Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Complex mechanisms of cryptic genetic variation
(USC Thesis Other)
Complex mechanisms of cryptic genetic variation
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
i
COMPLEX MECHANISMS OF CRYPTIC
GENETIC VARIATION
by:
Jonathan T Lee
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(MOLECULAR BIOLOGY)
MAY 2019
ii
Acknowledgements
I would like to thank my advisor Ian, my committee members, and my lab mates
for their invaluable input towards this degree. Thank you to my partner Christina
for being my constant support and companion throughout the years. We started
these PhDs on the same day and have grown together into the people and the
researchers we are today.
iii
Acknowledgements ii
List of Figures vii
List of Tables ix
Abstract 1
Chapter 1: Introduction 2
1.1 Cryptic genetic variation 2
1.2 Higher-order epistasis 2
1.3 Genotype-by-environment interaction 3
1.4 Molecular mechanisms of cryptic variation 4
1.5 Goals of this dissertation 5
1.6 Chapter summary 6
Chapter 2: Multi-locus genotypes underlying temperature sensitivity in
a mutationally induced trait 8
2.1 Abstract 8
2.2 Introduction 9
2.3 Results and Discussion 11
2.3.1 Screen for temperature sensitivity among rough BYx3S
ira2Δ2933 segregants 11
2.3.2 Bulk segregant mapping of the three temperature sensitivity
classes 14
2.3.3 Multiple genotypes underlie the HS and MS classes 16
2.3.4 Roles of detected alleles in modulating temperature sensitivity 19
2.4 Conclusion 21
2.5 Materials and Methods 24
2.5.1 Generation of backcross segregants for the preliminary screen
and genetic mapping 23
2.5.2 Phenotyping of rough segregants at multiple temperatures 24
2.5.3 Bulk segregant mapping of temperature sensitivity 25
iv
2.5.4 Genotyping of MS individuals using PCR and restriction
digestion 26
2.5.5 Detailed genetic characterization of the HS class 26
2.5.6 Genetic engineering 28
2.6 Supporting Materials 29
Chapter 3: Diverse genetic architectures lead to the same cryptic
phenotype in a yeast cross 40
3.1 Abstract 40
3.2 Introduction 41
3.3 Results 43
3.3.1 Genetic mapping of 17 independent cases of rough morphology 43
3.3.2 Genes that can lead to rough morphology when mutated 46
3.3.3 Genetics of mutation-independent rough colony morphology 47
3.4 Discussion 49
3.5 Methods 52
3.5.1 Phenotyping of yeast colony morphology 52
3.5.2 Generation of rough segregants 52
3.5.3 Generation of backcross segregants 52
3.5.4 Sequencing of mapping populations 53
3.5.5 Genetic mapping using Multipool 53
3.5.6 Identification of de novo mutations 54
3.5.7 Genetic engineering experiments 54
3.6 Supporting Materials 56
Chapter 4: Unraveling the layers of cryptic genetic variation in a yeast
gene regulatory network 75
4.1 Abstract 75
4.2 Introduction 76
4.3 Results 79
4.3.1 Isolation of a rough flo8∆ sfl1∆ segregant 79
v
4.3.2 The three GPs vary in their potential to express the phenotype
across temperatures 79
4.3.3 The GPs uncover distinct loci 81
4.3.4 The GPs alter genotype-environment interaction 83
4.3.5 The GPs affect additivity and epistasis among identified loci 84
4.3.6 Coding and regulatory variation in FLO11 plays an essential
role in the trait 86
4.3.7 GPa and GPb rough segregants utilize different sub-pathways
that impact Ras 87
4.3.8 Cryptic variation in several pathways underlies the phenotype
in GPc segregants 90
4.4 Discussion 93
4.5 Methods 96
4.5.1 Knockout of FLO8 and SFL1 in BY and 3S 96
4.5.2 Isolation of a rough BYx3S segregant with GPc 96
4.5.3 Generation of backcross segregants 96
4.5.4 Phenotyping at multiple temperatures 97
4.5.5 Genotyping of GPb and GPc rough segregants 97
4.5.6 Genetic mapping of loci underlying the rough phenotype 98
4.5.7 Testing for genotypic heterogeneity 98
4.5.8 Exploration of additivity and epistasis 99
4.5.9 Genetic engineering at the FLO11 promoter and other loci 99
4.6 Supporting Materials 100
Chapter 5: Concluding Remarks 112
5.1 Impact of my work 112
5.2 Future directions 114
References 116
vi
Appendix A: Genetic suppression – Extending our knowledge from lab
experiments to natural populations 132
A.1 Summary 132
A.2 Introduction 133
A.3 Main Text 135
A.3.1 High-throughput techniques for identifying suppressor mutations
in lab experiments 135
A.3.2 Functional mechanisms that cause genetic suppression 139
A.3.3 Examples of genetic suppression in natural populations 140
A.3.4 The genetic and molecular basis of naturally occurring genetic
suppression 142
A.4 Conclusion and outlook 145
vii
List of Figures
Figure 2.1. Segregants that carry ira2Δ2933 show different temperature
sensitivities 10
Figure 2.2 Bulk segregant mapping results for different temperature
sensitivity classes 13
Figure 2.3 FLO11
3S
is required for trait expression at 37°C 15
Figure 2.4 The HS and MS classes are each specified by two predominant
multi-locus genotypes 17
Figure 2.5 Comparison of multi-locus genotypes underlying differences in
temperature sensitivity among BYx3S FLO8
3S
ira2Δ2933 segregants 20
Figure S2.1 Bulk segregant mapping results for each temperature sensitivity
class 29
Figure S2.2 Two multi-locus genotypes express the rough phenotype
exclusively at 21 and 30°C 31
Figure S2.3 Genome-wide scan for pairs of loci that show correlated allele
states among sequenced individuals from the HS class 32
Figure S2.4 Multiple alleles are required for rough morphology in a MSS11
3S
HS individual 33
Figure S2.5 MGA1
BY
is required for trait expression at 37°C in an NS genetic
background 34
Figure S2.6 Cloning of the causal variant in MSS11 35
Figure 3.1 Overview of study 42
Figure 3.2 Genetic mapping results 45
Figure 3.3 Locations of independent recombination events within the FLO11
locus 48
Figure 3.4 Genes involved in the rough phenotype in the BY × 3S
cross largely regulate signaling and transcriptional control by the
Ras pathway 50
Figure S3.1. Phenotypes of the 17 identified rough segregants at 21°C 56
viii
Figure S3.2. Verification that identified de novo mutations have phenotypic
effects in rough segregants 1 through 8 57
Figure S3.3. Rough segregants 1 and 2 possess the same SSN8 lesion, but
are likely distinct individuals obtained from different matings of BY and 3S 58
Figure S3.4. Complete deletion of mutated genes from rough segregants 1
through 8 59
Figure 4.1 Different genetic perturbations (‘GPs’) interact with cryptic
variation to cause rough morphology 80
Figure 4.2. Loci identified in the presence of the GPs 83
Figure 4.3. Loci detected across GPs and temperature sensitivity classes 85
Figure 4.4. A transcription factor binding site polymorphism is required for
GPb and GPc rough segregants to express the trait 87
Figure 4.5. GPs unmask cryptic genetic variation in parallel signaling
pathways and sub-pathways 93
Fiugre S4.1 GPb and GPc individuals differ in their temperature sensitivities 100
Figure S4.2 Scheme to generate GPb backcross mapping populations 101
Figure S4.3 Genetic mapping for loci interacting with GPb and GPc 102
Figure S4.4 Key corresponding to gene position in Figure 4.3A 103
Figure S4.5 Scan for loci with correlated allele states 104
Figure S4.6 Variation in the FLO11 coding region 105
Figure S4.7 Additional prFLO11 fine mapping details 106
Figure S4.8 The SRV2 gene underlies a locus on chromosome XIV that
interacts with GPb 107
Figure S4.9 Deletions support interactions between GPc and identified
candidate genes 108
Figure A.1 Genetic basis of suppression in lab versus natural environments 134
Figure A.2 Identifying intragenic interactions across an entire gene 135
Figure A.3 Techniques for identifying intergenic suppressors 137
Figure A.4 Combinations of genetic variants may cause suppression in
natural populations 144
ix
List of Tables
Table S2.1. Phenotypes of BYx3S ira2Δ2933 backcross segregants in
preliminary and secondary screens 36
Table S2.2. Initial screen for rough morphology among segregants isolated
at three different temperatures 36
Table S2.3. Classification of BY backcross segregants obtained from the
preliminary screen into the three temperature sensitivity classes 36
Table S2.4. Bulk segregant mapping populations were generated for each
temperature sensitivity class 37
Table S2.5. Primers used in this study 37
Table S3.1. Number of individuals in each pooled sequencing experiment 61
Table S3.2. Loci detected in rough backcross pools 62
Table S3.3. Loci detected in control backcross pools 66
Table S3.4. Predominant genotypes in mapping populations derived from
each rough segregant 68
Table S3.5. Information regarding causal de novo mutations 69
Table S3.6. SRA accession numbers for each pooled sequencing
experiment 71
Table S3.7. Primers used in this study 73
Table 4.1. Causal genes underlying identified loci 89
Table S4.1. Loci detected across GPs and temperatures 109
1
Abstract
Cryptic variation is genetic variation that does not typically influence phenotype
but can cause unexpected trait outcomes when genetic and environmental
perturbations are introduced into a system. Despite its influence on the
heritability of many evolutionary and medically significant traits, the role of cryptic
variation is often inferred rather than explicitly defined. This largely stems from
their effects being highly conditional, making cryptic genetic variants difficult to
identify and study.
In this dissertation, I define the complex genetic and molecular mechanisms of
cryptic variation in a yeast colony morphology trait. In chapter two, I show that
sets of cryptic variants in a conserved signaling pathway regulate the trait’s
expression in an environment-dependent manner. In chapter three, I
demonstrate that multiple genetic perturbations can cause the trait through
interaction with different combinations of cryptic variants in the same pathway. In
chapter four, I describe how regulatory rewiring unmasks cryptic variation in
alternative signaling pathways and modifies the quantitative genetic architecture
underlying the trait.
2
Chapter 1: Introduction
1.1 Cryptic genetic variation
Determining the link between genotype and phenotype is rarely a simple
endeavor in genetics. This is in part because many traits are genetically complex,
resulting from the collective efforts of multiple genetic and environmental factors
(Mackay et al. 2009, Phillips 2008). Identifying these factors can be challenging
and is not always comprehensive. For instance, while genome-wide association
studies in humans have identified a multitude of loci associated with various
phenotypes, the loci identified in this manner only explain a portion of trait
heritability (Manolio et al. 2009). Part of this discrepancy stems from genetic
variants that only exhibit an effect on a phenotype in certain environmental or
genetic backgrounds. Such alleles are termed ‘cryptic variants’, and cryptic
genetic variation (or cryptic variation) has been found to shape evolutionary
trajectories and influence genetically complex heritable diseases (Gibson 2009,
Paaby and Rockman 2014). Despite having important implications on both
evolution and health, cryptic variation is difficult to identify and characterize due
to its highly conditional nature. Nevertheless, an improved understanding of the
genetic and molecular mechanisms underlying cryptic variation is necessary to
further comprehend the genotype-phenotype map.
1.2 Higher-order epistasis
One mechanism by which cryptic variation occurs is through gene-gene
interactions, or ‘epistasis’ (Carlborg and Haley 2004). While the majority of
genetic variants are thought to have additive effects on complex trait expression
(Mackay 2014), the combined phenotypic effect of multiple alleles can potentially
deviate from expectations based on a purely additive model. As a result, certain
loci may only influence a trait’s expression in specific genetic backgrounds that
carry their interactors. In an extreme case, two alleles may show no phenotypic
effect on their own but do so when present together. Indeed, yeast synthetic
genetic arrays have identified many such interactions between cryptic variants
3
that otherwise have little to no phenotypic consequences on their own. These
types of studies have provided functional insights into previously uncharacterized
genes based on their interaction networks (Costanzo et al. 2016, Dowell et al.
2010). While pairwise epistasis has been studied extensively, less is known
about higher-order epistasis between three or more genetic variants. This lack of
study is largely due to the much greater statistical power needed to detect
higher-order interactions, which increases exponentially with each additional
locus. Despite this challenge, the presence of higher-order interactions has been
described in the literature (Carlborg et al. 2006, Gerke et al. 2010, Wang et al.
2013).
Such higher-order epistasis is important to our understanding of cryptic variation,
as this has been posited to be a major mechanism by which cryptic variants are
uncovered (Gibson and Dworkin 2004, Paaby and Rockman 2014). Additionally,
pairwise interactions can often give rise to higher-order interactions (Kuzmin et al.
2018, Mullis et al. 2018), suggesting that pairwise interactions that have been
identified only describe a subset of the underlying gene interaction network.
Studies in systems including mice (Shao et al. 2008), worms (Gaertner et al.
2012), and yeast (Gerke et al. 2010) have described sets of cryptic variants
acting collectively in a non-additive manner. However, these studies do not
explicitly define all the cryptic variants involved in a higher-order interaction.
Rather, their conclusions are inferred through genetic background effects, where
the effects of cryptic loci are genotype-dependent. An improved understanding of
complex epistasis, as it relates to cryptic variation, requires higher resolution
mapping of the genes underlying these background effects
1.3 Genotype-by-environment interaction
Cryptic variation is also uncovered through changes in environment. Genotype-
by-environment interaction (GxE) occurs when genetic variants have different
phenotypic effects depending on the environment that an organism is presented
with. Although GxE contributes to the heritability of many evolutionarily and
4
medically relevant traits (Gibson 2009, Gutteling et al. 2007), studies of GxE
rarely describe the genetic basis of these traits to full completion and the
mechanisms by which cryptic variants engage in GxE is not well understood.
This is partially because natural environments are complex, and attributing GxE
to a measured factor is difficult in the absence of controlled environments (Moffitt
et al. 2005). Study of GxE can also be hampered by statistical limitations, a
problem that is further complicated by epistasis. Because cryptic variants often
function as combinations of factors, instances of GxE may only be found among
individuals with certain rare genotypes. When accounting for both genetic and
environmental background, the number of statistical tests needed to detect
interactions therefore increases dramatically. Moving forward, studies regarding
the genetic basis of GxE may require methodologies that overcome these issues
to emphasize the effects that combinations of cryptic variants have across
environments.
1.4 Molecular mechanisms of cryptic variation
A comprehensive understanding of cryptic variation extends beyond genetic
mapping to address how combinations of variants function collectively at the
molecular level. Two proposed mechanisms include activation thresholds and
alternative pathways (Gibson and Dworkin 2004). For instance, multiple cryptic
variants in the same transcriptional signaling cascade can result in a threshold
level of gene expression that is needed to induce a change in phenotype. This
expression threshold may be unachievable in the presence of each individual
variant on its own. Alternatively, genetic and environmental perturbations can
cause genes in pathways not normally responsible for a trait to instead exhibit
functional variation in regard to that phenotype.
Following that they act collectively in a variety of ways, multiple combinations of
cryptic variants may specify the same trait. This phenomenon, ‘genotypic
heterogeneity’, can impede analysis of a trait’s genetic basis because the alleles
involved in different genotypes can mask one another in a heterogeneous
5
population (Manchia et al. 2013). As a result, cryptic variants that take part in
less common genotypes may not be detectable using population-level statistics.
At a mechanistic level, genotypic heterogeneity can occur if individuals carry
cryptic variants in the same pathways, or similar pathways, that cause the same
outcome downstream. Studies in model organisms (Taylor and Ehrenreich 2015)
as well as clinical studies (Walsh and King 2007, Walsh et al. 2008) have
demonstrated this mechanism for genotypic heterogeneity.
1.5 Goals of this dissertation
In the course of my PhD, my goal has been to comprehensively dissect the
genetic basis of a model complex trait that results from combinations of cryptic
variants interacting with each other and environment. In doing so, I sought to
address the following questions: 1. How do genetic perturbations unmask sets of
cryptic variants to cause unexpected trait outcomes? 2. What is the relationship
between the cryptic variants that enable a trait’s expression and those that
exhibit GxE? 3. By what mechanisms do different sets of cryptic variants cause
the same trait?
To this end, I helped develop, and employed a yeast colony morphology system
for investigating the effects of cryptic variants. In a cross of two Saccharomyces
cerevisiae strains, BY4716 (BY) and 322134S (3S), both strains and their cross-
progeny form ‘smooth colonies’ when grown on solid media. However, certain
genetic perturbations can enable BYx3S segregants to potentially express an
alternative ‘rough’ colony phenotype. As described in previous work from the lab
(Taylor and Ehrenreich 2014, Taylor and Ehrenreich 2015), discrete
combinations of cryptic variants originating from BY and 3S are responsible for
the trait.
This rough colony system serves as a powerful method for studying cryptic
variation. Yeast have a high recombination rate and a short generation time,
enabling screens of extremely large populations of genetically distinct individuals
6
to be performed in a matter of weeks. The rough phenotype itself provides an
identifiable marker for selection within these large populations, which can be
screened across many distinct and defined environments. Yeast also have a
relatively small genome and are easy to genetically engineer compared to other
eukaryotes. As a result, they are ideal for high-throughput genome sequencing,
genotype analysis, and cloning. Furthermore, the yeast genome is well annotated,
with most genes having known functions. This makes it possible to determine the
molecular mechanisms by which the genes underlying detected cryptic variants
interact with one another.
1.6 Chapter summary
In chapter two, I determine the genetic basis of the rough colony trait across
three growth temperatures. In each environment, I identify sets of cryptic variants
that are unmasked by a loss-of-function mutation in the IRA2 gene and form
higher-order epistatic interactions involving up to eight genes. I demonstrate that
genotypic heterogeneity occurs in multiple environments through the
identification of multiple discrete genotypes within populations. I further show that
some of these genotypes carry genetic variants that enable GxE and cause
additional genes to engage in higher-order epistasis.
In chapter three, I explore the genotypic space of the rough phenotype across 17
independent instances of the trait. In doing so, I find that different genetic
perturbations enable unique combinations of cryptic variants to cause the same
phenotype. These perturbations include eight de novo mutations as well as six
cases of recombination between the promoter and coding region of the cell
surface adhesion gene FLO11. Genotype comparisons reveal that the majority of
involved loci are shared across genetic backgrounds and act primarily through
the Ras pathway, with each perturbation causing different combinations of genes
in this pathway to exhibit functional variation in the FLO11 regulatory network.
7
In chapter four, I compare the genetic architecture of the trait in ira2∆ and
FLO11-recombinant genetic backgrounds across environments and show that
each genetic perturbation causes cryptic variants in different Ras sub-pathways
to exhibit GxE. I also map the genetic basis of the trait when decoupled from Ras
signaling, in which cryptic variation in alternative signaling pathways regulate
expression of FLO11. By analyzing the distribution of variants within populations,
I reveal that the relative contribution of additive and epistatic effects varies by
genetic perturbation based on the cryptic variants they each interact with.
In chapter 5, I summarize the findings I have accrued using this rough colony
system. I discuss the impacts of my work on the field and potential future
directions in utilizing this system to better understand cryptic variation and
complex traits.
8
Chapter 2: Multi-locus genotypes underlying temperature sensitivity in a
mutationally induced trait
This work appears as published in PLOS Genetics, 2016. 12(3):e1005929
2.1 Abstract
Determining how genetic variation alters the expression of heritable phenotypes
across conditions is important for agriculture, evolution, and medicine. Central to
this problem is the concept of genotype-by-environment interaction (or ‘GxE’),
which occurs when segregating genetic variation causes individuals to show
different phenotypic responses to the environment. While many studies have
sought to identify individual loci that contribute to GxE, obtaining a deeper
understanding of this phenomenon may require defining how sets of loci
collectively alter the relationship between genotype, environment, and phenotype.
Here, we identify combinations of alleles at seven loci that control how a
mutationally induced colony phenotype is expressed across a range of
temperatures (21, 30, and 37°C) in a panel of yeast recombinants. We show that
five predominant multi-locus genotypes involving the detected loci result in trait
expression with varying degrees of temperature sensitivity. By comparing these
genotypes and their patterns of trait expression across temperatures, we
demonstrate that the involved alleles contribute to temperature sensitivity in
different ways. While alleles of the transcription factor MSS11 specify the
potential temperatures at which the trait can occur, alleles at the other loci modify
temperature sensitivity within the range established by MSS11 in a genetic
background- and/or temperature- dependent manner. Our results not only
represent one of the first characterizations of GxE at the resolution of multi-locus
genotypes, but also provide an example of the different roles that genetic
variants can play in altering trait expression across conditions.
9
2.2 Introduction
Genotype-by-environment interaction (or ‘GxE’) occurs when genetically distinct
individuals exhibit different phenotypic responses to the environment (Falconer
and Mackay 1996, Lynch and Walsh 1998). Work to date suggests that GxE is
an important contributor to heritable variation in many agriculturally, evolutionarily,
and medically relevant phenotypes (Baye et al. 2011, Mackay et al. 2009, Rauw
and Gomez-Raya 2015, Zeng 2005). However, although ‘GxE’ has been
extensively studied, there are few, if any, traits for which the underlying genetic
basis of GxE is fully understood. This lack of detailed case studies may have a
technical basis, as causal loci involved in GxE can act in an environment- and
genetic background-dependent manner (Bhatia et al. 2014, Gerke et al. 2010),
making them difficult to detect. Improving understanding of GxE could there- fore
require characterizing how combinations of alleles, rather than individual loci,
influence phenotype across environments.
We recently described a trait in Saccharomyces cerevisiae that can serve as a
useful model for studying the complex genetic basis of GxE. In a cross of the
BY4716 (‘BY’) lab strain and a derivative of the 322134S (‘3S’) clinical isolate (Liti
et al. 2009), individuals typically exhibit ‘smooth’ colony morphology (Taylor and
Ehrenreich 2014, Taylor and Ehrenreich 2015) (Fig 2.1). However, we showed
that a spontaneous frameshift mutation in IRA2, a negative regulator of the Ras-
cAMP-PKA (Ras) pathway (Tanaka et al. 1990), enables certain BYx3S
segregants to express an alternative, ‘rough’ colony phenotype (Taylor and
Ehrenreich 2014, Taylor and Ehrenreich 2015) (Fig 2.1). This mutation
(ira2Δ2933) results in a truncation of the cognate Ira2 protein by 117 amino acids
and causes a partial loss of Ira2 function (Taylor and Ehrenreich 2014). However,
ira2Δ2933 is insufficient to induce the rough phenotype on its own, as particular
higher-order combinations of epistatically interacting alleles at the vesicle
component END3 (Bénédetti et al. 1994, Tang et al. 2000), the transcriptional
activators FLO8 (Kobayashi et al. 1996), MGA1 (Lorenz and Heitman 1998,
Robertson and Fink 1998), and MSS11 (Gagiano et al. 2002), the transcriptional
10
repressor SFL1 (Atsushi et al. 1989, Robertson and Fink 1998), and the
thioredoxin reductase TRR1 (Pedrajas et al. 1999) are also needed (Taylor and
Ehrenreich 2014, Taylor and Ehrenreich 2015, Taylor and Ehrenreich 2015).
Specifically, we identified two multi-locus genotypes — END3
BY
FLO8
3S
ira2Δ2933 MSS11
BY
TRR1
3S
and END3
3S
FLO8
3S
ira2Δ2933 MGA1
BY
MSS11
BY
SFL1
BY
— that can cause the rough phenotype.
Figure 2.1: Segregants that carry ira2Δ2933 show different
temperature sensitivities. BY4716 (BY), 322134S (3S), and their
recombinant offspring typically exhibit ‘smooth’ colony morphology.
However, in the presence of the ira2Δ2933 frameshift mutation, some
BYx3S segregants are capable of expressing an alternative, ‘rough’
colony phenotype. Examination of rough BYx3S ira2Δ2933 segregants at
21, 30, and 37°C revealed that these individuals largely express the
phenotype in a temperature sensitive manner. For the most part,
11
individuals exhibit the phenotype only at 21°C, only at 21°C and 30°C, or
at all examined temperatures. We refer to these three classes as highly
sensitive (HS), moderately sensitive (MS), and non-sensitive (NS) to
temperature, respectively.
The aforementioned results stem from work performed exclusively at 30°C, the
standard temperature used to culture S. cerevisiae in the lab. Here, we extend
our research on the rough phenotype to two additional temperatures: 21°C and
37°C. In doing so, we find that many BYx3S ira2Δ2933 segregants express the
rough phenotype in a temperature sensitive manner (Fig 1). To determine the
genetic basis of this GxE, we perform genetic mapping of several temperature
sensitivity classes using a combination of bulk segregant analysis (Ehrenreich et
al. 2010, Michelmore et al. 1991, Wenger et al. 2010) and selective genotyping of
individual cross progeny (Matsui et al. 2015, Taylor and Ehrenreich 2014). These
efforts lead to the identification of seven environmentally responsive loci, and five
specific multi-locus genotypes involving these loci, that influence the expression
of the rough phenotype across temperatures. As we describe below, comparison
of these multi-locus genotypes provides detailed insights into the genetic
architecture of temperature-dependent GxE in our system, and also sheds light
on the distinct roles that the causal alleles play in modifying the rough
phenotype’s expression at different temperatures.
2.3 Results and Discussion
2.3.1 Screen for temperature sensitivity among rough BYx3S ira2Δ2933
segregants
We screened for rough BYx3S ira2Δ2933 segregants at three temperatures—21,
30, and 37°C— in a backcross of the rough ira2Δ2933 segregant described in
(Taylor and Ehrenreich 2014) to BY (S2.1 and S2.2 Tables; Materials and
Methods). ~3,000 segregants (30 random spore plates of ~100 colonies per
plate) were examined at each temperature and ~9,000 segregants were
screened in total across the three temperatures (Materials and Methods).
12
Among the 252 rough individuals obtained from this screen, 173, 107, and 72
were recovered from 21, 30, and 37°C, respectively (S2.1 and S2.2 Tables). The
majority of these rough segregants were capable of expressing the phenotype at
temperatures other than the one from which they were collected. When these
rough individuals were individually examined at each of the three initially
employed temperatures, they largely fell into three classes: rough at 21°C only,
rough at 21 and 30°C only, or rough at all three temperatures (Fig 2.1; S2.1 and
S2.3 Tables; Materials and Methods). This implies that the major form of
temperature-dependent GxE in our system is temperature sensitivity. For the
remainder of the paper, we refer to the three aforementioned classes as highly
sensitive (‘HS’), moderately sensitive (‘MS’), and non-sensitive (‘NS’) to
temperature, respectively (Fig 2.1).
We sought to determine the underlying genetic basis of the temperature-
dependent GxE by selectively genotyping individuals in each temperature
sensitivity class. To generate the necessary populations for this genetic mapping
strategy, we screened an additional 60 plates of random spores from the BY
backcross mentioned above, and also mated the rough ira2Δ2933 segregant
described in (Taylor and Ehrenreich 2014) to its 3S parent and screened 90
random spore plates from this 3S backcross (Materials and Methods). Given
that spores were plated at a density of ~100 colonies per plate, we estimate that
we examined an additional ~15,000 individuals as a part of this second screen.
This second screen was exclusively conducted at 21°C, as all three temperature
sensitivity classes can be recovered from this condition (S2.2 Table). Collected
individuals were then stringently phenotyped at 21, 30, and 37°C (Materials and
Methods). In combination with our preliminary screen, we recovered 544 and
466 rough backcross progeny from the BY and 3S backcrosses, respectively,
with 78.4% of these individuals classified as HS, MS, or NS (S2.1 and S2.4
Tables; Materials and Methods).
13
Figure 2.2: Bulk segregant mapping results for different temperature
sensitivity classes. (A) A rough BYx3S ira2Δ2933 F2 segregant was
backcrossed to the BY and 3S strains. Progeny from both of these
backcrosses were screened at 21, 30, and 37°C to identify individuals in
each temperature sensitivity class. Bulk segregant mapping was then
14
performed on each temperature sensitivity class in each backcross. (B)
Eight total loci were detected, seven of which we had previously resolved
to specific genes and one of which we cloned in the current paper. The
identities of these genes are stated above their corresponding loci. Alleles
detected from BY and 3S are shown in blue and orange, respectively,
while the ira2Δ2933 mutation is denoted in red. The color intensity of a
locus corresponds to its allele frequency in the bulk segregant mapping
data. A legend with the correspondence between allele frequency and
color intensity is provided at the bottom of the figure. To aid in
visualization, loci are depicted in the main text figures as having the same
widths.
2.3.2 Bulk segregant mapping of the three temperature sensitivity classes
We first attempted to determine the genetic bases of the three temperature
sensitivity classes using bulk segregant mapping by sequencing (Ehrenreich et al.
2010, Michelmore et al. 1991, Wenger et al. 2010) (Fig 2.2A). Between 51 and
126 individuals were pooled per backcross and temperature sensitivity class, and
each pool was sequenced to at least 114X coverage (S2.4 Table; Materials and
Methods). Across the six pools, eight distinct loci were detected using
MULTIPOOL (Edwards and Gifford 2012) (Figs 2.2B and S2.1; Materials and
Methods). Seven of these loci overlapped ira2Δ2933 or causal alleles that we
previously identified as contributors to the phenotype at 30°C: END3
BY
, FLO8
BY
,
MGA1
BY
, MSS11
BY
, SFL1
BY
, and TRR1
3S
(Taylor and Ehrenreich 2014, Taylor
and Ehrenreich 2015) (Fig 2.2B). We used genetic engineering to show that the
final locus, which was detected on Chromosome IX in the NS class, corresponds
to the 3S allele of FLO11, which encodes a cell surface glycoprotein that is
required for the rough phenotype (Taylor and Ehrenreich 2015) (Materials and
Methods). Specifically, we replaced the 3S version of the FLO11 coding region
with the BY allele in an NS individual from the BY backcross, and found the
resulting allele swap strain only exhibited rough morphology at 21 and 30°C (Fig
2.3).
15
Figure 2.3: FLO11
3S
is required for trait expression at 37°C. We
replaced FLO11
3S
with FLO11
BY
in a rough BY backcross segregant from
the NS class with the genotype END3
BY
FLO8
3S
FLO11
3S
ira2∆2933
MGA1
BY
MSS11
BY
SFL1
BY
TRR1
3S
. This allele replacement resulted in a
conversion from rough to smooth colony morphology specifically at 37°C.
In this picture, the FLO11
3S
individual is a segregant that has not been
genetically modified, while the FLO11
BY
individual is the same strain with
its FLO11 allele swapped. The phenotypes of both of these strains at 21,
30, and 37°C are shown.
Loci that contributed to temperature sensitivity could be distinguished from those
that did not based on the bulk segregant mapping results. FLO8
3S
and ira2Δ2933
were detected in every temperature sensitivity class (Fig 2.2B; S2.1 Note),
suggesting they are involved in the general expression of the rough phenotype
but do not influence its temperature sensitivity. In contrast, the other six involved
alleles — END3
BY
, FLO11
3S
, MGA1
BY
, MSS11
BY
, SFL1
BY
, and TRR1
3S
— were
each detected in just one or two of the classes, indicating they contribute to the
observed differences in temperature sensitivity (Fig 2.2B). Of these six alleles,
zero, three, and six were detected in the HS, MS, and NS classes, respectively,
and the three that were detected among MS individuals—END3
BY
, MSS11
BY
, and
TRR1
3S
— were also identified among NS individuals. These findings indicate
that the temperature sensitivity of the rough phenotype is largely controlled by
the same loci that were originally determined to underlie the trait’s expression at
16
30°C (Taylor and Ehrenreich 2014, Taylor and Ehrenreich 2015), and that the
rough phenotype’s temperature sensitivity can be reduced or even eliminated by
combining particular alleles at these loci. In particular, we note that individuals
with the genotype END3
BY
FLO8
3S
FLO11
3S
ira2Δ2933 MGA1
BY
MSS11
BY
SFL1
BY
TRR1
3S
did not exhibit any temperature sensitivity in our experiments.
2.3.3 Multiple genotypes underlie the HS and MS classes
Although bulk segregant mapping is known to be a statistically powerful
technique when large numbers of cross progeny are examined (Ehrenreich et al.
2010), it can fail to detect causal loci if distinct combinations of alleles that
interact with each other or the environment exhibit indistinguishable phenotypes
(Matsui et al. 2015, Taylor and Ehrenreich 2014). This phenomenon, which we
refer to here as ‘genotypic heterogeneity’, clearly occurred in the present data for
the MS class. As described in the introduction, we previously showed that two
specific multi-locus genotypes — END3
BY
FLO8
3S
ira2Δ2933 MSS11
BY
TRR1
3S
and END3
3S
FLO8
3S
ira2Δ2933 MGA1
BY
MSS11
BY
SFL1
BY
— express the rough
phenotype at 30°C (Taylor and Ehrenreich 2014) (Fig 2.4A). Phenotyping of
previously described segregants (Matsui et al. 2015, Taylor and Ehrenreich 2014)
revealed that both of these allele combinations lead to moderate temperature
sensitivity—i.e., expression of the trait at 21 and 30°C, but not 37°C (S2.2 Fig).
In our current data for the MS class, alleles required for both genotypes were
fixed, while alleles involved in just one of the two genotypes were merely
enriched or not even detected (Fig 2.2B; S2.2 Note). To directly show that our
sample of MS individuals was comprised of both previously identified multi-locus
genotypes, we genotyped 19 random MS segregants using restriction markers
for END3 and MGA1. This effort revealed that 14 and 5 of these individuals
possessed the END3
BY
FLO8
3S
ira2Δ2933 MSS11
BY
TRR1
3S
and END3
3S
FLO8
3S
ira2Δ2933 MGA1
BY
MSS11
BY
SFL1
BY
genotypes, respectively (S2.3
Note; Materials and Methods). After recognizing the genotypic heterogeneity
underlying the MS class (Fig 2.4A), we investigated whether the HS class, for
which only FLO8
3S
and ira2Δ2933 were detected by bulk segregant mapping
17
(Fig 2.2A), might also be genotypically heterogeneous. To examine this
possibility, we individually genotyped each HS segregant by performing low
coverage whole genome sequencing (Materials and Methods). We then used c
2
tests to scan the genomes of the ira2Δ2933 segregants in the HS class for pairs
of loci that exhibited correlated allele states (Materials and Methods). Such
associations might be expected if alleles at two loci participate in the same multi-
locus genotype. At a 1% false discovery rate threshold, we detected no
significant pairs of loci in the BY backcross and two significant pairs of loci in the
3S backcross: Chromosome XII-Chromosome XIII and Chromosome XIII-
Chromosome XV (S2.3 Fig; Materials and Methods). Both of these significant
locus pairs included a region of Chromosome XIII that overlapped MSS11. As for
the other two detected loci, the Chromosome XV region overlapped SFL1 and
the Chromosome XII region was novel relative to our past work (Taylor and
Ehrenreich 2014, Taylor and Ehrenreich 2015). As we have yet to determine the
causal gene at the Chromosomes XII locus (S2.4 Note), we hereafter refer to it
by its chromosome number: ‘XII’.
18
Figure 2.4: The HS and MS classes are each specified by two
predominant multi-locus genotypes. (A) We previously showed that two
distinct multi-locus genotypes underlie trait expression at 30°C (Taylor and
Ehrenreich 2014, Taylor and Ehrenreich 2015), and in this paper we
determined that these same genotypes underlie the MS class. In addition
to ira2Δ2933, these genotypes both involve FLO8
3S
and MSS11
BY
.
However, individuals carrying END3
BY
also require TRR1
3S
(‘END3
BY
-
dependent genotype’), while individuals carrying END3
3S
instead
require MGA1
BY
and SFL1
BY
(‘END3
3S
-dependent genotype’). We refer to
the fact that both of these multi-locus genotypes specify the same trait as
‘genotypic heterogeneity’. When such genotypic heterogeneity is present,
alleles involved in only one of the multi-locus genotypes can be masked.
For example, in the bulk segregant mapping data for the MS class, only
the alleles involved in the genotype that involves END3
BY
were detected.
(B) In the current data, we found evidence for genotypic heterogeneity in
the HS class. By partitioning individuals in the HS class from the 3S
backcross population and calculating their allele frequencies across the
genome, we determined that two predominant multi-locus genotypes
underlie this temperature sensitivity class. These genotypes
are FLO8
3S
ira2Δ2933 MSS11
BY
(‘MSS11
BY
-dependent genotype’) and
XII
3S
END3
BY
FLO8
3S
ira2Δ2933 MGA1
BY
MSS11
3S
SFL1
BY
(‘MSS11
3S
-
dependent genotype’). As described in the main text, XII
3S
refers to a
locus that was detected specifically among HS individuals
carrying MSS11
3S
. The same coloring scheme is used in this figure as
in Fig 2.2.
Detection of correlated loci among individuals from the 3S backcross suggested
that genotypic heterogeneity might have led to only FLO8
3S
and ira2Δ2933 being
identified by bulk segregant mapping focused on the HS class (Fig 2.2B).
Because MSS11 was present in both significant locus pairs, we split the 3S
backcross progeny by individuals’ genotypes at MSS11 and separately examined
19
genome-wide allele frequencies in the two resulting groups (Materials and
Methods). This analysis revealed that individuals with MSS11
BY
only need the
specific alleles ira2Δ2933 and FLO8
3S
to express the phenotype at 21°C,
whereas individuals with MSS11
3S
require a number of additional alleles to show
rough morphology under the same condition (Fig 2.4B; Materials and Methods).
Specifically, XII
3S
, END3
BY
, MGA1
BY
, and SFL1
BY
collectively enable FLO8
3S
ira2Δ2933 MSS11
3S
individuals to express the trait at 21°C (Fig 2.4B; Materials
and Methods). We validated this finding by performing allele replacements of
END3, MGA1, and SFL1 in a rough segregant from the 3S backcross that
possessed the XII
3S
END3
BY
FLO8
3S
ira2Δ2933 MGA1
BY
MSS11
3S
SFL1
BY
genotype (Materials and Methods). These replacements each resulted in the
engineered strain being incapable of expressing the rough phenotype at 21°C,
implying that the detected alleles have biologically meaningful effects on the trait
in the HS background involving MSS11
3S
(S2.4 Fig). Thus, two predominant
genotypes underlie the HS class: FLO8
3S
ira2Δ2933 MSS11
BY
and XII
3S
END3
BY
FLO8
3S
ira2Δ2933 MGA1
BY
MSS11
3S
SFL1
BY
.
2.3.4 Roles of detected alleles in modulating temperature sensitivity
Based on our genetic mapping efforts described in this paper, we have identified
seven environ- mentally responsive loci that influence the expression of the
rough phenotype across temperatures: XII, END3, FLO11, MGA1, MSS11, SFL1,
and TRR1. Furthermore, through our current and past efforts (Taylor and
Ehrenreich 2014, Taylor and Ehrenreich 2015), we have characterized five
predominant multi-locus genotypes involving these loci that exhibit different
levels of temperature sensitivity. These include two HS genotypes, two MS
genotypes, and a single NS genotype (Fig 2.5).
Comparison of the alleles involved in these five multi-locus genotypes suggests
that the seven environmentally responsive loci play different roles in modifying
the rough phenotype’s temperature sensitivity. In particular, MSS11 appears to
determine the range of temperatures at which the phenotype can be expressed
20
(Fig 2.5). Indeed, while both MSS11 alleles can facilitate expression of rough
morphology at 21°C, only individuals that carry MSS11
BY
have the genetic
potential to express the trait at 30 or 37°C (Fig 2.5). In contrast, the other
identified alleles appear to collectively increase or decrease temperature
sensitivity within the range established by MSS11 (Fig 2.5). These modifier
alleles show phenotypic effects that can depend on both genetic background and
temperature, although the degree of such dependencies varies. For example,
END3
BY
, MGA1
BY
, and SFL1
BY
were detected in one multi-locus genotype in
each temperature sensitivity class (Fig 2.5), and appear to influence temperature
sensitivity in a genetic back- ground-dependent manner (see S2.4 and S2.5 Figs,
as well as (Taylor and Ehrenreich 2014), for genetic engineering results
supporting this point for MGA1
BY
). The remaining alleles appear to have effects
on temperature sensitivity that depend on both genetic background and
temperature. Specifically, XII
3S
and FLO11
3S
were each detected in a single
multi-locus genotype and temperature sensitivity class, while TRR1
3S
was only
identified in the END3
BY
MSS11
BY
genetic background among MS and NS
individuals (Fig 2.5).
Figure 2.5: Comparison of multi-locus genotypes underlying
differences in temperature sensitivity among
BYx3S FLO8
3S
ira2Δ2933 segregants. Across our current study and past
work, we have identified five predominant multi-locus genotypes that
underlie the three temperature sensitivity classes. Excluding FLO8
3S
,
which is generally required for expression of the trait in the presence
21
of ira2Δ2933, these genotypes involve specific alleles at seven
environmentally responsive loci. Respectively, blue and orange indicate
the BY or 3S allele of a given gene participates in a specific multi-locus
genotype, while grey indicates that neither allele is required. Additionally,
brown lines emphasize how adding particular combinations of alleles to
a FLO8
3S
ira2Δ2933 MSS11
BY
genetic background can lead to reduction
or elimination of temperature sensitivity.
2.4 Conclusion
Across the current manuscript and our previous work (Taylor and Ehrenreich
2014, Taylor and Ehrenreich 2015), we have now described eight loci and five
predominant multi-locus genotypes that influence whether BYx3S segregants
carrying ira2Δ2933 can express the rough phenotype in at least one temperature.
Of the eight identified loci, only one—the transcription factor FLO8—does not
contribute to temperature sensitivity. This likely reflects the fact that FLO8
encodes a transcriptional activator that is required for expression of the rough
phenotype (Taylor and Ehrenreich 2014), and BY harbors a nonfunctional
version of this gene (Liu et al. 1996, Matsui et al. 2015). For the remaining seven
loci, it is difficult to distinguish between their effects on trait expression in general
and their effects on temperature sensitivity. Indeed, our results indicate that
genetic background and temperature together determine which alleles are
required for expression of the rough phenotype.
Our findings also suggest molecular mechanisms that might underlie the rough
phenotype’s temperature sensitivity. For instance, some of the alleles involved in
temperature sensitivity may have reduced biochemical activity and/or diminished
structural stability at 30 or 37°C. The most striking example of this is MSS11
3S
,
which can only facilitate trait expression at 21°C. Given that Mss11 acts as a
heterodimer with Flo8 (Kim et al. 2014), it might be that Flo8-Mss11
3S
does not
dimerize well, poorly binds DNA, or is unable to stimulate RNA polymerase II
activity at 30 and 37°C. Supporting such possibilities, we determined that the
22
causal variant in MSS11 is an isoleucine to serine amino acid change that occurs
in the LisH motif required for dimerization with Flo8 (Kim et al. 2014) (S2.6 Fig).
3S and roughly half of the other available sequenced S. cerevisiae isolates carry
the derived version of this site, which is the serine allele that results in
temperature sensitivity (S2.6 Fig). The data also suggest FLO11, which plays a
crucial role in cell-cell and cell- surface adhesion (Bruckner and Mosch 2012, Lo
and Dranginis 1996), harbors a temperature sensitive polymorphism in its coding
region. Instability of Flo11
BY
at 37°C could result in temperature-dependent
suppression of the rough phenotype among FLO11
BY
individuals and would
explain why FLO11
3S
is required by the NS class. Temperature sensitivity at the
phenotypic level may also be caused by the combined effects of genetic
background and temperature on Ras signaling and Ras-dependent gene
regulation. We note that Flo8-Mss11, Mga1, and Sfl1 are all Ras-regulated
transcription factors (Robertson and Fink 1998). Furthermore, Flo8-Mss11 and
Sfl1 are known to play particularly important roles in the expression of
multicellular phenotypes in yeast, as they compete to bind DNA and are
regulated in an antagonistic manner at the posttranslational level by Protein
Kinase A, the effector kinase of the Ras cascade (Bruckner and Mosch 2012,
Pan and Heitman 2002, Toda et al. 1987). In fact, we previously showed that
ira2Δ2933 reveals the rough phenotype in certain genetic backgrounds by
reducing Sfl1-mediated transcriptional repression (Taylor and Ehrenreich 2015).
Viewing our current work in light of our past findings suggests that the
temperature sensitivity described in this paper results in part from genotype-
temperature combinations that conditionally reduce Ras signaling and/or Ras-
regulated gene expression. Consistent with this possibility, others have noted
functional relationships between Ras signaling and both End3 (Gourlay and
Ayscough 2006) and the oxidative stress response, of which Trr1 is a component,
in yeast (Charizanis et al. 1999, Gourlay and Ayscough 2006). Additionally,
temperature has been reported to affect levels of Ras signaling in human
fibroblast cultures (Chan et al. 1999).
23
Our results also shed light on the genetic basis of phenotypic capacitance—i.e.,
the uncovering of cryptic genetic variation by environmental or mutational
perturbation (Bergman and Siegal 2003, Gibson and Dworkin 2004, Hermisson
and Wagner 2004, Paaby and Rockman 2014, Queitsch et al. 2002, Rutherford
and Lindquist 1998). As we previously described, all of the polymorphisms that
influence rough morphology in the BYx3S ira2Δ2933 cross can be considered
cryptic genetic variants, as they do not cause the rough phenotype under
standard conditions in the absence of the IRA2 mutation (Taylor and Ehrenreich
2015). Here, we have shown the multi-locus genotypes that provide the genetic
potential for ira2Δ2933 to uncover the rough phenotype differ across
temperatures. Moreover, we have demonstrated that these multi-locus genotypes
that facilitate phenotypic capacitance vary not only in their initial temperature
sensitivities, but also in their potential to reduce their temperature sensitivities
through segregating genetic variation (Fig 2.5). This latter finding has potential
relevance for our understanding of genetic assimilation, the process by which
environmentally induced traits are converted into constitutively expressed
phenotypes by natural selection (Badyaev 2005, Ehrenreich and Pfennig 2016,
Hermisson and Wagner 2004, Lande 2009, Masel and Siegal 2009, Masel and
Trotter 2010, Pfennig and Ehrenreich 2014, Pigliucci and Murren 2003, Pigliucci
et al. 2006, Siegal and Bergman 2002, Siegal and Leu 2014, Visser et al. 2003,
Waddington 1942), as our results provide an example of the genetic architecture
that might underlie this phenomenon.
Lastly, our study provides technical insights into research aimed at determining
the genetic basis of GxE. First, we have shown that selective genotyping of
individuals that exhibit particular levels of environmental sensitivity can identify
multi-locus genotypes that cause GxE, rather than just individual contributing loci.
Second, we have demonstrated that genotypic heterogeneity can complicate
efforts to genetically dissect GxE and have described a strategy to overcome this
challenge. Third, we have illustrated how characterizing the genetic basis of GxE
at the resolution of multi-locus genotypes can clarify the different roles that
24
contributing loci play in altering trait expression across conditions. These
technical insights that have emerged from our work will likely be relevant to future
studies of GxE in other species and traits.
2.5 Materials and Methods
2.5.1 Generation of backcross segregants for the preliminary screen and genetic
mapping
The rough BYx3S MATa F2
segregant used for genetic mapping in (Taylor and
Ehrenreich 2014) was mated to MATα versions of both BY and 3S. Diploid
zygotes were obtained from each backcross mating using micro- dissection, and
then sporulated at 21°C using standard yeast sporulation methods (F 1991).
Spore cultures were digested with β-glucoronidase and random MATa spores
were selected on yeast nitrogen base (YNB) plates containing canavanine using
the Synthetic Genetic Array (SGA) marker system (Tong et al. 2001), as
described previously (Matsui et al. 2015, Taylor and Ehrenreich 2014).
Implementation of the SGA system was possible because the MATa F2
segregant used in backcrossing possessed the markers can1Δ::STE2pr- SpHIS5
and his3Δ (Tong et al. 2001), and matings were performed to BY and 3S MATα
his3Δ strains. Spores were plated at low density (~100 spores per plate) so that
individual colonies could be easily distinguished. After five days of growth on
YNB + canavanine plates, colonies were replicated onto yeast extract-peptone-
ethanol (YPE) plates. These YPE plates were incubated at the specified
temperature (21, 30, or 37°C) for five days and then screened by eye for colonies
exhibiting the rough phenotype. Strains identified as rough were picked from the
plates, inoculated in liquid yeast extract-peptone-dextrose (YPD) media, and
grown overnight at 30°C. Freezer stocks of the rough backcross segregants were
generated by mixing equal volumes of 40% glycerol solution with a portion of the
liquid YPD cultures and then storing these cultures at -80°C.
2.5.2 Phenotyping of rough segregants at multiple temperatures
Cells from the freezer stocks described above were inoculated into 800 μl of
25
liquid YPD media. These YPD cultures were grown for two days at 30°C and
then manually pinned onto three YPE plates. Of these three plates, one was
incubated at 21°C, one was incubated at 30°C, and one was incubated at 37°C.
Individuals were scored manually on a scale of 0 to 5 based on their degree of
expression of the rough phenotype. The low and high ends of this scale
correspond to individuals that were smooth and rough, respectively. Individuals
with intermediate scores were bumpy to varying degrees, which may reflect
weaker expression of the phenotype. Three biological replicate cultures were
screened for each segregant. For the purposes of the paper, we viewed an
individual’s phenotype at a given temperature as their median score obtained for
that temperature. The results described in the paper are based on individuals
that clearly and consistently showed smooth or rough phenotypes at a given
temperature. We considered median scores between 0 and 1 or 4 and 5 as
clearly indicating the smooth and rough phenotypes, respectively. Phenotype
data are reported in S1 Table.
2.5.3 Bulk segregant mapping of temperature sensitivity
Cultures were inoculated from the freezer stocks described above into 800 μl of
liquid YPD. After two days of growth at 30°C, 100 μl of each culture from the
same backcross and temperature sensitivity class were mixed together, and DNA
was extracted from these pools using the Qiagen DNeasy Blood and Tissue kit.
DNA sequencing libraries were then constructed using the Illumina Nextera kit.
Each library was barcoded with a distinct sequence tag to facilitate multiplex
sequencing. Libraries were mixed in equimolar fractions and sequenced on an
Illumina NextSeq machine using 75x75 or 150x150 base pair reads. Sequencing
reads were mapped to both the reference genome for S288c, which is the
progenitor of BY, and the 3S draft genome (available through
http://www.yeastgenome.org) using the Burrows-Wheeler Aligner (BWA) version
7 with options mem -t 20 (Li and Durbin 2009). We then used SAMtools to obtain
mpileup files for each sample (Li et al. 2009). Based on these mpileups, we
determined that the pools were sequenced to an average per site coverage of at
26
least ~114X (S2.4 Table). Genome-wide allele frequencies were determined at
36,756 high confidence SNPs that were previously identified by mapping Illumina
sequencing reads for 3S to the S288c reference genome (Taylor and Ehrenreich
2014, Taylor and Ehrenreich 2015). Because roughly half of the genome is fixed
in each backcross, we subsetted out the data for segregating regions in each
backcross and analyzed each of these regions individually. To identify
significantly enriched loci, we used MULTIPOOL (Edwards and Gifford 2012)
with the settings: replicates mode, 3,300 bp centimorgans, 100 bp bins.
Significant loci were defined as genomic regions that had a maxi- mum LOD
score of at least 5 for a span of at least 20 kb. Confidence intervals were
estimated as the bounds of a locus that correspond to a 2 LOD drop from the
point of maximal significance. To generate allele frequency plots, the data was
smoothed by averaging allele frequency over sliding windows of 25 SNPs.
2.5.4 Genotyping of MS individuals using PCR and restriction digestion
DNA was extracted from 19 randomly chosen BY backcross segregants from the
MS class using the Qiagen DNeasy Blood and Tissue kit. Small regions of these
genes that contained a SNP were amplified by PCR and then digested with
restriction enzymes (see S2.5 Table for a description of specific reagents). The
amplified regions were chosen so that one parental allele would be cut a single
time while the other parental allele would not be cut at all. Each diagnostic
restriction digest was tested on BY and 3S. Digested PCR products were
examined on a 1.5% agarose gel containing ethidium bromide to determine each
individual’s genotype at a given locus.
2.5.5 Detailed genetic characterization of the HS class
Respectively, 68 and 89 HS segregants were independently prepared for
sequencing from the backcrosses to BY and 3S. Individual cultures were
generated by inoculating 800 μl of liquid YPD with cells from the freezer stocks
described above. After two days of growth at 30°C, DNA was extracted from
each individual culture using the Qiagen DNeasy 96 Blood and Tissue kit.
27
Libraries were prepared for sequencing using the Illumina Nextera kit, with each
individual receiving a unique sequence barcode. Sequencing was performed on
an Illumina NextSeq machine using 75x75 base pair reads. Illumina reads were
then mapped to the S288c genome, and mpileups were generated from these
alignments using BWA (Li and Durbin 2009) and SAMtools (Li et al. 2009) in the
same manner described earlier for the bulk segregant mapping data. Individuals
with an average per site cover- age below 1.02X were excluded from subsequent
analyses. Respectively, one and five individuals were excluded from the BY and
3S backcrosses due to low coverage. Furthermore, 16 sequenced individuals
from the BY backcross were excluded from downstream analysis because they
possessed a wild type IRA2 (S2.1 Note). As previously described (Taylor and
Ehrenreich 2014), a Hidden Markov Model (HMM) was used to determine the
haplotype of each segregant from the aforementioned 36,756 SNP differences
between the BY and 3S genomes. As with the bulk segregant data, we only per-
formed our statistical tests on regions of the genome that segregated in a given
backcross. Furthermore, to reduce our total number of statistical tests, we
collapsed linked SNPs that showed the same pattern of inheritance across
backcross segregants into unique segregating regions. There were 1,399 and
767 such segregating regions in the BY and 3S backcross populations,
respectively. We used the R statistical programming environment to identify pairs
of loci on different chromosomes that showed correlated alleles states based on
χ
2
tests. Pairs of loci were considered statistically significant if they exhibited a
point-wise false discovery rate (q-value) of 1% or less, as determined when p-
values for all segregating regions were converted into q-values by the qvalue()
package in R (Dabney et al. 2015, Storey and Tibshirani 2003). Regions within
20,000 bases of the ends of chromosomes were excluded from this analysis due
to the problems in mapping Illumina reads to telomeric regions. Based on the
results from the initial scan for correlated loci, HS individuals from the 3S
backcross were subsetted by their genotype at MSS11 and genome-wide allele
frequency plots were generated from the aforementioned HMM tables. Allele
frequencies were averaged over 25 SNP sliding windows, and loci were called as
28
significant in a given MSS11 background if they exhibited allele frequencies
below 10% or greater than 90%.
2.5.6 Genetic engineering
Adaptamer-mediated allele replacement (Matsui et al. 2015, Reid et al. 2002)
was used to alter the allele state at MGA1 and FLO11 in a NS segregant with the
genotype END3
BY
FLO8
3S
FLO11
3S
ira2Δ2933 MGA1
BY
MSS11
BY
SFL1
BY
TRR1
3S
, as well as END3, MGA1, and SFL1 in a HS segregant with the genotype
XII
3S
END3
BY
FLO8
3S
ira2Δ2933 MGA1
BY
MSS11
3S
SFL1
BY
. Using PCR, an
amplicon of the gene of interest was tailed at the 3’ end with the 5’ end of the
kanMX cassette, and an amplicon of kanMX was tailed on the 3’ end with the
region immediately downstream of the gene (Matsui et al. 2015). The two PCR
products were co-transformed into a given strain using the lithium acetate
method (Daniel Gietz and Woods 2002) and plated on YPD agar containing
G418 to screen for successful integration. Colonies that showed G418 resistance
were then checked by PCR and Sanger sequencing to ensure they harbored the
allele replacement. To clone the causal nucleotide in MSS11, adaptamer-
mediated allele replacement was performed multiple times using MSS11
amplicons that spanned only a part of the gene’s coding region. These specific
engineerings are shown in more detail in S2.6 Fig. Each of these engineerings to
clone the causal variant in MSS11 were checked by Sanger sequencing. The
gene deletions described in S2.4 Note were performed by replacing a gene of
interest with the CORE cassette (Storici et al. 2001). Regions corresponding to
60 bases upstream and downstream of the target gene were tailed to the CORE
cassette using PCR. This product was transformed into cells using the lithium
acetate method (Daniel Gietz and Woods 2002), and selection with G418 was
used to screen for integration of the cassette. PCR was then used to verify that
deletion strains recovered from the G418 selection lacked the gene of interest.
29
2.6 Supporting Materials
S2.1 Figure: Bulk segregant mapping results for each temperature
sensitivity class. Genome-wide allele frequencies are presented for the
(A) HS, (B) MS, and (C) NS mapping populations. For each of the three
plots, allele frequencies in the BY and 3S backcrosses are depicted in the
top and bottom panels, respectively. Approximately half of the genome
segregates in each back- cross and the regions that segregate in one
30
backcross are fixed in the other. Loci that differ significantly from the
expected frequency of 0.5 (Materials and Methods) are labeled with
highlighted bars: significantly enriched loci from the BY and 3S parents
are highlighted in blue and orange, respectively. Two selectable markers
on Chromosomes III and V correspond to MATa and can1Δ::STE2-
SpHIS5, respectively, and are highlighted in grey (Materials and
Methods). The ira2Δ2933 allele was a spontaneous mutation that
occurred on the 3S chromosome of a BY/3S diploid and is highlighted in
red. The results shown in this figure are summarized in the main text in
Fig 2.2B.
31
S2.2 Figure: Two multi-locus genotypes express the rough
phenotype exclusively at 21 and 30°C. Only alleles involved in the
END3
BY
FLO8
3S
ira2Δ2933 MSS11
BY
TRR1
3S
genotype were detected in
the bulk segregant mapping data for the MS class. However, we have
previously shown that the END3
3S
FLO8
3S
ira2Δ2933 MGA1
BY
MSS11
BY
SFL1
BY
genotype can also lead to expression of the rough phenotype at
30°C. Here, we provide examples of strains carrying the (A) END3
BY
FLO8
3S
ira2Δ2933 MSS11
BY
TRR1
3S
and (B) END3
3S
FLO8
3S
ira2Δ2933
MGA1
BY
MSS11
BY
SFL1
BY
genotypes. Both of these strains express the
rough phenotype at 21 and 30°C, but not 37°C.
32
S2.3 Figure: Genome-wide scan for pairs of loci that show correlated
allele states among sequenced individuals from the HS class. We
performed χ
2
tests on all possible pairs of segregating genomic segments
in the (A) BY and (B) 3S backcross populations. No locus pairs were
detected in the BY backcross, while two pairs of loci were detected in the
3S backcross. One of these pairs corresponds to MSS11 and SFL1, while
the other corresponds to MSS11 and a new locus on Chromosome XII
that was not identified in our past work.
33
S2.4 Figure: Multiple alleles are required for rough morphology in a
MSS11
3S
HS individual. Here, we performed allele replacements to verify
that END3
BY
, MGA1
BY
, and SFL1
BY
play causal roles in enabling the XII
3S
END3
BY
FLO8
3S
ira2Δ2933 MGA1
BY
MSS11
3S
SFL1
BY
genotype to
express the rough phenotype exclusively at 21°C. We used genetic
engineering to swap the BY allele of END3, MGA1, or SFL1 with the 3S
allele in a segregant carrying the aforementioned MSS11
3S
-dependent HS
genotype (Materials and Methods). Each allele swap resulted in loss of
the rough phenotype at 21°C.
34
S2.5 Figure: MGA1
BY
is required for trait expression at 37°C in an NS
genetic background. We replaced MGA1
BY
with MGA1
3S
in an END3
BY
FLO8
3S
FLO11
3S
ira2Δ2933 MGA1
BY
MSS11
BY
SFL1
BY
TRR1
3S
genetic
background. This allele replacement resulted in a conversion from rough
to smooth colony morphology specifically at 37°C.
35
S2.6 Figure: Cloning of the causal variant in MSS11. (A) We fine-mapped the
causal nucleotide in MSS11 by performing multiple genetic engineerings in which
only part of the gene was replaced in an END3
3S
FLO8
3S
ira2Δ2933 MGA1
BY
MSS11
BY
SFL1
BY
genetic background, as indicated by the portion of the gene
shown in orange. Vertical bars indicate the locations of SNPs differentiating BY
and 3S. The causal SNP is denoted by a black triangle and a scale bar is
provided in base pairs. (B) The causal nucleotide in MSS11 results in an
isoleucine to serine amino acid substitution in the LisH domain required for Flo8-
Mss11 dimerization. 3S carries the derived, serine allele of this amino acid.
Inspection of the MSS11 genotypes of other sequenced S. cerevisiae isolates
revealed that roughly 56% of strains also harbor the serine allele. Mss11 protein
sequence data were obtained from the Saccharomyces Genome Database
36
(http:// www.yeastgenome.org). A scale bar is provided in amino acids.
S2.1 Table. Phenotypes of BYx3S ira2Δ2933 backcross segregants in
preliminary and secondary screens.
Available at https://doi.org/10.1371/journal.pgen.1005929
S2.2 Table. Initial screen for rough morphology among segregants isolated at
three different temperatures. Random spore plates from the BY backcross were
screened for rough colonies at 21, 30, or 37°C (Materials and Methods). Rough
individuals isolated from each of the temperatures were then examined at all
three temperatures for the ability to express the phenotype.
Collection
Temperature
(°C)
Rough
Segregants
Collected
Rough at
21°C
Rough at
30°C
Rough at
37°C
21 173 173 66 56
30 107 102 107 49
37 72 68 47 72
S2.3 Table. Classification of BY backcross segregants obtained from the
preliminary screen into the three temperature sensitivity classes. These data are
based on the same individuals and phenotyping results described in S2.2 Table.
Collection
Temperature
(°C)
Rough
Segregants
Collected
Number
of HS
Segregants
Number
of MS
Segregants
Number
of NS
Segregants
% Segregants
in HS, MS, or
NS classes
21 173 43 27 31 60.7
30 107 40 48 91.6
37 72 47 65.3
37
S2.4 Table. Bulk segregant mapping populations were generated for each
temperature sensitivity class. Subsets of segregants from each backcross and
sensitivity class were gathered for pooled sequencing. Of the 1,010 collected
segregants at 21°C, 78.4% of them exhibited the HS, MS, or NS phenotype. DNA
from between 51 and 131 individuals of each backcross and class was combined
to form six pools. Each pool was sequenced to a minimum coverage of 114X.
Sensitivity Class
Collected Sequenced Coverage
BY 3S BY 3S BY 3S
HS 257 217 82 131 342.88 378.46
MS 74 139 55 101 209.63 406.09
NS 53 52 51 51 114.79 397.37
Other 160 58 0 0 NA NA
S2.5 Table. Primers used in this study.
Available at https://doi.org/10.1371/journal.pgen.1005929
38
S2.1 Note. The ira2Δ2933 allele was highly enriched (86.1% frequency) but not
completely fixed among HS individuals in the BY backcross. Based on whole
genome sequencing described later in the paper, we identified 16 IRA2
BY
individuals. These individuals were excluded from further consideration, as our
goal in this paper was to characterize GxE in an ira2Δ2933 background. We also
note that the number of these individuals was too low to enable detection of loci
that enable rough morphology in the absence of ira2Δ2933.
S2.2 Note. Because the END3
BY
- and END3
3S
-dependent genotypes require five
and six alleles, respectively, the latter genotype is expected to only occur half as
often. In practice, the bias towards the END3
BY
-dependent genotype is typically
even higher. This is because a locus that confers a selective advantage during
random spore isolation in the BYx3S cross is closely linked to END3, with the BY
allele of this locus conferring a benefit (Taylor and Ehrenreich 2014; Taylor and
Ehrenreich 2015).
S2.3 Note. Nineteen MS individuals from the 3S backcross population were
randomly chosen, and genotyped at END3 and MGA1. Fourteen of these
individuals carried END3
BY
, and the BY and 3S alleles of MGA1 were present in
equal frequencies among these END3
BY
MS individuals. In contrast, all five
segregants carrying END3
3S
also harbored MGA1
BY
. These are consistent with
our past results that two genotypes—END3
BY
FLO8
3S
ira2Δ2933 MSS11
BY
TRR1
3S
and END3
3S
FLO8
3S
ira2Δ2933 MGA1
BY
MSS11
BY
SFL1
BY
—underlie the
MS class, as MGA1
BY
co-segregates with END3
3S
but exhibits no such
association with END3
BY
.
S2.4 Note. The Chromosome XII interval was delimited to a 34,519 base region.
Within this interval, the candidate genes HAP1, HSP60, GSY2, LCB5, PDR8,
SYM1, YLR257W, and YPT6 were independently deleted in a 3S backcross
segregant expressing the HS phenotype and carrying the XII
3S
END3
BY
FLO8
3S
ira2Δ2933 MGA1
BY
MSS11
3S
SFL1
BY
genotype (Materials and Methods). None
39
of these gene deletions resulted in a loss of rough morphology. This indicates
that either the causal allele at XII
3S
is a loss-of-function polymorphism or none of
the tested genes are the causal gene at this locus.
40
Chapter 3: Diverse genetic architectures lead to the same cryptic
phenotype in a yeast cross
This chapter appears as published in Nature Communications, 2016. 7:11669
3.1 Abstract
Cryptic genetic variants that do not typically influence traits can interact
epistatically with each other and mutations to cause unexpected phenotypes. To
improve understanding of the genetic architectures and molecular mechanisms
that underlie these interactions, we comprehensively dissected the genetic bases
of 17 independent instances of the same cryptic colony phenotype in a yeast
cross. In eight cases, the phenotype resulted from a genetic interaction between
a de novo mutation and one or more cryptic variants. The number and identities
of detected cryptic variants depended on the mutated gene. In the nine remaining
cases, the phenotype arose without a de novo mutation due to two different
classes of higher-order genetic interactions that only involve cryptic variants. Our
results may be relevant to other species and disease, as most of the mutations
and cryptic variants identified in our study reside in components of a partially
conserved and oncogenic signaling pathway.
41
3.2 Introduction
Cryptic genetic variants are standing polymorphisms that only show phenotypic
effects under atypical conditions, such as when specific genes are mutated, rare
combinations of segregating alleles are generated, or the environment markedly
changes (Bergman and Siegal 2003, Gibson 2009, Gibson and Dworkin 2004,
Jarosz and Lindquist 2010, Le Rouzic and Carlborg 2008, Paaby and Rockman
2014, Queitsch et al. 2002, Rutherford and Lindquist 1998). Given a stable
environment, the uncovering of cryptic variation can be viewed as a form of
epistatic (or genetic) interaction, in which the phenotypic effects of cryptic
variants depend on the mutations and standing polymorphisms with which they
co-occur (Masel 2013, Richardson et al. 2013, Taylor and Ehrenreich 2015).
Empirical and theoretical evidence suggests that these epistatic interactions may
involve large numbers of loci that do not individually influence phenotype, but
collectively exhibit significant trait effects – so-called ‘higher-order genetic
interactions’ (Chandler et al. 2014, Hermisson and Wagner 2004, Richardson et
al. 2013, Taylor and Ehrenreich 2014, Taylor and Ehrenreich 2015). However,
the genetic architectures and molecular mechanisms that underlie these epistatic
interactions have yet to be characterized in detail (Dworkin 2005).
We have developed an experimental system in the budding yeast
Saccharomyces cerevisiae that provides a powerful resource for identifying and
studying epistatic interactions among cryptic variants and mutations (Lee et al.
2016, Taylor and Ehrenreich 2014, Taylor and Ehrenreich 2015). Specifically,
under the standard temperature used for culturing yeast (30°C), the lab strain
BY4716 (‘BY’) (Liti et al. 2009), a haploid derivative of the clinical isolate
322134S (‘3S’) (Liti et al. 2009), and their wild-type recombinant progeny show a
‘smooth’ colony phenotype (Taylor and Ehrenreich 2014). However, induced and
spontaneous mutations can cause BYx3S recombinants possessing particular
combinations of segregating cryptic variants to exhibit an alternative, ‘rough’
colony morphology (Lee et al. 2016, Taylor and Ehrenreich 2014, Taylor and
Ehrenreich 2015)
(Fig. 3.1a). Because the rough phenotype is only expressed
when epistatic interactions among cryptic variants and mutations occur, the
42
phenotype can be used as a reporter to detect these interactions. Furthermore,
once an individual rough segregant is obtained, the specific combinations of
cryptic variants and mutations that cause this individual to express the phenotype
can be determined through genetic mapping in backcrosses (Fig. 3.1b).
Figure 3.1: Overview of study. (a) Representative images of rough and
smooth colonies in the BY × 3S cross. (b) Backcrossing strategy used to
determine the genetic basis of the rough phenotype in a given
F2 segregant.
In the most thoroughly characterized example of the rough phenotype’s genetic
basis, we demonstrated that a de novo frameshift mutation in the Ras negative
regulator IRA2 (ira2∆2933) can alter colony morphology when it co-occurs with
one of two specific combinations of cryptic variants in six genes (Lee et al. 2016,
Taylor and Ehrenreich 2014, Taylor and Ehrenreich 2015). In more recent work,
we showed that the genetic architectures that enable ira2∆2933 to induce the
rough phenotype vary across temperatures (Lee et al. 2016). For instance, a
single, specific combination of seven cryptic variants facilitates ira2∆2933-
dependent rough morphology at 37°C. However, at room temperature (21°C),
two BYx3S ira2∆2933 multi-locus genotypes—one involving two cryptic variants
and the other involving six cryptic variants—show rough morphology. Moreover,
some BYx3S segregants appear to express the phenotype at 21°C independent
43
of any mutation (Lee et al. 2016). These results suggest that, out of the three
temperatures we have previously examined, the greatest diversity of genetic
architectures underlying the rough phenotype occurs at 21°C.
Here we take advantage of the rough colony morphology system to conduct the
first large-scale examination of epistatic interactions among cryptic variants and
de novo mutations. To perform this work, we screen 4100 independent BYx3S F2
crosses for the rough phenotype at 21°C. Through this screen, we obtain 17
independent occurrences of the trait. We then use genetic mapping in
backcrosses to comprehensively identify genetic factors that contribute to each
instance of the rough phenotype. In addition, we clone the specific genes
underlying most of the detected loci, thereby obtaining new insights into the
molecular mechanisms that uncover cryptic variation.
On the basis of these efforts, we find that the genetic architectures that can lead
to the rough phenotype at 21°C are quite diverse. About half of the instances of
the rough phenotype represent epistatic interactions among cryptic variants and
de novo mutations. These cases involve six different mutated genes, as well as
between one and nine cryptic variants. The remaining instances of the rough
phenotype do not require de novo mutations and instead arise due to higher-
order genetic interactions that only involve cryptic variants. These mutation-
independent cases fall into two classes that are distinguished by the occurrence
of recombination within a specific ~1.3-kb genomic interval in the promoter of a
required cell surface protein. We also demonstrate that the vast majority of
genetic factors involved in our study influence Ras signaling or Ras-dependent
transcriptional regulation. Thus, our results not only shed light on the forms of
epistatic interactions that can enable cryptic variants to influence phenotype, but
also implicate complex changes in gene regulation in a partially conserved and
disease-associated signaling pathway as the main source of these interactions.
3.3 Results
3.3.1 Genetic mapping of 17 independent cases of rough morphology
44
We mated BY and 3S 106 independent times, and screened >100,000 haploid F2
segregants derived from these matings at 21°C (Methods). Through this screen,
we obtained 17 rough segregants that were descended from different BYx3S
diploids and thus represented biologically independent occurrences of the same
phenotype (Fig. S3.1; Methods). To determine the genetic bases of these
distinct instances of the rough phenotype, we backcrossed each rough segregant
to both BY and 3S. Bulk segregant mapping by sequencing (Ehrenreich et al.
2010, Michelmore et al. 1991, Wenger et al. 2010)
was performed on pools
containing between 46 and 95 rough F2B progeny (Fig. 3.1b; Table S3.1). These
pools of rough backcross segregants were sequenced to an average genomic
coverage of 246 (Methods). Control pools were also generated for each
backcross using lawns containing millions of random progeny that had not been
selected based on their colony morphologies (Methods). These control pools
were sequenced to an average genomic coverage of 162 (Methods).
We separately analyzed the rough and control populations for each backcross
using MULTIPOOL (Edwards and Gifford 2012)
and excluded any loci detected
in the control pools from further consideration (Methods; Table S3.2; Table 3.3).
On the basis of this procedure, we identified between 2 and 10 loci per rough
segregant, with an average of 6.6 loci (Table S3.4). These detected genomic
regions corresponded to 8 de novo mutations and 18 distinct segregating loci
that harbor cryptic variants (Fig. 3.2a,b; Methods; Fig. S3.2). Only one of these
cryptic variants, which we previously localized to the 3S version of the Ras-
regulated transcriptional activator FLO8 (Taylor and Ehrenreich 2014), was fixed
in all our genetic mapping experiments (Fig. 3.2a,b). Given that BY carries a null
allele of FLO8 (Liu et al. 1996, Matsui et al. 2015), this finding indicates that a
functional copy of FLO8 is necessary for the phenotype’s expression. The
remaining loci were detected on average 4.1 times, with a range between 1 and
15.
45
Figure 3.2: Genetic mapping results. (a) Genetic mapping data for
each rough segregant is shown horizontally. The mapping results
correspond to data from backcrosses of each segregant to BY and 3S,
as depicted in Fig. 3.1b. Vertical bars represent detected loci, with blue
and orange coloring indicating cryptic variants from BY and 3S,
respectively. The width of each locus corresponds to the region of that
locus that exhibits a logarithm of odds score of at least 5. De
novo mutations are shown in red, with the specific mutated gene noted
to the right of the panel. The allele frequencies of detected loci in a
given backcross mapping population are provided, with the color scale
46
illustrated to the right of the figure. (b) The number of times each cryptic
variant was detected across the different mapping populations is plotted.
Counts were determined by summing the number of times each
segregating marker was detected in backcrosses to a particular parent.
Results corresponding to de novo mutations were excluded.
3.3.2 Genes that can lead to rough morphology when mutated
The eight de novo lesions were comprised of six small deletions and two point
mutations (Fig. S3.3; Table S3.5). The six genes harboring these changes fell
into three functional classes (Table S3.5): negative regulators of Ras signaling
(GPB1, IRA1 and IRA2), non-essential components of RNA polymerase II that
act in the mediator complex (SSN3 and SSN8) and a gene of unknown function
whose product localizes to bud tips during cell division (IRC8) (Cherry et al.
2012). On the basis of gene deletion experiments (Methods), seven of the
mutations were found to be null alleles (Fig. S3.4). The only partial loss-of-
function mutation was a point mutation in GPB1 (Fig. S3.4; Table S3.5). This is
consistent with our previous finding that ira2∆2933 is also a partial loss-of-
function allele (Taylor and Ehrenreich 2014), as Gpb1 and Ira2 physically interact
to coregulate Ras signaling, and these lesions in GPB1 and IRA2 fall within
(GPB1) or truncate (IRA2) protein–protein interaction domains needed for Gpb1–
Ira2 binding (Phan et al. 2010)
(Table S3.5).
When considered with our previous work, in which ira2∆2933 and complete
knockout of the Ras-regulated transcriptional repressor SFL1 were shown to
reveal the rough phenotype (Taylor and Ehrenreich 2014, Taylor and Ehrenreich
2015), we have now identified seven genes with the potential to alter colony
morphology when mutated. The identities of these genes, as well as the fact that
the mediator complex is regulated by Ras signaling (Chang et al. 2004)
and
physically interacts with Sfl1 to inhibit transcription (Song and Carlson 1998),
suggests that most of the mutations we have identified influence transcriptional
regulation by the Ras pathway. This finding supports our recent discovery that
expression of the rough phenotype in the BYx3S cross requires transcriptional
47
derepression of one or more Ras target genes (Taylor and Ehrenreich 2015).
However, even though the identified mutations each likely lead to derepression,
they show significant differences in the combinations of cryptic variants they
interact with to exert their effects. Specifically, between one and nine cryptic
variants were detected in backcross populations derived from the mutants (Fig.
3.2a; Table S3.4). As we discuss later, these differences in genetic complexity
among rough segregants may relate to the rough phenotype’s underlying gene
regulatory network.
3.3.3 Genetics of mutation-independent rough colony morphology
The nine other rough segregants did not harbor de novo mutations. This supports
a previous finding that some individuals in the BYx3S cross may show rough
morphology despite lacking ira2∆2933 or other mutations (Lee et al. 2016). Five
or more loci were detected in each of these cases, implying they occurred due to
higher-order genetic interactions that only involve cryptic variants. All of the
mapping populations lacking mutations were fixed for MSS11
BY
, which encodes
an activator that heterodimerizes with Flo8 (Kim et al. 2014, Kim et al. 2004).
Two-thirds of these individuals also possessed intragenic recombinations within a
1.3-kb region preceding the transcription start site of FLO11, which encodes a
cell surface glycoprotein that must be transcribed for the rough phenotype to be
expressed (Taylor and Ehrenreich 2015)
(Figs 3.2a and 3.3a). Specifically, the
BY promoter and the 3S coding region of FLO11 harbor cryptic variants that
together enable expression of the rough phenotype in the presence of other
cryptic variants that segregate in the BYx3S cross (Fig. 3.3b). The remaining
third of the wild-type rough segregants did not exhibit dependence on particular
FLO11 haplotypes. Instead, mapping data for these cases consistently showed
enrichment for cryptic variants on chromosomes V, VII, XIV and XV, as well as
other loci that differed among the individuals (Fig. 3.2a).
48
Figure 3.3: Locations of independent recombination events within
the FLO11 locus. (a) Genotypes of rough segregants that possess
a FLO11 intragenic recombination event are shown, along with
annotations of this locus from the Saccharomyces Genome Database
(Cherry 2012). ICR1 and PWR1 encode noncoding RNAs. BY and 3S
segments are shown in blue and orange, respectively. (b)
Representative images of allele replacements in the coding (‘c’) and
promoter (‘p’) regions of FLO11, which verify the presence of at least
two cryptic variants in different regions of this gene, are provided.
Most identified cryptic variants affect the Ras pathway. We next sought to define
the specific genes that harbor cryptic variation and found that most are involved
in Ras signaling or Ras-dependent transcriptional regulation. Among the 16 loci
not corresponding to the FLO11 promoter or coding region, seven contain genes
that we previously showed to harbor cryptic variation or be capable of uncovering
the rough phenotype when mutated (Lee et al. 2016, Taylor and Ehrenreich 2014,
Taylor and Ehrenreich 2015). In addition to FLO8, IRA2, MSS11 and SFL1, we
detected loci overlapping the vesicle component END3, the activator MGA1 and
the redox stress detoxifier TRR1 (Fig. 3.2a,b). As with Flo8–Mss11 heterodimer
and Sfl1, Mga1 acts downstream of the Ras pathway and regulates FLO11 and
other genes that are important for yeast colony morphology traits (Bruckner and
Mosch 2012)
(Fig. 3.4). Furthermore, although End3 and Trr1 are not
49
components of the Ras pathway, functional relationships between these genes
and Ras signaling likely exist (Charizanis et al. 1999, Gourlay and Ayscough
2006). We also cloned the causal genes underlying three loci that had not
previously been characterized. By performing allele replacements in multiple
rough segregants, we successfully resolved loci on chromosome V, VII and XI to
GPA2, MDS3 and TPK3, respectively (Fig. 3.2a,b; Fig. S3.5; Methods). These
genes encode a G protein subunit that is required for the recruitment of Ras-GTP
(GPA2) (Cherry et al. 2012), a component of the target of rapamycin pathway
that has also been shown to influence Ras signaling (MDS3) (McDonald et al.
2009), and a subunit of the Ras effector kinase protein kinase A (TPK3) (Cherry
et al. 2012).
3.4 Discussion
In summary, we have determined the genetic architectures underlying 17
independent instances of the same cryptic phenotype. These different
occurrences of rough colony morphology vary significantly in their numbers of
involved cryptic variants and also in whether they require a de novo mutation.
Among cases involving de novo mutations, our work suggests that the rough
phenotype’s genetic architecture depends on the gene that is mutated. This
relationship is likely tied to Ras signaling and the transcriptional control of Ras
target genes (Fig. 3.4). For example, Ssn3 and Ssn8 act the most proximally to
transcription, and rough SSN3 and SSN8 mutants show the lowest genetic
complexity in our study (Figs 3.2a and 3.4). In contrast, Gpb1, Ira1 and Ira2 act
in the upstream portion of the Ras cascade, and rough GPB1, IRA1 and IRA2
mutants exhibit relatively high numbers of detected loci (Figs 3.2a and 3.4).
Furthermore, the IRC8 mutant also shows a high number of detected loci (Fig.
3.2a), suggesting this uncharacterized gene might also act upstream of the Ras
pathway.
50
Figure 3.4: Genes involved in the rough phenotype in the BY × 3S
cross largely regulate signaling and transcriptional control by the
Ras pathway. A simplified portrait of the Ras pathway is presented.
Components of the Ras pathway and other identified proteins are color
coded by whether they can result in the rough phenotype when mutated
(red), harbor cryptic variants (green) or are unlikely to harbor
functionally distinct alleles that affect colony morphology in the cross
(black). IRA2 and SFL1 are colored as both green and red because they
fall into two of the aforementioned classes. Most of the genes that
uncover the rough phenotype when mutated or possess cryptic variants
function in Ras signaling and Ras-dependent transcriptional regulation.
Certain proteins, namely End3, Mds3 and Trr1, do not act directly in the
Ras pathway, but have been shown to either influence or be influenced
51
by the Ras activity. Our results also suggest that Irc8 might have some
functional relationship to the Ras pathway.
In addition, our study illustrates how the rough phenotype can arise in certain
environments, here 21°C, due to higher-order genetic interactions that only
involve cryptic variants (Fig. 3.2a). These combinations of cryptic variants
presumably recapitulate the molecular and systems level effects of epistatic
interactions that involve both cryptic variants and de novo or induced mutations
(Lee et al. 2016, Taylor and Ehrenreich 2014, Taylor and Ehrenreich 2015). This
scenario is supported by the fact that certain genes, namely IRA2 and SFL1,
appear to both possess cryptic variants and have the potential to uncover the
rough phenotype when mutated (Fig. 3.4). Tied to this point, we note that while
GPB1, IRA1, IRC8, SSN3 and SSN8 do not seem to harbor cryptic variation in
the BYx3S cross, it is possible that other S. cerevisiae strains carry cryptic
variation in these genes.
Given that most of the genes involved in the rough phenotype act in or are
influenced by the Ras pathway (Fig. 3.4), which has components that are
evolutionarily conserved (Cox and Der 2010), our findings might extend to other
species and traits. In fact, cryptic variation in the Ras pathway is known to impact
the development in Caenorhabditis elegans (Milloz et al. 2008), and perturbation
of Ras pathway components in humans can lead to cancer and other diseases
(Krauthammer et al. 2015). Thus, further characterizing cryptic variation in the
Ras pathway in yeast might provide valuable new insights into the mechanisms
that give rise to genetically complex phenotypes, which are relevant to health and
evolution.
52
3.5 Methods
3.5.1 Phenotyping of yeast colony morphology
Strains were grown at 30°C overnight in liquid media comprised of yeast extract
and peptone (YP) with 2% dextrose as the carbon source (YPD). Stationary
phase cultures were pinned onto YP agar media with 2% ethanol as the carbon
source (YPE) and grown for 7 days at 21°C, unless otherwise noted.
3.5.2 Generation of rough segregants
The rough segregants examined in this work come from a cross between the
BY4716 (‘BY’) lab strain, a derivative of S288C, and 322134S (‘3S’), a clinical
isolate. Strains used in this work possess the Synthetic Genetic Array marker
system (Tong et al. 2001), which facilitated rapid generation of MATa haploid
progeny from diploid strains. This system allows for selection of MATa spores by
plating of sporulated cultures onto yeast nitrogen base (YNB) media containing
canavanine. A MATa BY strain and a MATα 3S strain were mixed on YPD agar
media and incubated for 4 hours at 30°C. 106 zygotes were obtained from this
cross by microdissection. Each of these was sporulated as in (Taylor and
Ehrenreich 2014), after which segregants were plated onto YNB containing
canavanine to an average density of ~200 per plate. In total, >500 such plates
and >100,000 segregants were produced. Once colonies were visible on these
YNB plates, they were replicated onto YPE plates. After 7 days of growth at room
temperature, segregants were screened for rough morphology. To ensure all
instances of the rough phenotype were biologically independent, no more than
one rough segregant was gathered from any particular diploid.
3.5.3 Generation of backcross segregants
Each rough segregant was backcrossed to both BY and 3S to generate diploids
as described above. These were sporulated and plated onto YNB media
containing canavanine to select for MATa haploid progeny. Between 46 and 95
segregants were obtained from each of these backcrosses to create mapping
populations (Table S3.1). Segregants from each backcross were grown to
53
stationary phase in YPD at 30°C and mixed together in equal volumes. These
stationary cultures were then mixed together in equal volumes and DNA was
extracted from these pools of rough segregants using the Qiagen Genomic-tip
100/G kit. To generate control populations for these backcrosses, sporulations
were plated at high density onto YNB with canavanine and grown at 21 °C for 2
days to produce lawns containing millions of random backcross segregants.
These lawns were scraped directly off the plates and DNA was extracted from
these pools of cells, using the Qiagen Genomic-tip 100/G kit.
3.5.4 Sequencing of mapping populations
Sequencing libraries were generated from each backcross pool, using the
Illumina Nextera kit. These libraries were sequenced on an Illumina NextSeq
machine by the USC Epigenome Center, using 75 × 75 base reads. Rough
pools were sequenced to an average of 246-fold and control pools were
sequenced to an average of 162-fold coverage. Reads were aligned to either a
BY or 3S reference genome, using the Burrows–Wheeler Aligner version 7 with
option mem −t 20 (Li and Durbin 2009) and mpileup files were generated with
SAMtools (Li et al. 2009).
3.5.5 Genetic mapping using Multipool
Genome-wide allele frequencies at 36,756 high confidence SNPs were
determined by a custom Python script (Taylor and Ehrenreich 2014, Taylor and
Ehrenreich 2015). Loci were detected in each mapping population using
MULTIPOOL (Edwards and Gifford 2012) with settings: replicates mode, 3,300
bp centimorgans, 100 bp bins. Segregating genomic intervals in each population
were independently analyzed, with a region considered significant in a given
backcross if it had a LOD score >5 across a region of at least 30 kb. We
considered the span of each locus as the 90% confidence interval surrounding
the point of maximal significance, as outputted by MULTIPOOL. Loci that were
detected in control backcross populations were ignored from their corresponding
populations of rough segregants. All loci identified in mapping and control
54
experiments can be found in Supplemental Tables 3 and 4. In a few instances,
detected loci were broad and overlapped two cryptic variants that were
previously identified (Lee et al. 2016, Taylor and Ehrenreich 2014, Taylor and
Ehrenreich 2015) or were cloned in the current study. In such cases, these broad
regions were counted as two loci.
3.5.6 Identification of de novo mutations
Sequence data for the mapping populations were examined for genetic
differences relative to the BY and 3S reference genomes, using a combination
of custom Python scripts and manual inspection. Identified lesions were then
validated by PCR and Sanger sequencing of a potentially mutated site in the
corresponding rough segregant. Primers for these sequencing experiments are
provided in Supplementary Table 7. In addition, we investigated the role of
chromosomal anomalies, such as aneuploidies, inversions and translocation,
and did not see evidence of such events playing a role in the phenotype (Note
S3.1).
3.5.7 Genetic engineering experiments
Allele replacements were performed using a modified form of adaptamer
mediated allele replacement (Erdeniz et al. 1997, Matsui et al. 2015, Taylor and
Ehrenreich 2015). Transformations were conducted with two partially overlapping
PCR products—a full-length amplicon of a gene of interest that was tailed at the
3’ end with the 5’ portion of the kanMX cassette and a copy of the kanMX
cassette that was tailed on the 3’ end with part of the intragenic region
downstream of the gene (Matsui et al. 2015, Taylor and Ehrenreich 2015).
Knock-ins were identified using selection on G418 and verified by Sanger
sequencing. At least three knock-in strains were screened per allele
replacement. Gene deletions were performed by replacing a gene of interest
with the CORE cassette (Storici et al. 2001). Regions corresponding to 60
bases upstream and downstream of the target gene were tailed to the CORE
cassette using PCR. This product was transformed into cells using the lithium
55
acetate method (Daniel Gietz and Woods 2002), and selection with G418 was
used to screen for integration of the cassette. PCR was then used to verify that
deletion strains recovered from the G418 selection lacked the gene of interest.
Primers used in genetic engineering experiments can be found in Table S3.7.
56
3.6 Supporting Materials
Figure S3.1: Phenotypes of the 17 identified rough segregants at
21°C.
57
Figure S3.2. Verification that identified de novo mutations have
phenotypic effects in rough segregants 1 through 8. Allele
replacements were performed in the rough segregant in which a de novo
mutation was detected. For a gene harboring a de novo mutation in a
given rough segregant, we performed three genetic engineerings. The
mutant allele was replaced with itself, the wild type BY allele, and the wild
type 3S allele. These replacements are referred to in superscript as ‘m’,
‘BY’, and ‘3S’, respectively. For each of the eight de novo mutations,
replacement of the lesion with wild type BY or 3S alleles resulted in a
conversion from rough to smooth colony morphology. This implies that the
de novo mutations play causal roles in a given individual’s expression of
the rough phenotype.
58
Figure S3.3. Rough segregants 1 and 2 possess the same SSN8
lesion, but are likely distinct individuals obtained from different
matings of BY and 3S. Genome-wide haplotypes are shown below
mapping data from Fig. 2a. Blue horizontal bars represent regions in
which the segregant possessed the BY allele and orange horizontal bars
represent regions in which the segregant possessed the 3S allele. The red
vertical bars indicate the SSN8 mutations. As can be seen from their
haplotypes, rough segregants 1 and 2 do not share any recombination
breakpoints, indicating they are not spores from the same tetrad. Because
each rough segregant was obtained from independent matings of BY and
3S, it is unlikely that these individuals’ SSN8 lesions arose in the same
mutational event.
59
Figure S3.4. Complete deletion of mutated genes from rough
segregants 1 through 8. To test whether the identified de novo mutations
were null or partial loss-of-function alleles, we deleted these genes in their
entirety from the corresponding rough segregant. Lack of effect of these
complete deletions on colony morphology indicates that a given de novo
mutation is likely a null allele. In contrast, a conversion from rough to
smooth colony morphology suggests that a particular de novo mutation is
a partial loss-of-function allele.
60
Figure S3.5. Allele replacements verify that GPA2
3S
, MDS3
3S
, and
TPK3
BY
harbor cryptic variants with genetic background-dependent
effects. We replaced the causal allele of a given gene with the alternative
allele in four different rough segregants. As a control, causal alleles were
also replaced with themselves. Representative allele replacement results
are shown for each allele replacement experiment in each rough
segregant. Please note that the pictures are ordered so that allele
replacements involving the BY and 3S alleles are always shown to the left
and right, respectively.
61
Table S3.1. Number of individuals in each pooled sequencing
experiment.
RS1xBY 92
RS1x3S 95
RS2xBY 84
RS2x3S 83
RS3xBY 82
RS3x3S 92
RS4xBY 95
RS4x3S 86
RS5xBY 93
RS5x3S 90
RS6xBY 85
RS6x3S 89
RS7xBY 84
RS7x3S 48
RS8xBY 92
RS8x3S 52
RS9xBY 69
RS9x3S 69
RS10xBY 89
RS10x3S 68
RS11xBY 86
RS11x3S 74
RS12xBY 90
RS12x3S 53
RS13xBY 88
RS13x3S 72
RS14xBY 90
RS14x3S 74
RS15xBY 69
RS15x3S 46
RS16xBY 70
RS16x3S 50
RS17xBY 48
RS17x3S 70
62
Table S3.2. Loci detected in rough backcross pools. Start and stop
positions were called as the 90% confidence interval around the point of
maximum significance. Allele frequencies were calculated as the
maximum within a 25-SNP window within a locus.
Segregant Cross Chromosome Start Stop Frequency
1 3S 3 157990 235590 0.99
1 3S 5 6639 64439 1.00
1 BY 5 349582 413382 0.99
1 BY 14 458048 649748 1.00
2 3S 3 175351 212951 0.99
2 3S 5 6639 89139 1.00
2 BY 5 337336 461436 0.99
2 3S 13 547996 652396 0.87
2 BY 14 504670 686470 1.00
3 3S 3 78378 217178 0.99
3 BY 4 1172541 1220841 0.85
3 3S 5 5035 352535 1.00
3 BY 5 353654 460054 1.00
3 BY 14 441083 488583 0.86
3 3S 16 427572 563072 1.00
4 3S 3 90850 228750 1.00
4 BY 4 1173122 1259022 0.87
4 3S 5 1435 79335 0.99
4 BY 5 277713 497113 1.00
4 3S 13 514890 620490 0.88
4 3S 14 319549 485449 0.98
4 3S 15 88326 237926 0.99
4 3S 15 586426 733426 0.81
5 3S 2 450356 602056 1.00
5 3S 3 92377 219377 1.00
5 3S 5 6535 64435 1.00
5 BY 5 336513 446913 1.00
5 3S 12 56412 118112 0.80
5 3S 13 510149 686349 0.96
5 3S 14 360204 480404 0.95
6 BY 2 481372 547972 1.00
6 3S 2 548339 657239 0.82
6 3S 3 105147 207047 1.00
6 3S 5 1435 81635 1.00
6 BY 5 341045 467745 0.99
63
6 BY 12 617236 672536 0.82
6 3S 13 534632 626532 0.88
6 3S 14 350144 570944 0.90
6 BY 15 34977 191977 0.88
7 3S 3 132313 219813 0.98
7 3S 5 1435 70235 0.99
7 3S 7 100573 195573 0.85
7 3S 9 366656 418556 0.87
7 3S 11 119772 366972 0.88
7 3S 13 552176 667676 0.96
7 3S 14 373335 499935 0.91
7 BY 15 107986 215386 0.07
7 3S 15 451598 689798 0.92
8 3S 3 112256 214056 0.99
8 3S 5 28135 78135 1.00
8 BY 5 101481 271781 0.87
8 BY 5 286081 461781 0.99
8 BY 7 85900 234000 0.88
8 3S 8 441 142341 0.07
8 3S 9 383751 432851 0.95
8 3S 10 299283 464883 0.99
8 3S 13 25332 140232 0.83
8 3S 13 545805 702505 0.98
8 3S 15 82512 228312 0.92
8 3S 15 486712 624312 0.84
9 3S 3 97868 220168 0.99
9 3S 5 6639 69839 0.98
9 BY 5 319074 470274 1.00
9 BY 9 320560 394160 0.99
9 3S 9 395108 434008 0.99
9 3S 13 511018 655418 0.94
9 3S 15 126 197426 0.95
10 3S 3 105083 218983 1.00
10 BY 4 1169284 1269984 0.92
10 3S 5 4635 71335 0.99
10 BY 5 323425 456025 0.99
10 BY 9 317767 393767 1.00
10 3S 9 395108 432108 0.98
10 3S 11 94872 326772 0.86
10 3S 13 584908 654908 0.92
10 3S 15 95414 250914 0.98
64
11 3S 3 69577 232977 1.00
11 3S 5 1935 73635 1.00
11 BY 5 215722 472022 1.00
11 BY 7 124322 195022 0.81
11 BY 9 313835 393935 1.00
11 3S 9 394410 432610 1.00
11 3S 13 492981 670681 0.93
11 3S 14 319549 425249 0.92
11 3S 15 496529 724429 0.85
12 3S 3 92477 207077 0.99
12 3S 5 6539 57439 0.99
12 BY 5 308100 465000 0.98
12 3S 7 932650 994350 0.89
12 BY 9 327892 393492 0.98
12 3S 9 395543 438943 0.99
12 BY 12 708936 741736 0.81
12 3S 13 531055 620455 0.92
12 3S 14 363239 471839 0.83
12 3S 15 317537 744837 0.89
13 3S 3 137844 219644 0.99
13 BY 4 1140684 1259184 0.95
13 3S 5 6535 76835 1.00
13 BY 5 326731 462131 0.99
13 3S 8 5198 123698 0.20
13 BY 9 344049 394149 0.99
13 3S 9 395108 432208 0.99
13 3S 13 518383 691383 0.94
13 3S 15 113329 195529 0.91
13 3S 15 465529 548929 0.84
14 3S 3 111173 217873 0.98
14 3S 5 1435 62835 1.00
14 BY 5 153019 226919 0.81
14 BY 5 193362 272662 0.83
14 BY 5 285062 467062 1.00
14 BY 5 337672 413672 0.98
14 BY 7 87383 142783 0.78
14 BY 7 100573 195573 0.85
14 3S 7 961406 1001106 0.80
14 3S 9 361620 432620 0.99
14 3S 13 514075 735675 0.99
14 3S 14 415078 483578 0.85
65
14 3S 15 49129 227029 0.93
14 BY 15 981009 1057509 0.99
15 3S 3 91977 211777 0.99
15 BY 4 962405 999505 0.80
15 3S 5 5335 72735 1.00
15 BY 5 139920 410720 0.99
15 BY 7 68009 257109 0.87
15 BY 10 394723 462323 0.81
15 3S 13 544486 684186 0.98
15 3S 14 374874 497074 0.90
15 3S 15 451405 713805 1.00
16 3S 3 100677 219177 0.99
16 3S 5 6639 43939 1.00
16 BY 5 143401 301901 0.95
16 BY 5 363002 457002 0.99
16 BY 7 99973 191773 0.88
16 3S 11 79889 366889 0.86
16 3S 13 567537 688837 1.00
16 3S 14 344174 524474 0.96
16 3S 15 126 221726 1.00
16 3S 15 449982 664382 1.00
17 3S 3 96650 211250 1.00
17 BY 4 1055912 1150512 0.93
17 BY 4 1169012 1240512 1.00
17 3S 5 6535 68135 1.00
17 BY 5 197225 297825 0.89
17 BY 5 323125 420925 1.00
17 BY 7 81340 142540 0.93
17 3S 8 59041 154741 0.07
17 3S 11 122072 372272 0.95
17 BY 12 643938 729338 0.93
17 3S 13 538573 681373 0.99
17 3S 14 374177 482077 0.94
17 3S 15 141726 218926 1.00
17 3S 15 491269 659569 0.96
66
Table S3.3. Loci detected in control backcross pools. Start and stop
positions were called as the 90% confidence interval around the point of
maximum significance. Allele frequencies were calculated as the
maximum within a 25-SNP window within a locus.
Segregant Cross Chromosome Start Stop Frequency
2 3S 3 175351 222251 0.99
2 3S 5 6539 99439 1.00
2 BY 13 25470 83870 0.80
3 3S 3 95839 221139 0.99
3 3S 5 4935 352535 1.00
3 3S 16 451372 522072 0.85
4 3S 3 91477 228777 0.99
4 BY 4 1172722 1238422 0.82
4 3S 5 1335 79335 0.99
4 3S 14 319549 465949 0.97
5 3S 3 90277 226577 0.99
5 3S 3 157990 222790 0.99
5 BY 4 1095315 1264915 0.88
5 3S 5 6539 64439 0.99
5 3S 5 6539 64439 0.99
5 3S 14 312304 464204 0.98
5 3S 14 343473 427173 0.93
6 3S 3 105147 204847 0.97
6 3S 5 6539 89039 0.94
6 3S 14 324739 430039 0.91
7 3S 3 92213 221213 0.99
7 3S 5 1335 83635 0.99
7 3S 14 336535 428935 0.93
8 3S 3 70256 214056 1.00
8 3S 5 26335 97935 1.00
8 3S 14 354444 427244 0.95
9 3S 3 91468 220668 0.99
9 3S 4 127296 538696 0.83
9 3S 5 25139 78039 0.98
9 3S 14 341751 430651 0.93
10 3S 3 74057 218957 0.96
10 BY 4 1097984 1254184 0.86
10 3S 5 1335 117535 0.97
10 3S 14 312239 445239 0.93
11 3S 3 93677 219177 0.99
67
11 3S 5 1435 90435 1.00
11 3S 14 319649 441049 0.93
12 3S 3 91077 207077 0.99
12 BY 4 1091855 1221755 0.81
12 3S 5 6539 57439 1.00
12 3S 14 337139 441539 0.92
13 3S 3 92244 222244 0.99
13 BY 4 1171184 1255684 0.84
13 3S 5 1435 80735 1.00
14 3S 3 92073 222673 1.00
14 BY 4 1084535 1246235 0.82
14 3S 5 1435 59235 1.00
15 3S 3 91157 211857 0.99
15 3S 5 2835 78635 0.99
15 3S 14 313274 421374 0.95
15 BY 15 12077 98977 0.96
16 3S 3 100577 215577 0.99
16 BY 4 1095215 1220415 0.85
16 3S 5 6539 43939 1.00
16 3S 14 313274 424574 0.95
17 3S 3 104933 212933 0.99
17 BY 4 1172876 1214876 0.90
17 3S 5 6635 73435 1.00
17 3S 14 355180 412980 0.97
68
Table S3.4. Predominant genotypes in mapping populations derived
from each rough segregant.
Segregant
# loci
detected Predominant genotype
1 2
FLO8
3S
SSN8
mut
2 3
FLO8
3S
MSS11
BY
SSN8
mut
3 4
TRR1
3S
FLO8
3S
END3
3S
SSN3
mut
4 5
FLO8
3S
MSS11
BY
END3
BY
IRA2
mut
SFL1
BY
5 5
IRA1
mut
FLO8
3S
XII-1
BY
MSS11
BY
END3
BY
6 6
IRA1
mut
FLO8
3S
XII-2
3S
MSS11
BY
END3
BY
IRA2
3S
7 10
GPA2
3S
FLO8
3S
MDS3
3S
FLO11
BY
TPK3
BY
MSS11
BY
END3
BY
IRA2
BY
SFL1
BY
GPB1
mut
8 10
GPA2
3S
FLO8
3S
MDS3
3S
VIII
3S
FLO11
BY
IRC8
mut
XIII
BY
MSS11
BY
IRA2
BY
SFL1
BY
9 5
FLO8
3S
FLO11
3S
FLO11
BY
MSS11
BY
IRA2
BY
10 6
FLO8
3S
FLO11
3S
FLO11
BY
TPK3
BY
MSS11
BY
IRA2
BY
11 7
GPA2
3S
FLO8
3S
MDS3
3S
FLO11
3S
FLO11
BY
MSS11
BY
SFL1
BY
12 8
FLO8
3S
MGA1
BY
FLO11
3S
FLO11
BY
XII-2
3S
MSS11
BY
END3
BY
SFL1
BY
13 7
FLO8
3S
VIII
3S
FLO11
3S
FLO11
BY
MSS11
BY
IRA2
BY
SFL1
BY
14 9
GPA2
3S
FLO8
3S
MDS3
3S
MGA1
BY
FLO11
3S
FLO11
BY
MSS11
BY
END3
BY
IRA2
BY
15 8
IV
3S
GPA2
3S
FLO8
3S
MDS3
3S
X
3S
MSS11
BY
END3
BY
SFL1
BY
16 8
GPA2
3S
FLO8
3S
MDS3
3S
TPK3
BY
MSS11
BY
END3
BY
IRA2
BY
SFL1
BY
17 10
GPA2
3S
FLO8
3S
MDS3
3S
VIII
3S
TPK3
BY
XII-2
3S
MSS11
BY
END3
BY
IRA2
BY
SFL1
BY
69
Table S3.5. Information regarding causal de novo mutations.
Functional consequences were assessed by deleting the mutant allele
from the rough segregant carrying a given lesion. The IRA2 mutation with
an asterisk refers to the one identified in Taylor 2014. PLOS Genetics.
Mutant
gene
Function of
gene
Lesion Functional
consequence
Impact on
gene
sequence
GPB1 Multistep
regulator of
cAMP-PKA
signaling
Single
nucleotide
mutation
720G->T
Loss-of-
function
May disrupt
Kelch domain
(Phan et al.
2010)
IRA1 GTPase-
activating
protein;
negatively
regulates
RAS
Single base
deletion
1160∆G
Null Nonsense;
Loss of Gpb2
binding site
and GAP-
related
domain
(Harashima
2006);
Truncates
2694 amino
acids, 87% of
protein
IRA1 GTPase-
activating
protein;
negatively
regulates
RAS
Single base
deletion
7719∆G
Null Nonsense;
Loss of Gpb2
binding site
and
disruption of
GAP-related
domain
(Harashima
2006);
Truncates
503 amino
acids, 16% of
protein
IRA2 GTPase-
activating
protein;
negatively
regulates
RAS
Single
nucleotide
mutation
-60C->T
Null Likely
disrupts TBP
binding site
(Venters
2011)
IRA2* GTPase-
activating
Single base
deletion
Loss-of-
function
Nonsense;
Loss of Gpb1
70
protein;
negatively
regulates
RAS
8801∆A binding site
(Harashima
2006);
Truncates
117 amino
acids, 4% of
protein
IRC8 Bud tip
localized
protein
Single base
deletion
938∆A
Null Nonsense;
Truncates
487 amino
acids, 59% of
protein
SSN3 Cyclin-
dependent
protein
kinase;
component of
RNA pol II
holoenzyme
Single base
deletion
111∆T
Null Nonsense;
Truncates
508 amino
acids, 92% of
protein
SSN8 Cyclin-like
component of
RNA
polymerase II
holoenzyme
22 base
deletion
40∆22
Null Nonsense;
Loss of cyclin
box domain
(Kuchin
1995);
Truncates
291 amino
acids, 90% of
protein
71
Table S3.6. SRA accession numbers for each pooled sequencing
experiment.
RS1xBY4716 SAMN04126845
RS1x322134S SAMN04126846
RS2xBY4716 SAMN04126847
RS2x322134S SAMN04126848
RS3xBY4716 SAMN04126849
RS3x322134S SAMN04126850
RS4xBY4716 SAMN04126851
RS4x322134S SAMN04126852
RS5xBY4716 SAMN04126853
RS5x322134S SAMN04126854
RS6xBY4716 SAMN04126855
RS6x322134S SAMN04126856
RS7xBY4716 SAMN04126857
RS7x322134S SAMN04126858
RS8xBY4716 SAMN04126859
RS8x322134S SAMN04126860
RS9xBY4716 SAMN04126861
RS9x322134S SAMN04126862
RS10xBY4716 SAMN04126863
RS10x322134S SAMN04126864
RS11xBY4716 SAMN04126865
RS11x322134S SAMN04126866
RS12xBY4716 SAMN04126867
RS12x322134S SAMN04126868
RS13xBY4716 SAMN04126869
RS13x322134S SAMN04126870
RS14xBY4716 SAMN04126871
RS14x322134S SAMN04126872
RS15xBY4716 SAMN04126873
RS15x322134S SAMN04126874
RS16xBY4716 SAMN04126875
RS16x322134S SAMN04126876
RS17xBY4716 SAMN04126877
RS17x322134S SAMN04126878
RS1xBY4716_control SAMN04126879
RS1x322134S_control SAMN04126880
RS2xBY4716_control SAMN04126881
RS2x322134S_control SAMN04126882
RS3xBY4716_control SAMN04126883
72
RS3x322134S_control SAMN04126884
RS4xBY4716_control SAMN04126885
RS4x322134S_control SAMN04126886
RS5xBY4716_control SAMN04126887
RS5x322134S_control SAMN04126888
RS6xBY4716_control SAMN04126889
RS6x322134S_control SAMN04126890
RS7xBY4716_control SAMN04126891
RS7x322134S_control SAMN04126892
RS8xBY4716_control SAMN04126893
RS8x322134S_control SAMN04126894
RS9xBY4716_control SAMN04126895
RS9x322134S_control SAMN04126896
RS10xBY4716_control SAMN04126897
RS10x322134S_control SAMN04126898
RS11xBY4716_control SAMN04126899
RS11x322134S_control SAMN04126900
RS12xBY4716_control SAMN04126901
RS12x322134S_control SAMN04126902
RS13xBY4716_control SAMN04126903
RS13x322134S_control SAMN04126904
RS14xBY4716_control SAMN04126905
RS14x322134S_control SAMN04126906
RS15xBY4716_control SAMN04126907
RS15x322134S_control SAMN04126908
RS16xBY4716_control SAMN04126909
RS16x322134S_control SAMN04126910
RS17xBY4716_control SAMN04126911
RS17x322134S_control SAMN04126912
73
Table S3.7. Primers used in this study. For marked replacements, a
gene of interest is amplified using primers GENE_mark1 and
GENE_mark2. An MX cassette is amplified using primers universal3 and
GENE_mark4. These amplicons are mixed together and transformed into
a strain of interest. The two cassettes are joined by homologous
recombination and replace the gene of interest, leaving behind a
selectable MX cassette marker. Every mark2 primer begins with the
sequence TAAATGTACGGGCGACAGTCACATCATGCCCCTGAGCTGC
GCACGTCAAGACTGTCAAGG, which is homologous to all MX cassettes
and allows for a recombination between the gene cassette and the MX
cassette. Every mark4 primer ends with the sequence
CGCACTTAACTTCGCATCTG, which allows for the amplification of an
MX cassette.
The following primers can be used to generate allele swap strain using 'marked allele replacement', as in Matsui 2015. Genetics. doi: 10.1534/genetics.115.180661.
universal 3 CCTTGACAGTCTTGACGTGC
FLO11_coding_mark1 TCGCTTATTTGGTCCTTTCG
FLO11_coding_mark2 TAAATGTACGGGCGACAGTCACATCATGCCCCTGAGCTGCGCACGTCAAGACTGTCAAGGTTGCCAATGTATGAGAGTGG
FLO11_coding_mark4 CGTGTAAATCAAGTATGTCATTTCAGGAAACTTCCTCAATTCTTGGCGTACTATATTGAGCGCACTTAACTTCGCATCTG
FLO11_promoter_mark1 TGCGTATATGGATTTTTGAGG
FLO11_promoter_mark2 TAAATGTACGGGCGACAGTCACATCATGCCCCTGAGCTGCGCACGTCAAGACTGTCAAGG GCAGATGCAAACAAAAAGCA
FLO11_promoter_mark4 CACCACCACGATCGGAGAAGCGCTATTAGTAGCAATGGCTAAGCCTTGCCAGAACATGTAA CGCACTTAACTTCGCATCTG
GBP1_mark1 CCGTACCAATTCTTCTACAT
GBP1_mark2 TAAATGTACGGGCGACAGTCACATCATGCCCCTGAGCTGCGCACGTCAAGACTGTCAAGGCCCCGCGGAATTAATTAGTT
GBP1_mark4 GAAAAAATTTTCTCGTTTTCCTTTAGTCACTCTTGTCACATAAGGATTATCCGAACCCCGCGCACTTAACTTCGCATCTG
GPA2_mark1 AAGATATACCATATATTACG
GPA2_mark2 TAAATGTACGGGCGACAGTCACATCATGCCCCTGAGCTGCGCACGTCAAGACTGTCAAGGAACGGTTGTGCTTAATACAG
GPA2_mark4 CGGGATAATAACTATAATGACTACAATAATATAGTGGTATAACGCTATAAATTAAAAAATCGCACTTAACTTCGCATCTG
IRA1_mark1 GCCAATAAAATGATCAAAGG
IRA1_mark2 TAAATGTACGGGCGACAGTCACATCATGCCCCTGAGCTGCGCACGTCAAGACTGTCAAGGAAAACGTATATAATCACTGC
IRA1_mark4 AAAACAAAATATAATTATAAGGAAAAACGTATATAATCACTGCAATACTCTAATTTAAAACGCACTTAACTTCGCATCTG
IRA2_mark1
GGACATGCTTCTCCCTGAAG
IRA2_mark2 TAAATGTACGGGCGACAGTCACATCATGCCCCTGAGCTGCGCACGTCAAGACTGTCAAGGTACCCTTGTAAGTGTCATCC
IRA2_mark4 GCACAGATCCCAGAGAAAAGCAGGGAAACAAGAAAATAAGAAAACAAGAAAAACAGTAGTCGCACTTAACTTCGCATCTG
IRC8_mark1 GCAAAACGCACATACCCACA
IRC8_mark2 TAAATGTACGGGCGACAGTCACATCATGCCCCTGAGCTGCGCACGTCAAGACTGTCAAGGATATGAATAAGTCGGTTGGT
IRC8_mark4 ATCGTGTTCTACGGGATCAAAATAGTTGCTTTTAGCAGTTCCCATAGGTATCTTTGATACCGCACTTAACTTCGCATCTG
MDS3_mark1 GTGGCTAAGGCAGACTCCGT
MDS3_mark2 TAAATGTACGGGCGACAGTCACATCATGCCCCTGAGCTGCGCACGTCAAGACTGTCAAGGGTGCGAGTAACTATCCTGGG
MDS3_mark4 GGAAGCAATCCGTTTTGAGATGTATAGCAGCATATTCTTGGATTATTAAGAACTTTATATCGCACTTAACTTCGCATCTG
SSN3_mark1 TGTGGCTTAAGTTGCGTTTC
SSN3_mark2 TAAATGTACGGGCGACAGTCACATCATGCCCCTGAGCTGCGCACGTCAAGACTGTCAAGGCTATCTTCTGTTTTTCTTTC
SSN3_mark4 GAATATAATAGTGACAGTGCTGTGGAATGAAAAATTCCAAATATATATAAAAATAGAAGCCGCACTTAACTTCGCATCTG
SSN8_mark1 CTGAATCTCAAAAGTTAGAC
SSN8_mark2 TAAATGTACGGGCGACAGTCACATCATGCCCCTGAGCTGCGCACGTCAAGACTGTCAAGGGTTTTTAAATTTATTCTTCG
SSN8_mark4 GACGAAACATTTCCAAAACGGATCATCACCACCATAATGATTGAATTTACAGGCTTAACGCGCACTTAACTTCGCATCTG
TPK3_mark1 AGACACTTTGTGCAGTCGTC
TPK3_mark2 TAAATGTACGGGCGACAGTCACATCATGCCCCTGAGCTGCGCACGTCAAGACTGTCAAGGATGGCGTATATGAATGCTCC
TPK3_mark4 GAAAAGTCCTAGATCACTTTGAACGTCCCAGTCTTCTGAGGACGCAAATGTAGTCACAATCGCACTTAACTTCGCATCTG
74
The following primers can be used to amplify the CORE cassette (Storici 2001. Nature Biotechnology. 19: 773-776) for targeted gene deletion.
gpb1∆_F TTTCTCGTTTTCCTTTAGTCACTCTTGTCACATAAGGATTATCCGAACCCCGCCCCGCGGGAGCTCGTTTTCGACACTGG
gpb1∆_R TGAGCCGACCTCCCTATATCGGCTACTTTAAGGCTTTCCGTACCAATTCTTCTACATAAGTCCTTACCATTAAGTTGATC
ira1∆_F TTAGGAGCACGACATTCTTGCCAGTATCATTGTTGCTAATCTTTTTCTCTCATAAATTGCTCCTTACCATTAAGTTGATC
ira1∆_R GTTAAGCTATTTAACGAAAGCGTATAAAGTCAAGTGATCATCTTTTGCCCTGCAAATAGAGAGCTCGTTTTCGACACTGG
ira2∆_F GCATATAGCATTGTCCTCTGTTATTCGTTTTGCTTTTCTCCTTTAGTGTTACTTTTCCCCCAACGGAGCTCGTTTTCGACACTGG
ira2∆_R ATGTACATTCATGCTTACAGATAGATATTGATATTTCTTTCATTAGTTTATGTAACACCTTCCTTACCATTAAGTTGATC
irc8∆_F CTTCTTACGTATCAGAACAAGAAAGCATTTCCAAAGTAATTGCATTTGCCCTTGAGCAGTGAGCTCGTTTTCGACACTGG
irc8∆_R TTAATGAAATAAATCGTGTTCTACGGGATCAAAATAGTTGCTTTTAGCAGTTCCCATTCCTTACCATTAAGTTGATC
ssn3∆_F GAATATAATAGTGACAGTGCTGTGGAATGAAAAATTCCAAATATATATAAAAATAGAAGCTCCTTACCATTAAGTTGATC
ssn3∆_R AAAGGTTTATAGGAAAGAAAAAAGGCGGAAGGGTATACTGAAGTTAGTAATTTTGCTTCCGAGCTCGTTTTCGACACTGG
ssn8∆_F ATCATCACCACCATAATGATTGAATTTACAGGCTTAACGGTTTTTAAATTTATTCTTCGCGAGCTCGTTTTCGACACTGG
ssn8∆_R AAATGCCCTCTCAAACTTTAGTTGAAGAGCGATAAGGCATCTGAATCTCAAAAGTTAGACTCCTTACCATTAAGTTGATC
Note S3.1. Aneuploidies and chromosomal rearrangements do not
contribute to our current results. To determine if aneuploidies caused rough
morphology in any individuals, the average coverage of each chromosome was
examined in the rough and control backcross pools. Only mapping populations
descended from rough segregant 12 possessed an aneuploidy, which occurred
on Chromosome I. 38 and 34% enrichment of this chromosome was observed
relative to average genome-wide coverage in the BY and 3S backcrosses,
respectively. However, similar enrichment in coverage for Chromosome I was
seen in control pools descended from rough segregant 12, suggesting that the
aneuploidy does not contribute to the rough phenotype. Furthermore, analyses
aimed at identifying chromosomal rearrangements failed to identify any such
events.
75
Chapter 4: Unraveling the layers of cryptic genetic variation in a yeast gene
regulatory network
This work is currently under review at Genetics
4.1 Abstract
Cryptic genetic variation may be an important contributor to heritable traits, but its
extent and regulation are not fully understood. Here, we investigate the cryptic
variation underlying a Saccharomyces cerevisiae colony phenotype that is
typically suppressed in a cross. Mapping of the trait’s genetic basis after
induction by three different genetic perturbations identifies 21 total loci. The
perturbations largely uncover distinct loci, and strongly influence additivity,
epistasis, and genotype-environment interaction among these loci. Only a single
locus— the coding region of the cell surface gene FLO11—is common to all
three perturbations. While the remaining loci mostly influence FLO11
transcription in cis or trans, we detect loci in different pathways and sub-
pathways in the presence of each perturbation. Our work shows that complex
gene regulatory networks can harbor multiple layers of cryptic variation, each of
which may only be visible in highly specific genotypic and environmental contexts.
76
4.2 Introduction
Most research on complex traits focuses on characterizing the genetic basis of
phenotypic diversity that is visible within populations (Atwell et al. 2010, Aylor et
al. 2011, Bloom et al. 2013, Mackay et al. 2012). Yet, these same populations
can also harbor cryptic genetic variation that does not typically impact phenotype,
and is only observable when particular genetic or environmental perturbations
occur (Bergman and Siegal 2003, Dworkin et al. 2003, Geiler-Samerotte et al.
2016, Gibson and Dworkin 2004, Jarosz and Lindquist 2010, Mullis et al. 2018,
Queitsch et al. 2002, Rutherford and Lindquist 1998). This cryptic variation may
be an important source of phenotypic variability in medically and evolutionarily
significant traits (Le Rouzic and Carlborg 2008, McGuigan and Sgro 2009, Paaby
and Rockman 2014). Thus, it is imperative that we determine the extent of cryptic
variation within populations, as well as the mechanisms that toggle this cryptic
variation between silent and visible states. However, such work is inherently
difficult because the specific perturbations needed to uncover cryptic variation, as
well as the exact identities of the cryptic variants that are affected by these
perturbations, are rarely known. Such information is critical to obtaining a more
complete, mechanistic understanding of cryptic variation.
In previous papers, we developed an experimental system that can be used to
systematically identify cryptic variants influencing a colony phenotype in
Saccharomyces cerevisiae (Lee et al. 2016, Taylor and Ehrenreich 2014, Taylor
and Ehrenreich 2015, Taylor et al. 2016). Upon mating the lab strain BY4716 (BY)
to a haploid derivative of the clinical isolate 322134S (3S), we found that BY, 3S,
and nearly all of their recombinant haploid progeny form smooth colonies when
grown on solid media. However, certain genetic perturbations (GPs) can enable
some BYx3S segregants to express a rough colony phenotype at particular
temperatures (Fig 4.1A and 4.1B). Expression of the trait results from complex
interactions between the GPs, segregating loci, and temperature (Lee et al. 2016,
Taylor and Ehrenreich 2014, Taylor and Ehrenreich 2015, Taylor et al. 2016).
Crucially, genetic mapping and genetic engineering techniques can be used to
77
comprehensively identify the specific cryptic variants that enable rough
segregants to express the trait.
Our initial work on colony morphology in the BYx3S cross focused on a de novo
loss-of-function mutation in IRA2 (‘GPa’), a negative regulator of Ras signaling
(Lee et al. 2016, Taylor and Ehrenreich 2014, Taylor and Ehrenreich 2015) (Fig
4.1C). Across several papers, we demonstrated that GPa causes trait expression
through temperature-dependent, higher-order genetic interactions involving
cryptic variants inherited from both BY and 3S (Lee et al. 2016, Taylor and
Ehrenreich 2014, Taylor and Ehrenreich 2015). Subsequently, we identified other
de novo and introduced mutations that also facilitate expression of the rough
phenotype in the BYx3S cross (Lee et al. 2016, Taylor and Ehrenreich 2015).
These other mutations tended to also disrupt negative regulation of signaling and
transcription within the Ras pathway, and mostly interacted with the same alleles
found in the studies focused on GPa.
In the same study, we also found that some rough BYx3S segregants lacked any
detectable de novo mutations (Taylor et al. 2016). Instead, the majority of these
individuals inherited independent recombination events within a 1.3 kb region in
the promoter of the cell surface gene FLO11 (‘GPb’, Fig 4.1D). FLO11 encodes a
flocculin whose display on the cell surface facilitates cell-cell adhesion and is
required for expression of the rough phenotype (Lo and Dranginis 1996, Lo and
Dranginis 1996, Taylor and Ehrenreich 2015)(Lo and Dranginis 1996, Taylor and
Ehrenreich 2015). Genetic engineering experiments showed that the
recombination event brought at least two linked polymorphisms at FLO11 onto
the same chromosome, resulting in a new FLO11 haplotype that behaves like the
mutations described in our earlier studies (Lee et al. 2016). Specifically,
segregants with a BY promoter and a 3S coding region had the potential to
express the trait. Despite discovering GPb in this past study, we neither resolved
the causal variant(s) in the FLO11 promoter nor comprehensively mapped the
loci enabling this GP to exert a phenotypic effect. Thus, it was not possible to
78
compare the genetic basis of the rough phenotype in the presence of GPb to our
initial work on GPa.
Here, we use the rough colony system to determine how different types of GPs
interact with distinct cryptic variants in the BYx3S cross to produce the same trait.
In addition to GPa and GPb, we also examine a third GP that facilitates
expression of the rough phenotype. To generate ‘GPc,’ we knocked out the
activator Flo8 and the repressor Sfl1, which are the main transcription factors
acting downstream of the Ras pathway to regulate colony morphology in the
cross (Fig 4.1E). Previously, we showed that deletion of SFL1 is sufficient to
enable Ras-dependent cryptic variants to express FLO11 and that transcription
of FLO11 in these sfl1∆ segregants is Flo8-dependent (Lee et al. 2016). By
eliminating these key regulators and screening for segregants expressing the
trait (Fig 1F), we sought to uncover previously unidentified cryptic variants that
can also give rise to the rough phenotype.
In this paper, we comprehensively determine the genetic basis of the rough
phenotype across temperatures for GPb and GPc, and compare these results to
our past work on GPa. Across the three GPs, we identify 21 loci that contribute to
the rough phenotype. Of these loci, 20 show conditional phenotypic effects that
are influenced by particular combinations of GP, other loci, and temperature.
Although all three factors prove important, GP is by far the strongest determinant
of which loci show phenotypic effects, impacting nearly all identified loci. In fact,
12 of the detected loci only exert phenotypic effects in the presence of a single
GP. Additionally, we find that the identified loci exhibit varying degrees of
additivity, epistasis, and genotype-environment interaction depending on the GP
that uncovers them. At the molecular level, most, if not all, of the identified loci
influence FLO11 regulation in cis or trans, suggesting that our findings result
from complex genetic and environmental effects on the FLO11 regulatory
network. These findings enhance our understanding of both the extent of cryptic
79
variation within populations and the mechanisms by which GPs reveal cryptic
variants.
4.3 Results
4.3.1 Isolation of a rough flo8∆ sfl1∆ segregant
Our past work showed that expression of the rough phenotype in the BYx3S
cross mainly occurs due to genetic interactions between GPs and polymorphisms
in the Ras pathway (Lee et al. 2016, Taylor and Ehrenreich 2014, Taylor and
Ehrenreich 2015, Taylor et al. 2016). To examine whether genetic variation
beyond the Ras pathway might also be able to contribute to the trait, we knocked
out FLO8 and SFL1 in both BY and 3S using CRISPR/Cas9 (Methods). We then
employed random spore techniques to generate and screen > 100,000 flo8∆
sfl1∆ (GPc) cross progeny for the trait at 21°C (Methods). We used this condition
because the trait is less genetically complex at 21°C than at higher temperatures,
making it easier to find rough segregants in a screen (Lee et al. 2016). The GPc
screen produced a single rough segregant, implying that expression of the rough
phenotype in the absence of Flo8 and Sfl1 requires a complex combination of
alleles inherited from both BY and 3S.
4.3.2 The three GPs vary in their potential to express the phenotype across
temperatures
Our GPb and GPc rough segregants were both obtained at 21°C, but were
limited in their abilities to express the phenotype at higher temperatures.
Following from our previous work showing that GPa segregants have the
potential to express rough morphology across temperatures (Lee et al. 2016), we
assessed whether GPb and GPc strains also have the potential to express the
trait at higher temperatures. To check this, we backcrossed GPb and GPc F2
segregants to both BY and 3S, and then phenotyped the resulting progeny at 21,
30, and 37°C (Fig 4.1F). Note, such backcrossing generates new multi-locus
genotypes that may have the potential to express the rough phenotype at higher
temperatures than the F2 segregants recovered from initial screens.
80
Upon examining 768 GPb and GPc backcross progeny at higher temperatures,
we found that some GPb backcross segregants expressed the trait at higher
temperatures whereas GPc backcross segregants did not (Fig S4.1). We
observed varying degrees of temperature sensitivity among the rough GPb
segregants, with only a minority of rough individuals capable of expressing the
trait at 30 and 37°C (Fig S4.1). These results suggest that the trait has a greater
genetic complexity among GPb segregants that express it across temperatures
than among GPb segregants that only express it at low temperature. They also
show that GPc provides an inherently limited potential for trait expression across
temperatures, perhaps because Flo8 and Sfl1 are necessary for expression of
the trait at higher temperatures.
81
Figure 4.1: Different genetic perturbations (‘GPs’) interact with
cryptic variation to cause rough morphology. (a) Certain GPs can
cause BYx3S segregants to express a rough colony phenotype. (b)
Expression of the colony trait can be temperature-sensitive, with
individuals varying in their abilities to express the phenotype at 30 and
37°C. (c) GPa is a loss-of-function mutation in IRA2, which encodes a Ras
negative regulator. (d) GPb results from recombination between the BY
(blue) FLO11 promoter and 3S (orange) FLO11 coding region. (e) GPc is
a double knockout of FLO8 and SFL1, which encode the main Ras-
dependent transcriptional regulators of FLO11. (f) GPa and GPb occurred
spontaneously in the cross, while GPc was genetically engineered.
Genetic mapping populations were generated using backcrossing, as well
as screening for the trait at 21, 30, and 37°C. Loci associated with the trait
were identified in each backcross for each GP-temperature combination.
Loci detected in the presence of GP were then compared.
4.3.3 The GPs uncover distinct loci
To map loci involved in expression of the rough phenotype across temperatures,
we generated > 60,000 and > 12,000 GPb and GPc backcross segregants,
respectively. First-generation backcross segregants were used exclusively in
mapping, except in the case of the 3S backcross of GPb at high temperature.
Determining the genetic basis of the trait among GPb 3S backcross segregants
required combining information from the first-generation and second-generation
backcrosses (Fig S4.2). GPb segregants were phenotyped at 21, 30 and 37°C,
while GPc segregants were only phenotyped at 21°C. Low-coverage whole
genome sequencing was performed on between 48 and 151 individuals for each
GP–backcross–temperature combination, and loci associated with the trait were
identified based on their enrichment among genotyped segregants (Methods).
82
In past work, we identified eight loci that act in different multi-locus genotypes to
enable GPa segregants to express the trait at particular temperatures (Fig 4.2).
BY and 3S contributed the causal alleles at two and four of these loci,
respectively, while the remaining two loci were detected in both the BY and 3S
states in different multi-locus genotype and temperature contexts (Fig 4.3A).
Here, genetic mapping focused on GPb segregants detected a total of 11 loci
across the three environments (Fig 4.2, Fig S4.3, Table 4.S1). Among these loci,
six and three were detected in the BY and 3S allele states, respectively (Fig
4.3A). The other two loci were identified in both the BY and 3S allele states.
Eight of these loci influenced the trait’s expression independent of temperature.
Of the remaining loci, two were detected among individuals that could express
the trait at up to 30°C, while the third was identified among individuals that could
express the trait at up to 37°C. In addition, we identified 12 loci in the presence of
GPc (Fig 4.2, Fig S4.3, Table S4.1). Among the loci found in the presence of
GPc, seven were contributed by BY and five were contributed by 3S (Fig 4.3A).
Comparison of the loci detected in the presence of each of the three GPs
revealed that 21 distinct genomic regions were identified in total (Fig 4.2, Table 1,
Methods). Among these loci, one was found in the presence of all three GPs, 8
were found in the presence of two GPs, and 12 were found in the presence of a
single GP. This indicates that nearly all of the detected loci show differential
responsiveness to the GPs and that the majority of the loci we have identified are
cryptic variants that only act in the presence of specific GPs.
83
Figure 4.2: Loci identified in the presence of the GPs. Twenty-one loci
were identified in total. Circles indicate cryptic variants, whereas stars
indicate GPs.
4.3.4 The GPs alter genotype-environment interaction
We assessed how identified loci interact with particular GPs, each other, and
temperature to produce the rough phenotype. Our past work on GPa found that
at particular levels of temperature sensitivity, distinct sets of epistatic cryptic
variants form multi-locus genotypes that produce the phenotype (Lee et al. 2016).
Although we could not analyze the GPc segregants in the same manner,
examination of GPb segregants produced results comparable to our findings for
GPa. Among GPb segregants expressing the trait exclusively at 21°C, we
detected a pair of interacting loci on chromosomes XIII and XV, which then
allowed us to identify distinct multi-locus genotypes present among these
individuals (Fig 4.3A, Fig S4.5, Note S4.1, Methods). Similar to our work on
GPa (Lee et al. 2016), only one of the GPb multi-locus genotypes found at 21°C
provided a foundation upon which additional allele substitutions can facilitate trait
expression at higher temperatures (Fig 4.3A, black lines). Among these GPb
segregants exhibiting the trait at higher temperatures, single multi-locus
genotypes indicative of higher-order epistasis facilitated trait expression at
84
temperatures up to 30 and 37°C (Fig 4.3A). Comparison our results for GPb to
our past work on GPa found that more than half of the loci and the exact multi-
locus genotypes differed between the two GPs (Fig 4.3A). These results show
that the GPs significantly modify genotype-environment interactions underlying
the rough phenotype.
4.3.5 The GPs affect additivity and epistasis among identified loci
The only temperature at which all three GPs enable trait expression is 21°C. To
compare the quantitative genetic architectures enabling the three GPs to induce
the rough phenotype at this temperature, we used allele and multi-locus
genotype frequency data from backcrosses (Fig 4.3B through 4.3D, Methods). If
loci act in a predominantly additive manner in the presence of a given GP,
observed multi-locus genotype frequencies should match expected multi-locus
genotype frequencies. In contrast, if epistasis meaningfully contributes to the trait,
observed multi-locus genotype frequencies should depart from expected multi-
locus genotype frequencies. Based on these tests, we observed significant
deviation from expected frequencies for GPa (Fig 4.3B) and GPb (Fig 4.3C),
confirming that epistasis plays a significant role in the trait in the presence of
these GPs. However, data for GPc suggested that trait expression in the
presence of this GP is entirely additive (Fig 4.3D). These findings imply that
although the three GPs in this study can each enable expression of the same
rough phenotype, they do so not only through largely distinct cryptic variants but
also through fundamentally different quantitative genetic architectures.
85
Figure 4.3: Loci detected across GPs and temperature sensitivity
classes. (a) Loci interacting with GPa and GPb at 21 and 30°C act in
different multi-locus genotypes. As shown with black lines, certain
temperature-sensitive genotypes provide the potential for additional alleles
to enable trait expression at higher temperatures. Loci interacting with
GPc only have effects at 21°C. Genes underlying each locus are provided
in Figure S4.4 (b) For GPa individuals expressing the trait at only 21°C,
observed multi-locus genotype frequencies at the four loci segregating in
the 3S backcross do not reflect expected frequencies. This can occur if
certain alleles exhibit epistasis and co-segregate in a non-random manner.
(c) Observed and expected genotype frequencies based on four loci in
86
GPb individuals expressing the trait at only 21°C. (d) Observed genotype
frequencies in GPc individuals closely match expected frequencies,
suggesting involved loci act in an additive manner.
4.3.6 Coding and regulatory variation in FLO11 plays an essential role in the trait
Only a single locus exhibited a phenotypic effect in the presence of all three GPs.
This locus corresponds to the 3S allele of the FLO11 coding region (Lee et al.
2016). The BY and 3S alleles of FLO11 possess not only 57 synonymous and 29
nonsynonymous SNP differences, but also a length polymorphism of ~581
nucleotides (Fig S4.6). FLO11
3S
encodes a longer cell-surface protein than
FLO11
BY
, an attribute that is known to favor the expression of multi-cellular traits
like the rough phenotype (Fidalgo et al. 2006, Fidalgo et al. 2008, Hope and
Dunham 2014, Matsui et al. 2015, Verstrepen et al. 2005, Zara et al. 2009).
Although this length polymorphism is likely causal for the trait, we cannot rule out
the possibility that some of the SNPs also play a role.
In addition to being a required component of GPb, we found the BY allele of the
FLO11 promoter is necessary for GPc segregants to express the trait (Figure
4.4A, Table 1). FLO11 has one of the largest and most complex promoters in S.
cerevisiae, with more than 17 transcription factors and 6 signaling cascades
capable of influencing its regulation (Bruckner and Mosch 2012). Through
genetic engineering experiments, we localized the causal variant in the FLO11
promoter to a Rim101 binding site that is present in 3S but not BY (Fig 4.4B, Fig
4.4C, Fig S4.7). This finding is consistent with the important role that
transcriptional derepression of FLO11 plays in expression of the rough
phenotype (Taylor and Ehrenreich 2015, Taylor et al. 2016). We note that
although Rim101 has been described as a FLO11 activator in other strains of S.
cerevisiae, this role is indirect and mediated through its role in silencing NRG1
(Barrales et al. 2008, Kuchin et al. 2002, Kuchin et al. 2002, Lamb and Mitchell
2003, Lamb and Mitchell 2003), which encodes a repressor that directly binds the
FLO11 promoter. The Rim101 binding site in the 3S FLO11 promoter most likely
87
results in direct repression of FLO11 by Rim101, which reinforces Sfl1-mediated
repression. These findings not only speak to the critical role of Flo11 in
expression of the rough phenotype in the BYx3S cross, but also illustrate how
regulation of this gene by multiple pathways determines the phenotypic effects of
cryptic variation.
Figure 4.4: A transcription factor binding site polymorphism is
required for GPb and GPc rough segregants to express the trait. (a)
The GPb and GPc FLO11 promoters contain portions of both the BY (blue)
and 3S (orange) versions. (b) Genetic engineering experiments identified
a 534 nucleotide segment of the BY-derived promoter as being
responsible for the trait. (c) A Rim101 transcription factor binding site that
is absent in BY was found to be the causal polymorphism.
4.3.7 GPa and GPb rough segregants utilize different sub-pathways that impact
Ras
Excluding the FLO11 coding region and promoter, all remaining loci act in the
presence of only one or two of the GPs. To better understand the highly
contextual effects of these cryptic variants, we attempted to resolve detected loci
to specific genes. Seven of the GPb-responsive loci detected in this study were
previously found to interact with either GPa or other GPs in previous papers, in
which they were mapped to individual genes (Lee et al. 2016, Taylor et al. 2016)
88
(Table 4.1). These genes encode Ras-regulated transcription factors (FLO8,
MSS11, MGA1, SFL1), a protein kinase A subunit (TPK3), a TOR pathway
component (MDS3), and IRA2. Two of the loci found in the current study among
GPb segregants were located on chromosomes II and XIV, and had never been
detected in our past work (Fig 4.2, Table 4.1). Through genetic engineering
experiments, we resolved the chromosome XIV locus to SRV2 (Fig S4.8), which
encodes a post-translational activator of adenylate cyclase. At the chromosome II
locus, the most likely candidate is SRB6, which encodes an essential subunit of
the RNA polymerase II mediator complex. Previously, we showed that mutations
disrupting other mediator components, Srb10 and Srb11 (also known as Ssn3
and Ssn8, respectively), can induce the rough phenotype by interacting with a
subset of the alleles that have a phenotypic effect in the presence of GPa (Taylor
et al. 2016) (Note S4.2).
Although loci interacting with GPa and GPb primarily act through the Ras
pathway and many loci interact with both GPs, some are specific to one or the
other. Among these are TRR1 and END3, which were detected in GPa
segregants but not GPb segregants, as well as MDS3, SRV2, and TPK3, which
were identified in GPb segregants but not GPa segregants. Notably, these genes
play a role in activating the Ras pathway through oxidative stress and actin
organization (Fig 4.5). Actin cytoskeleton stability is required for cell polarity and
yeast-adhesion traits, and is regulated in part by the effects of End3 and Srv2 on
Ras signaling (Du and Ayscough 2009). This process results in increased
production of reactive oxygen species (ROS) through the activity of Tpk3
(Gourlay and Ayscough 2006). ROS accumulation is then influenced by Trr1
(Charizanis et al. 1999) and the TOR pathway, of which Mds3 is a component.
Together, Mds3, Srv2, and Tpk3 form a well-described sub-pathway that
influences Ras activity (Du and Ayscough 2009). Thus, although different loci
cause the trait in the presence of GPa and GPb, they appear to reflect different
sub-pathways affecting the same cellular processes, which ultimately impact how
Ras-regulated transcription factors influence FLO11 activity.
89
Table 4.1: Causal genes underlying identified loci
Genetic
Perturbatio
n Chr. Gene Function Allele
Initial
detection
GPc 2 IRA1
Ras negative regulator,
IRA2 paralog 3S Taylor , 2016
GPb 2 SRB6* RNA Pol II subunit BY this paper
GPc 4 GPI8
GPI transamidase
complex subunit 3S this paper
GPa 4 TRR1 redox state regulator 3S
Taylor and
Ehrenreich, 2014
GPc 5 GCN4
stress response
transcription factor 3S this paper
GPa,b 5 FLO8
Ras activated transcription
factor 3S
Taylor and
Ehrenreich, 2014
GPc 5 PMD1
growth regulator; MDS3
paralog BY this paper
GPb 7 MDS3
TOR pathway growth
regulator 3S Taylor , 2016
GPa,b 7 MGA1 transcriptional activator BY
Taylor and
Ehrenreich, 2015
GPc 8 GPA1 G-protein BY this paper
GPa,b,c 9 FLO11 cell surfaace glycoprotein 3S Lee , 2016
GPb,c 9 prFLO11 FLO11 promoter BY/3S Taylor , 2016
GPb 11 TPK3 Protein Kinase A subunit BY Taylor , 2016
GPc 12 SDC25 Ras GEF BY this paper
GPa,c 12 YPT6 Rab GTPase 3S Lee , 2016
GPa,b 13 MSS11
Ras activated transcription
factor BY
Taylor and
Ehrenreich, 2014
GPb 14 SRV2
activator of adenylate
cyclase BY this paper
GPa,c 14 END3 cell wall morphogenesis BY, 3S
Taylor and
Ehrenreich, 2014
GPb,c 15 IRA2 Ras negative regulator BY Taylor , 2016
GPa,b 15 SFL1
Ras inactivated
transcription factor BY
Taylor and
Ehrenreich, 2015
GPc 16 DIG1
MAPK-regulated
transcription factor 3S this paper
*most likely candidate
4.3.8 Cryptic variation in several pathways underlies the phenotype in GPc
segregants
90
Lastly, we sought to determine the genes harboring cryptic variants that interact
with GPc. Regarding GPc, 3 of the 10 identified loci also interact with either GPa
or b. These loci correspond to END3, IRA2, and a locus on chromosome XII. For
chromosome XII and the remaining seven loci, we identified likely candidate
genes based on our highly resolved genetic mapping data and publicly available
research on these genes’ functions and phenotypic effects. One of these loci
corresponds to IRA1, a Ras negative regulator and paralog of IRA2 (Taylor et al.
2016) (Table 1). To obtain additional support for the remaining candidate genes,
we deleted them from GPc rough segregants and determined whether their loss
affected the colony trait. All seven knockouts had an effect, suggesting genetic
variation in these genes plays a causal role in the rough phenotype (Fig S4.9).
Detection of END3, IRA1, and IRA2 in the absence of Flo8 and Sfl1 indicates
that Ras still contributes to FLO11 regulation in the absence of these
transcription factors, possibly by impacting the activities of other transcription
factors (Estruch 2000). The remaining loci implicate alternative signaling
pathways as being responsible for trait expression in GPc segregants. The genes
we identified in the other intervals were a TOR pathway gene and MDS3 paralog
(PMD1), components of the MAP kinase (MAPK) signaling cascade (DIG1,
GPA1) (Metodiev et al. 2002), an environmentally-responsive transcriptional
activator (GCN4), a Rab GTPase that influences Ras (Costanzo et al. 2010),
MAPK (Costanzo et al. 2016), and Rim101 (Zheng et al. 2010) signaling (YPT6),
a stress responsive guanine exchange factor (SDC25), and an enzyme that post-
translationally modifies proteins to help them anchor into the cell wall (GPI8). Of
particular note for GPA1, the BY strain is known to carry a lab-derived allele that
is an expression QTL hotspot (Yvert et al. 2003), supporting the possibility that
GPA1 allele state might also impact expression of the rough phenotype in the
presence of GPc. Additionally, GPA1
BY
was previously shown to influence other
FLO11-dependent traits through its downstream transcriptional activator Ste12
(Matsui et al. 2015). Nearly all of the genes identified in rough GPc segregants
have the potential to directly or indirectly influence FLO11 transcription. The lone
91
exception is GPI8, which could still influence Flo11 at the protein level, as it is
responsible for adding glycosylphosphatidylinositol (GPI) anchors to new proteins
and could affect Flo11’s binding to the cell surface (Benghezal et al. 1996).
These results show the abundant cryptic variation that exists within the FLO11
gene regulatory network.
4.4 Discussion
We have detailed how different genetic perturbations can reveal distinct cryptic
variants, thereby enabling a multitude of genotype, environment, and phenotype
relationships to produce the same trait. Across the 21 loci that can contribute to
the rough phenotype in the presence of GPa, GPb, and GPc, all but one of them
show phenotypic effects that depend on the GP that is present. This single locus
corresponds to the coding region of FLO11, supporting the notion that regulation
of Flo11 levels and stability is the central determinant of the rough phenotype’s
expression across GPs, combinations of segregating loci, and temperature.
Bolstering the importance of variability in FLO11 regulation to our findings, nearly
all of the loci that exhibit phenotypic effects influence FLO11 in trans. Moreover,
many of these loci are only visible in the absence of a repressive Rim101 binding
site in the FLO11 promoter. These results highlight the potentially important role
that cis-regulatory polymorphisms can play in enabling trans-regulatory
polymorphisms to exert phenotypic effects. Indeed, cis-regulatory polymorphisms
have been shown to modify the effects of trans variants in other systems (Reddy
et al. 2012, Wong et al. 2017), thereby altering how genetic differences impact
traits (Payne and Wagner 2014). More generally, our findings imply that changes
at the level of an organismal phenotype—the rough colony trait—reflect complex
genetic and environmental effects on a key gene’s regulatory network, here
FLO11.
In addition to showing that different GPs enable distinct cryptic variants to have
phenotypic effects, we also demonstrate that the GPs modify how genotype-
92
environment interactions impact the trait. For both GPa and GPb, we find that
certain combinations of epistatic alleles both enable expression of the trait at a
permissive temperature of 21°C, as well as provide the genetic potential for the
phenotype at higher temperatures. However, different loci and multi-locus
genotypes enable GPa and GPb segregants to express the trait at higher
temperatures. In contrast, GPc individuals are unable to express the phenotype
at temperatures above 21°C, implying their potential to express the trait across
environments is constrained. These findings support the concept that not all
genotypes specifying the same trait possess comparable phenotypic robustness
(Ehrenreich and Pfennig 2016, Payne et al. 2014, Pfennig and Ehrenreich 2014,
Siegal and Leu 2014, Wagner 2012). In the case of the rough phenotype, our
genetic mapping and genetic engineering results imply that differences in
phenotypic robustness relate to changes in signaling and transcription factor
activity across multiple pathways and sub-pathways influencing FLO11.
Furthermore, whereas the rough phenotype in GPa and GPb segregants involves
complex epistatic effects, we find no evidence for epistasis in GPc rough
segregants. This implies that by eliminating the main transcriptional regulators of
FLO11, we not only uncovered a different set of cryptic variants but also
converted the trait’s genetic architecture from mainly epistatic to additive.
Perhaps this additivity at the phenotypic level reflects the cumulative effect of
multiple pathways influencing FLO11 expression at the molecular level. While the
majority of the loci that interact with GPa and GPb correspond to components of
the Ras pathway, many of the loci found in the presence of GPc are involved in
other signaling pathways that may have compensatory functions (Fig 4.5). This is
consistent with the idea that eliminating Flo8 and Sfl1 might enable other
pathways to play a stronger role in expression of FLO11 and the rough colony
phenotype.
93
94
Figure 4.5: GPs unmask cryptic genetic variation in parallel signaling
pathways and sub-pathways. (a) Loci that interact with GPa influence
FLO11 regulation by the Ras pathway. (b) GPb uncovers loci that
primarily act through a different Ras sub-pathway involving Mds3, Srv2,
and Tpk3. (c) Loci interacting with GPc function in a number of different
pathways that are capable of regulating FLO11 activity. The locations of
transcription factor binding sites are not intended to reflect specific
positioning along the FLO11 promoter.
Although not an explicit focus of the current manuscript, our results also provide
valuable insights into genetic background effects, the phenomenon in which GPs
show different phenotypic effects in distinct individuals (Chandler et al. 2013,
Ehrenreich 2017, Nadeau 2001). A number of recent studies have shown that
these background effects often result from higher-order epistasis between a GP
and multiple segregating loci (Chandler et al. 2014, Kuzmin et al. 2018, Lee et al.
2016, Miotto et al. 2015, Taylor and Ehrenreich 2014, Taylor and Ehrenreich
2015). Despite supporting an important role for higher-order epistasis in
background effects, our current work, in particular on GPc, also shows that
background effects can have much simpler underpinnings. In the context of GPc,
identified loci each show pairwise epistasis with the GP, but exhibit no epistasis
with each other and thus appear to act additively. These differences in
quantitative genetic architecture again tie back to differences in FLO11 regulation,
consistent with theoretical work suggesting that how GPs perturb gene regulatory
networks can impact whether loci show additive or epistatic effects (Gjuvsland et
al. 2007).
In summary, our study demonstrates the large amount of cryptic variation that
can underlie a single phenotype and strongly implicates complex changes in
transcription as a mechanism regulating this cryptic variation. Our findings also
illustrate how systematically characterizing the impacts of GPs on genotype-
95
environment-phenotype relationships can significantly advance understanding of
some of the most difficult, yet fundamental problems in complex trait genetics
4.5 Methods
4.5.1 Knockout of FLO8 and SFL1 in BY and 3S
FLO8 and SFL1 targeting gRNA sequences were cloned into the pML104
plasmid vector, which carries Cas9 and URA3 (Laughery 2015). Each resulting
plasmid was transformed into the BY and 3S strains alongside a double stranded
oligo to be used as a repair template for homologous recombination (HR). Two
sequential knockouts were performed for each strain to modify both genes. The
lithium acetate method (Daniel Gietz and Woods 2002) was used for
transformation. Repair template oligos were 90 bases long with homology to
either FLO8 or SFL1, but with three bases replaced in frame with a stop codon to
truncate the gene. A one base frameshift deletion was also introduced following
the stop codon to further ensure loss of gene function. Stop codons were
introduced at amino acid 155 and 39 in FLO8 and SFL1, respectively.
Transformed cells were plated on solid yeast nitrogen base (YNB) media lacking
uracil to select for retention of the pML104 plasmid. Presence of the intended
genetic modifications was checked in the transformants using Sanger
sequencing.
4.5.2 Isolation of a rough BYx3S segregant with GPc
The BY and 3S strains engineered with GPc were mated to one another. The
resulting diploid was sporulated and plated at low density (~300 colonies per
plate) onto yeast nitrogen base (YNB) agar containing canavanine to select for
MATa haploids using the synthetic genetic array system (Tong et al. 2001). After
five days at 30°C, colonies were replicated onto yeast extract-peptone-ethanol
(YPE) plates. A single rough colony was identified after four days of growth at
21°C.
96
4.5.3 Generation of backcross segregants
The GPb rough segregant used in this study was obtained through a screen
described in (Taylor et al. 2016), in which it was referred to as ‘Rough segregant
13.’ To generate backcross progeny for genetic mapping and segregation
analysis, we mated the GPb and GPc rough segregants to obtain diploids. The
GPb segregant was backcrossed to wild type BY and 3S, while the GPc
segregant was backcrossed to flo8∆ sfl1∆ BY and 3S strains. Diploids were
sporulated and plated at low density onto YNB plates containing canavanine.
After five days of growth at 30°C, haploid colonies were replicated onto YPE
plates and incubated at 21°C. Segregants expressing rough morphology after
four days of growth on YPE were inoculated into liquid yeast extract-peptone-
dextrose (YPD) media and grown overnight at 30°C. Freezer stocks of rough
segregants were made by mixing aliquots of these cultures with 20% glycerol
solution and storing these glycerol stocks at -80°C.
4.5.4 Phenotyping at multiple temperatures
Cells from freezer stocks were inoculated into liquid YPD media, grown for two
days at 30°C, and then pinned onto three YPE plates. Each plate was then
incubated at a single, constant temperature: 21, 30, or 37°C. The colonies were
then phenotyped for expression of the colony trait after four days of growth. Due
to the low frequency of the rough phenotype in the GPb x 3S backcross at 30
and 37°C, another round of mating was performed to enrich for individuals
expressing the trait in these conditions. Specifically, a 3S backcross segregant
was mated to the BY strain, and segregants were selected and screened for the
trait in the same manner described above.
4.5.5 Genotyping of GPb and GPc rough segregants
For each backcross, segregants expressing the rough phenotype at up to 21, 30,
or 37°C were inoculated into liquid YPD. DNA was extracted from overnight
cultures using the Qiagen DNeasy 96 Blood and Tissue kit. Illumina sequencing
libraries were then prepared using the Illumina Nextera kit, with a unique pair of
97
dual-indexed barcodes for each individual. Between 48 and 156 segregants from
each combination of GP, backcross, and temperatures sensitivity were
sequenced. Sequencing was performed at the Beijing Genomics Institute on an
Illumina HiSeq 4000 using paired-end 100x100 bp reads. Each segregant was
sequenced to an average per site coverage of at least 2.5x. Reads were mapped
onto the S288c reference genome and a 3S draft genome, from the
Saccharomyces Genome Database, using the Burrows-Wheeler Aligner (BWA)
version 7 with options mem -t 20 (Li and Durbin 2009). SAMtools was used to
generate mpileup files (Li et al. 2009), and genome-wide allele frequencies were
calculated at 36,756 SNPs that were previously identified between BY and 3S
(Taylor et al. 2016).
4.5.6 Genetic mapping of loci underlying the rough phenotype
Each segregant’s genotype at 36,756 SNPs was determined using a Hidden
Markov model, implemented in the HMM package in R. Loci associated with the
trait for each combination of GP, backcross, and temperature sensitivity were
identified using binomial tests. Sites were considered statistically significant at a
Bonferroni-corrected threshold of p < .01. Multiple-testing correction was
performed on each backcross population on its own, as the number of unique
tests varied 837 to 1,526 among these populations. We defined the interval
surrounding identified loci by computing the -log10(p-value) at each linked SNP
and determining the SNPs at which this statistic was 2 lower than the peak
marker for a locus. These bounds were used in fine-mapping and to assess
overlap among loci detected in different combinations of GP, backcross, and
temperature sensitivity.
4.5.7 Testing for genotypic heterogeneity
Observed and expected haplotype frequencies for each pair of SNPs were
compared using c
2
tests in a custom Python script. Expected haplotype
frequencies were calculated as the product of the individual allele frequencies at
each of the two sites. To reduce the number of statistical tests, SNPs containing
98
the same information across all segregants in a given backcross population were
collapsed into a single marker. The Benjamini-Hochberg method for False
Discovery Rate (FDR) (Benjamini and Hochberg 1995) was then implemented
using the statsmodels Python module (Seabold and Perktold 2010). A stringent
FDR of 0.0001 was employed and regions of the genome within 30,000 bases of
the ends of chromosomes were excluded to avoid genotyping errors.
4.5.8 Exploration of additivity and epistasis
Expected multi-locus genotype frequencies were calculated as the product of the
frequencies of the individual alleles segregating in a given backcross. For each
population, observed and expected genotype frequencies were compared using
a c
2
test with degrees of freedom equal to one less than the number of possible
genotypes. These analyses were only performed on backcross populations in
which three or more loci segregated.
4.5.9 Genetic engineering at the FLO11 promoter and other loci
Gene deletions at GP-interacting loci were accomplished by replacement with the
kanMX cassette (Wach et al. 1994). Using the lithium acetate transformation
method (Daniel Gietz and Woods 2002), a genomic region of interest was
replaced with a PCR amplicon of the kanMX cassette with 60 bases of homology
to the ends of the region. Transformed cells were plated on YPD + G418 to
select for integration of kanMX and insertion was verified using PCR.
Allele replacements were performed using a two-step, CRISPR/Cas9-mediated
approach. A kanMX deletion strain, generated as described above, was
transformed with the pML104 plasmid (Laughery et al. 2015) carrying the Cas9
gene and a gRNA sequence targeting kanMX, along with a PCR product repair
template that should result in excision of kanMX and result in an allele swap.
Cells were plated on YNB plates lacking uracil to select for retention of pML104.
Replacement of the kanMX cassette was verified by PCR, and transformants
were plated and screened on YPE to determine the phenotypic effects of allele
99
replacement. All replacements were confirmed by Sanger sequencing. Also, in
parallel with each allele replacement, we generated control strains where kanMX
was replaced with the original sequence at a given site. The phenotypes of allele
replacement strains were assessed based on comparison to control strains that
had been generated in parallel.
100
4.6 Supporting Materials
Figure S4.1: GPb and GPc individuals differ in their temperature
sensitivities. (a) A subset of GPb backcross segregants express rough
morphology at 30 and 37°C. (b) GPc backcross segregants are incapable of
expressing the trait at 30 and 37°C. Ninety-six backcross segregants were
generated and phenotyped for each GP-backcross parent combination.
101
Figure S4.2: Scheme to generate GPb backcross mapping populations. A
rough BYx3S F2 segregant carrying GPb was backcrossed to BY and 3S. In the
case of the cross to 3S, a second-generation backcross to BY was then
performed, which was necessary to enrich for segregants that express the rough
phenotype at 37°C.
102
Figure S4.3: Genetic mapping for loci interacting with GPb and GPc. For
each mapping population based on GP, backcross, and temperature sensitivity, a
binomial test was performed for each SNP to identify regions of the genome
associated with the rough phenotype. Results in are plotted in -log10(p-values) in
the following order: GPb (BY backcross, 21, 30, 37°C), GPb (3S backcross,
21°C), GPb (3S backcross, 21, 30°C), GPb (2
nd
generation backcross, 21°C),
GPb (2
nd
generation backcross, 21, 30°C), GPb (2
nd
generation backcross, 21,
30, 37°C), GPc (BY backcross, 21°C), GPc (3S backcross, 21°C).
103
Figure S4.4: Key corresponding to gene position in Figure 4.3A.
104
Figure S4.5: Scan for loci with correlated allele states. Significant p-values
for c
2
tests between each pair of uniquely segregating genomic regions: GPb – (a)
backcross to BY, (b) backcross to 3S, and GPc – (c) backcross to BY, (d)
backcross to 3S. Results are only plotted for the upper triangle. Significant
results were only found in the GPb backcross to 3S.
105
Figure S4.6: Variation in the FLO11 coding region. The 3S strain carries a
longer FLO11 allele relative to BY. This is due to a ~500-600 nucleotide length
polymorphism in the middle of the gene. Nucleotide positions are provided based
on the BY gene sequence.
106
Figure S4.7: Additional prFLO11 fine mapping details. (a) Engineering of the
six polymorphisms within the promoter region shown in Fig 3B revealed that
replacement of a TT BY sequence at position 394,846 to 394,847 with the 3S CC
allele ablates the phenotype. (b) Replacement of the same polymorphism in a
strain with GPc also causes loss of the rough phenotype.
107
Fig S4.8: The SRV2 gene underlies a locus on chromosome XIV that
interacts with GPb. Replacement of the BY SRV2 allele with the 3S version
results in loss of the rough phenotype. A similiar loss of phenotype is also
observed when SRV2 is deleted.
108
Fig S4.9: Deletions support interactions between GPc and identified
candidate genes. Complete deletion of the candidate genes in a GPc rough
segregant was performed using tailed kanMX cassettes. Deletion of six
candidate genes causes loss of the phenotype. For the seventh candidate gene,
deletion enhances the phenotype. Notably, DIG1 encodes a repressor of the
FLO11 activator Ste12. Thus, the phenotypic enhancement seen upon its
deletion is consistent with cryptic variation promoting transcription of FLO11.
109
Table S4.1: Loci detected across GPs and temperatures
GP Chromosome
Maximum
temperature of
trait expression Allele
Upstream
bound
Downstream
bound
-Log10(p)-
value
b 2 21 BY 723181 734674 9.371
b 2 30 BY 705006 734674 10.706
b 2 37 BY 705006 734674 4.056
b 5 30 3S 368771 394911 10.709
b 5 37 3S 356220 383540 8.826
b 5 21, 30, 37 3S 353654 400481 6.665
b 7 30 3S 87038 214328 4.821
b 7 30 BY 894815 985698 6.917
b 7 37 3S 81381 143359 5.078
b 7 37 BY 852002 1000433 4.374
b 9 21 BY 396020 400489 8.101
b 9 21 BY 396027 439742 14.214
b 9 30 3S 386815 394651 10.709
b 9 30 BY 395106 416603 11.929
b 9 37 3S 386817 394001 10.127
b 9 37 BY 395106 396891 6.92
b 9 21, 30, 37 3S 386511 394651 6.665
b 11 21 BY 97364 242623 4.898
b 11 21 BY 97394 243150 5.854
b 11 30 BY 154545 257237 3.703
b 13 21 BY 319105 924189 4.898
b 13 21 BY 395951 664756 5.854
b 13 21 BY 570969 605817 14.214
b 13 30 BY 548950 609784 10.706
b 13 37 BY 554372 923918 4.98
b 14 37 BY 356409 394076 7.711
b 15 21 BY 100250 202290 4.18
b 15 21 BY 111105 223620 4.283
b 15 21 BY 129314 193871 8.708
b 15 21 BY 553016 631585 4.853
b 15 30 BY 131162 299083 3.703
b 15 30 BY 559403 628841 9.617
b 15 37 BY 498862 623604 6.731
c 2 21 3S 503310 531496 40.859
c 4 21 3S 961848 1151234 16.42
110
c 5 21 3S 105743 16630 11.564
c 5 21 BY 377154 438640 8.564
c 8 21 BY 98357 122190 13.319
c 9 21 3S 387502 394427 44.464
c 9 21 BY 395106 397560 26.391
c 12 21 BY 109399 173876 6.429
c 12 21 3S 642815 680926 20.06
c 14 21 BY 445983 473648 17.815
c 15 21 BY 147945 182906 12.549
c 16 21 3S 429704 477416 16.42
111
Note S4.1: For GPb individuals expressing the rough phenotype at 21°C, we
detected two multi-locus genotypes among segregants from the backcross to 3S.
We also identified a third genotype in the second-generation backcross
population, involving a different combination of alleles than those from the initial
3S backcross.
Note S4.2: We deleted each non-essential gene at the GPb chromosome II locus
– RRT2, HIS7, ARO4, MRPS5, MTC4, SHG1, and YBR259W – and found that
none of these deletions caused loss of the rough phenotype. This implies that
loss-of-function in a non-essential gene is not responsible for the trait at this
locus, and the causal genetic variant is either a gain-of-function polymorphism or
is an essential gene. Among the essential genes at this locus, SRB6 was
determined to be the strongest candidate.
112
Chapter 5: Concluding Remarks
5.1 Impact of my work
In this dissertation, I have utilized a yeast colony morphology system to
investigate the mechanisms of cryptic genetic variation at a previously
unexplored level of detail and completion. Across multiple projects, I
characterized the genetic and molecular bases of a model complex trait across
multiple genetic and environmental backgrounds. In doing so, I addressed
important questions in the field:
1. How do genetic perturbations unmask sets of cryptic variants to cause
unexpected trait outcomes?
In chapters 2 through 4, I demonstrated that perturbing IRA2 or other genes
enable sets of cryptic variants in the BYx3S cross to influence the expression of
rough colony morphology. Through genetic engineering experiments, these
variants were found to primarily correspond to genes involved in the Ras
pathway to influence the expression of FLO11. The most thoroughly explored
case occurs in chapter 2, where I identify specific combinations of up to 7 alleles
that cause individuals with a loss-of-function mutation in IRA2 to express the trait.
Here, general derepression of Ras resulting from the ira2∆2933 mutation
modifies the activity of downstream signaling components and causes them to
act as cryptic variants. Furthermore, this general Ras derepression is insufficient
to cause the trait on its own and requires higher-order epistasis involving
standing genetic variants originating from both BY and 3S. This is best seen in
the interaction between ira2∆2933 and FLO8
3S
, the only functional allele of the
FLO8 transcription factor in the BYx3S cross. While FLO8
3S
is required for the
trait in all detected genotypes, its effect on rough colony morphology is only
observed among segregants with ira2∆2933 and also requires at least one
cryptic variant from the BY parent. My work provides one of the most detailed
examples of how introducing a new allele (ira2∆2933) into a population causes
standing alleles that do not normally influence trait variation to have unexpected
113
phenotypic effects. This work also illustrates the degree of higher-order epistasis
that can underlie a phenotype, which may be underexplored in many
experimental systems.
2. What is the relationship between the cryptic variants that enable a trait’s
expression and those that exhibit GxE?
In chapters 2 and 4, I explicitly defined the genes and gene variants that exhibit
GxE in the rough colony system. My findings suggest that the majority of cryptic
variants depend not only on genetic perturbation, but also on environment in that
they may only display functional variation in certain environmental conditions.
Furthermore, GxE only occurs in a subset of individuals that carry these variants,
depending on their genetic background. In my ira2∆2933 and prFLO11
BY
-
FLO11
3S
recombinant segregants, I attributed these genetic background effects
to a polymorphism in the MSS11 transcription factor. MSS11
BY
enables
individuals to express rough colony morphology at 30 and 37°C if they carry the
requisite combination of interacting cryptic variants at additional loci. In
segregants with MSS11
3S
, these interactions are not detected and the additional
loci have no capacity to exhibit GxE. These loci occur in addition to the
genotypes responsible for the trait at 21°C, providing an example of how higher-
order epistasis modifies existing interactions that are less complex.
3. By what mechanisms do different sets of cryptic variants cause the same trait?
Across chapters 2 through 4, I mapped the genetic basis of the rough colony trait
as its results from 16 independent genetic perturbations. In doing so, I
demonstrate that cryptic variation is abundant in this system and reveals itself to
different degrees depending on the nature of the perturbation. In chapter 3, I
show that while the trait’s expression is primarily caused by cryptic variation in
the Ras pathway, each mutation or intragenic recombination unmasks different
combinations of cryptic variants. Here, the number of involved cryptic variants
likely depends on the degree of FLO11 transcriptional derepression caused by
the perturbation itself. For instance, mutations in the RNA polymerase subunits
114
SSN3 and SSN8 were found to interact with 1-2 cryptic variants, whereas
mutations near the top of the Ras cascade in genes such as IRA2 result in more
complex interactions. This is also consistent for individuals with a prFLO11
BY
-
FLO11
3S
recombination, which alters a transcription factor binding site, as the
genetic complexity of the trait is even greater in this background.
In chapter 4, I found that decoupling the trait from the Ras pathway by deleting
FLO8 and SFL1 reveals an alternative set of cryptic variants in parallel signaling
pathways. This appeared to result in the greatest genetic complexity for the trait,
with 12 cryptic variants detected in flo8∆ sfl1∆ rough segregants. Further analysis
revealed that this apparent complexity stems from a different quantitative genetic
architecture. Whereas variants interacting with other genetic perturbations exhibit
higher-order epistasis with one another, these variants have additive effects that
presumably cumulate in a requisite level of FLO11 expression. These findings
provide an explicit example of a threshold model by which sets of cryptic variants
can cause a trait’s expression, as well the ways that genes in alternative
pathways can specify the same trait through different mechanisms.
5.2 Future Directions
Although my results implicate derepression of Ras signaling and FLO11
transcriptional activation as the driving mechanisms of the rough colony trait,
neither of these is explicitly measured in these studies. To further define the
effects of cryptic variants on signaling, transcriptome analysis and DNA
occupancy assays can be utilized to compare the expression and activity of
genes responsible for this signaling across genetic backgrounds. Additionally,
while the rough colony phenotype is defined here as a binary trait, FLO11
expression may have a more continuous range when comparing individuals that
do and do not express the trait. By measuring levels of FLO11 transcription, the
requisite expression threshold for rough colony morphology can be determined
alongside the relative contribution of each cryptic variant. This may be
115
particularly useful in the case of variants that interact with flo8∆ sfl1∆ (GPc),
which are more likely to have an additive effect on FLO11 expression.
While my project provides one of the most thorough examples of how cryptic
variation occurs and functions, the scope of the work is limited based on its
design. The results from chapter 4 imply that other genetic architectures for the
trait may exist but were not detected. Detection of all the genes that can
potentially act as cryptic variants in this system would require a larger scale
screen involving targeted perturbation of many genes, including each gene that
contributes to FLO11 regulation. Screening for the trait in other environments
may also reveal additional environment-responsive cryptic variants.
Although Ras signaling is the primary mechanism for the trait in BYx3S
segregants, this may not be the case for other yeast strains. Introducing alleles
from additional strains into the system may cause different alleles from BY and
3S to act as cryptic variants. Finally, the genetic and molecular mechanisms
described in my work provides a framework for other complex traits, but the
genes involved are specific to a single model phenotype and perhaps closely
related traits. Applying the techniques of conditional genetic mapping to other
traits will provide a broader understanding of the mechanisms behind epistasis,
GxE, and cryptic variation.
116
References
Atsushi, F., K. Yoshiko, K. Satoru, et al. (1989). "Domains of the SFL1 protein of
yeasts are homologous to Myc oncoproteins or yeast heat-shock transcription
factor." Gene 85(2): 321-328.
Atwell, S., Y. S. Huang, B. J. Vilhjalmsson, et al. (2010). "Genome-wide association
study of 107 phenotypes in Arabidopsis thaliana inbred lines." Nature 465(7298):
627-631.
Aylor, D. L., W. Valdar, W. Foulds-Mathes, et al. (2011). "Genetic analysis of
complex traits in the emerging Collaborative Cross." Genome Res 21(8): 1213-
1222.
Badyaev, A. V. (2005). "Stress-induced variation in evolution: from behavioural
plasticity to genetic assimilation." Proc Biol Sci 272(1566): 877-886.
Barrales, R. R., J. Jimenez and J. I. Ibeas (2008). "Identification of novel activation
mechanisms for FLO11 regulation in Saccharomyces cerevisiae." Genetics
178(1): 145-156.
Baryshnikova, A., M. Costanzo, Y. Kim, et al. (2010). "Quantitative analysis of fitness
and genetic interactions in yeast on a genome scale." Nat Methods 7(12): 1017-
1024.
Baye, T. M., T. Abebe and R. A. Wilke (2011). "Genotype-environment interactions
and their translational implications." Per Med 8(1): 59-70.
Bénédetti, H., S. Raths, F. Crausaz, et al. (1994). "The END3 gene encodes a
protein that is required for the internalization step of endocytosis and for actin
cytoskeleton organization in yeast." Mol Biol Cell. 5(9): 1023-1037.
Benghezal, M., A. Benachour, S. Rusconi, et al. (1996). "Yeast Gpi8p is essential for
GPI anchor attachment onto proteins." The EMBO Journal 15(23): 6575-6583.
Benjamini, Y. and Y. Hochberg (1995). "Controlling The False Discovery Rate - A
Practical And Powerful Approach To Multiple Testing." Journal of the Royal
Statistical Society. Series B: Methodological 47: 289-300.
Bergman, A. and M. L. Siegal (2003). "Evolutionary capacitance as a general feature
of complex gene networks." Nature 424(6948): 549-552.
117
Bergstrom, A., J. T. Simpson, F. Salinas, et al. (2014). "A high-definition view of
functional genetic variation from natural yeast genomes." Mol Biol Evol 31(4):
872-888.
Bhatia, A., A. Yadav, C. Zhu, et al. (2014). "Yeast growth plasticity is regulated by
environment-specific multi-QTL interactions." G3 (Bethesda) 4(5): 769-777.
Bloom, J. S., I. M. Ehrenreich, W. T. Loo, et al. (2013). "Finding the sources of
missing heritability in a yeast cross." Nature 494(7436): 234-237.
Boone, C., H. Bussey and B. J. Andrews (2007). "Exploring genetic interactions and
networks with yeast." Nat Rev Genet 8(6): 437-449.
Breslow, D. K., D. M. Cameron, S. R. Collins, et al. (2008). "A comprehensive
strategy enabling high-resolution functional analysis of the yeast genome." Nat
Methods 5(8): 711-718.
Bruckner, S. and H. U. Mosch (2012). "Choosing the right lifestyle: adhesion and
development in Saccharomyces cerevisiae." FEMS Microbiol Rev 36(1): 25-58.
C, C., J. H, K. B, et al. (1999). "The oxidative stress response mediated via
Pos9/Skn7 is negatively regulated by the Ras/PKA pathway in Saccharomyces
cerevisiae." Mol Gen Genet. 261(4-5): 740-752.
Carlborg, O. and C. S. Haley (2004). "Epistasis: too often neglected in complex trait
studies?" Nat Rev Genet 5(8): 618-625.
Carlborg, O., L. Jacobsson, P. Ahgren, et al. (2006). "Epistasis and the release of
genetic variation during long-term selection." Nat Genet 38(4): 418-420.
Chan, E. Y., S. L. Stang, D. A. Bottorff, et al. (1999). "Hypothermic stress leads to
activation of Ras-Erk signaling." J Clin Invest 103(9): 1337-1344.
Chandler, C. H., S. Chari and I. Dworkin (2013). "Does your gene need a
background check? How genetic background impacts the analysis of mutations,
genes, and evolution." Trends Genet 29(6): 358-366.
Chandler, C. H., S. Chari, D. Tack, et al. (2014). "Causes and consequences of
genetic background effects illuminated by integrative genomic analysis."
Genetics 196(4): 1321-1336.
118
Chang, Y. W., S. C. Howard and P. K. Herman (2004). "The Ras/PKA signaling
pathway directly targets the Srb9 protein, a component of the general RNA
polymerase II transcription apparatus." Mol Cell 15(1): 107-116.
Chari, S. and I. Dworkin (2013). "The conditional nature of genetic interactions: the
consequences of wild-type backgrounds on mutational interactions in a genome-
wide modifier screen." PLoS Genet 9(8): e1003661.
Chen, R., L. Shi, J. Hakenberg, et al. (2016). "Analysis of 589,306 genomes
identifies individuals resilient to severe Mendelian childhood diseases." Nat
Biotechnol 34(5): 531-538.
Cherry, J. M., E. L. Hong, C. Amundsen, et al. (2012). "Saccharomyces Genome
Database: the genomics resource of budding yeast." Nucleic Acids Res
40(Database issue): D700-705.
Ciliberti, S., O. C. Martin and A. Wagner (2007). "Robustness can evolve gradually
in complex regulatory gene networks with varying topology." PLoS Comput Biol
3(2): e15.
Cong, L., F. A. Ran, D. Cox, et al. (2013). "Multiplex genome engineering using
CRISPR/Cas systems." Science 339(6121): 819-823.
Costanzo, M., A. Baryshnikova, J. Bellay, et al. (2010). "The genetic landscape of a
cell." Science 327(5964): 425-431.
Costanzo, M., B. VanderSluis, E. N. Koch, et al. (2016). "A global genetic interaction
network maps a wiring diagram of cellular function." Science 353(6306).
Cox, A. D. and C. J. Der (2010). "Ras history: The saga continues." Small GTPases
1(1): 2-27.
D., B. (2015). "Decoding the language of genetics." Cold Spring Harbor Laboratory,
NY: Cold Spring Harbor Laboratory Press.
Dabney, A., J. Storey and W. GR. (2015). "qvalue: q-value estimation for false
discovery rate control."
Daniel Gietz, R. and R. A. Woods (2002). "Transformation of yeast by lithium
acetate/single-stranded carrier DNA/polyethylene glycol method." 350: 87-96.
119
DiCarlo, J. E., J. E. Norville, P. Mali, et al. (2013). "Genome engineering in
Saccharomyces cerevisiae using CRISPR-Cas systems." Nucleic Acids Res
41(7): 4336-4343.
Dixon, S. J., M. Costanzo, A. Baryshnikova, et al. (2009). "Systematic mapping of
genetic interaction networks." Annu Rev Genet 43: 601-625.
Dominguez, A. A., W. A. Lim and L. S. Qi (2016). "Beyond editing: repurposing
CRISPR-Cas9 for precision genome regulation and interrogation." Nat Rev Mol
Cell Biol 17(1): 5-15.
Dowell, R. D., O. Ryan, A. Jansen, et al. (2010). "Genotype to phenotype: a complex
problem." Science 328(5977): 469.
Du, W. and K. R. Ayscough (2009). "Methyl beta-cyclodextrin reduces accumulation
of reactive oxygen species and cell death in yeast." Free Radic Biol Med 46(11):
1478-1487.
Dworkin, I. (2005). "Towards a genetic architecture of cryptic genetic variation and
genetic assimilation: The contribution of K. G. Bateman." Journal of Genetics
84(3): 223-226.
Dworkin, I., A. Palsson, K. Birdsall, et al. (2003). "Evidence that Egfr Contributes to
Cryptic Genetic Variation for Photoreceptor Determination in Natural Populations
of Drosophila melanogaster." Current Biology 13(21): 1888-1893.
Edwards, M. D. and D. K. Gifford (2012). "High-resolution genetic mapping with
pooled sequencing." BMC Bioinformatics 13 Suppl 6: S8.
Ehrenreich, I. M. (2017). "Epistasis: Searching for Interacting Genetic Variants Using
Crosses." G3 (Bethesda) 7(6): 1619-1622.
Ehrenreich, I. M. and D. W. Pfennig (2016). "Genetic assimilation: a review of its
potential proximate causes and evolutionary consequences." Ann Bot 117(5):
769-779.
Ehrenreich, I. M., N. Torabi, Y. Jia, et al. (2010). "Dissection of genetically complex
traits with extremely large pools of yeast segregants." Nature 464(7291): 1039-
1042.
Erdeniz, N., U. H. Mortensen and R. Rothstein (1997). "Cloning-Free PCR-Based
Allele Replacement Methods." Genome Research 7(12): 1174-1183.
120
Estruch, F. (2000). "Stress-controlled transcription factors, stress-induced genes and
stress tolerance in budding yeast." FEMS Microbiology Reviews 24(4): 469-486.
F, S. (1991). Guide to yeast genetics and molecular biology. San Diego, California,
Elsevier Academic Press.
Falconer, D. and T. Mackay (1996). Introduction to quantitative genetics (4th
edition). Harlow, England, Pearson Education Limited.
Felix, M. A. and A. Wagner (2008). "Robustness and evolution: concepts, insights
and challenges from a developmental model system." Heredity (Edinb) 100(2):
132-140.
Fidalgo, M., R. R. Barrales, J. I. Ibeas, et al. (2006). "Adaptive evolution by
mutations in the FLO11 gene." Proc Natl Acad Sci U S A 103(30): 11228-11233.
Fidalgo, M., R. R. Barrales and J. Jimenez (2008). "Coding repeat instability in the
FLO11 gene of Saccharomyces yeasts." Yeast 25(12): 879-889.
Forsberg, S. K., J. S. Bloom, M. J. Sadhu, et al. (2017). "Accounting for genetic
interactions improves modeling of individual quantitative trait phenotypes in
yeast." Nat Genet 49(4): 497-503.
Gaertner, B. E., M. D. Parmenter, M. V. Rockman, et al. (2012). "More than the sum
of its parts: a complex epistatic network underlies natural variation in thermal
preference behavior in Caenorhabditis elegans." Genetics 192(4): 1533-1542.
Gagiano, M., M. Bester, D. Van Dyk, et al. (2002). "Mss11p is a transcription factor
regulating pseudohyphal differentiation, invasive growth and starch metabolism
in Saccharomyces cerevisiae in response to nutrient availability." Molecular
Microbiology 47(1): 119-134.
Geiler-Samerotte, K. A., Y. O. Zhu, B. E. Goulet, et al. (2016). "Selection Transforms
the Landscape of Genetic Variation Interacting with Hsp90." PLoS Biol 14(10):
e2000465.
Genomes Consortium. Electronic address, m. n. g. o. a. a. and C. Genomes (2016).
"1,135 Genomes Reveal the Global Pattern of Polymorphism in Arabidopsis
thaliana." Cell 166(2): 481-491.
Gerke, J., K. Lorenz, S. Ramnarine, et al. (2010). "Gene-environment interactions at
nucleotide resolution." PLoS Genet 6(9): e1001144.
121
Gibson, G. (2009). "Decanalization and the origin of complex disease." Nat Rev
Genet 10(2): 134-140.
Gibson, G. and I. Dworkin (2004). "Uncovering cryptic genetic variation." Nat Rev
Genet 5(9): 681-690.
Gilbert, L. A., M. H. Larson, L. Morsut, et al. (2013). "CRISPR-mediated modular
RNA-guided regulation of transcription in eukaryotes." Cell 154(2): 442-451.
Gjuvsland, A. B., B. J. Hayes, S. W. Omholt, et al. (2007). "Statistical epistasis is a
generic feature of gene regulatory networks." Genetics 175(1): 411-420.
Gourlay, C. W. and K. R. Ayscough (2006). "Actin-induced hyperactivation of the
Ras signaling pathway leads to apoptosis in Saccharomyces cerevisiae." Mol
Cell Biol 26(17): 6487-6501.
Guarente, L. (1993). "Synthetic enhancement in gene interaction: a genetic tool
come of age." Trends Genet 9(10): 362-366.
Gutteling, E. W., A. Doroszuk, J. A. Riksen, et al. (2007). "Environmental influence
on the genetic correlations between life-history traits in Caenorhabditis elegans."
Heredity (Edinb) 98(4): 206-213.
Hermisson, J. and G. P. Wagner (2004). "The population genetic theory of hidden
variation and genetic robustness." Genetics 168(4): 2271-2284.
Hope, E. A. and M. J. Dunham (2014). "Ploidy-regulated variation in biofilm-related
phenotypes in natural isolates of Saccharomyces cerevisiae." G3 (Bethesda)
4(9): 1773-1786.
Hou, J., A. Friedrich, J. S. Gounot, et al. (2015). "Comprehensive survey of
condition-specific reproductive isolation reveals genetic incompatibility in yeast."
Nat Commun 6: 7214.
Hou, J. and J. Schacherer (2017). "Fitness Trade-Offs Lead to Suppressor
Tolerance in Yeast." Mol Biol Evol 34(1): 110-118.
Hou, J., A. Sigwalt, T. Fournier, et al. (2016). "The Hidden Complexity of Mendelian
Traits across Natural Yeast Populations." Cell Rep 16(4): 1106-1114.
Ip, C. L., M. Loose, J. R. Tyson, et al. (2015). "MinION Analysis and Reference
Consortium: Phase 1 data release and analysis." F1000Res 4: 1075.
J., H. (2005). "Genetic Suppression." Wormbook: 1-13.
122
Jain, M., I. T. Fiddes, K. H. Miga, et al. (2015). "Improved data analysis for the
MinION nanopore sequencer." Nat Methods 12(4): 351-356.
Jarosz, D. F. and S. Lindquist (2010). "Hsp90 and environmental stress transform
the adaptive value of natural genetic variation." Science 330(6012): 1820-1824.
JL., H., B. Garvick and H. L. (2001). "Principles for the buffering of genetic variation."
Science 291: 1001-1004.
Jones, G. M., J. Stalker, S. Humphray, et al. (2008). "A systematic library for
comprehensive overexpression screens in Saccharomyces cerevisiae." Nat
Methods 5(3): 239-241.
Jordan, D. M., S. G. Frangakis, C. Golzio, et al. (2015). "Identification of cis-
suppression of human disease mutations by comparative genomics." Nature
524(7564): 225-229.
Julien, P., B. Minana, P. Baeza-Centurion, et al. (2016). "The complete local
genotype-phenotype landscape for the alternative splicing of a human exon." Nat
Commun 7: 11558.
Kim, H. Y., S. B. Lee, H. S. Kang, et al. (2014). "Two distinct domains of Flo8
activator mediates its role in transcriptional activation and the physical interaction
with Mss11." Biochem Biophys Res Commun 449(2): 202-207.
Kim, T. S., H. Y. Kim, J. H. Yoon, et al. (2004). "Recruitment of the Swi/Snf complex
by Ste12-Tec1 promotes Flo8-Mss11-mediated activation of STA1 expression."
Mol Cell Biol 24(21): 9542-9556.
Kobayashi, O., H. Suda, T. Ohtani, et al. (1996). "Molecular cloning and analysis of
the dominant flocculation gene FLO8 from Saccharomyces cerevisiae." Mol Gen
Genet 251(6): 707-715.
Konermann, S., M. D. Brigham, A. E. Trevino, et al. (2015). "Genome-scale
transcriptional activation by an engineered CRISPR-Cas9 complex." Nature
517(7536): 583-588.
Krauthammer, M., Y. Kong, A. Bacchiocchi, et al. (2015). "Exome sequencing
identifies recurrent mutations in NF1 and RASopathy genes in sun-exposed
melanomas." Nat Genet 47(9): 996-1002.
123
Kuchin, S., V. K. Vyas and M. Carlson (2002). "Snf1 Protein Kinase and the
Repressors Nrg1 and Nrg2 Regulate FLO11, Haploid Invasive Growth, and
Diploid Pseudohyphal Differentiation." Molecular and Cellular Biology 22(12):
3994-4000.
Kuchin, S., V. K. Vyas and M. Carlson (2002). "Snf1 protein kinase and the
repressors Nrg1 and Nrg2 regulate FLO11, haploid invasive growth, and diploid
pseudohyphal differentiation." Mol Cell Biol 22(12): 3994-4000.
Kuzmin, E., B. VanderSluis, W. Wang, et al. (2018). "Systematic analysis of complex
genetic interactions." Science 360(6386).
Lamb, T. M. and A. P. Mitchell (2003). "The Transcription Factor Rim101p Governs
Ion Tolerance and Cell Differentiation by Direct Repression of the Regulatory
Genes NRG1 and SMP1 in Saccharomyces cerevisiae." Molecular and Cellular
Biology 23(2): 677-686.
Lamb, T. M. and A. P. Mitchell (2003). "The transcription factor Rim101p governs ion
tolerance and cell differentiation by direct repression of the regulatory genes
NRG1 and SMP1 in Saccharomyces cerevisiae." Mol Cell Biol 23(2): 677-686.
Lande, R. (2009). "Adaptation to an extraordinary environment by evolution of
phenotypic plasticity and genetic assimilation." J Evol Biol 22(7): 1435-1446.
Larson, M. H., L. A. Gilbert, X. Wang, et al. (2013). "CRISPR interference (CRISPRi)
for sequence-specific control of gene expression." Nat Protoc 8(11): 2180-2196.
Laughery, M. F., T. Hunter, A. Brown, et al. (2015). "New vectors for simple and
streamlined CRISPR-Cas9 genome editing in Saccharomyces cerevisiae." Yeast
32(12): 711-720.
Le Rouzic, A. and O. Carlborg (2008). "Evolutionary potential of hidden genetic
variation." Trends Ecol Evol 23(1): 33-37.
Lee, J. T., M. B. Taylor, A. Shen, et al. (2016). "Multi-locus Genotypes Underlying
Temperature Sensitivity in a Mutationally Induced Trait." PLoS Genet 12(3):
e1005929.
Lehner, B., C. Crombie, J. Tischler, et al. (2006). "Systematic mapping of genetic
interactions in Caenorhabditis elegans identifies common modifiers of diverse
signaling pathways." Nat Genet 38(8): 896-903.
124
Li, C., W. Qian, C. J. Maclean, et al. (2016). "The fitness landscape of a tRNA gene."
Science 352.
Li, H. and R. Durbin (2009). "Fast and accurate short read alignment with Burrows-
Wheeler transform." Bioinformatics 25(14): 1754-1760.
Li, H., B. Handsaker, A. Wysoker, et al. (2009). "The Sequence Alignment/Map
format and SAMtools." Bioinformatics 25(16): 2078-2079.
Liti, G., D. M. Carter, A. M. Moses, et al. (2009). "Population genomics of domestic
and wild yeasts." Nature 458(7236): 337-341.
Liu, G., M. Y. Yong, M. Yurieva, et al. (2015). "Gene Essentiality Is a Quantitative
Property Linked to Cellular Evolvability." Cell 163(6): 1388-1399.
Liu, H., C. Styles and G. Fink (1996). "Saccharomyces cerevisiae S288C has a
mutation in FLO8, a gene required for filamentous growth." Genetics 144(3): 967-
978.
Lo, W. S. and A. M. Dranginis (1996). "FLO11, a yeast gene related to the STA
genes, encodes a novel cell surface flocculin." Journal of Bacteriology 178(24):
7144-7151.
Lo, W. S. and A. M. Dranginis (1996). "FLO11, a yeast gene related to the STA
genes, encodes a novel cell surface flocculin." J Bacteriol 178(24): 7144-7151.
Lorenz, M. and J. Heitman (1998). "Regulators of pseudohyphal differentiation in
Saccharomyces cerevisiae identified through multicopy suppressor analysis in
ammonium permease mutant strains." Genetics 150(4): 1443-1457.
Lynch, M. and B. Walsh (1998). Genetics and analysis of quantitative traits.
Sunderland, Massachusetts, Sinauer Associates, Inc.
Mackay, T. F. (2014). "Epistasis and quantitative traits: using model organisms to
study gene-gene interactions." Nat Rev Genet 15(1): 22-33.
Mackay, T. F., S. Richards, E. A. Stone, et al. (2012). "The Drosophila melanogaster
Genetic Reference Panel." Nature 482(7384): 173-178.
Mackay, T. F., E. A. Stone and J. F. Ayroles (2009). "The genetics of quantitative
traits: challenges and prospects." Nat Rev Genet 10(8): 565-577.
125
Magtanong, L., C. H. Ho, S. L. Barker, et al. (2011). "Dosage suppression genetic
interaction networks enhance functional wiring diagrams of the cell." Nat
Biotechnol 29(6): 505-511.
Mali, P., L. Yang, K. M. Esvelt, et al. (2013). "RNA-guided human genome
engineering via Cas9." Science 339(6121): 823-826.
Manchia, M., J. Cullis, G. Turecki, et al. (2013). "The impact of phenotypic and
genetic heterogeneity on results of genome wide association studies of complex
diseases." PLoS One 8(10): e76295.
Mani, R., R. P. St Onge, J. L. t. Hartman, et al. (2008). "Defining genetic interaction."
Proc Natl Acad Sci U S A 105(9): 3461-3466.
Manolio, T. A., F. S. Collins, N. J. Cox, et al. (2009). "Finding the missing heritability
of complex diseases." Nature 461(7265): 747-753.
Masel, J. (2013). "Q&A: Evolutionary capacitance." BMC Biol 11: 103.
Masel, J. and M. L. Siegal (2009). "Robustness: mechanisms and consequences."
Trends Genet 25(9): 395-403.
Masel, J. and M. V. Trotter (2010). "Robustness and evolvability." Trends Genet
26(9): 406-414.
Matsui, T., R. Linder, J. Phan, et al. (2015). "Regulatory Rewiring in a Cross Causes
Extensive Genetic Heterogeneity." Genetics 201(2): 769-777.
McDonald, C. M., M. Wagner, M. J. Dunham, et al. (2009). "The Ras/cAMP pathway
and the CDK-like kinase Ime2 regulate the MAPK Smk1 and spore
morphogenesis in Saccharomyces cerevisiae." Genetics 181(2): 511-523.
McGuigan, K. and C. M. Sgro (2009). "Evolutionary consequences of cryptic genetic
variation." Trends Ecol Evol 24(6): 305-311.
Metodiev, M. V., D. Matheos, M. D. Rose, et al. (2002). "Regulation of MAPK
function by direct interaction with the mating-specific Galpha in yeast." Science
296(5572): 1483-1486.
Michelmore, R. W., I. Paran and R. V. Kesseli (1991). "Identification of markers
linked to disease-resistance genes by bulked segregant analysis: a rapid method
to detect markers in specific genomic regions by using segregating populations."
Proceedings of the National Academy of Sciences 88(21): 9828-9832.
126
Milloz, J., F. Duveau, I. Nuez, et al. (2008). "Intraspecific evolution of the intercellular
signaling network underlying a robust developmental system." Genes Dev
22(21): 3064-3075.
Miotto, O., R. Amato, E. A. Ashley, et al. (2015). "Genetic architecture of artemisinin-
resistant Plasmodium falciparum." Nat Genet 47(3): 226-234.
Moffitt, T. E., A. Caspi and M. Rutter (2005). "Strategy for investigating interactions
between measured genes and measured environments." Arch Gen Psychiatry
62(5): 473-481.
Mullis, M. N., T. Matsui, R. Schell, et al. (2018). "The complex underpinnings of
genetic background effects." Nat Commun 9(1): 3548.
Nadeau, J. H. (2001). "Modifier genes in mice and humans." Nat Rev Genet 2(3):
165-174.
Narasimhan, V. M., K. A. Hunt, D. Mason, et al. (2016). "Health and population
effects of rare gene knockouts in adult humans with related parents." Science
352(6284): 474-477.
Paaby, A. B. and M. V. Rockman (2014). "Cryptic genetic variation: evolution's
hidden substrate." Nat Rev Genet 15(4): 247-258.
Paaby, A. B., A. G. White, D. D. Riccardi, et al. (2015). "Wild worm embryogenesis
harbors ubiquitous polygenic modifier variation." Elife 4.
Pan, X. and J. Heitman (2002). "Protein Kinase A Operates a Molecular Switch That
Governs Yeast Pseudohyphal Differentiation." Molecular and Cellular Biology
22(12): 3981-3993.
Payne, J. L., J. H. Moore and A. Wagner (2014). "Robustness, evolvability, and the
logic of genetic regulation." Artif Life 20(1): 111-126.
Payne, J. L. and A. Wagner (2014). "The robustness and evolvability of transcription
factor binding sites." Science 343(6173): 875-877.
Pedrajas, J. R., E. Kosmidou, A. Miranda-Vizuete, et al. (1999). "Identification and
Functional Characterization of a Novel Mitochondrial Thioredoxin System
inSaccharomyces cerevisiae." Journal of Biological Chemistry 274(10): 6366-
6373.
127
Perez-Pinera, P., D. D. Kocak, C. M. Vockley, et al. (2013). "RNA-guided gene
activation by CRISPR-Cas9-based transcription factors." Nat Methods 10(10):
973-976.
Pfennig, D. W. and I. M. Ehrenreich (2014). "Towards a gene regulatory network
perspective on phenotypic plasticity, genetic accommodation and genetic
assimilation." Mol Ecol 23(18): 4438-4440.
Phan, V. T., V. W. Ding, F. Li, et al. (2010). "The RasGAP proteins Ira2 and
neurofibromin are negatively regulated by Gpb1 in yeast and ETEA in humans."
Mol Cell Biol 30(9): 2264-2279.
Phillips, P. C. (2008). "Epistasis--the essential role of gene interactions in the
structure and evolution of genetic systems." Nat Rev Genet 9(11): 855-867.
Pigliucci, M. and C. J. Murren (2003). "Perspective: Genetic Assimilation and a
Possible Evolutionary Paradox: Can Macroevolution Sometimes Be So Fast as to
Pass Us By?" Evolution 57(7): 1455-1464.
Pigliucci, M., C. J. Murren and C. D. Schlichting (2006). "Phenotypic plasticity and
evolution by genetic assimilation." J Exp Biol 209(Pt 12): 2362-2367.
Prelich, G. (1999). "Suppression mechanisms: themes from variations." Trends
Genet 7.
Puchta, O., B. Cseke, H. Czaja, et al. (2016). "Network of epistatic interactions within
a yeast snoRNA." Science 352.
Qi, L. S., M. H. Larson, L. A. Gilbert, et al. (2013). "Repurposing CRISPR as an
RNA-guided platform for sequence-specific control of gene expression." Cell
152(5): 1173-1183.
Queitsch, C., T. A. Sangster and S. Lindquist (2002). "Hsp90 as a capacitor of
phenotypic variation." Nature 417(6889): 618-624.
Rauw, W. M. and L. Gomez-Raya (2015). "Genotype by environment interaction and
breeding for robustness in livestock." Front Genet 6: 310.
Reddy, T. E., J. Gertz, F. Pauli, et al. (2012). "Effects of sequence variation on
differential allelic transcription factor occupancy and gene expression." Genome
Res 22(5): 860-869.
128
Reid, R. J. D., M. Lisby and R. Rothstein (2002). "Cloning-free genome alterations in
saccharomyces cereuisiae using adaptamer-mediated PCR." 350: 258-277.
Richardson, J. B., L. D. Uppendahl, M. K. Traficante, et al. (2013). "Histone variant
HTZ1 shows extensive epistasis with, but does not increase robustness to, new
mutations." PLoS Genet 9(8): e1003733.
Robertson, L. S. and G. R. Fink (1998). "The three yeast A kinases have specific
signaling functions in pseudohyphal growth." Proceedings of the National
Academy of Sciences 95(23): 13783-13787.
Rutherford, S. L. and S. Lindquist (1998). "Hsp90 as a capacitor for morphological
evolution." Nature 396(6709): 336-342.
Sandrock, T. M., J. L. O'Dell and A. E. M. Adams (1997). "AlleleSpecific
Suppression by Formation of New Protein-Protein Interactions in Yeast."
Genetics 147.
Sarkisyan, K. S., D. A. Bolotin, M. V. Meer, et al. (2016). "Local fitness landscape of
the green fluorescent protein." Nature 533(7603): 397-401.
Schell, R., M. Mullis and I. M. Ehrenreich (2016). "Modifiers of the Genotype-
Phenotype Map: Hsp90 and Beyond." PLoS Biol 14(11): e2001015.
Seabold, S. and J. Perktold (2010). Statsmodels: Econometric and Statistical
Modeling with Python.
Shao, H., L. C. Burrage, D. S. Sinasac, et al. (2008). "Genetic architecture of
complex traits: large phenotypic effects and pervasive epistasis." Proc Natl Acad
Sci U S A 105(50): 19910-19914.
Siegal, M. L. and A. Bergman (2002). "Waddington's canalization revisited:
developmental stability and evolution." Proc Natl Acad Sci U S A 99(16): 10528-
10532.
Siegal, M. L. and J. Y. Leu (2014). "On the Nature and Evolutionary Impact of
Phenotypic Robustness Mechanisms." Annu Rev Ecol Evol Syst 45: 496-517.
Song, W. and M. Carlson (1998). "Srb/mediator proteins interact functionally and
physically with transcriptional repressor Sfl1." EMBO J 17(19): 5757-5765.
Sopko, R., D. Huang, N. Preston, et al. (2006). "Mapping pathways and phenotypes
by systematic gene overexpression." Mol Cell 21(3): 319-330.
129
St Onge, R. P., R. Mani, J. Oh, et al. (2007). "Systematic pathway analysis using
high-resolution fitness profiling of combinatorial gene deletions." Nat Genet 39(2):
199-206.
Storey, J. D. and R. Tibshirani (2003). "Statistical significance for genomewide
studies." Proc Natl Acad Sci U S A 100(16): 9440-9445.
Storici, F., L. K. Lewis and M. A. Resnick (2001). "In vivo site-directed mutagenesis
using oligonucleotides." Nat Biotechnol 19(8): 773-776.
Tanaka, K., M. Nakafuku, F. Tamanoi, et al. (1990). "IRA2, a second gene of
Saccharomyces cerevisiae that encodes a protein with a domain homologous to
mammalian ras GTPase-activating protein." Mol Cell Biol. 10(8): 4303-4313.
Tanenbaum, M. E., L. A. Gilbert, L. S. Qi, et al. (2014). "A protein-tagging system for
signal amplification in gene expression and fluorescence imaging." Cell 159(3):
635-646.
Tang, H., J. Xu and M. Cai (2000). "Pan1p, End3p, and Sla1p, Three Yeast Proteins
Required for Normal Cortical Actin Cytoskeleton Organization, Associate with
Each Other and Play Essential Roles in Cell Wall Morphogenesis." Mol Cell Biol.
20(1): 12-25.
Taylor, M. B. and I. M. Ehrenreich (2014). "Genetic interactions involving five or
more genes contribute to a complex trait in yeast." PLoS Genet 10(5): e1004324.
Taylor, M. B. and I. M. Ehrenreich (2015). "Higher-order genetic interactions and
their contribution to complex traits." Trends Genet 31(1): 34-40.
Taylor, M. B. and I. M. Ehrenreich (2015). "Transcriptional Derepression Uncovers
Cryptic Higher-Order Genetic Interactions." PLoS Genet 11(10): e1005606.
Taylor, M. B., J. Phan, J. T. Lee, et al. (2016). "Diverse genetic architectures lead to
the same cryptic phenotype in a yeast cross." Nat Commun 7: 11669.
Tischler, J., B. Lehner, N. Chen, et al. (2006). "Combinatorial RNA interference in
Caenorhabditis elegans reveals that redundancy between gene duplicates can
be maintained for more than 80 million years of evolution." Genome Biol 7(8):
R69.
130
Toda, T., S. Cameron, P. Sass, et al. (1987). "Three different genes in S. cerevisiae
encode the catalytic subunits of the cAMP-dependent protein kinase." Cell 50(2):
277-287.
Tong, A. H., M. Evangelista, A. B. Parsons, et al. (2001). "Systematic genetic
analysis with ordered arrays of yeast deletion mutants." Science 294(5550):
2364-2368.
van Leeuwen, J., C. Pons, J. C. Mellor, et al. (2016). "Exploring genetic suppression
interactions on a global scale." Science 354(6312).
Verstrepen, K. J., A. Jansen, F. Lewitter, et al. (2005). "Intragenic tandem repeats
generate functional variability." Nat Genet 37(9): 986-990.
Visser, J. A. G. M., J. Hermisson, G. P. Wagner, et al. (2003). "Perspective:
Evolution and Detection of Genetic Robustness." Evolution 57(9): 1959-1972.
Wach, A., A. Brachat, R. Pöhlmann, et al. (1994). "New heterologous modules for
classical or PCR-based gene disruptions inSaccharomyces cerevisiae." Yeast
10(13): 1793-1808.
Waddington, C. H. (1942). "Canalization of Development and the Inheritance of
Acquired Characters." Nature 150(3811): 563-565.
Wagner, A. (2005). "Distributed robustness versus redundancy as causes of
mutational robustness." Bioessays 27(2): 176-188.
Wagner, A. (2012). "The role of robustness in phenotypic adaptation and
innovation." Proc Biol Sci 279(1732): 1249-1258.
Walsh, T. and M. C. King (2007). "Ten genes for inherited breast cancer." Cancer
Cell 11(2): 103-105.
Walsh, T., J. M. McClellan, S. E. McCarthy, et al. (2008). "Rare structural variants
disrupt multiple genes in neurodevelopmental pathways in schizophrenia."
Science 320(5875): 539-543.
Wang, Y., C. D. Arenas, D. M. Stoebel, et al. (2013). "Genetic background affects
epistatic interactions between two beneficial mutations." Biol Lett 9(1): 20120328.
Wenger, J. W., K. Schwartz and G. Sherlock (2010). "Bulk segregant analysis by
high-throughput sequencing reveals a novel xylose utilization gene from
Saccharomyces cerevisiae." PLoS Genet 6(5): e1000942.
131
Wong, E. S., B. M. Schmitt, A. Kazachenka, et al. (2017). "Interplay of cis and trans
mechanisms driving transcription factor binding and gene expression evolution."
Nat Commun 8(1): 1092.
Yvert, G., R. B. Brem, J. Whittle, et al. (2003). "Trans-acting regulatory variation in
Saccharomyces cerevisiae and the role of transcription factors." Nat Genet 35(1):
57-64.
Zara, G., S. Zara, C. Pinna, et al. (2009). "FLO11 gene length and transcriptional
level affect biofilm-forming ability of wild flor strains of Saccharomyces
cerevisiae." Microbiology 155(Pt 12): 3838-3846.
Zeng, Z. B. (2005). "QTL mapping and the genetic basis of adaptation: recent
developments." Genetica 123(1-2): 25-37.
Zheng, J., J. J. Benschop, M. Shales, et al. (2010). "Epistatic relationships reveal the
functional organization of yeast transcription factors." Mol Syst Biol 6: 420.
132
Appendix A: Genetic suppression – Extending our knowledge from lab
experiments to natural populations
This chapter appears as published in Bioessays, 2017. 39(7)
A.1 Summary
Many mutations have deleterious phenotypic effects that can be alleviated by
suppressor mutations elsewhere in the genome. High-throughput approaches
have facilitated the large-scale identification of these suppressors and have
helped shed light on core functional mechanisms that give rise to suppression.
Following reports that suppression occurs naturally within species, it is important
to determine how our understanding of this phenomenon based on lab
experiments extends to genetically diverse natural populations. Although
suppression is typically mediated by individual genetic changes in lab
experiments, recent studies have shown that suppression in natural populations
can involve combinations of genetic variants. This difference in complexity
suggests that sets of variants can exhibit similar functional effects to individual
suppressors found in lab experiments. In this review, we discuss how
characterizing the way in which these variants jointly lead to suppression could
provide important insights into the genotype-phenotype map that are relevant to
evolution and health.
133
A.2 Introduction
Genetic interactions occur when combinations of mutations show phenotypic
effects that differ from expectations based on individual mutations (Boone et al.
2007, Carlborg and Haley 2004, Costanzo et al. 2010, Costanzo et al. 2016,
Forsberg et al. 2017, JL. et al. 2001, Mackay 2014, Mani et al. 2008, Phillips
2008, Schell et al. 2016, Taylor and Ehrenreich 2015, van Leeuwen et al. 2016).
Among the types of genetic interactions that can occur (Breslow et al. 2008, St
Onge et al. 2007, van Leeuwen et al. 2016), suppression represents an extreme
case in which one or more genetic changes at other sites in the genome (i.e.,
‘suppressors’) reverse a mutation’s deleterious effects (D. 2015, Guarente 1993).
Identifying these suppressors can provide valuable insights into the functional
mechanisms by which mutations jointly affect phenotype (D. 2015, Dowell et al.
2010, Prelich 1999). As we describe below, high-throughput sequencing and
genomics strategies have led to new approaches for rapidly identifying mutation-
suppressor combinations on a large scale. These methods have helped produce
general insights into the mechanisms that can cause suppression within and
between genes in lab experiments.
Suppression has also become a point of interest for researchers focused on
understanding how naturally occurring genetic differences among individuals
alter the effects of mutations, thereby leading to incomplete penetrance and
variable expressivity. For example, recent studies in yeast (Dowell et al. 2010,
Hou et al. 2015, Taylor et al. 2016) and humans (Chen et al. 2016, Jordan et al.)
have shown that the ability to suppress particular large effect and Mendelian
mutations can segregate within populations. At present, the extent to which lab
studies on suppression relate to these cases of naturally occurring suppression is
unclear. Notably, while suppression in the lab typically entails one suppressor
interacting with a mutation (van Leeuwen et al. 2016), suppression in natural
contexts may involve combinations of genetic variants that collectively revert the
effect of a mutation (Fig. A1).
134
In this review, we attempt to broadly synthesize work on suppression and
leverage knowledge gained from lab experiments to provide insights into the
genetic and molecular basis of suppression in natural populations.
Figure A1. Genetic basis of suppression in lab versus natural
environments. A: In lab experiments, populations start out as genetically
identical and suppressors arise as new single mutations on the
background of the original mutation. B: Large amounts of genetic variation
segregate within natural populations and some of these variants can
individually or collectively act as suppressors in a similar fashion to the
large-effect mutations identified in lab experiments. In both A and B, each
chromosome represents a haploid individual. Grey chromosomes exhibit
the mutant phenotype, while black chromosomes do not. Stars represent a
mutation of interest, while vertical bars show other genetic changes in the
population. Red vertical bars individually or collectively act as suppressors,
while black vertical bars have no effect on the mutant phenotype.
135
A.3 Main Text
A.3.1 High-throughput techniques for identifying suppressor mutations in lab
experiments
High-throughput approaches have facilitated the comprehensive identification of
suppressors in experimental systems. In the following section, we summarize
these techniques, differentiating between approaches that primarily enable
identification of suppressors that occur in the same gene as the mutation whose
effect they alleviate (‘intragenic’ suppressors) and suppressors that occur in
different genes (‘intergenic’ or ‘extragenic’ suppressors):
Figure A2. Identifying intragenic interactions across an entire gene.
A: Using directed or random mutagenesis, a library of mutants can be
constructed such that every pairwise (or potentially higher-order)
combination of nucleotides in a gene (YFG – ‘your favorite gene’) is
mutated. B: Each single and double mutant can be examined in order to
identify all of the potential genetic interactions that can occur within a gene.
Colors refer to the difference between the observed fitness of each double
mutant minus the expected value given the fitnesses of the two single
mutants. Instances of suppression may be observed if the fitness of a
double mutant is higher than expected.
136
Comprehensive mutagenesis of individual genes: Researchers attempting to
identify intragenic suppressors can employ site-directed or random mutagenesis
to generate a large subset, if not all, of the possible single and double mutants of
a given gene (Julien et al. 2016, Li et al. 2016, Puchta et al. 2016, Sarkisyan et al.
2016) (Fig. A2a). Comparison of single and double mutant phenotypes can be
used to identify cases of intragenic suppression (Fig. A2b). Studies of this type
can examine a large fraction of the possible genetic interactions within a given
gene, especially if a gene is small. Such projects are presently constrained by
the read lengths of short read sequencing technologies and the throughputs and
error rates of long read sequencing technologies (Ip et al. 2015, Jain et al. 2015).
However, as long read sequencing technologies increase in throughput and
accuracy, larger genes, and potentially even sets of functionally related genes,
may become amenable to gene-specific mutagenesis techniques.
Mapping induced or spontaneous suppressors: Suppressors can be obtained by
screening for induced or spontaneous mutations that revert the phenotype of the
original mutant (Fig. A3a). However, revertants recovered from these screens
typically carry multiple mutations and which mutation is the suppressor is not
always clear-cut (D. 2015, van Leeuwen et al. 2016). A straightforward strategy
to distinguish a mutation with a phenotypic effect from its co-occurring
‘passenger’ mutations is by using crosses in combination with whole genome
sequencing (Taylor et al. 2016, van Leeuwen et al. 2016). Recently, such an
approach was used to identify more than 200 mutation-suppressor pairs in a
single study (van Leeuwen et al. 2016).
Genome-wide knockout screens: Large-scale screens for genetic interactions
have been performed by systematically deleting each gene in the genome in a
mutant background (Fig. A3b). In yeast for instance, this has been accomplished
by crossing strains that carry a query mutation to a genome-wide collection of
gene deletion strains to generate every possible double deletion mutant
137
(Baryshnikova et al. 2010, van Leeuwen et al. 2016). With the availability of
CRISPR/Cas9 gene editing technologies (Cong et al. 2013, DiCarlo et al. 2013,
Mali et al. 2013), similar genome-wide genetic interaction screens can now be
implemented in other species. A caveat here is that some suppressors are gain-
of–function mutations (Sopko et al. 2006, van Leeuwen et al. 2016), which may
not be identifiable in a screen focused on gene knockouts.
Genome-wide screens involving overexpression or silencing: Dosage
suppression occurs when a mutant phenotype is rescued by overexpression of
another gene (Magtanong et al. 2011) (Fig. A3c). High-copy plasmid libraries
have been used to successfully identify dosage suppressors (Jones et al. 2008,
Magtanong et al. 2011), and similar screens are now possible using
CRISPR/Cas9 activation (CRISPRa) (Dominguez et al. 2016, Konermann et al.
2015, Perez-Pinera et al. 2013). Much like overexpression screens, genes that
act as suppressors when they are downregulated can also be identified using
approaches that instead repress transcription at interacting genes using RNA
interference (RNAi) (Lehner et al. 2006, Tischler et al. 2006) or CRISPR/Cas9
interference (CRISPRi) (Gilbert et al. 2013, Larson et al. 2013, Qi et al. 2013,
Tanenbaum et al. 2014).
Figure A3. Techniques for identifying intergenic suppressors.
Multiple types of genetic screens have been utilized to identify
138
suppressors in lab experiments. A: New mutations generated either
through mutagenesis or spontaneous mutation can result in suppressors
(red). This approach usually results in the identification of multiple
mutations, from which the actual suppressor must be distinguished. B:
Genome-wide suppressor screens can be accomplished by systematically
knocking out each gene in a mutant background. This can be
accomplished by targeted gene deletion or recombination with a collection
of knockout strains. C: Dosage suppression interactions can be screened
using different strategies for altering the regulation of each gene in the
genome using either plasmid-based overexpression, RNAi, or
CRISPRa/CRISPRi.
In the next section, we discuss some of the general insights that have been
gained from studies of genetic suppression to date.
A.3.2 Functional mechanisms that cause genetic suppression
Past studies, including recent work using the approaches mentioned above, have
described a number of mechanisms that can lead to genetic suppression (J.
2005, Jones et al. 2008, Prelich 1999, van Leeuwen et al. 2016). These
mechanisms include:
Reverting a mutated amino acid: Intragenic suppressors in coding regions may
change a mutated codon so that it specifies an amino acid that is structurally or
biochemically similar to the one that was initially present (Prelich 1999).
Intergenic suppressors in tRNAs can have a similar effect by changing the
codons that specify a particular amino acid (D. 2015).
Restoring the structural conformation or dosage of an mRNA, protein, protein
complex, or cellular component: Intragenic suppressors can impact transcript or
protein stability, thereby allowing a mutated gene to function at closer to wild type
levels (J. 2005, Prelich 1999). Intergenic suppressors might affect the physical
139
interaction between proteins by altering sites of protein-protein interaction
(Sandrock et al. 1997), increasing the levels of available binding partners
(Magtanong et al. 2011), or even enabling a protein complex or cellular
component to function in the absence of the originally mutated gene’s protein
product (Liu et al. 2015).
Changing the dosage of a mutated gene’s cognate mRNA or protein: Both
intragenic and intergenic suppressors may directly alter the levels of a mutated
gene’s cognate mRNA or protein (J. 2005, Magtanong et al. 2011, Prelich 1999,
van Leeuwen et al. 2016). Changing dosage in this way may compensate for the
reduced activity of a gene product due to the destabilizing effect of an initial
mutation. Intragenic suppressors that act in this way may occur in cis regulatory
elements, whereas intergenic suppressors could occur in transcription factors
and their regulators (D. 2015).
Modified activity within a pathway: An intergenic suppressor may occur in a gene
that is in the same pathway as the original mutation, thereby restoring wild type
activity levels within that pathway (van Leeuwen et al. 2016). For instance, one
gene may activate downstream targets of the pathway, while the other represses
these targets. A mutation in the activator that disrupts the balance between these
regulators could inactivate the pathway. Loss-of-function in the corresponding
repressor or gain-of-function in a parallel activator in the same pathway could
then suppress the effect of the first mutation and restore the pathway’s function.
Changes in activity between pathways: Intergenic suppressors can also occur in
pathways that perform functions related to that of the pathway containing the
original mutation, which may be able to functionally compensate for altered
activity in the initial mutant. However, research suggests this form of genetic
suppression is less prevalent than some of the other mechanisms described
above (Baryshnikova et al. 2010, Dixon et al. 2009, van Leeuwen et al. 2016).
This could in part be due to the negative effects of network rewiring, as the gain-
140
of-function mutation in the suppressor allele may disrupt the original function of
the gene harboring the suppressor, resulting in a negative fitness effect.
Global changes in transcription, translation, or other cellular processes:
Suppressors may act in a more general manner by influencing overall levels of
transcription and translation in cells (van Leeuwen et al. 2016). For example, a
mutation that reduces the expression of a gene might be suppressed by another
mutation that decreases protein degradation.
As described in this section, lab experiments have been used to
comprehensively determine the mechanisms that can give rise to genetic
suppression. Information from these experiments is a valuable research tool for
considering how genetic suppression might occur in other contexts, such as in
natural populations.
A.3.3 Examples of genetic suppression in natural populations
Although genetic suppression has historically been a focal point for researchers
interested in dissecting pathways and genetic networks (van Leeuwen et al.
2016), it is now becoming increasingly important to scientists who study heritable
phenotypic variation within natural populations. Multiple examples of suppression
have been identified in different species, particularly yeast (Dowell et al. 2010,
Hou et al. 2015) and humans (Chen et al. 2016). Such naturally occurring
suppression might play an important role in evolution and disease.
One of the most striking examples of naturally occurring suppression comes from
genes that are essential in only certain individuals within a species. Essential
genes encode fundamental cellular functions that are required for viability.
However, which genes are essential varies from individual to individual because
of differences in their genetic backgrounds. To demonstrate this point, Dowell
knocked out nearly all of the roughly 5,000 genes in two strains of the budding
yeast Saccharomyces cerevisiae (Dowell et al. 2010). By doing this, they found
141
57 genes that were essential in only one strain or the other. This represents
nearly 6% of the genes that were essential in either of the strains.
In another study focused on yeast, Hou found that naturally occurring
suppressors can have a large effect on the viability of segregants in S. cerevisiae
crosses (Hou et al. 2015). Specifically, they identified a genetic variant in a
tyrosine tRNA that allows read-through of TGA stop codons. This tRNA variant
suppresses a nonsense allele in a mitochondrial cytochrome c-oxidase (COX15),
thereby reverting the respiratory defect shown by individuals with the COX15
nonsense allele.
Large-scale whole genome resequencing studies in humans have found similar
results to the work in yeast. For example, an analysis of ~590,000 genomes
identified 13 adults who were healthy even though they possessed disease-
associated genotypes at fully penetrant early onset Mendelian disease loci (Chen
et al. 2016). In another study, sequencing of exomes from 3,222 highly related
individuals identified 1,111 homozygous variants with predicted loss of function in
781 genes (Narasimhan et al. 2016). Despite the fact that some of the variants
had previously been associated with diseases, no significant correlation was
observed between an individual’s genotype and health record (Narasimhan et al.
2016). Resequencing studies have shown that other species also large amounts
of rare loss-of-function variation in phenotypically important genes (e.g.,
(Bergstrom et al. 2014, Genomes Consortium. Electronic address and Genomes
2016, Liti et al. 2009)).
Additionally, Jordan used comparative genomics to discover intragenic
suppressors of human disease mutations that were present in other species
(Jordan et al. 2015). They first characterized the extent to which disease alleles
have fixed in 100 non-human vertebrates. Up to 12% of the queried variants
showed fixation in at least one non-human genome, suggesting that other
genetic differences in these outgroups ameliorated the effects of the disease
142
alleles. Examination of orthologous amino acid sequences using a computational
model facilitated the identification of potential intragenic suppressors, several of
which were experimentally validated.
These findings in yeast, humans, and other species reflect a broader reality that
genetic background often plays a strong role in influencing how large effect
mutations impact phenotype (Chandler et al. 2013, Nadeau 2001). In some
cases, alleles that appear to have Mendelian effects in some genetic
backgrounds can show more complicated phenotypic impacts when other
backgrounds are considered (Hou et al. 2016). Also, some mutations may affect
viability in a more probabilistic manner, with different genotypes varying in their
propensities to survive a particular genetic perturbation. For example, Paaby
found extensive genetic variation in embryonic lethality among Caenorhabditis
elegans strains by using RNAi knockdown of important developmental regulators
(Paaby et al. 2015).
Following upon the results in this section, as well as similar studies that were not
discussed, it is important to determine the genetic and molecular mechanisms
that cause individuals to show different responses to the same large effect
mutations.
A.3.4 The genetic and molecular basis of naturally occurring genetic suppression
Characterizing the mechanisms that underlie naturally occurring suppression can
provide novel insights into how genetic variation within populations can modify
the responses of individuals to new mutations. As a starting point for considering
this problem, one must appreciate that populations often harbor large amounts of
genetic variation, which can rewire the pathways and networks that give rise to
phenotype (Ciliberti et al. 2007, Felix and Wagner 2008, Masel and Siegal 2009
2014, Matsui, 2015, Taylor and Ehrenreich 2015, Wagner 2005, Wagner 2012).
These changes can cause individuals to differ in the genes they require to be
healthy or viable, or to express a given trait.
143
As with most traits that segregate within species, suppression in natural
populations may have a complex genetic basis that involves multiple variants
(Fig. A1). Supporting this view, crossing experiments in budding yeast indicate
that most conditional essentialities are mediated by two or more genetic variants
(Dowell et al. 2010, Hou and Schacherer 2017). More generally, work on other
cases in which a mutation shows different effects across genetic backgrounds
has shown that response to mutations can be mediated by higher-order sets of
variants that interact not only with a mutation, but also with each other (Chandler
et al. 2014, Chari and Dworkin 2013, Lee et al. 2016, Taylor and Ehrenreich
2014, Taylor and Ehrenreich 2015, Taylor et al. 2016) and potentially even the
environment (Lee et al. 2016). However, only a small number of examples of this
phenomenon have been comprehensively teased apart at the genetic level
(Chandler et al. 2013), leaving open the possibility that other genetic
architectures, such as those in which a mutation acts as a hub of pairwise
genetic interactions with many different variants (Forsberg et al. 2017, Schell et
al. 2016), could also be important.
Moving forward, it is necessary to better determine the genetic and molecular
basis of suppression in natural populations. Such work could be difficult in
humans, but is feasible in model systems that facilitate comprehensive genetic
dissection of complex traits (Bloom et al. 2013, Ehrenreich et al. 2010). Research
along these lines can answer questions about the number and molecular
functions of involved genes and genetic variants, and can clarify the relationship
between the functional mechanisms of suppression in lab experiments and
natural populations.
144
Figure A4. Combinations of genetic variants may cause suppression
in natural populations. A: A signaling network may regulate the
expression of a transcript that is required for viability. B: Knockdown of a
key component of the network (red) can result in loss of a required
transcript. C: As is typically observed in lab experiments, a large effect
mutation (blue) elsewhere in the pathway can suppress the deleterious
effects of the initial mutation and restore activity. D: In natural populations,
multiple variants with small functional effects within the pathway (light blue)
can collectively result in suppression if their combined effect on network
activity is sufficient to achieve levels of pathway output required for
viability.
Given the genetic complexity that can underlie suppression in natural populations,
intergenic suppressors may be more likely to contribute in nature than intragenic
suppressors because of their significantly larger target space for accumulating
genetic variation. Beyond this distinction, categorizing the functional mechanisms
that are responsible may not be as straightforward as in lab experiments.
145
Naturally occurring suppression might involve a number of variants with small
effects on molecular function, which may individually alter the structures of
mRNAs and peptides, enzymatic activities, or transcript and protein levels in
subtle ways (e.g., (Taylor et al. 2016)). Some of these variants may influence
individual genes, whereas others may have more global effects, in some cases
influencing the expression of a large fraction of the genome. Combinations of
these variants could then collectively achieve a similar functional effect to the
individual suppressors typically seen in lab experiments (Fig. A4).
Characterizing the genetic and functional mechanisms underlying naturally
occurring suppression will provide valuable insights into how combinations of
variants can alter the susceptibility of biological systems to genetic perturbations.
This problem has a fundamental bearing on our understanding of the ways in
which new mutations and pre-existing genetic variation jointly determine the
relationship between genotype and phenotype.
A.4 Conclusion and outlook
Lab experiments enabled by high-throughput genetic and genomic approaches
have helped provide detailed insights into the distinct functional mechanisms that
give rise to genetic suppression. Given that suppression appears to segregate
within species, determining how findings from the lab relate to natural
populations is important. Because of the large amount of genetic variation in
these populations, naturally occurring suppression may in some cases be more
complex at the genetic and molecular levels than suppression studied in the lab.
Work that is able to tease apart this complexity may provide new insights into
how genetic variation within species alter the susceptibilities of individuals to
large effect mutations.
Abstract (if available)
Abstract
Cryptic variation is genetic variation that does not typically influence phenotype but can cause unexpected trait outcomes when genetic and environmental perturbations are introduced into a system. Despite its influence on the heritability of many evolutionary and medically significant traits, the role of cryptic variation is often inferred rather than explicitly defined. This largely stems from their effects being highly conditional, making cryptic genetic variants difficult to identify and study. ❧ In this dissertation, I define the complex genetic and molecular mechanisms of cryptic variation in a yeast colony morphology trait. In chapter two, I show that sets of cryptic variants in a conserved signaling pathway regulate the trait’s expression in an environment-dependent manner. In chapter three, I demonstrate that multiple genetic perturbations can cause the trait through interaction with different combinations of cryptic variants in the same pathway. In chapter four, I describe how regulatory rewiring unmasks cryptic variation in alternative signaling pathways and modifies the quantitative genetic architecture underlying the trait.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Genetic architectures of phenotypic capacitance
PDF
Exploring the genetic basis of quantitative traits
PDF
The complex genetic and molecular basis of oxidative stress tolerance
PDF
Genetic and molecular insights into the genotype-phenotype relationship
PDF
Evolutionary mechanisms responsible for genetic and phenotypic variation
PDF
Understanding the genetic architecture of complex traits
PDF
Understanding the genetics, evolutionary history, and biomechanics of the mammalian penis bone
PDF
Genome-scale insights into the underlying genetics of background effects
PDF
Robustness and stochasticity in Drosophila development
PDF
Studies in bivalve aquaculture: metallotoxicity, microbiome manipulations, and genomics & breeding programs with a focus on mutation rate
PDF
The evolution of gene regulatory networks
PDF
Evolutionary genomic analysis in heterogeneous populations of non-model and model organisms
PDF
Genetic architecture underlying variation in different traits in the Pacific oyster Crassostrea gigas
PDF
Investigating the potential roles of three mammalian traits in female reproductive investment
PDF
Genome sequencing and transcriptome analysis of the phenotypically plastic spadefoot toads
PDF
The molecular mechanism underlying the autoimmune-associated PTPN22 R620W variation and the quest for therapeutics
PDF
Cellular level bottlenecks: genetic diversity, population dynamics, and technology development
PDF
Application of genetic association methods in mice to understand phenotypes with a complex etiology
PDF
Understanding genetics of traits critical to the domestication of crops using Mixed Linear Models
PDF
From gamete to genome: evolutionary consequences of sexual conflict in house mice
Asset Metadata
Creator
Lee, Jonathan Tak
(author)
Core Title
Complex mechanisms of cryptic genetic variation
School
College of Letters, Arts and Sciences
Degree
Doctor of Philosophy
Degree Program
Molecular Biology
Publication Date
02/13/2019
Defense Date
12/18/2018
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
cell signaling,CRISPR,cryptic variation,genetic engineering,genetic mapping,genetics,OAI-PMH Harvest,quantitative trait,yeast
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Ehrenreich, Ian (
committee chair
), Dean, Matthew (
committee member
), Gracey, Andrew (
committee member
), Nuzhdin, Sergey (
committee member
)
Creator Email
jonathantaklee@gmail.com,lee212@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c89-118339
Unique identifier
UC11675618
Identifier
etd-LeeJonatha-7057.pdf (filename),usctheses-c89-118339 (legacy record id)
Legacy Identifier
etd-LeeJonatha-7057.pdf
Dmrecord
118339
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Lee, Jonathan Tak
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
cell signaling
CRISPR
cryptic variation
genetic engineering
genetic mapping
genetics
quantitative trait
yeast