NON-RANDOM MUTATIONS IN YEAST

NON-RANDOM MUTATIONS IN YEAST:
AN EXAMINATION OF YEAST GENETICS DURING SELECTION CYCLING
by
CHRISTOPHER MCATEE
ERIC J. SORSCHER, COMMITTEE CHAIR
JOHN L. HARTMAN
STEVE A. WATTS
AUBREY E. HILL
A THESIS
Submitted to the graduate faculty of The University of Alabama at Birmingham,
in partial fulfillment of the requirements for the degree of
Master of Science
BIRMINGHAM, ALABAMA
2013
Copyright by
CHRISTOPHER MCATEE
2013
NON-RANDOM MUTATIONS IN YEAST:
AN EXAMINATION OF YEAST GENETICS DURING SELECTION CYCLING
CHRISTOPHER MCATEE
BIOLOGY
ABSTRACT
The appearance of genomic mutation is generally considered a random process,
during which single nucleotide polymorphisms (SNPs) and other DNA alterations are
selected and fixed by virtue of beneficial effects on fitness. Recent studies have provided
examples of induced mutations in response to stress, including the suggestion of biased
SNP formation that favors certain adaptations. This project analyzed the appearance and
accumulation of mutations in S. cerevisiae following exposure to two lethal forms of
selective pressure. A mutation/reversion cycle was established in which 5-fluoroorotic
acid (5-FoA) served as an environmental stress to identify the functional loss of URA3 (a
gene required to decarboxylate 5-FoA and necessary for 5-FoA cellular toxicity),
followed by –URA growth (i.e. absence of uracil, which is lethal if URA3 function has
been disabled) to detect restoration of URA3. During selection cycling, we observed a
tendency towards increased drug resistance and appearance of yeast strains that could
survive under these two otherwise lethal selective pressures. Such adaptation might be
genetic or epigenetic, and might involve specific gene defects or epistatic interactions
(i.e. among many distinct genes). A full genomic analysis revealed a logical pattern of
mutations that help explain the phenotypes observed. In addition, when complete DNA
sequencing of yeast tested here was compared to the published reference genome, we
noted a mutation bias in the founder sample, suggesting the action of a strong selective
pressure and possibly including a preference in the way SNPs are originally generated in
iii
yeast. One mechanism that might contribute to strong synonymous SNP bias likely
involves a preponderance of transition polymorphisms, which would affect SNP
formation and favor synonymous nucleotide replacement. The information described
here will contribute to understanding mutation accrual in eukaryotic cells, as well as
possible ramifications of biased patterns of mutation in yeast.
Keywords: selection cycling, S. cerevisiae, mutation accrual, transition bias
iv
ACKNOWLEDGEMENTS
First, I would like to thank my committee members, Drs. Eric Sorscher, John
Hartman, Steve Watts, and Aubrey Hill, for all of their advice and guidance throughout
my project. I would also like to thank Jingyu Guo, Zack Plyler, and John Rogers for their
assistance and discussions.
I would like to thank the Heflin Genetics Center for technical support. I would
also like to thank everyone in the UAB Gregory Fleming James Cystic Fibrosis Research
Center for their help.
Finally, I would like to thank my family and friends for all of the support they
have shown me over the years.
v
DEDICATION
I would like to dedicate this manuscript to my parents who have motivated me
throughout my academic training. Without their love and support, none of this would
have been possible.
vi
TABLE OF CONTENTS
Page
ABSTRACT...................................................................................................................... iii
ACKNOWLEDGEMENTS ................................................................................................v
DEDICATION .................................................................................................................. vi
LIST OF TABLES ............................................................................................................ ix
LIST OF FIGURES ............................................................................................................x
INTRODUCTION ..............................................................................................................1
Background and Significance .................................................................................2
Spontaneous Mutation ................................................................................2
Fidelity of DNA Replication in Yeast ........................................................3
Non-Random SNP Accrual in Yeast...........................................................4
Mutation Rates and Genomic Evolution .....................................................4
URA3 Mutation Model ...............................................................................5
Overview of the Present Study ...............................................................................6
METHODS .........................................................................................................................8
Media and Reagents ................................................................................................8
Screening and Selection of Strains .........................................................................8
Mutation and Reversion Cycle.............................................................................. 10
Statistics ................................................................................................................ 12
RESULTS ......................................................................................................................... 13
Impact of Selection Cycling on Survival .............................................................. 13
Modeling Yeast Mutation Rates during Selection Cycling
as a Test of Directed Evolution................................................................. 15
Genomic Diversity following YPD Expansion......................................... 16
Genomic DNA Analysis ....................................................................................... 20
Comparison of Yeast Strain Sequence with Reference Genome .............. 23
DISCUSSION ................................................................................................................... 26
vii
SNP Formation following Exposure to Lethal Environmental Stress .................. 26
Adaptive SNP Formation ...................................................................................... 27
Global SNP Patterns Observed in this Study ........................................................ 29
Considerations Regarding the “Dual Survivor” Phenotype .................................. 32
SNP Accumulation in the Setting of Recurrent Environmental Stress ................. 33
CONCLUSION ................................................................................................................. 35
LIST OF REFERENCES .................................................................................................. 37
APPENDIX
A
SUPPLEMENTAL TABLES ............................................................................... 42
B
SUPPLEMENTAL FIGURES .............................................................................. 44
viii
LIST OF TABLES
Table
Page
1
Relationship between final mutation frequency and the generation
number at which mutation occurs during YPD growth .............................................. 16
2
Mutations in strain 2B8 observed during three cycles of selection ............................ 22
3
Mutations in strain 1B8 observed during three cycles of selection
to reach “dual survivor” phenotype ............................................................................ 22
4
SNP frequency for full genome data compared to S. cerevisiae
reference genome ........................................................................................................ 24
5
Exonic SNP frequency for full genome data compared to
S. cerevisiae reference genome ................................................................................... 25
ix
LIST OF FIGURES
Figure
Page
1
Initial screening protocol and generation of experimental strains .............................. 10
2
S. cerevisiae mutation and reversion cycle ................................................................. 11
3
Colony numbers following multiple rounds of selection cycling ............................... 14
4
Selection of yeast surviving both 5-FoA treatment and –URA .................................. 15
5
Frequency for which a single base substitution would be
expected in a final YPD culture of ~10 x 109 yeast; compiled
from 500K simulations ............................................................................................... 18
6
Model of mutation accumulation under varied mutation rates ................................... 28
x
INTRODUCTION
The appearance of genomic mutation is generally considered a random process,
during which single nucleotide polymorphisms (SNPs) and other DNA alterations are
selected, and allele frequency is determined by virtue of variable effects on fitness (or
other features such as genetic drift). Recent studies have provided examples of induced
mutations in response to stress, including the suggestion of directional adaptation, in
which SNPs are formed in a manner that favors certain advantageous phenotypic. This
project will test the stochastic nature of DNA variation in the context of exposure to
selective cycles involving two lethal stresses (chosen because they are thought to be
mutually exclusive and related to the function of a single gene) and examining the
changes in DNA sequence that result. The Specific Aims are as follows:
Specific Aim 1: Investigate selection cycling with 5-flouroorotic acid (5-FoA) and –URA
in Saccharomyces cerevisiae. We will obtain genetic information at each step of an
iterative selection cycle and determine the mutations in URA3 that are associated with
the adaptive phenotype(s).
Specific Aim 2: Investigate genomic patterns of mutation appearance or accrual during
selection cycling. Complete genomic data will be probed to identify the location and
incidence of new SNPs and calculate mutation frequency following exposure to lethal
environmental stress.
1
The proposed experiments will address a number of topical issues regarding yeast
genomic evolution including 1) the extent to which permanent DNA modification (as
opposed to epigenetic alteration) is responsible for adaptation after an acute and
otherwise lethal perturbation in yeast, 2) preferences in mutation accrual in association
with specific environmental stress, 3) whether repetitive exposure to the same external
challenge confers a measure of “learned” genomic adaptation, and 4) the degree to which
randomness accounts for new mutations and their rate of accumulation following
exposure to an extremely adverse environment. The specific hypothesis being pursued
by these studies is as follows:
Multiple rounds of selective cycling will induce an adaptive yeast phenotype that
becomes increasingly resistant against further exposures to the same (repeated)
environmental insult. Such an adaptation might be genetic or epigenetic, and might
involve specific gene defects or epistatic interactions (i.e. among many distinct genes).
Background and Significance
Spontaneous Mutation
Mutations in genomic DNA are traditionally viewed as appearing in a random
fashion, with subsequent natural selection allowing for variation and evolution of novel
phenotypes. A conventional perspective has been that sites of mutation occur
independently of environmental stress (1), although overall mutation rates can be strongly
influenced by certain extrinsic circumstances, and several reports have suggested possible
directionality to patterns of mutation accrual and genomic evolution (2-11). In
2
prokaryotes, for example, the rates and positions of stress-induced mutation are
increasingly viewed as non-random, with SNP patterns that occur in a significantly
adaptive fashion (2). In a classic study of mutation reversion in yeast, cells were
subjected to a selective pressure for several days, and new revertant colonies continued to
arise long after the stress exposure had ended. The findings were ascribed to nonrandom,
adaptive (and directional) reversion of a mutated allele (6). Non-Mendelian inheritance
of genetic information has also been noted in Arabidopsis, where progeny from
homozygous recessive parents exhibited a dominant, fitness-enhancing allele at high
frequency (7). It should also be noted that, among both eukaryotes and prokaryotes,
stressful environments often confer higher rates of mutation through effects on the
fidelity of DNA polymerase that may otherwise be mistaken for directional SNP
formation (3-4). Hypermutable strains of Pseudomonas aeruginosa, for example, are
often observed in the lungs of cystic fibrosis patients and contribute to bacterial survival
under very severe extrinsic pressure (8).
Fidelity of DNA Replication in Yeast
Mutations in eukaryotic genomes are typically caused by errors during DNA
replication, such as erroneous base insertion or faulty proofreading of newly synthesized
DNA associated with mismatch repair. In Saccharomyces cerevisiae, POL3 DNA
polymerase δ (12) recognizes errors in replication of new DNA strands and makes
corrections. Several studies have demonstrated that mutated POL3 polymerase δ is
correlated with higher rates of mutation (12-14). Another polymerase involved in
proofreading, polymerase ε, has also been implicated in hypermutable strains (13).
3
Additional genes critical to mismatch repair include PMS1, MSH2, and MLH1,
mutations of which lead to higher rates of genomic SNP formation when compared with
non-mutants of the same genetic background (12, 13).
Non-Random SNP Accrual in Yeast
Although mutations typically are viewed as occurring in stochastic fashion, bias
in the location and surrounding context of mutations is well recognized. Mutational ‘hot
spots,’ regions with an increased probability of DNA polymorphism, are commonly
observed in eukaryotes. The mechanisms underlying mutational predisposition of this
kind are not well understood, but studies of mammalian DNA indicate that surrounding
genomic context may confer single nucleotide polymorphism at specific loci (15). In
addition, transition mutations (cytosine/thymine or adenine/guanine) appear much more
frequently in most genomic settings compared with their transversional counterparts.
Transitions have been highly correlated with CpG islands – areas prone to methylation –
a finding compatible with the observation that cytosine methylation and aberrant
deamination contribute strongly to SNP formation (16). Although such methylation
patterns may not occur as frequently in yeast as in higher organisms (17), a robust
transition bias is evident in Saccharomyces cerevisiae (18).
Mutation Rates and Genomic Evolution
Mutation rate reflects the probability that a specific base will change over a
discrete time interval or during a specific number of DNA replications. Several studies
have estimated the mutation frequency of various species under defined conditions,
4
including both prokaryotic and eukaryotic genomes. Earlier reports employed colony
counts of polymorphic genomes from bacteria (1) or yeast (19-20) in the setting of a
selective perturbation to measure mutation rate. For example, a previous study of S.
cerevisiae analyzed diploid cell divisions and included an experimental group prone to
more frequent mutation following exposure to a chemical mutagen. Yeast with defective
mismatch repair were investigated by the same protocol. An estimated mutation rate of
1.1 x 10-3 mutations per diploid cell division, or approximately 9.2 x 10-11 mutations per
nucleotide position per cell division, was reported based on the appearance of colonies
resistant to an otherwise lethal environmental stress (19).
High volume DNA sequencing techniques provide a new methodology for
mutation rate analysis (18, 21). In one study, yeast strains were propagated in a fashion
independent of the incidence of lethal nucleotide replacement or reversion in order to
detect all SNPs, regardless of effects on fitness. The experimentally measured mutation
rate (3.3 x 10-10 per base pair per cell division; 18) was viewed as nonbiased since, in
contrast to studies that monitor phenotypically important mutations, the approach also
identified neutral base substitutions appearing in subsequent generations. These
estimates of yeast mutation rate are in good agreement with earlier reports, differing by
only approximately threefold (19).
URA3 Mutation Model
A well characterized URA3 mutation model was chosen for the present
experiments. URA3 is crucial to the pathway of uracil synthesis (22), and defects in this
gene are lethal unless uracil is provided in growth media. 5-fluorootic acid (5-FoA) is a
5
compound converted to the cellular toxin 5-fluorouracil (5-FU) by URA3. 5-FU replaces
uracil during RNA synthesis, causing disruption of cellular machinery and cell death
(23). Therefore, the combination of 5-FoA with uracil in growth media allows for
selection of yeast that have mutated to a nonfunctional URA3 (24). In order to study
phenotypic reversion, 5-FoA resistant cells were grown in media lacking both uracil and
5-FoA. Only URA3 corrective revertants would be expected to survive nutrient
restriction; i.e. yeast that recover the ability to generate uracil via the URA3 pathway can
be selected in this manner.
Overview of the Present Study
This project was intended to develop a novel protocol for environmental selection
cycling and to evaluate mutation patterns of S. cerevisiae under transient, alternating
selective pressures. In particular, we used the selection cycle methodology to address the
following questions: 1) Does the mutation rate for yeast under non-stressed conditions, as
detected by an imposed selective pressure, conform to what has been published
previously?, 2) Will a specific protective locus (and/or the entire genome) become
hypermutable in the setting of repetitive cycles of a lethal stress (in this case, the locus
being URA3)?, 3) Do observed mutation frequencies exhibit patterns in any sense, or are
they strictly random? Do mutations appear to be biased in a fashion that is functionally
responsive to a specific environmental insult?, and 4) How does mutation accrual in the
setting of a strong selective pressure impact overall fitness, i.e. can yeast that must
‘mutate or die’ prevent rampant accumulation of SNPs throughout the genome and
maintain their genomic integrity despite the ongoing accumulation of new DNA
6
polymorphisms (the so-called ‘ratchet-type’ genomic attrition described historically by
Mueller and colleagues (25))? Accumulation of SNPs in this context was considered
from the perspective of mutation frequency and yeast growth rate in laboratory culture.
7
METHODS
Media and Reagents
YPD liquid media, 5-FoA media plates, –URA plates, and DNA extraction were
prepared and utilized as described by Burke et al (26). Primers for PCR amplification (to
define URA3) were 5’ TTGGTATATATACGCATATGTGGTGTT 3’ and 5’ GACCGA
GATTCCCGGG 3’. PCR was conducted with an initializing step at 94°C for 120 sec,
thirty-five cycles of denaturing at 94°C for 20 sec, annealing at 58.5°C for 20 sec, and
elongation at 72°C for 80 sec. The final elongation step was performed at 72°C for 300
sec. PCR utilized the ExoSAP purification kit according to manufacturer instructions
(USB Corporation). DNA for complete genome sequencing was extracted using the ZR
Fungal/Bacterial DNA MicroPrep™ according to manufacturer’s protocol (Zymo
Research) and provided at a concentration of 500 ng/μl to the Heflin Genetics Center for
analysis on an Illumina HiSeq2000. A total of 11 yeast isolates (2 “dual survivor” strains
in the initial screen and 2 strains that underwent a total of 9 individual selection steps in
the cycling protocol) were subjected to full genomic analysis.
Screening and Selection of Strains
Saccharomyces cerevisiae encoding a functional URA3 gene (designated here as
“founder strain”) was grown in 5 ml YPD liquid media for two days at 30° C, after which
a sample was taken for DNA extraction. In parallel, six hundred μL of liquid culture was
8
combined with 200 μL of 80% glycerol to establish a glycerol stock and stored at –80° C.
Four samples of the remaining liquid culture (0.5 mL each) were applied separately to
four media plates containing 5-FoA and uracil. In order to ensure that yeast exposed to
5-FoA were suitable for reversion, 96 random colonies from the initial 5-FoA plates were
selected and purified, and each colony placed individually in 200 μl YPD liquid media of
a 96 well plate. After two days of growth, 5-FoA resistant strains were stamped onto
−URA plates and incubated for seven days at 30° C. Of the 96 original strains, four
cultures included yeast capable of growth in −URA media. A dilution of the liquid
culture for each of these strains (on 5-FoA) was used to inoculate 4 separate tubes of 5
mL YPD liquid media. After two days of growth, DNA was extracted and a glycerol
stock established for each 5-FoA resistant strain that exhibited subsequent growth on –
URA media. To collect adequate data, 10 replicates of each strain (totaling 40 unique
cultures) underwent mutation and reversion cycling until either a 1) failure to survive or
2) survival in both 5-FoA and –URA media was observed (Figure 1).
9
Figure 1: Initial screening protocol and generation of experimental strains. 1) Selection
of 96 random 5-FoA resistant colonies, 2) Growth in YPD media, 3) Testing for survival
in –URA media†, 4) Glycerol stocks established for the 4 5-FoA resistant strains that
survived –URA stress, 5) 10 separate cultures established for each strain (40 cultures
total), and 6) Growth on –URA media followed by additional rounds of selection as in
Figure 2.
Mutation and Reversion Cycle
To begin the mutation and reversion cycle, each of 4 5-FoA resistant glycerol
stocks described here were used to inoculate 10 separate 5 mL YPD liquid media (Step 5
on Figure 1). After two days of aerated growth at 30° C, liquid cultures (0.5 mL) were
spread onto −URA media to obtain revertant colonies which, after seven days of growth
at 30° C, were counted (Figure 2). Single colonies from the –URA plate were chosen at
random, purified, and used to inoculate 5 mL YPD liquid media. The liquid cultures
†
Two “dual survivors” were also identified during this step as described in the body of
this thesis.
10
were grown at 30° C for 2 days, followed by glycerol preservation and DNA extraction.
As described above, 0.5 mL samples of liquid culture were spread on fresh 5-FoA plus
uracil media plates, followed by seven days growth at 30° C. Surviving colonies were
counted (Figure 2).
Figure 2: S. cerevisiae mutation and reversion cycle. Blue circles represent media type,
green boxes depict protocol steps, and red boxes indicate relevant yeast genomic changes
during growth in YPD.
To complete each selection cycle, single colonies from 5-FoA plates were chosen
at random, purified, and used to inoculate 5 mL of YPD liquid media. The liquid cultures
were grown at 30° C for 2 days, followed by glycerol preservation and DNA extraction.
Next, 0.5 mL samples of liquid culture were spread on fresh –URA media, followed by
11
seven days of growth at 30° C. Surviving colonies were counted, and iterative selection
cycles repeated as depicted in Figure 2.
Statistics
Student’s t test was applied to analysis of colony count data. Observed SNP
counts and ratios were compared to expected values using chi-square goodness of fit.
DNA sequences were aligned using the ClustalW2 multiple sequence alignment program
(http://www.ebi.ac.uk/Tools/clustalw2/index.html).
12
RESULTS
A mutation/reversion cycle was established in which 5-fluoroorotic acid (5-FoA)
served as an environmental stress to identify the functional loss of URA3 (a gene
required to decarboxylate 5-FoA and necessary for 5-FoA cellular toxicity), followed by
–URA growth (i.e. absence of uracil, which is lethal if URA3 function has been disabled)
to detect restoration of URA3. Yeast were expanded in standard YPD media (an
“unstressed” environment) prior to imposing selective pressure, and colonies surviving
the selective environment were quantified following each bottleneck. Glycerol stocks of
resulting strains were generated. The URA3 gene was amplified using PCR, product size
confirmed by gel electrophoresis, and sequenced using URA3 specific primers.
Impact of Selection Cycling on Survival
Colony counts obtained from increasing numbers of cycles of –URA stress
trended significantly lower compared to the first round in –URA media, whereas survival
frequency in 5-FoA increased in early rounds (before decreasing in later rounds, Figure
3). Overall, the majority of yeast strains (29 of 40) ended the mutation/reversion protocol
with a failure to survive on one of the selective media, signifying failure to adapt against
the imposed extrinsic stress. The remaining 11 S. cerevisiae cultures exhibited an
unexpected adaption in which cells became both 5-FoA resistant and uracil prototrophic
(termed a “dual survivor” phenotype) as a consequence of selection cycling. Yeast with
13
this phenotype exhibited a propensity to secrete uracil into the medium which promoted
survival of neighboring cells lacking URA3 (Figure 4). The mechanism underlying this
phenotype is considered below.
Figure 3: Colony numbers following multiple rounds of selection cycling. Asterisks
denote p < 0.05 as compared to Round 1. One-tenth of the 5 mL YPD growth media was
studied at each time point for each of the two selective pressures. Error bars = SEM.
“Dual survivors” were noted in Rounds 2-4, 6, and 7 of –URA media and Round 3 of 5FoA media. N = number of surviving strains, each of which produced colonies that were
counted on selective media. † = Average of 10 separate experiments that generated 5FoA resistant colonies from the founder strain.
14
Figure 4: Selection of yeast surviving both 5-FoA treatment and –URA. Two
experimentally derived strains shown above were tested for “dual survivor” phenotype
and ability to secrete uracil to neighboring (URA3 deficient) cells. (A) survival on both
5-FoA and –URA media, and evidence for supporting proliferation of a surrounding yeast
lawn on –URA plates. (B) survival only in 5-FoA media, with single colony growth on
–URA, but no surrounding yeast lawn in cells deficient for URA3.
Modeling Yeast Mutation Rates during Selection Cycling as a Test of Directed Evolution
One obstacle to estimating expected numbers of colonies resulting from protective
mutations in URA3 involves timing of SNP appearance during YPD growth. For
example, if a protective mutation appears during a very early generation, the change will
be propagated and extensively represented in the final yeast culture. On the other hand,
protective mutations that occur in the final round of cell division may be represented by
only one resistant colony on selective media (Table 1). The likelihood of a protective
15
mutation is also dependent on the number of cell divisions characterizing a particular
generation.
Total
Generation during which
Yeast Cells
Mutation Occurs
1.67E+05
1
3.34E+05
2
6.69E+05
3
1.34E+06
4
2.67E+06
5
5.35E+06
6
1.07E+07
7
2.14E+07
8
4.28E+07
9
8.56E+07
10
1.71E+08
11
3.42E+08
12
6.85E+08
13
1.37E+09
14
2.74E+09
15
5.48E+09
16
1.10E+10
Number of Cells with
Mutation in Final Culture
1.67E+05
32768
3.34E+05
16384
6.69E+05
8192
1.34E+06
4096
2.67E+06
2048
5.35E+06
1024
1.07E+07
512
2.14E+07
256
4.28E+07
128
8.56E+07
64
1.71E+08
32
3.42E+08
16
6.85E+08
8
1.37E+09
4
2.74E+09
2
5.48E+09
1
NG
Table 1: Relationship between final mutation frequency and the generation number at
which mutation occurs during YPD growth. NG is the number of cell divisions occuring
in each generation shown and corresponds to the number of yeast in liquid culture. The
initial inoculum (1.67 x 105± 4.88 x 104 yeast cells) and the final population size in
culture at the conclusion of YPD expansion (1.35 x 1010 ± 2.63 x 109 yeast cells) were
measured experimentally by limiting dilution. The number of generations to achieve a
fully grown YPD culture under conditions used here was estimated as 16. Highlight
indicates 1) starting inoculum and 2) final yeast density that approximates the
experimentally derived value.
Genomic Diversity following YPD Expansion
To evaluate the expected appearance of SNPs during a single round of selection
cycling, we developed methods to simulate occurrence of an individual mutation in a
single yeast strain. First, an estimate of expected mutations per each nucleotide position
was calculated for a fully expanded liquid culture using the reported mutation rate (3.3 x
16
10-10 mutations/ nucleotide/cell division) multiplied by the total number of cell divisions
measured in YPD (~1.4 x 1010), and obtained a value of roughly 4-5 mutations for each
position of the yeast genome. This indicates that every possible SNP that can occur
would be represented approximately once in a final YPD culture containing ~10 x 109
yeast cells (Table 1). Therefore, adaptive mutations and reversions observed in the same
URA3 codon might be expected during repetitive rounds of selective pressure and would
not necessarily represent evidence of directional SNP accumulation.
A random number generator was next used to place a single SNP in one yeast cell
arbitrarily within one of the ~10 x 109 cell divisions shown in Table 1. Computer
simulation of this process was performed during 500,000 individual runs, and the
program tallied the number of computer-generated mutations expected under these
conditions in the final population (~10 x 109 yeast). Because the vast majority of cell
divisions occur towards the end of yeast expansion in this setting, most of the simulation
runs indicated (as expected) that only one cell would carry the mutation (Figure 5).
17
Frequency of Observations
1000000
100000
10000
1000
100
10
1
Number of Cells Enocoding the New Mutation
Figure 5: Frequency for which a single base substitution would be expected in a final
YPD culture of ~10 x 109 yeast; compiled from 500K simulations. A computer program
randomly selected 1 of 16 possible generations for placement of a random SNP. The
number of cells in the final culture (after 16 generations) carrying that mutation was
calculated and tallied as a histogram.
An average number of cells possessing the computer generated mutation was
determined by multiplying the frequency of observations in each generation by the
number of expected (SNP-containing) cells for that generation, totaling these values, and
dividing by the number of runs (500,000). We found that approximately 8 cells
containing a mutation would be expected per run. In other words, if a single mutation
were to occur during any given cell division of YPD yeast expansion shown in Table 1,
we would expect (on average) eight yeast in the final YPD culture to carry this defect. If
the mutation was capable of protecting yeast from 5-FoA, we would expect to observe
eight 5-FoA resistant colonies in the final yeast culture.†
In Figure 3, one tenth volume samples of YPD expansion cultures led to an
average of 65 colonies during the first round of 5-FoA selection cycling. Based on the
†
Note that, although this model does not incorporate SNP effects that result in alterations
of yeast growth rate/fitness, it does provide a rough estimate for mutation accrual in
culture.
18
analysis shown above, between 8 and 9 mutations in URA3 that disrupt function would
account for 65 new colonies (e.g. 8 colonies/mutation x 8 mutations = 64 colonies), or 80
independent SNPs mediating 5-FoA resistance would be expected across an entire
population of 10 x 109 yeast (since only one tenth of each culture was plated). As
described above, a model of random (non-directional) SNP formation indicates that every
possible SNP in the 804 nucleotide open reading frame of URA3 should be represented
within 10 x 109 yeast following YPD expansion, or a total of ~2400 single nucleotide
substitutions. A previous study (20) examined the mutational spectra within URA3
mediating resistance to 5-FoA and found that approximately 10% of all possible SNPs in
the gene conferred sufficient disruption of enzymatic activity to permit survival in lethal
concentrations of the compound. Among 2400 possible URA3 SNPs, therefore, a lower
limit estimate of 10% (or 240) might be expected in a final broth culture of 10 x 109
organisms. Our estimate of ~80 SNPs that disrupt URA3 activity is therefore in
reasonable agreement with published data regarding the plasticity of URA3 (i.e. within a
factor of roughly 3).†
Figure 3 depicts numbers of colonies observed across numerous rounds of
selection cycling. The trend towards increased numbers of 5-FoA resistant strains
(observed in later cycles) could be taken to indicate escalating damage to URA3, itself,
and that diminishing numbers of –URA resistant strains reflect the increasing difficulty of
rescuing a progressively more damaged URA3 (due to multiple destructive SNPs in
URA3 or other genetic determinants). In other words, the increase in 5-FoA resistant
strains observed during later rounds of selection cycling could represent a URA3 protein
†
with a caveat that the fraction of URA3 mutations which disrupt gene function may be
higher than 10%, see below.
19
that becomes increasingly difficult to repair, rather than an adaptive process that
directionally targets URA3 mutations after sequential rounds of 5-FoA selection. Other
interpretations include a more complex genetic or epigenetic explanation, such as
multiple gene defects at loci other than URA3 characterized by epistasis.
Genomic DNA Analysis
We conducted URA3 sequencing for all 40 strains tested in this project (Figure 1),
and full genomic sequence analysis for representative strains undergoing the iterative
selection protocol. For example, strain 1B8 (found to exhibit a “dual survivor”
phenotype after the third round of 5-FoA selection) and strain 2B8 (demonstrating
mutations localized to URA3 after the third round of 5-FoA) were investigated in detail
(Tables 2 and 3). In general, we observed concordance between phenotype and
underlying genotype for yeast isolates evaluated in this fashion. For example, as shown
in Tables 2 and 3, 5-FoA resistance in both strains was clearly attributable to premature
termination codons (PTCs) that appeared in URA3 following the initial round of 5-FoA
stress. Reversion mutations in URA3 were not observed during that specific cycle,
suggesting other genes (or URA3 regulatory elements) were contributing uracil
production. In this setting, complete genomic sequencing was found informative.
Although the total number of mutations observed for individual strains was very small
(19 total new mutations in Tables 2 and 3, using stringent criteria such as a Phred score
≥1000 and sequencing depth >100, genome-wide across multiple rounds of selection
cycling), in a number of instances, our results provide a likely explanation for observed
phenotype. For example, PTCs in URA3 were overcome by altering tRNA anticodons in
20
chromosomes VI, VII, and XI to place the correct amino acid at an aberrant nonsense
position. Another mutation of interest occurred in SEC22, a gene important for
nucleotide metabolism implicated as a mediator of chemotherapy responsiveness (27).
The relationship of SEC22 to tRNA suppression is not known, but can be investigated in
the future using protocols such as those shown here. Complete genomic sequencing in
later cycles of the protocol also demonstrated URA3 point mutations that overcame PTCs
in URA3, including new SNPs in exactly the same codon (Table 3). The full genomic
sequence data also points to URA6 as a possible target underlying the “dual survivor”
phenotype (see Discussion).
From a total of 11 yeast genomes sequenced in their entirety, 26 new SNPs were
observed. This represents a total of 616 generations (16 generations in YPD media +
appx. 40 generations on selection plate x 11 samples)† for 12 million nucleotide positions
per genome, and a measured mutation rate of 2.3 x 10-9 (MSS maximum likelihood
method, 28). The rate estimated here is therefore appx. seven fold higher than the
published mutation rate for yeast studied under minimally selective conditions (18).
†
The estimate incorporates two rounds of yeast colony growth on solid media. Note that
survival under selective growth conditions would likely influence the total number of
mutations observed. Also, mutations occurring later in the growth expansion phase
would be missed due to the nature of Next Gen DNA sequencing technology, which only
detects SNPs present in the majority of DNA samples being evaluated.
21
chrom
chrIV
chrV
chrIV
chrX
chrVI
chrXII
chrVI
chrII
chrXI
chrXVI
chrV
position SGD
369197 G
116590 A
844066 G
589942 G
226712 A
680584 A
226693 A
667758 T
202678 T
928063 G
116345 C
Founder
G
A
G
G
A
A
A
T
T
G
C
Round 1 5-FoA
C
T
A
T
A
A
A
T
T
G
C
Round 1 -URA
C
T
A
T
T
C
A
T
T
G
C
Round 2 5-FoA
C
T
A
T
T
C
T
A
T
G
C
Round 2 -URA
C
T
A
T
T
C
T
A
A
T
C
Round 3 5-FoA
C
T
A
T
T
C
T
A
A
T
A
SNP Effect Gene
Intergenic
STOP GAIN URA3
NON-SYN NUP42
NON-SYN BIR1
tRNA
tK(CUU)F
NON-SYN SEC22
tRNA
tK(CUU)F
NON-SYN PCS60
tRNA
tK(CUU)K
SYN
YPR195C
NON-SYN URA3
Table 2: Mutations in strain 2B8 observed during three cycles of selection. SGD
designates nucleotide in reference genome (Saccharomyces Genome Database). Founder
denotes sequence data of original laboratory strain tested here, with “round” and “media
type” indicating genomic sequence observed during iterative cycles of mutation and
reversion. Color coding indicates the type of SNP effect (see legend).
chrom
chrIV
chrIV
chrX
chrV
chrVII
chrVII
chrVII
chrV
chrX
chrXI
chrXII
chrXVI
position SGD
369197 G
844066 G
589942 G
116590 A
122308 T
122312 C
271803 G
116590 A
50691 C
392641 C
856163 C
134129 A
Founder
G
G
G
A
T
C
G
A
C
C
C
A
Round 1 5-FoA
C
A
T
T
T
C
G
T
C
C
C
A
Round 1 -URA
C
A
T
T
A
C
G
T
C
C
C
A
Round 2 5-FoA
C
A
T
T
A
A
A
T
C
C
C
A
Round 2 -URA
C
A
T
G
A
A
A
G
C
C
C
A
Round 3 5-FoA
C
A
T
G
A
A
A
G
T
G
A
C
SNP Effect Gene
Intergenic
NON-SYN NUP42
NON-SYN BIR1
STOP GAIN URA3
tRNA
tK(CUU)G1
tRNA
tK(CUU)G1
SYN
SCS3
NON-SYN URA3
NON-SYN RCY1
NON-SYN URA6
5' REG
RPS22B
NON-SYN FLC1
Table 3: Mutations in strain 1B8 observed during three cycles of selection to reach “dual
survivor” phenotype. SGD designates nucleotide in reference genome (Saccharomyces
Genome Database). Founder denotes sequence data of original laboratory strain tested
here, with “round” and “media type” indicating genomic sequence observed during
iterative cycles of mutation and reversion. Color coding indicates the type of SNP effect
(see legend).
Color Legend
Intergenic Occurs in Intergenic Region
5' REG
Occurs in 5' Regulatory Region
NON-SYN Non-Synonymous
SYN
Synonymous
STOP GAIN Stop Codon
tRNA
Occurs in tRNA
22
Comparison of Yeast Strain Sequence with Reference Genome
Full genomic sequence from the founder strain was also compared to the S.
cerevisiae reference genome (http://downloads.yeastgenome.org/sequence/S288C_
reference/genome_releases/). The analysis revealed strong bias in the patterns of
mutation accrual (Table 4). For example, the ratio of SNPs occurring in the exonic
versus intergenic DNA compartments was 1.5:1 (a ratio of 3:1 would be expected for
random SNP formation based on the proportion of coding to non-coding DNA in yeast
(~3:1)). In addition, the proportion of non-synonymous to synonymous SNPs was 1:1
(the expected ratio being 3:1 for stochastic SNP formation and tabulation of all possible
single nucleotide changes for all possible codons). Additionally, transition mutations
occurred much more frequently (observed ratio of 1.8:1 transition to transversion, versus
an expected ratio of 1:2 for random SNP formation; i.e. transition changes (T ↔ C or
A ↔ G) would otherwise be expected to occur half as often as transversions (A ↔ T,
A ↔ C, C↔ G, and G ↔ T, Table 4).
23
Table 4: SNP frequency for full genome data compared to S. cerevisiae reference
genome. Asterisks denote p < 0.05. Expected ratio of exonic:intergenic SNPs was
calculated based on total numbers of exonic and intergenic bases in S. cerevisiae.
Expected non-synonymous:synonymous and transition:transversion ratios were
calculated based on all possible single nucleotide replacements in the genetic code.
Exonic SNPs in the above analysis were further categorized based on their
frequencies in essential versus non-essential yeast genes (29). An essential gene is
defined as vital for S. cerevisiae propagation on standard growth media (i.e. gene deletion
leads to inviability). Locations of all SNPs that differed from the reference (and their
frequencies in essential versus non-essential genes) were compared against either the
SNP consequence (non-synonymous versus synonymous) or type (transition versus
transversion). The category of genes in which SNPs were observed (essential or nonessential) was found to be independent of both SNP consequence (p = 0.37) and type (p =
0.98) in this analysis. The non-synonymous:synonymous and transition:transversion
ratio were similar in both gene categories, suggesting that natural selection, by itself, does
24
not fully explain the ratios seen here. The implications of this finding with regard to
adaptiveness and/or directionality of SNP formation are considered further in the next
section.
Non-Synonymous:Synonymous
Essential Non-Essential
Synonymous
108
650
Non-Synonymous
96
661
Ratio
0.89
1.02
Transition:Transversion
Essential Non-Essential
Transition
133
856
Transversion
71
455
Ratio
1.87
1.88
Table 5: Exonic SNP frequency for full genome data compared to S. cerevisiae reference
genome. SNPs were analyzed by gene category (essential or non-essential), SNP
consequence (synonymous or non-synonymous), and SNP type (transition or
transversion). The gene category was independent of SNP consequence (p = 0.37) and
type (p = 0.98). Note the yeast genome contains approximately 5,300 non-essential and
1,300 essential genes (30).
25
DISCUSSION
SNP Formation following Exposure to Lethal Environmental Stress
This study was intended to observe SNP formation under non-selective conditions
and following lethal bottlenecks imposed by repetitive, mutually exclusive, and
alternating selective pressure. We found the numbers of adapting colonies (5-FoA
treatment and –URA growth conditions) were close to what might have been expected
based on published rates of SNP formation in yeast, data regarding plasticity of URA3,
and the population sizes evaluated here (with assumptions described above regarding
additional levels of complexity (i.e. epistatic gene interaction) that might also contribute
to survival in selective media). The types of mutations observed indicate a
straightforward explanation for many of the resulting survival phenotypes (eg. PTCs in
URA3, suppressor mutations in tRNAs, reversion mutations in the same genes). From
among 40 strains subjected to selection cycling and multiple rounds of selection, all
exhibited genotypic changes very likely to account for the resulting phenotype (see also
below). It therefore does not appear necessary to invoke epigenetic or related factors as
an explanation for results such as those shown in Figure 3.
An increasing number of studies in both prokaryotes and eukaryotes suggests a
measure of directional SNP formation. In this context, the current studies were designed
to provide one test of directionality, including the hypothesis that with repeat exposures
to an environmental stress, yeast might become increasingly able to develop protective
26
mutations and promote their own survival. As described here, although colony counts
did increase modestly with progressive 5-FoA exposure, DNA sequence analysis and
intrinsic mutation rates (above) suggest this finding could be due to progressive
impairment of the URA3 gene product, and that increasing numbers of mutations in the
gene itself (or relevant tRNAs) lead to cumulative impairment of the pathway mediating
5-FoA toxicity (Figure 3). In the same fashion, decreasing numbers of surviving colonies
in –URA most likely reflect progressive difficulty rescuing a URA3-dependent pathway
with multiple deleterious mutations (Figure 3). Other possibilities include more complex
genetic or epigenetic alterations, such as specific gene defects at loci other than URA3 or
epistatic interactions (i.e. among many distinct genes).
Adaptive SNP Formation
We also considered a question concerning the balance that must exist between
new SNP formation (necessary to generate adaptive diversity under conditions such as
those described here) and long term genomic survival despite the inexorable
accumulation of SNPs within crucial yeast open reading frames. It is well established
that among humans and other higher eukaryotes, each generation adds a discrete number
of single nucleotide polymorphisms to the gene pool. Over hundreds of millions of years
of gene evolution, higher complexity genomes might therefore become compromised due
to ongoing mutation accrual. This might be the case regardless of natural selection or
recombination, since each new generation would be subject to a steadily increasing SNP
burden that would not be overcome simply by removing unfit individuals. Note that even
if specific individuals (or entire species) with detrimental mutations were expunged from
27
the DNA pool by purifying selection, drift, shift, etc., this would have no effect (and
would not reverse) an accumulating mutation burden among all survivors and their
constituent genomes. Based on these considerations, we modeled genomic SNP
accumulation in yeast using mutation rates reported here and by others as a test of
whether long term DNA stability should be expected. As shown in Figure 6, a mutation
rate at or below 0.5 mutations/genome/generation would always retain a population of
yeast with DNA identical to the founder strain (a “reservoir” of yeast preserving the
founder DNA sequence). This would have the effect of averting so-called “mutational
meltdown”.
Number of Yeast Cells Identical to
Founder Strain
1.E+16
1.E+14
1.E+12
µ=0
µ = 0.35
µ = 0.5
µ = 0.65
1.E+10
1.E+08
1.E+06
1.E+04
1.E+02
1.E+00
0
10
20
Generations
30
40
Figure 6: Model of mutation accumulation under varied mutation rates. The number of
yeast cells retaining the founder genome has been mathematically modeled. The
mutation rate (µ) is given in mutations/genome/generation. The number of cells at
generation 0 reflects the initial inoculum of YPD media (Table 1). Note that, if µ = 0 (no
new mutations occur), the number of yeast cells identical to the founder strain is simply
the total number of yeast under logarithmic growth conditions.
In general, the number of yeast cells in a culture identical to the founder genome can be
calculated by the following equation:
28
1:
=
∗ (2 − 2μ) , 0 < µ < 1
where N0 is the initial number of cells, µ is the mutation rate, T is the number of
generations, and NT is the total number of cells with no mutation at generation T.
The proportion of cells that contain the founder genome versus those with at least one
mutation is not dependent on the initial number of cells and can be calculated as follows:
2: = (1 − μ) , 0 < µ <1
where µ is the mutation rate, T is the number of generations, and PT the proportion
of non-mutated cells versus those with at least one mutation at generation T.
Our model indicates that, for a mutation rate ≤ 0.5 mutations/genome/generation, a
significant reservoir preserving the founder genome will be maintained. Because the
yeast mutation rate even in the setting of a significantly stressful environment is far
below this value (2.8 x 10-2 mutations/genome/generation), a significant reservoir of
yeast encoding DNA identical to the founder should be preserved.†
Global SNP Patterns Observed in this Study
One notable finding from the present study involved genomic comparisons
between a standard laboratory strain of S. cerevisiae and the published reference genome
for this species. Although different evolutionary/environmental pressures have been
experienced by the two strains, it was problematic to reconcile the observed SNP
frequency as attributable to purifying natural selection, particularly when one considers
available estimates regarding fitness in yeast. For example, note the ratio of exonic to
†
These calculations do not account for mutations that dramatically enhance fitness,
which would lead to strains that outcompete other genomes and eventual loss of the
original (founder) DNA sequence (Tables 2 and 3). In other words, considerations such
as 1) ways in which fitness may be gradually selected, 2) recurring bottlenecks and/or
positive selection, which can be purifying, and 3) interactions between variant alleles,
robustness, and buffering (which can mask detrimental effects) may represent important
contributions to the rate at which mutations accumulate in an overall yeast population.
29
intergenic SNPs in the laboratory strain tested here is appx. 1.5:1 (Table 4). If one
assumes a common ancestor for the two strains, the expected ratio of exonic:intergenic
(omitting purifying selection for the moment) would be 3:1 (i.e. appx. 75% of the yeast
genome is protein coding). Our findings therefore point to the somewhat surprising
conclusion that appx. 1 of every 2 randomly placed exonic SNPs in S. cerevisiae resulted
in a remarkable effect on fitness – i.e. 1 of every 2 random exonic SNPs conferred loss of
a yeast genome from the gene pool. This is in contrast to previous estimates indicating
that 0.1% – 2% of randomly placed mutations should have measurable effects on fitness
among laboratory yeast strains (18). The notion that 1 of every 2 random SNPs should
reduce fitness also seems at odds with the large number of yeast genes that are nonessential to laboratory growth and survival.
In a similar fashion, we observed a very strong bias in synonymous versus nonsynonymous SNPs when a laboratory strain was compared to the reference genome. In
particular, while the predicted ratio (based on random mutation accrual and all possible
permutations of the genetic code) indicates ~3:1 non-synonymous to synonymous single
base replacements as expected for stochastic SNP formation, our sequence analysis
indicates a ratio very close to unity (Table 4). The findings therefore suggest that appx. 2
of every 3 (66%) non-synonymous SNPs conferred a very significant effect on survival,
leading to removal of a sizable majority of randomly placed exonic SNPs (and their
respective yeast genomes) from the DNA pool. It is difficult to reconcile such findings
with the known plasticity of yeast DNA and the previous estimates of fitness ascribed to
random mutation (18).
30
A similar analysis has been conducted by our laboratory for both human and
murine genomic DNA. In each case, the enrichment of synonymous and intronic SNPs
seemed disproportionate to fitness effects observable in laboratory experiments alone. In
human SNPs downloaded from 1000 Genomes, measured ratios led to a conclusion
concordant with that in yeast: approximately half of all single nucleotide changes across
the human exome (assuming a random distribution of mutation) must have had effects on
fitness leading to removal from the gene pool. The ratio of synonymous to nonsynonymous SNPs reported in 1000 Genomes and by the Sanger Institute (for 16 strains
of congenic mice) (31,32) led to similar conclusions – approximately 2 of every 3 nonsynonymous mutations were expunged from the gene pool due to effects on fitness. This
sort of purifying selection does not seem commensurate with the known plasticity (and
tolerance to single nucleotide polymorphism) that are well established for human,
murine, and yeast genes. The likelihood that a solitary, randomly placed SNP would
cause early death or undermine propagation in 66% of cases seems at odds with a large
body of evidence regarding genomic resilience in hominid, murine, and yeast DNA.
Findings such as these indicate the possibility of biased SNP formation when
evaluating genome-wide SNP patterns and suggest that single nucleotide polymorphisms
could arise biologically (by yet to be determined mechanisms) in a non-random fashion
that favors intronic and synonymous locations (as opposed to being selected after random
appearance). In two earlier papers (33, 34), we emphasize the importance of transition
type mutations as a mechanism favoring synonymous nucleotide replacement, since
transition defects are strongly favored in virtually every genomic context studied to date,
and transitions (compared to transversion SNPs) strongly augment synonymous
31
substitution. Transition SNPs are also remarkably favored in yeast studied here (Table 4)
and provide new evidence that a bias with regard to synonymous SNP formation (as
opposed to natural selection) is responsible for a synonymous SNP preference. The
finding of similar SNP distributions in both essential and non-essential yeast genes (Table
5) provides evidence that purifying selection is not primarily responsible for the observed
enrichment of synonymous and non-coding SNPs (Table 4). Therefore, although our data
with regard to URA3 and selection cycling do not indicate directionality, full genomic
sequence analysis can be taken to suggest a very prominent bias regarding the
mechanisms by which yeast SNPs are generated (intergenic, synonymous) that strikingly
resembles the same SNP patterns in higher eukaryotes.
Considerations Regarding the “Dual Survivor” Phenotype
5-fluorouracil (5-FU), the active product of 5-FoA, does not exert a toxic effect
on cells without conversion to 5-fluorouridine triphosphate through the pyrimidine
salvage pathway. Specifically, 5-FU is converted to 5-fluorouridine monophosphate (5FUMP) by FUR1 (35) and subsequently phosphorylated to 5-flourouridine diphosphate
by URA6. The enzyme encoded by URA6 is integral to uracil synthesis and converts the
product of URA3, uridine monophosphate (UMP), to uridine diphosphate (UDP) (36).
Disruption of FUR1 has been linked to 5-FU resistance (35), but no mutations in FUR1
were detected by full genomic sequencing reported here. On the other hand, mutations in
URA6 were noted in association with the “dual survivor” phenotype in 3 of 3 strains
tested in our studies (examples shown in Table 3), including strains from the initial
screen of 5-FoA resistant colonies that possessed a wild type copy of URA3
32
(Supplemental Table 1). URA6 is an essential gene in yeast, and we hypothesis that
changes in substrate preference (or partial gene activity) could lead to a “dual survivor”
phenotype. For example, URA6 mutations that retain affinity for UMP but not 5-FUMP
would serve to protect yeast cells from 5-FoA while allowing production of uracil (and,
at least in principle, might support a small lawn on –URA media, Figure 4). Enzymatic
measurements to investigate conversion of UMP and 5-FUMP by URA6 and
transformation of wild type yeast with experimentally observed mutations in URA6 (e.g.
C → G mutation at position 392641 of chromosome XI, see Table 3) will be necessary to
more fully investigate the “dual survivor” strains described here.
SNP Accumulation in the Setting of Recurrent Environmental Stress
The mutation rate calculated using full genomic sequence data (2.3 x 10-9
mutations/base pair/cell division) is at least 7 fold higher than that reported previously
(3.3 x 10-10 mutations/base pair/cell division, 18). This difference might be attributable to
a number of factors, including error in estimates due to assumptions that differ inherently
between independent studies measuring mutation rates. For example, our experimental
design actively selects for mutations by exposure to an environmental stress, which might
inflate the mutation rate above results obtained under entirely non-stressed conditions.
The founder strain used in this study might also carry an intrinsic defect leading to
increased SNP formation (although full genomic sequence analysis did not reveal any
polymorphisms shown previously to alter the intrinsic mutation rate). Although it
remains possible that the cycling protocol selected for a strain with an increased
propensity towards SNP formation, our analysis to date does not provide conclusive
33
evidence for a directional or otherwise biased SNP formation insofar as URA3 and the
specific environmental perturbations described here are concerned.
34
CONCLUSION
Genomic polymorphism represents a fundamental basis for the evolutionary
process, but mechanisms that underlie the formation and frequency of single nucleotide
replacements are not fully understood. This project analyzed the appearance and
accumulation of SNPs in S. cerevisiae following exposure to two lethal forms of selective
pressure. The study, although limited in its scope, did not provide compelling evidence
of directionality with regard to appearance of new SNPs in URA3. That being said,
further experiments are merited to better test whether the diversity of mutations expected
in such large populations of yeast is actually achieved and whether or not discrete URA3
defects (or their genetic revertants) are repeatedly observed. During iterative rounds of
selection cycling, we also found SNP accumulation often led to eventual inviability
unless a novel (“dual survivor”) phenotype was achieved. One interesting observation
from the studies was that – prior to selection cycling – a strong mutation bias was evident
in a standard laboratory strain of yeast, and that SNP patterns observed here were not
readily explained by purifying selection, based on the magnitude of fitness effects that
would otherwise be required. One mechanism contributing to the strong synonymous
SNP bias described by our studies likely involves a preponderance of transition SNPs in
yeast, which would influence SNP formation and favor synonymous nucleotide
35
replacement.† The findings described here contribute to the understanding of SNP
accrual in eukaryotic cells, as well as possible ramifications of an intrinsically biased
pattern of mutation accrual in yeast.
†
As part of further studies, it will also be important to incorporate a non-stressed control
to compare mutation distributions under selective versus non-selective conditions.
36
LIST OF REFERENCES
1. Luria, S. E., & Delbruck, M. (1943). Mutation of bacteria from virus sensitivity to
virus resistance. Genetics, 6, 491-511.
2. Martincorena, I., & Luscombe, N. (2013). Non-random mutation: the evolution of
targeted hypermutation and hypomutation. Bioassays, 35(2), 123-130.
3. Kang, J. M., Iovine, N. M., & Blaser, M. J. (2006). A paradigm for direct stressinduced mutation in prokaryotes. The Journal of the Federation of American Societies
for Experimental Biology, 20, 2476-2485.
4. Goho, S., & Bell, G. (2000). Mild environmental stress elicits mutations affecting
fitness in Chlamydomonas. Proceedings of the Royal Society of London, Series B,
267, 123–129.
5. Cairns, J., & Foster, P. L. (1991). Adaptive reversion of a frameshift mutation in
Escherichia coli. Genetics, 128, 695-701.
6. Steele, D. F., & Jinks-Robertson, S. (1992). An Examination of adaptive reversion in
Saccharomyces cerevisiae. Genetics, 132, 9-21.
7. Lolle, S. J., Victor, J. L., Young, J. M., & Pruitt, R. E. (2005). Genome-wide nonmendelian inheritance of extra-genomic information in Arabidopsis. Nature, 434,
505-509.
8. Oliver, A., et. al. (2000). High frequency of hypermutable Pseudomonas aeruginosa
in cystic fibrosis lung infection. Science, 289, 381-382.
37
9. Sniegowski, P., Gerrish, P., Johnson, T., & Shaver, A. (2000). The evolution of
mutation rates: separating causes from consequences. BioEssays, 22 (12), 1057-1066.
10. Mitchell, A., et. al. (2009). Adaptive prediction of environmental changes by
microorganisms. Nature, 460, 220-224.
11. Shenton, D., et. al. (2006). Global translational responses to oxidative stress impact
upon multiple levels of protein synthesis. The Journal of Biological Chemistry, 281
(39), 29011-29021.
12. Morrison, A., Johnson, A. L., Johnston, L. H., & Sugino, A. (1993). Pathway
correcting DNA replication errors in Saccharomyces cerevisiae. The EMBO Journal,
12 (4), 1467-1473.
13. Kunz, B. A., Ramachandran, & K., Vonarx, E. J. (1998). DNA sequence analysis of
spontaneous mutagenesis in Saccharomyces cerevisiae. Genetics 148, 1491–1505.
14. Nick McElhinny, S. A., Stith, C. M., Burgers, P. M., & Kunkel, T. A. (2007).
Inefficient proofreading and biased error rates during inaccurate DNA synthesis by a
mutant derivative of Saccharomyces cerevisiae DNA polymerase delta. Journal of
Biological Chemistry, 282(4), 2324-2332.
15. Zhang, F., & Zhao, Z. (2004). The influence of neighboring-nucleotide composition
on single nucleotide polymorphisms (SNPs) in the mouse genome and its comparison
with human SNPs. Genomics, 84(5), 785-795.
16. Sved, J., & Bird, A. (1990). The expected equilibrium of the CpG dinucleotide in
vertebrate genomes under a mutation model. Proceedings of the National Academy of
Sciences, 87(12), 4692-4696.
38
17. Tang, Y., et. al. (2012). Widespread existence of cytosine methylation in yeast DNA
measured by gas chromatography/mass spectrometry. Analytical Chemistry, 84(16),
7249-7255.
18. Lynch, M., et al. (2008). A genome-wide view of the spectrum of spontaneous yeast
mutations. Proceedings of the National Academy of Sciences, 105 (27), 9272-9277.
19. Wloch, D. M., Szafraniec, K., Borts, R. H., & Korona, R. (2001). Direct estimate of
the mutation rate and the distribution of fitness effects in the yeast Saccharomyces
cerevisiae. Genetics 159, 441–452.
20. Lang, G., & Murray, A. (2008). Estimating the per-base-pair mutation rate in the
yeast Saccharomyces cerevisiae. Genetics, 178 (1), 67-82.
21. Xue, Y. et. al. (2009). Human Y chromosome base-substitution mutation rate
measured by direct sequencing in a deep-rooting pedigree. Current Biology, 19 (17),
1453-1457.
22. Lacroute, F. (1968). Regulation of pyrimidine biosynthesis in Saccharomyces
cerevisiae. Journal of Bacteriology, 95 (3), 824-832.
23. Ko, N., Nishihama, R., & Pringle, J. R. (2008). Control of 5-FOA and 5-FU
resistance by Saccharomyces cerevisiae YJL055W. Yeast, 25 (2), 155-160.
24. Boeke, J. D., Trueheart, J., Natsoulis, G., & Fink, G. R. (1987). 5-Fluoroorotic acid
as a selective agent in yeast molecular genetics. Methods in Enzymology, 154, 164175.
25. Muller, H. (1964). The relation of recombination to mutational advance. Mutation
Research, 106, 2-9.
39
26. Burke, D., Dawson, D., & Stearns, T. (2000). Methods in Yeast Genetics, Cold Spring
Harbor Laboratory Press.
27. Indira, S., et. al. (2009). Stringent mating-type-regulated auxotrophy increases the
accuracy of systematic genetic interaction screens with Saccharomyces cerevisiae
mutant arrays. Genetics, 181(1), 289-300.
28. Hall, B.M., Ma, C., Liang, P. & Singh, K.K. (2009). Fluctuation AnaLysis
CalculatOR (FALCOR): a web tool for the determination of mutation rate using
Luria-Delbruck fluctuation analysis. Bioinformatics, 25(12), 1564-1565.
29. Yu, L., et. al. (2006). A survey of essential gene function in the yeast cell division
cycle. Molecular Biology of the Cell, 17(11), 4736-4747.
30. Giaever, G., et. al. (2002). Functional profiling of the Saccharomyces cerevisiae
genome. Nature, 418, 387-391.
31. The 1000 Genomes Project Consortium. (2012). An integrated map of genetic
variation from 1,092 human genomes. Nature, 491, 56-65.
32. Yalcin, B., Adams, D.J., Flint, J., & Keane, T.M. (2012). Next-generation sequencing
of experimental mouse strains. Mammalian Genome, 23, 490-498.
33. Plyler, Z., et. al. (2013). The significance of contextual SNP patterns in the murine
and hominid genome, and evidence for rapid parallel evolution in mice. Submitted for
publication.
34. Hill, A., et. al. (2013). Evolutionary teleos: an argument for non-randomness and
adaptive design during genomic SNP accumulation in CFTR. Submitted for
publication.
40
35. Kern L, et al. (1991). Regulation of the pyrimidine salvage pathway by the FUR1
gene product of Saccharomyces cerevisiae. Current Genetics, 19(5), 333-337
36. Liljelund, P., & Lacroute, F. (1986). Genetic characterization and isolation of the
Saccharomyces cerevisiae gene coding for uridine monophosphokinase. Molecular
and General Genetics, 205(1),74-81
41
APPENDIX A
SUPPLEMENTAL TABLES
42
SUPPLEMENTAL TABLES
chrom
III
IV
VII
IX
XI
chrom
III
VII
VIII
IX
XI
XV
position SGD
303798 G
17231 A
530034 A
437368 A
392987 G
position SGD
303798 G
530034 A
413608 T
437368 A
392987 G
60240 C
Founder
G
A
A
A
G
Founder
G
A
T
A
G
C
Round 1 5-FoA
A
G
C
T
T
Round 1 5-FoA
A
C
C
T
T
T
SNP Effect
Intergenic
Intergenic
SYN
Pseudo
NON-SYN
SNP Effect
Intergenic
SYN
Intergenic
Pseudo
NON-SYN
Intergenic
Gene
MTL1
YIR043C
URA6
Gene
MTL1
YIR043C
URA6
Supplemental Table 1: Mutations in two strains observed during one round of 5-FoA
selection to reach “dual survivor” phenotype. SGD designates nucleotide in reference
genome (Saccharomyces Genome Database). Founder denotes sequence data of original
laboratory strain tested here, with “Round 1 5-FoA” indicating genomic sequence
observed during the first cycle of mutation. Color coding indicates the type of SNP effect
(see legend).
Color Legend
Intergenic Occurs in Intergenic Region
NON-SYN Non-Synonymous
SYN
Synonymous
Pseudo
Occurs in Pseudogene
43
APPENDIX B
SUPPLEMENTAL FIGURES
44
SUPPLEMENTAL FIGURES
Non-synonymous:Synonymous
Ratio
6
5
4
Hypothesized
Gene of
Same Size
URA3
3
2
1
0
Transtion SNPs
Transversion SNPs
Supplemental Figure 1: Expected ratio for non-synonymous:synonymous SNPs
considering transition and transversion SNPs in URA3 and a hypothesized gene of the
same size. Every possible SNP at every nucleotide position was incorporated in this
model. In both cases, transition SNPs were much more likely to be associated with
synonymous mutations. URA3 SNPs were based on codon usage of URA3 and all
possible SNPs (~2400) in URA3. The hypothesized gene was established based on equal
representation of every codon, with the exception of only one stop codon in order to
mimic a functional open reading frame. Note that, in our experimental analysis of appx.
96 URA3 SNPs (identified during selection cycling by this project), the ratio of
transverion:transition polymorphisms was ~1.5:1.
45