 and 111.2 ± 1.2 gL-1 respectively. The maximum ethanol concentration 220

achieved in MTPs was significantly less at 92.4 ± 4.2 gL-1 (P < 0.05). Ethanol 221

formation was most rapid in AeFs (Figure 3A and B) which can largely be attributed 222

to the higher amount of biomass in this flask type. This is indicated by the maximum 223

specific ethanol production rates which are not significantly different from that 224

achieved in MTPs. The maximum specific ethanol production rates in both AeFs and 225

MTPs were significantly higher than that achieved in SAFs (P < 0.001) suggesting 226

that the efficiency of ethanol production is assisted by access to oxygen. We attribute 227

the reduced lower absolute concentration of ethanol in MTPs compared with flask 228

cultures to loss through evaporation even though evaporative water loss is not a 229

problem in the humidified incubators used here. 230

231

L

Maximum glycerol concentrations and rates of production did not differ significantly 232

between the SAFs and MTPs, P > 0.05 (Figure 2E and Figure 3). The maximum 233

concentration of glycerol achieved in AeFs was slightly higher than the other two 234

fermentation types (P < 0.05) although its rate of production did not differ 235

significantly. Maximum acetic acid concentrations were highest in AeFs, reaching 236

1.211 ± 0.02 gL-1 at 120 h, lowest in SAFs (0.478 ± 0.08 gL-1) and intermediate in 237

MTPs (0.727 ± 0.02). Rates of acetic acid production did not differ significantly 238

between any of the flask types, although the time during which the maximum rate of 239

acetic acid accumulation occurred was dependent on the flask type. Although acetic 240

acid concentrations in MTP ferments began to increase at the final time point, this 241

occurred well after sugar depletion and never reached the levels observed in AeFs. 242

Application of microplate fermentation to screening of wine yeast for 243

fermentation performance 244

It has previously been shown, and reproduced in this work, that it is possible to 245

reproduce flask type growth and biomass profiles in MTP cultures (Stitt et al. 2002; 246

Warringer et al. 2003), however screening methods that enable other performance 247

parameters to be assessed are also desirable. Sugar utilization is a key parameter of 248

interest and although the exact SAF sugar utilization profiles could not be reproduced 249

in MTP ferments we explored whether sugar utilization trends in MTP fermentations 250

could be used as predictors of performance in larger scales. 251

252

MTP ferments of 15 different wine yeast strains were evaluated using a sacrificial 253

plate format in which a single MTP was set up for each required time point. The 254

sugar utilization curves of these are shown in Figure 4A. A subset comprising 4 255

M

strains (Shown in bold in Figure 4A) whose sugar utilization profiles covered the full 256

range observed in sacrificial MTP assays was chosen for more detailed 257

characterization in 200 ml ferments (Figure 4B). A comparison of Figures 4A and 4B 258

demonstrates that although the total duration of fermentation is reduced in MTP 259

format, the overall profile of sugar utilization for the different strains is consistent 260

irrespective of the fermentation scale. For example, strain 4 continues to ferment 261

strongly until a residual sugar concentration of 50 gL-1 before its fermentation rate 262

slows, whereas the rate of sugar utilization for the other strains slows at > 100 gL-1. 263

When strains are ranked by time taken to reach a target sugar of < 2 gL-1, strain order 264

does not change when ferments are performed on a larger scale (Table 1). 265

Discussion 266

Results reported in this paper indicate that MTPs can be used to follow growth and 267

metabolite production in fermentations of grape juice by wine yeast. This format 268

shows great promise for high-throughput screening of industrial yeast; in many 269

respects an unshaken MTP grape-juice fermentation is indistinguishable from a self-270

induced anaerobic Erlenmeyer flask fermentation. Fermentations performed in these 271

vessels exhibit similar maximum growth rates and biomass formation, utilize glucose 272

and accumulate glycerol and acetic acid to the same degree and with similar 273

maximum rates, many of which are consistent with previous observations (Berthels et 274

al. 2008; Fleet and Heard 1993; Ooi et al. 2008). 275

276

In other respects ferments performed in MTPs were different from those in either 277

SAFs or AeFs and this reflects, at least in some regards, the limitations inherent in the 278

use of MTPs. For example, while the maximum rate of ethanol accumulation in 279

N

MTPs was similar to that in SAFs during the early phase of production, the maximum 280

specific rate of ethanol production was similar to that in AeFs. Later in fermentation 281

evaporation rates of ethanol from MTPs presumably exceeded production and were 282

particularly pronounced during the final stages of fermentation. Thus, the microplate 283

ferments never attained concentrations of ethanol equivalent to those of flask cultures. 284

285

Significantly higher biomass production is the primary distinguishing feature of AeF 286

fermentations. The higher accumulation of biomass in the AeFs is consistent with 287

increased access to oxygen in this vessel type (Henzler and Schedel 1991). Aside 288

from biomass formation, the impact of culture vessel type was most evident in the 289

profiles of glucose and fructose consumption. Total sugar consumption occurred 290

most rapidly in AeFs with both glucose and fructose utilization effectively complete 291

within 75 hours. The faster completion of fermentation is most likely related to 292

higher biomass accumulation (Varela et al. 2004). Sugar consumption was slowest in 293

the SAFs which showed both the slowest total utilization rate and the largest 294

discrepancy between the time required to ferment glucose and fructose. This 295

discrepancy in glucose and fructose utilization is typical of wine ferments with 296

residual fructose representing a frequent difficulty for the wine industry (Berthels et 297

al. 2004; Bisson 1999; Fleet et al. 1993; Reynolds et al. 2001; Schutz and Gafner 298

1995; Varela et al. 2004). Cultures in MTPs were intermediate in their ferment 299

completion times with fructose utilization accelerating only once glucose 300

consumption was complete. Total sugar consumption was complete in 120 hours, a 301

full 70 hours earlier than in SAFs. This accelerated sugar consumption in MTP 302

ferments indicates that agitation is not required either for access to nutrients other than 303

oxygen or for reproducible fermentation. It is likely that the reduced accumulation of 304

O

ethanol in MTP ferments is at least in part responsible for the accelerated use of sugar 305

compared to SAF ferments. The faster fermentation times in MTPs, however, did not 306

change the overall profile of sugar utilization. We were therefore able to use MTP 307

fermentations as a method to screen wine yeast strains for their sugar utilization 308

capacity, a trait that translated well to large fermentation scales despite the difference 309

in overall fermentation duration between MTP and flask fermentations. 310

Conclusions 311

Grape-juice fermentations performed in MTPs imitate standard, SAFs when 312

comparing most of the major criteria used to assess yeast performance and product 313

profiling, with the exception of time to dryness (i.e. completion of sugar consumption) 314

and ethanol concentration. A simple humidified incubator prevented excessive water 315

evaporation however, evaporation of ethanol still occurred suggesting that micro-316

scale ferments of the design described in this paper would be of limited utility in 317

studies focusing on highly volatile metabolites. The data illustrates that microplate 318

cultures can be used as a replacement for SAFs in some instances and provide a useful 319

and economical platform for the screening of industrial strains and culture media. 320

Authors’ contributions 321

SS, TT and TL carried out fermentations and data acquisition. SS and DC performed 322

data analysis. SS, TT, VJ and PC participated in the design of the study and helped to 323

draft the manuscript. All authors read and approved the final manuscript. 324

P

Acknowledgements 325

This project was supported by Australia’s grape growers and winemakers through 326

their investment body, the Grape and Wine Research and Development Corporation, 327

with matching funds from the Australian Government. We would also like to thank 328

Maurizio Ugliano and Richard Gawel for their comments during the preparation of 329

this manuscript. The AWRI and UA are part of the Wine Innovation Cluster, 330

Adelaide, South Australia. 331

References 332

Baranyi J, Roberts TA (1994) A dynamic approach to predicting bacterial 333

growth in food. International Journal of Food Microbiology 334

23:277-294 335

Berthels NJ, Otero RRC, Bauer FF, Pretorius IS, Thevelein JM (2008) 336

Correlation between glucose/fructose discrepancy and hexokinase 337

kinetic properties in different Saccharomyces cerevisiae wine yeast 338

strains. Applied Microbiology and Biotechnology 77:1083-1091 339

Berthels NJ, Otero RRC, Bauer FF, Thevelein JM, Pretorius IS (2004) 340

Discrepancy in glucose and fructose utilisation during fermentation 341

by Saccharomyces cerevisiae wine yeast strains. Fems Yeast 342

Research 4:683-689 343

Bisson LF (1999) Stuck and sluggish fermentations. American Journal of 344

Enology and Viticulture 50:107-119 345

Blank LM, Kuepfer L, Sauer U (2005a) Large-scale C-13-flux analysis 346

reveals mechanistic principles of metabolic network robustness to 347

null mutations in yeast. Genome Biology 6:R49 348

Q

Blank LM, Lehmbeck F, Sauer U (2005b) Metabolic-flux and network 349

analysis in fourteen hemiascomycetous yeasts. Fems Yeast 350

Research 5:545-558 351

Börner J, Buchinger S, Schomburg D (2007) A high-throughput method 352

for microbial metabolome analysis using gas chromatography/mass 353

spectrometry. Analytical Biochemistry 367:143-151 354

Duetz WA (2007) Microtiter plates as mini-bioreactors: miniaturization 355

of fermentation methods. Trends in Microbiology 15:469-475 356

Duetz WA, Ruedi L, Hermann R, O'Connor K, Buchs J, Witholt B (2000) 357

Methods for intense aeration, growth, storage, and replication of 358

bacterial strains in microtiter plates. Applied and Environmental 359

Microbiology 66:2641-2646 360

Duetz WA, Witholt B (2004) Oxygen transfer by orbital shaking of 361

square vessels and deepwell microtiter plates of various 362

dimensions. Biochemical Engineering Journal 17:181-185 363

Ferreira-Torres C, Micheletti M, Lye GJ (2005) Microscale process 364

evaluation of recombinant biocatalyst libraries: application to 365

Baeyer-Villiger monooxygenase catalysed lactone synthesis. 366

Bioprocess and Biosystems Engineering 28:83-93 367

Fleet GH, Heard GA (1993) Yeasts - Growth during fermentation. In: 368

Fleet GH (ed) Wine microbiology and biotechnology, Harwood 369

Academic Publishers, Chur, Switzerland, pp 27-54 370

Frayne RF (1986) Direct analysis of the major organic components in 371

grape must and wine using high-performance liquid-372

chromatography. American Journal of Enology and Viticulture 373

37:281-287 374

R

Henzler HJ, Schedel M (1991) Suitability of the shaking flask for oxygen 375

supply to microbiological cultures. Bioprocess Engineering 7:123-376

131 377

Hermann R, Lehmann M, Buchs J (2003) Characterization of gas-liquid 378

mass transfer phenomena in microtiter plates. Biotechnology and 379

Bioengineering 81:178-186 380

Kensy F, Zimmermann HF, Knabben I, Anderlei T, Trauthwein H, 381

Dingerdissen U, Buchs J (2005) Oxygen transfer phenomena in 48-382

well microtiter plates: Determination by optical monitoring of 383

sulfite oxidation and verification by real-time measurement during 384

microbial growth. Biotechnology and Bioengineering 89:698-708 385

Maresova L, Sychrova H (2007) Applications of a microplate reader in 386

yeast physiology research. Biotechniques 43:667-672 387

Micheletti M, Barrett T, Doig SD, Baganz F, Levy MS, Woodley JM, 388

Lye GJ (2006a) Fluid mixing in shaken bioreactors: Implications 389

for scale-up predictions from microlitre-scale microbial and 390

mammalian cell cultures. Chemical Engineering Science 61:2939-391

2949 392

Micheletti M, Lye GJ (2006b) Microscale bioprocess optimisation. 393

Current Opinion in Biotechnology 17:611-618 394

Ooi BG, Wanamaker LE, Markuszewski BM, Chong NS (2008) Genetic 395

and enological analysis of selected Saccharomyces cerevisiae 396

strains for wine production. International Journal of Food Science 397

and Technology 43:1111-1120 398

Reynolds AG, Edwards CG, Cliff MA, Thorngate JH, Marr JC (2001) 399

Evaluation of yeast strains during fermentation of Riesling and 400

Chenin blanc musts. American Journal of Enology and Viticulture 401

52:336-344 402

S

Schutz M, Gafner J (1995) Lower fructose uptake capacity of genetically 403

characterized strains of Saccharomyces bayanus compared to 404

strains of Saccharomyces cerevisiae - A likely cause of reduced 405

alcoholic fermentation activity. American Journal of Enology and 406

Viticulture 46:175-180 407

Stitt DT, Nagar MS, Haq TA, Timmins MR (2002) Determination of 408

growth rate of microorganisms in broth from oxygen-sensitive 409

fluorescence plate reader measurements. Biotechniques 32:684-689 410

Varela C, Pizarro F, Agosin E (2004) Biomass content governs 411

fermentation rate in nitrogen-deficient wine musts. Applied and 412

Environmental Microbiology 70:3392-3400 413

Warringer J, Blomberg A (2003) Automated screening in environmental 414

arrays allows analysis of quantitative phenotypic profiles in 415

Saccharomyces cerevisiae. Yeast 20:53-67 416

Zimmermann HF, John GT, Trauthwein H, Dingerdissen U, Huthmacher 417

K (2003) Rapid evaluation of oxygen and water permeation 418

through microplate sealing tapes. Biotechnology Progress 19:1061-419

1063 420

Zimmermann HF, Rieth J (2007) A fully automated robotic system for 421

high throughput fermentation. Clinics in Laboratory Medicine 422

27:209-214 423

424

425

Figures 426

Figure 1 Impact of culture vessel type on culture physiology 427

Chardonnay fermentations using strain AWRI 1493 were performed in self-induced 428

anaerobic flasks (SAF), aerobic flasks (AeF) or microtiter plates (MTP). Growth 429

T

curves showing biomass accumulation (A) and maximum specific growth rates (B) 430

were estimated from optical density measurements as described in the materials and 431

methods. Error bars show ± standard deviation. Columns show means from two 432

independent experiments. Each treatment was performed in triplicate within each 433

experiment. 434

Figure 2 Effect of culture vessel type on sugar utilization and metabolite 435

production 436

Chardonnay fermentations using strain AWRI 1493 were performed in SAFs (�), 437

AeFs (�) or microtiter plates (�). Glucose (A), fructose (B), ethanol (C), glycerol 438

(D) and acetic acid (E) concentrations were determined by HPLC as described in 439

materials and methods. Data points show the means from two independent 440

experiments. Each treatment was performed in triplicate in each experiment. Error 441

bars show ± standard deviation. 442

Figure 3 Estimated sugar utilization and major metabolite production rates in 443

different culture vessels 444

Maximal rates (A) and maximum specific rates (B) were estimated from the first 445

derivative of curves fitted to each data set shown in Figure 2. Specific rates were 446

calculated from by dividing maximum rates by the biomass concentration estimated at 447

the time the maximum rate was achieved. Self-induced anaerobic flasks (SAF), 448

aerobic flasks (AeF) and microtiter plates (MTP). 449

U

Figure 4 Screening of yeast strain fermentation performance and validation 450

using selected strains at a larger scale. 451

The fermentation performance of 20 industrial strains was evaluated in Chardonnay 452

juice using MTP sacrificial plates (A). One plate was prepared for each time point, 453

each strain was fermented in four wells on each plate. Each time point in (A) is the 454

average sugar concentration from four wells on one plate. Four strains, representing a 455

cross-section of performance profiles, were selected for further characterization using 456

200 ml air-locked fermenters (SAF) in the same Chardonnay juice (B). Each time 457

point in B is the average total sugar concentration from 3 fermentations. Error bars 458

indicate ± standard deviation. 459

460

Table 1 Strain ranking based on the time taken to reach a target sugar of < 2 gL-1 461

Strain Time (MTP) Rank MTP Time (SAF) Rank (SAF)

1 120 4 280 4

2 110 2,3 265 3

3 110 2,3 209 2

4 95 1 180 1

462

463

464

V

Additional files 465

Additional file 1 466

File Format : PDF 467

Title: Optical density of settled and suspended cells 468

Description: We compared the effect of cell settling on the measurement of optical 469

density in MTPs. The optical density of settled cells was measured at 630 nm. 470

Immediately following measurement, settled cells were suspended by pipetting 471

followed by vortexing. Surface bubbles were removed with a flame and the optical 472

density was again determined. 473

Additional file 2 474

File Format : PDF 475

Title: The effect of MTP well volume on optical density of settled cells 476

Description: Graph 2 shows the effect of reducing well volume on the optical density 477

of settled cells. 200 μl of cell a flask culture was pipetted into a MTP, the cells we 478

allowed to settle and the optical density was determined at 630 nm. 20 μl was 479

removed from each well and the MTP was measured again. This was repeated another 480

3 times. The graph shows the optical density of all wells in the MTP following each 481

subsequent removal of 20 μl. 482

483

484

485

486

487

W

0 50 100 150 2000

2

4

6

8

10

AeFSAF

MTP

time (hours)

dry

cell w

eigh

t (gL

-1)

SAF AeF MTP0.00

0.05

0.10

0.15

0.20

0.25

grow

th ra

te (h

-1)

A

B

488

489

Figure 1 490

491

492

493

494

495

496

497

498

X

0 50 100 150 2000

20

40

60

80

100

120SAFAeFMTP

[Glu

cose

] gL-1

0 50 100 150 2000

20

40

60

80

100

120

[Fru

ctos

e] g

L-1

0 50 100 150 2000

20

40

60

80

100

120

[Eth

anol

] gL-1

0 50 100 150 200012345678

[Gly

cero

l] gL

-1

0 50 100 150 2000.00.20.40.60.81.01.21.41.6

[Ace

tic a

cid]

gL

-1

time (hours)

A

B

D

E

C

499

Figure 2 500

Y

Glucos

e

Fructos

e

Ethano

l

Glycero

l

Acetat

e0

1

2

3

4

5

0.00

0.05

0.10

0.15

0.20

0.25SAF AeF MTPG

luco

se, F

ruct

ose,

Eth

anol

gL-1

h-1

Glycerol, Acetic acid

gL-1h

-1

Glucos

e

Fructos

e

Ethano

l

Glycero

l

Acetat

e0.0

0.5

1.0

1.5

2.0

0.00

0.05

0.10

0.15

Glu

cose

, Fru

ctos

e, E

than

olgL

-1h-1

gDC

W-1

Glycerol, Acetic acidgL

-1h-1gD

CW

-1

A

B

501

Figure 3 502

503

504

Z

0 100 200 3000

50

100

150

200

250

time (hours)

tota

l sug

ar [g

L-1]

0 50 100 1500

50

100

150

200

250 Strain 1

Strain 4Strain 3Strain 2

Other strains in screen

time (hours)

tota

l sug

ar [g

L-1]

505

Figure 4 506

507

508

509

110

REFERENCES

111

Adams, A., Gottschling, D. E., Kaiser, C. and Stearns, T. (1997). Methods in Yeast Genetics. Cold Spring Harbor Laboratory Press.

Alexandre, H. and Charpentier, C. (1998). Biochemical aspects of stuck and sluggish fermentation in grape must. Journal of Industrial Microbiology and Biotechnology 20, 20-27.

Arroyo-Lopez, F. N., Querol, A. and Barrio, E. (2009). Application of a substrate inhibition model to estimate the effect of fructose concentration on the growth of diverse Saccharomyces cerevisiae strains. Journal of Industrial Microbiology & Biotechnology 36, 663-669.

Barre, P., Vezinhet, F., Dequin, S. and Blondin, B. (1993). Genetic improvement of wine yeast. In Wine Microbiology and Biotechnology, Fleet, G. H. (Ed) Hardwood Academic Publishers, Sydney, pp. 265-287.

Berthels, N. J., Cordero Otero, R. R., Bauer, F., F., Thevelein, J. M. and Pretorius, I. S. (2004). Discrepancy in glucose and fructose utilisation during fermentation by Saccharomyces cerevisiae wine yeast strains. FEMS Yeast Research 4, 683-689.

Berthels, N. J., Cordero Otero, R. R., Bauer, F. F., Pretorius, I. S. and Thevelein, J. M. (2008). Correlation between glucose/fructose discrepancy and hexokinase kinetic properties in different Saccharomyces cerevisiae wine yeast strains. Applied Microbiology and Biotechnology 77, 1083-1091.

Bisson, L. F. (1988). High-affinity glucose transport in Saccharomyces cerevisiae is under general glucose repression control. Journal of Bacteriology 170, 4838-4845.

Bisson, L. F. (1999). Stuck and sluggish fermentations. American Journal of Enology and Viticulture 50, 107-119.

Bisson, L. F. and Block, D. E. (2002). Ethanol tolerance in Saccharomyces. Biodiversity and Biotechnology of Wine Yeast, M. Ciani. Trivandrum, Kerala, India, Research Signpost, 85-98.

Bisson, L. F. and Butzke, C. E. (2000). Diagnosis and rectification of stuck and sluggish fermentations. American Journal of Enology and Viticulture 51, 168-177.

Bisson, L. F. and Fraenkel, D. G. (1983). Involvement of kinases in glucose and fructose uptake by Saccharomyces cerevisiae. Proceedings of the National Academy of Sciences of the United States of America 80, 1730-1734.

Boehringer-Mannheim (1989). D-glucose/D-fructose. In Methods of Biochemical Analysis and Food Analysis, Boehringer Mannheim, pp. 50-55.

Boulton, R. B., Singleton, V. L., Bisson, L. F. and Kunkee, R. E. (1998). Principles and Practices of Winemaking. Aspen Publishers, Gaithersburg.

112

Brock, T. D., Madigan, M. T., Mertinko, J. M. and Parker, J. (1994). Biology of Microorganisms, 7th edition. Prentice Hall, Englewood Cliffs.

Brown, C. J., Todd, K. M. and Rosenzweig, R. F. (1998). Multiple duplications of yeast hexoxe transport gene in response to selection in a glucose-limited environment. Molecular Biology and Evolution 15, 931-942.

Brown, S. W. and Oliver, S. G. (1982). Isolation of ethanol-tolerant mutants of yeast by continuous selection. European Journal of Applied Microbiology and Biotechnology 16, 119-122.

Butler, P. R., Brown, M. and Oliver, S. G. (1996). Improvement of antibiotic titers from Streptomyces bacteria by interactive continuous selection. Biotechnology and Bioengineering 49, 185-196.

Cairns, J., Overbaugh, J. and Miller, S. M. (1988). The origin of mutants. Nature 335, 142-145.

Cavazza, A., Poznanski, E. and Trioli, G. (2004). Restart of fermentation of simulated stuck wines by direct inoculation of active dry yeasts. American Journal of Enology and Viticulture 55, 160-167.

Chambers, P. J., Bellon, J. R., Schmidt, S. A., Varela, C. and Pretorius, I. S. (2007). Non-genetic engineering approaches for isolating and generating novel yeast for industrial applications. In Diversity and Potential Biotechnological Application of Yeast, Kunze, G. and Satyanarayana, T. (Eds).

Ciriacy, M. and Reifenberger, E. (1997). Hexose transport. In Yeast Sugar Metabolism, Zimmermann, F. K. and Entian, K. D. (Eds), Technomic, Lancaster, Pennsylvania, pp. 45-65.

De Deken, R. H. (1966). The Crabtree effect: a regulatory system in yeast. Journal of General Microbiology 44, 149-156.

Diderich, J. A., Schepper, M., van Hoek, P., Luttik, M. A., van Dijken, J. P., Pronk, J. T., Klaassen, P., Boelens, H. F., de Mattos, M. J., van Dam, K. and Kruckeberg, A. L. (1999). Glucose uptake kinetics and transcription of HXT genes in chemostat cultures of Saccharomyces cerevisiae. Journal of Biological Chemistry 274, 15350-15359.

Diderich, J. A., Schuurmans, J. M., Van Gaalen, M. C., Kruckeberg, A. L. and Van Dam, K. (2001). Functional analysis of the hexose transporter homologue HXT5 in Saccharomyces cerevisiae. Yeast 18, 1515-1524.

Dumont, A., Raynal, C., Raginel, F. and Ortiz-Julien, A. (2009). The ability of wine yeast to consume fructose. The Australian & New Zealand Grapegrower & Winemaker 543, 52-56.

Elbing, K., Larsson, C., Bill, R. M., Albers, E., Snoep, J. L., Boles, E., Hohmann, S. and Gustafsson, L. (2004). Role of hexose transport in control of glycolytic flux in Saccharomyces cerevisiae. Applied and Environmental Microbiology 70, 5323-5330.

113

Fink, G. R. (1970). The biochemical genetics of yeast. In Methods in Enzymology, Volume 17, Part 1. Herbert, T. and Celia White, T. (Eds), Academic Press, pp. 59-78.

Fleet, G. H. (1992). Wine Microbiology and Biotechnology. Harwood academic publishers, Sydney.

Fleet, G. H. (2008). Wine yeasts for the future. FEMS Yeast Research 8, 979-995.

Fleet, G. H., Lafon Lafourcade, S. and Ribereaugayon, P. (1984). Evolution of yeasts and lactic-acid bacteria during fermentation and storage of Bordeaux wines. Applied and Environmental Microbiology 48, 1034-1038.

Foster, P. L. (1992). Directed mutation: between unicorns and goats. Journal of Bacteriology 174, 1711-1716.

Foster, P. L. and Cairns, J. (1992). Mechanisms of directed mutation. Genetics 131, 783-789.

Gafner, J. and Schütz, M. (1996). Impact of glucose-fructose-ratio on stuck fermentations: practical experiences to restart stuck fermentations. Wein-Wissenschaft, Vitic. Enol. Sci. 51, 214-218.

Giudici, P., Solieri, L., Pulvirenti, A. M. and Cassanelli, S. (2005). Strategies and prospectives for genetic improvement of wine yeasts. Applied Microbiolical Biotechnology 66, 622-628.

Gonçalves, P., Rodrigues de Sousa, H. and Spencer-Martins, I. (2000). FSY1, a novel gene encoding a specific fructose/H+ symporter in the type strain of Saccharomyces carlsbergensis. Journal of Bacteriology 182, 5628-5630.

Guillaume, C., Delobel, P., Sablayrolles, J. M. and Blondin, B. (2007). Molecular basis of fructose utilization by the wine yeast Saccharomyces cerevisiae: a mutated HXT3 allele enhances fructose fermentation. Applied and Environmental Microbiology 73, 2432-2439.

Hall, B. G. (1988). Adaptive evolution that requires multiple spontaneous mutations. I. Mutations involving an insertion sequence. Genetics 120, 887-897.

Hall, B. G. (1989). Selection, adaptation, and bacterial operons. Genome 31, 265-271.

Hall, B. G. (1992). Selection-induced mutations occur in yeast. Genetics 89, 4300-4303.

Hall, B. G. (1997). On the specificity of adaptive mutation. Genetics 145, 39-44.

Heard, G. M. and Fleet, G. H. (1985). Growth of natural yeast flora during the fermentation of inoculated wines. Applied and Environmental Microbiology 50, 727-728.

Helling, R. B., Vargas, C. N. and Adams, J. (1987). Evolution of Escherichia coli during growth in a constant environment. Genetics 116, 349-358.

114

Henschke, P. A. and Jiranek, V. (1993). Yeast - Metabolism of nitrogen compound. In Wine Micorobiology and Biotechnology, Fleet, G. H. (Ed) Harwood academic publisher, Chur, Switzerland, pp. 77-164.

Hornsey, I. (2007). The Chemistry and Biology of Winemaking. The Royal Society of Chemistry, Cambridge.

Jiranek, V., Langridge, P. and Henschke, P. A. (1995). Amino-Acid and ammonium utilization by Saccharomyces cerevisiae wine yeasts from a chemically-defined medium. American Journal of Enology and Viticulture 46, 75-83.

Júnior, M. M., Batistote, M. and Ernandes, J. R. (2008). Glucose and fructose fermentation by wine yeasts in media containing structurally complex nitrogen sources. Journal of the The Institute of Brewing and Distilling 114, 199-204.

Kao, K. C. and Sherlock, G. (2008). Molecular characterization of clonal interference during adaptive evolution in asexual populations of Saccharomyces cerevisiae. Nature Genetics 40, 1499-1504.

Karpel, J. E., Place, W. R. and Bisson, L. F. (2008). Analysis of the major hexose transporter genes in wine strains of Saccharomyces cerevisiae. American Journal of Enology and Viticulture 59, 265-275.

Kunkee, R. E. and Amerine, A. M. (1970). Yeast in winemaking. In The Yeast, Volume 3. Rose, A. H. and Harrison, J. S. (Eds), Academy Press, New York, pp. 5-57.

Lambrechts, M. G. and Pretorius, I. S. (2000). Yeast and its importance to wine aroma: A review. South African Journal of Enology and Viticulture 21, 97-129.

Lane, P. G., Oliver, S. G. and Butler, P. R. (1999). Analysis of a continuous-culture technique for the selection of mutants tolerant to extreme environmental stress. Biotechnology and Bioengineering 65, 397-406.

Lee, C. K. (1987). The chemistry and biochemistry of the sweetness of sugars. Advances in Carbohydrate Chemistry and Biochemistry 45, 199–351.

Lema, C., GarciaJares, C., Orriols, I. and Angulo, L. (1996). Contribution of Saccharomyces and non-Saccharomyces populations to the production of some components of Albarino wine aroma. American Journal of Enology and Viticulture 47, 206-216.

Luria, S. E. and Delbruck, M. (1943). Mutations of bacteria from virus sensitivity to virus resistance. Genetics 28, 491-511.

Luyten, K., Riou, C. and Blondin, B. (2002). The hexose transporters of Saccharomyces cerevisiae play different roles during enological fermentation. Yeast 19, 713-726.

115

Malherbe, S., Bauer, F. F. and Du Toit, M. (2007). Understanding problem fermentations - A review. South African Journal of Enology and Viticulture 28, 169-186.

Martinez, J. A., Melgosa, M., Perez, M. M., Hita, E. and Negueruela, A. I. (2001). Note. Visual and instrumental color evaluation in red wines. Food Science and Technology International 7, 439-444.

McBryde, C., Gardner, J. M., Lopes, M. d. B. and Jiranek, V. (2006). Generation of novel wine yeast strains by adaptive evolution. American Journal of Enology and Viticulture 57, 423-430.

Meneses, F. J. and Jiranek, V. (2002). Expression patterns of genes and enzymes involved in sugar catabolism in industrial Saccharomyces cerevisiae strains displaying novel fermentation characteristics. Journal of the Institute of Brewing 108, 322-335.

Meneses, F. J., Henschke, P. A. and Jiranek, V. (2002). A survey of industrial strains of Saccharomyces cerevisiae reveals numerous altered patterns of maltose and sucrose utilisation. Journal of the Institute of Brewing 108, 310-321.

Mestre, A. C. and Mestre, J. A. (1946). Fermentaciones comparativos con diferentes levaduras. Minist Agric Inst Nac Invest Agron Estacion Vitic. y Enol., Villagranca del Panades, Cuarderno 68, 1-28.

Muller, H. J. (1932). Some genetics aspect of sex. American Naturalist 66, 118-138.

Ness, F., Lavallee, F., Dubourdieu, D., Aigle, M. and Dulau, L. (1993). Identification of yeast strains using the polymerase chain-reaction. Journal of the Science of Food and Agriculture 62, 89-94.

Novak, S., D'Amore, T. and Stewart, G. G. (1990). 2-Deoxy-D-glucose resistant yeast with altered sugar transport activity. FEBS Letters 269, 202-204.

Orr, H. A. and Otto, S. P. (1994). Does diploidy increase the rate of adaptation? Genetics 136, 1475-1480.

Ozcan, S., Dover, J., Rosenwald, A. G., Wolfl, S. and Johnston, M. (1996). Two glucose transporters in Saccharomyces cerevisiae are glucose sensors that generate a signal for induction of gene expression. Proceedings of the National Academy of Sciences of the United States of America 93, 12428-12432.

Ozcan, S. and Johnston, M. (1995). Three different regulatory mechanisms enable yeast hexose transporter (HXT) genes to be induced by different levels of glucose. Molecular and Cellular Biology 15, 1564-1572.

Ozcan, S. and Johnston, M. (1999). Function and regulation of yeast hexose transporters. Microbiology and Molecular Biology Reviews 63, 554-569.

116

Paquin, C. and Adams, J. (1983). Frequency of fixation of adaptive mutations is higher in evolving diploid than haploid yeast populations. Nature 302, 495-500.

Perez, M., Luyten, K., Michel, R., Riou, C. and Blondin, B. (2005). Analysis of Saccharomyces cerevisiae hexose carrier expression during wine fermentation: both low- and high-affinity Hxt transporters are expressed. FEMS Yeast Research 5, 351-361.

Postma, E., Verduyn, C., Scheffers, W. A. and Van Dijck, J. P. (1989). Enzymic analysis of Crabtree effect in glucose-limited chemostat cultures of Saccharomyces cerevisiae. Applied and Environmental Microbiology 55, 468-477.

Pretorius, I. S. (2000). Tailoring wine yeast for the new millenium; novel approaches to the ancient art of winemaking. Yeast 16, 675-729.

Pretorius, I. S. and van der Westhuizen, T. J. (1991). The impact of yeast genetics and recombinant DNA technology on the wine industry. South African Journal of Enology and Viticulture 12, 3-29.

Rainieri, S. and Pretorius, I. S. (2000). Selection and improvement of wine yeasts. Annals of Microbiology 50, 15-31.

Ramos, J., Szkutnicka, K. and Cirillo, V. P. (1988). Relationship between low- and high-affinity glucose transport systems of Saccharomyces cerevisiae. Journal of Bacteriology 170, 5375-5377.

Rapp, A. (1998). Volatile flavour of wine: correlation between instrumental analysis and sensory perception. Nahrung 42, 351-363.

Reifenberger, E., Freidel, K. and Ciriacy, M. (1995). Identification of novel HXT genes in Saccharomyces cerevisiae reveals the impact of individual hexose transporters on glycolytic flux. Molecular Microbiology 16, 157-167.

Ribereau-Gayon, P., Dubourdieu, D. and Doneche, B. (2005). Handbook of Enology, Volume 1, 2nd edition. Wiley-Blackwell.

Robinson, J. (1994). The Oxford Companion to Wine. Oxford University Press, Oxford.

Rolland, F., Wanke, V., Cauwenberg, L., Ma, P. S., Boles, E., Vanoni, M., de Winde, J. H., Thevelein, J. M. and Winderickx, J. (2001a). The role of hexose transport and phosphorylation in cAMP signalling in the yeast Saccharomyces cerevisiae. Fems Yeast Research 1, 33-45.

Rolland, F., Winderickx, J. and Thevelein, J. M. (2001b). Glucose-sensing mechanisms in eukaryotic cells. Trends in Biochemical Sciences 26, 310-317.

Romano, P., Fiore, C., Paraggio, M., Caruso, M. and Capece, A. (2003). Function of yeast species and strains in wine flavour. International Journal of Food Microbiology 86, 169-180.

117

Rose, A. H. and Harrison, J. S. (1969). The Yeast, Volume 1. Academic Press, New York.

Salmon, J.-M. (1989). Effects of sugar transport inactivation in Saccharomyces cerevisiae on sluggish and stuck enological fermentations. Applied and Environmental Microbiology 55, 953-958.

Salmon, J. M., Vincent, O., Mauricio, J. C., Bely, M. and Barre, P. (1993). Sugar-transport inhibition and apparent loss of activity in Saccharomyces cerevisiae as a major limiting factor of enological fermentations. American Journal of Enology and Viticulture 44, 56-64.

Samuel, D. (1996). Investigation of ancient Egyptian baking and brewing methods by correlative microscopy. Science 273, 488-490.

Santos, J., Sousa, M. J., Cardoso, H., Inacio, J., Silva, S., Spencer-Martins, I. and Leao, C. (2008). Ethanol tolerance of sugar transport, and the rectification of stuck wine fermentations. Microbiology 154, 422-430.

Schreier, P. and Jennings, W. G. (1979). Flavor composition of wines - Review. Critical Reviews in Food Science and Nutrition 12, 59-111.

Schutz, M. and Gafner, J. (1993). Sluggish alcoholic fermantation in relation to alterations of the glucose-fructose ratio. Chemie, Mikrobiologie, Technologie der Lebensmittel 15, 73-78.

Serrano, R. and Delafuente, G. (1974). Regulatory properties of the constitutive hexose transport in Saccharomyces cerevisiae. Molecular and Cellular Biochemistry 5, 161-171.

Spencer, J. F. T. and Spencer, D. M. (1983). Genetic improvement of industrial yeasts. Annual Review of Microbiology 37, 121-142.

Stanley, D., Bandara, A., Fraser, S., Chambers, P. J. and Stanley, G. A. (2010). The ethanol stress response and ethanol tolerance of Saccharomyces cerevisiae. Journal of Applied Microbiology, doi:10.1111/j.1365-2672.2009.04657.x.

Steele, D. F. and Jinks-Robertson, S. (1992). An examination of adaptive reversion in Saccharomyces cerevisiae. Genetics 132, 9-21.

Tronchoni, J., Gamero, A., Arroyo-Lopez, F. N., Barrio, E. and Querol, A. (2009). Differences in the glucose and fructose consumption profiles in diverse Saccharomyces wine species and their hybrids during grape juice fermentation. International Journal of Food Microbiology 134, 237-243.

Ugliano, M., Fedrizzi, B., Siebert, T., Travis, B., Magno, F., Versini, G. and Henschke, P. A. (2009). Effect of nitrogen supplementation and Saccharomyces species on hydrogen sulfide and other volatile sulfur compounds in Shiraz fermentation and wine. Journal of Agricultural and Food Chemistry 57, 4948-4955.

118

Ugliano, M. and Henschke, P. A. (2010). Comparison of three methods for accurate quantification of hydrogen sulfide during fermentation. Analytica Chimica Acta 660, 87-91.

Varela, C., Cardenas, J., Melo, F. and Agosin, E. (2005). Quantitative analysis of wine yeast gene expression profiles under winemaking conditions. Yeast 22, 369-383.

Verwaal, R., Paalman, J. W., Hogenkamp, A., Verkleij, A. J., Verrips, C. T. and Boonstra, J. (2002). HXT5 expression is determined by growth rates in Saccharomyces cerevisiae. Yeast 19, 1029-1038.

Wahl, L. M. and Krakauer, D. C. (2000). Models of experimental evolution: the role of genetic chance and selective necessity. Genetics 156, 1437-48.

Walker, M. E., Gardner, J. M., Vystavelova, A., McBryde, C., de Barros Lopes, M. and Jiranek, V. (2003). Application of the reuseable, KanMX selectable marker to industrial yeast: construction and evaluation of heterothallic wine strains of Saccharomyces cerevisiae, possessing minimal foreign DNA sequences. FEMS Yeast Research 4, 339-347.

Wang, D., Xu, Y., Hu, J. and Zhao, G. (2004). Fermentation kinetics of different sugars by apple wine yeast Saccharomyces cerevisiae. J. Inst. Brew. 110, 340-346.

Wick, L. M., Weilenmann, H. and Egli, T. (2002). The apparent clock-like evolution of Escherichia coli in glucose-limited chemostats is reproducible at large but not at small population sizes and can be explained with Monod kinetics. Microbiology 148, 2889-28902.

Wieczorke, R., Krampe, S., Weierstall, T., Freidel, K., Hollenberg, C. P. and Boles, E. (1999). Concurrent knock-out of at least 20 transporter genes is required to block uptake of hexoses in Saccharomyces cerevisiae. FEBS Letters 464, 123-128.

Zambonelli, C. (1998). Microbiologia e Biotecnologia dei Vini. Edagricole, Bologna.

Zeyl, C. (2004). Capturing the adaptive mutation in yeast. Research in Microbiology 155, 217-223.

Zeyl, C. (2005). The number of mutations selected during adaptation in a laboratory population of Saccharomyces cerevisiae. Genetics 169, 1825-1831.

Zeyl, C., Vanderford, T. and Carter, M. (2003). An evolutionary advantage of haploidy in large yeast populations. Science 299, 555-558.