|Home | About | Journals | Submit | Contact Us | Français|
Conceived and designed the experiments: WG HJ LMS YL ZG. Performed the experiments: WG HJ XG LX. Analyzed the data: WG HJ LX YL ZG. Contributed reagents/materials/analysis tools: WG HJ EM LX YL ZG. Wrote the paper: WG HJ ZG.
Fungal pathogens can be lethal, especially among immunocompromised populations, such as patients with AIDS and recipients of tissue transplantation or chemotherapy. Prolonged usage of antifungal reagents can lead to drug resistance and treatment failure. Understanding mechanisms that underlie drug resistance by pathogenic microorganisms is thus vital for dealing with this emerging issue. In this study, we show that dramatic sequence changes in PDR5, an ABC (ATP-binding cassette) efflux transporter protein gene in an opportunistic fungal pathogen, caused the organism to become hypersensitive to azole, a widely used antifungal drug. Surprisingly, the same mutations conferred growth advantages to the organism on polyenes, which are also commonly used antimycotics. Our results indicate that Pdr5p might be important for ergosterol homeostasis. The observed remarkable sequence divergence in the PDR5 gene in yeast strain YJM789 may represent an interesting case of adaptive loss of gene function with significant clinical implications.
Opportunistic fungal infections are a global health threat . Widespread use of antifungal drugs in the immunocompromised population has been associated with the emergence of clinically significant drug resistance among patients who have been exposed to such antimycotics for prolonged periods , . Understanding the processes that underlie the emergence of such resistance is vital for dealing with this critical issue. Known molecular mechanisms of drug resistance in fungi include overexpression of genes that encode drug-efflux pumps belonging to the ABC (ATP-binding cassette) family of transporter proteins , overexpression or mutation of the target enzyme, and alteration of other enzymes in the same biosynthetic pathway as the target enzyme .
The budding yeast, Saccharomyces cerevisiae, widely used in baking and ethanol production for industrial usage and human consumption, in general is non-pathogenic. Strain YJM789, however, was derived from a clinical S. cerevisiae isolate (YJM128) collected from the lung of an AIDS patient , . The YJM789 strain has many phenotypes that are relevant to its pathogenicity, including high-temperature growth , pseudohyphae , and deadly virulence in mouse models , . Its capability of crossing with laboratory strains of S. cerevisiae makes it an excellent tool to study genetic systems that underlie these complex phenotypes . As fungal infections are common among immunocompromised individuals, AIDS patients are routinely treated with antifungal drugs in general clinical therapy , . Therefore, YJM789 represents an excellent tool for understanding how an organism can survive in antifungal drug environments.
Our initial determination of its genome sequence showed that the PDR5 gene is highly polymorphic in YJM789 . It has an amino acid difference of 5.3% from the lab strain, S288c, whereas at the whole genome level the difference is only 0.43%. It is particularly noteworthy that most of the amino acid differences occurred in two transmembrane domain regions (TMDRs). In yeast, PDR5 is an important ABC transporter that actively exports various xenobiotic compounds , , , , such as azole antifungal drugs. Loss-of-function mutants for the PDR5 gene in lab strains show hypersensitivity to a spectrum of antifungal drugs and overexpression of this gene product results in resistance to a variety of chemicals , , , . The mechanism by which Pdr5p recognizes so many structurally and functionally unrelated substrates remains an enigma . Previous studies indicate that the transmembrane domains of the ABC transporters may play major roles in recognizing substrates , . Because the hyper-variable regions co-localize with the transmembrane regions of Pdr5p from YJM789, we have investigated the functional consequence of these dramatic changes in this important gene.
In this study, several representative antifungal drugs were used to treat YJM789 and BY4741 (a lab strain that is isogenic with S288c) and the outcomes after drug treatment were compared. In contrast to BY4741, deletion of the PDR5 gene in YJM789 appeared to have no impact on growth in the presence of both azole and polyene, two antifungal drugs widely used in clinical practice. Interestingly, although YJM789 is hypersensitive to azoles, as expected by loss of Pdr5p-exporter function, it is hyposensitive to polyene antimycotics. Our experiments, using both a laboratory strain (BY4741) and YJM789, show that loss of the PDR5 gene indeed confers a small, but significant growth advantage to the organism in the presence of polyenes.
Pdr5p belongs to the ABC gene family, a large and important group of proteins that are conserved from bacteria to humans . As shown in Figure 1A, Pdr5p has two transmembrane domain regions (TMDRs) and two nucleotide-binding domains (NBDs). Each TMDR has six stretches of amino acids that span cell membranes and five linkers that connect the transmembrane segments. According to genomic sequences, PDR5 is a highly polymorphic ORF in YJM789 . Table 1 shows the number of amino acid changes in each part of the gene. Consistent with our previous report, TMD regions have significantly more amino acid changes than do NBD regions in Pdr5p of YMJ789 (chi-square test, P<0.0001). Interestingly, within TMDR2, the transmembrane segments also have higher percentage of amino acid changes than the linker regions (Fisher's Exact Test, P=0.04).
To see if PDR5 is uniquely different in YJM789, we compared the polymorphism and divergence of PDR5 genes among S. cerevisiae strains and other Saccharomyces species. Figure 1B and 1C show the pairwise Pdr5p amino acid differences among eight sequenced S. cerevisiae strains and S. paradoxus, S. bayanus, two species in the Saccharomyces sensu stricto complex. For the entire Pdr5p sequence, all seven S. cerevisiae strains except YJM789 shared more than 99% similarity, but the similarities of YJM789 Pdr5p to those of the other S. cerevisiae strains were less than 95% (bright green lane), which are even lower than those between S. paradoxus and other S. cerevisiae strains (>96%, Figure 1B).
Figure 1C shows pairwise differences for the transmembrane domain regions. The sequence similarities are about 100% among seven S. cerevisiae strains except YJM789, which had similarities of only ~87% to the other strains. In comparison, the average sequence similarities for S. paradoxus and S. bayanus to other S. cerevisiae strains were 96% and 94%, respectively. It is noteworthy also that, except for YJM789, the TMDRs in Pdr5p were more conserved than the rest of the gene within or between species, indicating that TMDRs are more important for gene function. Our results imply that functions of Pdr5p in YJM789 might have been dramatically changed during its evolution.
In the United States, approximately ten antifungal drugs are currently approved by the Food and Drug Administration (FDA) for the therapy of systemic fungal infections. Azoles and polyenes are two principal classes . The azoles can block the ergosterol biosynthesis pathway by inhibiting the enzyme 14-α-demethylase (ERG11), which converts lanosterol to ergosterol and is required in fungal cell membrane synthesis. The polyene antimycotics bind to sterol (ergosterol preferentially) in cell membranes and create holes that lead to ion leakage and fungal death.
It is well known that azoles are substrates of Pdr5p in yeast , . To investigate the functional consequence of dramatic changes in the YJM789 PDR5 gene, we compared the growth of YJM789 with that of BY4741 in three selected azoles: itraconazole, ketoconazole, and fluconazole. As shown in Figure 2, YJM789 was much more sensitive than BY4741 to all these drugs. We tagged PDR5 genes with GFP in both BY4741 and YJM789. Although Pdr5p still localizes in cell membranes and has an intact protein expression (Figure 3), growth patterns of YJM789 were similar to those of the BY4741Δpdr5 strain when grown in the presence of azoles (Figure 2), which would be expected if Pdr5p drug efflux functions were impaired in YJM789. Considering that YJM789 and BY4741 have significant genetic differences, it is risky to conclude that from this evidence that hypersensitivity of YJM789 to azoles is solely due to its highly divergent PDR5 gene.
To determine if the PDR5 gene is responsible for the observed drug phenotype in YJM789, we conducted the following two experiments, with fluconazole used to represent azole antimycotics. First, growth profiles of YJM789 were measured before and after the deletion of the PDR5 gene. If Pdr5p had indispensable drug-exporter function, deletion of the gene would lead to reduced growth of YJM789 in fluconazole. As shown in Figure 4, however, YJM789 and YJM789Δpdr5 grew similarly in fluconazole. The fact that Pdr5p played no significant role in YJM789 for resistance to fluconazole suggests that the protein might have lost its ability to export azoles. In the presence of fluconazole, both YJM789 and YJM789Δpdr5 grew slightly better than BY4741Δpdr5 (Figure 4), indicating that YJM789 may have acquired residual azole-export capability elsewhere in the genome.
Second, to test if BY4741 Pdr5p could accomplish its normal function in the YJM789 background, we replaced the YJM789 PDR5 gene with the allele from BY4741. The growth profiles for BY4741, YJM789 and YJM789 expressing BY4741 PDR5 were compared. As shown in Figure 4, YJM789 with BY4741 PDR5 grew as well as BY4741 WT over the range of fluconazole concentrations, indicating that in the YJM789 background, BY4741 Pdr5p confers normal azole-exporter function. It also indicates that hypersensitivity of YJM789 to azoles was not due to inherent changes in the plasma membrane of YJM789. Our additional results indicated that it is more likely that the TMDR2 domain in the YJM789 PDR5 gene cannot carry out its original function (Figure 5), which is consistent with the observation that the transmembrane segments in this region have a higher percentage of amino acid changes than other regions (Table 1).
Resistance to antifungal drugs represents a serious health threat. YJM789 is a common yeast strain increasingly isolated from AIDS patients who have received prolonged courses of prophylactic antifungal drug treatment . Because these isolates of YJM789 presumably have survived intensive exposure to antimycotic drugs, particularly the azoles, it was intriguing to observe that YJM789 lost the ability to grow in the presence of antifungal drugs that target ergosterol synthesis.
To further investigate the reason underlying the high mutations of PDR5 gene in YJM789, we examined the influence of another group of antifungal drugs: the polyenes. Polyene antimycotics can bind to sterols (ergosterol preferentially) in fungal cell membranes, promoting leakage that contributes to cell death. Amphotericin B (AmB) and nystatin are commonly used polyene antimycotics . Interestingly, YJM789 is much more resistant to AmB (Figure 6A, 6C) and nystatin (data not shown) than is BY4741. Several lines of evidence indicate that mutations in PDR5 gene of YJM789 are related to its improved growth in this drug environment. First, the BY4741Δpdr5 (green bars in Figure 6A) strain was more resistant to AmB than BY4741WT (yellow bars in Figure 6A), indicating that loss of PDR5 gene function led to a growth advantage in the presence of AmB. Second, strain YJM789 showed similar growth profiles both before and after PDR5 deletion in the presence of AmB (Figure 6A, 6C). Third, and most importantly, YJM789 strains containing a functional BY4741 PDR5 gene became more sensitive to AmB than did YJM789 strains without a functional PDR5 gene (Figure 6B). Interestingly, this growth difference was significant only at 37°C, the temperature of the environment from which YJM789 was isolated, whereas it was difficult to see this difference at 30°C. Of note, YJM789 with a functional BY4741 PDR5 gene showed a growth advantage over wild-type YJM789 in the presence of fluconazole at both at 30°C (Figure 4) and 37°C (Figure S1).
Genetic variation among individuals contributes to the fitness landscape of a population. Loss of gene function can have dramatic fitness consequences for individuals. Can loss of a gene be adaptive? The antagonistic pleiotropy hypothesis states that certain genes, when functional, are beneficial in some conditions and deleterious in others , , , . This genetic pleiotropic effect is a result of interactions between genes and environments, which can lead to a trade off for organismal fitness under certain conditions. When environments change, the antagonistic pleiotropy of gene functions can lead to adaptive gene loss in evolution. Indeed, this “less is more” scenario was proposed as one model for phenotypic evolution , . Accumulating evidence for adaptive gene loss indicates that antagonistic pleiotropy of gene function may play an important role in species adaptation , , , .
In this study, we discovered an interesting case of possible adaptive functional loss that has clinical relevance. Our data show that the PDR5 gene in YJM789 has lost the ability to facilitate growth in the presence of azoles, presumably due to the inability to recognize and export members of this class of antifungal drug. It has been reported that mutations in PDR5 gene in YJM789 may be responsible for its hypersensitivity to cycloheximide . We showed that deletion of PDR5 gene, however, improves organism growth in the presence of polyenes. Interestingly, among recently genotyped S. cerevisiae strains , several different pathogenic S. cerevisiae isolates showed large sequence divergence in the PDR5 gene, which was not observed in any non-pathogenic strains, indicating that dramatic changes in this gene could be important for the pathogenicity of S. cerevisiae (Figure S2). We are currently conducting experiments to address this issue in these pathogenic yeast strains.
Because at least two azole drugs were brought to market (ketoconazole and miconazole) before the isolation of YJM789 and other clinical strains , it is intriguing to observe that a yeast strain with high azole sensitivity could still survive in the pathogenic S. cerevisiae population. One possibility is that the patients from whom the strains were isolated didn't receive intensive treatment with azole antifungal drugs. However, if the gene has been in the population for a considerable time, it seems less likely that azoles were not used extensively in this population. YJM789 is a haploid derivative of the heterozygous diploid clinical isolate YJM128 . The highly mutated allele we observed could be protected by another copy due to the rare occurrence of meiosis in these clinical strains. Indeed, both the complicated life history of pathogenic yeast strains and adaptation to antifungal drugs (polyenes) might function together to keep this highly mutated PDR5 allele at a certain frequency in pathogenic population of S. cerevisiae.
The origin of the highly polymorphic regions in the YJM789 PDR5 gene is not clear. The synonymous distance (KS) of two transmembrane domain regions between S288c and YJM789 are 0.3 and 0.5, respectively (Figure 7A), which are similar to the synonymous distance between S288c and S. paradoxus, another Saccharomyces species, whereas KS for the rest of the gene between S288c and YJM789 is 0.071, similar to the genome average . AmB, a polyene antimycotic, was made available in the early 1960s, and is widely used in human immunodeficiency virus (HIV)-seropositive patients . Since antifungal drug therapy has a short history, we have to emphasize that adaptation to polyenes by modification of the PDR5 gene, on the assumption that it did occur, likely represents an evolutionary force that selected a pre-existing genetic variation.
The closest ABC-family member to the PDR5 gene in S. cerevisiae genome (PDR15 gene) has a 24% amino acid difference from Pdr5p and the synonymous differences between PDR5 and PDR15 genes are saturated (Ks>5), indicating that the two TMDRs in the YJM789 PDR5 gene could not have originated from ectopic gene conversion within the same species. A phylogenetic tree was built for the two TMD regions within orthologs of Pdr5p for all sequenced strains of S. cerevisiae and S. paradoxus . As shown in Figure S3, TMD regions in Pdr5p from YJM789 have the most divergent sequences among strains of S. cerevisiae, which is not the case for other genes in the YJM789 genome . Furthermore, TMDRs in the YJM789 PDR5 gene also show great sequence divergence from strains of S. paradoxus, indicating that these two domains were not from S. paradoxus, or at least from strains of that species with available genome sequences. No matter what are the origins of these two TMDRs in the YJM789 PDR5 gene, the ratio of nonsynonymous to synonymous changes across the whole PDR5 gene significantly increased in YJM789 (Figure 7B), indicating functional relaxation of this gene, which is consistent with our experimental data for inactivation of important functions in Pdr5p of strain YJM789.
The observation of a gain in AmB resistance as a result of the loss of PDR5 gene is surprising. Ergosterol is an important component of fungal membranes and serves the same function as cholesterol in animal cells. When ergosterol content decreases in fungal plasma membranes, AmB-binding sites might decrease, leading to AmB resistance , . If this is indeed the mechanism for AmB resistance after PDR5 deletion, our results imply that Pdr5p is involved in ergosterol homeostasis. Consistent with our speculation, some studies have reported that Pdr5p may transport lipids, such as phospholipids , sphingolipids  or glycerophospholipids , although the mechanisms of transport are unclear. There are no reports of ergosterol transport by Pdr5p. Further experiments are needed to elucidate the functional role of Pdr5p in ergosterol homeostasis.
It is important to point out that the PDR5 gene in YJM789 might not be correctly called a pseudogene. The fact that Pdr5p could localize in cell membranes and has an intact protein expression in YJM789 indicates that the pleiotropic protein Pdr5p might conduct unrecognized functions. Because Pdr5p is a multi-substrate transporter , our observation could be caused by changes in substrate specificity as a result of mutations in the Pdr5p TMDRs of YJM789. Indeed, this conclusion is supported by the observation that the ratio of non-synonymous to synonymous distances between S288c and YJM789 is less than 0.2 for the PDR5 gene (Figure 7B).
It is also important to note that other genetic changes contribute to the YJM789 AmB-resistance phenotype. At higher concentrations of AmB, growth of BY4741 and BY4741Δpdr5 was inhibited, whereas YJM789 and YJM789Δpdr5 still showed significant growth, indicating that changes in other genes in YJM789 contributed to AmB adaptation. It was shown that mutations in ERG6, an important gene in the ergosterol biosynthesis pathway, can lead to AmB resistance in various pathogenic species of yeast , . Our analysis indicated that ERG6 gene and its ~300 bp 5′ UTR region are identical in BY4741 and YJM789, indicating that the gene is not likely involved in AmB resistance of YJM789. By using a genotyped progeny panel (~200 strains) from a cross between BY and YJM789 , we are currently conducting QTL-mapping to identify genetic loci that underlie this interesting phenotype.
Regardless of the evolutionary history of the PDR5 gene and the mechanism for the AmB-sensitivity change in YJM789, our experimental results might imply a new drug-resistance strategy in pathogenic yeasts, i.e., sacrificing an important function in one drug condition for a minor fitness gain in other drug conditions. This clinically important showcase with interesting evolutionary implication will, hopefully, lead to better understanding of the emergence of drug resistance not only in pathogenic fungi, but also in microbes in general.
All antifungal drugs were obtained from Sigma-Aldrich and Fisher Scientific. AmB, fluconazole, cycloheximide, nystatin dihydrate and itraconazole were reconstituted with water to appropriate concentrations. Ketoconazole was prepared in dimethyl sulfoxide. All stock dilutions were stored at –20°C for up to 2 months.
The strains of S. cerevisiae, listed in Table 2, were grown at 30°C and maintained on yeast extract/peptone/dextrose medium (YPD). YPD and synthetic media (SD) were prepared as described by Rose et al. , YPD media containing G418 (200 µg/mL) was used for selection of strains with kanMX4 dominant drug-resistance markers. SD uracil-deficient medium was used for selection of strains with URA3 marker.
Transformation of yeast with plasmid DNA was achieved according to the procedure previously described . Bacterial transformation was performed using the calcium chloride procedure as described by Sambrook et al. .
The strains GY01 (YJM789Δpdr5) and GB01 (BY4741Δpdr5) were constructed by replacing the PDR5 gene from YJM789 and BY4741 parental strains using a URA3 marker, respectively. Correct integration of the deletion constructs and proper looping-out were confirmed by PCR analysis.
Strain GY02 was constructed by inserting BY4741 PDR5 gene (BPDR5) into the PDR5 locus of GY01. The transformants were selected on 5-FOA plates . Because the viability of the mutants (GY02) in antifungal drugs might be influenced by 5-FOA, the BPDR5 gene was disrupted in GY02 by targeted insertion with URA3 as the selection marker to reconstruct the pdr5 null mutants (GY03).
Fresh S. cerevisiae colonies were inoculated in 5 mL of YPD or SD selective medium, and grown overnight at 30°C. The cells were diluted to an OD600 of 0.2, and 4 µL was spotted with serial dilutions on solid medium containing drugs (AmB, fluconazole, itraconazole or ketoconazole) in the agar. The plates were incubated at 30°C for 48 h.
To test drug resistance, strains were grown overnight, diluted in 90 µL of rich medium (YPD), treated with 10 µL of water or a pharmacological compound, and then grown for 20 to 24 h. Samples were grown in a microplate spectrophotometer (OD 600) (model Uμ-MQX200R; Bio-T Instrument Inc.). Drug resistance was estimated from the optical density after incubating for 20 to 24 h (20 h for AmB and 24 h for azoles).
The 5.18 kb PDR5 fragment including 404 bp upstream and 224 bp downstream sequences was amplified from YJM789 or BY4741 genomic DNA (using primers Xba I_upper/Xba I_lower) and then inserted into the Xba I site of vector YEplac195, generating YEplac195-YPDR5 or YEplac195-BPDR5. For the replacement of YJM789 PDR5 transmembrane domains (TMDR1 or TMDR2) with the BY4741 PDR5 TMDRs, 2.7KB (1–2764 bp) and 1.8 KB (2764–4535 bp) fragments of PDR5 ORF which include BY4747 PDR5 TMDR1 or TMDR2 encoding sequences were amplified by using primer combinations Sac I-upper/BstP I-lower and BstP I-upper/Sal I-lower, and then cloned into Sac I-BstP I and BstP I-Sal I digested YEplac195-YPDR5, respectively. All the above reconstructed plasmids were confirmed by colony PCR and DNA sequencing. Primer sequences used in this work (the underlined sequences indicate restriction sites, the restriction site of BstP I is inside the PCR product) are listed in Table 3.
Pdr5p-GFP protein fusions were constructed as previous described. The 2.6kb GFP-kanMX6 cassette was amplified from pFA6a-GFP(S65T)-kanMX6 with a pair of chimeric primers: TGGCTAGCAAGGGTACCTAAAAGGAATGGTAAACTCTCCAAGAAAGcttcgt acgCtgcaggtc, CGATGAGATAACCTAGAAATAAAATTCTCGGAATTCTTTCGGACcc aatacgcaaaccgcctct. The PCR product was transformed into YJM789 WT and the transformants were selected on YPD plates containing G418. Integration of the GFP fusion at the PDR5 locus was confirmed by PCR using combinations of primers flanking PDR5 gene and GFP-kanMX6 cassette-specific primers. Fluorescence analysis was done using exponentially growing cells. Cells expressing Pdr5p-GFP fusion protein were examined by using fluorescence and differential interference contrast (DIC) microscopy (OLYMPUS BX51 microscope).
To confirm PDR5-1 (YPDR5 TMDR1 + BPDR5 TMDR2) and PDR5-2 (BPDR5 TMDR1 + YPDR5 TMDR2) could express in BY4741Δpdr5 background, total RNA was prepared as described in an earlier study. The cDNA was synthesized by PrimeScript™ RTase (Takara). Primer pairs (TGAGGTCTTCATCTCCTT/AAACCAACCTCTCCGTGT for PDR5-1, the PCR product length is 1111 bp, GCAAGACCTTCCTCTCCT/CAATTTCTCCGTAAGT ATCG for PDR5-2, the PCR product length is 1114 bp) were used to detect transcription of TMDR1 and TMDR2 constructs.
To calculate evolutionary distance between two strains, we used PAML program to calculate the substitution rates of synonymous sites (KS) and nonsynonymous sites (KA) as previously described .
All sequenced S. cerevisiae strains and four S. paradoxus strains, one from each of four clades in this species , were used for tree reconstruction. S. bayanus sequence was used as outgroup. The combined TMDRs in each strain were aligned by Muscle  and tree was reconstructed by Clustalw . The phylogenetic tree is shown in Figure S3.
Growth difference between YJM789Δpdr5::BPDR5 (GY02) and YJM789Δpdr5 (GY03) mutants in media containing fluconazole at 37°C. GY02(•) and GY03(-) were grown in YPD overnight at 30°C and reinoculated to OD600=0.1. 90 µl above media of strains were treated with 10 µl of water containing fluconazole (Final concentration is 20 µg/ml), and then grown for 24 h at 37°C in a microplate spectrophotometer (model Uμ-MQX200R; Bio-T Instrument Inc.). OD600 value was sampled every 20 minutes.
(0.03 MB DOC)
Multiple pathogenic S. cerevisiae strains have divergent sequence in PDR5 gene. SNP distributions for PDR5 gene from all 63 strains were generated from http://gbrowse.princeton.edu/cgi-bin/gbrowse/yeast_strains_snps/ (YSB, Yeast SNPs Browser ). Eleven strains were shown in this figure, and the pathogenic strains were highlighted in red rectangular. Among all 63 strains, five pathogenic strains (including YJM145 that is isogenic to YJM789), which were from different clinical isolation , display dramatic sequence divergence in PDR5 gene, while the remaining clinical strains (YJM421 was shown as an example) show only a few SNPs. In contrast, none of the non-clinical strains (five were shown as example) show dramatic sequence divergence in PDR5 gene.
(0.04 MB DOC)
Phylogenetic tree for the combined two transmembrane domain regions (TMDR) in PDR5 (YOR153W) gene. All sequenced S. cerevisiae strains and four S. paradoxus strains (marked in green), one from each of four clades in this species , were used for tree reconstruction. S. bayanus sequence was used as outgroup. The combined TMDRs in each strain were aligned by Muscle  and tree was reconstructed by Clustalw . Bootstrap values are shown on the tree. As indicated in the figure, the PDR5 gene in YJM789 (indicated by red arrow) is clearly an outgroup in all S. cerevisiae strains, which is not true for whole genome sequence comparison(3). The PDR5 gene in YJM789 is also very divergent from all S. paradoxus strains (including all sequenced strains, only four were shown here). For S. mikatae, S. kudriavzevii, S. cariocanus, as they either are more divergent than S. paradoxus from S. cerevisiae, or their PDR5 gene sequences are not available, we didn't include them in the phylogenetic analysis.
(0.05 MB DOC)
We thank Hong Chen, Wu Wei for help and discussions, and André Goffeau, Paul Soloway, Lixin Zhang, and Allan Eaglesham for reading the manuscript. Helps during experiments from Eric Alani and Koodali Nishant are also very much appreciated.
Competing Interests: The authors have declared that no competing interests exist.
Funding: The study was supported by the faculty startup funds from Cornell University awarded to Z.G. and the Future Academic Stars project of Zhejiang University to W.G. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.