Oxidative Stress Survival in a Clinical Saccharomyces cerevisiae Isolate Is Inﬂuenced by a Major Quantitative Trait Nucleotide Stephanie Diezmann1,2 and Fred S. Dietrich Institute for Genome Sciences and Policy and Department of Molecular Genetics and Microbiology, Duke University Medical Center, Durham, North Carolina 27710
ABSTRACT One of the major challenges in characterizing eukaryotic genetic diversity is the mapping of phenotypes that are the cumulative effect of multiple alleles. We have investigated tolerance of oxidative stress in the yeast Saccharomyces cerevisiae, a trait showing phenotypic variation in the population. Initial crosses identiﬁed that this is a quantitative trait. Microorganisms experience oxidative stress in many environments, including during infection of higher eukaryotes. Natural variation in oxidative stress tolerance is an important aspect of response to oxidative stress exerted by the human immune system and an important trait in microbial pathogens. A clinical isolate of the usually benign yeast S. cerevisiae was found to survive oxidative stress signiﬁcantly better than the laboratory strain. We investigated the genetic basis of increased peroxide survival by crossing those strains, phenotyping 1500 segregants, and genotyping of high-survival segregants by hybridization of bulk and single segregant DNA to microarrays. This effort has led to the identiﬁcation of an allele of the transcription factor Rds2 as contributing to stress response. Rds2 has not previously been associated with the survival of oxidative stress. The identiﬁcation of its role in the oxidative stress response here is an example of a speciﬁc trait that appears to be beneﬁcial to Saccharomyces cerevisiae when growing as a pathogen. Understanding the role of this fungal-speciﬁc transcription factor in pathogenicity will be important in deciphering how fungi infect and colonize the human host and could eventually lead to a novel drug target.
ontinuous phenotypic variation is the norm rather than the exception for most eukaryotic traits. Unlike Mendelian traits that are governed by a single gene, quantitative traits are controlled by multiple genes, typically unlinked, described as quantitative trait loci (QTL) (Geldermann 1975). Locating these QTL and pinpointing the responsible genes (QTGs) and nucleotides (QTNs) are central challenges of genetics that are being pursued with sophisticated genotyping and mapping methods. Selective genotype data for progeny with one of the two extreme phenotypes have been found to be most informative (Lander and Botstein 1989). Individuals can be genotyped separately via single
Copyright © 2011 by the Genetics Society of America doi: 10.1534/genetics.111.128256 Manuscript received March 1, 2011; accepted for publication April 18, 2011 Supporting information is available online at http://www.genetics.org/cgi/content/ full/genetics.111.128256/DC1. 1 Present address: Department of Molecular Genetics, University of Toronto, Toronto, ON M5S 1A8, Canada. 2 Corresponding author: Department of Molecular Genetics, University of Toronto, Medical Sciences Bldg., Room 4368, 1 King's College Circle, Toronto, ON M5S 1A8, Canada. E-mail: [email protected]
segregant analysis (SSA) or by pooling their genomic DNA and performing bulk segregant analysis (BSA) (Arnheim et al. 1985; Michelmore et al. 1991; Quarrie et al. 1999). Once a QTL has been located in the genome, the next level of analysis is the identiﬁcation of the responsible QTG and QTN, which are typically veriﬁed by the homologous replacement of the candidate gene or polymorphism between both parental strains using site-directed mutagenesis (Sinha et al. 2008b). Microbial QTL that modulate virulence are of particular scientiﬁc interest due to their impact on human health. Virulence-associated QTL have been identiﬁed in the parasites Toxoplasma gondii (Su et al. 2002) and Trypanosoma brucei (Morrison et al. 2009) and the opportunistic pathogen S. cerevisiae (Steinmetz et al. 2002), which has rapidly become an exquisite model system for the study of quantitative genetics. Successfully mapped S. cerevisiae quantitative traits include sporulation efﬁciency (Deutschbauer and Davis 2005; Ben-Ari et al. 2006), regulation of gene expression (Brem et al. 2002), DNA damage repair (Demogines et al. 2008), cell morphology (Brauer et al. 2006), genetic
Genetics, Vol. 188, 709–722
changes in experimentally evolved populations (Segre et al. 2006), and ethanol tolerance (Hu et al. 2007). The level of resolution to which S. cerevisiae quantitative traits have been mapped varies. Some traits, like resistance to small molecules and ethanol, have been mapped to the level of candidate QTL without independent marker veriﬁcation for this region (Perlstein et al. 2006; Hu et al. 2007). For others, such as sporulation efﬁciency (Deutschbauer and Davis 2005), high-temperature growth (Sinha et al. 2008a), and DNA damage repair (Demogines et al. 2008), the causative QTNs have been experimentally validated. The small number of documented QTNs highlights the magnitude of the challenge of mapping quantitative traits even in S. cerevisiae. The only other eukaryotic QTN that has been experimentally conﬁrmed to date is a nonsynonymous substitution in the Caenorhabditis elegans calpain-like protease gene, which attenuates growth at lower temperatures (Kammenga et al. 2007). In addition to being a prominent genetic model system, S. cerevisiae has also been described as an emerging pathogen with case report numbers steadily increasing since the late 1950s (Cimolai et al. 1987; Hazen 1995; Skovgaard 2007). Notably, S. cerevisiae infections are indistinguishable from those caused by the most widely recognized fungal pathogen Candida albicans (Sobel et al. 1993; McNeil et al. 2001; Zaoutis et al. 2005). As with C. albicans, S. cerevisiae can cause a wide spectrum of diseases ranging from superﬁcial cutaneous infections and vaginitis to life-threatening systemic infections of the blood stream and vital organs in immunocompetent and immunocompromised individuals (Enache-Angoulvant and Hennequin 2005; McCusker 2006). While C. albicans is the most common cause of fatalities due to mycoses, with rates up to 45% (Tortorano et al. 2004), it lacks a complete sexual cycle, complicating the genetic dissection of clinically relevant traits. A common characteristic of human pathogens is their ability to overcome and neutralize the host’s oxidative immune defenses including reactive oxygen species (ROS) such as hydrogen peroxide (H2O2) and superoxide radicals (Iyer et al. 1961; Abshire and Neidhardt 1993; Rea et al. 2004; Riboulet et al. 2007). Upon entering the blood stream, microbial invaders face the cellular immune response made up of neutrophils and macrophages. Both respond to the ingestion of foreign particles with increased O2 uptake and ROS production (Sbarra and Karnovsky 1959; Iyer et al. 1961). The primary product of this oxidative burst is superoxide radical that is immediately converted to H2O2, which nonspeciﬁcally damages nucleic acids, lipids, and proteins (Halliwell and Gutteridge 2007). S. cerevisiae’s close kinship with C. albicans (Diezmann et al. 2004) and its facilities as a genetic model system make this yeast an attractive model to study the oxidative stress response in fungal pathogens (McCusker 2006). Genetic dissection of the oxidative stress response yielded two transcription factors required for the survival of oxidative stress not only in S. cerevisiae but also in C. albicans (Singh
S. Diezmann and F. S. Dietrich
et al. 2004), and other pathogenic fungi, such as Aspergillus fumigatus (Kusch et al. 2007; Lamarre et al. 2007) and Cryptococcus neoformans (Wormley et al. 2005). Yap1 and Skn7 are activated via oxidation by H2O2 and initiate transcription of a whole suite of genes whose products detoxify ROS (Krems et al. 1996; Delaunay et al. 2000; Causton et al. 2001). H2O2 furthermore initiates transcription of the catalase gene CTT1, turning the catalase genes into prominent candidates in any investigation into the fungal oxidative stress response (Marchler et al. 1993). Catalases, however, are dispensable for virulence in C. albicans (Wysong et al. 1998), C. neoformans (Giles et al. 2006), and A. fumigatus (Calera et al. 1997; Paris et al. 2003). This observation raises the question as to which molecules or pathways protect pathogenic fungi from the human oxidative defense system. Two investigations reported that oxidative stress resistance is a quantitative trait in S. cerevisiae (Perlstein et al. 2006; Witten et al. 2007). Witten et al. (2007) evolved two yeast strains for oxidative stress resistance and showed that both strains acquired the same unknown major-effect mutation but differed in the nature of unidentiﬁed segregating modiﬁers. Perlstein et al. (2006) tested growth inhibition in F1 segregants in response to small molecules, including H2O2, and came to the conclusion that two as of yet undiscovered loci affect oxidative stress resistance. When we examined the evolutionary relationships among .100 S. cerevisiae strains collected from diverse environmental niches, we showed that clinical isolates survive oxidative stress signiﬁcantly better than vineyard, soil, or fruit isolates (Diezmann and Dietrich 2009). On the basis of these results, we concluded that oxidative stress resistance is an adaptive property of virulent strains (Diezmann and Dietrich 2009). We furthermore noticed a striking ﬁvefold difference between a diploid derivative of the laboratory reference strain S288c (S344) and the clinical isolate YJM789 (YJK1272). S344 and YJK1272 have successfully been used for the genetic dissection of another clinically important quantitative trait, high-temperature growth (Steinmetz et al. 2002). The genomes of both strains have been sequenced (Goffeau et al. 1996; Wei et al. 2007), which is a prerequisite for the design of strain-speciﬁc probes for microarray genotyping. Given the dramatic difference in peroxide survival, the excellent track record of S344 and YJK1272 for QTL mapping and the availability of genome data we set out to identify the genetic basis of oxidative stress survival. Here we describe a QTL mapping strategy that combines BSA and SSA with recombination mapping that led to the identiﬁcation of a QTN causing the clinical isolate to survive oxidative stress signiﬁcantly better than the laboratory strain. We exploited the naturally occurring genetic variation between both strains in our mapping effort. By combining microarray-assisted genotyping with classic mapping, we were able to identify a QTL on chromosome XVI. Site-directed mutagenesis validated a nonsynonymous polymorphism in
a fungal-speciﬁc transcription factor as the oxidative stress QTN. The polymorphic site encodes a neutral amino acid in the low-survival laboratory strain but a negatively charged amino acid in the high-survival clinical isolate. The negatively charged residue is conserved across the fungal tree of life. We show that the zinc cluster transcription factor, RDS2, a gene previously identiﬁed in regulating gluconeogenesis (Soontorngun et al. 2007) plays a role in oxidative stress survival in S. cerevisiae.
Materials and Methods Yeast strains and media
All isolates (supporting information, Table S1) were permanently kept in cryostorage in 15% (v/v) glycerol at 280 and cultured for 48 hr on rich media plates [(YPD) 1% (w/v) yeast extract, 2% (w/v) peptone, 2% (w/v) dextrose, 2% (w/v) agar] prior to experimentation (Burke et al. 2000). Synthetic-deﬁned media plates for selection after transformation with the LYS5MX4 cassette (Wach et al. 1994) contained 2% (w/v) glucose, 37.8 mM (NH4)2SO4, 1.7 g/liter YNB without amino acids and ammonium sulfate, and 2% (w/v) agar. For high-throughput phenotyping of backcross segregants, 1 and 1.2 mM H2O2 were added to autoclaved and cooled synthetic-deﬁned media. Concentration of H2O2 in the stock solution was determined spectrometrically at l ¼ 240 nm (e ¼ 39.4 M21 cm21) and adjusted accordingly (Lushchak and Gospodaryov 2005). Plates were stored in the dark at 4 for no more than 5 days. a-Aminoadipate plates were prepared as described before (Zaret and Sherman 1985) with two modiﬁcations: the YNB-dextrose-lysine mixture was not autoclaved but ﬁlter sterilized and the pH of the a-aminoadipate solution was adjusted to 5.5 instead of 6. Lithium acetate transformation of S. cerevisiae using MX4 cassettes
The high efﬁciency lithium acetate transformation protocol (Gietz and Schiestl 1991), with modiﬁcations, was used to generate allele replacement strains (Table S1). The washed and pelleted cells were diluted in 400 ml 0.1 M LiAc and 50– 100 ml of this suspension used for transformation. After adding the transformation mix, samples were incubated for 30 min at 30 and for 15–20 min at 42 in water baths. Transformed cells were centrifuged for 15 sec at 6000 rpm and gently resuspended in 600 ml YPD, when transformed with hphMX4, kanMX4, or pAG26 (hphMX4CEN), or 600 ml sterile water, when transformed with LYS5MX4 or a PCR product. MX4 cassettes were PCR ampliﬁed as follows: initial denaturation at 94 for 3 min was followed by 10 cycles of 94 for 20 sec, 55 for 15 sec, and 72 for 1 min 30 sec. These were followed by 25 cycles of 94 for 20 sec and 72 for 1 min 30 sec (modiﬁed from Goldstein and McCusker 1999). Cells transformed with MX4 cassettes or plasmids were allowed to recover for 2 hr at 30 while shaking (200 rpm) before being spread onto selective media. Cells transformed with LYS5MX4 were spread onto synthetic-
deﬁned media and cells transformed with PCR products onto a-aminoadipate plates. Plates were incubated for 3 days at 30. Measurement of peroxide survival
To assess the ability of yeast strains to survive peroxide stress, survival in the presence of the organic H2O2 analog tert-Butyl hydroperoxide (TBHP; Sigma) was measured as follows: Cells were grown overnight in liquid YPD and washed twice with 1· phosphate buffered saline before being diluted to 2000 cells/ml and exposed to 20 mM TBHP for 60 min. Treated and untreated samples were then plated on YPD agar and colony forming units (CFUs) counted after 48 hr of incubation at 30. Survival is the ratio between treated and untreated cells. Heritability of peroxide survival
The genetic basis of peroxide survival was investigated by sporulating YJK1272 and S344 in liquid 0.5% (w/v) potassium acetate without glucose or other additives at 25 (S344) or 30 (YJK1272) (Codon et al. 1995) and mating them to generate the hybrid strain YJS (Table S1). To assess the phenotypic nature of peroxide survival and identify its mode of inheritance, YJK1272, S344, YJS, and 200 F1 segregants were phenotyped using the CFU assay described above. To estimate the effective number of loci segregating in the S344 · YJK1272 cross, the Castle–Wright estimator (ne) and its standard error were calculated on the basis of the phenotypic variance observed in the F1 as described by Lynch and Walsh 1998, p. 234. Diagnostic PCR and RFLP assays for ROS detoxifying candidate gene genotyping and QTL ﬁne mapping
A total of 200 F1 segregants were genotyped for the candidate genes CTA1, CTT1, CCP1, and SOD1 using restriction length polymorphisms (Table S2) on PCR products derived from genomic DNAs extracted with the CTAB method (Gardes and Bruns 1993). Fine mapping of the QTL on chromosome XVI was conducted on a panel of 22 high- and 20 low-survival segregants from backcross line I (BCI). High-survival segregants were scored for the presence of YJK1272 and S344 alleles at seven markers spanning the QTL and low-survival segregants for four of the seven markers (Table S2). The seven markers comprised three diagnostic RFLPs and four allele-speciﬁc PCR reactions. Diagnostic RFLPs were also tested on 47 high-survival and 16 low-survival F1 segregants. A set of allele-speciﬁc PCR primers was composed of three primers, two of which are speciﬁc due to a mismatch at the ﬁnal 39 position and a third general primer to amplify in the opposite direction from the mismatch primer. Two PCR reactions per segregant were analyzed, each containing one allele-speciﬁc and the general primer. Amplicons were electrophoretically separated and scored for presence/absence of fragments in both reactions. Strains S344 and YJK1272 served as positive controls.
S. cerevisiae oxidative stress QTN
Backcross design and segregant phenotyping
Three backcross lines (BCI, BCII, and BCIII) were created by crossing high-survival F1 segregants F1–2D (survival ¼ 0.79, drug marker hphMX4), F1–10B (0.91, hphMX4), F1–19C (0.99, kanMX4), with strain S344 (kanMX4) or S347 (hphMX4) (Table S1). Double-drug resistant hybrids were sporulated and tested for drug marker segregation. These backcross segregants were phenotyped using the quantitative CFU assay (Diezmann and Dietrich 2009) and a spot dilution assay for high-throughput preliminary screening (R. Strich, personal communication). A total of 1246 backcross segregants were phenotyped in the spot-dilution screen. For this, 5 ml of 10-fold dilutions of S344, YJK1272, and eight backcross segregants (two complete tetrads), ranging from 107 to 102 cells/ml, were spotted onto one synthetic-deﬁned media plate that contained 1 mM or 1.2 mM H2O2. Plates were incubated for 48 hr at 30 and the phenotype of putative high-survival segregants, those resembling growth of YJK1272, was conﬁrmed and quantiﬁed using the CFU assay. Microarray-assisted QTL mapping
The mapping population comprised 47 F1 and 22 BCI highsurvival segregants. Pools of F1 and BCI DNAs were hybridized to the Affymetrix S98 array and compared to parental hybridizations. The 22 BCI segregants were separately hybridized to the 4x2k Combimatrix custom-designed array and compared to both parental strains. For detailed description of the microarray design and analyses see Supporting Methods in File S1. Characterization of RDS2
The effect of the candidate gene RDS2 on peroxide survival was investigated using three different comparisons: (i) Wild isolates of S. cerevisiae that have S344 or YJK1272 like RDS2 alleles were compared with each other and S344 and YJK1272, (ii) homozygous RDS2 deletions in S344 and YJK1272 were examined for peroxide survival, and (iii) S344 and YJK1272 strains with single nucleotide polymorphism replacements in RDS2 were compared to peroxide survival of the wild-type strains in TBHP. The Saccharomyces_cerevisiae_WGS database at GenBank has been mined using BLASTn with 50-bp long sequences of RDS2 of either parental allele containing one of the two nonsynonymous polymorphisms using MEGABLAST (Altschul et al. 1990). The wild S. cerevisiae strains identiﬁed as having a S344- or YJK1272-like allele, were tested three times for survival in 20 mM TBHP using the CFU assay. The following comparisons were analyzed with one-way ANOVA with Bonferroni correction: S344 vs. S344-like strains, YJK1272 vs. YJK1272-like strains, S344 vs. YJK1272-like strains, and YJK1272 vs. S344-like strains. To identify patterns of evolutionary conservation, Rds2 was BLAST searched against the protein translations of the genomes of C. glabrata, Ashbya gossypii, Kluyveromyces lactis, Clavispora lusitaniae, C. albicans, Yarrowia lipolytica,
S. Diezmann and F. S. Dietrich
A. fumigatus, A. nidulans, Penicillium marneffei, Magnaporthe grisea, Giberella zea, Neurospora crassa, Coprinopsis cinerea, Malassezia globosa, Ustilago maydis, and C. neoformans deposited at the National Center for Biotechnology Information (NCBI) GenBank and the alignment visually analyzed. Homozygous RDS2 deletions in either parental background were generated by mating two strains with heterozygous deletions of RDS2 with hphMX4 and kanMX4 with each other and screening the diploids for lysine prototrophy and double drug resistance. Two transformants of each parental background were analyzed and the four homozygous deletion strains tested for survival in 16 mM TBHP three times. Means and standard error were calculated and compared with Student’s t-test. Site-directed mutagenesis of the C751G single nucleotide polymorphism (SNP) in RDS2 was carried out in the lysine auxotrophic haploid strains S1288 and YHS633 (Table S1), employing the lysine/a-aminoadipate counterselection system (Ito-Harashima and McCusker 2004). The two-step process entailed replacement of the region between nucleotides 701 and 830 inside the RDS2 open reading frame (ORF) with LYS5MX4 (pSA39) (Ito-Harashima and McCusker 2004) and removal of LYS5MX4 with a PCR product containing the other parent’s SNP. A total of 1 ml of a 1:100 dilution of pSA39 was added to reactions amplifying LYS5MX4 with primers JM41_AS751-828_F and JM42_AS751-828_R (Table S2) (Goldstein and McCusker 1999; Ito-Harashima and McCusker 2004). Successful integration of LYS5 and removal of the RDS2 target region was tested with primer pairs FT167/RDS2_2500F and JM37/RDS2_1500R at TA ¼ 50 and primers RDS2_701F and RDS2_830R at TA ¼ 53 (Table S2). A total of 1 ml genomic DNA, prepared using the 10-min protocol (Hoffman and Winston 1987), was added to generate PCR products for homologous allele replacement and to test for successful replacement and marker removal. Homologous allele-replacement PCR products were generated by combining two short fragments that were ampliﬁed off parental DNA with the following primer combinations: RDS2_550F/RDS2_ 755_Y_R (i) and RDS2_751_Y_F/RDS2_1000R (ii) were used on S1288 and primers RDS2_550F/RDS2_755_S_R (iii) and RDS2_751_S_F/RDS2_1000R (iv) on YHS633. PCR products (i) and (iv) and (ii) and (iii) were mixed and 1 ml ampliﬁed with primers RDS2_550F and RDS2_ 1000R (Table S2). All reactions were done at TA ¼ 52. The PCR product worked less efﬁciently in the Y background. Hence, a second PCR product, with longer regions of homology, was produced by replacing primer RDS2_ 500F with RDS2_102F and RDS2_1000R with RDS2_ 1315R (Table S2), resulting in transformant YSD52 (Table S1). The LYS5MX4 replacement transformation was done as cotransformation with 2 ml plasmid pAG26 (hphMX4 CEN URA3) (Goldstein and McCusker 1999). After 3 days of incubation at 30, drug-resistant transformants were replica
plated onto a-aminoadipate plates and incubated at 30 for 2 days. Putative transformants were tested for correct SNP replacement by restriction digest with BbsI and sequenced with BigDye chemistry version 3, according to the instructions supplied by Applied Biosystems. PCR amplicons for the BbsI digest were generated using primers RDS2_102F and RDS2_966R at TA ¼ 50 and 8-ml PCR product were digested with 10· reaction buffer and 1 unit BbsI. Sequencing was carried out using primers RDS2_102F, RDS2_966R, RDS2_550F, RDS2_1000R, RDS2_701F, and RDS2_1315R (Table S2). Two independent transformants were created in both genetic backgrounds. To diploidize haploid strains with allele replacements (YSD1, -3, -36, and -52) and the wild types (S1288 and YHS633), all of which are MATa lys5, two crosses were conducted (Table S1). S1288 and YSD1 were crossed with YSD-S1 and YHS633 and YSD36, were crossed with YSD-Y1. Diploids were selected for hygromycin B drug resistance and lysine prototrophy and single colonies isolated, sporulated, and between 40 and 60 tetrads dissected. Those tetrads were subjected to tester matings on SD and YPD 1 hygromycin B. The MATa lys2 segregants were then mass mated to the transformed and wild-type strains to generate diploids (Table S1). Additionally, matings were carried out to generate strains that are heterozygous for the polymorphism and hybrid strains that are homozygous for one or the other C751G SNP (Table S1). All strains were screened for survival in 16 mM TBHP up to 10 times. Differences between genotypes were assessed using one-way ANOVA with Bonferroni correction and Dunnett’s test and the Wilcoxon signed rank test to detect differences in peroxide survival between strains with different genotypes. The percentage of phenotypic contribution was calculated as 100 · (Survival(YJK1272 C*/C*) 2 Survival (YJK1272 G/G))/(Survival(S344 C/C) 2 Survival(YJK1272 G/G)), where the * denotes allele replacement (Deutschbauer and Davis 2005).
Results Peroxide survival is a dominant quantitative trait
Crosses between S344 and YJK1272 indicate that enhanced peroxide tolerance is a dominant quantitative trait (Figure 1). The S344 · YJK1272 hybrid (YJS) phenotype (survival ¼ 0.93 6 0.098) resembled that of the clinical strain YJK1272 (0.83 6 0.124). Both strains survived signiﬁcantly better than the laboratory background S344 (0.15 6 0.062) (P , 0.01) (Figure 1A). To establish whether peroxide survival is a Mendelian or quantitative trait, YJS was sporulated and a total of 68 tetrads, each of which gave rise to four viable spores exhibiting 2:2 segregation of the drug resistance markers hygromycin B and G418, were obtained. A total of 200 spores originating from 50 of these F1 tetrads were tested for peroxide survival (Figure 1B). Survival ranged from 0 to 1 and the plot resembled that of a bellshaped curve, a diagnostic feature of quantitative traits. On
both extremes of the curve, F1 segregants were observed whose phenotypes exceeded that of the parental means, suggesting transgressive segregation. A total of 47 F1 segregants (23.5%) exhibited highsurvival $0.8 (YJK1272-like), and 16 segregants (8%) low-survival #0.2 (S344-like). On the basis of the phenotypic variance among F1 segregants (0.0587) we estimated that 10 6 2 oxidative stress survival loci segregate in this cross, conﬁrming the quantitative nature of peroxide survival. The high- and low-survival F1 segregants were included in the ensuing QTL mapping effort. To determine whether the survival phenotype is linked to genes that encode the ROS detoxifying enzymes catalases CTA1 and CTT1, peroxidase CCP1, or superoxide dismutase SOD1, a primary candidate gene screen was conducted. The parental alleles of those genes differ by a total of ﬁve nonsynonymous and 14 synonymous SNPs in their ORFs and an additional 38 polymorphisms in the 1000-bp up- and downstream regions (Figure 1C). An RFLP assay was applied to genotype segregants for the presence of either parental allele. However, no association between parental allele and phenotype could be detected. The genes segregated at random with respect to peroxide survival; therefore, the phenotype is not linked to any of the ROS detoxifying candidate genes (Figure 1C). Generating a selective mapping population from two independent backcross lines
After establishing that peroxide survival is inherited quantitatively, we pursued the identiﬁcation of the genetic factors responsible by establishing a selective QTL mapping population for use in SSA and BSA microarray hybridizations. Each individual selected for the mapping population survived peroxide as well as or better than the high-survival parent YJK1272. Due to the potentially large number of loci segregating in the S344 · YJK1272 cross, a backcross strategy was implemented. Two independent backcrosses were set up to reduce the number of segregating loci while aiming to maintain most of the genetic variation and ensuring that at least one line yielded sufﬁcient numbers of segregants for selective genotyping. The BC lines were established by backcrossing two highsurvival F1 segregants to the low-survival S344 parent. As expected, peroxide survival declines in the backcross hybrids (Figure S1B) that were subsequently sporulated, giving rise to BCI and BCII. Survival of 199 BCI and 197 BCII randomly chosen segregants was measured using the quantitative CFU assay. The phenotypic variance and frequencies of highsurvival segregants declined in each backcross line with respect to the F1 (Figure S1B). While 1 in 4 F1 segregants displayed high peroxide survival, only 1 in 12 and 1 in 22 did so in BCI and BCII. A total of 1246 spores were tested for peroxide survival using a high-throughput H2O2 plate assay. The high-survival phenotype of 15 (BCI) and 63 (BCII) segregants was veriﬁed using the CFU assay and 8 and 11 met the cut-off survival of 0.8.
S. cerevisiae oxidative stress QTN
Figure 1 Peroxide survival is a dominant quantitative trait. (A) Survival of parental strains S344 (light blue) and YJK1272 (dark blue) and their hybrid YJS (green) as determined in the CFU assay with 20 mM TBHP. The bar graph represents the mean of three experimental replicates per strain and its standard error. (B) The histogram shows CFU data obtained for 200 F1 segregants that are clustered by survival in bins of size 0.1. Parental means (boldface vertical line) and standard errors (horizontal lines) are indicated above the histogram. (C) Oxidative stress candidate gene polymorphism distribution and genotypes. Listed are the synonymous (Ks) and nonsynonymous (Ka) changes as well as the number of polymorphisms 1000 bp upstream/downstream of the respective candidate gene ORF. The presence of either parental allele of the four candidate genes CTA1, CTT1, CCP1, and SOD1 in each of the 200 F1 segregants analyzed is indicated by applying the appropriate parental color—light blue or dark blue—to each cell. Each row represents data for one gene and each column shows the genotype of each segregant. Segregants are sorted by survival from lowest (0) to highest (1), reﬂecting the phenotypic data from the histogram shown in B. Arrows below the chart point at segregants whose phenotypes are closest to the parental means. Arrows above the chart indicate the two F1 segregants that were crossed to S344 to generate backcross lines BCI and BCII. Gaps mark missing genotype data.
The mapping population was derived from two independent backcross lines (BCI and BCII) and the F1 (Figure S1A). In total, 47 F1 and 22 BCI high-survival segregants were included in the microarray-assisted QTL mapping and 16 F1 and 20 BCI low-survival segregants for linkage analysis and QTL ﬁne mapping. An unconventional combination of BSA and SSA mapping strategies yields seven candidate QTL
Our approach to dissecting the genetic basis of peroxide survival was to (i) identify at least one QTL derived from the high-survival parent YJK1272, (ii) conﬁrm genetic linkage, and (iii) characterize a gene or even nucleotide that affects the phenotype by applying an economic mapping approach that combined genotyping data obtained from the more costly Affymetrix S98 array with the inexpensive Combimatrix 4x2k array that not only contains four arrays per chip but can be stripped and reprobed twice. Candidate QTL were mapped via SSA to the 4x2k array and BSA to the S98 array. The high-survival mapping population comprised 47 F1 spores and 22 backcross segregants. Hybridization signals were normalized for melting temperature and values of the segregants compared to those of the parental strains S344 and YJK1272. Probes were considered informative when their signal resembled that of YJK1272 and further analyzed as single feature polymorphisms (SFPs). The S98 array used for BSA hybridizations contained 10,160 probes that matched S344 perfectly and had one base pair mismatch in YJM789 and 1075 probes that had two mismatches. A total of 478 probes had to be excluded due to sequencing errors in the S288c (S344) genome. The
S. Diezmann and F. S. Dietrich
remaining 9776 one-mismatch (analysis A) and 981 twomismatch (analysis B) probes were considered SFPs and included in the analysis. As expected, spacing between SFPs differed between the one-mismatch analysis A (0.4 cM) and the two-mismatch analysis B (4 cM). Note, that due to the design of the S98 array, which was originally generated for gene expression proﬁling, markers are not evenly spaced across the genome. This situation was remedied by custom designing a Combimatrix 4x2k array speciﬁcally for the genotyping of segregants from a cross between S344 and YJK1272. The Combimatrix 4x2k array designed for SSA contains a total of 1127 oligonucleotide pairs, one oligonucleotide from each parental genome. A total of 37 oligonucleotide pairs had to be excluded from the analyses due to sequencing errors, yielding a total of 1090 SFPs. These were on average spaced 3.5 cM apart and covered about 11.5 Mb of the genome. Genomic regions that were not incorporated, by design, included the rDNA array, the telomeres, and other repetitive sequences. From the pool of 1090 SFPs, blocks of consecutive SFPs derived from YJK1272 were extracted for analyses C and D. Blocks of 4 out of 5 YJK1272-derived SFPs (analysis C) and $3 consecutive YJK1272-derived SFPs (analysis D) were recorded and analyzed. Candidate QTL based on SFPs from BSA hybridizations to the S98 (analyses A and B) array and SSA hybridizations from the 4x2k Combimatrix array (analyses C and D) were mapped onto the S288c reference genome and compared with each other. Both BSA analyses A and B and SSA analysis C yielded similar numbers of regions in the genome that were derived from the high-survival parent YJK1272 (21, 18, and 19) (Figure 2 A–C). In SSA analysis D, 10 YJK1272-derived loci were identiﬁed (Figure 2D). Given
Figure 2 Peroxide survival candidate QTL as identiﬁed by BSA and SSA. Shown is a combination of all four S98 BSA and 4x2k SSA genotyping analyses. (A) S98 one-mismatch SFPs, (B) S98 two-mismatch SFPs, (C) 4x2k SFPs with gaps, (D) 4x2k $3 consecutive SFPs. The black horizontal line at the bottom is the S. cerevisiae genome with its chromosomes indicated by roman numerals between the beginning (left) and end (right) of each chromosome (short vertical bars). Chromosome lengths are drawn to scale. Candidate QTL are marked by short horizontal bars. Red bars represent loci that were found in the most conservative analysis (B) and at least one other analysis. These regions are highlighted by red vertical lines and their approximate position on the chromosome is indicated at the bottom. Black arrows point to candidate QTL that have been tested with diagnostic RFLPs (Table 2).
the stringency of BSA analysis B (two base pair mismatches SFPs), QTL were considered high quality when they were identiﬁed in analysis B and one or more other analyses (Table 1). Seven candidate QTL matched this criterion. Those QTL are located on seven chromosomes and differ in length from 25 cM to 137 cM (Mortimer et al. 1989). None of them included a ROS candidate gene from the primary gene screen (Table 1). To assess the reliability of BSA and SSA and detect differences in their ability to identify QTL, ﬁve of the seven candidate QTL were tested with diagnostic RFLPs (Figure 2, Table 2). These ﬁve candidate QTL were identiﬁed in different combinations of analyses. Two were identiﬁed only in B (Chr XI) or C (Chr IX), one was identiﬁed in A and C (Chr X), one in A and B (Chr X), and one in A, B, and C (Chr XVI). RFLPs were scored for 22 high-survival and 20 low-survival BCI segregants. About 50% of the high- and low-survival segregants inherited their alleles from YJK1272 for the QTL on chromosomes IX, X, and XI (Table 2). The genotype data for the QTL on chromosome XVI exhibited a different pattern. A total of 77% of the high-survival segregants inherited this genomic region from YJK1272, while only 21% of the low-survival segregants inherited the YJK1272
allele, resulting in a signiﬁcant chi-square test (x2 ¼ 144, P , 0.0001). In summary, our economic mapping approach identiﬁed linkage to one QTL by using a total of three Affymetrix S98 arrays and three 4x2k Combimatrix arrays. The allele distribution at the Chr XVI locus was furthermore determined in high- and low-survival segregants in the F1 and BCII mapping populations. A total of 60% of the high-survival F1 segregants (x2 ¼ 4, P , 0.05) carried a YJK1272 allele as opposed to 19% of the low-survival segregants (x2 ¼ 38, P , 0.0001) in this line. A total of 66% of the high-survival BCII spores (x2 ¼ 90, P , 0.0001) had the YJK1272 allele but only 3% of the low-survival segregants (x2 ¼ 26, P , 0.0001) tested positive for the YJK1272 allele. Fine mapping of the QTL on Chr XVI in combination with a secondary candidate screen reveals a gene previously unlinked to the oxidative stress response
Following the positive linkage analysis, we carried out ﬁner scale mapping and identiﬁed a 250-kb (86 cM) interval on chromosome XVI (Mortimer et al. 1989). Recombination mapping using RFLPs and PCR was applied to the region, reducing it to 8 cM. The mapping population comprised 41 high-survival and 16 low-survival F1 segregants and 22
Table 1 Physical locations, sizes, and map lengths of seven putative QTL identiﬁed in different analyses Chromosome II VII XI XII XIII XIV XVI IV (CTA1) VII (CTT1) X (SOD1) XI (CCP1)
Physical size (kb)
Map length (cM)
368,678–474,302 769,508–1,071,337 72,059–285,440 77,543–277,809 195,745–598,203 403,929–634,658 220,020–291,835 969,677–968,130 654,638–656,326 623,003–622,539 566,840–565,755
105.6 301.9 213.4 200.3 402.5 230.7 71.8
35.9 102.6 78.7 68.1 136.8 78.4 24.4
B and D A–C A–C A–D A–D A–D A–C
Physical sizes and map lengths of putative QTL that were identiﬁed in analysis B and at least one more are summarized. For comparison, the chromosomal and physical locations of ROS detoxifying candidate genes are indicated.
S. cerevisiae oxidative stress QTN
Table 2 PCR primers, diagnostic RFLPs, and Y allele distribution in QTL linkage analysis between candidate QTL in BCI high- and low-survival segregants Chr
Primer sequence (59–39)
ATAGCGAGCCACCAACACCA TACTCCTTGGAATCACTGGA ATGCCTGTGAGGTGTATGACC GATTTCCACTGCATTACCCA ACCCAGCAGAAAGCAAACAA GGACGTGAAGAACAAAAGGAA GGTAATGATGGTGATTTTAGAT CGGTAGTTTGCTTTATGGATTCA ATGCTGGAGAGAAACCCAAA AAGAAAGAAATCCGTCCTATGA
chrIX_NotI_F chrIX_NotI_R chrX_AﬂII_F chrX_AﬂII_F chrX_DrdI_F chrX_DrdI_R chrXI_NsiI_F chrXI_NsiI_R chrXVI_EcoNI_F chrXVI_EcoNI_R
Fragment sizea S: 906 Y: 626, 280 S: 826 Y: 245, 581 S: 530 Y: 153, 377 S: 275, 570 Y: 845 S: 852 Y: 338, 514
High (% Y)b
Low (% Y)c
a Fragment size in S344 (S) and YJK1272 (Y). b Percentage of high-survival BCI segregants that carry a Y allele at the respective locus. c
Percentage of low-survival BCI segregants that carry a Y allele at the respective locus.
high- and 20 low-survival segregants from the BCI line (Table S2). Genotyping by allele-speciﬁc PCR was unambiguous and restriction fragments of segregants resembled that of either parental control. Seven recombination events in the BCI high-survival mapping population and two in the BCI low-survival mapping population were detected. Two recombination events in the high-survival group at markers ChrXVI_300 and ChrXVI_322 narrowed the region of interest to 23.2 kb (7.9 cM). The region identiﬁed by recombination mapping contains 14 genes (Figure 3A). In a secondary candidate gene screen, all genes in the QTL were compared for DNA sequence similarity between the parental strains and gene functions related to stress response. Unlike the ﬁrst candidate gene screen that exclusively included genes functioning in the oxidative stress response, this screen considered genes that are part of a locus that had been shown to be genetically linked to peroxide survival. This provided a solid rationale for ﬁnding the quantitative trait gene. Four genes are identical between the parental strains, four contain only synonymous polymorphisms, and six contain nonsynonymous polymorphisms in addition to synonymous polymorphisms (Figure 3A). The strongest candidate based on sequence and function was RDS2 (regulator of drug sensitivity), which is a zinc cluster transcriptional activator involved in resistance to ketoconazole (Akache and Turcotte 2002) and the null mutant is calcoﬂuor white sensitive (Soontorngun et al. 2007). A nonsynonymous polymorphism in RDS2 signiﬁcantly affects peroxide survival in YJK1272
The impact of RDS2 on peroxide survival was investigated by applying a combination of analytical and experimental methods, such as homology searches against sequence archives, gene deletions, and allele replacements via sitedirected mutagenesis. The S. cerevisiae RDS2 ORF is 1341 nucleotides long and contains four synonymous SNPs and two nonsynonymous SNPs (Figure 3B). The ﬁrst nonsynonymous SNP (A505G) results in a substitution from threonine (T) in S344 to ala-
S. Diezmann and F. S. Dietrich
nine (A) in YJK1272 (T169A), both of which have neutral side chain charges. The second SNP (C751G) leads to the replacement of a histidine (H) in S344 with an aspartic acid (D) in YJK1272 (H251D). Given the nonconservative nature of H251D—histidine carries a neutral side chain charge and aspartic acid a negative one—we designed subsequent experiments to verify this SNP as the peroxide survival QTN. BLAST searches of both RDS2 alleles against the S. cerevisiae sequences of Liti et al. (2009) yielded 30 strains with sequence data for A505G and C751G. Twelve strains have either the S344 allele or the YJK1272 allele. This indicates that both SNPs cosegregate in the S. cerevisiae population. Six other strains have the S344-derived A on position 505 and a YJK1272-derived G on position 751 (Figure 3C). The absence of the fourth possible combination (G at 505 and C at 751) suggests lack of recombination between the two sites. On the basis of these data, a subset of six strains with the S344 allele and two with the YJK1272 allele were selected (Figure 3C) and treated with TBHP to measure their survival phenotype (Figure 3D). Interestingly, strains varied in peroxide survival depending on their RDS2 conﬁguration (P , 0.001, Figure 3D). RM11-1A was the only S344-like strain that survived TBHP treatment better than S344 or any other S344-like strain. The RM11-A phenotype resembled that of YJK1272. The other ﬁve S344-like strains were statistically indistinguishable from S344 and demonstrated less survival than YJK1272 (P , 0.001). The two YJK1272-like strains exhibited intermediate phenotypes. Next, we extended our search from within S. cerevisiae to 16 additional fungal species. Given the large evolutionary distances covered with this kind of taxa sampling, the translated ORF instead of the nucleotide sequence of RDS2 was searched against the NCBI protein database. Rds2 sequences from 12 ascomycetous and 4 basidiomycetous fungi were retrieved and the alignment screened for conserved regions that would offer insights as to which part of the molecule, including which of the two nonsynonymous SNPs, is of functional importance (Figure 4). Rds2 is highly conserved in most
Figure 3 The candidate QTL on chromosome XVI and RDS2 allele distribution and its contribution to peroxide survival in wild strains. (A) A brief summary of the arrangement and function of the 14 genes that form the peroxide survival QTL on the left arm of chromosome XVI. Genes are color coded depending on the identiﬁed polymorphic differences between S344 and YJK1272. (B) Schematic of the 1341-bp long RDS2 ORF in S344 and YJK1272. Synonymous and nonsynonymous polymorphisms are indicated and their exact position within the gene is given below each SNP. Polymorphic changes are color coded for easier identiﬁcation. (C) Results from a search of RDS2 sequences as submitted by the Sanger Institute. Twelve strains are identical with S344 on positions 505 and 751, the sites of the nonsynonymous changes, and 12 are identical with YJK1272. Six strains share the A with S344 on position 505 and the G from YJK1272 on position 751. Strains in boldface type were subjected to the CFU assay and their survival measured and compared to S344 and YJK1272. (D) The peroxide survival phenotypes of these eight strains, that are identical with either S344 or YJK1272 with respect to positions 505 and 751.
Saccharomycetales, C. cinerea and M. grisea with blast alignment scores .200 over almost the entire sequence length. The C terminus displays a high degree of conservation across all species, strongly indicating an important role in the molecule’s function.
C751G, which translates into the residue H251D, is part of this highly conserved region. A most striking pattern emerged at H251D in comparison with the other fungal species. The three Saccaromycetaceae species share the aspartic acid residue with YJK1272 and every
S. cerevisiae oxidative stress QTN
Figure 4 Evolutionary conservation of the Rds2 C terminus and the acidic amino acid residue on position 251. Rds2 blastp results for 16 different fungal species, 12 ascomycetes and 4 basidiomycetes. The Smith Waterman alignment emphasizes the high degree of evolutionary conservation around amino acid position 251. All species, except C. neoformans, have either an aspartic acid (D) or glutamic acid (E) residue on that position.
other fungal species, except for C. neoformans, has a glutamic acid residue (E), which also has a negatively charged side chain. The evolutionary conservation of an amino acid with a negatively charged side chain in the C-terminal region of Rds2 strongly suggests this residue to be of functional importance, adding further incentive to investigate C751G. To verify the effect of RDS2 and speciﬁcally C751G on peroxide survival, we deleted the entire ORF and conducted site-directed mutagenesis of C751G in S344 and YJK1272 and looked for changes in peroxide survival in the transformants. Homozygous deletion of RDS2 resulted in slightly reduced peroxide survival of YJK1272 (strains YSD87 and YSD88) (P ¼ 0.0810) but had no effect on S344 (strains YSD85 and YSD86) (Table S1, Figure 5A). Both YJK1272 deletion strains resembled the phenotype of the hybrid YJS, which survived less well than YJK1272 in 16 mM TBHP (data not shown). The reduced survival of YJK1272 deletion mutants but not S344 deletion mutants, indicated that RDS2 is the gene underlying this QTL and that its downstream targets or interactions differ between S344 and YJK1272. Only nucleotides C and G at position 751 were exchanged between S344 and YJK1272. Two independent transformants in each background (YSD26, YSD30, YSD58, and YSD70) (Table S1) were compared with their respective parental strains. Introducing the YJK1272-derived G into the S344 background did not alter the survival phenotype (Figure 5B). Both diploidized transformants were indistinguishable from S344. Yet, introducing the S344-derived C into YJK1272 lead to signiﬁcantly reduced survival in TBHP (P , 0.05). Two independent transformants were tested, and both exhibited reduced survival when compared to YJK1272 (Figure 5B). Furthermore, testing homozygous YJK1272 and S344 backgrounds that were heterozygous for C751G (YSD62, YSD66, YSD55, and YSD72) demonstrated decreased survival in the YJK1272-derived strains (Figure 5B). A homozygous YJK1272 background that is
S. Diezmann and F. S. Dietrich
heterozygous for the C751G site survived treatment better than the homozygous S344 background that is heterozygous (P , 0.001) but less well than the wild-type YJK1272 (P , 0.05). This result shows that, although the phenotype appears to be dominant, the causative QTN in RDS2 is recessive. Testing hybrid backgrounds that were homozygous for either polymorphism conﬁrmed the ﬁndings from the allele replacement experiments (Figure 5C). Hybrids homozygous for the YJK1272-derived G survived treatment better than hybrids homozygous for the S344-derived C (P , 0.01, P , 0.001) or the hybrid YJS itself (P , 0.05). Hybrids homozygous for the S344-derived C allele were statistically indistinguishable from YJS. On the basis of survival data of YJK1272 that is homozygous for the S344-derived C allele, it can be concluded that the S344-derived allele decreased survival in the YJK1272 background by 15%. Combined evidence from survival data of homozygous deletion and allele replacement strains demonstrated that C751G is a QTN affecting peroxide survival.
Discussion We identiﬁed a QTN that affects survival of oxidative stress, a trait predominantly associated with clinical S. cerevisiae isolates (Diezmann and Dietrich 2009). A synthesis of selective genotyping methods together with classical genetic approaches and a candidate gene screen implementing evolutionary conservation and gene function proved to be successful in identifying a novel transcription factor that affects survival of oxidative stress. RDS2 encodes a zinc cluster protein of the fungal-speciﬁc GAL4 family (Akache and Turcotte 2002). Zinc cluster proteins are intimately involved in the development of pleiotropic drug resistance (PDR), which is deﬁned as the rapid acquisition of resistance to multiple drugs (Kane et al.
Figure 5 C751G is the RDS2 QTN that affects peroxide survival in YJK1272. (A) Homozygous deletion of RDS2 in two independent transformants in the S344 (YSD85 and YSD86) and YJK1272 (YSD87 and YSD88) backgrounds only affects survival in YJK1272, which exhibits decreased survival in 16 mM TBHP. The bar graph presents the mean of three replicates and its standard error (SEM) survival. Wild-type strains (S344 and YJK1272) and their respective two deletion mutants (rds2D_1 and rds2D_2) are color coded as before. The strain numbers as referenced in Table S1 are indicated below each chart. Statistical signiﬁcance was assessed using one-way ANOVA and the nonparametric Dunnetts’ multiple comparison test, which compares the group of the transfomant to its respective parental reference. (B) The nonsynonymous SNP on position 751 was exchanged between S344 and YJK1272. Plotted are the means and SEMs for six (S344) and eight (YJK1272) biological replicates done on two independent transformants. Strains are color coded by background and for clarity the nucleotide combination on position 751 is given below the bars. *, allele replacement. Statistical signiﬁcance was assessed using one-way ANOVA and the nonparametric Dunnetts’ multiple comparison test and the Wilcoxon signed rank test. (C) Testing survival of homozygous SNP conﬁgurations in the hybrid background. Shown is the mean of six biological replicates and SEMs in strains derived from two independent transformations. The combination of nucleotides on position 751 is indicated below the bars. Statistical differences were assessed by comparing transformants with each other using the Wilcoxon signed rank test.
1990). S. cerevisiae PDR genes encode transcriptional activators, such as Pdr1 and Pdr3 (Moye-Rowley 2003), that initiate transcription of genes whose products have detoxifying function, such as drug transporters. Since neither deletion of RDS2 nor allele replacement affected survival of S344, it can be speculated that those downstream targets diverged between S344 and YJK1272. It is also possible that the YJK1272-derived allele can no longer interact with targets in S344. Yet, these might be the genes whose products are required for stress response. Previous work showed that the rds2 deletion mutant is sensitive to the cell wall stressor calcoﬂuor white and the antifungal drug ketoconazole but resistant to osmotic stress and unable to grow on the nonfermentable carbon sources (Akache et al. 2001; Akache and Turcotte 2002; Moreno et al. 2008). The null mutant’s cell wall composition changed depending on the growth stage, suggesting Rds2 regulates drug sensitivity by altering cell surface permeability (Moreno et al. 2008). This ﬁnding offers an explanation why YJK1272 and S344 survived peroxide exposure equally well in stationary phase but exhibited different survival phenotypes in exponential phase (Diezmann 2009). Rds2 binds to the promoters of metabolic genes when cells are grown in glucose (Soontorngun et al. 2007). Upon shifting cells to the nonfermentable carbon source ethanol, however, Rds2 is phosphorylated by Snf1 and activates transcription of genes involved in gluconeogenesis, the tricarboxylic acid cycle, and glucose metabolism (Soontorngun et al. 2007). Although the phosphorylation site(s) remain(s)
unknown, previous work on Skn7, in which an aspartic acid residue is phosphorylated by Ypd1, indicates a critical role for the aspartic acid residue Rds2. H251D is an evolutionarily conserved acidic amino acid residue, suggesting that the histidine in S344 is an evolved character state, while aspartic acid is the ancestral version. The presence of both alleles in the S. cerevisiae population implies that these alleles function differently, rather than that one allele is defective. The basis for the successful mapping of RDS2 was provided by combining SSA with BSA selective genotyping of high-survival segregants. SSA via custom-designed Combimatrix 4x2k arrays provides an economic application that facilitates genotyping of segregants of any cross between strains whose genomes have been sequenced. This method will be increasingly applicable as additional S. cerevisiae whole genome sequences become available. The ecological and genetic diversity captured by this massive sequencing effort, together with ongoing sequencing projects at other institutions, is unsurpassed in eukaryote genomics. More than 1000 potential crosses between those genome strains can be exploited for the study of diverse phenotypes that are important to yeast biology or for yeast as a model system for human diseases. Candidate QTL, other than the one on chromosome XVI, were screened for genes, whose functions would suggest an involvement in the defense against reactive oxygen species. One candidate QTL contains MKT1 and END2 (Chr XIV) and another HSP104 (Chr XII). QTNs in the ﬁrst two genes have been shown to affect high-temperature growth
S. cerevisiae oxidative stress QTN
and sporulation efﬁciency and HSP104 is a chaperone that refolds proteins in response to heat, ethanol, and sodium arsenite (Sanchez et al. 1992; Steinmetz et al. 2002; Deutschbauer and Davis 2005). These ﬁndings suggest that S. cerevisiae responds to multiple stressors with a select set of proteins and emphasizes that the oxidative stress response may be more complex than previously appreciated. Transgression in F1 segregants suggests segregation of multiple modiﬁers in the S344 and YJK1272 parental genomes. The presence of segregant phenotypes exceeding parental phenotypes indicates either nonadditive gene action, i.e., epistasis and/or overdominance or the presence of particular alleles with opposing effects in the parental lines (Lynch and Walsh 1998). Epistatic interactions between QTNs have been identiﬁed in high-temperature growth (Sinha et al. 2008a) and sporulation efﬁciency (Deutschbauer and Davis 2005). We compared strains that are homozygous for either background but heterozygous for the C751G polymorphism. The YJK1272 background exhibits reduced survival when heterozygous for the QTN, strongly suggesting epistasis between the S344-derived C allele and an as of yet undetermined locus in YJK1272. This ﬁnding is supported by survival data from hybrids homozygous for either allele. A hybrid homozygous for the S344 allele resembles the heterozygous hybrid but a hybrid homozygous for the YJK1272 allele exhibited increased survival. This study provides genetic evidence for a novel function of RDS2 in the oxidative stress response. The quantitative trait nucleotide C751G inﬂuences the oxidative stress response in an as of yet unknown mechanism or pathway. Since Rds2 is a fungal-speciﬁc transcription factor that regulates cell wall architecture and pleiotropic drug resistance it could provide a promising drug target. Two-component and phosphorelay signaling systems appear to be absent from mammals and are consequently considered potential drug targets in other fungi and the cell wall synthesis pathway has proven to be a powerful drug target that is being exploited successfully (Hoch 2000; Chauhan and Calderone 2008; Cowen and Steinbach 2008). The identiﬁcation of a QTN affecting a virulence-related trait furthers our understanding of yeast biology, the evolution of fungal virulence, and how eukaryotic cells cope with imbalances in their redox homeostasis.
Acknowledgments S.D. thanks John McCusker, Joanne Kingsbury, and Ludo Müller for advice on QTL mapping and technical assistance. The authors are grateful to Heath O’Brien, Tim James, James Fraser, and Leah Cowen for comments on the manuscript.
Literature Cited Abshire, K. Z., and F. C. Neidhardt, 1993 Analysis of proteins synthesized by Salmonella thyphimurium during growth within a host macrophage. J. Bacteriol. 175: 3734–3743.
S. Diezmann and F. S. Dietrich
Akache, B., and B. Turcotte, 2002 New regulators of drug sensitivity in the family of yeast zinc cluster proteins. J. Biol. Chem. 277: 21254–21260. Akache, B., K. Wu, and B. Turcotte, 2001 Phenotypic analysis of genes encoding yeast zinc cluster proteins. Nucleic Acids Res. 29: 2181–2190. Altschul, S. F., W. Gish, W. G. Miller, E. W. Myers, and D. J. Lipman, 1990 Basic local alignment search tool. J. Mol. Biol. 215: 403– 410. Arnheim, N., C. Strange, and H. Erlich, 1985 Use of pooled DNA samples to detect linkage disequilibrium of polymorphic restriction fragments and human disease: studies of the HLA class II loci. Proc. Natl. Acad. Sci. USA 82: 6970–6974. Ben-Ari, G., D. Zenvirth, A. Sherman, L. David, M. Klutstein et al., 2006 Four linked genes participate in controlling sporulation efﬁciency in budding yeast. PLoS Genet. 2: e195. Brauer, M. J., C. M. Christianson, D. A. Pai, and M. J. Dunham, 2006 Mapping novel traits by array-assisted bulk segregant analysis in Saccharomyces cerevisiae. Genetics 173: 1813–1816. Brem, R. B., G. Yvert, R. Clinton, and L. Kruglyak, 2002 Genetic dissection of transcriptional regulation in budding yeast. Science 296: 752–755. Burke, D., D. Dawson, and T. Stearns, 2000 Methods in Yeast Genetics. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Calera, J. A., S. Paris, M. Monod, A. J. Hamilton, J.-P. Debeaupuis et al., 1997 Cloning and disruption of the antigenic catalase gene of Aspergillus fumigatus. Infect. Immun. 65: 4718–4724. Causton, H. C., B. Ren, S. S. Koh, C. T. Harbison, E. Kanin et al., 2001 Remodeling of yeast genome expression in response to environmental changes. Mol. Biol. Cell 12: 323–337. Chauhan, N., and R. Calderone, 2008 Two-component signal transduction proteins as potential drug targets in medically important fungi. Infect. Immun. 76: 4795–4803. Cimolai, N., M. J. Gill, and D. Church, 1987 Saccharomyces cerevisiae fungemia: case report and review of the literature. Diagn. Micr. Infec. Dis. 8: 113–117. Codon, A. C., J. M. Gasent-Ramirez, and T. Benitez, 1995 Factors which affect the frequency of sporulation and tetrad formation in Saccharomyces cerevisiae baker’s yeast. Appl. Environ. Microb. 61: 630–638. Cowen, L. E., and W. J. Steinbach, 2008 Stress, drugs, and evolution: the role of cellular signaling in fungal drug resistance. Eukaryot. Cell 7: 747–764. Delaunay, A., A.-D. Isnard, and M. B. Toledano, 2000 H2O2 sensing through oxidation of the Yap1 transcription factor. EMBO J. 19: 5157–5166. Demogines, A., E. Smitch, L. Kruglyak, and E. Alani, 2008 Identiﬁcation and dissection of a complex DNA repair sensitivity phenotype in baker's yeast. PLoS Genet. 4: e1000123. Deutschbauer, A. M., and R. W. Davis, 2005 Quantitative trait loci mapped to single nucleotide resolution in yeast. Nat. Genet. 37: 1333–1340. Diezmann, S., 2009 Evolutionary implications and genetic basis of peroxide survival in Saccharomyces cerevisiae, pp. 214 in University Program in Genetics and Genomics. Duke University, Durham, NC. Diezmann, S., C. J. Cox, G. Schönian, R. J. Vilgalys, and T. G. Mitchell, 2004 Phylogeny and evolution of medical species of Candida and related taxa: a multigenic analysis. J. Clin. Microbiol. 42: 5624–5635. Diezmann, S., and F. S. Dietrich, 2009 Saccharomyces cerevisiae: population divergence and resistance to oxidative stress in clinical, domesticated and wild isolates. PLoS ONE 4: e5317. Enache-Angoulvant, A., and C. Hennequin, 2005 Invasive Saccharomyces infection: a comprehensive review. Clin. Inf. Dis. 41: 1559–1568.
Gardes, M., and T. D. Bruns, 1993 ITS primers with enhanced speciﬁcity for basidiomycetes - application to the identiﬁcation of mycorrhizae and rusts. Mol. Ecol. 2: 113–118. Geldermann, H., 1975 Investigations on inheritance of quantitative characters in animals by gene markers. I. Methods. Theor. Appl. Genet. 46: 319–330. Gietz, R. D., and R. H. Schiestl, 1991 Applications of high efﬁciency lithium acetate transformation of intact yeast cells using single-stranded nucleic acids as carrier. Yeast 7: 253–263. Giles, S. S., J. E. Stajich, C. Nichols, Q. D. Gerrald, J. A. Alspaugh et al., 2006 The Cryptococcus neoformans catalase gene family and its role in antioxidant defense. Eukaryot. Cell 5: 1447– 1459. Goffeau, A., B. G. Barrell, H. Bussey, R. W. Davis, B. Dujon et al., 1996 Life with 6000 genes. Science 274: 546–567. Goldstein, A. L., and J. H. McCusker, 1999 Three new dominant drug resistance cassettes for gene disruption in Saccharomyces cerevisiae. Yeast 15: 1541–1553. Halliwell, B., and J. M. C. Gutteridge, 2007 Free Radicals in Biology and Medicine. Oxford University Press, Oxford. Hazen, K. C., 1995 New and emerging yeast pathogens. Clin. Microbiol. Rev. 8: 462–478. Hoch, J. A., 2000 Two-component and phosphorelay signal transduction. Curr. Opin. Microbiol. 3: 165–170. Hoffman, C. S., and F. Winston, 1987 A ten-minute DNA preparation from yeast efﬁciently releases autonomous plasmids for transformation of Escherichia coli. Gene 57: 267–272. Hu, X. H., M. H. Wang, T. Tan, R. Li, H. Yang et al., 2007 Genetic dissection of ethanol tolerance in the budding yeast Saccharomyces cerevisiae. Genetics 175: 1479–1487. Ito-Harashima, S., and J. H. McCusker, 2004 Positive and negative selection LYS5MX gene replacement cassettes for use in Saccharomyces cerevisiae. Yeast 21: 53–61. Iyer, G. Y. N., M. F. Islam, and J. H. Quastel, 1961 Biochemical aspects of phagocytosis. Nature 192: 535–541. Kammenga, J. E., A. Doroszuk, J. A. G. Riksen, E. Hazendonk, L. Spiridon et al., 2007 A Caenorhabditis elegans wild type deﬁes the temperature-size rule owing to a single nucleotide polymorphism in tra-3. PLoS Genet. 3: e34. Kane, S. E., I. Pastan, and M. M. Gottesman, 1990 Genetic basis of multidrug resistance of tumor cells. J. Bioenerg. Biomembr. 22: 593–618. Krems, B., C. Charizanis, and K.-D. Entian, 1996 The response regulator-like protein Pos9/Skn7 of Saccharomyces cerevisiae is involved in oxidative stress resistance. Curr. Genet. 29: 327– 334. Kusch, H., S. Engelmann, D. Albrecht, J. Morschhäuser, and M. Hecker, 2007 Proteomic analysis of the oxidative stress response in Candida albicans. Proteomics 7: 686–697. Lamarre, C., O. Ibrahim-Granet, C. Du, R. Calderone, and J.-P. Latgé, 2007 Characterization of the SKN7 ortholog of Aspergillus fumigatus. Fungal Genet. Biol. 44: 682–690. Lander, E. S., and D. Botstein, 1989 Mapping mendelian factors underlying quantitative traits using RFLP linkage maps. Genetics 121: 185–199. Liti, G., D. M. Carter, A. M. Moses, J. Warringer, L. Parts et al., 2009 Population genomics of domestic and wild yeasts. Nature 458: 337–341. Lushchak, V. I., and D. V. Gospodaryov, 2005 Catalases protect cellular proteins from oxidative modiﬁcation in Saccharomyces cerevisiae. Cell Biol. Int. 29: 187–192. Lynch, M., and B. Walsh, 1998 Genetics and Analysis of Quantitative Traits. Sinauer Associates, Sunderland, MA. Marchler, G., C. Schüller, G. Adam, and H. Ruis, 1993 A Saccharomyces cerevisiae UAS element controlled by protein kinase A activates transcription in resopnse to a variety of stress conditions. EMBO J. 12: 1997–2003.
McCusker, J. H., 2006 Saccharomyces cerevisiae: an emerging and model pathogenic fungus. In: Molecular Principles of Fungal Pathogenesis, edited by Heitman, J., S. G. Filler, J. E. Edwards Jr., and A. P. Mitchell, pp. 245–260. ASM Press, Washington, DC. McNeil, M. M., S. L. Nash, R. A. Hajjeh, M. A. Phelan, L. A. Conn et al., 2001 Trends in mortality due to invasive mycotic diseases in the United States, 1980–1997. Clin. Inf. Dis. 33: 641– 647. 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. Proc. Natl. Acad. Sci. USA 88: 9828–9832. Moreno, I., N. Tutrone, R. Sentandreu, and E. Valentin, 2008 Sacharomyces cerevisiae Rds2 transcription factor involvement in cell wall composition and architecture. Int. Microbiol. 11: 57–63. Morrison, L. J., A. Tait, S. McLellan, L. Sweeney, C. M. Turner et al., 2009 A major genetic locus in Trypanosoma brucei is a determinant of host pathology. PLoS Negl. Trop. Dis. 3: e557. Mortimer, R. K., D. Schild, C. R. Contopoulou, and J. A. Kans, 1989 Genetic map of Saccharomyces cerevisiae, edition 10. Yeast 5: 321–403. Moye-Rowley, W. S., 2003 Transcriptional control of multidrug resistance in the yeast Saccharomyces. Prog. Nucleic Acid Res. 73: 251–279. Paris, S., D. Wysong, J.-P. Debeaupuis, K. Shibuya, B. Philippe et al., 2003 Catalases of Aspergillus fumigatus. Infect. Immun. 71: 3551–3562. Perlstein, E. O., D. M. Ruderfer, G. Ramachandran, S. J. Haggarty, L. Kruglyak et al., 2006 Revealing complex traits with small molecules and naturally recombinant yeast strains. Chem. Biol. 13: 319–327. Quarrie, S., V. Lazic-Jancic, D. Kovacevic, A. Steed, and S. Pekic, 1999 Bulk segregant analysis with molecular markers and its use for improving drought resistance in maize. J. Exp. Bot. 50: 1299–1306. Rea, R. B., C. G. Gahan, and C. Hill, 2004 Disruption of putative regulatory loci in Listeria monocytogenes demonstrates a signiﬁcant role for Fur and PerR in virulence. Infect. Immun. 72: 717– 727. Riboulet, E., N. Verneuil, S. La Carbona, N. Sauvageot, Y. Auffray et al., 2007 Relationships between oxidative stress response and virulence in Enterococcus faecalis. J. Mol. Microb. Biotech. 13: 140–146. Sanchez, Y., J. Taulien, and S. Lindquist, 1992 Hsp104 is required for tolerance to many forms of stress. EMBO J. 11: 2357–2364. Sbarra, A. J., and M. L. Karnovsky, 1959 The biochemical basis of phagocytosis. I. metabolic changes during the ingestion of particles by polymorphnuclear leukocytes. J. Biol. Chem. 234: 1355–1362. Segre, A. V., A. W. Murray, and J. Leu, 2006 High-resolution mutation mapping reveals parallel experimental evolution in yeast. PLoS Biol. 4: e256. Singh, P., N. Chauhan, A. Ghosh, F. Dixon, and R. Calderone, 2004 SKN7 of Candida albicans: Mutant Construction and Phenotype Analysis. Infect. Immun. 72: 2390–2394. Sinha, H., L. David, R. C. Pascon, S. Clauder-Munster, S. Krishnakumar et al., 2008a Sequential elimination of major-effect contributors identiﬁes additional quantitative trait loci conditioning high-temperature growth in yeast. Genetics 180: 1661– 1670. Sinha, H., B. P. Nicholson, L. M. Steinmetz, and J. H. McCusker, 2008b Complex genetic interactions in a quantitative trait locus. PLoS Genet. 2: e13.
S. cerevisiae oxidative stress QTN
Skovgaard, N., 2007 New trends in emerging pathogens. Int. J. Food Microbiol. 120: 217–224. Sobel, J. D., J. Vazquez, M. Lynch, C. Meriwether, and M. J. Zervos, 1993 Vaginitis due to Saccharomyces cerevisiae: epidemiology, clinical aspects, and therapy. Clin. Inf. Dis. 16: 93–99. Soontorngun, N., M. Larochelle, S. Drouin, F. Robert, and B. Turcotte, 2007 Regulation of gluconeogenesis in Saccharomyces cerevisiae is mediated by activator and repressor functions of Rds2. Mol. Cell. Biol. 27: 7895–7905. Steinmetz, L. M., H. Sinha, D. R. Richards, J. I. Spiegelman, P. J. Oefner et al., 2002 Dissecting the architecture of a quantitative trait locus. Nature 416: 326–330. Su, C., D. K. Howe, J. P. Dubey, J. W. Ajioka, and L. D. Sibley, 2002 Identiﬁcation of quantitative trait loci controlling acute virulence in Toxoplasma gondii. Proc. Natl. Acad. Sci. USA 99: 10753–10758. Tortorano, A. M., J. Peman, H. Bernhardt, L. Klingspor, C. C. Kibbler et al., 2004 Epidemiology of candidaemia in Europe: results of 28-month European Confederation of Medical Mycology (ECMM) hospital-based surveillance study. Eur. J. Clin. Microbiol. Infect. Dis. 23: 317–322. Wach, A., A. Brachat, R. Pöhlmann, and P. Philippsen, 1994 New heterologous modules for classical or PCR-based gene disruptions in Saccharomyces cerevisiae. Yeast 10: 1793–1808.
S. Diezmann and F. S. Dietrich
Wei, W., J. H. McCusker, R. W. Hyman, T. Jones, Y. Ning et al., 2007 Genome sequencing and comparative analysis of Saccharomyces cerevisiae strain YJM789. Proc. Natl. Acad. Sci. USA 104: 12825–12830. Witten, J. T., C. T. L. Chen, and B. A. Cohen, 2007 Complex genetic changes in strains of Saccharomyces cerevisiae derived by selection in the laboratory. Genetics 177: 449–456. Wormley, F. L. Jr., G. Heinrich, J. L. Miller, J. R. Perfect, and G. M. Cox, 2005 Identiﬁcation and Characterization of an SKN7 Homologue in Cryptococcus neoformans. Infect. Immun. 73: 5022– 5030. Wysong, D. R., L. Christin, A. M. Sugar, P. W. Robbins, and R. D. Diamond, 1998 Cloning and sequencing of a Candida albicans catalase gene and effects of disruption of this gene. Infect. Immun. 66: 1953–1961. Zaoutis, T. E., J. Argon, J. Chu, J. A. Berlin, T. J. Walsh et al., 2005 The epidemiology and attributable outcomes of candidemia in adults and children hospitalized in the United States: a propensity analysis. Clin. Inf. Dis. 41: 1232–1239. Zaret, K., and F. Sherman, 1985 a-Aminoadipate as a primary nitrogen source for Saccharoyces cerevisiae mutants. J. Bacteriol. 162: 579–583. Communicating editor: J. Borevitz
GENETICS Supporting Information http://www.genetics.org/cgi/content/full/genetics.111.128256/DC1
Oxidative Stress Survival in a Clinical Saccharomyces cerevisiae Isolate Is Inﬂuenced by a Major Quantitative Trait Nucleotide Stephanie Diezmann and Fred S. Dietrich
Copyright © 2011 by the Genetics Society of America DOI: 10.1534/genetics.111.128256
! ) $&' *(() & '#)..&' ).-& ' $ )# ! % $&'
! ) +,, ! ! $ ! ) $
$" $ $
/-.)',,3F2E%)!)(/&).#'#,),,3- ,)')'#'.,#2 -,0-!().3*#(!*&. ),':#,-) )!)(/&).#-7)() 1"#"'."-)(*,(.1"#&#.-*,.(,"-E#(.,(& '#-'."-71,#(.# #3-+/(.#&)'*,#-)(-) )."*,(.&!()'-? DKKH8 [email protected]
:" IJK!()'?((%--#)((/',[email protected]
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
'/$*0#$$4-% %5/%*) 2.*1!-'''*2!-7*)!:) /2*:(%.(/$+-*!.$4-% %5! '(*./.2!''./$!5!-*:(%.(/$ +-*!9$! %""!-!)!%)$4-% %5/%*).%#)'!/2!!)/$!5!-*:(%.(/$*)/-*') /$!%)"*-(/%1!+-*!.2..(''!-/$) %) @AFA) @7-!.!('%)#/$!BCC+//!-)) -!"'!/%)#/$!&-*..)/0-!*"/$%.'%)!9 ) % /! .2!-!% !)/%"%! .! *)"*0-#!)*/4+! /.!/.;'!@++! - >++!
>++ - >++
>++! - >++
>++! - >++
>++! - >++
>++ - >++!
>++! - >++
++! - >++
++! - >++
56++ '>++ '