AN INTEGRATED APPROACH FOR THE GENETIC

DISSECTION OF COMPLEX TRAITS IN YEAST

THESIS SUBMITTED FOR THE DEGREE "DOCTOR OF PHILOSOPHY"

BY

GAL-HAGIT ROMANO

SUBMITTED TO THE SENATE OF TEL-AVIV UNIVERSITY

OCTOBER 2009

This work was carried out under the supervision o

Professor Martin Kupiec

"For the truth is that what one really needs is not Nobel laureates but love." "For the truth is that what one really needs is not Nobel laureates but love." "For the truth is that what one really needs is not Nobel laureates but love." "For the truth is that what one really needs is not Nobel laureates but love."

George Wald, George Wald, George Wald, George Wald,

Nobel laureate,Nobel laureate,Nobel laureate,Nobel laureate, Physiology and Physiology and Physiology and Physiology and Medicine, 1967Medicine, 1967Medicine, 1967Medicine, 1967

Acknowledgments

I would like to thank all the people that have been part of my journey during my PhD

studies. This thesis could not have been accomplished without the support of my

parents, my sisters, my friends, my colleagues and my teachers.

Above all I am grateful to Martin Kupiec for his scientific guidance, for his patient,

intelligence and humor, and for our challenging and fruitful discussions.

I would also like to thank my friends at the lab for the wonderful long hours together.

I am indebted to Prof. Moshe Mevarech and Prof. Danny Segal for the encouragement

and for their great insights and wisdom.

Finally, I feel fortunate to travel side by side with Rakefet Bar-Kama. I thank her for

the immeasurable love and for the drops of 'inner peace' in my life.

Table of Contents

Abstract

Introduction 1

1. Mendelian Traits and Complex Traits 1

2. Evolutionary Aspects of Quantitative Traits 3

3. The Challenges in Genetic Dissection of Complex Traits 5

4. Strategies for Genetic Dissection of Complex Traits 7

4.1 QTL Detection by Marker-Trait Association in a Whole Genome Scan 7

4.1.1. High Resolution Genotyping Methods 8

4.2. Advanced Methods for Improving the Mapping Resolution 12

4.2.1. Selective Genotyping 12

4.2.2. Selective Genotyping of DNA Pools 13

4.2.3. Congenic Lines 14

4.3. High Resolution Fine Mapping and Proof of Causation 15

5. Yeast as a Model Organism 17

5.1. The Yeast Deletion Library 18

6. Experimental Evolution with Yeast 19

7. The Ability of Yeast Cells to Grow at High pH as a Complex Trait 21

8. Working Hypothesis 23

Materials and Methods 24

Results 44

1. Characterizing the Ability to Grow at High pH 47

1.1. The Ability to Grow Under Alkali Stress is a Quantitative Trait 48

1.2. Estimating the Heritability and the Number of QTLs that Affect the Trait 50

2. Dissecting the Genetic Network 51

2.1. In-Lab Evolution 52

2.1.1. Enrichment for Beneficial Mutation in Several ILE Lines 52

2.1.2. Several Genetic Networks have Evolved During the ILE 53

2.2. Identifying QTLs Using Congenic Lines 54

2.2.1. Constructing the Congenic Lines 54

2.2.2. Only One Genetic Network Contributes to the MP Phenotype in the

Congenic Lines 55

3. Genotyping 56

3.1. Genotyping Using Genomic Mismatch Scanning (GMS) Method 56

3.2. Genotyping the Congenic Lines Using Y98 Oligonucleotide Arrays 57

3.3. Genotyping Using Tiling Arrays 60

3.4. QTLs Identification in the ILE 60

4. Validation of the Genes Set Identified in the ILE Experiment 65

4.1. Reciprocal Hemizygotes 66

4.2. One SNP Can Affect the Phenotype Via Two QTLs 68

4.3. Allele Swapping 69

5. QTLs Identification in the Congenic Lines 70

6. Fine Mapping in Congenic Lines 74

7. Validation of the Results Obtained from the Congenic Lines 89

Discussion 92

1. QTLs Dissection – Challenges and Achievements 93

1.1. The Advantages of a Combined Strategy for QTLs Dissection 94

1.2. Fine Mapping and Proof of Causation 95

1.3. Revealing Genetic Interactions Among Different Loci 97

2. Adaptation to Alkali Stress 100

2.1. The Role of Metal Transporters in Alkali Stress Resistance 101

2.2. Genes Encoding Cell Wall Proteins Affect Growth at High pH 102

2.3. The role of Ubiquitin Ligases in Alkali Stress Resistance 103

2.4. Additional Genes Affecting Alkali Stress Resistance 105

3. Adaptation - A Lesson from QTLs Dissection 106

3.1. Mutations in Regulatory Genes Shape the Architecture of Complex Traits 106

3.2. The Number of QTLs Affecting a Trait and Their Size Effect 107

3.3. Dynamics of Adaptation 109

3.4. Is the Adaptation Process Repeatable? 111

Bibliography 115

Abstract

Many traits in nature as well as several traits with practical importance, such

as crop yield in plants and susceptibility to various diseases in humans, are affected

by multiple genes and environmental factors. These traits are termed complex traits

and in contrast to mongenic or Mendelian traits, usually have quantitative phenotypes

with subtle differences among individuals. However, it is becoming increasingly

evident that even Mendelian phenotypes can differ in subtle or profound ways due to

genetic background or modifier genes. The propensity of genetic background to

modify the phenotypic expression of most Mendelian traits suggests that few if any

traits are truly monogenic and

that instead most are genetically complex.

Understanding the architecture of complex traits has become the new frontier of

genetic research, and many studies have greatly contributed to this field using

different strategies. However, the identification of genes underlying complex traits

[Quantitative Trait Loci (QTLs)] has encountered significant difficulties, and, despite

major efforts during the past decade, is still a major challenge. By contrast, the

mapping of genes that underlie Mendelian phenotypes has been spectacularly

successful. Not surprisingly, it has been suggested that alternative approaches for the

molecular dissection of complex traits should be developed. The challenge in QTLs

mapping resides in the fact that each QTL only marginally contributes to the studied

phenotype. In addition, each complex phenotype can result from different allelic

combinations, thus the phenotypic-genotypic correlation is low. To make things more

complicated, epistatic interactions and genetic background can alter or mask the

contribution of some QTLs.

This work presents a new global strategy for mapping the whole QTLs

network determining the genetic variation of a single phenotypic trait. In order to

develop our methodology for QTLs dissection, we have chosen the ability of the

budding yeast Saccharomyces cerevisiae to grow at high pH as a model trait for QTLs

mapping. This method presents a major advancement in the field of QTL research, as

we have succeeded in identifying, without a-priori knowledge, several gene sets that

contribute to the survival at alkali stress. As a first step we have estimated the

heritability and the number of QTLs controlling this trait. Then, we used two

independent methods for identifying the QTLs; In-Lab Evolution and Congenic lines.

During the in-lab evolution (ILE) process a laboratory strain, usually unable to

grow at high pH, was grown under gradually increasing alkali stress to acquire

beneficial mutations improving its fitness under these conditions. Following this

process, we isolated several clones whose fitness under alkali stress had improved

considerably. Genetic analysis of these clones showed that diverse mutations occurred

in each clone, and that both dominant and recessive mutations, differing between

selected lines, contribute to the studied phenotype. We then used tiling arrays to

identify the beneficial mutations. In total, we identified SNPs in 15 genes. We

confirmed each QTL by direct sequencing, and tested the effect of each mutation

using Allele Swapping, Reciprocal Hemizygosity and deletion analyses and used our

results to obtain a quantitative estimate of the contribution of each QTL. The sum of

the individual effects measured was greater than the effect observed in the selected

High MP strain. These results show that a simple additive model alone is insufficient

to explain the phenotype, as in this case, the whole was less than the sum of its parts.

Thus, our results are consistent with epistasis models, in which genetic interactions

between contributing genes affect the final phenotype.

In addition to the ILE, we used a clinically isolated strain able to grow at high

pH and a standard laboratory strain with a limited ability to sustain high pH as the

parents of series of backcrosses to construct congenic lines up to the 8th generation.

We identified 17 genomic intervals that are candidates to contain QTLs. In order to

detect the contributing QTL in each interval we used bio-informatic tools. We applied

a predictive algorithm and scored the candidate genes in each genomic interval based

on their interactions and similarity to the ILE genes. The algorithm was validated by

testing the effect of the predicted candidate gene's deletions on the phenotype. 12 out

of 29 deletions were found to affect the trait (p-value 0.023).

Interestingly, our results show that almost all beneficial mutations affected

regulatory genes, and not structural components of the pH homeostasis machinery

(such as proton pumps). The genes identified affect global regulators, such as

ubiquitin ligases, proteins involved in GPI anchoring and copper sensing and transport

factors. Thus, we show that adaptive changes tend to occur in genes with wide

influence, rather than in genes narrowly affecting the phenotype selected for.

One example for results obtained by ILE and also by congenic lines is the

copper-sensing transcription factor MAC1, and its downstream targets CTR1 and

CTR3, which encode copper transporters. Mutations at the same residue (Cys 271)

were found in four out of five independent ILE lines. This mutation inactivates a

copper-sensing region of Mac1 and causes up-regulation of its target genes. MAC1

and also its targets CTR1 and CTR3 were identified in the congenic lines. Moreover,

we found that a Ty transposable element is responsible for the decreased expression

of CTR3 in some strains, and its excision caused transcriptional activation, affecting

the ability to thrive at high pH

Our results also give some insight regarding evolutionary issues. Previous

studies have discussed the issue of accumulation of mutations and the development of

genetic networks that affect complex traits. Our results suggest that at each stage of

the selection, more than one mutation appears at the same time in the population.

When mutations are abundant enough, such that the population contains multiple

clones, each with new mutations, additional mutations will also arise in some of these

clones while they are still competing. This can lead, as described in our work, to the

development of several genetic networks in parallel.

In summary, this work provides insights on both evolutionary and genetic

issues (such as the appearance of adaptive mutations and the architecture of complex

traits), while at the same time providing information about the mechanisms that

contribute to growth at high pH, a subject with ramifications for cell physiology,

pathogenicity, and stress response. With the availability of new methodologies to map

and identify QTLs, such as the one presented here, the traditional distinction between

Mendelian and quantitative traits will probably blur, and what is considered today a

trait with a Mendelian inheritance pattern will be probably regarded as a complex trait

in which one of the contributing QTLs has a strong phenotypic effect. Thus mysterious

“modifier genes” and “genetic background” are soon to be regarded as part of the

genetic network that determines the observed phenotype.

1

Introduction

1. Mendelian Traits and Complex Traits

The experiments carried out by Mendel at the nineteenth century set the ground for

the very basic principles of genetics and genetic mapping. Yet, already at the

beginning of the twentieth century Mendel’s theory was attacked on the ground that

the simple rules he set did not apply to the variation typically seen in nature (East

1909; Castle 1914; East 1916). The majority of phenotypes in nature do not follow the

rules of simple Mendelian monogenic inheritance. Most traits are characterized by a

complex pattern of heredity and their phenotypes are determined by multiple genes.

These traits are termed Complex traits. Actually, the first indication that variation in

nature is controlled by a complex genetic was already suggested by Charles Darwin

himself. Although Darwin had no clear understanding of the nature of inheritance or

of complex traits as we understand them today, he argued in his book that “Natural

selection can act only by taking advantage of slight successive variations; she can

never take a leap, but must advance by the shortest and slowest steps”. [Darwin, C.

R. The Origin of Species (J. Murray, London, 1859)]. During adaptation mutations

accumulate to increase the fitness of organisms. Therefore, fitness, just like many

other traits, is affected by many genes and not by one gene. The Darwinian view was

later adapted and developed by the Biometric School led by Karl Pearson and Walter

Weldon (Pearson 1903). At 1930 Fisher successfully fused the model supported by

the Mendelian school with the Biometric school and suggested a working hypothesis

for analyzing complex traits. Fisher suggested that instead of following the effects of

individual genes on a selected phenotype, one could analyze their aggregate effects

2

under the assumption that a character is under-laid by an infinite number of genes,

each unlinked to all others, each having no epistatic interactions with the others, and

each having a small effect on the character (Fisher 1919; Fisher 1930). In fact, his

model describes the basic additive model of complex traits in which there are no

interactions between genes. During the last two decades a major effort has been made

to study and better understand complex traits. Complex traits are shaped by multiple,

often intricately interacting genetic and environmental factors (Lander and Schork

1994; Darvasi and Pisante-Shalom 2002; Abiola, Angel et al. 2003). Therefore, their

phenotypes are quantitative, and according to the central limit theorem, occasionally,

exhibit a characteristic normal distribution in nature (Lander and Botstein 1989). Loci

affecting such traits are generally termed Quantitative Trait Loci (QTLs). Quantitative

traits include economically and medically important determinants, including

agricultural crop yields and propensity to diseases such as cancer, schizophrenia or

myocardial arrest in humans (Lander and Schork 1994; Eshed and Zamir 1996;

Fijneman, de Vries et al. 1996; Flint and Mott 2001; Glazier, Nadeau et al. 2002;

Shifman, Bronstein et al. 2002; Hirschhorn and Daly 2005; Holland 2007). Moreover,

variation in the phenotype of individuals carrying the same Mendelian allele is also a

common observation. This natural variability is often explained by the presence of

“genetic modifiers” or variations in the “genetic background”, that is, by the effects

that particular combinations of alleles have on the expression of the gene in question

(Lander and Botstein 1989; Lander and Schork 1994; Lander and Kruglyak 1995;

Glazier, Nadeau et al. 2002). In fact, since natural diversity observed among

organisms is often continuous and characterized by differences in degree rather than

in kind, complex traits involve every aspect of life.

3

2. Evolutionary Aspects of Quantitative Traits

Adaptation is characterized by the movement of a population towards a phenotype that

best fits the present environment (Fisher 1930). Adaptation increases the genotypic and

phenotypic diversity of organisms and determines their fitness by affecting their ability

to survive and multiply in a variety of ecologic niches and under different, often

changing, environmental conditions. The fitness of an individual in any condition is

usually determined by complex traits that are influenced by many quantitative trait loci

(QTLs) and their combinations. Many questions regarding the evolutionary forces that

shape genetic variability and the development of quantitative genetic networks have

been addressed (Johnson and Barton 2005; Zeyl 2005; Mitchell-Olds and Schmitt 2006;

Desai, Fisher et al. 2007; Gresham, Desai et al. 2008). Among them are; how do

beneficial mutations arise, and how are they maintained? How many QTLs contribute

complex traits and what is the effect of each QTL?

Basically, during adaptation three types of mutations can randomly occur [reviewed

by (Mitchell-Olds and Schmitt 2006)]:

1. Mutations with neutral effect which do not affect fitness; their frequency in a

population will be mainly determined by genetic drift.

2. Deleterious mutations that decrease the fitness and therefore natural selection will

act to eliminate them.

3. Beneficial mutations which, due to their beneficial effect on fitness, will tend to be

established in the population.

However, the real situation is much more complicated. For example, some mutations

are beneficial under specific conditions and deleterious under different conditions

(Pepin, Samuel et al. 2006). Some deleterious mutations or neutral mutations can

become established within the population by hitchhiking a beneficial mutation

4

provided that the combined contribution of both mutations to fitness is higher than

that of the wild type alleles. Therefore, the accumulation of beneficial mutations

during adaptation is very complicated and not well understood. Several models have

been suggested regarding the accumulation of beneficial mutations to assemble a

complex phenotype. All models agree that a precise adaptation is possible only if

organisms can fit their environments by carrying out several adjustments. The first

view, named 'micromutational' view is based on Fisher's work (Fisher 1930) who

suggested that many mutations with minute additive effect accumulate during

adaptation, as described above. This view emphasized the extreme gradualness of

phenotypic evolution. However, empirical results could not support the

'micromuational' view alone (Tanksley 1993; Orr 2003). Instead, an alternative

approach suggests that evolution often involves genetic changes of relatively large

effect and the total number of mutations seems to be modest [reviewed in (Tanksley

1993)]. The basic rational behind this theory is that mutations with small beneficial

effect can be eliminated accidentally by genetic drift or displaced due to clonal

interference when rare, while mutations with intermediate or large effects would

become easily established within the population. More accurate models regarding the

type of mutations that occur during adaptation present views intermediate between the

two extreme approaches. It is possible that during adaptation different combinations

of mutations with small, large and intermediate effect may appear.

In order to study and test the theoretical views, an experimental approach should be

taken. In principle, if we could find all the QTLs affecting one phenotype and

measure the effect of each QTL, we could address these theoretical views with

experimental data. However, as I will describe next, QTLs identification is limited

and some of the methods are biased to the identification of QTLs with large effect.

5

3. The Challenges in Genetic Dissection of Complex Traits

Dissecting genetic components of complex traits and studying the intricate

interactions among them are some of the major challenges facing modern biology.

Traditional approaches, which work so well in the context of Mendelian traits, met

only limited success when applied to polygenic traits. The basic strategy of traditional

mapping approaches is to detect single genes based on their effect on the expression

of the phenotype. This strategy is very powerful for the identification of major genes

with a substantial effect. It has a much lower likelihood of success when the

contribution of each gene is small or when the variability in the phenotype stems

mainly from the interaction between genes. The challenge in QTLs mapping resides

in the fact that each QTL only marginally contributes to the studied phenotype and

explains only a small fraction of the phenotypic variation. In addition, each complex

phenotype can result from different allelic combinations, thus the phenotypic-

genotypic correlation is low (Darvasi and Pisante-Shalom 2002; Abiola, Angel et al.

2003; Page, George et al. 2003; Lloret, Dragileva et al. 2006). To make matters

worse, complex phenotypes are determined by both genetic and environmental

factors, and by subtle interactions among these. This has hampered progress in the

mapping and identification of QTLs in any organism. Therefore, so far only very few

affecting genes have been identified (Figure 1) (Glazier, Nadeau et al. 2002; Shifman,

Bronstein et al. 2002; Mitchell-Olds and Schmitt 2006; Weedon, Lango et al. 2008).

Moreover, usually the QTLs identified by large mapping efforts account for only a

very small fraction of the variability seen in nature (Maher 2008). Mendelian traits, on

the other hand, are controlled by a single locus, and their inheritance can be clearly

traced from generation to generation due to a strong correlation between genotype and

phenotype. Therefore, many Mendelian genes have been identified while dissecting

6

the genetic component of the phenotypic diversity accrued by polygenic traits is still a

major challenge (Lander and Schork 1994; Lander and Kruglyak 1995; Glazier,

Nadeau et al. 2002; Linney, Murray et al. 2003).

Figure 1 [adapted from (Glazier, Nadeau et al. 2002): Identification of genes

underlying human Mendelian traits and genetic complex traits in humans and other

species. Cumulative data for human Mendelian trait genes (to 2001) include all major

genes causing a Mendelian disorder in which causal variants have been identified

(Glazier, Nadeau et al. 2002). Until 2001 the molecular basis of ~1700 human

Mendelian traits was found, while only ~9 QTLs affecting human complex traits

were found.

7

4. Strategies for Genetic Dissection of Complex Traits

During the last two decades several methods for QTLs mapping have been developed

taking into consideration the obstacles described above. Generally, genetic dissection

of complex traits can be divided into three stages: (1) Correlating genetic markers

with the studied phenotype. The markers define chromosomal interval which are

candidate to contain the QTLs. (2) Narrowing down the chromosomal interval to a

few genes. Usually it is impossible to narrow down each interval to one gene; several

methods are combined to narrow the intervals as much as possible. (3) Proof of

causation. At this stage the QTLs found need to be tested to demonstrate their effect

on the phenotype. Each stage is usually divided into several steps which can be

carried out by different strategies.

Here the main methods are described with some examples revealing their advantages

and disadvantages.

4.1 QTL Detection by Marker-Trait Association in a Whole Genome Scan

The basic rational behind marker-trait association studies is that if a factor contributes

an increase risk for phenotype occurrence, then that factor should be found at higher

frequency in individuals showing that phenotype compared to individuals showing the

alternative phenotype. Similarly, if a genetic marker is linked to a gene that

contributes to a certain phenotype, then that marker should also be found at higher

frequency in individuals showing that phenotype. If an association between a given

phenotype and markers on a genomic interval is found, then this genomic interval is

candidate to contain the gene affecting that phenotype (Greenberg 1993). Thus, the

aim is to find those chromosomal intervals that tend to be shared among

phenotypically similar individuals and differ from dissimilar individuals. This is done

by scanning the entire genome of many individuals with a dense collection of genetic

8

markers. The frequency of each marker is calculated in each phenotypic group.

Genomic intervals containing markers with significant deviation from what would be

expected under the assumption of independent assortment are defined as associated

with the phenotype (Shifman, Bronstein et al. 2002; Hirschhorn and Daly 2005). The

ability to identify the causative genes is largely dependent on the size of the intervals.

The aim is then to narrow down the chromosomal interval so that it will contain only

a few genes (and ideally a single one).

The size of the intervals thus defined is affected by the density of the markers, by the

sample size and by the frequency of recombination (Darvasi, Weinreb et al. 1993;

Darvasi 1998; Fan and Jung 2002). Recombination breaks apart the correlation

between adjacent loci in the genome and affects the association between the genetic

markers, the causative gene and the phenotype. Therefore, high recombination

frequency will decrease the interval size by expelling non-relevant markers. The

frequency of meiotic recombination is inherent to each organism. For example, in

yeast, the extremely high recombination frequency allows very high resolution

mapping with relatively few meioses. The ratio between physical and genetic map is

3kbp/cM compared to the mouse, where 1cM corresponds to 1.6Mkb (Glazier,

Nadeau et al. 2002). Thus, a dense set of polymorphic markers will have to be typed

in a large number of individuals in order to narrow down the intervals. Methods for

high resolution genotyping which were recently developed allow testing high density

genetic markers and can be applied for QTLs identification.

4.1.1. High Resolution Genotyping Methods

Genetic markers such as microsatellites and single nucleotide polymorphisms (SNPs)

can be identified by PCR or direct sequencing. However, with these classic methods

labor increases in proportion to the density of the markers typed. In recent years some

9

new whole genome technologies were developed that enable high resolution genotyping

(Gresham, Desai et al. 2008). Similar to classic methods, the new technologies detect

genetic polymorphisms (generally, these polymorphisms are termed 'Single Feature

Polymorphisms' or SFPs). Conveniently, the new methods do not require prior map

knowledge and they can uncover thousands of SFPs within the whole genome.

Genomic Mismatch Scanning (GMS)

GMS is a high-resolution mapping technique (Nelson, McCusker et al. 1993; Brown

1994). This method is designed to recognize differences (SFPs) between two genomes.

In GMS heteroduplexes are formed from DNA fragments of two individuals. The

mismatch-containing fragments are depleted by Escherichia coli mismatch repair

proteins and an exonuclease, while identical fragments are enriched by PCR. This is

followed by hybridization to a spotted microarray. Since fragments containing

mismatches were depleted, these fragments will be under-represented on the arrays.

This leads to the identification of genomic regions that differ between the tested

individuals.

High resolution oligonucleotide arrays

Oligonucleotide arrays designed for expression analysis can be used to detect and score

allelic variation in yeast via direct hybridization of labeled genomic DNA. These

expression arrays contain a total of 157,112 25-mer probes derived from the yeast

genome coding sequences and they cover 21.8% of the nonrepetitive regions of the

yeast genome (Winzeler, Richards et al. 1998; Winzeler, Castillo-Davis et al. 2003).

The hybridization intensity of a target sequence to an oligonucleotide probe depends on

the homology between the two sequences. Therefore, allelic variation between any two

genomes can be detected by hybridizing genomic DNA from the two strains to the

arrays and analyzing the hybridization differences (Figure. 2). Since the frequency of

10

variation between common laboratory strains of yeast can be as high as 1% (Nelson,

McCusker et al. 1993), a new genetic map containing a large number of closely spaced

markers could be constructed using those arrays. Several studies have applied

genotyping by oligonucleotide arrays for the identification of candidate regions to

contain QTLs in yeast (Deutschbauer and Davis 2005; Ben-Ari, Zenvirth et al. 2006).

The development of DNA oligonucleotide arrays made a major breakthrough in marker-

trait association studies. The use of microarray for genotyping enables the use of

thousands of markers using a single hybridization.

Figure 2 [adapted from (Winzeler, Richards et al. 1998)]: Detecting allelic variation

with high density arrays. Each 25-mer probes derived from yeast genome coding

sequences. In addition to probes designed to be perfectly complementary to regions of

yeast coding sequence (PM), probes containing a single base mismatch in the central

position of the oligonucleotide were also synthesized. The mismatch probes serve as

background and nonspecific hybridization controls. Several SFPs (*) are found in

strain 2. Each SFP decreases the hybridization to the probe and the signal intensity

relative to strain 1 that carry no SFPs. The reduction in signal will depend on several

factors, such as initial probe intensity and whether the probed fragment is completely

absent in strain 1 or contains a small substitution. The location of the polymorphism

within the probe sequence will also affect the observed decrease in intensity.

11

Tiling arrays

Similarly to the oligonucleotide arrays described above, high-density Affymetrix

yeast tiling microarrays (YTMs) can be used to detect and score allelic variation in

yeast via

direct hybridization of labeled genomic DNA. Tiling arrays provide

complete coverage of the genome by tiling short probes immediately adjacent to one

another (Figure 3).

Figure 3 [adapted from (Gresham, Desai et al. 2008)]: Multiple overlapping

probes cover each nucleotide. Hybridization of the sample to tiling arrays can be

used to identify SFPs. Mismatches resulting from SFPs in the sample DNA will

result in a lower hybridization efficiency to each probe compared with hybridization

to a sample with complete sequence complementarity. A. A SFP at the site

indicated in red perturbs hybridization of the sample to all probes. The colors (from

yellow to blue) represent the probe intensity. B. The effect of a SNP on

hybridization is related to its corresponding position in a probe. More central

positions result in the greatest decrease, whereas SNPs positioned at the end of

probes are much less likely to result in a significant decrease in hybridization.

12

Each tiling array has ~2.6 million perfect match probes and ~2.6 million

corresponding mismatch probes. 25- nucleotide oligomers spaced 5 base pairs (bp)

apart to provide complete and 5-fold redundant coverage of the entire S. cerevisiae

genome. This design provides five to seven measurements of a given nucleotide's

effect on hybridization efficiency, which we exploited to predict the presence and

location of SFPs throughout the entire yeast genome (Gresham, Ruderfer et al. 2006).

4.2. Advanced Methods for Improving the Mapping Resolution

High resolution genotyping methods such as oligonucleotide arrays exploit the use of

thousands of markers. However the resolution of QTL mapping by association studies

is limited also by the recombination frequency and the sample size. Some models

have been designed to improve mapping resolution. Several examples of these

models and some studies in which they were developed and used are presented here.

4.2.1. Selective Genotyping

Selective genotyping (Lander and Botstein 1989) is a method in which only

individuals from the extreme phenotypic groups of the population are genotyped. For

complex traits, a cross between two parents with different phenotype is expected to

yield a progeny population with a variety of phenotypes. This population will usually

exhibit a characteristic normal distribution of the phenotype. It has been shown that

genotyping only individuals at the phenotypic extremes will not reduce the detection

power (e.g., by loosing QTLs or increasing the interval containing the QTL) (Darvasi

and Soller 1994; Darvasi 1997).

A pioneering work for finding genes underlying a complex phenotype using a marker-

trait association approach with selective genotyping by hybridization to DNA

oligonucleotide array was published in 2002 (Steinmetz, Sinha et al. 2002). Three

linked genes that affect high temperature growth (Htg) were identified in that study.

13

Steinmetz et al. choose a strategy of a cross between two S. cerevisiae strains from

different genetic background and selective genotyping of the offspring. The parents

were selected according to their high temperature growth (Htg) phenotype; one parent

was able to grow at high temperature (Htg+) and the other had a poor such ability

(Htg-). Each strain was hybridized separately to high density oligonucleotide arrays to

create a genetic map based on SFPs. The Htg phenotype of the progeny was scored

and the progeny from the extreme phenotypic groups were genotyped using

hybridization to oligonucleotide array. Htg+ segregants and the Htg+ parent shared a

genomic interval of 32kbp. Despite the use of selective genotyping and the high

density oligonucleotide arrays, a further step of direct sequencing was needed in order

to map the genes in the candidate interval.

4.2.2. Estimating SFPs alleles frequencies in DNA Pools

Even while using selective genotyping, the number of individuals that needs to be

genotyped is large. To increase the sample size and significantly decrease the interval

and make the fine mapping stage less complicated, a method of selective genotyping

of DNA pools can be efficient (Darvasi and Soller 1994). In this procedure two DNA

pools are formed. One pool includes DNA from individuals with high phenotypic

value for the trait of interest and the other includes DNA from individuals with low

phenotypic value. Instead of genotyping one individual at a time, the whole DNA

pool is genotyped in one hybridization or sequencing. Selective genotyping of DNA

pools to identify QTLs affecting sporulation efficiency in yeast was first performed by

Ben Ari et. al (Ben-Ari, Zenvirth et al. 2006). After a cross between two strains that

differ in their sporulation efficiency the progeny were tested for their sporulation

efficiency phenotype. DNA was extracted from 21 segregants with high sporulation

efficiency and 21 with low sporulation efficiency. DNA from each extreme group was

14

assembled into a DNA pool. DNA pools were hybridized to oligonucleotide arrays

and probe intensities were compared to evaluate the frequency of SFP alleles in each

pool. Three intervals had differences in the allele's frequency between the two pools

indicating the presence of QTLs. Even using DNA pools the interval around each

QTL was too large and a further step of fine mapping was inevitable. As in many

other studies, the difficulties involved in QTL fine mapping forced the researchers to

focus only on one or few intervals containing genes for which prior knowledge exist.

Correlations were thus tested for a few selected candidate genes and the observed

phenotypic variation. In the work of Ben Ari et al., for example, a candidate interval

containing about 100 genes was chosen. Based on previous knowledge, twelve genes

were chosen for further analyses. Yet, the main aspiration of discovering genes which

contribute to a complex phenotype without any biological clue as to how they

function remains inaccessible even by estimating SFPs alleles frequencies in DNA

pools. Only by significantly decreasing the intervals found by marker association

studies, QTLs can be dissect without previous knowledge.

4.2.3. Congenic Lines

A complementary strategy that can decrease the interval and allow for fine mapping is

the construction of Congenic lines. A Congenic strain is generated by selective

mating, in order to place the QTL interval from one strain (the donor) within the

genetic background of another strain (the background). Starting with two parents that

differ in the phenotype of an investigated trait, a cross between the parents is carried

out. At each generation, a segregant with the phenotype of the donor parent is selected

for, and backcrossed to the background parent. After a series of several such

backcrosses, the congenic strains are very similar to the background parent, but carry

intervals originating in the donor strain. Since the congenic strain retains the donor’s

15

phenotype, the QTL must be located within those intervals. Because of recombination,

after every backcross, from one generation to another the interval inherited from the

donor parent is reduced. The use of Yeast Congenic Lines (YCL) has many

advantages. Due to their small size, their simplicity and their short generation time it is

easy to screen hundreds of yeast cells for the desired phenotype and to backcross them

to the background parent in a few days. The fact that it is possible to use the exact

same strain as the background parent (e.g., it does not get old like model animals)

prevents disturbing noise. Taking into consideration that recombination frequency in

the yeast genome is the highest of any other model organism, the use of YCL is highly

efficient. Deutschbauer et al. in 2005 used a backcross strategy to identify QTLs

associated with the differences in sporulation efficiency between two strains. They

backcrossed a high-sporulating F1 segregant to a low-efficiency parent. A high-

sporulation backcrossed segregant was genotyped by hybridization to oligonucleotide

arrays as described above in order to find the intervals inherited from the high-

sporulation parent (the donor), which were candidates to contain the QTLs. After one

backcross the intervals were still too large to map the genes affecting the trait within

them. Therefore, another backcross was required. After the second backcrosses, three

candidate intervals were chosen for further analysis and sequenced. Almost all genes

within each interval had non-synonymous SFPs so the sequence alone could not be

used for fine mapping. Only 4th

generation congenic lines enabled more accuracy

mapping. Yet, the identified intervals of the congenic lines are too large and contain

many genes. Therefore, further steps of fine mapping are required.

4.3. High Resolution Fine Mapping and Proof of Causation

As demonstrated here, selective genotyping, DNA pools and congenic lines increase

the mapping resolution but yet have some limitations, and the intervals identified by

16

them alone cannot usually pinpoint the exact contributing alleles. Moreover, in

association studies such as those presented here, some hindrances can mask the

contributing genes making them almost impossible to be found. For example, closely

linked QTL with effects in the same direction cannot be separated and will be

considered as one QTL. In contrast, closely linked QTL with effects in the opposite

direction will mask each other and usually, and none of them will be found. In both

cases, recombination, the natural mechanism that breaks apart the correlation between

adjacent loci in the genome, is not sufficient, in most instances, to deliver the necessary

accuracy for QTL localization. Therefore, the high resolution fine mapping stage should

be carried out by other methods. Currently, only genetic engineering methods can help

to overcome these obstacles: After identifying a candidate interval, the identification of

the causative gene is carried out by deleting or knocking down individual genes (either

systematically deleting all genes in an interval, or only candidate genes). The genetic

amenability and the sophisticated molecular genetics techniques available for yeast

make it an ideal model organism for this type of experiments.

Even using the methods described above, QTLs analysis certainly has many

limitations. The number of QTLs estimated is usually lower than the real number due

to linkage between several QTLs. There is also likely a bias in the estimation of QTLs

effect distribution as only QTLs with relatively large effects are likely to be detected.

Analysis of epistasis may only detect a fraction of the gene interactions and there

could be many other hidden interactions between detected and undetected QTLs.

Thus, QTL fine mapping is very rare. In most cases the QTLs are mapped into a large

interval and identifying the genes that are responsible for the variation is difficult.

17

5. Yeast as a Model Organism

In order to develop our approach for genetic dissection of complex traits, we use the

budding yeast Saccharomyces cerevisiae. Yeast has played a central role in genetics and

is emerging as an excellent model organism for studying quantitative traits. The yeast

system offers many unique advantages for studying quantitative genetics and for

experimental evolution. Yeast cells grow extremely fast, enabling propagating

populations of a wide range of sizes while controlling their environmental variables.

Being a single-celled organism, it is possible to use clones derived from a single cell, thus

ensuring that the whole population shares exactly the same genome. Microbial evolution

experiments have exploited the ease with which very large populations can be maintained

(allowing faster adaptation), and the fact that samples of evolving populations can be

frozen for later analysis and compared with their ancestor and with each other. Yeast cells

can grow as diploids or haploids, and undergo meiosis when transferred to a particular

starving regime. Thus it is easy to carry out crosses and to genotype meiotic products

(gametes) rather than diploids obtained after the gametes have fused.

During the last 50 years, a vast body of knowledge on the genetic and biochemical

properties of these cells has accumulated. The yeast genome sequence, the first complete

DNA sequence of a eukaryotic genome, is known since 1996 (Goffeau, Barrell et al.

1996). In addition, isogenic strains are available, which differ only in a single trait.

Moreover, a vast number of natural isolates exist, and many of them have been sequenced

(McCusker, Clemons et al. 1994; Winzeler, Richards et al. 1998; Schacherer, Ruderfer et

al. 2007; Liti, Carter et al. 2009; Schacherer, Shapiro et al. 2009).

Due to its advantages and its genetic amenability many powerful sophisticated genetic

tools have been developed for this organism. Those tools can be exploited for

quantitative genetic researches.

18

5.1. The Yeast Deletion Library

One of the most powerful yeast genetic tools currently available is a collection of

4850 deletion mutants, each carrying a single gene knocked out (Giaever, Chu et al.

2002). This collection is routinely used for genome-wide systematic screenings of

many types. For example, the phenotype of all mutants can be characterized and

scored under different conditions, such as alkali stress (Serrano, Bernal et al. 2004) or

in the presence of Ethanol (Fujita, Matsuyama et al. 2006). A phenotypic change

indicates the contribution of the deleted gene to the trait. However, these screens

usually do not give quantitative data; the mutants either grow or fail to do so.

Moreover, most phenotypes detected in this way behave as Mendelian traits. Only

few studies combine results of such screening into quantitative results and networks

analyses (Gatbonton, Imbesi et al. 2006; St Onge, Mani et al. 2007; Hillenmeyer,

Fung et al. 2008; Shachar, Ungar et al. 2008).

The major advantage of the systematic screenings of the library is that they do not

require any crosses and once a difference in the phenotype is recognized, the gene

responsible is already known. This method is therefore an easy and fast way of

identifying all genes affecting a certain phenotype. However it is important to be

aware that some qualities of the library itself and the nature of complex trait decrease

the power of this method for identifying QTLs:

• QTLs with small effect cannot be detected.

• Essential genes cannot be identified, since the library contains only viable

deletion mutants.

• Since the library contains mutants with only a single deletion at a time, it is

impossible to analyze genetic interactions, and QTLs that have an effect only

in the presence of additional mutation(s) will not be detected.

19

• It is known that the effect of the majority of QTLs is dependent on the genetic

background. Using this library eliminates the possibility of monitoring QTLs

in different genetic backgrounds.

These drawbacks make simple systematic screening of deletion library a limited tool

for QTLs analyses. However, the use of the deletion library can be combined with

other strategies for QTLs mapping or to serve as an initial step in 'proof of causation'

for candidate QTLs which were already identified. Furthermore, results obtained from

screening the deletion library can identify candidate genes that can potentially be

affecting QTLs. In this case, a further step of testing their direct effect on the

phenotype should be taken.

6. Experimental Evolution with Yeast

Microbial experimental evolution, such as In Lab Evolution (ILE) can shed light on

several questions regarding adaptation and natural variation. In ILE works, yeast are

introduced into a new environment allowing adaptation; genetic and molecular tools

are then used to identify the genetic changes that underlie this adaptation. During the

adaptation samples of each population are periodically frozen for analysis. This

method, as shown in this work, allows not only to address evolutionary issues, but

also to identify QTLs affecting the studied phenotype. Two central methods are cited

in the literature: 1. Serial transfer, in which a sample of each population is transferred

to a tube or plate of fresh medium at regular intervals (Figure 4), and 2. Chemostats,

which balance inflow of fresh medium with outflow of used medium at constant rates

(Paquin and Adams 1983). Each method has its advantages and disadvantages, for

example, in chemostats, population sizes and environments are held constant, but

serial transfer allows for greater replication within an experimental treatment.

20

However, both methods allow selective pressure in which rare mutations that enhance

fitness will appear and establish. Applying this method to QTLs dissection gives rise

to many benefits. The in-lab evolution method enables tracking the mutations that

occur during adaptation and then, to measure their effect on the fitness. Since the

identification of the mutations is independent of the analysis of their effect, mutations

are identified regardless of their strength or epistastic interactions. Therefore, this

method permits to overcome some of the major problems in dissecting complex traits.

Figure 4: Serial transfers. The experiment starts with a yeast culture grown at

basic optimal conditions. An aliquote of cells is transferred to mild selective

conditions. Each time the culture reaches maximal density, a sample is transfer to

fresh medium with either the same or to more severe stress conditions. From one

transfer to another, mutations accumulate and beneficial mutations become

established within the population of the adapted strains.

21

7. The Ability of Yeast Cells to Grow at High pH as a Complex

Trait

We have chosen to dissect the genetic network responsible for the ability of the

budding yeast Saccharomyces cerevisiae to grow at high pH as a model for a complex

trait. Extracellular pH represents one of the most important factors influencing cell

physiology and growth. Environmental pH serves as a potent inducer of

differentiation and development. In the opportunistic fungal pathogen Candida

albicans, for example, high pH induces hyphal growth and pathogenesis (Davis

2003). Moreover, appropriate responses to high pH govern fungal virulence in plants,

insects and animals (Penalva, Tilburn et al. 2008).

The yeast Saccharomyces cerevisiae grows better at acidic pH than in neutral or

alkaline media, and maintenance of an acidic environment is mainly based

on the

activity of the plasma membrane H+-ATPase, which actively

extrudes protons.

Maintenance of a proton gradient is crucial for the uptake of diverse nutrients and

cations (van der Rest, Kamminga et al. 1995; Serrano 1996). Consequently, even a

transient exposure to mild alkaline pH represents a stressful situation to which the

yeast must adapt to survive and proliferate (Penalva and Arst 2002).

Environmental pH has dramatic effects on the physiology of the cell, and affects

nutrient availability and protein activity. Adaptation to a wide environmental pH

range requires not only an internal pH homeostatic system but also a means of

ensuring that molecules directly exposed to the environment such as permeases,

secreted enzymes and exported metabolites are synthesized only at pH values in

which they can function. Adaptation of Saccharomyces cerevisiae to alkaline pH

involves a change in its expression profile. DNA microarray analysis identified

hundreds of genes that are induced/repressed at least 2-fold when cells are transferred

22

from acidic medium to one of pH 7.6–8.0 (Causton, Ren et al. 2001; Lamb, Xu et al.

2001). Some of these genes are, as expected, part of the environmental stress response

(ESR) genes (Gasch, Spellman et al. 2000; Causton, Ren et al. 2001), which responds

to any type of stress, and some of them are unique to alkali stress, such as the Rim101

pathway (Lamb, Xu et al. 2001). Screening of a the yeast Systematic Deletion Mutant

Library revealed 118 genes, whose absence results in reduced growth at alkaline

conditions (Serrano, Bernal et al. 2004). Interestingly, the genes found are involved in

numerous cellular functions. For example, that list includes BCK1 and SLT2/MPK1,

which are components of the MAPK2 cascade involved in cell wall remodeling and

maintenance of cell integrity (Heinisch, Lorberg et al. 1999; Martin-Yken,

Dagkessamanskaia et al. 2003; Serrano, Martin et al. 2006). In addition, the

systematic deletion mutant screen also revealed genes involved in transport of copper

and iron (Serrano, Bernal et al. 2004). This is consistent with the transcriptional

profile observed after alkaline stress, which suggests that alkalinization of the medium

affects the availability of different nutrients, such as phosphate, and that of specific

cations,(Lamb, Xu et al. 2001).

The involvement of alkali stress in various pathways and its effect on several

mechanisms supported our assumption that the ability to grow at high pH is a

complex trait that is influenced by a large genetic network.

23

8. Working Hypothesis

All the information and methods described above demonstrate the importance of

understanding the architecture of quantitative variation. This field has become the

new frontier of genetic research, and several studies that use different whole-genome

strategies have greatly contributed to our understanding of this subject. Some of these

approaches have provided technical break-throughs and others have resulted in the

“re-discovery” of known biological mechanisms. That background has led me to

develop, during my PhD, an integrated approach for dissecting complex traits. This

approach combines in-lab evolution (ILE) with an association study, utilizing several

genotyping procedures. Using ILE process it is possible to allow a laboratory strain,

usually unable to grow at high pH, to acquire beneficial mutations, improving its

fitness under these conditions. The ILE strategy was combined with an association

study carried out on a wild type clinically isolated strain revealing data on natural

genetic variability. Today, with the advanced techniques for genotyping we can use

high throughput methods such as GMS, high density olignucleotides array and tiling

arrays to identify the QTLs affecting the phenotype. This integrated approach enables

mapping of QTLs even with relatively small effect. It can also accomplish accurate

fine mapping at a high resolution. Once the fine mapping is achieved, we can estimate

the effect of each QTL on the studied phenotype. These data will not only enable

complex traits dissection, but will also shed light on some evolutionary questions

regarding adaptation and the assembly of complex phenotypes during the evolution.

While combining these approaches, complex traits and evolutionary issues can be

studied at a greater scale and depth than is possible using either technique alone.

24

Materials and Methods

PCR - Polymerase Chain Reaction,

RT – Reverse Transcript

RH – Reciprocal Hemizygotes

AS – Allele Swapping

1. Growth Media

YEAST:

YPD (yeast rich medium) - 1% Bacto yeast extract, 2% Bacto peptone, 2% Glucose.

Yeast cells were grown in YPD at pH 6 at 30oC (standard growth conditions). The pH

was adjusted and kept constant using 100 mM Tris buffer.

SD (yeast defined medium) - 0.67% Bacto yeast nitrogen base w/o amino acids (DIFCO),

2% Glucose. Amino acids were added according to requirement (sigma).

YPD G418 - YPD containing 200mg/l G418 Geneticin (sigma).

YPD Hygromycin - YPD containing 300mg/l Hygromycin (sigma).

5FOA: 0.67% Bacto yeast nitrogen base without amino acids, 2% glucose, 50mg/l

Uracil, 0.8 gr/l 5FOA.

SPO - 1% Potassium acetate, 0.1% Bacto yeast extract, 0.05% Glucose + 10% of all the

necessary amino acids.

SM - 1% Potassium acetate

BACTERIA:

LB - 1% Bacto Tryptone, 0.5% Bacto yeast extract, 0.17M NaCl.

Ampicilin (Amersham) 50 mg/l was added to LB+Amp plates.

25

2. The Ability to Grow at High pH:

2.1. Determining the MP Phenotype Using a 'Drop Assay'

Ten fold serial dilutions were spotted on YPD plates at various pH levels. The MP

phenotype was defined as the solid media of highest pH at which each strain could grow

after five days at 30oC. To avoid inaccuracy resulting from differences between plates we

used the same plate batch for every experiment and we included a high MP strain and a

low MP strain with a known phenotype as controls on every plate.

2.2. Fitness Measurements

Strains were inoculated into 2 ml of YPD at pH6 and incubated until exponential growth

(~107 cells per ml) was reached. Optical density at 600 nm (OD600) of cultures was

measured and all strains were diluted to OD 0.1 in 2 ml YPD at either pH 6 or high pH (8

or 7.9). Cultures were grown to OD 1.5. The growth rate of each culture was monitored

by measuring the OD600 every 30 min. and the doubling time was calculated by fitting

an exponential curve to the measurements. The Relative fitness of a strain was defined as

the ratio of the doubling times at pH8 and pH6.

2.3. Heritability Estimation

Heritability was estimated using the formula:

Where Var(seg) and Var(p) stand for the variance in MP phenotype of the segregants and

the parents, respectively (Brem, Yvert et al. 2002).

2.4. Estimating the Number of QTLs Affecting the Trait

Where Pex stand for proportion of the extreme phenotype and Q stand for the number of

QTLs

26

3. S. cerevisiae Strains and Yeast Genetics

3.1. Yeast Strains:

BY4741a: MATa ura3∆ met15∆ leu2∆ his3∆. This strain is of S288c background and was

used as a standard for the sequencing project and for the deletion collection (Brachmann,

Davies et al. 1998).

BY4741α: MATα ura3∆ met15∆ leu2∆ his3∆. Created by transformation with a URA3-

marked plasmid carrying the HO gene which enabled mating type switch.

BY4741d: Diploid created by mating BY4741a and BY4741alpha.

EM39: Wild type diploid, parent of S288c (Mortimer and Johnston 1986).

W303: MATa leu2-3,112 trp1-1 can1-100 ade2-1 ura3-1 his3-11,15. (Thomas and

Rothstein 1989).

TBR strains from Sigma 1278b background, kindly supplied by G. Fink (Reynolds

and Fink 2001):

TBR1: MAT@ ura3-52 his3::hisG leu2::hisG.

TBR2: MATa ura3-52 his3::hisG leu2::hisG.

TBR3:MATa/MAT@ ura3-52/ura3-52 his3::hisG/his3::hisG leu2::hisG/leu2::hisG.

TBR5: MAT@ ura3-52 his3::hisG leu2::hisG flo11del::KanMX.

TBR8: 10560-23c flo10 MAT@ ura3-52 his3::hisG leu2::hisG flo10::KanMX.

TBR12: MATa ura3-52 his3::hisG leu2::hisG flo11del::KanMX.

Wild type clinical isolates from immune-compromised patients, kindly provided by

G. Fink (originally isolated from human patients' blood by RH Rubin, Harvard

Medical School, Boston, USA):

F1411, F1622, F1623 (MKF12), F1624, F1627, F1630, F1632, F1633, F1635, F1637,

F1639, F1642, F1644, F1646, F1649, F1651, F1652, F1653, F1657, F1658, F1659,

F1660, F1661, F1662.

27

All strains in this collection are prototrophs. Most are homothallic diploids, although still

maintain a high degree of heterozygozity (M.K., unpublished).

Clinical Isolates (McCusker, Clemons et al. 1994):

YJM128, YJM222, YJM273, YJM309, YJM310, YJM311

3.2. Construction of the High MP Strain GRA2:

GRA2 is a haploid spore of the diploid wild type strain F1623 (MKF12). Natural isolates

of yeast are usually diploid, prototrophic and homothallic (self-mating), all of which

preclude control crosses. To form a haploid spore from MKF12 we first put MKF12

through meiosis in sporulation medium. Tetrads were separated by micromanipulation.

Since wild type isolates are homothallic, the spores switched their mating type and self-

mated to create diploid homozygote strains. We tested the ability of these strains to grow

at high pH and chose GRA2-dip as the highest MP strain. We then deleted one of the HO

alleles in this strain using targeted disruption with a KanMX4 cassette. Since only one of

the alleles was deleted, half of the spores were heterotallic. We picked for further

analyses MATa and MATalpha haploid G418R segregants, GRA2a and GRA2alpha,

which exhibited a high MP phenotype.

3.3. In-Lab Evolved Strains

Strain BY4741 (MATa ∆ura3, ∆leu2, ∆his3, ∆met15) grows extremely slowly at pH7.4.

Five independent lines were established; line A, line B, line C, line G and line I. When

cultures reached saturation, aliquots (5 ul) were transferred to fresh medium (5 ml) at a

slightly higher pH (0.1 pH unit steps) and were similarly incubated. This procedure was

repeated until populations able to grow at pH 8.5 or 8.6 were obtained. Samples were

taken at each passage, the genotype of the strain tested and the population kept in -80oC.

28

At the end of this procedure cells were streaked on YPD at pH 6, and individual colonies

were tested for their ability to grow at high pH using a drop assay. For each line,

populations were frozen after each selection step, involving increasing pH (starting from

pH 7.4, pH7.5, pH7.6, etc., until they were able to grow at pH 8.6 (lines B, C and G) or

pH 8.5 (lines A and I).

3.4. Mating Between the Ancestor and the Evolved Strains:

In order to carry out a cross between the ancestor and the evolved strains we first had to

switch the mating type of the ancestor. A URA3-marked plasmid carrying the HO

endonuclease was introduced into this strain. This led to a mating type switch. The

plasmid was then lost by growing the transformants in media containing 5-FOA, which

selects against Ura+ cells.

3.5. Mating:

Two haploid colonies were mixed on YPD plate and incubated overnight at 25º. Zygotes

were then separate under micro-manipulator.

3.6. Sporulation:

Diploids were grown to logarithmic phase at YPD liquid media and then transferred to

YPA to 24 hours at 30o. After 24 hours cells were transferred to SPO or SM for 3 days at

25o.

3.7. Deletions:

To test candidate genes in the BY4741 background we use the Saccharomyces Genome

Deletion Project (Giaever, Chu et al. 2002) in which each strain was deleted for a single

ORF (replaced by the KanMX4 cassette, which confers G418 resistance). When we

wanted to create deletions at different background we picked the KanMX4 cassette from

the relevant strain from the deletion library and transformed it into the desired

background. The KanMX4 cassette would replace the targeted ORF. Alternatively, some

29

of the ORFs were deleted by 'Delitto Perfetto' (Storici and Resnick 2003) (see "allele

swapping" below).

3.8. Allele Swapping:

We carried out allele swapping using the Delito Perfetto methodology (Storici and

Resnick 2003). For each candidate QTL, we deleted the region containing the mutation

by insertion of a fragment carrying the URA3 and HygBR genes (which complement

uracil auxotrophy and confers resistance to Hygromycin, respectively) and the gene

encoding the I-SceI nuclease with its restriction site [pCORE plasmid's cassette, a gift

from F. Storici (Storici and Resnick 2003)]. Transformants were selected by resistance to

hygromycin and ability to grow on media without uracil. These transformants were

confirmed by colony PCR. At least two pCORE transformants were selected for each

experiment. We then induced a double strand break at the I-SceI site to enhance

homologous recombination and replaced the Hyg-URA3 cassette with a PCR fragment

carrying the desired changes (single nucleotide replacements). Replacement of the Hyg-

URA3 cassette with the new allele was selected on 5FOA plates. The sequence of the

inserted alleles was confirmed by sequencing of PCR amplicons of the targeted region.

Essential genes (GPI17 and CDC23) and the telomeric gene YFR057w were not tested

using allele swapping due to technical difficulties.

BY4741 Mac1 C271W

BY4741 Mac1 C271S, M386 silent

BY4741 Mac1 C271Y

BY4741 Ecm21 193Nonsense

BY4741 Ctf3 silent

BY4741 Oac1 I48F

BY4741 Rri2 A138P

BY4741 Sgt2 M1I

BY4741 Hrd1 L19F

BY4741 Gtt2/Mmp1 3'UTR XII 20872

30

BY4741 Gph1 3'UTR XVI 864306, C to A

BY4741 YFR057w Base substitution at promoter (-161bp) G to A

BY4741 Nmd4 Base substitution at promoter (-262bp) C to A

BY4741 Yhr140w/SPS100 Base substitution at SPS100 promoter (-384bp) G to T

BY4741 Ies2 Base substitution at promoter (-65bp) T to C

3.9. Reciprocal Hemizygote Strains:

For a particular candidate gene, we first deleted the gene in each parent using a KanMX

cassette or the pCORE plasmid's cassette [URA3, Hyg, I-Sce-I; (Storici and Resnick

2003)]. We carried out reciprocal crosses between the ancestor (BY4741) and the

evolved strain carrying the deletion of the particular candidate allele. For essential genes,

which were impossible to delete, we first carried a cross between the ancestor and the

evolved strain and then deleted one of the alleles. We isolated five independent

transformants for each gene knockout in both parental backgrounds and confirmed these

transformants carried the desired deletion by colony PCR.

BY4741 ∆gtt2/mmp1::core x C8.6

C8.6 ∆gtt2/mmp1::core x BY4741

BY4741 ∆gtt2/mmp1::core x C8.0

C8.0 ∆gtt2/mmp1::core x BY4741

BY4741 ∆ecm21::core x C8.6

C8.6 ∆ecm21::core x BY4741

BY4741 ∆ecm21::kanMX x C8.6

C8.6 ∆ecm21::kanMX x BY4741

BY4741 ∆gtt2/mmp1::core x C8.6

C8.6 ∆gtt2/mmp1::core x BY4741

BY4741 ∆ nmd4/ylrw-a::core x C8.6

C8.6 ∆ nmd4/ylrw-a::core x BY4741

BY4741 ∆ nmd4::kanMX x C8.6

C8.6 ∆ nmd4::kanMX x BY4741

BY4741 ∆gpi17::core x C8.6

C8.6 ∆gpi17::core x BY4741

31

BY4741 ∆yhr140w/sps100::core x C8.6

C8.6 ∆yhr140w/sps100::core x BY4741

BY4741 ∆mac1::core x C8.6

C8.6 ∆mac1::core x BY4741

BY4741 ∆mac1::kanMX x C8.6

C8.6 ∆mac1::kanMX x BY4741

BY4741 ∆mac1::core x A8.5

A8.5 ∆mac1::core x BY4741

BY4741 ∆ctf3::core x A8.5

A8.5 ∆ctf3::core x BY4741

BY4741 ∆cdc23::core x A8.5

A8.5 ∆cdc23::core x BY4741

BY4741 ∆oac1::core x C8.0

C8.0 ∆oac1::core x BY4741

BY4741 ∆rri2::core x C8.0

C8.0 ∆rri2::core x BY4741

BY4741 ∆gph1::core x C8.0

C8.0 ∆gph1::core x BY4741

BY4741 ∆ies2::core x C8.0

C8.0 ∆ies2::core x BY4741

BY4741 ∆sgt2::core x C8.0

C8.0 ∆sgt2::core x BY4741

BY4741 ∆hrd1::core x C8.0

C8.0 ∆hrd1::core x BY4741

3.10. Congenic Strains

A cross between BY4741 and GRA2 was performed, and after meiosis, haploid spores

were plated on medium at pH 8.4. Spores that could grow were then re-tested in a drop

assay on media at several pH (as described above) and the best-growing segregants

backcrossed to the BY4741 strain. This procedure was repeated for 8 generations,

establishing four independent congenic lines.

32

Table 1: Congenic Strains

Strain Description

GRA2 High MP parent

BY4741 Low MP parent

N1HA Haploid spore of GRA2 x BY4741 - First generation, line A

N1DA Diploid N1HA x BY4741, line A

N2HA Haploid spore of N1HA x BY4741 - Second generation, line A

N2DA Diploid N2HA x BY4741, line A

N3HA Haploid spore of N2HA x BY4741 - Third generation, line A

N3DA Diploid N3HA x BY4741, line A

N4HA Haploid spore of N3HA x BY4741 - Forth generation, line A

N4DA Diploid N4HA x BY4741, line A

N5HA Haploid spore of N4HA x BY4741 - Fifth generation, line A

N5DA Diploid N5HA x BY4741, line A

N6HA Haploid spore of N5HA x BY4741 - Sixth generation, line A

N6DA Diploid N6HA x BY4741, line A

N7HA Haploid spore of N6HA x BY4741 - Seventh generation, line A

N7DA Diploid N7HA x BY4741, line A

N8HA Haploid spore of N7HA x BY4741 - Eighth generation, line A

N1HB Haploid spore of GRA2 x BY4741 - First generation, line B

N1DB Diploid N1HB x BY4741, line B

N2HB Haploid spore of N1HB x BY4741 - Second generation, line B

N2DB Diploid N2HB x BY4741, line B

N3HB Haploid spore of N2HB x BY4741 - Third generation, line B

N3DB Diploid N3HB x BY4741, line B

N4HB Haploid spore of N3HB x BY4741 - Forth generation, line B

N4DB Diploid N4HB x BY4741, line B

N5HB Haploid spore of N4HB x BY4741 - Fifth generation, line B

N5DB Diploid N5HB x BY4741, line B

N6HB Haploid spore of N5HB x BY4741 - Sixth generation, line B

N6DB Diploid N6HB x BY4741, line B

N7HB Haploid spore of N6HB x BY4741 - Seventh generation, line B

N7DB Diploid N7HB x BY4741, line B

N8HB Haploid spore of N7HB x BY4741 - Eighth generation, line B

N1HC Haploid spore of GRA2 x BY4741 - First generation, line C

N1DC Diploid N1HC x BY4741, line C

N2HC Haploid spore of N1HC x BY4741 - Second generation, line C

N2DC Diploid N2HC x BY4741, line C

N3HC Haploid spore of N2HC x BY4741 - Third generation, line C

N3DC Diploid N3HC x BY4741, line C

N4HC Haploid spore of N3HC x BY4741 - Forth generation, line C

N4DC Diploid N4HC x BY4741, line C

N5HC Haploid spore of N4HC x BY4741 - Fifth generation, line C

N5DC Diploid N5HC x BY4741, line C

N6HC Haploid spore of N5HC x BY4741 - Sixth generation, line C

N6DC Diploid N6HC x BY4741, line C

N7HC Haploid spore of N6HC x BY4741 - Seventh generation, line C

N7DC Diploid N7HC x BY4741, line C

33

N8HC Haploid spore of N7HC x BY4741 - Eighth generation, line C

N1HO Haploid spore of GRA2 x BY4741 - First generation, line O

N1DO Diploid N1HO x BY4741, line O

N2HO Haploid spore of N1HO x BY4741 - Second generation, line O

N2DO Diploid N2HO x BY4741, line O

N3HO Haploid spore of N2HO x BY4741 - Third generation, line O

N3DO Diploid N3HO x BY4741, line O

N4HO Haploid spore of N3HO x BY4741 - Forth generation, line O

N4DO Diploid N4HO x BY4741, line O

N5HO Haploid spore of N4HO x BY4741 - Fifth generation, line O

N5DO Diploid N5HO x BY4741, line O

N6HO Haploid spore of N5HO x BY4741 - Sixth generation, line O

N6DO Diploid N6HO x BY4741, line O

N7HO Haploid spore of N6HO x BY4741 - Seventh generation, line O

N7DO Diploid N7HO x BY4741, line O

N8HO Haploid spore of N7HO x BY4741 - Eighth generation, line O

N1HR Haploid spore of GRA2 x BY4741 - First generation, line R

N1DR Diploid N1HR x BY4741, line R

N2HR Haploid spore of N1HR x BY4741 - Second generation, line R

N2DR Diploid N2HR x BY4741, line R

N3HR Haploid spore of N2HR x BY4741 - Third generation, line R

N3DR Diploid N3HR x BY4741, line R

N4HR Haploid spore of N3HR x BY4741 - Forth generation, line R

N4DR Diploid N4HR x BY4741, line R

N5HR Haploid spore of N4HR x BY4741 - Fifth generation, line R

N5DR Diploid N5HR x BY4741, line R

N6HR Haploid spore of N5HR x BY4741 - Sixth generation, line R

N6DR Diploid N6HR x BY4741, line R

N7HR Haploid spore of N6HR x BY4741 - Seventh generation, line R

N7DR Diploid N7HR x BY4741, line R

N8HR Haploid spore of N7HR x BY4741 - Eighth generation, line R

4. Genotyping

4.1. Yeast Genomic DNA Extraction (Phenol Method)

A 5ml overnight culture was harvested, and the supernatant was removed. To the tube we

added ~0.3g acid-washed glass beads, 300µl lysis buffer (2% Triton, 1% SDS, 0.1M NaCl,

10mM TrisHCl pH8, 1mM EDTA) and 300µl 25:24:1 phenol:chlorophorm:isoamyl. Tubes

were vortexed for 20’, and centrifuged for 10’ at 14 krpm. The top phase was carefully

transferred to a new tube containing 3 volumes of ethanol in order to precipitate the DNA.

Tubes were incubated for 20' at RT and centrifuged for 10’ at 14 krpm. The pellet was

34

resuspended in 300µl TE x1 (10mM TrisHCl pH 7.5, 1mM EDTA) with RNase (Sigma)

50µg/ml in 65ºC for 10'. Then DNA was precipitated by adding 0.3 volumes of 10M

Ammonium acetate and 3 volumes of ethanol 100% and incubated at -20ºC for 30 min-16

hrs. Tubes were centrifuged for 10' at 14krpm at 4ºC. Pellet DNA was washed with cold

ethanol 70%, and resuspended in 50µl H2O or 10mM Tris-HCL pH 7.5. This protocol yield

~10µg DNA. When larger quantities of DNA were necessary, we started with higher

volumes cultures and used the appropriate ratios of materials.

4.2 Genomic Mismatch Scanning (GMS) Method

Genomic DNA was isolated using phenol. We followed the GMS protocol as described

by D. Smirnov (Smirnov, Bruzel et al. 2004). Briefly, 20µg of genomic DNA samples

were digested with PstI (New England Biolabs, MA, USA) restriction enzyme. DNA

samples from the progenitors were methylated with dam methylase (10 units/µg of

genomic DNA). PstI digested genomic DNA from one individual was mixed with PstI

digested and methylated genomic DNA from the other individual in a. DNA was

denatured and reanealed Samples were digested with DpnI and MboI, followed by

addition of exonuclease III. After heat inactivation DNA was EtOH precipitated. The

DNA mixture was incubated with MutS, MutL and MutH. Exonuclease III was added

with incubation at 37°C for 15 minutes followed by heat inactivation. The eluted DNA

was amplified by PCR.

4.3. Hybridization and Analysis of GMS

DNA from each GMS reaction was labeled with Cy3–dCTP and a reference genomic

DNA was labeled with Cy5–dCTP (Amersham Pharmacia Biotech). The labeled samples

were hybridized to arrayed slides overnight at 60°C. Slides were washed once with 2×

SSC, 0.2% SDS for 10 minutes. Slides were spin dried and scanned.

4.4. Sample Preparation and Hybridization to Oligonucleotide Array

35

Yeast strains were routinely grown in yeast extract, peptone, and dextrose (YPD) medium

pH 6. Genomic DNA was purified from 2*1010

cells using a phenol. 15µg of yeast

genomic DNA was fragmented to an average size of 25 bp with 1 unit of DNase I

(Promega) for 5 min at 37°. DNase I was inactivated by 15 min boiling. After

heat

inactivation of DNase I, the DNA fragments were end-labeled by the addition of 37 units

of terminal transferase and 1.5 nmol Biotin-N6-ddATP for 90min at 37°C. Each sample

was hybridized to the array in 300 µl containing 1x MES buffer [100 mM

MES, 1 M

(Na+), 20 mM EDTA, 0.01% Triton X-100], 33µg

herring sperm DNA, 150 µg BSA, and

15 nmol of control oligonucleotide that hybridizes to control features on the gene array.

Samples were heated to 95° for 10 min, placed on ice for 5 min, and then

applied to the

gene array. Hybridizations were carried out at 45° for 24 hr with mixing on a rotisserie at

60 rpm. Following hybridization, the solutions were removed according to standard

Affymetrix protocols. All washes were automated on a fluidics station (Affymetrix).

Gene arrays were then scanned at an emission wavelength of 560 nm at 3 µm

resolution

using a specially designed confocal scanner (Affymetrix). The hybridization intensity for

each 25-bp probe from each scan was computed using Affymetrix GeneChip software.

4.5. Analyzing Tiling Arrays Data

We first applied SNPScanner (Gresham, Ruderfer et al. 2006) for detecting SNPs in a

single chip. SNPScanner performs model training based on two sets of arrays representing

strains with known genomic sequences. In our case, we used as one set the replicates from

the reference strain BY4741, and the other set contained the replicate arrays for the strain

RM11 [taken from (Gresham, Ruderfer et al. 2006)].

Once trained, the SNPScanner model predicts mutations in each chip (and replicate)

separately. Nucleotide positions that manifest a significant drop in the intensity of

hybridization given the model are predicted to be mutations. This detection takes into

36

account probe nucleotide content and the SNP position within the probe, and integrates

input from all ~5 probes that cover each position. Our further analysis focused only on

positions that attained a SNPScanner score >=5 in at least 50% of the replicates of the same

strain. In the second phase, we estimated the signal loss at each candidate SNP position

using the Hodges-Lehamn estimator for every genomic position (Hollander 1999), testing

whether or not the perfect match probes of a mutated strain have significantly lower signal

than their corresponding mismatch probes. This test sacrifices some information as it

ignores the probes’ content and the position within the probe, which are accounted for in

SNPScanner. However, unlike SNPScanner, it takes into consideration repeated

measurements and can measure signal loss between any two strains, not requiring that one

of them is the original reference strain (BY4741 in our case). In particular, this is important

for the congenic lines analysis, as we can compare the High MP parent strain with its

congenic descendants.The signal loss at each candidate position was identified using the

Affymetrix Tiling Analysis Software, which computes the significance of the hybridization

intensity drop using the Wilcoxon signed-rank test. Since the reference genome for the

Affymetrix chips is similar to BY4741, a drop in the signal suggests a potential SNP in the

given position. We filtered all putative SNPs that did not reach a p-value of 0.01, and also

discarded positions that attained p<0.01 in the BY4741 strain, implying that the SNP was

present already in our reference strain. We also filtered some additional putative SNPs

based on manual inspection of their signals.

An additional analysis was applied to the congenic strains in order to identify regions that

originated from the GRA2 strain. For each congenic line, we compared the 8th

generation

strains to BY4741, using the two-sample-analysis as implemented in the Affymetrix Tiling

Array Software. For each nucleotide, the comparison resulted in an estimation of the

intensity drop between the congenic strain and BY4741 as a reference. In order to detect

37

regions that are likely to originate from the GRA2 strain, we first selected in each

chromosome the top 0.05% nucleotides that showed the highest drop. Next, every two

selected nucleotides at a distance <8000bp were defined as an interval that is suspect of

originating from the GRA2 strain, and overlapping intervals were merged. Last, isolated

selected regions of length <1000bp were discarded. The result is a collection of intervals

H(1),…H(n).

Using the same method we compared the same congenic strain to the GRA2 strain,

resulting with regions L(1),…L(m) that are likely to originate from the BY4741 strain.

If all inference was perfect and complete, the two sets of regions H and L should have

formed a complementary cover of the genome, and the endpoints of the intervals in each

would give the recombination points. In reality, in order to handle noise, we combined the

two information sources as follows. Let H(i)=[r,l], and let L(j)=[r’,l’] be the leftmost

interval in L s.t. r<r’. If l=<r’ we defined the recombination point to be in the middle

between the two regions, i.e. (r’-l)/2. If r’<l, and l’<l, we discarded L(j); otherwise, we

defined the recombination point to be in the middle of the overlapping region, i.e, (l-r’)/2.

After applying this method, less than 10% of the genome was found to be inherited from

the GRA2 parent.

We repeated this procedure on the 4th and 6th generations of the congenic lines, and as

expected, virtually every region in a late generation was contained in some region in an

early generation of the same line. Validation of all SNPs acquired during in-lab evolution

was done by PCR and sequencing. Some of the regions found by congenic lines were

also determined by PCR and sequencing of several SNPs in the genotyped lines and in

other independent congenic lines.

38

5. PCR and Sequencing Analysis

All SNPs found at the ILE strains were verified by PCR and sequencing. Candidate

congenic regions were also verified by sequencing of 1kb inside the region. PCR reaction

was performed with an annealing temperature of 52oC, using primers at a final

concentration of 0.4µM. Elongation time was 1 minute for each Kb of the longer

predicted product.

Table 2.1: List of primers that were used for verification of SNPs found in ILE strains.

NAME Sequence from 5' to 3' ORF Verification

GY2_F AGTGCTTGAACGGATGATTCC MMP1/GTT2 TRUE

GY2_R AAGAGAATGCAACAGCGCCC MMP1/GTT2 TRUE

GY3_F TGCAAACGTCGGCTTCTTTA GPH1 TRUE

GY3_R ACGGACCCAACAATAGTTCCA GPH1 TRUE

GY4_F ATGCCGGAAACGGTCTTAG OAC1 TRUE

GY4_R TGTAGTGCGTTTGTTCACCA OAC1 TRUE

GY5_F CTATTGAAACGTTTGACGCG RRI2 TRUE

GY5_R CAATGCACTTCTTAAACAACC RRI2 TRUE

GY6_F CATGTCCGTAAATCGGGTTG ECM21 TRUE

GY6_R AACCCTCTCCCACCAAGAA ECM21 TRUE

GY7_F AAGATGTCCTTACGTGGAGC GPI17 TRUE

GY7_R TCGTTGTTTGTAGAAACGCC GPI17 TRUE

GY8_F TCATTAAAGACACCGCCAAG YFR057w TRUE

GY8_R TCGGATCACTACACACGGAA YFR057w TRUE

GY9_F CAATGTTAATCCTCCTCCCA IES2 TRUE

GY9_R ATAGCCTCCAGATTCCCCG IES2 TRUE

GY10_F GGGTGCATTTTCATTCCTGT MAC1 TRUE

GY10_R CTCCGGACTCTACCAAGTC MAC1 TRUE

GY11_F TTTGTTTGGAGTCTCACCAG CTF3 TRUE

GY11_R AGCGCACTACTGAGTCCAA CTF3 TRUE

GY12_F AAGGGTTTTCATTGTGGTGG HRD1 TRUE

GY12_R ATATTCCCAACGATAGCGCA HRD1 TRUE

GY13_F AAAAACCCCAGTGCCAGTG SKT5 FALSE

GY13_R GAGCAGAACCTGCTTTTGAA SKT5 FALSE

GY14_F ACCCGCTAGGTTACAAAACG YBL028c FALSE

GY14_R CGCAATATTCATCCAGTTCC YBL028c FALSE

GY15_F TCACATCCGAACTATCCCAA MRPS9 FALSE

GY15_R TATGCCTGAACCCGGTGTTA MRPS9 FALSE

GY16_F TTCAGTCTGCACCCTGTTTC FET5 FALSE

GY16_R TCAAGCATCGTCTCATCCAT FET5 FALSE

GY17_F TGGAAAAACTAAGGACGGGA ECM32 FALSE

GY17_R TTGTTTGCCACGACTTGCAT ECM32 FALSE

GY18_F AGAAAATCGCCAAAACGAGG CUP2 FALSE

GY18_R GCAATTCGACAATTGAAGAGA CUP2 FALSE

39

GY19_F ACCTTCAACG TAGTCTTTAG KEM1 FALSE

GY19_R CTCACGATTAATGGTCTTTTTC KEM1 FALSE

GY20_F CTCATATGAAAAAGTAATGG AST2 FALSE

GY20_R TCCCCTCTACCCGAGAAATT AST2 FALSE

GY21_F ACCGAGGATCATCAGCGTTA SPS100 TRUE

GY21_R ATCTAAGCGGCCAATAAACG YHR140w TRUE

GY22_F CCATACTTTCCTGGCTTTTC CDC23 TRUE

GY22_R TCCCTGTGTAAATCAGGCAA CDC23 TRUE

GY23_F ACCTGCTTTTTGAGGAGACC NMD4 TRUE

GY23_R GCAAACCTGTGTCGATTTTTT NMD4 TRUE

GY24_F GCAAGATGTTGACCCTTGAA SGT2 TRUE

GY24_R AAGGGCTATAGGCGGCATAA SGT2 TRUE

Table 2.2: List of primers that were used for verification of congenic regions

NAME Sequence from 5' to 3' Ch. Congenic region

GY72_F TGGTTATGACTTTGGGTTCG III 168966-200800

GY72_R TTTGGATCGTCCGGTGAAAT III

GY73_F TGGCGGCATAGTAACATCAA V 1-29961

GY73_R TCATTCGTCAAAACCTGCAC V

GY16_F TTCAGTCTGCACCCTGTTTC VI 30356-51606

GY16_R TCAAGCATCGTCTCATCCAT VI

GY19_F ACCTTCAACG TAGTCTTTAG VII 143903-200324

GY19_R CTCACGATTAATGGTCTTTTTC VII

GY74_F ATCCCATTTATCGAAGGGGAA VII 325256-396816

GY74_R ATCAAACTGCAAGGGTAAGGC VII

GY40_F AAATTGGCCACTGCAAATCTC VIII 478122-517418

GY40_R GGTTTTGCAACTCGCTAAGA VIII

GY41_F TATAGAACCGATGACGCACTG IX 424859-431838

GY41_R AAGAGCCTTTGCATCTTCTGA IX

GY75_F GCTTGCCCTGTTTAATGTACG X 633129-683951

GY75_R AAAATACAGCCGTGATCAAGG X

GY43_F GCCTCATTTCGATATGCCAA XI 533219-574982

GY43_R ACTCTCAACGTGGTGCGTTT XI

GY76_F AGGACACCCATAGCTGAAGC XII 92342-114724

GY76_R ATTTCCGAGACTCCAGAAGCA XII

GY81_F AAAGCCACCAGCTTAATCCA XII 262946-280320

GY81_R TTGCCTACAGTTTCACCGTAA XII

GY77_F TGTTCAGCCTTACATCGAAGA XII 606720-775305

GY77_R AATGGAGACAGGATCCGGAA XII

CTR3_F CGAGCAAATTAGCGCCATAA XII 937245-949420

CTR3_R CATTTTTGTCTTTCTACAAGCAG XII

GY25_F ATGGATATGCACACTTCCCCA XIII 307000-355700

GY25_R ACGGAATGAGGCCGATAGAA XIII

GY78_F TCCACACCAAATCTTGATGAA XIV 453937-509171

GY78_R TGGACAATCTGCGTGACAAT XIV

GY64_F ACGGTATGGAATTGGGAAAC XVI 785313-804496

GY64_R GGCTGCCACACTTGTTATTCT XVI

40

Table 2.3: List of primers that were used to construct the cassette for the allele swapping

NAME Sequence from 5' to 3'

GY36_F_gpi17 atgtccaatgcaaatctaagaaaatgggttggtttttgctttgttgccatttatctagggataacagggtaatttggatggacgcaaagaagt

GY36_R_gpi17 gtaacgacatgaaactggttgccttcactttccatctgctcaatagtctcttcgtacgctgcaggtcgac

GY37_F_mac1 ccgcttttaacgatattttacaagaaaactacaatagttctgttcctggtgctagggataacagggtaatttggatggacgcaaagaagt

GY37_R_mac1 ccatcacaagtgcagttatccggaggacatatgcattccttgtcagtgcatttacttcgtacgctgcaggtcgac

GY38_F_gph1 cgcttgtgtaaaaccgagtgtcctgaaggtgggtaaggcggtccttattatctagggataacagggtaatttggatggacgcaaagaagt

GY38_R_gph1 ctatgtaatctatcacgttttgtcactgtctcgctcttttcaaacaatccttcgtacgctgcaggtcgac

GY44_F_oac1 atgtcatctgacaactctaaacaagataaacaaattgaaaaaacagccgctagggataacagggtaatttggatggacgcaaagaagt

GY44_R_oac1 ggctcataaaaccctaatctggaaccatttagcccaatttgatagatatattcgtacgctgcaggtcgac

GY45_F_rri2 aagacgaacgatataaagaggctcgcgatttatttttaaagatatattactagggataacagggtaatttggatggacgcaaagaagt

GY45_R_rri2 ctttctcgatgtcgaacaagaatactctttcccacttgggtgccaataccttcgtacgctgcaggtcgac

GY46_F_ies2 ttatgtaaacttttacgtattttaattttcaaaattcttacaatgattaatagggataacagggtaatttggatggacgcaaagaagt

GY46_R_ies2 gctgttcttctagatgatttggctgtgactatctgatcgtcaatatcttcttcgtacgctgcaggtcgac

GY47_F_sgt2 aattttgccaccatataaaatgagtacgagcgatataatcggacaactgatagggataacagggtaatttggatggacgcaaagaagt

GY47_R_sgt2 ttaaagaatcggcgccatcttctgaaatttctttcttttccacaatggaattcgtacgctgcaggtcgac

GY48_F_nmd4 atggggcagcttttctcaaatttttttgtttctttgtgaccctacgggggtagggataacagggtaatttggatggacgcaaagaagt

GY48_R_nmd4 tctgagattatgaaagtcgtcacgttttacgaaggtattgaatttttcagttcgtacgctgcaggtcgac

GY49_F_ctf3 gaaacggccgaaaataccaaaataaaattcaccagtgggatcataaatgatagggataacagggtaatttggatggacgcaaagaagt

GY49_R_ctf3 ccacaagtaatcatatgacgccatgaatttaccactagtgcttgtaaatagttcgtacgctgcaggtcgac

GY50_ F_ecm21 ccgccaaacgcaaggcgacacagtaccactgccatacagggctctatcagtagggataacagggtaatttggatggacgcaaagaagt

GY50_ R_ecm21 atcattgagtccgtttagtctggacaggtactcatcgtcattcgaggatattcgtacgctgcaggtcgac

41

GY51_ F_hrd1 actgtcgactttctaccgatttgttggttcctttttataaccagaaaatctagggataacagggtaatttggatggacgcaaagaagt

GY51_ R_hrd1 gttaggagttgccatagtaaggtagaatttaataagatgaatatcgacaattcgtacgctgcaggtcgac

GY52_ F_yhr140w gggtacagtgagattccccaacaggaattcgcttctgtacatcagccggttagggataacagggtaatttggatggacgcaaagaagt

GY52_ R_yhr140w cgcctttttcaggtcagcgaacgcttataattccaagccatctcatgttattcgtacgctgcaggtcgac

GY53_ F_gtt2 tttgaaatcttcttaaatttctagtcggcaagcagcaaagaatttcagtatagggataacagggtaatttggatggacgcaaagaagt

GY53_ R_gtt2 aaaatttaggggttcttcgcggcgccattctgtcaagtaaaaaaatagctttcgtacgctgcaggtcgac

GY56_ F_yfr057w ctagttgcactaggcgcaaaaaattcctttgacaatagcctttcaaagcatagggataacagggtaatttggatggacgcaaagaagt

GY56_ R_yfr057w taatataaaatccacctaataatcaccgttaaactcagctaaacgtaaaattcgtacgctgcaggtcgac

GY79_F_cdc23 cctgtacctccttggttctacgttgtttgatgctaaagagtttgatcgatgcgtagggataacagggtaatttggatggacgcaaagaagt

GY79_R_cdc23 gtttgatcaacatcaatagcttctgcaagacctgctagcgcctctgctgcccacttcgtacgctgcaggtcgac

42

6. Estimating the Fixation Stage

For each mutation a population of cells from each selection stage was PCR-amplified

and sequenced. The sequencing results were then viewed as a chromatogram, in which

certain positions contained two overlapping graphs. The surface under these curves was

calculated, producing an approximate estimate of the representation of each allele in the

PCRed population. For each line, populations were amplified at different stages of the

selection procedure. The stage at which only one of the two alleles was detectable was

defined as the fixation stage.

7. Estimating the Relative Effect of Each QTL Using Allele Swapping

In order to evaluate the contribution of each QTL, we used the information from the

allele swapping experiment. We divided the relative fitness of the ancestor carrying the

swapped allele (RF Allele swap) by the differences between the relative fitness of the Low

MP ancestor (RFAncestor) and the relevant evolved strain (RF Evolved strain) and calculated

the percent of the contribution of each swapped allele.

8. Molecular Biology Techniques

8.1. Yeast Chemical Transformation

Yeast overnight cultures were diluted 1/20 into 50ml of YPD and incubated at 30ºC, 220

RPM for 4hrs. Cells were harvested at a density of 2*107 cells/ml, washed twice with

cold H2O and resuspended at a final volume of 1ml 0.1mM LiAcetate. 100µl of the cell

suspension was then added to 360µl transformation mix containing 33% PEG3500,

0.1M LiAcetate, 0.27 mg/ml Salmon Sperm Single Stranded DNA. 1µg of DNA was

43

added and the mixture was vortexed gently and incubated at 42ºC for 40 min. Following

heat shock cells were spinned down and resuspended in 1 ml of YPD. Cells were

incubated at RT for 30 min and then plated on selective plates (Gietz and Woods 2006).

8.2. E. coli Plasmid Extraction

Plasmids were extracted from E. coli using the DNA-spin plasmid DNA purification kit.

8.3. E. coli Transformation

Standard protocol to prepare competent cells was used (Chung and Miller 1993).

Competent cells were thawed on ice. 50-100 µl bacteria were mixed with 1µl DNA, and

incubated at 42ºC for 2'. Following heat shock cells were immediately transferred to ice

for 2' and afterward resuspended in 1ml LB and were allowed to grow at 37º for 1hr.

Appropriate dilutions were plated on selective plates (LB+ Amp) and allowed to grow

over night at 37ºC.

9. Systematic Prediction of QTLs in Each Genomic Region

Identified by Congenic Lines

We combined information about protein-protein interactions from various sources

(Breitkreutz, Stark et al. 2008; Yu, Braun et al. 2008; Matthews, Gopinath et al. 2009).

We then calculated, for each gene in the identified intervals, the minimal network

distance from any gene uncovered using in lab evolution (ILE gene set) and picked the

gene for which this distance was minimal. In case several such genes were found, we

picked the gene that had a path of minimal length to more ILE genes. Finally, in case of

ties, we picked the gene that was most similar to some ILE gene based on GO semantic

similarity (Lord, Stevens et al. 2003) (Taking into account only "biological process"

annotations). Validations were carried out with the two higher- and lower- ranking

deletion strains available (essential genes were not tested).

44

Results

The goal of this project was to characterize and identify all the components of a

complex genetic network that affect a particular quantitative trait. We have chosen the

ability of yeast to grow under alkali stress as the trait for dissection. We first

characterized several aspects of this trait and then we used in-lab evolution and

constructed congenic lines in order to map the genes and to estimate their effect on the

fitness at alkali stress. Figure 5 summarizes our combined strategy for QTLs dissection

and the results obtained. Briefly, we first characterized the trait, estimated the

heritability and the number of QTLs affecting it. Then we used two different strategies

for dissecting the network: A. An in-lab-evolution (ILE) strategy to enrich beneficial

mutations. B. The construction of congenic lines, starting with a clinically isolated strain

and a laboratory strain as a parent. Genotyping of ILE strains and congenic lines was

performed by several methods, including Genome Mismatch Scanning (GMS), Y98

Affymatrix arrays, tiling arrays and direct sequencing. Our results reveal several sets of

QTLs that confer ability to grow under alkali stress. We mapped several mutations and

estimated their effect on the trait. We show that although some mechanisms are shared

by all strains, different sets of genes can evolved to confer similar stress resistance.

45

46

Figure 5: General strategy for mapping quantitative trait loci (QTLs) affecting

genetic variability. See text for details. A. In-lab evolution (ILE). A strain with poor

ability to grow under alkali stress was serially transferred to media of increasing pH.

B. Construction of congenic lines up to the 8th

generation. C. Serially transferred

cultures accumulate beneficial mutations. After long-term selection at increasing

pH levels, cells were isolated from low pH media and individual colonies were grown

and tested for their ability to grow at high pH. The ancestor (BY4741) and colonies

derived from individual selection lines are shown. D. Map of QTLs affecting the MP

phenotype. Schematic representation of the sixteen chromosomes of S. cerevisiae

(black circles represent centromeres). Regions that were inherited from the High MP

parent in the congenic lines and were identified by hybridization to oligonucleotide

arrays are marked in red. QTLs showing mutations in the in-lab evolved strains are

marked green, yellow and orange.

47

1. Characterizing the Ability to Grow at High pH

Within genetically diverse populations, individuals may exhibit large disparities in

fitness, and these differences may vary in diverse environments. We examined the

capacity of 37 clinically-isolated and laboratory Saccharomyces cerevisiae strains

(Materials and Methods 3.1) to grow on media of varying pH and determined the

maximal pH (MP) at which each strain was able to grow. The MP phenotypes of this

population exhibited a normal distribution ranging between 7.4 and 8.6, with a mean

around ph 8 (Figure 6).

pH 8.3 pH 6 pH 8.2 A

B

Figure 6: A. An example of ten-fold serial dilutions of 8 strains from a mixed population

of wild type laboratory and clinically- isolated strains at pH 6, 8.2 and 8.3. B. A normal

distribution of phenotypes regarding the ability to grow under alkali conditions (maximal

pH, MP) of this mixed population.

48

1.1. The Ability to Grow under Alkali Stress is a Quantitative Trait

BY4741 (Brachmann, Davies et al. 1998), a standard laboratory strain, grows poorly at

high pH, whereas the clinically isolated strain F1623 (MKF12) grows at pH8.6. F1623

is a diploid homothallic strain. After deleting its HO gene, we subjected it to meiosis

and chose one of highest MP spores, GRA2, which grows extremely fast at pH8.6. We

compared the relative fitness at high pH of GRA2 and BY4741. As expected, GRA2

exhibits higher relative fitness at high pH (Figure 7 and 8). The relative fitness at high

pH was estimated by measuring the doubling time of the strain at optimal and at high

pH. Hereafter, we refer to BY4741 as the Low MP and to GRA2 as the High MP strains.

We have also compared the differences in the general fitness of the two strains. GRA2

exhibits a shorter doubling time under standard growth conditions (Figure 7) (measured

as in reference (St Onge, Mani et al. 2007)). We took this advantage under consideration

when we estimated the relative fitness at high pH (see Materials and Methods 2.2).

Figure 7: Fitness measurement under optimal conditions of strains GRA2 and

BY4742. The doubling time (St Onge, Mani et al. 2007) of each strain was measured

while growing on YPD pH 6 during logarithmic phase, using an optical density

reader (see Methods).

49

A cross between the strains yielded a diploid hybrid (BY4741 X GRA2) whose relative

fitness at high pH media was slightly higher than that of diploid strains isogenic to

either parent, a phenomenon known in quantitative genetics as heterosis (Darvasi and

Pisante-Shalom 2002) (Figure 8). The phenotypic superiority of the hybrid implies that

both parents carry alleles contributing to the phenotype. Heterosis may be determined

by several mechanisms, including dominance complementation, overdominance and

epistasis (Lippman and Zamir 2007).

The hybrid was subjected to meiosis and tetrads analysis. For a trait controlled by only

one locus, we expect all the tetrads of that cross to exhibit a 2:2 ratio for the ability to

grow at high pH. Instead, we found tetrads for which only one spore could grow on high

pH (1:3 ratio) and tetrads in which all spores or some of the spores had an intermediate

phenotype while other had low or high phenotype. (examples in Figure 9). These results

confirm that the ability to grow at high pH is ruled by many QTLs and pinpoint some

genetic interactions among them.

Figure 8: Heterosis

A cross between the Low MP parent (BY4741) and the High MP parent (GRA2)

results in hybrid vigor (heterosis). A. Four serial ten-fold dilutions of diploid strains

were plated on regular and high-pH media. The parents and the hybrid are presented.

B. The graph shows the relative fitness of the three strains, defined as the doubling

time at optimal pH divided by the doubling time at pH 7.9.

50

1.2. Estimating the Heritability and the Number of QTLs that Affect the Trait

In order to estimate heritability and the number of QTLs that contribute to the MP

phenotype, the ability of 258 haploid segregants to grow at various pH levels was

monitored. The MP phenotypes of this population also exhibit normal distribution

(Figure 10) Heritability was calculated as the ratio between the genotypic and the

phenotypic variance [see Methods, (Brem, Yvert et al. 2002)]. Our results show that an

estimated 88% of the phenotypic variance is genetic. For a quantitative trait each allele

is neither necessary nor sufficient to produce the final phenotype; therefore, prediction

of the exact number of loci is impossible. However, from the proportion of segregants

exhibiting the most extreme phenotypic characteristic, a minimal estimate of QTLs can

be obtained [see Methods, (Steinmetz, Sinha et al. 2002)]. Only 1.9% of the segregants

were able to grow at pH 8.6, (as their High MP parent). This inheritance ratio indicates

Figure 9: Tetrad analysis: Tetrad analysis for a cross between the High MP GRA2

and the laboratory Low MP BY4741. Here we present ten-fold dilution assay for two

tetrads from this cross at YPD and pH 8.6

51

that at least 5.7 independent unlinked loci determine the variation in the ability to grow

at high pH between the two wild type strains analyzed.

2. Dissecting the Genetic Network

The common strategies for dissecting complex traits include linkage disequilibrium and

association studies, both aiming to detect association between the presence of a

particular genetic genotype or haplotype and a certain phenotype (Cardon and Abecasis

2003; Binder 2006). The central principle of our methodology is to analyze the genetic

network as a whole rather than trying to identify individual contributors. A genetic

network is defined as all the genes, and the interactions among them, that contribute to

the expression of a phenotype. We used a combination of two independent methods to

dissect the architecture of the quantitative trait loci underlying the trait. The first

method, in-lab evolution (ILE), is based on the evolutionary process in which beneficial

mutations are acquired during selection. The second method, the construction of

Figure 10 - Normal distributions of the MP phenotype in a population of 258

progeny from a cross between the high MP clinical isolated strain (GRA2) and the

Low MP laboratory strain BY4741.

52

congenic lines, is based on the principles of QTL detection by marker-trait association

in a whole genome scan. Figure 5 illustrates the combined strategy.

2.1. In-Lab Evolution (ILE)

2.1.1. Enrichment for Beneficial Mutation in Several ILE Lines

The ILE experiment was carried out by a serial transfer method. The Low MP

laboratory strain BY4741 grows slowly at pH 7.4. BY4741 was subjected to a stepwise

selection procedure lasting 120-150 generations, which enabled enrichment for

beneficial mutations that allow growth at high pH. We started cultures from single

colonies grown at pH 7.4, and successively transferred samples of the population to rich

medium at higher and higher pH levels up to pH 8.6 (Figure 5A). Five independent lines

went through this procedure in parallel to create five separate populations (A-C, F-G).

At the end of this long-term selection, individual clones were isolated from each

population and their ability to grow at high pH was examined after nonselective growth;

all the cells tested retained their high MP phenotype (Figure 5C). We crossed yeast cells

able to grow at pH 8.6 to the Low MP ancestor and examined the ability of the haploid

progeny to grow on media of various pH levels. As expected, a normal distribution of

phenotypes was observed in each case (Figure 11A). Here too, analysis of tetrads

showed different High MP:Low MP spore ratios (examples in Figure 11B).

53

2.1.2. Several Genetic Networks have Evolved During the ILE

In order to see if different mutations had evolved in different lines, we carried out

crosses between clones from different populations and the ancestor. Based on the

phenotype of each hybrid we conclude that the mutations selected included both

recessive and dominant (or at least co-dominant) alleles (examples are shown in Figure

12 A,B). Interestingly, different types of mutations were acquired at each clone. For

example the cross between A8.5-1 and the ancestor had an intermediate phenotype

(higher than the ancestor, but lower than A8.5-1) indicating that some of the mutations

in A8.5-1 are recessive while some are dominant. The majority of mutations in clone

C8.6-1 were dominant, while mutations in C8.6-8 were mostly recessive. We also

carried out crosses between clones from different lines. The MP phenotypes observed in

the hybrids were sometimes different from those of the parents indicating that different

genetic networks had evolved in each clone (Figure 12C).

Figure 11: A. Normal distributions of the MP phenotype in a population of 100

progeny from a cross between the Ancestor BY4741 and one of the strains that went

through the ILE process. B. A drop assay at high pH media of two tetrads from this

cross.

54

2.2. Identifying QTLs Using Congenic Lines

2.2.1. Constructing the Congenic Lines

Using ILE we were able to uncover new mutations affecting several networks of genes

selected in-lab. In order to explore additional variability that may have developed in

nature (under different environmental conditions), we have also dissected the genetic

network in GRA2, a naturally occurring yeast isolate able to grow at high pH. To dissect

Figure 12: Several genetic networks were selected for during the in-lab evolution.

Ten-fold dilutions of diploid yeast cells were plated on optimal (pH 6) and high pH solid

media. A and B – One clone from population A8.5 and two clones from population C8.6

were crossed to the Low MP ancestor. In each case, the hybrid had a different phenotype

relatively to the parent. A. The hybrid from a cross between the ancestor and a clone

from population A8.5 has an intermediate to high MP phenotype. B. Two independent

clones from strain C8-6 were crossed to the Low MP ancestor. The hybrid derived from

clone C8.6-8 has an intermediate to Low MP phenotype whereas the one derived from

C8.6-1 from the same population has a High MP phenotype. C. The phenotype of two

high MP clones from population C8.6 and F8.6 and the hybrid from a cross between

them. The hybrid has a Low MP phenotype indicating that most of the mutations that

occurred in these lines are recessive and affect different QTLs.

55

the network of genes contributing to the high MP phenotype of this strain, we created

congenic lines up to the 8th

generation (Figure 5B). Starting with the High MP GRA2

and the Low MP BY4741 parents, a series of backcrosses was carried out. At each

generation, segregants able to grow at pH 8.6 were selected for, and were backcrossed

to the Low MP parent. Four independent congenic lines were thus created in parallel.

After 8 rounds of such backcrosses, all segregants with high MP should have inherited

the genetic network from the High MP parent, whereas the rest of the genome

predominantly reflects that of the Low MP parent (Darvasi 1997). At each generation,

random recombination events are expected to reduce the size of the chromosomal

intervals inherited from the High MP parent. However, the reduction in size decreases

with each successive generation. Simulation showed that after 8 generations additional

backcrosses did not significantly reduce interval size.

2.2.2. Only One Genetic Network Contributes to the MP Phenotype in the

Congenic Lines

To investigate whether all congenic lines carried the same genetic network, or whether

each one acquired a different sub-network able to allow growth at high pH, we carried

out crosses between individuals from the four different congenic lines at the fifth

generation. If the genetic networks of two lines are identical, we expect all the progeny

to be able to grow at high pH. If, however, each parent inherited from the High MP

strain a different sub-network, then we should encounter, among the progeny, spores

with poor alkali resistance. The proportion of high MP segregants was 0.85-0.9 for all

inter-line crosses. This frequency is similar to the estimated heritability (0.88),

indicating that only one genetic network controls the MP phenotype in the wild strain

tested.

56

3. Genotyping

In order to identify the QTLs both in the ILE strains and in the congenic lines, a

genotyping process must be done. In recent years some new technologies were

developed that enable high resolution genotyping of the whole genome (Winzeler,

Castillo-Davis et al. 2003). Those methods use genetic polymorphisms such as

microsatellites and single nucleotide polymorphisms (SNPs). We have used three

different technologies.

3.1. Genotyping Using Genomic Mismatch Scanning (GMS) Method

The Genomic Mismatch Scanning (GMS) method (Nelson, McCusker et al. 1993;

Smirnov, Bruzel et al. 2004) is a high-resolution mapping technique. In GMS

heteroduplexes are formed from DNA fragments of two samples (by denaturation-

renaturation). The mismatch-containing fragments are depleted by the Escherichia coli

mismatch repair proteins MutL, MutH and MutS and exonuclease, while those bearing

no mismatch are spared. The bulk of the DNA is then amplified by PCR using universal

primers. The PCR products are then hybridized to DNA microarrays. Regions containing

mismatches are expected to be under-represented at the arrays. GMS was applied in

order to identify the mutations acquired during ILE in three strains. A correlation

analysis of the results shows that the three independently evolved lines were very

similar, and different from the original Low MP strain (Figure 13), suggesting that the

majority of the genome have acquired mutations during the ILE. However, during 120-

150 generations of ILE we expect that strains will acquire only a very small number of

mutations (Gresham, Desai et al. 2008) so the high correlation in this case could indicate

a technical problem. For example, too much degradation of the samples by the Exo1

nuclease may result in biased results. We therefore sequenced some of the regions that

were underrepresented in the arrays. As expected, many of the regions did not contain

57

any mutations. These results were mainly false positive. Since the GMS results were too

noisy, we had to repeat the genotyping stage using different methods. GMS was also

applied on the congenic lines and here too, the results were too noisy.

3.2. Genotyping the Congenic Lines Using Y98 Oligonucleotide Arrays

To determine the regions inherited from the High MP parent in the congenic lines we

used oligonucleotide arrays (Affymatrix Y98) designed for expression analysis. These

expression arrays contain a total of 157,112 probes (25-meres) derived from yeast

genome coding sequences (Winzeler, Richards et al. 1998; Winzeler, Castillo-Davis et

al. 2003). First, genomic DNA from both the High MP and the Low MP parents of the

congenic lines was hybridized to those arrays (Winzeler, Richards et al. 1998;

Deutschbauer and Davis 2005; Ben-Ari, Zenvirth et al. 2006) to detect allelic variation

and to generate a map of Single Feature Polymorphisms (SFPs) containing more than

2000 markers. These SFPs allowed us to genotype four strains (from two independent

congenic lines) obtained after eight successive backcrosses and one strain obtained at an

Figure 13: Correlation between BY4741 (the Ancestor) and three independently

selected strains (A, B and C). Each axis represents the Cy5/Cy3 intensity ratio for

every ORFs in one of the analyzed strains. Each spot represents the ratios for an

individual ORF. Differences in the Cy5/Cy3 ratio indicate the presence of SFPs. A

strong correlation is observed among all the selected strains, whereas low

correlation is observed between the selected strains and the ancestor.

58

intermediate (fourth) backcross (Figure 14A). As expected, the strains from the eighth

generation analyzed shared most of their genomes with the Low MP parent, whereas

less than 10% of their genomes were inherited from the High MP parent. The genetic

information potentially containing the High MP QTLs was present in chromosomal

intervals that were larger in the fourth generation backcross than in the strains that

underwent eight backcrosses, illustrating the reduction in size of genomic regions

originating from the High MP parent with each backcross (Figure 14).

As explained in the Introduction, SFPs detection is limited by probe density, which is

typically a few oligonucleotides per gene at the Affimatrix Y98 arrays. Even a

complete single coverage of the genome is unlikely to be sufficient for finding all

mutations, because statistically detectable reductions in hybridization intensity usually

require that a variant nucleotide fall within the central 15 bases of a 25-base probe

(Ronald, Brem et al. 2005). Therefore, the Affymatrix Y98 arrays could not be used for

identifying the mutations acquired during the ILE. Moreover, the 2000 SFPs which were

found between the Low MP BY4741 and the High MP GRA2 were insufficient to

identify candidate QTLs. Large chromosome intervals were not covered by those arrays,

and therefore, the borders of the regions were undetectable. We therefore decided to use

a higher resolution genotyping technique which was just developed and published

several month before that stage (Gresham, Ruderfer et al. 2006).

59

60

3.3. Genotyping Using Tiling Arrays

In order to achieve high genotyping resolution, both the congenic lines and the selected

lines (ILE) were genotyped by hybridization to 25mer oligonucleotide microarrays

(Affymetrix yeast tiling arrays) that provide complete and redundant coverage of the

~12 Mb S. cerevisiae genome. This design provides for multiple measurements of each

nucleotide's contribution to hybridization efficiency and therefore has the ability to

detect the presence and location of single nucleotide polymorphisms and deletion events

throughout the entire yeast genome with near nucleotide precision (Gresham, Ruderfer

et al. 2006; Schacherer, Ruderfer et al. 2007). The results were analyzed using the

SNPscanner (Gresham, Ruderfer et al. 2006) which calculates the log of likelihood ratio

(the prediction signal) for the presence of a SNP at each nucleotide position in the

genome using measurements from all probes that cover that site (Figure 15).

3.4. QTLs Identification in the ILE

In order to identify the beneficial mutations that occurred during the ILE we hybridized

to the tiling arrays the Low MP ancestor and three evolved strains: two clones from two

independent lines that had acquired the ability to grow at pH 8.5 and pH 8.6 (A8.5 and

C8.6) and one clone from an intermediate stage of line C (C8.0). Two potential

Figure 14: Genomic inheritance of segregants from congenic lines as obtained

from hybridization to oligonucleotide arrays. Chromosome numbers are plotted

along the y axis and position along the chromosome in the x axis. For each

chromosome one segregant from the 4th

generation and two segregants from the 8th

generation are shown, followed by the High MP and Low MP parents. Green

intervals were inherited from the Low MP parent and red intervals were inherited

from the High MP parent. A. Each 'plus' represent one SFP. B. Final map obtained.

61

problems may affect the detection of SFPs by this technology in the evolved strains.

First, the results obtained from the hybridization procedure include many false positive

results, depending on the threshold chosen for the analysis (Gresham, Ruderfer et al.

2006; Schacherer, Ruderfer et al. 2007). Second, during the selective growth, random

mutations (not related to the ability to grow at high pH) may accumulate in these strains.

The presence of these mutations may affect our ability to identify the relevant QTLs.

We therefore verified all mutations by sequencing and then tested the effect of each

QTL that found.

In order to identify the SFPs using the tilling arrays, we used two different thresholds of

prediction: First we used threshold 5, which is expected to avoid all false positives and

to give 77.5% of all real SNPs (Gresham, Ruderfer et al. 2006). Using threshold 5 we

Figure 15: An example of the SNPscanner results. The decrease in hybridization is

used to estimate the log of the likelihood ratio of the presence of a polymorphism versus

the absence of a polymorphism. The presence of a SNP typically results in a region of

positive prediction signal with a peak defined as the predicted SNP. Using this approach,

we detected single–base pair substitutions. Each color represents a different strain after

the in-lab evolution. Here, all three evolved strains, but not the ancestor BY4741, had the

same substitution. All SNPs were confirmed by sequencing.

62

identified 15 mutations (see Table 1). All of them were verified by sequencing and

found to be true positive (i.e., they contained SFPs). We then decreased the threshold to

3 which is expected to detect more than 90% of the SNPs with less than 5% false

positive (Gresham, Ruderfer et al. 2006). We found again all SNPs which were found

with threshold 5; however, this low threshold resulted in dozens of additional predicted

SNPs. We picked 8 SNPs that were found at that threshold in all triplicates and

sequenced the relevant regions after PCR amplification in the ancestor and the evolved

strains. We found no differences in the sequenced regions, indicating that the predicted

SFPs by threshold 3 were false. We therefore used only the results obtained with

threshold 5.

In total, 15 mutations were found (Table 3) in clones from selected lines C8.0, C8.6 and

A8.5. Table 3A presents a list of all mutations detected. Table 3B presents the false

positive results – SNPs which were detected by SNPScanner but sequencing of the

region revealed no mutation.

63

Table 3A: Location and nature of the mutations acquired during selection for the ability to grow at high pH in the ILE procedure.

Clone Gene Mutation Function Fixation stage1 Validation

2

C8.0, C8.6 GTT2/

MMP1 Converging 3’UTRs

Glutathione S-transferase/

S-methylmethionine permease. pH 7.7 No significant effect found

C8.0, C8.6 YFR057w Base substitution at

promoter Unknown. pH 7.8 Could not be tested

3

C8.6 ECM21 Nonsense at codon 193 Ubiquitin-ligase adaptor. No fixation Deletion, RHA, AS

C8.6 NMD4 /

YLR363w-a Base substitution at promoter

Nonsense-mediated mRNA decay/

Unknown. pH 8.4 Deletion, RHA, AS

C8.6 GPI17 S63 to L GPI-anchor transamidase. No fixation Essential gene

C8.6 YHR140w/

SPS100 Base substitution at promoter

Unknown/

spore wall maturation. No fixation Deletion, RHA, AS

C8.6

A8.5

G8.6

B8.6

MAC1

C271 to W

C271 to S, M386 Silent

C271 to Y

C271 to Y

Copper-sensing transcription factor.

pH 8.4

pH 7.6

pH 8.6

pH 8.64

Deletion, RHA, AS

Deletion, RHA, AS

Deletion, RHA, AS

Deletion, RHA, AS

A8.5 CDC23 M71 to I Ubiquitin ligase (APC/C). pH 8.4 RHA

A8.5 CTF3 R697 Silent Outer kinetochore protein. No fixation Deletion (mild effect)

C8.0 OAC1 I48 to F Oxaloacetate carrier. No fixation No significant effect found

C8.0 RRI2 A138 to P Ubiquitin ligase regulation. No fixation No significant effect found

C8.0 GPH1 3'UTR Glycogen phosphorylase. No fixation Deletion, AS

C8.0 IES2 Base substitution at promoter Chromatin remodeling. No fixation Deletion, AS

C8.0 SGT2 Start codon M1 to I Glutamine-rich chaperone. No fixation AS

C8.0 HRD1 L19 to F Ubiquitin ligase. No fixation AS

1 Stage at the ILE procedure at which the given allele is observed as the sole allele in the cell population.

2Assay that demonstrated a biological effect on pH homeostasis for the gene analyzed. RHA- Reciprocal Hemizygosity Assay. AS – Allele Swapping Assay

3 YFR057w is a telomeric ORF. All attempts to delete or replace it failed.

4 Approximately 80 percent of population in B8.6 has the mutation in MAC1

64

Table 3B: Location of false positive SNPs identified by SNPScanner

Clone Gene Chromosome Position

C8.0, C8.6, A8.5 SKT5 II 105413

C8.0, C8.6, A8.5 ybl028c II 167712

C8.0, C8.6, A8.5 MRPS9 II 535709

C8.6 ECM32 V 543250

C8.6 AST2 V 361590

C8.0, C8.6, A8.5 FET5 VI 49730

C8.0 CUP2 VII 191911

C8.0, C8.6, A8.5 KEM1 VII 179079

The mutations obtained in the independently selected lines C an A affect different genes

(with the sole exception of MAC1, which was mutated in both).

Moreover, the mutations observed in strain C8.0, which represents an intermediate stage

in the evolution of line C (to which C8.6 also belongs), only partially overlap those of

clone C8.6. Thus, some of the mutations that appeared in the population selected for

growth at pH 8.0 did not reach fixation in the population and, therefore, do not appear at

later stages of selection (pH 8.6). In order to pinpoint the stage at which each mutation

attained fixation within the population and to test if other lines had acquired the same

mutations we sequenced all the relevant SNPs (including upstream and downstream

sequences) in populations from all intermediate stages of the evolution in lines A and C

and from the last selection stage of lines B, F and G. In line A the mutation in MAC1

attained fixation at the stage that selected for the ability to grow at pH8 and the one in

CDC23 several steps later, at pH 8.3. In line C the mutation in MAC1 reached fixation at

pH 8.5, the one in GTT2 at pH 8.0 and that in NMD4 at pH 8.4. Interestingly, similar

mutations in MAC1 appeared also in populations B8.6 and G8.6 (C271 to Y) but not in

F8.6. However, none of the other genes in Table 3 were affected in lines B, F and G

despite the ability of those strains to grow at high pH. Thus, different mutations were

selected, with different kinetics, in each independent selection line, giving raise to the

individual genetic networks.

65

4. Validation of the Genes Set Identified in the ILE Experiment

As a first step, to determine whether all 15 genes mutated during ILE contribute to the

ability to thrive at high pH, we deleted the individual genes in the Low MP ancestor and

in each of the evolved clones and tested their ability to grow at high pH. In principle it is

not to be expected that a full deletion of the gene should display the same phenotype as

the point mutations found during the in-lab evolution procedure. Yet, some of the

deleted strains such as ∆mac1 and ∆gph1 exhibited a reduction in their ability to grow at

high pH, while ∆ecm21 showed improved fitness under these conditions (Figure 16).

This is consistent with the fact that a nonsense mutation in ECM21 was acquired by

clone C8.6 (Table 3). In agreement, deletion of the gene from the Low MP background

increased the fitness at high pH, whereas the same deletion did not improve the

phenotype of the C8.6 strain (Figures 16A, B).

Figure 16: Ten-fold dilutions of haploid yeast cells on optimal (pH 6) and high pH solid

media. A. Deletion of candidate genes at BY4741 background. B. Deletion of candidate

genes at C8.6 and A8.5 background. C. Deletion of candidate genes at C8.2

background.

66

4.1. Reciprocal Hemizygotes

Reciprocal hemizygosity is a tool for analyzing the contribution of each allele in the

same (hybrid) genetic background (Steinmetz, Sinha et al. 2002). Isogenic pairs of

strains were constructed, all with the same hybrid genetic background (Low MP x

Selected High MP), and differing by which of the two alleles of each gene was deleted

(Figure 17).

We tested the differences in the ability to grow at high pH between 9 pairs of reciprocal

hemizygotes of QTLs found in A8.5 and C8.6. Six out of the seven mutations tested in

C8.6 and one out of two mutations in A8.5 displayed a noticeable contribution to the

ability to grow at high pH (Figure 18A, B). However, there were no differences in the

phenotype between reciprocal hemizygotes pairs from the C8.0 background (Figure

18C).

Figure 17 [adapted from (Darvasi and Pisante-Shalom

2002)]: Each block represents a gene identified by ILE. A

pair of hemizygote hybrids is made for each gene, one with

the allele from the High MP evolved strain deleted and one

with the allele from the Low MP ancestor deleted.

67

68

4.2. One SNP Can Affect the Phenotype Via Two QTLs

One of the mutations identified in C8.6 occurs in chromosome XII, at the joint

regulatory region of two divergently transcribed genes, NMD4 and YLR363w-A.

Whereas NMD4 is a well-characterized gene that affects nonsense-mediated decay and

mRNA stability (He and Jacobson 1995), YLR363w-A encodes a nuclear protein of

unknown function. To determine which of the two genes is the one contributing to the

high MP phenotype we constructed two pairs of reciprocal hemizygotes. One pair with

reciprocal deletions of the NMD4 ORF, and the other pair with reciprocal deletions of

ORF YLR363w-A. The ability of the two pairs to grow on media of different pH was

tested. Figure 19 shows that deletion of any of the ORFs in the C8.6 background leads

to reduced MP, indicating that both genes contribute to the phenotype. We thus

conclude that a single mutation in the promoter of these divergent genes was selected in

C8.6, affecting two neighboring genes that contribute to survival under alkali stress.

Figure 19: A single SNP affects two genes that contribute to the ability to grow at

high pH. Drop assay for reciprocal hemizygote pairs of NMD4 and its adjacent,

divergently transcribed ORF YLR363w-A (of unknown function). Both genes

contribute to the ability to grow under alkali stress. Under the drop assay there is a

schematic view of the interval containing both genes.

69

4.3. Allele Swapping

Finally, we wanted to estimate the contribution of each mutation to the phenotype. In

order to do so, we created a set of 14 strains from the Low MP haploid background in

which one of its alleles was swapped by an allele from the evolved strain (11 different

QTLs and 3 independent alleles of MAC1). We measured the fitness of these strains (as

well as that of the evolved strains and the Low MP wild type control) in YPD and at

high pH (Figure 20). Ten out of the 14 mutations exhibited a significant effect on the

phenotype (Figure 20). Among them, the three different alleles of MAC1 (C271Y,

C271W, C271S) had the highest effect. Under this genetic configuration (single allele

swapping), most of the ability to grow at high pH can be attributed to MAC1, whereas

other QTLs have more modest contributions (Figure 20C).

70

71

5. QTLs Identification in the Congenic Lines

In order to genotype the segregants obtained from the congenic approach we took

advantage of the naturally occurring variation in sequence between the parents GRA2

and BY4741. The DNA microarrays were designed to match the DNA sequence of the

latter, and therefore regions inherited from the high MP strain containing SFPs are

easily detected. Four strains from two independent congenic lines were genotyped by

tiling arrays and candidate SFPs were then verified by PCR and sequencing. Two of the

strains (N8HA, N8HB) were obtained after eight successive backcrosses and two

additional strains represented intermediate stages (N4HA, N6HB - fourth and sixth

backcrosses, respectively). As expected, the region inherited from the High MP parent

decreased from one generation to another and those regions were much larger at the

fourth generation than at the eight generation (Figure 21). Strains from the eighth

generation analyzed shared most of their genomes with the Low MP parent, whereas

less than 10% of their genomes were inherited from the High MP parent. The genetic

information potentially containing the High MP QTLs was present in chromosomal

Figure 20: Estimating the relative effect of each QTL using allele swapping - 14

strains from the Low MP ancestor background were created. In each strain one gene

was swapped by the mutated allele from a High MP evolved strain. A. A schematic

illustration of allele swapping. Each block represents a candidate QTL. The ancestor

was used as a background. In each strain one QTL was replaced with the allele from

the evolved strain. B. The relative fitness (defined as the doubling time at optimal pH

divided by the doubling time at high pH) of 14 constructed strains, the ancestor, and

the evolved strains were measured at pH8. C. The relative effect of each allele on the

low ancestor background. The effect was calculated as the ratio between the

contribution of the allele to the phenotype and the difference between the evolved

strain and the ancestor.

72

intervals that were larger in the fourth and in the sixth generation backcross than in the

strains that underwent eight backcrosses, illustrating the reduction of genomic regions

originating from the High MP parent with each backcross.

Figure 5D and Table 4 present the 17 regions derived from the High MP parent found in

at least two of the congenic lines genotyped, which are candidates to contain QTLs.

Once the regions were identified by the tiling arrays, we confirmed them by sequencing.

We also sequenced the same regions in two additional independent congenic lines to see

whether they had also inherited the same regions from the high MP parent. Most of the

regions identified were conserved in more than one independently-obtained congenic

line (Table 5). On average, each region is 42 kb long and contains 17.6 genes. MAC1

was the only gene identified by the ILE procedure present in one of the congenic

regions.

73

Figure 21: Examples of results obtained with SNPscanner for chromosome V, VII and

X in congenic strain N8HA, N4HA, N8HB, N6HB (line A 8th

and 4th

generation and line

B 6th

and 8th

generation, respectively). Red lines represent SFPs that exist in the

congenic lines but not in the Low MP parent. Therefore red intervals represent regions

which were inherited from the High MP parent. Green lines represent SFPs which exist

in the congenic lines but not in the High MP parent. Therefore, green intervals represent

regions which were inherited from the Low MP parent. Red lines at low density in green

intervals were found to be false positive. Chromosome V- It is possible to see that the

region inherited from the High MP parent at the 4th

generation (green and red lines

N4HA) decreased significantly at the 8th

generation (green and red lines in N8HA)

Chromosome VII- example of two regions which were inherited in both lines (N6HB

and N4HA) Chromosome X- An example of a region which was inherited in all lines.

Higher resolution is obtained from the two independent lines A and B.

74

6. Fine Mapping in Congenic Lines using a predictive algorithm

In order to systematically predict the QTL in each genomic region identified, we took

advantage of the large amount of information available about the function of the yeast

genome. We applied a predictive algorithm (see Materials and Methods) that takes into

consideration the network proximity of each gene within the defined intervals to the

genes identified in the ILE procedure, and their similarity in GO annotation, generating

ranked probabilities of being a relevant QTL (Table 4). This method allowed us to

predict, for each interval, the genes most likely to affect the MP phenotype in the high

MP strain (Table 2). Five out of the predicted genes (MAC1, CTR1, FEN1, KEM1 and

PMR1) were identified in a systematic genome-wide screen as required for growth at

high pH (Serrano, Bernal et al. 2004). We have confirmed this phenotype (Figure 22).

From each region we picked the candidates with the highest and the lowest scores and

tested the ability of their deletion to affect growth at high pH. Twelve of the 29

deletions with the best scores tested decreased the ability to grow at high pH (Table 6,

Figure 22) while only four out of the 28 lowest-ranking candidates affected the

phenotype (p-value =0.023).

75

Figure 22: Validation of the predictive algorithm applied on the congenic

regions. Ten-fold dilutions of haploid yeast strains on optimal (pH 6) and high pH

solid media. Each strain carries a single deletion of a gene found at the congenic

regions and was predicted by our algorithm to be a good candidate to affect the

ability to grow at high pH. Out of 29 deletions tested, twelve have an MP

phenotype.

76

Table 4: Regions inherited from the High MP parent to the different congenic lines (A4, A8 –line A generation 4th

and 8th

and B6, B8 - line

B generation 6th

and 8th

, respectively), and candidate genes in those regions. N represents the number of genes in each region. Genes in bold

were also identified in the ILE experiments.

Ch. Position N Candidate Function

III 168966-200800 13 FEN1 Fatty acid elongase that affects cell wall synthesis. Deletion confers sensitivity to high pH

(Serrano, Bernal et al. 2004).

V 1-29961 7 SIT1 Ferrioxamine B transporter, induced during iron deprivation.

VI 30356-51606 8 ALR2 Probable Mg(2+) transporter.

VII 143903-200324 26 PMR1

KEM1

High affinity Ca2+/Mn2+ P-type ATPase. Deletion confers sensitivity to high pH (Serrano,

Bernal et al. 2004)

Exonuclease involved in mRNA stability. Deletion confers sensitivity to high pH (Serrano,

Bernal et al. 2004).

VII 325256-396816 31 HNM1 Choline transporter.

VIII 478122-517418 20 GPI161

RPN10

Subunit of GPI transamidase complex.

Subunit of the 19S regulatory particle (RP) of the 26S proteasome

IX 424859-431838 1 YPS6 GPI-anchored aspartic protease.

X 633129-683951 20 ILM1 Unknown function

XI 533219-574982 18 TIF1 Translation initiation factor eIF4A.

XII 92342-114724 7 KNS1 Nonessential putative protein kinase

XII 262946-280320 8 RPL22A Protein component of the large (60S) ribosomal subunit

XII 345291-374049 7 MDN11 Peroxyredoxin, protects against oxidative damage.

XII 606720-775305 67 MMS22 Protein involved in resistance to ionizing radiation

XII 937245-949420 2 CTR3 High affinity Copper transporter.

XIII 307000-355700 24 MAC1 Transcriptional activator of Cu transporters. Deletion confers sensitivity to high pH (Serrano,

Bernal et al. 2004)

XIV 453937-509171 28 RPL16B N-terminally acetylated protein component of the large (60S) ribosomal subunit

XVI 785313-804496 12 CTR1 High affinity Copper transporter. Deletion confers sensitivity to high pH (Serrano, Bernal et

al. 2004)

77

Table 5: Sequencing verification of regions inherited from the High MP parent in the congenic lines.

Ch. Position N N4HA N8HA N6HB N8HB N6HO N7HO N6HC N7HC

III 168966-200800 13 + + - - + + + +

V 1-29961 7 + + - - - -

VI 30356-51606 8 + + - - - -

VII 143903-200324 26 + - + - - -

VII 325256-396816 31 + - + + + + -

VIII 478122-517418 20 - - + + - -

IX 424859-431838 1 + + - - - -

X 633129-683951 20 + + + + + + + +

XI 533219-574982 18 - - + + + + -

XII 92342-114724 7 + + - - - -

XII 262946-280320 8 + + + - - -

XII 345291-374049 7 + + + + + + -

XII 606720-775305 67 + + - - + + -

XII 937245-949420 2 + + - - - - - -

XIII 307000-355700 24 - - + - - - - -

XIV 453937-509171 28 + + - - - -

XVI 785313-804496 12 - - + + - + +

78

Table 6:

No. ORF Name Chr Close ILE genes

Number

of close

ILE

genes

network

proximity

p-value of

minimal

distance

GO- Most

similar

ILE

gene(s)

Best GO

semantic

similarity

score

Alkalai

sensitive

1 YCR034W FEN1 3 YHR140W, MMP1 2 1 0.005566 GPI17 3.104507 Yes

YCR037C PHO87 3 YHR140W, MMP1 2 2 0.233099 MMP1 1.770543 No

YCR028C FEN2 3 YHR140W, MMP1 2 2 0.233099 MMP1 1.280903 No

YCR031C RPS14A 3 YLR363WA, HRD1 2 2 0.233099 GPI17 0.949776 No

YCR028CA RIM1 3 HRD1 1 2 0.447864 ECM21 0.767537 No

YCR030C SYP1 3 GTT2, YLR363WA, YFR057W, YHR140W, HRD1,

MAC1, GPI17 7 3 0.564904 ECM21 1.069253 No

YCR035C RRP43 3 GTT2, YLR363WA, ECM21, YHR140W, HRD1,

GPI17 6 3 0.593551 NMD4 1.651343 No

YCR036W RBK1 3 GTT2, YHR140W, HRD1, MAC1, GPI17 5 3 0.62809 NMD4,

GPI17 0.587945 No

YCR033W SNT1 3 GTT2, YLR363WA, HRD1, MAC1 4 3 0.668195 GPI17 1.077951 No

YCR040W MATAL

PHA1 3 YHR140W, MMP1, GPI17 3 3 0.709281 NMD4 1.098794 No

YCR038C BUD5 3 GTT2, YHR140W, HRD1 3 3 0.709281 SPS100 0.991166 No

YCR032W BPH1 3 YHR140W, HRD1 2 3 0.748404 ECM21 1.489095 No

YCR039C MATAL

PHA2 3 HRD1, MAC1 2 3 0.748404 NMD4 0.941824 No

2 YEL070W DSF1* 5 HRD1 1 2 0.447864 NMD4,

GPI17 0.176642 No

YEL066W HPA3 5 GTT2, YHR140W, HRD1, GPI17 4 3 0.668195 GPI17 1.172541 No

YEL065W SIT1 5 YFR057W, HRD1 2 3 0.748404 MMP1 1.057544 No

YEL071W DLD3 5 GTT2, HRD1 2 3 0.748404 HRD1 0.514057 No

YEL064C AVT2 5 HRD1 1 3 0.789164 MMP1 1.14926 No

YEL072W RMD6 5 GTT2, HRD1, MAC1 3 4 0.823212 SPS100 0.783444 No

YEL069C HXT13 5 MMP1 1.280903 No

79

3 YFL050C ALR2 6 MMP1 1 1 0.041087 MMP1 1.280903 No

YFL048C EMP47 6 YHR140W, HRD1, MMP1 3 2 0.133246 MMP1 1.280903 No

YFL044C OTU1 6 HRD1 1 2 0.447864 HRD1 1.592784 No

YFL041W FET5 6 YHR140W 1 2 0.447864 MMP1 1.280903 No

YFL045C SEC53 6 GPI17 1 2 0.447864 GPI17 0.691198 No

YFL049W SWP82 6 HRD1, MAC1, MMP1 3 3 0.709281 NMD4 1.276863 No

YFL047W RGD2 6 YHR140W 1 3 0.789164 SPS100 0.457247 No

YFL046W FMP32 6 GTT2, YLR363WA, ECM21, YHR140W, HRD1,

MAC1, GPI17 7 4 0.812081 No

4 YGL167C PMR1 7 YHR140W, MMP1 2 2 0.233099 MMP1 1.015743 Yes

YGL173C KEM1 7 YLR363WA, YHR140W 2 2 0.233099 NMD4 0.921069 Yes

YGL189C RPS26A 7 YLR363WA, HRD1 2 2 0.233099 GPI17 0.909881 No

YGL181W GTS1 7 HRD1, GPI17 2 2 0.233099 SPS100 0.598567 No

YGL190C CDC55 7 HRD1, MAC1 2 2 0.233099 SPS100 0.563913 No

YGL172W NUP49 7 GTT2 1 2 0.447864 MMP1 1.087802 No

YGL166W CUP2 7 YHR140W 1 2 0.447864 MAC1 0.945897 No

YGL178W MPT5 7 GTT2, YFR057W, ECM21, YHR140W, HRD1,

GPI17 6 3 0.593551 NMD4 1.005863 No

YGL175C SAE2 7 GTT2, YHR140W, HRD1, MAC1, GPI17 5 3 0.62809 HRD1 1.449395 No

YGL174W BUD13 7 GTT2, YLR363WA, ECM21, HRD1 4 3 0.668195 NMD4 1.078136 No

YGL170C SPO74 7 GTT2, ECM21, YHR140W, HRD1 4 3 0.668195 SPS100 0.911406 No

YGL171W ROK1 7 GTT2, YLR363WA, HRD1, MAC1 4 3 0.668195 NMD4 0.880077 No

YGL179C TOS3 7 GTT2, YHR140W, HRD1, MMP1 4 3 0.668195 GPI17 0.859914 No

YGL161C YIP5 7 YHR140W, HRD1, MMP1 3 3 0.709281 MMP1 1.280903 No

YGL192W IME4 7 GTT2, YLR363WA, HRD1 3 3 0.709281 SPS100 1.049073 No

YGL163C RAD54 7 GTT2, HRD1, MAC1 3 3 0.709281 MAC1 0.780333 No

YGL187C COX4 7 YLR363WA, HRD1, MAC1 3 3 0.709281 NMD4,

GPI17 0.310847 No

YGL162W SUT1 7 HRD1 1 3 0.789164 NMD4 1.090946 No

YGL164C YRB30 7 GTT2 1 3 0.789164 MMP1 0.853936 No

YGL180W ATG1 7 HRD1 1 3 0.789164 MMP1 0.702209 No

YGL168W HUR1 7 GPI17 0.981358 Yes

80

YGL184C STR3 7 GPI17 0.829456 No

YGL183C MND1 7 SPS100 0.767978 No

YGL186C TPN1 7 MMP1 0.747106 No

YGL169W SUA5 7 ECM21 0.616679 No

YGL191W COX13 7 NMD4,

GPI17 0.384219 No

5 YGL077C HNM1 7 YHR140W, HRD1, GPI17 3 2 0.133246 MMP1 1.745628 No

YGL065C ALG2 7 YHR140W, MMP1 2 2 0.233099 GPI17 1.651358 No

YGL092W NUP145 7 GTT2, HRD1 2 2 0.233099 MMP1 1.07616 No

YGL076C RPL7A 7 HRD1 1 2 0.447864 GPI17 1.655656 No

YGL084C GUP1 7 YHR140W 1 2 0.447864 GPI17 1.523756 No

YGL066W SGF73 7 HRD1 1 2 0.447864 NMD4 1.172365 No

YGL078C DBP3 7 HRD1 1 2 0.447864 ECM21 1.069253 No

YGL068W MNP1 7 MAC1 1 2 0.447864 GPI17 0.808031 No

YGL056C SDS23 7 YHR140W 1 2 0.447864 MAC1 0.736439 No

YGL075C MPS2 7 YHR140W 1 2 0.447864 ECM21 0.616679 No

YGL070C RPB9 7 GTT2, YLR363WA, ECM21, YHR140W, HRD1,

MAC1, MMP1, GPI17 8 3 0.538222 MAC1 0.873383 Yes

YGL093W SPC105 7 GTT2, YFR057W, ECM21, HRD1, MAC1, GPI17 6 3 0.593551 ECM21 1.069253 No

YGL095C VPS45 7 YLR363WA, YHR140W, HRD1, MAC1, MMP1,

GPI17 6 3 0.593551 MMP1 0.80663 No

YGL094C PAN2 7 GTT2, YLR363WA, ECM21, YHR140W, HRD1 5 3 0.62809 NMD4 1.630298 No

YGL063W PUS2 7 GTT2, YHR140W, HRD1, MMP1, GPI17 5 3 0.62809 GPI17 1.171309 No

YGL067W NPY1 7 GTT2, YLR363WA, YHR140W, HRD1, MMP1 5 3 0.62809 NMD4 0.938998 No

YGL062W PYC1 7 GTT2, YHR140W, HRD1, MAC1, GPI17 5 3 0.62809 NMD4 0.878746 No

YGL073W HSF1 7 GTT2, YFR057W, YHR140W, HRD1, MMP1 5 3 0.62809 MAC1 0.724681 No

YGL090W LIF1 7 GTT2, YFR057W, YHR140W, HRD1, MAC1 5 3 0.62809 MAC1 0.722988 No

YGL060W YBP2 7 GTT2, YHR140W, HRD1, MAC1 4 3 0.668195 SPS100 0.709658 No

YGL086W MAD1 7 GTT2, YFR057W, HRD1, MAC1 4 3 0.668195 SPS100 0.54715 No

YGL061C DUO1 7 GTT2, HRD1, MAC1, GPI17 4 3 0.668195 NMD4 0.464992 No

YGL058W RAD6 7 GTT2, ECM21, HRD1 3 3 0.709281 GPI17 1.070258 No

YGL071W AFT1 7 HRD1, MAC1, GPI17 3 3 0.709281 NMD4 0.943249 Yes

81

YGL087C MMS2 7 HRD1, GPI17 2 3 0.748404 GPI17 1.192587 No

YGL083W SCY1 7 YLR363WA, HRD1 2 3 0.748404 GPI17 0.995898 No

YGL096W TOS8 7 ECM21, HRD1 2 3 0.748404 NMD4 0.966169 No

YGL064C MRH4 7 GTT2, YLR363WA, YHR140W, HRD1, MAC1,

MMP1, GPI17 7 4 0.812081 GPI17 0.94601 No

YGL091C NBP35 7 GTT2, YHR140W, HRD1, MAC1, MMP1 5 4 0.81781 GTT2 0.804848 No

YGL089C MF(ALP

HA)2 7 YHR140W, HRD1, MMP1 3 4 0.823212 SPS100 0.822313 No

YGL080W FMP37 7 No

6 YHR188C GPI16 8 GPI17 1 1 0.041087 GPI17 5.82206 No

YHR200W RPN10 8 HRD1, GPI17 2 2 0.233099 HRD1 2.248239 No

YHR195W NVJ1 8 YHR140W 1 2 0.447864 GTT2 1.517664 No

YHR203C RPS4B 8 HRD1 1 2 0.447864 GPI17 1.347025 No

YHR196W UTP9 8 HRD1 1 2 0.447864 NMD4 0.911793 No

YHR199C FMP34 8 GTT2 1 2 0.447864 No

YHR197W RIX1 8 GTT2, YLR363WA, YFR057W, ECM21, YHR140W,

HRD1, MAC1 7 3 0.564904 ECM21 1.069253 No

YHR193C EGD2 8 GTT2, YLR363WA, YHR140W, HRD1, MAC1,

MMP1, GPI17 7 3 0.564904 MMP1 0.683241 No

YHR190W ERG9 8 YLR363WA, YHR140W, HRD1, MAC1, MMP1,

GPI17 6 3 0.593551 GPI17 3.364245 No

YHR206W SKN7 8 GTT2, YLR363WA, YHR140W, HRD1, MMP1,

GPI17 6 3 0.593551 NMD4 0.543401 No

YHR201C PPX1 8 GTT2, ECM21, YHR140W, HRD1, GPI17 5 3 0.62809 HRD1 1.555774 No

YHR186C KOG1 8 GTT2, YLR363WA, HRD1, MAC1, GPI17 5 3 0.62809 SPS100 0.556152 No

YHR187W IKI1 8 ECM21, YHR140W, HRD1, MMP1 4 3 0.668195 NMD4 0.959815 No

YHR204W MNL1 8 YLR363WA, HRD1, MAC1 3 3 0.709281 HRD1 1.191309 No

YHR198C FMP22 8 YHR140W, HRD1, GPI17 3 3 0.709281 No

YHR194W MDM31 8 GTT2, HRD1 2 3 0.748404 ECM21 0.767537 No

YHR191C CTF8 8 YLR363WA, HRD1 2 3 0.748404 SPS100 0.447795 No

YHR205W SCH9 8 HRD1 1 3 0.789164 SPS100 0.757975 No

YHR207C SET5 8 HRD1 1 3 0.789164 No

YHR189W PTH1 8 GPI17 1.655656 No

82

7 YIR039C YPS6 9 HRD1 0.932952 No

8 YJR117W STE24 10 YHR140W 1 1 0.041087 HRD1 1.082824 No

YJR132W NMD5 10 YHR140W, HRD1, MAC1, GPI17 4 2 0.079718 MMP1 1.09477 No

YJR118C ILM1 10 YHR140W, MMP1, GPI17 3 2 0.133246 ECM21 1.069253 Yes

YJR121W ATP2 10 GTT2, HRD1 2 2 0.233099 MMP1 0.942874 No

YJR127C RSF2 10 MAC1 1 2 0.447864 NMD4 1.449254 No

YJR123W RPS5 10 HRD1 1 2 0.447864 GPI17 1.158472 No

YJR125C ENT3 10 MAC1 1 2 0.447864 MMP1 0.908637 No

YJR130C STR2 10 HRD1 1 2 0.447864 GPI17 0.884485 No

YJR134C SGM1 10 GTT2 1 2 0.447864 No

YJR112W NNF1 10 GTT2, YFR057W, YHR140W, HRD1, MAC1, GPI17 6 3 0.593551 ECM21 0.422719 No

YJR113C RSM7 10 GTT2, YHR140W, HRD1, MAC1, GPI17 5 3 0.62809 GPI17 1.655656 No

YJR131W MNS1 10 GTT2, YHR140W, HRD1, MMP1, GPI17 5 3 0.62809 HRD1 1.496891 No

YJR133W XPT1 10 GTT2, YHR140W, HRD1, GPI17 4 3 0.668195 NMD4 0.807849 No

YJR135C MCM22 10 GTT2, YHR140W, MAC1, MMP1 4 3 0.668195 SPS100 0.57765 No

YJR119C JHD2 10 GTT2, HRD1 2 3 0.748404 GPI17 1.486577 No

YJR110W YMR1 10 MAC1 1 3 0.789164 GPI17 2.133196 No

YJR122W CAF17 10 HRD1 1 3 0.789164 GTT2 0.804848 No

YJR135WA TIM8 10 GTT2, YHR140W, HRD1, MMP1 4 4 0.820756 MMP1 1.280903 No

YJR126C VPS70 10 MMP1 1.057988 No

YJR137C ECM17 10 GPI17 0.594083 No

9 YKR059W TIF1 11 HRD1 1 1 0.041087 GPI17 1.655656 No

YKR053C YSR3 11 YHR140W, GPI17 2 2 0.233099 GPI17 2.119516 No

YKR067W GPT2 11 YHR140W 1 2 0.447864 GPI17 2.42734 No

YKR057W RPS21A 11 HRD1 1 2 0.447864 GPI17 1.655656 No

YKR068C BET3 11 HRD1 1 2 0.447864 MMP1 1.280903 No

YKR062W TFA2 11 HRD1 1 2 0.447864 NMD4 0.966763 No

YKR054C DYN1 11 MAC1 1 2 0.447864 ECM21 0.808784 No

YKR060W UTP30 11 GTT2, YLR363WA, YFR057W, YHR140W, HRD1,

MAC1 6 3 0.593551 ECM21 0.842966 No

YKR069W MET1 11 GTT2, YHR140W, HRD1, GPI17 4 3 0.668195 GPI17 0.936166 No

83

YKR056W TRM2 11 GTT2, MMP1, GPI17 3 3 0.709281 GPI17 0.969575 No

YKR066C CCP1 11 GTT2, HRD1, MAC1 3 3 0.709281 GTT2 0.937785 No

YKR061W KTR2 11 GTT2, HRD1 2 3 0.748404 GPI17 1.675974 No

YKR058W GLG1 11 YHR140W, HRD1 2 3 0.748404 GPI17 1.556205 No

YKR064W OAF3 11 GTT2, HRD1 2 3 0.748404 GPI17 1.1201 No

YKR071C DRE2 11 YHR140W, GPI17 2 3 0.748404 NMD4 0.542651 No

YKR063C LAS1 11 HRD1 1 3 0.789164 SPS100 0.655061 No

YKR055W RHO4 11 HRD1 1 3 0.789164 ECM21 0.615561 Yes

YKR065C PAM17 11 GTT2, YHR140W, HRD1, MMP1 4 4 0.820756 MMP1 1.280903 No

10 YLL022C HIF1* 12 GTT2, HRD1, MAC1 3 2 0.133246 NMD4 0.948465 No

YLL024C SSA2 12 HRD1 1 2 0.447864 GPI17 0.808054 No

YLL019C KNS1 12 GTT2, YLR363WA, YFR057W, YHR140W, HRD1,

MAC1, MMP1, GPI17 8 3 0.538222 GPI17 1.991796 No

YLL021W SPA2 12 GTT2, YLR363WA, ECM21, YHR140W, HRD1,

MAC1, MMP1, GPI17 8 3 0.538222 SPS100 1.114811 No

YLL018C DPS1 12 GTT2, ECM21, YHR140W, HRD1, MAC1, GPI17 6 3 0.593551 GPI17 1.635766 No

YLL018CA COX19 12 YHR140W, HRD1, MAC1, MMP1 4 4 0.820756 MMP1 0.750018 No

YLL025W PAU17 12 GTT2 0.188088 No

11 YLR061W RPL22A 12 HRD1 1 2 0.447864 GPI17 1.655656 No

YLR069C MEF1 12 HRD1 1 2 0.447864 GPI17 1.655656 No

YLR067C PET309 12 ECM21 1 2 0.447864 GPI17 1.007078 No

YLR071C RGR1 12 GTT2, YLR363WA, YFR057W, ECM21, HRD1,

GPI17 6 3 0.593551 MAC1 0.736281 No

YLR066W SPC3 12 YLR363WA, YHR140W, MMP1, GPI17 4 3 0.668195 GPI17 1.077951 No

YLR070C XYL2 12 YHR140W, HRD1, MAC1 3 4 0.823212 GPI17 0.771345 No

YLR068W FYV7 12 NMD4 1.238062 No

YLR062C BUD28 12 No

12 YLR106C MDN1 12 HRD1, MAC1 2 2 0.233099 NMD4 0.998364 No

YLR109W AHP1 12 GTT2, HRD1 2 2 0.233099 GTT2 0.817093 No

YLR103C CDC45 12 YFR057W 1 2 0.447864 NMD4 0.734434 No

YLR113W HOG1 12 YFR057W, ECM21, YHR140W, HRD1, MAC1,

MMP1 6 3 0.593551 MAC1 0.76413 No

84

YLR105C SEN2 12 GTT2, ECM21, HRD1, MAC1, MMP1 5 3 0.62809 NMD4 1.238062 No

YLR110C CCW12 12 YHR140W, MMP1, GPI17 3 3 0.709281 ECM21 0.810891 No

YLR107W REX3 12 NMD4 0.880077 No

13 YLR249W YEF3 12 HRD1 1 1 0.041087 GPI17 1.158472 No

YLR320W MMS22 12 GTT2, YLR363WA, HRD1, MAC1 4 2 0.079718 MAC1 0.57226 No

YLR277C YSH1 12 GTT2, ECM21, HRD1 3 2 0.133246 NMD4 1.63309 No

YLR291C GCD7 12 YHR140W, HRD1, MAC1 3 2 0.133246 GPI17 1.337797 No

YLR292C SEC72 12 YHR140W, HRD1, MMP1 3 2 0.133246 MMP1 1.280903 No

YLR314C CDC3 12 YHR140W, HRD1, MMP1 3 2 0.133246 SPS100 0.687233 No

YLR231C BNA5 12 GTT2, YFR057W 2 2 0.233099 NMD4 1.285388 No

YLR268W SEC22 12 YHR140W, GPI17 2 2 0.233099 MMP1 1.280903 Yes

YLR260W LCB5 12 YHR140W, GPI17 2 2 0.233099 GPI17 1.225683 No

YLR262C YPT6 12 YHR140W, HRD1 2 2 0.233099 MMP1 1.056649 No

YLR274W CDC46 12 GTT2, YFR057W 2 2 0.233099 NMD4 1.031722 No

YLR259C HSP60 12 YHR140W, HRD1 2 2 0.233099 GPI17 0.838886 No

YLR303W MET17 12 ECM21, HRD1 2 2 0.233099 GTT2 0.712703 No

YLR304C ACO1 12 GTT2, HRD1 2 2 0.233099 NMD4 0.667868 No

YLR310C CDC25 12 GTT2, HRD1 2 2 0.233099 SPS100 0.551016 No

YLR242C ARV1 12 GPI17 1 2 0.447864 GPI17 1.788748 Yes

YLR239C LIP2 12 MAC1 1 2 0.447864 GPI17 1.749277 No

YLR287CA RPS30A 12 HRD1 1 2 0.447864 GPI17 1.655656 No

YLR258W GSY2 12 HRD1 1 2 0.447864 GPI17 1.556205 No

YLR295C ATP14 12 GTT2 1 2 0.447864 MMP1 1.084933 No

YLR256W HAP1 12 HRD1 1 2 0.447864 NMD4 1.019619 No

YLR264W RPS28B 12 HRD1 1 2 0.447864 GPI17 0.909881 No

YLR245C CDD1 12 MAC1 1 2 0.447864 NMD4 0.889043 No

YLR293C GSP1 12 GTT2 1 2 0.447864 MMP1 0.800998 No

YLR288C MEC3 12 GTT2 1 2 0.447864 NMD4 0.746506 No

YLR247C IRC20 12 MAC1 1 2 0.447864 GTT2 0.188088 No

YLR276C DBP9 12 GTT2, YLR363WA, ECM21, YHR140W, HRD1,

MAC1, MMP1, GPI17 8 3 0.538222 ECM21 1.069253 No

85

YLR233C EST1 12 GTT2, YLR363WA, YFR057W, ECM21, YHR140W,

HRD1, MAC1, MMP1 8 3 0.538222 ECM21 0.767537 No

YLR305C STT4 12 GTT2, ECM21, YHR140W, HRD1, MAC1, MMP1,

GPI17 7 3 0.564904 GPI17 1.326477 No

YLR298C YHC1 12 GTT2, YLR363WA, ECM21, YHR140W, HRD1,

GPI17 6 3 0.593551 NMD4 1.912204 No

YLR240W VPS34 12 GTT2, YHR140W, HRD1, MAC1, MMP1, GPI17 6 3 0.593551 GPI17 1.459541 Yes

YLR237W THI7 12 GTT2, YFR057W, YHR140W, MAC1, MMP1,

GPI17 6 3 0.593551 MMP1 1.280903 No

YLR263W RED1 12 YLR363WA, ECM21, YHR140W, HRD1, MMP1,

GPI17 6 3 0.593551 SPS100 0.878792 No

YLR246W ERF2 12 GTT2, YHR140W, HRD1, MMP1, GPI17 5 3 0.62809 GPI17 2.481376 No

YLR248W RCK2 12 GTT2, YHR140W, HRD1, MAC1, MMP1 5 3 0.62809 GPI17 0.771845 No

YLR315W NKP2 12 GTT2, YLR363WA, YFR057W, HRD1, MAC1 5 3 0.62809 SPS100 0.436051 No

YLR275W SMD2 12 GTT2, YLR363WA, ECM21, HRD1 4 3 0.668195 NMD4 1.39614 No

YLR238W FAR10 12 YHR140W, HRD1, MAC1, MMP1 4 3 0.668195 SPS100 1.105314 No

YLR299W ECM38 12 GTT2, YHR140W, HRD1, GPI17 4 3 0.668195 GTT2 1.060218 No

YLR309C IMH1 12 GTT2, YHR140W, HRD1, MAC1 4 3 0.668195 MMP1 1.024723 No

YLR312WA MRPL15 12 GTT2, YLR363WA, YHR140W, HRD1 4 3 0.668195 GPI17 1.007078 No

YLR254C NDL1 12 YHR140W, HRD1, MMP1, GPI17 4 3 0.668195 MMP1 0.610824 No

YLR270W DCS1 12 YLR363WA, HRD1, MAC1 3 3 0.709281 NMD4 3.248047 No

YLR284C ECI1 12 YHR140W, HRD1, GPI17 3 3 0.709281 GPI17 2.44763 No

YLR300W EXG1 12 YHR140W, HRD1, GPI17 3 3 0.709281 ECM21 0.966093 No

YLR313C SPH1 12 GTT2, YHR140W, HRD1 3 3 0.709281 SPS100 0.839495 No

YLR272C YCS4 12 GTT2, YHR140W, HRD1 3 3 0.709281 ECM21 0.666965 No

YLR319C BUD6 12 GTT2, YHR140W, HRD1 3 3 0.709281 SPS100 0.663775 No

YLR267W BOP2 12 GTT2, YFR057W, HRD1 3 3 0.709281 No

YLR289W GUF1 12 GTT2, HRD1 2 3 0.748404 GPI17 1.655656 No

YLR250W SSP120 12 GTT2, YHR140W 2 3 0.748404 MMP1 1.263511 No

YLR266C PDR8 12 YFR057W, MAC1 2 3 0.748404 MAC1 1.021008 No

YLR306W UBC12 12 HRD1 1 3 0.789164 GPI17 1.660526 No

YLR273C PIG1 12 HRD1 1 3 0.789164 GPI17 1.092172 No

YLR244C MAP1 12 HRD1 1 3 0.789164 HRD1 0.729881 Yes

86

YLR265C NEJ1 12 MAC1 1 3 0.789164 MAC1 0.722988 No

YLR234W TOP3 12 YHR140W 1 3 0.789164 GPI17 0.690706 No

YLR312C QNQ1 12 HRD1 1 3 0.789164 No

YLR318W EST2 12 GTT2, YLR363WA, YFR057W, ECM21, YHR140W,

HRD1, MAC1, MMP1 8 4 0.808479 SPS100 0.578303 No

YLR316C TAD3 12 GTT2, YLR363WA, ECM21, HRD1 4 5 0.828941 GPI17 1.41484 No

YLR262CA TMA7 12 GPI17 1.655656 No

YLR285W NNT1 12 NMD4 1.122502 No

YLR307W CDA1 12 SPS100 1.098477 No

YLR308W CDA2 12 SPS100 1.098477 No

YLR286C CTS1 12 HRD1 0.911495 No

YLR251W SYM1 12 NMD4,

GPI17 0.384219 No

YLR261C VPS63 12 No

14 YLR411W CTR3 12 YHR140W 1 2 0.447864 MMP1 1.280903 No

YLR410W VIP1 12 HRD1 1 2 0.447864 GPI17 1.655656 No

15 YMR021C MAC1 13 MAC1 1 0 0.002783 MAC1 1.120227 No

YMR022W UBC7 13 HRD1 1 1 0.041087 HRD1 1.609072 No

YMR041C ARA2 13 HRD1 1 2 0.447864 GPI17 1.357859 No

YMR039C SUB1 13 ECM21 1 2 0.447864 MAC1 1.237248 No

YMR043W MCM1 13 MAC1 1 2 0.447864 MAC1 0.935413 No

YMR032W HOF1 13 YHR140W 1 2 0.447864 SPS100 0.314468 No

YMR024W MRPL3 13 GTT2, YLR363WA, ECM21, YHR140W, HRD1,

MAC1, MMP1, GPI17 8 3 0.538222 NMD4 1.238062 No

YMR033W ARP9 13 GTT2, YLR363WA, ECM21, YHR140W, HRD1,

MAC1, MMP1 7 3 0.564904 NMD4 1.045711 No

YMR029C FAR8 13 GTT2, YLR363WA, ECM21, YHR140W, HRD1,

MAC1 6 3 0.593551 SPS100 1.105314 No

YMR044W IOC4 13 GTT2, YHR140W, HRD1, MAC1, MMP1 5 3 0.62809 MAC1 0.995102 No

YMR019W STB4 13 YLR363WA, YHR140W, MAC1, MMP1, GPI17 5 3 0.62809 MAC1 0.713734 No

YMR036C MIH1 13 GTT2, YFR057W, ECM21, HRD1, MAC1 5 3 0.62809 SPS100 0.513358 No

YMR028W TAP42 13 GTT2, YHR140W, HRD1, MMP1, GPI17 5 3 0.62809 NMD4 0.473226 No

YMR017W SPO20 13 ECM21, YHR140W, MMP1 3 3 0.709281 SPS100 1.060747 No

87

YMR026C PEX12 13 GTT2, YHR140W, HRD1 3 3 0.709281 MMP1 1.001704 No

YMR025W CSI1 13 YLR363WA, YHR140W, HRD1 3 3 0.709281 GPI17 0.773335 No

YMR037C MSN2 13 YLR363WA, ECM21 2 3 0.748404 MAC1 0.602959 No

YMR023C MSS1 13 HRD1 1 3 0.789164 GPI17 1.13114 No

YMR020W FMS1 13 MAC1 1 3 0.789164 NMD4 1.107922 No

YMR042W ARG80 13 MAC1 1 3 0.789164 MAC1 1.028269 No

YMR038C CCS1 13 GTT2 1 3 0.789164 HRD1 0.578014 Yes

YMR030W RSF1 13 GTT2 1 4 0.827795 MAC1 0.801445 No

YMR035W IMP2 13 HRD1 1.775093 No

YMR040W YET2 13 MMP1 1.024723 No

16 YNL069C RPL16B 14 YLR363WA, HRD1 2 2 0.233099 GPI17 1.655656 No

YNL064C YDJ1 14 GTT2, HRD1 2 2 0.233099 GPI17 0.638493 No

YNL067W RPL9B 14 HRD1 1 2 0.447864 GPI17 1.655656 No

YNL073W MSK1 14 GTT2 1 2 0.447864 GPI17 1.282768 No

YNL070W TOM7 14 GPI17 1 2 0.447864 MMP1 1.132293 No

YNL071W LAT1 14 HRD1 1 2 0.447864 HRD1 1.070417 No

YNL088W TOP2 14 HRD1 1 2 0.447864 GPI17 0.981358 No

YNL076W MKS1 14 HRD1 1 2 0.447864 NMD4 0.974364 No

YNL084C END3 14 GTT2 1 2 0.447864 SPS100 0.7265 No

YNL087W TCB2 14 YHR140W 1 2 0.447864 No

YNL078W NIS1 14 GTT2, YLR363WA, ECM21, YHR140W, HRD1,

MAC1, MMP1, GPI17 8 3 0.538222 SPS100 0.681403 No

YNL085W MKT1 14 GTT2, YLR363WA, YHR140W, HRD1, MAC1,

MMP1, GPI17 7 3 0.564904 GPI17 0.659253 No

YNL075W IMP4 14 GTT2, YLR363WA, YHR140W, HRD1, MAC1,

MMP1 6 3 0.593551 ECM21 1.069253 No

YNL090W RHO2 14 GTT2, YHR140W, HRD1, MMP1 4 3 0.668195 ECM21 0.907218 No

YNL079C TPM1 14 GTT2, YLR363WA, HRD1, MAC1 4 3 0.668195 ECM21 0.635045 No

YNL091W NST1 14 GTT2, ECM21, YHR140W, HRD1 4 3 0.668195 GTT2 0.376175 No

YNL062C GCD10 14 GTT2, YHR140W, HRD1 3 3 0.709281 GPI17 1.505146 No

YNL068C FKH2 14 YLR363WA, HRD1, MAC1 3 3 0.709281 NMD4 0.948198 No

YNL063W MTQ1 14 GTT2, HRD1 2 3 0.748404 GPI17 1.823726 No

88

YNL066W SUN4 14 YLR363WA, HRD1 2 3 0.748404 ECM21 0.99192 No

YNL082W PMS1 14 HRD1 1 3 0.789164 MAC1 0.779775 No

YNL081C SWS2 14 GTT2, YLR363WA, YHR140W, HRD1, MAC1,

MMP1, GPI17 7 4 0.812081 SPS100 1.072889 No

YNL083W SAL1 14 YLR363WA, YHR140W, HRD1, MAC1, MMP1,

GPI17 6 4 0.814372 MMP1 1.280903 No

YNL065W AQR1 14 GTT2, YLR363WA, YHR140W, HRD1, MMP1,

GPI17 6 4 0.814372 MMP1 1.066535 No

YNL072W RNH201 14 GTT2, HRD1 2 4 0.825667 GPI17 1.23021 No

YNL080C EOS1 14 GPI17 1.34389 No

YNL074C MLF3 14 MAC1 0.553528 No

YNL077W APJ1 14 GPI17 0.409584 No

17 YPR124W CTR1 16 YHR140W, MMP1 2 2 0.233099 MMP1 0.941759 Yes

YPR129W SCD6 16 GPI17 1 2 0.447864 NMD4 1.437544 No

YPR132W RPS23B 16 HRD1 1 2 0.447864 GPI17 1.655656 No

YPR128C ANT1 16 YHR140W 1 2 0.447864 MMP1 0.92005 No

YPR135W CTF4 16 HRD1 1 2 0.447864 GPI17 0.651859 No

YPR137W RRP9 16 GTT2, YLR363WA, ECM21, YHR140W, HRD1,

MAC1 6 3 0.593551 GPI17 0.957792 No

YPR134W MSS18 16 GTT2, ECM21, HRD1 3 3 0.709281 NMD4 1.462776 No

YPR133C SPN1 16 ECM21, HRD1, MAC1 3 3 0.709281 NMD4 1.069704 No

YPR125W YLH47 16 YHR140W, HRD1 2 3 0.748404 GPI17 0.94601 No

YPR131C NAT3 16 YLR363WA, HRD1 2 3 0.748404 GPI17 0.933018 No

YPR133WA TOM5 16 YHR140W 1 3 0.789164 MMP1 1.057988 No

YPR122W AXL1 16 HRD1 1 3 0.789164 SPS100 0.812378 No

89

7. Validation of the Results Obtained from the Congenic Lines

As a proof of concept, we decided to concentrate on one of the identified regions in

chromosome XII. First, we determined its borders by using information obtained from

the tiling arrays and the Y98 arrays. Since different independent lines were hybridized

in each case, combining information from both was highly beneficial. Together with

sequencing the relevant region in one more independent line we were able to decrease

the size of the region from 124 kb (containing 54 candidate genes) to a region of ~12kb

carrying a single candidate gene. The small region contains the gene CTR3 and a Ty

element. CTR3 encodes a high-affinity copper transporter of the plasma membrane,

which is regulated by Mac1 (Pena, Puig et al. 2000). We sequenced the CTR3 ORF and

promoter region of the High MP and the Low MP parents and found no differences with

the exception of a transition at position 95 of the ORF, which led to a conservative

change between acidic amino acids (aspartic to glutamic acid). Notably, however, the

Low MP strain carries, ~100bp upstream of the CTR3 ORF, a Ty element (a yeast

retrotransposon), which is missing from the High MP strain (Figure 23A and 23B).

Analysis of several clones from the in-lab evolved lines C and A showed that they lack

the Ty element (originally present in the Low MP ancestor), which was excised by an

LTR-LTR recombination event (Kupiec and Petes 1988) during the selection procedure.

We used RT -PCR to measure the transcription level of CTR3 in both parents and found

that the mRNA level of CTR3 is 5 times higher in the High MP strain than in the Low

MP parental strain (Figure 24A). Deletion of the Ty in the Low MP strain led to a 2.5

fold increase in CTR3 transcription (Quantified using GELQUANT), but it is still lower

than the transcription level observed in the High MP strain (Figure 24A). These results

indicate that in addition to factors acting in cis, such as the presence or absence of the

Ty element, additional trans-acting factors play a role in the regulation of CTR3. A good

90

candidate for such a regulator is MAC1, identified in several of the ILE lines (Table 3)

and also found in one of the congenic regions (Table 6).

In order to confirm that the CTR3 locus affects growth at high pH, and to investigate a

possible regulatory role for the Ty insertion, we deleted the Ty element from the

upstream regulatory region in the Low MP strain. This led to a noticeable improvement

in its ability to grow at high pH, but, as expected, not to a complete high MP phenotype

(Figure 24B). Overexpression of CTR3 from a high copy number plasmid also improved

the ability to grow at high pH (Figure 24B).

Figure 23: Differences between the High MP and the Low MP in one of the

regions detected on Chr XII. A. Schematic maps of the CTR3 region in the Low MP

and the High MP parents. P1 and P2 represent the primers which were used to identify

the lack of the transposable element. B. PCR products of 7kb region containing the

transposable element and CTR3 ORF. The transposable element of 6kb is missing in

the High MP strain.

91

Figure 24: The expression level of CTR3 affects the MP phenotype. A. RT-PCR

showing the differences in transcription levels of CTR3 in different strains. ACT1 served

as a control. The products were quantified after 25 cycles using gelQuant and are

presented here as the ratio between the CTR3 products and the ACT1 products. B. The

ability of various strains to grow at regular or high pH media (ten-fold serial dilutions).

Overexpression of CTR3 at the Low MP strain BY4741 (Low MO/OE-CTR3) improves

significantly the MP phenotype.

92

Discussion

The fitness of an individual (its ability to survive and multiply) in any condition is

usually determined by complex traits that are influenced by many quantitative trait loci

(QTLs). However, to date the genetic basis of only a few complex traits has been

identified and many questions regarding the architecture of complex traits and the

accumulation of mutations during evolution still remain unanswered. Among them are:

How many QTLs affect complex phenotypes? What is the effect of each QTL? How do

complex traits change during the evolution? Is the adaptation process repeatable? etc.

In order to identify the QTLs that affect one of the important components of fitness

variability in yeast, and to answer some of the questions above, we combined in-lab

evolution (ILE) with the construction of Congenic lines to isolate and map several genes

sets that contribute the ability of yeast cells to survive under alkali stress.

We carried out an ILE experiment, in which we grew yeast populations under increasing

alkali stress to enrich for beneficial mutations. This process was followed by

hybridizations to tiling arrays (Gresham, Ruderfer et al. 2006) enabling the

identification of the mutations acquired during the laboratory selective process. The ILE

procedure revealed mutations in 15 genes. Identification of these mutations has defined

the QTLs and mechanisms that affect, in a quantitative fashion, the ability to cope with

alkali stress. We found that during ILE several populations had acquired different sets of

mutations that conferred the same phenotype. We identified each individual mutation in

these strains, and validated and estimated their contribution to the phenotype by three

independent methods (deletions, reciprocal hemizygosity and allele swapping). The total

additive effect of the QTLs was much larger than the difference between the ancestor

and the evolved strains, suggesting epistatic interactions between the QTLs.

93

In addition, we wanted to study the mechanisms regulating fitness variation at natural

habitats. In order to identify the QTLs that evolved in nature to confer alkali stress

resistance, we have also applied an association study using a clinically isolated wild

type yeast strain. To achieve high resolution mapping we have constructed Congenic

Lines up to the 8th

generation. As backcrosses are carried out till the eighth generation,

the portions of the contributing parent that do not add to the studied phenotype are

eliminated, producing strains with a cleaner genetic background. At the same time, the

size of the intervals carrying the candidate QTLs is reduced. With the Congenic Lines

we found 17 regions associated with the phenotype. Based on the results obtained from

the In-Lab Evolution, we applied a predictive algorithm on the 17 regions obtained from

the clinical isolates, and found 13 QTLs within those regions that their deletions affect

the trait.

Almost all beneficial mutations and candidate QTLs uncovered affected regulatory

genes, such as ubiquitin ligases, proteins involved in GPI anchoring, as well as

copper/iron sensing and transport factors, and not structural components of the pH

homeostasis machinery (such as proton pumps).

1. QTLs Dissection – Challenges and Achievements

Despite the tremendous success of genetics in the twentieth century, the mapping of

complex traits has lagged behind. This is mainly because of the difficult task of

mapping individual genes, which contribute little to the phenotype, and whose

individual inputs are often obscured by interactions with other genetic elements (Flint

and Mott 2001; Darvasi and Pisante-Shalom 2002; Abiola, Angel et al. 2003; Carlborg

and Haley 2004).

94

1.1. The Advantages of a Combined Strategy for QTLs Dissection

Extant strategies for QTLs dissection are not always designed to overcome the obstacles

involved in this mission. Therefore, there is no one complex trait for which all QTLs

affecting it have been found, and usually, the identified QTLs explain only a small

fraction of genetic variance. The fact that most of the genetic variance remains

unexplained is usually attributed to the fact that most methods are underpowered for

QTLs with smaller effects (Rockman and Kruglyak 2006). The most dominant methods

are based on association studies (Cardon and Bell 2001; Konig, Schafer et al. 2001).

Association studies search for regions of the genome with a higher-than-expected

number of shared alleles among individuals with a specific phenotype. These studies

generally identify broad genomic intervals that correlate with the phenotype, and these

intervals can encompass hundreds of candidate genes. Association studies are therefore

followed by fine mapping. Since fine mapping is usually complicated, only a few of the

most prominent QTL or, alternatively, QTLs containing candidate genes are analyzed at

a higher resolution (Darvasi 1998; Shalom and Darvasi 2002; Steinmetz, Sinha et al.

2002; Hirschhorn and Daly 2005; Ben-Ari, Zenvirth et al. 2006). Unfortunately, the

candidate gene approach can sometimes be misleading or mask some of the QTLs,

especially those with small effects, which may be responsible for most of the genetic

variance (Zeyl 2005). Therefore, association studies are limited and usually fail to

achieve fine mapping of all QTLs affecting one trait, and a more holistic approach or the

combination of several approaches is required.

To identify the QTLs conferring alkali stress resistance we used the combination of

congenic lines and In Lab Evolution (ILE). The congenic lines approach is essentially

an association study; however, due to the backcrosses, the resolution achieved is much

higher and the identified genomic intervals encompass a relatively small number of

95

genes. By definition, congenic lines enable the isolation of the whole genetic network

contributing to a certain phenotype in the tested strains. The stringent selection at each

generation maintains all QTLs, including those with small effects, or those that only

exert their influence when present in a particular combination of alleles. Together with

the ILE procedure and with the predictive algorithm we have applied, we achieved high

resolution fine mapping and were able to pinpoint the genes and the mutations affecting

the trait studied. Thanks to the combination of these two methods the QTLs were

identified regardless to their effect so we could also detect QTLs with minor effects.

1.2. Fine Mapping and Proof of Causation

Two additional major challenges in dissecting complex traits are the QTLs fine mapping

and the proof of causation (Darvasi and Pisante-Shalom 2002; Page, George et al.

2003). Using ILE we have identified the exact positions of each causative mutation. An

additional stage of fine mapping was not required. We obtained proof of causation and

estimated the effect of each mutation by three alternative methods:

i) We first deleted the candidate QTLs in the ancestor background and in the evolved

strain background. This enabled an initial indication of whether a certain QTL affects

the phenotype and gave some nice insights regarding the mutations found. For example,

ECM21 had acquired a nonsense mutation during the ILE. In accordance with that, the

deletion of the wt allele in the ancestor background conferred increased resistance to

alkali pH, while the deletion of the mutated allele (carrying the nonsense mutation) in

the C8.6 background had no effect.

ii) The next step we used to proof causation is reciprocal hemizygosity. This assay has

many advantages over the deletion assay. In reciprocal hemizygosity assay two isogenic

strains are compared. Each strain carries only one allele of the tested gene, enabling to

estimate the effect of each allele on the phenotype. If the mutation is beneficial, we

96

expect that the strain that carries the mutated allele will have a higher fitness than the

strain that carries the wild type allele.

iii) Finally, we wanted to estimate the contribution of each mutation to the phenotype.

This was done by swapping the wild type allele with the mutant one in the ancestor

background. We can then compare between the ability of two isogenic strains differing

only by a single SNP to grow at high pH. This method is highly efficient for estimating

the effect of each allele; however, since each allele is tested separately, the contributions

of any interactions between alleles are not measured.

The three methods enable us to validate and estimate the effect of each mutation that

appeared during ILE. The effect of some mutations, such as those in MAC1, was

detectable by all methods while others were detectable only by some of the methods.

For the Congenic Lines a stage of fine mapping was required. The Congenic Lines

provided 17 small genomic intervals inherited from the high MP parent, which are

candidates to contain QTLs. On average, each interval is 42 kb long and contains 17.6

genes. Here, a fine mapping stage was needed to pinpoint one candidate gene in each

interval. We therefore used the information obtained from the ILE. MAC1 was the only

gene identified by the ILE procedure present in one of the Congenic intervals. In order

to systematically predict the QTLs in each genomic interval identified, we combined

information from protein-protein interactions network (Breitkreutz, Stark et al. 2008;

Yu, Braun et al. 2008; Matthews, Gopinath et al. 2009) and GO annotations (Lord,

Stevens et al. 2003) to find similarity to the genes found by ILE. This allowed us to

generate ranked probabilities of being a relevant QTL (Table 6). For each interval, we

predicted the genes most likely to affect the MP phenotype in the high MP strain. Five

out of the predicted genes (MAC1, CTR1, FEN1, KEM1 and PMR1) have been

previously identified in a systematic genome-wide screen as required for growth at high

97

pH (Serrano, Bernal et al. 2004). That was a first indication that our algorithm can

predict the correct QTLs. However, a statistical proof was needed. Therefore, we picked

the two best predicted QTLs and two worse predictions from each interval and tested the

effect of their deletions on the trait. As explained in the Introduction, testing deletions

can be sometime misleading. However, here we just wanted to compare between two

groups of deletions to see if there is a difference between them that can either validate

the algorithm. Twelve of the 29 deletions with the best scores tested decreased the

ability to grow at high pH (Figure 22) while only four out of the 28 lowest-ranking

candidates affected the phenotype (p-value= 0.023). Since the algorithm took into

consideration the similarity and the interactions with genes found by ILE, these results

shows that the QTLs developed in nature have common mechanisms and functions with

the ones that mutated during ILE. The differences between the QTLs could be attributed

to the differences between laboratory conditions and the natural conditions. Yeast cells

growing in the wild must adapt to multiple and sequential stresses, a response that was

shown to be more complex (Tagkopoulos, Liu et al. 2008; Mitchell, Romano et al.

2009). However, even in a controlled environment under the same conditions, different

QTLs evolved during the ILE, reinforcing the 'multiple mutations' theory for adaptive

evolution (see below) (Desai, Fisher et al. 2007; Kao and Sherlock 2008). We have

identified several sets of QTLs determining the same phenotype. We conclude that

different sets of genes can evolve to confer similar phenotypes, although some of the

mechanisms are shared by all strains, indicating their importance in surviving alkali

stress.

1.3. Revealing Genetic Interactions Among Different Loci

When studying complex traits, most studies tacitly embrace an additive model, in which

each QTL has a constant, contributing effect, regardless of the presence or absence of

98

modifier alleles (Carlborg and Haley 2004). However, epistatic interactions were

shown to have a significant effect on complex traits (Eshed and Zamir 1996; Brem,

Storey et al. 2005; Sinha, Nicholson et al. 2006). Epistasis has been defined as a

deviation from the expected phenotype under the additive model, that is, the sum of the

independent effects of the individual genes (Fisher 1918; Cordell 2002). The existence

of complex genetic interactions between polymorphic QTLs can dramatically affect the

ability of genetic studies to detect QTLs (Fijneman, de Vries et al. 1996; Flint and Mott

2001; Carlborg and Haley 2004) . An allele in one locus can mask or alter the effect of

several other QTLs (Carlborg, Jacobsson et al. 2006). On the other hand, the presence of

two mutations may produce a phenotype that is stronger from the one predicted by

simple additive models. Therefore, the genetic background is a crucial factor and it is

better to capture all QTLs with a whole genome strategy, rather than focusing on one

QTL at a time. Both ILE and congenic lines result in strains carrying a genetic network

that determines the ability to grow at high pH. The use of whole genotyping strategies

such as tiling arrays to genotype these strains enables the identification of all QTLs

network as a whole, including interacting genes. Indeed, once the QTLs have been

identified, we have estimated the independent effect of each QTL and were able to infer

epistatic interactions among them. The sum of the independent effects of the QTLs was

much higher than the effect of all QTLs together at the high MP strain (calculated as the

difference between the relative fitness of the high MP and the low MP strain. Figure

20). For example MAC1 under the examined conditions explained ~80% of the

difference and ECM21 explained 40% of the difference. Even without taking into

consideration all other genes the sum of the effects of these two genes is higher than

100%. These results show that a simple additive model alone is insufficient to explain

the phenotype, as the sum of the individual effects measured is greater than the effect

99

observed in the selected High MP strain. Our results thus suggest that epistatic effects

among the various QTLs dampen the individual contribution measured as a single QTL.

Similarly, several of the mutations tested that failed to show a significant effect on the

MP phenotype may do so in the appropriate genetic background, i.e., when combined

with the appropriate QTLs. Furthermore, some genes were found to have different

effects in different backgrounds. For example, HRD1, IES2 and SGT2 were found by

allele swapping in a haploid background to affect the phenotype (Figure 20B) but no

effect was observed using the reciprocal hemizygosity assay in a hemizygote

background (Figure 18C). In light of the complex interactions shown here, a

combinatorial approach is required to study the way in which the identified genes

interact. For that purpose I have already created a combinatorial collection of mutants.

Future plans include using this collection together with expression data obtained from

some of these strains at high and optimal pH, in order to analyze the hierarchies of

regulation and the genetic relations between the mutations found.

100

2. Adaptation to Alkali Stress

Extracellular pH is a key environmental signal that influences growth, physiology, and

differentiation (Penalva, Tilburn et al. 2008). Yeast cells grow more rapidly in acidic

media than in neutral or alkaline media. In an acidic environment the plasma membrane

H+-ATPase hydrolyzes ATP to pump protons out of the cell. The proton gradient is used

for transport of amino acids, nucleotide bases, phosphate, and many other molecules in

symport reactions (van der Rest, Kamminga et al. 1995). One effect of external

alkalinity is the disruption of the membrane proton gradients (van der Rest, Kamminga

et al. 1995). Thus, alkali environment is a stressful perturbation which cells must

monitor, respond and adapt to using complex mechanisms, such as induction of

alternative ion pumps (Nelson and Nelson 1990; Munn and Riezman 1994; Stevens and

Forgac 1997). The major regulator responsible for gene expression in alkaline

environments and for pH homeostasis is RIM101 (Li and Mitchell 1997; Treton,

Blanchin-Roland et al. 2000), which, among other effects, promotes expression of the

genes encoding the ion pump. Our results discovered different sets of QTLs affecting

resistance to alkali stress. Inspection of these QTLs reveals a striking absence of

structural genes directly in charge of the cellular machinery responsible for pH

homeostasis, such as proton pumps (van der Rest, Kamminga et al. 1995; Forgac 1998;

MacLeod, Vasilyeva et al. 1998) and the RIM101 pathway. Instead, the majority of the

mutations identified affect global regulators, such as ubiquitin ligases, proteins involved

in cell wall maintenance and copper sensing/ transport factors. Such mutations affect the

phenotype in a global fashion, rather than specifically affecting the target cellular

machinery.

101

2.1. The Role of Metal Transporters in Alkali Stress Resistance

One effect of external alkalinity was suggested to be nutrient and ion limitation that

arises from disruption of the plasma membrane proton gradient (Lamb, Xu et al. 2001).

Copper and iron ions were shown to affect tolerance to high pH (Serrano, Bernal et al.

2004) and several metal transport genes are up-regulated under alkali stress (Lamb, Xu

et al. 2001). Four of five separate selection lines independently acquired different

mutations at the same position (Cysteine 271) of the copper-sensing transcription factor,

MAC1. Mac1p is a transcriptional activator of copper-responding genes, such as CTR1

and CTR3 (Jungmann, Reins et al. 1993), which were also found by the Congenic Lines

strategy. Copper binds the trans-activating Cysteine-rich domain (residues 264–279) of

Mac1p and inhibits its activity [Figure 25 (Jensen and Winge 1998)] . Mutations at

Cys271 are therefore likely to abrogate copper-binding, thus elevating transcription in

the Mac1 gene cohort.

We have shown that changes in the expression level of CTR3 can alter fitness at high

pH. CTR3 was shown by Northern blot to have a slightly higher expression under alkali

stress (Lamb, Xu et al. 2001). Yet, we were unable to detect that difference by RT-PCR

(maybe due to the low sensitivity of this method). However, we observed a significant

difference in the expression level of CTR3 between the clinical isolate GRA2 and the

laboratory strain BY4741. In the wild type strain GRA2 (and in several of the ILE

lines), the high expression of CTR3 was caused by excision of a transposon (Knight,

Labbe et al. 1996), underscoring the importance of mobile elements in evolution

(Dunham, Badrane et al. 2002; Gresham, Desai et al. 2008) and gene expression, as

suggested by McClintock (McClintock 1984).

102

2.2. Genes Encoding Cell Wall Proteins Affect Growth at High pH

A second group of regulatory genes affected cell wall maintenance. The fungal cell wall

is essential for maintaining the osmotic balance of the cell, for creating and maintaining

the shape of the cell, and for morphogenesis. A transcriptional analysis (Viladevall,

Serrano et al. 2004) pointed out to an overlap between the transcriptional profile of yeast

cells subjected to severe alkaline stress and that observed after mutation of

genes

encoding certain cell wall components or after exposure to cell wall-damaging agents

(Lagorce, Hauser et al. 2003). Consequently, alkaline pH leads, at least to some extent,

to cell wall damage (Serrano, Martin et al. 2006). Many cell surface proteins are

anchored to the plasma membrane via a glycosylphosphatidylinositol (GPI) residue. GPI

anchoring is essential for the expression of those proteins on the cell surface

(McConville and Ferguson 1993). GPI anchoring is also essential for the correct

Figure 25 [adapted from (Jensen and Winge 1998)]: Model of the Copper-induced

intramolecular interaction. DNA-binding and trans-activation domains are shown by

hatching and shaded boxes, respectively. The dots represent cysteine residues. Mac1p is

the transcription factor of copper transporters such as CTR1 and CTR3. Copper binds the

cycteines at the trans-activation domain to inactivate Mac1p. Mutations in these residues

prevent copper binding resulting in constitutive activation of Mac1 that leads to elevated

expression of, its target genes.

103

localization of many proteins acting in the assembly and remodeling of fungal cell wall

as a response to environmental pH (Ohishi, Inoue et al. 2001). Our results revealed

several GPI-anchored and -anchoring genes. GPI17, which was identified in the in-lab

evolution lines, as well as FEN2, YPS6 and GPI16 that were found in the congenic lines,

play important roles in this process. Moreover, both ECM21 and YHR140w show

physical and genetic interactions with members of the GPI addition pathway (Tong,

Lesage et al. 2004; Miller, Lo et al. 2005). In addition, transcription of some GPI-

anchored genes such as ZPS1, FIT2 and FIT3 is elevated under alkali stress (Lamb, Xu

et al. 2001). Taken together, these results underscore the important role played by GPI-

anchoring proteins in adaptation to high pH.

2.3. The role of Ubiquitin Ligases in Alkali Stress Resistance

A third group of QTLs was related to ubiquitin metabolism. They either encode

ubiquitin ligases [HRD1, CDC23 (Scrimale, Didone et al. 2009)], or adaptors/regulators

of ubiquitin ligases [ECM21/ART2 (Lin, MacGurn et al. 2008), RRI2 (Wee, Hetfeld et

al. 2002)]. CDC23 is an essential gene that encodes a subunit of the ubiquitin ligase

complex required for degradation of anaphase inhibitors (Zachariae and Nasmyth 1999).

A missense mutation occurred during ILE in CDC23. We were unable to see an effect in

our reciprocal hemizygozity assay (Figure 18B). However, since CDC23 is an essential

gene, we were unable to perform additional tests, such as allele swapping. We are thus

unable to establish definitely whether it affects the phenotype selected, or attained

fixation within the population due to genetic drift. HRD1 is a central member of the

ubiquitin ligase complex responsible for recognizing and ubiquitinating misfolded

proteins in the ER for degradation by the proteasome (Gauss, Jarosch et al. 2006). The

allele swapping assay of HRD1 showed that it has a major effect on alkali stress

resistance (Figure 19B). However, these results were inconsistent with the results

104

obtained from the reciprocal hemizygosity assay (Figure 18B) or with the deletion of

HRD1 that failed to show an effect on the phenotype (Figure 16A,B), therefore we are

repeating the allele swapping experiment for this gene.

We have not found previous evidence for an involvement of ubiquitin ligases in alkaline

stress resistance. Neither the systematic screening of the deletion library for mutants that

are sensitive to alkaline stress (Serrano, Bernal et al. 2004) nor the gene expression

surveys (Lamb, Xu et al. 2001) have found evidence for the involvement of ubiquitin

metabolism in alkaline stress response. However, it has been shown that some ubiquitin

ligases play an important role in general stress resistance. As a result from

environmental stresses many proteins are damaged or misfolded. Therefore, many

chaperones that protect native proteins from damage or refold defective polypeptides are

induced in response to stress in an attempt to restore the protein's native conformation

(Gasch, Spellman et al. 2000; Causton, Ren et al. 2001). When repair or refold is

impossible, damaged proteins must be removed from the cell. For this purpose,

ubiquitin–protein ligases label terminally misfolded polypeptides with a polyubiquitin

chain for proteasomal destruction (Hershko, Heller et al. 1983; Hershko and

Ciechanover 1998). In addition, the ubiquitination machinery is necessary for the

correct induction of the stress response SPI1 gene whose deletion causes

hypersensitivity to various stresses, such as heat shock, alkaline stress and oxidative

stress (Cardona, Aranda et al. 2009). Thus, we assume that the mutations found in

ubiquitin ligases or in ubiquitin-related genes are beneficial for resistance to general

environmental stress and not specific to resistance to alkali stress. In order to confirm

this theory it is possible to test whether the mutations in the ubiquitin-related and

chaperone-encoding genes provide fitness advantage under different stressful conditions

such as high osmolarity, high temperature or nutrient limitations. Since strains carrying

105

each single mutation were already constructed by allele swapping, testing their fitness

under several conditions is called for.

2.4. Additional Genes Affecting Alkali Stress Resistance

Most of the genes found have no known direct effect on alkali stress resistance. Their

potential effect on alkali stress may thus take place via additional global mechanisms.

For example, a missense mutation in the first methionine of SGT2 was identified in the

ILE strains. SGT2 has a similarity to human SGT, which is a co-chaperone that

negatively regulates Hsp70 (Angeletti, Walker et al. 2002). The mutation in the first

methionine of the ORF may reduce or abrogate its translation, resulting in up-regulation

of Hsp70.

An additional interesting example is ECM21/ART2. A nonsense mutation in

ECM21/ART2 was acquired during ILE by clone C8.6 (Table 3). In agreement, deletion

of the gene from the Low MP background increased fitness at high pH, whereas the

same deletion did not improve the phenotype of the C8.6 strain (Figures 16A, B).

ECM21/ART2 is part of the ART family (Lin, MacGurn et al. 2008). ARTs and the E3

ubiquitin ligase Rsp5 are recruited to the plasma membrane in response to

environmental stimuli that trigger the endocytosis of proteins such as permeases and

transporters. Several ART proteins interact with Rsp5 at a specific domain (Lin,

MacGurn et al. 2008) and possibly compete for binding. Thus, in the absence of one

ART protein, other ART proteins may work more efficiently. One member of the ART

family is RIM8/ART9 which is required for the alkaline pH signaling response (Penalva,

Tilburn et al. 2008). RIM8/ART9 was proposed to regulate the pH sensor membrane

protein Rim21 (Lin, MacGurn et al. 2008). It is possible that in the absence of

ECM21/ART2, the pathway of RIM8/ART9 acts more efficiently.

106

3. Adaptation - A Lesson from QTLs Dissection

Many general questions regarding the architecture and the evolution of complex traits

are beginning to be unraveled lately. Among them are; What is the number of loci that

underlie variation in heritable phenotypes (Brem and Kruglyak 2005)? What is the size

effect of each QTL and what is their molecular nature and mechanisms of action? How

do QTLs interact (St Onge, Mani et al. 2007; Shachar, Ungar et al. 2008)? How do

mutations evolve and accumulate (Fisher 1930; Orr 2003)? What types of genomic

variation are associated with adaptation and how repeatable is the process (Gresham,

Desai et al. 2008)? By dissecting and characterizing complex trait, our results shed light

on some of these questions.

3.1. Mutations in Regulatory Genes Shape the Architecture of Complex Traits

Inspection of the QTLs affecting resistance to high pH reveals a striking absence of

structural genes directly in charge of the cellular machinery responsible for pH

homeostasis, such as proton pumps. Instead, the majority of the mutations identified

affect global regulators, such as ubiquitin ligases, proteins involved in cell wall

maintenance and copper sensing/ transport factors. Mutations in genes with wide

regulatory functions seem therefore to be preferred to mutations with narrow functions

(Gerke, Lorenz et al. 2009; Romano, Gurvich et al. 2010) affecting solely the trait

selected. At the moment, we have only a few examples to base our conclusions, but, our

results may reflect the inherent architecture of the genomes, which places genes with

related functions under common regulatory circuits. Mutations in key regulators can

thus affect several genes at once, conferring an immediate effect on fitness. Such an

arrangement may be advantageous in the long run, despite the possibility of mutations

that may inactivate the whole network.

107

3.2. The Number of QTLs Affecting a Trait and Their Size Effect

Two basic questions about the architecture of complex traits are: how many QTLs

underlie a genetic variation of a quantitative trait, and how much of the heritable

variation in the trait does each QTL explain. Disagreement regarding these issues begun

with the inherent contradiction between basic Mendelism that suggests that each

phenotype is determined by a single gene, and the argument of Darwin that adaptation

must take place by many successive small steps. The general consensus today is that

several traits are mono-alleleic while the majority of phenotypes in nature are

determined by more than one gene. However, for complex traits general principles

regarding the distribution of QTL numbers, effect sizes, and combined effects of

multiple QTLs remain to be elucidated. Some studies support the prevailing view that

complex traits result from many mutations, each with small phenotypic effects (Fisher

1930; Barton and Turelli 1989; Brem and Kruglyak 2005). Others claim that most

phenotypes result from a few QTLs with relatively large effects (Orr and Coyne 1992;

Keightley 1998). The answer for those questions is not trivial since, as described in the

introduction, most association studies are biased to detect only QTLs with the largest

effect on the trait. Therefore, the observed number of loci is usually an extreme

underestimate of the actual number, and the observed effect sizes represent the high end

of the overall distribution of effect sizes.

Since our method combined results from In Lab Evolution (ILE) with an association

study they give us an opportunity to examine these issues. By crossing a High MP

clinical isolate strain to the laboratory Low MP strain we showed that at least six QTLs

contribute the High MP phenotype in that strain. The In Lab Evolution method reveals 3

different sets of QTLs contributing to the MP phenotype. In clone A8.5 three candidate

108

QTLs were found, in C8.0 eight and in C8.6 seven candidate QTLs were found. The

average is therefore of approximately six QTLs in each strain, matching previous

studies that characterized QTLs underlying inheritance of thousands of transcript levels

in a cross between two strains of Saccharomyces cerevisiae (Brem, Yvert et al. 2002;

Schadt, Monks et al. 2003; Yvert, Brem et al. 2003; Morley, Molony et al. 2004).

However, it is important to notice that different experimental designs will give different

answers to the question of how many QTLs determine the genetic variability. The

number of QTLs that can be detected is depending on the resolution power of each

experiment.

The abundance of a transcript, as measured with DNA microarrays, can be treated as

quantitative traits, allowing thousands of such traits to be studied simultaneously

[expression QTLs or eQTLs (Brem, Yvert et al. 2002; Schadt, Monks et al. 2003; Yvert,

Brem et al. 2003; Morley, Molony et al. 2004)]. These studies have demonstrated that

the levels of many transcripts vary among genetically diverse individuals in a species,

and they used linkage mapping to identify hundreds of QTLs that underlie this variation.

These studies suggest that the genetic architecture of most expression traits involves

multiple QTLs. The median of the variation explained by most QTLs (the size effect of

each eQTL) was estimated to be 27% (Brem and Kruglyak 2005; Rockman and

Kruglyak 2006). In another study yeast populations were subjected to ~200 generations

in chemostats under different metabolic limitations. After 200 generations the mutations

were detected by different methodologies. Most of these mutations appear to be

adaptive and provide a fitness advantage of 5–10% (Gresham, Desai et al. 2008). Our

results pinpoint an average of 6 QTLs meaning that under the simple additive model, the

average size effect is expected to be ~17%. One explanation for this difference can be

attributed to the difference in the experimental methods. Chemostats provide a

109

consistent environment across time and replicates, and the nutrient-limited conditions

presumably represent a constant selective pressure. Under these conditions we expect

adaptive mutations to appear early, and additional mutations appearing later are likely to

add only minimally to the general fitness (Barton 1998; Orr 1999). In contrast, during

the ILE experiment, we kept increasing the pH, thus applying stronger and stronger

selective forces. We expect that under these conditions beneficial mutations with a

stronger effect will be selected for. In addition, it is important to emphasize that the

causation of eQTLs was not experimentally proven and the effect of mutations found

under metabolic limitation was estimated only by fixation rate. Identifying the mutations

does not necessarily indicate their contribution to the trait. This contribution was

estimated based on the number of QTLs found and statistical modeling. Only by

empirical experiment such as a reciprocal hemizygosity and allele swapping can a

precise validation be obtained. Moreover, as explained before, our results indicate

epistatic interactions among various QTLs, as the sum of the individual effects

measured is greater than the effect observed in the selected High MP strain. Epistatic

interactions can mask or alter the estimated effect of each QTL in all experiments and

should be taken under consideration.

3.3. Dynamics of Adaptation

Many theories have been raised to explain how adaptive mutations accumulate during

adaptation. In their simplest version, beneficial mutations arise spontaneously in any

cell population. If they are adaptive, selection increases the frequency of individuals

carrying the adaptive mutation until all the population is composed of descendants from

the original mutant. This process of replacing the former wild type by an adaptive

mutation is called fixation. Support for this theory has come from studies of Escherichia

coli populations in which the data are consistent with the fixation of one adaptive

110

mutation before the appearance of subsequent ones (Atwood, Schneider et al. 1951).

Therefore, a complex phenotype can be developed by successively appearing mutations.

Each mutation contributes to the final phenotype. However, other theories suggest more

complex dynamics in which fixation of an adaptive mutation is dependent on multiple

factors such as mutation rate and population size (Helling, Vargas et al. 1987;

Rosenzweig, Sharp et al. 1994; Notley-McRobb and Ferenci 1999; Notley-McRobb and

Ferenci 2000). If mutations are not very rare, several mutations can occur in the

population at the same time. In asexual populations, where mutations cannot be

combined by mating, they will compete until the best fitted mutation attains fixation

[reviewed by (Zeyl 2007)]. According to this model, named clonal interference, one

mutation is fixed and the rest are lost (Kim and Orr 2005). Here too, once a mutation

attained fixation an additional beneficial mutation can occur resulting in a complex

phenotype.

The ILE experiment provides an opportunity to observe the evolutionary dynamics of

adaptation in real time. During ILE different sets of genes exhibited mutations in

different clones (Figure 12). The mutations found in a clone from C8.0 [which

represents an intermediate stage in the evolution of line C], only partially overlap those

of clones from the C8.6 population. Only two out of eight mutations found in clone C8.0

attained fixation and existed in the clone from C8.6. Moreover, crosses between

different clones from the same pool at the end point of the selection revealed that

different beneficial mutations exist in different clones isolated from the same pool

(Figure 12B). These observations demonstrate how several beneficial mutations can

exist at the same time and still be competing while other mutations already attained

fixation.

111

However, the possibility to estimate the effect of each mutation offers a new perspective

for the dynamic of adaptation. Some of our results do not fit the basic clonal

interference model but reinforce more complex models (Desai, Fisher et al. 2007; Kao

and Sherlock 2008).

For example the mutation in HRD1 had a major effect on the phenotype (Figure 20B).

Nevertheless, this mutation did not attain fixation and could not be found at clone C8.6.

This characterizes a situation in which, while two mutations are still competing, an

additional mutation can occur in one of the clones. The additional mutation will give an

advantage to this clone (Desai, Fisher et al. 2007). In this 'multiple mutations' model a

previously inferior clone that was headed for extinction can suddenly become the

favorite. The superior previous mutation will thus not attain fixation. The clone with

multiple mutations can be overtaken only by another multiple mutant (Desai, Fisher et

al. 2007). In this way, the fittest mutation will not always attain fixation and the

competition between clones will continue, resulting in a population with several

different clones, each containing different sets of beneficial mutations contributing to

one phenotype. In our case, it is possible that the mutation in HRD1 with the fitness

advantage Fhrd1 appeared in a population in which another lineage carrying the mutation

in gene X with fitness advantage Fx. Since Fhrd1>Fx, we expect that X will be eliminated

while HRD1 will attain fixation. Yet, if another mutation (Y) occurs in the lineage

containing the mutation in X and Fhrd1< Fx+ Fy then HRD1 will be eliminated while X

and Y will prevail.

The fact that different sets of genes have been found in different ILE clone address

additional question regarding the consistencies of adaptation in the formation of

complex traits.

112

3.4. Is the Adaptation Process Repeatable?

The vast number of genome-wide studies trying to understand the architecture of

quantitative variation offers an opportunity to address some old questions at a higher

level. One of these questions is whether adaptation process is repeatable or whether

during adaptation different genetic sets evolve to confer the same phenotype. Some

studies analyzing similar phenotypes have been published during the last years.

However, groups analyzing the same phenotype usually find little overlap between

QTLs. One example is the three main whole genome studies that identified QTLs

influencing height in human (Gudbjartsson, Walters et al. 2008; Lettre, Jackson et al.

2008; Weedon, Lango et al. 2008). These studies used genome-wide association data

from tens of thousands of individuals to identify loci associated with height. Together

they identified more than 50 new Loci. However, only three of them (ZBTB38, HIPP

and CDK6) were found in all studies. The small overlap can be attributed to the fact that

the loci discovered explain only ~5% of the total phenotypic variation in height. The

heritability of this trait is estimated at 80%-90%; thus, each set of loci that was found in

each study can explain a different part of the variability seen in nature. However, for the

next example this simple explanation does not fit. Three studies attempted to identify

QTLs controlling sporulation efficiency in budding yeast (Deutschbauer and Davis

2005; Ben-Ari, Zenvirth et al. 2006; Gerke, Lorenz et al. 2009). Each of them found

several QTLs and showed that they explain the majority of the differences in sporulation

efficiency. However, only one gene (RME1) was found in two (Deutschbauer and Davis

2005; Gerke, Lorenz et al. 2009) out the three studies. All other QTLs were unique to

each specific study. The small overlap is not necessarily a contradiction, but might

suggest that the variance in sporulation efficiency is due to the existence of several

alternative sets of genes, each conferring a similar phenotype. In our study we show that

113

at least three independent genetic networks have been developed during In Lab

Evolution. These results imply that even under a uniform selection regime in the same

controlled environment different combinations of mutations can appear, leading to

similar phenotypes. An additional genetic network developed in nature and was

identified using the congenic lines constructed by the clinical isolated strain. However,

some of the basic mechanisms appeared in all different sets. For example, high

expression of CTR3 is required for growing at high pH. In the clinical isolate CTR3 is

highly expressed due to the absence of a repressing transposable element. This

transposon exists in the laboratory strain used for the ILE. Therefore an alternative

mechanism to highly express CTR3 was required in strains carrying the repressing

transposon. Indeed, four out of five ILE strains tested acquired mutations in MAC1

which result in high expression of CTR3.

Our work provides insights on both evolutionary and genetic issues. We discuss the

dynamics of adaptation, characterize the architecture of complex traits and the

appearance of adaptive mutations. At the same time we uncovered information about the

mechanisms that contribute to growth at high pH, a subject with ramifications for cell

physiology, pathogenicity, and stress response. This approach can be extended to other

genetic traits and to additional organisms amenable to genotyping in which a certain

phenotype can be selected for.

With the availability of new methodologies to map and identify QTLs, such as the one

presented here, many issues can be addressed. One important lesson from the explosion

of whole genome studies during the last years is that 'simple' monogenic Mendelian

inheritance is often not so simple and the traditional distinction between Mendelian and

quantitative traits begins to blur. What was considered in the past a gene with a

114

Mendelian inheritance pattern, can be regarded today as a QTL with a strong phenotypic

effect. Mutational or genotypic heterogeneity can explain some of the phenotypic

variation observed in mono-genic traits, but usually not all (Botstein and Risch 2003).

The residual variation is probably due to modifier genes, genetic background and

environmental contributors. Identifying such modifiers is an important challenge for the

future. Our work provides new comprehension and novel tools for handling this

challenge and shed light on genetic basis of fitness variation.

115

Bibliography

Abiola, O., J. M. Angel, P. Avner, A. A. Bachmanov, J. K. Belknap, B. Bennett, E. P.

Blankenhorn, D. A. Blizard, V. Bolivar, G. A. Brockmann, K. J. Buck, J. F. Bureau, W.

L. Casley, E. J. Chesler, J. M. Cheverud, G. A. Churchill, M. Cook, J. C. Crabbe, W. E.

Crusio, A. Darvasi, G. de Haan, P. Dermant, R. W. Doerge, R. W. Elliot, C. R. Farber,

L. Flaherty, J. Flint, H. Gershenfeld, J. P. Gibson, J. Gu, W. Gu, H. Himmelbauer, R.

Hitzemann, H. C. Hsu, K. Hunter, F. F. Iraqi, R. C. Jansen, T. E. Johnson, B. C. Jones,

G. Kempermann, F. Lammert, L. Lu, K. F. Manly, D. B. Matthews, J. F. Medrano, M.

Mehrabian, G. Mittlemann, B. A. Mock, J. S. Mogil, X. Montagutelli, G. Morahan, J. D.

Mountz, H. Nagase, R. S. Nowakowski, B. F. O'Hara, A. V. Osadchuk, B. Paigen, A. A.

Palmer, J. L. Peirce, D. Pomp, M. Rosemann, G. D. Rosen, L. C. Schalkwyk, Z. Seltzer,

S. Settle, K. Shimomura, S. Shou, J. M. Sikela, L. D. Siracusa, J. L. Spearow, C.

Teuscher, D. W. Threadgill, L. A. Toth, A. A. Toye, C. Vadasz, G. Van Zant, E.

Wakeland, R. W. Williams, H. G. Zhang and F. Zou (2003). "The nature and

identification of quantitative trait loci: a community's view." Nat Rev Genet 4(11): 911-

6.

Angeletti, P. C., D. Walker and A. T. Panganiban (2002). "Small glutamine-rich protein/viral

protein U-binding protein is a novel cochaperone that affects heat shock protein 70

activity." Cell Stress Chaperones 7(3): 258-68.

Atwood, K. C., L. K. Schneider and F. J. Ryan (1951). "Periodic selection in Escherichia coli."

Proc Natl Acad Sci U S A 37(3): 146-55.

Barton, N. (1998). "Evolutionary biology. The geometry of adaptation." Nature 395(6704): 751-

2.

Barton, N. H. and M. Turelli (1989). "Evolutionary quantitative genetics: how little do we

know?" Annu Rev Genet 23: 337-70.

Ben-Ari, G., D. Zenvirth, A. Sherman, L. David, M. Klutstein, U. Lavi, J. Hillel and G. Simchen

(2006). "Four linked genes participate in controlling sporulation efficiency in budding

yeast." PLoS Genet 2(11): e195.

Binder, A. (2006). "Identification of genes for a complex trait: examples from hypertension."

Curr Pharm Biotechnol 7(1): 1-13.

Botstein, D. and N. Risch (2003). "Discovering genotypes underlying human phenotypes: past

successes for mendelian disease, future approaches for complex disease." Nat Genet 33

Suppl: 228-37.

Brachmann, C. B., A. Davies, G. J. Cost, E. Caputo, J. Li, P. Hieter and J. D. Boeke (1998).

"Designer deletion strains derived from Saccharomyces cerevisiae S288C: a useful set

of strains and plasmids for PCR-mediated gene disruption and other applications." Yeast

14(2): 115-32.

Breitkreutz, B. J., C. Stark, T. Reguly, L. Boucher, A. Breitkreutz, M. Livstone, R. Oughtred, D.

H. Lackner, J. Bahler, V. Wood, K. Dolinski and M. Tyers (2008). "The BioGRID

Interaction Database: 2008 update." Nucleic Acids Res 36(Database issue): D637-40.

Brem, R. B. and L. Kruglyak (2005). "The landscape of genetic complexity across 5,700 gene

expression traits in yeast." Proc Natl Acad Sci U S A 102(5): 1572-7.

Brem, R. B., J. D. Storey, J. Whittle and L. Kruglyak (2005). "Genetic interactions between

polymorphisms that affect gene expression in yeast." Nature 436(7051): 701-3.

Brem, R. B., G. Yvert, R. Clinton and L. Kruglyak (2002). "Genetic dissection of transcriptional

regulation in budding yeast." Science 296(5568): 752-5.

Brown, P. O. (1994). "Genome scanning methods." Curr Opin Genet Dev 4(3): 366-73.

Cardon, L. R. and G. R. Abecasis (2003). "Using haplotype blocks to map human complex trait

loci." Trends Genet 19(3): 135-40.

Cardon, L. R. and J. I. Bell (2001). "Association study designs for complex diseases." Nat Rev

Genet 2(2): 91-9.

116

Cardona, F., A. Aranda and M. del Olmo (2009). "Ubiquitin ligase Rsp5p is involved in the

gene expression changes during nutrient limitation in Saccharomyces cerevisiae." Yeast

26(1): 1-15.

Carlborg, O. and C. S. Haley (2004). "Epistasis: too often neglected in complex trait studies?"

Nat Rev Genet 5(8): 618-25.

Carlborg, O., L. Jacobsson, P. Ahgren, P. Siegel and L. Andersson (2006). "Epistasis and the

release of genetic variation during long-term selection." Nat Genet 38(4): 418-20.

Castle, W. E. (1914). "Multiple Factors in Heredity." Science 39(1010): 686-689.

Causton, H. C., B. Ren, S. S. Koh, C. T. Harbison, E. Kanin, E. G. Jennings, T. I. Lee, H. L.

True, E. S. Lander and R. A. Young (2001). "Remodeling of yeast genome expression

in response to environmental changes." Mol Biol Cell 12(2): 323-37.

Chung, C. T. and R. H. Miller (1993). "Preparation and storage of competent Escherichia coli

cells." Methods Enzymol 218: 621-7.

Cordell, H. J. (2002). "Epistasis: what it means, what it doesn't mean, and statistical methods to

detect it in humans." Hum Mol Genet 11(20): 2463-8.

Darvasi, A. (1997). "The effect of selective genotyping on QTL mapping accuracy." Mamm

Genome 8(1): 67-8.

Darvasi, A. (1997). "Interval-specific congenic strains (ISCS): an experimental design for

mapping a QTL into a 1-centimorgan interval." Mamm Genome 8(3): 163-7.

Darvasi, A. (1998). "Experimental strategies for the genetic dissection of complex traits in

animal models." Nat Genet 18(1): 19-24.

Darvasi, A. and A. Pisante-Shalom (2002). "Complexities in the genetic dissection of

quantitative trait loci." Trends Genet 18(10): 489-91.

Darvasi, A. and M. Soller (1994). "Selective DNA pooling for determination of linkage between

a molecular marker and a quantitative trait locus." Genetics 138(4): 1365-73.

Darvasi, A., A. Weinreb, V. Minke, J. I. Weller and M. Soller (1993). "Detecting marker-QTL

linkage and estimating QTL gene effect and map location using a saturated genetic

map." Genetics 134(3): 943-51.

Davis, D. (2003). "Adaptation to environmental pH in Candida albicans and its relation to

pathogenesis." Curr Genet 44(1): 1-7.

Desai, M. M., D. S. Fisher and A. W. Murray (2007). "The speed of evolution and maintenance

of variation in asexual populations." Curr Biol 17(5): 385-94.

Deutschbauer, A. M. and R. W. Davis (2005). "Quantitative trait loci mapped to single-

nucleotide resolution in yeast." Nat Genet 37(12): 1333-40.

Dunham, M. J., H. Badrane, T. Ferea, J. Adams, P. O. Brown, F. Rosenzweig and D. Botstein

(2002). "Characteristic genome rearrangements in experimental evolution of

Saccharomyces cerevisiae." Proc Natl Acad Sci U S A 99(25): 16144-9.

East, E. M. (1909). "A Note Concerning Inheritance in Sweet Corn." Science 29(742): 465-467.

East, E. M. (1916). "Studies on Size Inheritance in Nicotiana." Genetics 1(2): 164-76.

Eshed, Y. and D. Zamir (1996). "Less-than-additive epistatic interactions of quantitative trait

loci in tomato." Genetics 143(4): 1807-17.

Fan, R. and J. Jung (2002). "Association studies of QTL for multi-allele markers by mixed

models." Hum Hered 54(3): 132-50.

Fijneman, R. J., S. S. de Vries, R. C. Jansen and P. Demant (1996). "Complex interactions of

new quantitative trait loci, Sluc1, Sluc2, Sluc3, and Sluc4, that influence the

susceptibility to lung cancer in the mouse." Nat Genet 14(4): 465-7.

Fisher, R. A. (1918). "The correlation between relatives on the supposition of Mendelian

inheritance." Trans. R. Soc. Edin. 52: 399–433.

Fisher, R. A. (1919). "The Genesis of Twins." Genetics 4(5): 489-99.

Fisher, R. A. (1930). "The Genetical Theory of Natural Selection." Oxford University Press,

Oxford.

Flint, J. and R. Mott (2001). "Finding the molecular basis of quantitative traits: successes and

pitfalls." Nat Rev Genet 2(6): 437-45.

Forgac, M. (1998). "Structure, function and regulation of the vacuolar (H+)-ATPases." FEBS

Lett 440(3): 258-63.

117

Fujita, K., A. Matsuyama, Y. Kobayashi and H. Iwahashi (2006). "The genome-wide screening

of yeast deletion mutants to identify the genes required for tolerance to ethanol and

other alcohols." FEMS Yeast Res 6(5): 744-50.

Gasch, A. P., P. T. Spellman, C. M. Kao, O. Carmel-Harel, M. B. Eisen, G. Storz, D. Botstein

and P. O. Brown (2000). "Genomic expression programs in the response of yeast cells

to environmental changes." Mol Biol Cell 11(12): 4241-57.

Gatbonton, T., M. Imbesi, M. Nelson, J. M. Akey, D. M. Ruderfer, L. Kruglyak, J. A. Simon

and A. Bedalov (2006). "Telomere length as a quantitative trait: genome-wide survey

and genetic mapping of telomere length-control genes in yeast." PLoS Genet 2(3): e35.

Gauss, R., E. Jarosch, T. Sommer and C. Hirsch (2006). "A complex of Yos9p and the HRD

ligase integrates endoplasmic reticulum quality control into the degradation machinery."

Nat Cell Biol 8(8): 849-54.

Gerke, J., K. Lorenz and B. Cohen (2009). "Genetic interactions between transcription factors

cause natural variation in yeast." Science 323(5913): 498-501.

Giaever, G., A. M. Chu, L. Ni, C. Connelly, L. Riles, S. Veronneau, S. Dow, A. Lucau-Danila,

K. Anderson, B. Andre, A. P. Arkin, A. Astromoff, M. El-Bakkoury, R. Bangham, R.

Benito, S. Brachat, S. Campanaro, M. Curtiss, K. Davis, A. Deutschbauer, K. D. Entian,

P. Flaherty, F. Foury, D. J. Garfinkel, M. Gerstein, D. Gotte, U. Guldener, J. H.

Hegemann, S. Hempel, Z. Herman, D. F. Jaramillo, D. E. Kelly, S. L. Kelly, P. Kotter,

D. LaBonte, D. C. Lamb, N. Lan, H. Liang, H. Liao, L. Liu, C. Luo, M. Lussier, R.

Mao, P. Menard, S. L. Ooi, J. L. Revuelta, C. J. Roberts, M. Rose, P. Ross-Macdonald,

B. Scherens, G. Schimmack, B. Shafer, D. D. Shoemaker, S. Sookhai-Mahadeo, R. K.

Storms, J. N. Strathern, G. Valle, M. Voet, G. Volckaert, C. Y. Wang, T. R. Ward, J.

Wilhelmy, E. A. Winzeler, Y. Yang, G. Yen, E. Youngman, K. Yu, H. Bussey, J. D.

Boeke, M. Snyder, P. Philippsen, R. W. Davis and M. Johnston (2002). "Functional

profiling of the Saccharomyces cerevisiae genome." Nature 418(6896): 387-91.

Gietz, R. D. and R. A. Woods (2006). "Yeast transformation by the LiAc/SS Carrier DNA/PEG

method." Methods Mol Biol 313: 107-20.

Glazier, A. M., J. H. Nadeau and T. J. Aitman (2002). "Finding genes that underlie complex

traits." Science 298(5602): 2345-9.

Goffeau, A., B. G. Barrell, H. Bussey, R. W. Davis, B. Dujon, H. Feldmann, F. Galibert, J. D.

Hoheisel, C. Jacq, M. Johnston, E. J. Louis, H. W. Mewes, Y. Murakami, P. Philippsen,

H. Tettelin and S. G. Oliver (1996). "Life with 6000 genes." Science 274(5287): 546,

563-7.

Greenberg, D. A. (1993). "Linkage analysis of "necessary" disease loci versus "susceptibility"

loci." Am J Hum Genet 52(1): 135-43.

Gresham, D., M. M. Desai, C. M. Tucker, H. T. Jenq, D. A. Pai, A. Ward, C. G. DeSevo, D.

Botstein and M. J. Dunham (2008). "The repertoire and dynamics of evolutionary

adaptations to controlled nutrient-limited environments in yeast." PLoS Genet 4(12):

e1000303.

Gresham, D., D. M. Ruderfer, S. C. Pratt, J. Schacherer, M. J. Dunham, D. Botstein and L.

Kruglyak (2006). "Genome-wide detection of polymorphisms at nucleotide resolution

with a single DNA microarray." Science 311(5769): 1932-6.

Gudbjartsson, D. F., G. B. Walters, G. Thorleifsson, H. Stefansson, B. V. Halldorsson, P.

Zusmanovich, P. Sulem, S. Thorlacius, A. Gylfason, S. Steinberg, A. Helgadottir, A.

Ingason, V. Steinthorsdottir, E. J. Olafsdottir, G. H. Olafsdottir, T. Jonsson, K. Borch-

Johnsen, T. Hansen, G. Andersen, T. Jorgensen, O. Pedersen, K. K. Aben, J. A. Witjes,

D. W. Swinkels, M. den Heijer, B. Franke, A. L. Verbeek, D. M. Becker, L. R. Yanek,

L. C. Becker, L. Tryggvadottir, T. Rafnar, J. Gulcher, L. A. Kiemeney, A. Kong, U.

Thorsteinsdottir and K. Stefansson (2008). "Many sequence variants affecting diversity

of adult human height." Nat Genet 40(5): 609-15.

He, F. and A. Jacobson (1995). "Identification of a novel component of the nonsense-mediated

mRNA decay pathway by use of an interacting protein screen." Genes Dev 9(4): 437-

54.

118

Heinisch, J. J., A. Lorberg, H. P. Schmitz and J. J. Jacoby (1999). "The protein kinase C-

mediated MAP kinase pathway involved in the maintenance of cellular integrity in

Saccharomyces cerevisiae." Mol Microbiol 32(4): 671-80.

Helling, R. B., C. N. Vargas and J. Adams (1987). "Evolution of Escherichia coli during growth

in a constant environment." Genetics 116(3): 349-58.

Hershko, A. and A. Ciechanover (1998). "The ubiquitin system." Annu Rev Biochem 67: 425-

79.

Hershko, A., H. Heller, S. Elias and A. Ciechanover (1983). "Components of ubiquitin-protein

ligase system. Resolution, affinity purification, and role in protein breakdown." J Biol

Chem 258(13): 8206-14.

Hillenmeyer, M. E., E. Fung, J. Wildenhain, S. E. Pierce, S. Hoon, W. Lee, M. Proctor, R. P. St

Onge, M. Tyers, D. Koller, R. B. Altman, R. W. Davis, C. Nislow and G. Giaever

(2008). "The chemical genomic portrait of yeast: uncovering a phenotype for all genes."

Science 320(5874): 362-5.

Hirschhorn, J. N. and M. J. Daly (2005). "Genome-wide association studies for common

diseases and complex traits." Nat Rev Genet 6(2): 95-108.

Holland, J. B. (2007). "Genetic architecture of complex traits in plants." Curr Opin Plant Biol

10(2): 156-61.

Hollander, M. a. W., D.A. (1999). "Nonparametric Statistical Methods, 2nd edition." John

Wiley and Sons, Inc.

Jensen, L. T. and D. R. Winge (1998). "Identification of a copper-induced intramolecular

interaction in the transcription factor Mac1 from Saccharomyces cerevisiae." Embo J

17(18): 5400-8.

Johnson, T. and N. Barton (2005). "Theoretical models of selection and mutation on quantitative

traits." Philos Trans R Soc Lond B Biol Sci 360(1459): 1411-25.

Jungmann, J., H. A. Reins, J. Lee, A. Romeo, R. Hassett, D. Kosman and S. Jentsch (1993).

"MAC1, a nuclear regulatory protein related to Cu-dependent transcription factors is

involved in Cu/Fe utilization and stress resistance in yeast." Embo J 12(13): 5051-6.

Kao, K. C. and G. Sherlock (2008). "Molecular characterization of clonal interference during

adaptive evolution in asexual populations of Saccharomyces cerevisiae." Nat Genet

40(12): 1499-504.

Keightley, P. D. (1998). "Genetic basis of response to 50 generations of selection on body

weight in inbred mice." Genetics 148(4): 1931-9.

Kim, Y. and H. A. Orr (2005). "Adaptation in sexuals vs. asexuals: clonal interference and the

Fisher-Muller model." Genetics 171(3): 1377-86.

Knight, S. A., S. Labbe, L. F. Kwon, D. J. Kosman and D. J. Thiele (1996). "A widespread

transposable element masks expression of a yeast copper transport gene." Genes Dev

10(15): 1917-29.

Konig, I. R., H. Schafer, H. H. Muller and A. Ziegler (2001). "Optimized group sequential study

designs for tests of genetic linkage and association in complex diseases." Am J Hum

Genet 69(3): 590-600.

Kupiec, M. and T. D. Petes (1988). "Allelic and ectopic recombination between Ty elements in

yeast." Genetics 119(3): 549-59.

Lagorce, A., N. C. Hauser, D. Labourdette, C. Rodriguez, H. Martin-Yken, J. Arroyo, J. D.

Hoheisel and J. Francois (2003). "Genome-wide analysis of the response to cell wall

mutations in the yeast Saccharomyces cerevisiae." J Biol Chem 278(22): 20345-57.

Lamb, T. M., W. Xu, A. Diamond and A. P. Mitchell (2001). "Alkaline response genes of

Saccharomyces cerevisiae and their relationship to the RIM101 pathway." J Biol Chem

276(3): 1850-6.

Lander, E. and L. Kruglyak (1995). "Genetic dissection of complex traits: guidelines for

interpreting and reporting linkage results." Nat Genet 11(3): 241-7.

Lander, E. S. and D. Botstein (1989). "Mapping mendelian factors underlying quantitative traits

using RFLP linkage maps." Genetics 121(1): 185-99.

Lander, E. S. and N. J. Schork (1994). "Genetic dissection of complex traits." Science

265(5181): 2037-48.

119

Lettre, G., A. U. Jackson, C. Gieger, F. R. Schumacher, S. I. Berndt, S. Sanna, S. Eyheramendy,

B. F. Voight, J. L. Butler, C. Guiducci, T. Illig, R. Hackett, I. M. Heid, K. B. Jacobs, V.

Lyssenko, M. Uda, M. Boehnke, S. J. Chanock, L. C. Groop, F. B. Hu, B. Isomaa, P.

Kraft, L. Peltonen, V. Salomaa, D. Schlessinger, D. J. Hunter, R. B. Hayes, G. R.

Abecasis, H. E. Wichmann, K. L. Mohlke and J. N. Hirschhorn (2008). "Identification

of ten loci associated with height highlights new biological pathways in human growth."

Nat Genet 40(5): 584-91.

Li, W. and A. P. Mitchell (1997). "Proteolytic activation of Rim1p, a positive regulator of yeast

sporulation and invasive growth." Genetics 145(1): 63-73.

Lin, C. H., J. A. MacGurn, T. Chu, C. J. Stefan and S. D. Emr (2008). "Arrestin-related

ubiquitin-ligase adaptors regulate endocytosis and protein turnover at the cell surface."

Cell 135(4): 714-25.

Linney, Y. M., R. M. Murray, E. R. Peters, A. M. MacDonald, F. Rijsdijk and P. C. Sham

(2003). "A quantitative genetic analysis of schizotypal personality traits." Psychol Med

33(5): 803-16.

Lippman, Z. B. and D. Zamir (2007). "Heterosis: revisiting the magic." Trends Genet 23(2): 60-

6.

Liti, G., D. M. Carter, A. M. Moses, J. Warringer, L. Parts, S. A. James, R. P. Davey, I. N.

Roberts, A. Burt, V. Koufopanou, I. J. Tsai, C. M. Bergman, D. Bensasson, M. J.

O'Kelly, A. van Oudenaarden, D. B. Barton, E. Bailes, A. N. Nguyen, M. Jones, M. A.

Quail, I. Goodhead, S. Sims, F. Smith, A. Blomberg, R. Durbin and E. J. Louis (2009).

"Population genomics of domestic and wild yeasts." Nature 458(7236): 337-41.

Lloret, A., E. Dragileva, A. Teed, J. Espinola, E. Fossale, T. Gillis, E. Lopez, R. H. Myers, M.

E. MacDonald and V. C. Wheeler (2006). "Genetic background modifies nuclear mutant

huntingtin accumulation and HD CAG repeat instability in Huntington's disease knock-

in mice." Hum Mol Genet 15(12): 2015-24.

Lord, P. W., R. D. Stevens, A. Brass and C. A. Goble (2003). "Investigating semantic similarity

measures across the Gene Ontology: the relationship between sequence and annotation."

Bioinformatics 19(10): 1275-83.

MacLeod, K. J., E. Vasilyeva, J. D. Baleja and M. Forgac (1998). "Mutational analysis of the

nucleotide binding sites of the yeast vacuolar proton-translocating ATPase." J Biol

Chem 273(1): 150-6.

Maher, B. (2008). "Personal genomes: The case of the missing heritability." Nature 456(7218):

18-21.

Martin-Yken, H., A. Dagkessamanskaia, F. Basmaji, A. Lagorce and J. Francois (2003). "The

interaction of Slt2 MAP kinase with Knr4 is necessary for signalling through the cell

wall integrity pathway in Saccharomyces cerevisiae." Mol Microbiol 49(1): 23-35.

Matthews, L., G. Gopinath, M. Gillespie, M. Caudy, D. Croft, B. de Bono, P. Garapati, J.

Hemish, H. Hermjakob, B. Jassal, A. Kanapin, S. Lewis, S. Mahajan, B. May, E.

Schmidt, I. Vastrik, G. Wu, E. Birney, L. Stein and P. D'Eustachio (2009). "Reactome

knowledgebase of human biological pathways and processes." Nucleic Acids Res

37(Database issue): D619-22.

McClintock, B. (1984). "The significance of responses of the genome to challenge." Science

226(4676): 792-801.

McConville, M. J. and M. A. Ferguson (1993). "The structure, biosynthesis and function of

glycosylated phosphatidylinositols in the parasitic protozoa and higher eukaryotes."

Biochem J 294 ( Pt 2): 305-24.

McCusker, J. H., K. V. Clemons, D. A. Stevens and R. W. Davis (1994). "Genetic

characterization of pathogenic Saccharomyces cerevisiae isolates." Genetics 136(4):

1261-9.

Miller, J. P., R. S. Lo, A. Ben-Hur, C. Desmarais, I. Stagljar, W. S. Noble and S. Fields (2005).

"Large-scale identification of yeast integral membrane protein interactions." Proc Natl

Acad Sci U S A 102(34): 12123-8.

Mitchell-Olds, T. and J. Schmitt (2006). "Genetic mechanisms and evolutionary significance of

natural variation in Arabidopsis." Nature 441(7096): 947-52.

120

Mitchell, A., G. H. Romano, B. Groisman, A. Yona, E. Dekel, M. Kupiec, O. Dahan and Y.

Pilpel (2009). "Adaptive prediction of environmental changes by microorganisms."

Nature 460(7252): 220-4.

Morley, M., C. M. Molony, T. M. Weber, J. L. Devlin, K. G. Ewens, R. S. Spielman and V. G.

Cheung (2004). "Genetic analysis of genome-wide variation in human gene expression."

Nature 430(7001): 743-7.

Mortimer, R. K. and J. R. Johnston (1986). "Genealogy of principal strains of the yeast genetic

stock center." Genetics 113(1): 35-43.

Munn, A. L. and H. Riezman (1994). "Endocytosis is required for the growth of vacuolar H(+)-

ATPase-defective yeast: identification of six new END genes." J Cell Biol 127(2): 373-

86.

Nelson, H. and N. Nelson (1990). "Disruption of genes encoding subunits of yeast vacuolar

H(+)-ATPase causes conditional lethality." Proc Natl Acad Sci U S A. 87(9): 3503-7.

Nelson, S. F., J. H. McCusker, M. A. Sander, Y. Kee, P. Modrich and P. O. Brown (1993).

"Genomic mismatch scanning: a new approach to genetic linkage mapping." Nat Genet

4(1): 11-8.

Notley-McRobb, L. and T. Ferenci (1999). "The generation of multiple co-existing mal-

regulatory mutations through polygenic evolution in glucose-limited populations of

Escherichia coli." Environ Microbiol 1(1): 45-52.

Notley-McRobb, L. and T. Ferenci (2000). "Experimental analysis of molecular events during

mutational periodic selections in bacterial evolution." Genetics 156(4): 1493-501.

Ohishi, K., N. Inoue and T. Kinoshita (2001). "PIG-S and PIG-T, essential for GPI anchor

attachment to proteins, form a complex with GAA1 and GPI8." Embo J 20(15): 4088-

98.

Orr, H. A. (1999). "The evolutionary genetics of adaptation: a simulation study." Genet Res

74(3): 207-14.

Orr, H. A. (2003). "The distribution of fitness effects among beneficial mutations." Genetics

163(4): 1519-26.

Orr, H. A. and J. A. Coyne (1992). "The genetics of adaptation: a reassessment." Am Nat

140(5): 725-42.

Page, G. P., V. George, R. C. Go, P. Z. Page and D. B. Allison (2003). ""Are we there yet?":

Deciding when one has demonstrated specific genetic causation in complex diseases

and quantitative traits." Am J Hum Genet 73(4): 711-9.

Paquin, C. E. and J. Adams (1983). "Relative fitness can decrease in evolving asexual

populations of S. cerevisiae." Nature 306(5941): 368-70.

Pearson, K. (1903). "Biometry and Biometrika." Science 17(432): 592-594.

Pena, M. M., S. Puig and D. J. Thiele (2000). "Characterization of the Saccharomyces cerevisiae

high affinity copper transporter Ctr3." J Biol Chem 275(43): 33244-51.

Penalva, M. A. and H. N. Arst, Jr. (2002). "Regulation of gene expression by ambient pH in

filamentous fungi and yeasts." Microbiol Mol Biol Rev 66(3): 426-46, table of contents.

Penalva, M. A., J. Tilburn, E. Bignell and H. N. Arst, Jr. (2008). "Ambient pH gene regulation

in fungi: making connections." Trends Microbiol 16(6): 291-300.

Pepin, K. M., M. A. Samuel and H. A. Wichman (2006). "Variable pleiotropic effects from

mutations at the same locus hamper prediction of fitness from a fitness component."

Genetics 172(4): 2047-56.

Reynolds, T. B. and G. R. Fink (2001). "Bakers' yeast, a model for fungal biofilm formation."

Science 291(5505): 878-81.

Rockman, M. V. and L. Kruglyak (2006). "Genetics of global gene expression." Nat Rev Genet

7(11): 862-72.

Romano, G., Y. Gurvich, O. Lavi, I. Ulitsky, R. Shamir and M. Kupiec (2010). "Different sets

of QTLs influence fitness variation in yeast." Molecular Systems Biology 6:346.

Ronald, J., R. B. Brem, J. Whittle and L. Kruglyak (2005). "Local regulatory variation in

Saccharomyces cerevisiae." PLoS Genet 1(2): e25.

121

Rosenzweig, R. F., R. R. Sharp, D. S. Treves and J. Adams (1994). "Microbial evolution in a

simple unstructured environment: genetic differentiation in Escherichia coli." Genetics

137(4): 903-17.

Schacherer, J., D. M. Ruderfer, D. Gresham, K. Dolinski, D. Botstein and L. Kruglyak (2007).

"Genome-wide analysis of nucleotide-level variation in commonly used Saccharomyces

cerevisiae strains." PLoS ONE 2(3): e322.

Schacherer, J., J. A. Shapiro, D. M. Ruderfer and L. Kruglyak (2009). "Comprehensive

polymorphism survey elucidates population structure of Saccharomyces cerevisiae."

Nature 458(7236): 342-5.

Schadt, E. E., S. A. Monks, T. A. Drake, A. J. Lusis, N. Che, V. Colinayo, T. G. Ruff, S. B.

Milligan, J. R. Lamb, G. Cavet, P. S. Linsley, M. Mao, R. B. Stoughton and S. H.

Friend (2003). "Genetics of gene expression surveyed in maize, mouse and man."

Nature 422(6929): 297-302.

Scrimale, T., L. Didone, K. L. de Mesy Bentley and D. J. Krysan (2009). "The unfolded protein

response is induced by the cell wall integrity mitogen-activated protein kinase signaling

cascade and is required for cell wall integrity in Saccharomyces cerevisiae." Mol Biol

Cell 20(1): 164-75.

Serrano, R. (1996). "Salt tolerance in plants and microorganisms: toxicity targets and defense

responses." Int Rev Cytol 165: 1-52.

Serrano, R., D. Bernal, E. Simon and J. Arino (2004). "Copper and iron are the limiting factors

for growth of the yeast Saccharomyces cerevisiae in an alkaline environment." J Biol

Chem 279(19): 19698-704.

Serrano, R., H. Martin, A. Casamayor and J. Arino (2006). "Signaling alkaline pH stress in the

yeast Saccharomyces cerevisiae through the Wsc1 cell surface sensor and the Slt2

MAPK pathway." J Biol Chem 281(52): 39785-95.

Shachar, R., L. Ungar, M. Kupiec, E. Ruppin and R. Sharan (2008). "A systems-level approach

to mapping the telomere length maintenance gene circuitry." Mol Syst Biol 4: 172.

Shalom, A. and A. Darvasi (2002). "Experimental designs for QTL fine mapping in rodents."

Methods Mol Biol 195: 199-223.

Shifman, S., M. Bronstein, M. Sternfeld, A. Pisante-Shalom, E. Lev-Lehman, A. Weizman, I.

Reznik, B. Spivak, N. Grisaru, L. Karp, R. Schiffer, M. Kotler, R. D. Strous, M. Swartz-

Vanetik, H. Y. Knobler, E. Shinar, J. S. Beckmann, B. Yakir, N. Risch, N. B. Zak and

A. Darvasi (2002). "A highly significant association between a COMT haplotype and

schizophrenia." Am J Hum Genet 71(6): 1296-302.

Sinha, H., B. P. Nicholson, L. M. Steinmetz and J. H. McCusker (2006). "Complex genetic

interactions in a quantitative trait locus." PLoS Genet 2(2): e13.

Smirnov, D., A. Bruzel, M. Morley and V. G. Cheung (2004). "Direct IBD mapping: identical-

by-descent mapping without genotyping." Genomics 83(2): 335-45.

St Onge, R. P., R. Mani, J. Oh, M. Proctor, E. Fung, R. W. Davis, C. Nislow, F. P. Roth and G.

Giaever (2007). "Systematic pathway analysis using high-resolution fitness profiling of

combinatorial gene deletions." Nat Genet 39(2): 199-206.

Steinmetz, L. M., H. Sinha, D. R. Richards, J. I. Spiegelman, P. J. Oefner, J. H. McCusker and

R. W. Davis (2002). "Dissecting the architecture of a quantitative trait locus in yeast."

Nature 416(6878): 326-30.

Stevens, T. H. and M. Forgac (1997). "Structure, function and regulation of the vacuolar (H+)-

ATPase." Annu Rev Cell Dev Biol 13: 779-808.

Storici, F. and M. A. Resnick (2003). "Delitto perfetto targeted mutagenesis in yeast with

oligonucleotides." Genet Eng (N Y) 25: 189-207.

Tagkopoulos, I., Y. C. Liu and S. Tavazoie (2008). "Predictive behavior within microbial

genetic networks." Science 320(5881): 1313-7.

Tanksley, S. D. (1993). "Mapping polygenes." Annu Rev Genet 27: 205-33.

Thomas, B. J. and R. Rothstein (1989). "Elevated recombination rates in transcriptionally active

DNA." Cell 56(4): 619-30.

Tong, A. H., G. Lesage, G. D. Bader, H. Ding, H. Xu, X. Xin, J. Young, G. F. Berriz, R. L.

Brost, M. Chang, Y. Chen, X. Cheng, G. Chua, H. Friesen, D. S. Goldberg, J. Haynes,

122

C. Humphries, G. He, S. Hussein, L. Ke, N. Krogan, Z. Li, J. N. Levinson, H. Lu, P.

Menard, C. Munyana, A. B. Parsons, O. Ryan, R. Tonikian, T. Roberts, A. M. Sdicu, J.

Shapiro, B. Sheikh, B. Suter, S. L. Wong, L. V. Zhang, H. Zhu, C. G. Burd, S. Munro,

C. Sander, J. Rine, J. Greenblatt, M. Peter, A. Bretscher, G. Bell, F. P. Roth, G. W.

Brown, B. Andrews, H. Bussey and C. Boone (2004). "Global mapping of the yeast

genetic interaction network." Science 303(5659): 808-13.

Treton, B., S. Blanchin-Roland, M. Lambert, A. Lepingle and C. Gaillardin (2000). "Ambient

pH signalling in ascomycetous yeasts involves homologues of the Aspergillus nidulans

genes palF and paIH." Mol Gen Genet 263(3): 505-13.

van der Rest, M. E., A. H. Kamminga, A. Nakano, Y. Anraku, B. Poolman and W. N. Konings

(1995). "The plasma membrane of Saccharomyces cerevisiae: structure, function, and

biogenesis." Microbiol Rev 59(2): 304-22.

Viladevall, L., R. Serrano, A. Ruiz, G. Domenech, J. Giraldo, A. Barcelo and J. Arino (2004).

"Characterization of the calcium-mediated response to alkaline stress in Saccharomyces

cerevisiae." J Biol Chem 279(42): 43614-24.

Wee, S., B. Hetfeld, W. Dubiel and D. A. Wolf (2002). "Conservation of the COP9/signalosome

in budding yeast." BMC Genet 3: 15.

Weedon, M. N., H. Lango, C. M. Lindgren, C. Wallace, D. M. Evans, M. Mangino, R. M.

Freathy, J. R. Perry, S. Stevens, A. S. Hall, N. J. Samani, B. Shields, I. Prokopenko, M.

Farrall, A. Dominiczak, T. Johnson, S. Bergmann, J. S. Beckmann, P. Vollenweider, D.

M. Waterworth, V. Mooser, C. N. Palmer, A. D. Morris, W. H. Ouwehand, J. H. Zhao,

S. Li, R. J. Loos, I. Barroso, P. Deloukas, M. S. Sandhu, E. Wheeler, N. Soranzo, M.

Inouye, N. J. Wareham, M. Caulfield, P. B. Munroe, A. T. Hattersley, M. I. McCarthy

and T. M. Frayling (2008). "Genome-wide association analysis identifies 20 loci that

influence adult height." Nat Genet 40(5): 575-83.

Winzeler, E. A., C. I. Castillo-Davis, G. Oshiro, D. Liang, D. R. Richards, Y. Zhou and D. L.

Hartl (2003). "Genetic diversity in yeast assessed with whole-genome oligonucleotide

arrays." Genetics 163(1): 79-89.

Winzeler, E. A., D. R. Richards, A. R. Conway, A. L. Goldstein, S. Kalman, M. J. McCullough,

J. H. McCusker, D. A. Stevens, L. Wodicka, D. J. Lockhart and R. W. Davis (1998).

"Direct allelic variation scanning of the yeast genome." Science 281(5380): 1194-7.

Yu, H., P. Braun, M. A. Yildirim, I. Lemmens, K. Venkatesan, J. Sahalie, T. Hirozane-

Kishikawa, F. Gebreab, N. Li, N. Simonis, T. Hao, J. F. Rual, A. Dricot, A. Vazquez, R.

R. Murray, C. Simon, L. Tardivo, S. Tam, N. Svrzikapa, C. Fan, A. S. de Smet, A.

Motyl, M. E. Hudson, J. Park, X. Xin, M. E. Cusick, T. Moore, C. Boone, M. Snyder, F.

P. Roth, A. L. Barabasi, J. Tavernier, D. E. Hill and M. Vidal (2008). "High-quality

binary protein interaction map of the yeast interactome network." Science 322(5898):

104-10.

Yvert, G., R. B. Brem, J. Whittle, J. M. Akey, E. Foss, E. N. Smith, R. Mackelprang and L.

Kruglyak (2003). "Trans-acting regulatory variation in Saccharomyces cerevisiae and

the role of transcription factors." Nat Genet 35(1): 57-64.

Zachariae, W. and K. Nasmyth (1999). "Whose end is destruction: cell division and the

anaphase-promoting complex." Genes Dev 13(16): 2039-58.

Zeyl, C. (2005). "The number of mutations selected during adaptation in a laboratory population

of Saccharomyces cerevisiae." Genetics 169(4): 1825-31.

Zeyl, C. (2007). "Evolutionary genetics: a piggyback ride to adaptation and diversity." Curr Biol

17(9): R333-5.

תכונות מורכבות של גנטיניתוח גישה משולבת ל

בשמר ההנצה

"דוקטור לפילוסופיה"חיבור לש� קבלת התואר

:מאת

חגית רומנו�גל

אביב�הוגש לסנאט של אוניברסיטת תל

2009 ,אוקטובר

תקציר

תכונות החשובות לאיכות , השונות בפיטנס של אורגניזמי� ביניה ,תכונות רבות בטבע

הינ , מות וחומרת הסבירות לחלות במחלות מסויותכונות שונות בבני אד� כמו ,חקלאיהיבול ה

הפנוטיפ של תכונות מורכבות הינו . מושפעות ממספר גני� ומגורמי� סביבתיי�ה תכונות מורכבות

לתכונות מורכבות , לפיכ�. תור� מעט לביטוי התכונה �ה של מספר גני� אשר כל אחד מהתוצא

, תכונות מנדליות .באופ דיכוטומיזה מזה אינ� נבדלי� הפרטי� באוכלוסיה ו הדרגתיפנוטיפ לרוב

. ולכ ההבדלי� בי פרטי� באוכלוסיה צפויי� להיות דיכוטומי� ,דילעומת זאת מושפעות מג יח

ניכר כי ג� הפנוטיפ של תכונות , ההבדל המהותי בי תכונות מורכבות לתכונות מנדליותלמרות

שההבחנה עובדה זו מצביעה על כ� . אלא מושפע מרקע גנטי ,מנדליות לרוב אינו דיכוטומי

ג� התכונות ,למעשה. המסורתית בי תכונות מורכבות לתכונות מנדליות עשויה להולי� שולל

בשל תפוצת . מג יחיד באמת מושפעותהתכונות ות ה המעטו, מורכבותניהנחשבות מנדליות ה

נמצאי� ,מחקרי� שמטרת� ניתוח והבנת הארכיטקטורה הגנטית של תכונות מורכבות, העצומה

Quantitative Trait Loci]זיהוי גני� המשפיעי� על תכונות מורכבות .כיו� בחזית המחקר הגנטי

(QTLs)], בחובו קשיי� רבי�נחשב למשימה מאתג הכלי� הגנטיי� .רת ביותר שכ הוא טומ

אינ� מתאימי� לזיהוי גני� של תכונות , המתוחכמי� והיעילי� המאפשרי� זיהוי של גני� מנדליי�

בקורלציה טמו QTLsהאתגר העיקרי במיפוי . לכ נדרשות גישות אלטרנטיביות חדשות ,מורכבות

ג תרומה מועטה לפנוטיפ לכ� שלכציה נמוכה זו נובעת מקורל. הנמוכה בי הפנוטיפ לגנוטיפ

, זאת ועוד .זהה פנוטיפקומבינציות אלליות שונות עשויות להביא לביטוי של , בנוס�. הסופי

.ימי�מסו QTLsאינטראקציות אפיסטטיות ורקע גנטי עשויי� למס� או לשנות את התרומה של

פוי של כל הרשת הגנטית התורמת עבודה זו מציגה אסטרטגיה גלובלית המאפשרת מי

בחרנו ביכולת של שמר ההנצה , על מנת לפתח את האסטרטגיה. לתכונה מורכבת יחידה

Saccharomyces cerevisiae אסטרטגיה זו הינה . לגדול בסביבה בסיסית כמודל לתכונה מורכבת

ל התכונה ללא גני� המשפיעי� ע באמצעותה זוהו מספר מערכי� שוני� שלשכ , פורצת דר� בתחו�

השלב הראשו כלל אומד של מספר הגני� המשפיעי� על . ידע מוקד� על תפקיד ומיקו� הגני�

נעשה שימוש בשתי אסטרטגיות בלתי תלויות לזיהוי , לאחר מכ . התכונה ומידת התורשתיות שלה

� ה QTLs . קווי� 'ויצירת באמצעות סלקציה ארוכת טווח 'אבולוציה במבחנה' :אסטרטגיות אלו ה

.על ידי סדרה של הכלאות חוזרות (Congenic lines) 'קונגניי�

בתהלי� האבולוציה במבחנה גידלנו ז מעבדה שאינו יכול לגדול בסביבה בסיסית תחת

בתהלי� זה הועשרו מוטציות המקנות יכולת גידול בסביבה . בהדרגהלח! אלקאלי אשר הוחמר

מושבות אשר הפיטנס שלה בסביבה ותהלי� בודדבסו� ה. בסיסית בז מעבדה החסר יכולה זו

העלתה כי במושבות שונות , אנליזה גנטית של מושבות אלו. בסיסית השתפר באופ משמעותי

המקנות יתרו סלקטיבי בסביבה , התפתחו מערכי� שוני� של מוטציות דומיננטיות ורצסיוויות

�זיהוי המוטציות נעשה באמצעות היברידיזציה ל. בסיסיתtiling arrays . מיקו� המוטציות אומת

חסרי� רציפרוקליי� , באמצעות ריצו� ותרומתה של כל מוטציה נאמדה באמצעות יצירת חסרי�

אומד התרומה היחסית של כל מוטציה הצביע על כ� כי כל המוטציות ביחד .והחלפת אללי�

לפיו , טיביימודל האדכי ה מציעהתוצאה זו . תורמות פחות מסכו� התרומות של כל מוטציה בנפרד

, לעומת זאת. הסופי ק על מנת להסביר את הפנוטיפאינו מספ, תרומה בלתי תלויה QTLלכל

QTLהמשפיעי� על התרומה של כל ,יחסי� אפיסטטי� בי הגני� התוצאה מתאימה למודל המציע

".השל� שונה מסכו� חלקיו"כ� ש

בעלי בני אד� דמ� של ודד מהשתמשנו בשמר מז בר שב, בנוס� לאבולוציה במבחנה

לעומת ז המעבדה , גבוה pHלז זה יכולת לגדול בתנאי� קיצוניי� כדוגמת . מערכת חיסונית פגומה

לקבלת צאצאי� ע� שונות גדולה ז המעבדה הוכלא ע� ז הבר הקליני . הרגיש בתנאי� אלו

ההכלאות החוזרות בוצעו .כלאו שוב לז המעבדהוה בסביבה בסיסיתהצאצאי� העמידי� . בפנוטיפ

מאחר וכל ההכלאות . עד דור שמיני של צאצאי� עמידי� בסביבה בסיסית' קו קונגני'לקבלת

מז עברה בתורשהצפוי כי מרבית הגנו� של צאצאי הדור השמיני הו ,החוזרות היו ע� ז המעבדה

�לצאצאי הדור השמיני יכולת טובה לגדול ב, בניגוד לז המעבדה, אול�. זהpH גבוה .�נית , לפיכ

זיהוי . המקני� יכולת זו QTLsלהניח כי האזורי� הגנומיי� שהורשו מז הבר הקליני כוללי�

אזורי� החשודי� כמכילי� 17זוהו . tiling arraysאזורי� אלו נעשה בשיטות שונות כדוגמת

QTLs .על מנת לזהות את ה�QTL מידע שבידנוכלי� חישוביי� ע� ההרלוונטי בכל אזור שילבנו .

ישמנו אלגורית� המדרג את הגני� החשודי� בכל אזור על פי סבירות� לקיי� אינטראקציות או

�להיות בעלי פעילות דומה לזו של ה QTLs על מנת לאמת . "אבולוציה במבחנה"שנמצאו בשיטת ה

שפיע בחנו את ההשפעה של חסרי� בגני� שנחזו כבעלי סבירות גבוהה לה, את תוצאות האלגורית�

חסרי� 28מתו� 4לעומת [נמצאו כמשפיעי� על התכונה , חסרי� שנבחנו 29מתו� 12. על התכונה

. ](p-value=0.023)י האלגורית� "שלא נחזו ע

מרבית המוטציות שהתגלו אינ משפיעות על גני� מבניי� המעורבי� ישירות , באופ מעניי

א על גני� רגולטורי� כוללניי� כדוגמת גני� אל, האחראי להומיאוסטזיס בסביבה בסיסית נגנו במ

תוצאה זו מרמזת . גני� האחראי� לחישה של יוני� ולעיגו חלבוני דופ התא, במסלול היוביקיטי

מועדפי� על פני מוטציות ,לכ� ששינויי� אדפטיביי� בגני� רגולטורי� בעלי השפעה נרחבת

.בה� נעשתה הסלקציההמשפיעות באופ בלעדי על הפיטנס בתנאי� הספציפיי�

, MAC1, היא ג המקודד לפקטור טרנסקריפציה תלוי יוני נחושת אחת הדוגמאות שנבחנה

בג זה נוצרו . האבולוציה במבחנה וג� באמצעות ההכלאות החוזרות באמצעות ניסוישנמצא ג�

. יהשעברו אבולוציה בלתי תלו מושבות שחמבארבע מתו� (Cys271)מוטציות שונות באותה עמדה

של פקטור הטרנסקריפציה ולפיכ� לביטוי מוגבר של הגברת הפעילותמוטציה זו עשויה לגרו� ל

�ו CTR1שניי� מגני� אלו . הגני� המבוקרי� על ידוCTR3 ה� כחשודי� באמצעות �נמצאו א

בז המעבדה ובז הקליני חשפה כי בז המעבדה CTR3בדיקה מעמיקה של . ההכלאות החוזרות

טרנספוזו זה נמצא . שאינו נמצא בז הקליני של הג ) פרומוטר(זו באזור המקד� קיי� טרנספו

הוצאתו באופ מלאכותי העלתה את רמת הביטוי של .ת משמעותית את רמת הביטוי של הג כמפחי

.וכ שיפרה את יכולת הגידול בסביבה בסיסית, הג

מחקרי� . וניי� חשובי�מתוצאות המחקר עולות ג� מספר תובנות לגבי נושאי� אבולוצי

אופי המוטציות והתפתחות , קודמי� דני� באופ הצטברות המוטציות התורמות לתכונות מורכבות

תוצאות ניסוי האבולוציה מצביעות על כ� שבכל שלב סלקציה נוצרו . מסוג זה י�גנטי מערכי�

עוד מספר הביאה לכ� שבשל המוטציות שכיחות הגבוהה .מספר מוטציות במקביל באוכלוסיה

. נוצרת מוטציה נוספת באחד השבטי�, שבטי� בעלי יתרו סלקטיבי מתחרי� זה בזה באוכלוסיה

בעלת ער� , היווצרות המוטציה החדשה עשויה להטות את שיווי המשקל כ� שמוטציה קודמת

�זו מרתקת הדינאמיק. תתבסס באוכלוסיה למרות תרומתה הקטנה לפנוטיפ, אדפטיבי נמו

. ציונית ועשויה לתרו� רבות להבנת התהלי� בו מתפתחות רשתות גנטיותמבחינה אבולו

של הכדוגמת הדינאמיק �למחקר זה השלכות שונות ה בנושאי� אבולוציוניי, לסיכו�

וה בהבנת המכניז� המקנה , היווצרות מוטציות והארכיטקטורה הגנטית של תכונות מורכבות

הפתוגניות והתגובה למצבי , הפיסיולוגיה של התא מכניז� המשפיע על, עמידות בסביבה בסיסית

נית יהיה לחקור ולנתח כדוגמת אלו המובאות במחקר , ע� הנגישות של מתודולוגיות שונות. עקה

. בעתיד תכונות מורכבות נוספות