In Saccharomyces cerevisiae, transcription of several drug transporter genes, including the major transporter gene PDR5, has been shown to peak during mitosis. The significance of this observation, however, remains unclear. PDR1 encodes the primary transcription activator of multiple drug transporter genes in S. cerevisiae, including PDR5. Here, we show that in synchronized PDR1 and pdr1-3 (multidrug resistant) strains, cellular efflux of a known substrate of ATP-binding-cassette transporters, doxorubicin (a fluorescent anticancer drug), is highest during mitosis when PDR5 transcription peaks. A genetic screen performed to identify regulators of multidrug resistance revealed that a truncation mutation in ELM1 (elm1-300) suppressed the multidrug resistance of pdr1-3. ELM1 encodes a serine/threonine protein kinase required for proper regulation of multiple cellular kinases, including those involved in mitosis, cytokinesis, and cellular morphogenesis. elm1-300 as well as elm1Δ mutations in a pdr1-3 strain also caused elongated bud morphology (indicating a G2/M delay) and reduction of PDR5 transcription under induced and noninduced conditions. Interestingly, mutations in several genes functionally related to ELM1, including cla4Δ, gin4Δ, and cdc28-C127Y, also caused drastic reductions in drug resistance and PDR5 transcription. Collectively, these data show that ELM1, and genes encoding related serine/threonine protein kinases, are required for regulation of multidrug resistance involving, at least in part, control of PDR5 transcription.
IN Saccharomyces cerevisiae, transcriptional upregulation of transporters that belong to the ATP-binding-cassette (ABC) superfamily results in multiple drug resistance or in pleiotropic drug resistance (PDR). Transcriptional activation of many of these transporters is known to occur with drug exposure (e.g., cycloheximide) or in the presence of gain-of-function mutations in the transcriptional activators themselves (Balzi and Goffeau 1995; Wolfger et al. 2001; Moye-Rowley 2003). Transcription of most genes is known to be significantly reduced during mitosis, and this mitotic repression has been, at least in part, attributed to inactivation of the transcriptional machinery (Gottesfeld and Forbes 1997; Long et al. 1998). Surprisingly, microarray analysis indicates that transcription of several drug transporter genes, including PDR5, peaks during mitosis (Spellman et al. 1998; reviewed in Bahler 2005; Wittenberg and Reed 2005). However, this finding has not been thoroughly investigated, and the impact of PDR5 transcriptional upregulation during mitosis on multidrug resistance remains unknown.
Two major transcriptional activators, Pdr1 and Pdr3, control the level of many drug transporters in S. cerevisiae (Gao et al. 2004; Milgrom et al. 2005 and references therein). These homologous proteins belong to the Gal4 superfamily with Cys6-Zn(II) DNA-binding domains (Poch 1997; Kolaczkowski et al. 1998; Bauer et al. 1999). The DNA-binding domain of Pdr1 targets over a dozen transport gene promoters (most notably PDR5) with the pleiotropic drug resistance element (PDRE) 5′-TCCGCGGA-3′ (Balzi and Goffeau 1995; Kolaczkowska et al. 2002). The functions of Pdr1 and Pdr3 overlap; however, Pdr3, but not Pdr1, is subject to auto-regulation (Delahodde et al. 1995). Several substitution mutations within Pdr1 result in a hyperactive activator (e.g., F815S in the Pdr1-3 hyperactivator protein encoded by the pdr1-3 allele (Meyers et al. 1992; Carvajal et al. 1997) that increases the transcription of many genes encoding ABC transporters (including PDR5), as well as permeases and enzymes involved in lipid and cell-wall synthesis (DeRisi et al. 2000). Similarly to Pdr1, several hyperactive Pdr3 activators, including that encoded by the pdr3-2 allele, were identified (Nourani et al. 1997).
The Pdr5 transporter is a major plasma-membrane-associated ATPase regulated by Pdr1/Pdr3, and it is responsible for cellular detoxification of many agents, including the anticancer drug doxorubicin (Rogers et al. 2001). Yeast Pdr5 exhibits functional homology to the mammalian P-glycoproteins (Pgp), and overexpression of Pdr5 confers multidrug resistance. Extensive efforts have been made to identify small molecules that could reverse the drug resistance phenotype, primarily by inhibiting transporter activities (Lewis 2001). In this regard, the function of Pdr5, as well as similar drug transporters of the pathogenic yeast Candida albicans, are inhibited by the immunosuppressant FK506 (tacrolimus, Prograf) (Egner et al. 1998; Schuetzer-Muehlbauer et al. 2003).
We previously characterized the transcriptional regulation of PDR5 by comparing differences in the recruitment of activators and coactivators and the nucleosome structure in isogenic PDR1 and pdr1-3 strains. We demonstrated that Pdr1 is constitutively bound to the PDR5 promoter. Cycloheximide induction in the wild-type PDR1 strain alters the nucleosome structure at the PDR5 upstream activating sequence (UAS) region harboring PDRE. These alternations reflect changes in the interactions between Pdr1 and PDRE and are associated with PDR5 transcriptional activation (Gao et al. 2004). Moreover, we showed that proper interactions between histones and PDR5 coding sequences specifically require the transcription factor Spt-Ada-Gcn5-acetyltransferase (SAGA) (Milgrom et al. 2005). However, factors other than Pdr1 and SAGA that are required for proper transcriptional regulation of PDR5 remain to be identified.
In S. cerevisiae, mitotic entrance is coupled to morphogenesis (Sakchaisri et al. 2004). Entry into mitosis is initiated by activation of Clb2-Cdc28/Cdk1 cyclin-dependent kinase (CDK), which involves degradation of its inhibitor Swe1 kinase (McMillan et al. 2002). Mechanistically, Swe1 phosphorylation by Cdk1 activates Swe1, and this phosphorylation is required for the formation of a stable Swe1–Cdk1 complex that maintains Cdk1 in an inhibited state (Asano et al. 2005; Harvey et al. 2005). Additional kinases are involved in the regulation of transition from G2 to mitosis. ELM1 (elongated morphology 1) encodes a serine (Ser)/threonine (Thr) protein kinase, and cells harboring elm1Δ exhibit elongated filamentous growth, an indication of G2/M delay (Koehler and Myers 1997). The function of Elm1 kinase in mitotic signaling (Sreenivasan and Kellogg 1999) has been linked, in part, to regulators of septin organization, a key set of protein serine/threonine kinases encoded by GIN4 and CLA4, as well as an interactor of mitotic cyclin Clb2 encoded by NAP1 (Altman and Kellogg 1997; Benton et al. 1997; Carroll et al. 1998; Edgington et al. 1999; Longtine et al. 2000; Gladfelter et al. 2004). Support of Elm1 involvement in the G2/M transition includes the fact that Elm1 is required for hyperphosphorylation of Swe1 during mitosis (Sreenivasan and Kellogg 1999). Elm1 is also required for the regulation of bud emergence at the G1 phase (Sreenivasan et al. 2003). Independently of its roles in the cell cycle, Elm1 appears to function upstream of Snf1, a key AMP-dependent kinase pathway that regulates carbon metabolism in yeast (Hong et al. 2003; Sutherland et al. 2003) and mammalian cells (Woods et al. 2003).
In this study, we show that PDR5 transcription during cell cycle progression is inversely correlated with cellular accumulation of doxorubicin. A truncation and null mutation in ELM1 were identified as suppressors of pdr1-3. Yeast strains harboring mutations in genes encoding Elm1-related kinases (e.g., gin4Δ, cla4Δ, and cdc28-C127Y) similarly reversed the multidrug resistance of pdr1-3, exhibited elongated bud morphology and impaired PDR5 transcription (without affecting Pdr1-independent transcription) and abolished cellular doxorubicin elimination. Epistasis analysis suggested that ELM1 functions upstream of Pdr1-mediated PDR5 transcription. This ELM1–PDR5 genetic connection is independent of the SNF1 pathway. We also show that elm1Δ alters nucleosome structure upstream of the established Pdr1-binding sites in the PDR5 promoter (Katzmann et al. 1996). However, Myc-tagged Elm1 is not detectable on the PDR5 promoter. In summary, our studies indicate a novel link between regulation of multidrug resistance and cell cycle progression in S. cerevisiae, involving genes encoding key serine/threonine kinases acting during mitosis.
MATERIALS AND METHODS
Chemicals and solutions:
Doxorubicin HCl (MW 579.99) solution (3.45 mm) was obtained from GensiaSicor Pharmaceuticals (Irvine, CA). FK506 (tacrolimus, Prograf, MW 822) solution (5 mg/ml = 6.08 mm) was obtained from Fujisawa (Deerfield, IL). The remaining reagents were purchased from Sigma-Aldrich (St. Louis). Sodium methanesulfonate solution used for HPLC was prepared from 15.4 m methane sulfonic acid by addition of one equivalent of sodium hydroxide and dilution to 4.0 m.
Yeast strains, genetic manipulations, agar plate drug resistance assays, and measurements of cellular respiration:
Yeast cells were grown in rich (YPD) or synthetic media according to standard procedures (Sherman 1991). The genotypes of yeast strains used in this study are listed in Table 1 (Wolfger et al. 1997; Gao et al. 2004) and described previously for deletion strains derived from BY4741 (Milgrom et al. 2005). The null alleles introduced into pdr1-3 were carried out by PCR-mediated allele transfer from deletion alleles of nonessential genes available from the collection of synthetic genetic arrays (Tong et al. 2001) or by PCR-based gene deletion using modification cassettes as previously described (Longtine et al. 1998). The strain with the upstream PDR5 promoter region replaced (WCS651, Table 1) was generated by using the following primers to amplify a TRP1 fragment from the pRS404 vector: PDR5-F1, 5′-CTTTTGTACGATTTTAAACAGTAAAATCGATGCATATTAAGGGAGGCCCCGGATCCCCGGGTTAATTAA-3′ (with underlined sequences being PDR5 specific) and PDR5-R1, 5′-GGTAATTTGATGTTCTTTTTTTTCTTTGATTTGAACTTTTGTTCTCTCTCTGAATTCGAGCTCGTTTAAAC-3′. The WCS651 strain generated bears a deletion from −726 to −1123 (relative to the transcription start site) and was replaced with TRP1 (1049 bp). Myc or GFP tags were introduced at the 3′-end of the PDR1 and PDR5 coding sequences by PCR-mediated modification (Longtine et al. 1998). PDR5 mRNA was induced by treatment of cells for 45 min in YPD medium containing 0.2 μg/ml (0.71 μm) cycloheximide (CYH) as described (Gao et al. 2004; Milgrom et al. 2005).
Agar plate drug resistance assays were carried out as follows. Strains were spotted (at sequential 10-fold dilutions) on plates containing either YPD or complete synthetic medium with indicated amino acids omitted. Images were taken after incubation at 30° for 3 days. Plates containing 0.2 μg/ml CYH (Figure 4E), 1.0 μg/ml CYH (Figure 1A; Figure 4, A–C; Figure 6B), or 15 μg/ml fluconazole (FLU; Figure 1A) were used in the agar plate drug resistance assay.
Cellular respiration was measured at 25° in sealed vials containing 107 cells as described (Souid et al. 2003). The cells were suspended in 1.0 ml medium, containing 6.0 mm Na2HPO4, 10 mm glucose, 2.0 μm Pd phosphor [Pd(II) complex of meso-tetra-(4-sulfonatophenyl)-tetrabenzoporphyrin], and 1.0% (w/v) fat-free bovine serum albumin. The rate of cellular respiration was determined as the negative slope of the curve of [O2] vs. time (in micromolars O2/min/107 cells). These measurements reflected mainly mitochondrial respiration because addition of 1.0 mm NaCN completely inhibited oxygen consumption.
Ethylmethane sulfonate mutagenesis and identification of elm1-300 mutation:
The ethylmethane sulfonate (EMS) mutagenesis protocol was performed as described (Lawrence 1991; Milgrom et al. 2005), using a concentration of EMS (15 μl/ml of cell culture, final 15 mm) that gave 50% cell survival. Log-phase pdr1-3 cells (∼5 × 107 cells/ml of WCS261, Table 1) were treated with EMS, and 8 × 104 cells were screened for mutants that failed to grow on 1.0 μg/ml CYH by replica plating. Seven mutants showed a clear loss of CYH resistance. These were mated with the opposite mating type of the pdr1-3 strain (WCS347, Table 1). We noted that pdr1-3 is dominant over PDR1 in terms of drug resistance (Wolfger et al. 1997). Tetrad analysis of the resulting diploid progeny indicated that two of the seven mutants (WCS343 and WCS345) harbored a single recessive mutation responsible for the loss of CYH resistance. Strain WCS345 was transformed with a YEp13-based wide-type genomic DNA library to clone the corresponding wild-type gene of the suppressor mutation. The resulting transformants were selected for restoration of CYH resistance. To exclude the possibility that recovery of CYH resistance in WCS345 was due to overexpression of genomic DNA inserted into the YEp13 vector, we subcloned the insert into a low-copy-number pRS415 plasmid (Sikorski and Hieter 1989) and repeated the test for restoration of drug resistance in WCS345. Analysis of various subclones of genomic inserts confirmed ELM1 as the wild-type gene corresponding to the single recessive allele introduced by EMS in the WCS345 strain. The elm1 allele responsible for loss of CYH resistance was cloned by PCR from genomic DNA of the WCS345 strain and sequenced.
Mapping PDR5 chromatin structure by micrococcal nuclease and Northern and Western blots:
The detailed mapping of the PDR5 promoter in pdr1-3 and pdr1-3 elm1Δ strains (Figure 8A) was essentially as described (Gao et al. 2004). The following modifications were made for mapping data presented in Figure 8B. First, ClaI instead of HindIII digestion was used for mapping nucleosome structures upstream known as PDREs. Second, 32P-labeled probes for Southern blot analysis were generated from genomic PCR product (273 bp) of primers 5′-CGATGCATATTAAGGGAGGCC-3′ and 5′-CGCTTCCTTTGTATGATATC-3′. Preparation of total RNA and Northern blot analysis were performed as described (Shen and Green 1997; Gao et al. 2004). Probes for PDR5, SWI5, PDR12, and ADH1 transcripts were obtained by PCR amplification from genomic DNA. Preparation of whole-cell extract by glass bead disruption was performed as described (Walker et al. 1997) and protein samples were subjected to SDS–polyacrylamide gel electrophoresis and immunoblotting (Harlow and Lane 1988). The mouse monoclonal antibodies against c-Myc (9E10 clone), GFP antibody, glucose-6-phosphate dehydrogenase (G6PD) antibody, and antibody α-tubulin (12G10 clone) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA), Roche Diagnostics, Sigma, and the Developmental Studies Hybridoma Bank at University of Iowa, respectively. Quantification of relative mRNA and protein levels was performed using a PhosphorImager (model 425, Molecular Dynamics, Sunnyvale, CA).
Cell synchronization with nocodazole or α-factor was performed as described (Amon 2002). Yeast strains bearing the pdr1-3 allele or a 3′ Myc-tagged pdr1-3 allele were grown in YPD at 30° to A600 of ∼0.3. Samples were removed for the asynchronous controls. For metaphase arrest, nocodazole (20 μg/ml) was added to the cultures. When >90% of the cells showed the characteristic arrest (equal size of budded mother–daughter cells), cells were collected by centrifugation and washed twice with 2 vol of YPD, and the pellets were resuspended in an equal volume of YPD. Samples were taken at time zero for RNA preparation or incubated with 50 μm doxorubicin for 15 min for measurement of cellular doxorubicin. Microscopic examination of cells harvested throughout the experiments confirmed that cells sampled at time zero were arrested at mitosis and revealed approximation of various stages during a cell cycle. The synchronized cultures continued to grow at 30°, and samples were taken at 15-min intervals for preparations of RNA and measurement of intracellular doxorubicin after incubation with 50 μm doxorubicin for 15 min. There was no apparent cell cycle progression during the 15-min incubation with doxorubicin, consistent with the antiproliferative effect of the drug in mammalian cells (Harisi et al. 2005) and yeast (our observation).
Growth of cultures for G1 arrest by α-factor was essentially the same as metaphase arrest by nocodazole. The concentration of α-factor used for synchronization was 3 μg/ml, and the characteristic schmoo phenotype, indicating G1 arrest, was confirmed microscopically.
The condition for formaldehyde-based in vivo crosslinking and chromatin immunoprecipitation (ChIP) was as described (Gao et al. 2004; Milgrom et al. 2005). The immunoprecipitations were performed using anti-Myc antibody. The following sets of primers were used for the PCR analysis: PDR5 (UAS), 5′-TCGTGATCACGATTCAGCACC-3′ and 5′-GGAGAGGCCTTGTTTGTATTGC-3′; PDR5 (middle of the coding sequences, CDS), 5′-GAAAGCTCTGAAGAGGAATCC-3′ and 5′-CCCTTTCGGCCAAACAATCCA-3′; PDR12 (promoter), 5′-CACATTTTCTCGACGGTTC-3′ and 5′-GTAACTGGGAAAACAGAG-3′.
Cellular doxorubicin determination:
For cellular doxorubicin efflux, log-phase cells were incubated at 30° in YPD plus 50 μm doxorubicin for 60 min. The cells were collected by centrifugation, resuspended in drug-free YPD (prewarmed to 30°), and incubated at 30° for various periods of time. Aliquots of the cell suspension were then spun at specific time points. The pellets were suspended in 250 μl of 10% perchloric acid/2.0 m Na methanesulfonate. Equal volumes of glass beads were added, and the suspensions were vigorously vortexed for 3.0 min. The acid-soluble supernatants were separated on HPLC and analyzed as described below.
For cellular doxorubicin accumulation, log-phase cells were incubated in YPD media plus 50 μm doxorubicin at 30° for various periods of time. The cells were collected by centrifugation and the pellets were immediately chilled on ice. The cold pellets were washed twice with ice-cold double-distilled H2O containing 20 μm FK506, a known inhibitor of the ABC transporters (Egner et al. 1998). The washed pellets were then suspended in 250 μl of 10% perchloric acid/2.0 m sodium methanesulfonate. Equal volumes of glass beads were added, and the suspensions were vigorously vortexed as described above.
Doxorubicin peaks were detected by fluorescence (excitation, 480 nm; emission, 560 nm) as described (Fogli et al. 1999). The analysis was performed on a Beckman HPLC system. The solvent used was 60% of 50 mm NaH2PO4 (pH ∼3.5) and 40% of acetonitrile. The column (4.6 × 250-mm Beckman ultrasphere IP) was operated isocratically at 0.5 ml/min. Standards (10 μm doxorubicin in 10% perchloric acid/2.0 m sodium methanesulfonate) were included with each analytical run. Peaks were identified (doxorubicin retention time, ∼13.8 min) and quantitated using doxorubicin standards, with the minimal quantifiable level of doxorubicin (∼10 pmol) giving a signal:noise ratio of 3:1. Cellular doxorubicin was expressed in picomoles/107 cells on the basis of microscopic enumeration. Standard deviations were derived from at least three sets of experiments. Treating of PDR1 and pdr1-3 cells with 50 μm doxorubicin for 15–120 min did not induce PDR5 expression (our observation). Statistical significance between values in different strains was determined by paired Student's t-test analyses. A value of P < 0.05 was considered significant.
Characterization of Pdr1-mediated multidrug resistance—a critical role for PDR5 in cellular doxorubicin efflux:
The transcription of several ABC transporter genes (including PDR5) is upregulated in the pdr1-3 strain (WCS261), which is resistant to CYH and the antifungal drug FLU. Introducing pdr5Δ into the pdr1-3 strain resulted in loss of multidrug resistance, confirming PDR5 as the major transporter target of Pdr1 (Figure 1A). The impact of CYH treatment (0.2 μg/ml) in our study on the stability of PDR5 mRNA was investigated in the rpb1-1 strain, which carried a temperature-sensitive mutation in the largest subunit of RNA polymerase II. After thermal inactivation of the rpb1-1 allele, t1/2 for PDR5 mRNA was ∼20 min in either the presence or the absence of the CYH condition (our observation), consistent with the estimation by microarray analysis (Holstege et al. 1998). Thus, our Northern blot analyses of PDR5 reflect transcription levels rather than stability of mRNA.
Compared to wild-type PDR1, an ∼10-fold increase in PDR5 mRNA was found in the pdr1-3 strain (DeRisi et al. 2000; Gao et al. 2004), and further induction occurred with CYH treatment (Gao et al. 2004). To compare the level of Pdr5 protein in PDR1 vs. pdr1-3, we introduced sequences encoding a 13xMyc tag or GFP tag at the 3′-end of the PDR5 coding sequence. The tagged version of PDR5 was functional, indicated by normal resistance to several drugs (not shown). An ∼10-fold increase in Pdr5 protein was observed in pdr1-3 relative to PDR1 (Figure 1B; note the 10-fold lower G6PD loading control in the pdr1-3 lane). Pdr5 protein was inducible by CYH in both pdr1-3 (Figure 1C) and PDR1 strains (not shown). The comparable increases in both PDR5 mRNA and Pdr5 level in the pdr1-3 strain relative to the PDR1 strain confirmed transcriptional regulation as a crucial step for PDR5-mediated drug resistance in S. cerevisiae (Moye-Rowley 2003).
Doxorubicin is a known substrate of many ABC transporters (Eytan 2005) and deletion of PDR5 leads to hypersensitivity to doxorubicin and many other drugs (Rogers et al. 2001; Golin et al. 2003). Transport activities for doxorubicin were determined by analytically detecting the fluorescence of the drug in acid soluble supernatants extracted from cells and separated by HPLC (Figure 2A). In the presence of FK506, a known inhibitor of ABC transporters, including Pdr5 (Egner et al. 1998), cells accumulated approximately ninefold more doxorubicin than in its absence (compare doxorubicin peaks in Figure 2, A and B; note the different cell number count in each condition). FK506 appears to prevent the interactions between Pdr5 and its substrates directly. Changes of a single residue of the Pdr5 transmembrane domain 10 (for example, S1360F, T1364F, and T1364A) alter both substrate specificity and susceptibility to FK506 (Egner et al. 1998, 2000).
A time-course study was used to compare the relative accumulation of doxorubicin by the PDR1 and pdr1-3 strains. During the first 15 min of incubation with the drug, doxorubicin content was similar in the PDR1 and pdr1-3 cells (Figure 2C). Thereafter, the difference between the two strains increased, with accumulation being higher in PDR1. Cellular doxorubicin reached a plateau after ∼50 min, about twice as high in PDR1 as in pdr1-3 (Figure 2C). On the basis of this study, we preincubated the cells with 50 μm doxorubicin for 60 min to visualize the rates of drug elimination (Figure 2D), while drug accumulation was measured by incubating cells with 50 μm doxorubicin for 15 min (Figure 3B). The rate of drug efflux was determined by measuring the fraction of the drug remaining in the cells after incubation in drug-free medium. The resulting best-fit lines (Figure 2D) gave r2 values for pdr1-3 of 0.975 and for pdr1-3 pdr5Δ of 0.990, which is consistent with a kinetic study of P-glycoprotein (Ambudkar et al. 1997) and a process ensuring complete elimination of the drug. The zero-order rate constant (k), calculated as the negative slope of the best-fit straight line, was ∼48 × 10−4/sec for the pdr1-3 and <0.1 × 10−4/sec in the presence of 20 μm FK506. Cellular doxorubicin elimination was therefore negligible in the presence of the inhibitor FK506. The k values for pdr1-3 pdr5Δ and PDR1 were comparable, ∼1.4 × 10−4/sec and ∼1.5 × 10−4/sec, respectively, which were at least 30-fold lower than that for the pdr1-3 strain (Table 2). Significantly, the pdr1-3 pdr5Δ strain exhibited a drug efflux rate similar to that in PDR1 (Figure 2D), underscoring the pivotal role of Pdr5 in Pdr1-regulated cellular detoxification. The nonlinear relationship between the increased doxorubicin efflux (∼30-fold relative to PDR1, Table 2) and the upregulation of PDR5 expression (∼10-fold relative to PDR1, Figure 1) may reflect the collective activity of several transporters known to be overexpressed in the pdr1-3 strain (DeRisi et al. 2000). Other factors might include altered patterns of post-translational modifications on Pdr5, such as ubiquitination (Egner and Kuchler 1996), phosphorylation (Conseil et al. 2001), and glycosylation (Jakob et al. 2001) in the pdr1-3 strain, which in turn contributed to the effectiveness of the drug transporter functions.
Cyclic levels of PDR5 mRNA and cellular doxorubicin during cell cycle progression:
We next analyzed PDR5 mRNA fluctuation during the cell cycle. We synchronized pdr1-3 cells with nocodazole and monitored PDR5 transcript levels and cellular doxorubicin concentration during cell cycle progression (Figure 3). PDR5 transcripts varied considerably during the cell cycle and peaked concurrently with SWI5 mRNA accumulation, a known marker of mitosis (Figure 3A). PDR5 mRNA was approximately fourfold higher at the M phase than at the G1 phase (Figure 3B, diamonds and solid line). Cellular doxorubicin, on the other hand, was about fourfold lower at the M phase than at the G1 phase (Figure 3B, circles and dotted line). Similar phase-related behavior was seen with PDR5 mRNA and cellular doxorubicin during cell cycle progression in synchronized PDR1 cells (not shown). These results indicated that PDR5 mRNA accumulation was cell cycle controlled. Moreover, the inverse relationship between PDR5 mRNA and cellular doxorubicin suggested that PDR5 expression might alter cellular susceptibility to drugs during the cell cycle, with the greatest susceptibility during G1 phase and the least susceptibility during M phase. The possibility that regulatory mechanisms might modulate PDR5 expression during cell cycle progression prompted us to search for genes affecting this aspect of PDR5-mediated drug resistance.
Mutation of ELM1 suppresses PDR5 transcription and drug resistance:
We then performed a genetic screen for extragenic suppressors of CYH resistance in the pdr1-3 haploid strain (WCS 261). We searched for factors that regulate drug resistance mediated by PDR5. Screening of ∼80,000 EMS-treated colonies gave seven candidates showing loss of resistance. Two of the seven candidate haploid pdr1-3 strains harbored additional single recessive mutations as indicated by CYH resistance of the homozygous diploid pdr1-3 strains resulting from mating candidate suppressor strains to an isogenic pdr1-3 strain of the opposite mating type (WCS347) and from segregation of 2:2 resistant and sensitive spores derived from each tetrad. In one of the two mutants, WCS345, the corresponding wild-type gene of the extragenic suppressor allele was cloned by complementation from a high-copy genomic library. Sequencing the clones identified a 6-kb fragment of chromosome XI, carrying the two complete genes ELM1 and CSE4. Microscopic examination of the WCS345 strain demonstrated elongated bud morphology, indicating a G2 delay (Figure 4A, left, pdr1-3 elm1-300), a phenotype reminiscent of elm1 mutations (Koehler and Myers 1997). A low-copy plasmid pRS415 (Sikorski and Hieter 1989) harboring ELM1 (pELM1) restored CYH resistance (Figure 4A, bottom left) and rescued the elongated bud morphology of pdr1-3 elm-300 (not shown). The mutant elm1 allele from the WCS345 strain was cloned and sequenced. The allele harbored a C-to-T mutation in the coding region, converting glutamine 301 (CAG) to a stop codon (TAG) and was therefore termed elm1-300 (Figure 4A). This truncation deleted part of the kinase domain of Elm1, which spans amino acid residues 88–406 (Koehler and Myers 1997; Figure 4A). The Elm1 kinase domain contains two stretches of amino acid residues that match the consensus sequence of Ser/Thr kinases (Blacketer et al. 1993; Koehler and Myers 1997). The elm1-300 mutation eliminated completely the second consensus sequence, GTPAFIAPE (amino acid residues 309–317; Figure 4A).
The level of PDR5 mRNA in pdr1-3 elm-300 was significantly lower than in pdr1-3, and CYH failed to fully induce PDR5 transcription (Figure 4B), indicating that both noninduced (constitutive) and drug-induced transcriptions of PDR5 were reduced in pdr1-3 elm-300. These transcriptional defects were rescued by introducing wild-type ELM1 on a low-copy plasmid, pELM1, into the pdr1-3 elm-300 strain (Figure 4B). Therefore, ELM1 was required for proper constitutive and drug-induced PDR5 transcription.
To further test the effect of elm1, we introduced the elm1Δ allele into pdr1-3 (Figure 4C) and PDR1 (Figure 4D) strains. The pdr1-3 elm1Δ strain was more sensitive to CYH than pdr1-3 elm1-300 was (Figure 4C), suggesting a residual function of ELM1 in the elm1-300 allele. Loss of CYH resistance was also observed when the elm1Δ allele was introduced to another drug resistant strain, pdr3-2, that overexpressed PDR5 (Delaveau et al. 1994) (Figure 4C). These data indicated that the observed elm1Δ effects were not pdr1-3 allele specific. Compared with the PDR1 wild-type strain, the presence of the elm1Δ drastically reduced PDR5 mRNA under both noninduced and induced conditions (Figure 4D). We therefore conclude that ELM1 is required for Pdr1/Pdr3-regulated CYH resistance, which is mediated, at least partially, by PDR5. Epistasis analysis revealed that both pdr1-3 pdr5Δ elm1Δ and pdr1-3 pdr5Δ strains failed to grow in the presence of 0.2 μg/ml CYH or 10 μg/ml fluconazole whereas pdr-1-3 elm1Δ could survive under the same conditions (Figure 4E). These results provided genetic evidence that ELM1 functions upstream of PDR5 in regulation of CYH resistance.
Elm1 is an upstream activator of Snf1 kinase (Hong et al. 2003; Sutherland et al. 2003). Elm1 also regulates mitotic entrance, septin formation, and cytokinesis (Garrett 1997; Koehler and Myers 1997; Bouquin et al. 2000). To test the possibility that Elm1 affects PDR5 transcription via the Snf1-mediated pathway, PDR5 transcription (Figure 4F, left) and CYH resistance (Figure 4F, right) were analyzed in pairwise isogenic wild-type and snf1Δ strains. Deletion of SNF1 had no effect on either PDR5 transcription or CYH resistance. These data indicate that ELM1 regulates PDR5 transcription via a pathway(s) independent of SNF1.
Mutation of genes required for mitotic progression represses PDR5 transcription:
In S. cerevisiae, entry into mitosis is mediated by activating CDK. As described in the Introduction, additional serine/threonine kinases (e.g., ELM1, GIN4, and CLA4) are also involved in regulating the transition from G2 to mitosis. We next investigated the effects on PDR5-mediated drug resistance by other genes that are functionally related to ELM1 and required for proper mitotic progression. In addition to elm1Δ, we introduced individual null alleles of nap1Δ, gin4Δ, and cla4Δ into the haploid pdr1-3 strain and analyzed their effects on cell morphology, CYH resistance, PDR5 mRNA accumulation (Figure 5), and cellular doxorubicin efflux (Table 2). The nap1Δ, gin4Δ, cla4Δ, and elm1Δ alleles resulted in various degrees of elongated bud morphology (Figure 5A), decreased PDR5 transcription (Figure 5B, left), loss of CYH resistance (Figure 5B right), and reduced doxorubicin efflux rates (Table 2), with the general order of effect on drug resistance being elm1 > cla4 > gin4 > nap1 in both the pdr1-3 strain (Figure 5B) and the PDR1 strain (Figure 5C). Consistent patterns of reduced growth at lower drug concentrations were also observed; the minimal inhibitory concentrations for three drugs are listed in Table 3. The pdr1-3 elm1Δ strain exhibited the most pronounced effects on elongated morphology, PDR5 transcription, and drug resistance. The pdr1-3 nap1Δ strain exhibited relatively minor morphological changes, moderately decreased PDR5 transcription, and slightly reduced drug resistance. It is worth noting that there was no detectable doxorubicin efflux in the pdr1-3 elm1Δ and pdr1-3 cla4Δ strains (Table 2), which is virtually identical to that observed in the presence of the inhibitor FK506 (Figure 2D). This suggests that other drug transporter genes in addition to PDR5 are negatively affected by elm1Δ and cla4Δ, consistent with previous reports (Spellman et al. 1998). Significantly, however, elm1Δ, cla4Δ, gin4Δ, and nap1Δ did not affect sorbic-acid-induced PDR12 mRNA (Figure 5D, left) or growth in the presence of sorbate (Figure 5D, right). PDR12 encodes a weak acid anion transporter, whose transcription is independent of Pdr1 and Pdr3 (Piper et al. 1998; Kren et al. 2003). These results indicate a gene-specific transcriptional defect in these mutant strains.
Cdc28 is a master regulator of cell division in S. cerevisiae that controls mitotic entrance (Mendenhall and Hodge 1998). We hypothesized that a cdc28 mutation impairing mitotic progression would suppress the multidrug resistance of pdr1-3 in a manner comparable to that of elm1Δ. The cdc28-C127Y allele was previously shown to cause elongated bud morphology (Edgington et al. 1999). In contrast, the cdc28-Y19F allele (McMillan et al. 1999) is insensitive to Swe1-kinase-imposed inhibition of mitotic entrance and exhibits normal morphogenesis. We introduced these two mutant alleles into the pdr1-3 strain and investigated their effects on PDR5 mRNA levels and CYH resistance. As anticipated, pdr1-3 cdc28-C127Y exhibited reduced PDR5 mRNA (Figure 6C), loss of CYH resistance (Figure 6B), and elongated bud morphology (Figure 6A). Like the pdr1-3 elm1Δ strain, doxorubicin efflux in the pdr1-3 cdc28-C127Y strain was negligible (Table 2). In contrast, the pdr1-3 cdc28-Y19F strain behaved in a manner comparable to the pdr1-3 strain (Table 2; the difference in the efflux rate between the two strains was within the standard error). Reduced CYH resistance due to the presence of the cdc28-C127Y, but not cdc28-Y19F, allele was also observed in the PDR1 strain (Figure 6D), indicating that the effect is not pdr1-3 allele specific. We further analyzed the effects of Swe1 (Cdc28 kinase) and Mih1 (Cdc28 phosphatase) on PDR5 transcription. The swe1Δ strain can enter mitosis like wild-type cells, whereas the mihΔ strain exhibits a mitotic delay (Sia et al. 1996). Relative to pdr1-3 and pdr1-3 swe1Δ, the level of PDR5 mRNA in pdr1-3 mih1Δ was significantly reduced (Figure 6E). As both mihΔ and cdc28-C127Y mutants were defective in mitotic entrance and exhibited marked reduction of PDR5 transcription, these data reinforce the significance of PDR5 transcription peaking during normal mitosis (Figure 2). Therefore, mitotic progression is required for optimal PDR5 expression and development of drug resistance.
We then examined the possibility that other non-cell-cycle-related changes in these mutants (for instance, general sickness) might account for the total loss of doxorubicin efflux (Table 2). For example, it was reported that loss of signaling between nuclei and mitochondria reduces the level of PDR5 expression and drug resistance in rhoo cells (Hallstrom and Moye-Rowley 2000). We therefore measured the rates of cellular mitochondrial oxygen consumption (cellular respiration) in these mutant strains (strains shown in Figure 5 and Figure 6). The respiration was virtually identical, 4.6 ± 0.8 μm O2/min/107 cells, in all mutants studied. These results rule out that alteration of mitochondrial functions in these mutants contributes to the observed defects in PDR5 transcription and to related loss of drug resistance. Moreover, snf1Δ, which exhibits a growth defect on a nonfermentable (respiratory) carbon source, did not affect PDR5 transcription and CYH resistance (Figure 4F). Together, these data support the possibility that Elm1 and related serine/threonine kinases affect PDR5 transcription and drug resistance by a cell-cycle-derived mechanism.
Recruitment of Pdr1 to the PDR5 UAS is cell cycle independent:
We next explored the molecular events required for PDR5 transcriptional upregulation during mitosis. We investigated whether the level of the Pdr1-3 activator and its binding to the PDR5 promoter varied during the cell cycle. We monitored the level of Pdr1-3 during cell cycle progression by integrating a Myc-epitope tag at the 3′-end of the pdr1-3 open reading frame (ORF). The induction of PDR5 in this strain was identical to the nontagged parental strain (Gao et al. 2004). Myc-tagged Pdr1-3 was then analyzed during the cell cycle (Figure 7A). Western blot analysis of nocodazole synchronized cells showed constant Pdr1-3 levels during cell cycle progression (Figure 7A). Therefore, the fluctuating PDR5 mRNA levels in the cell cycle (Figure 3) were not due to fluctuating Pdr1-3 levels.
We then investigated how much the Pdr1-3 activator associates with the PDR5 promoter. A ChIP assay was used to analyze the recruitment of Myc-tagged Pdr1-3 to the PDR5 promoter. As PDR5 transcription is sensitive to cell cycle progression, it is possible that different synchronization protocols might interfere with activator recruitment (Shedden and Cooper 2002). To test this possibility, we conducted parallel ChIP assays from cells synchronized by nocodazole or α-factor treatment and then released. The two treatments gave the same result. Successful synchronization and release at various time points thereafter by either nocodazole (Figure 7A) or α-factor (Figure 7B) were validated by microscopic examination of distinct morphologies throughout the experiments (not shown). Pdr1-3 was recruited specifically to the promoter region of PDR5, and no detectable Pdr1-3 was recruited to the PDR5 CDSs (coding sequences, Figure 7B, synchronized by α-factor). Pdr1-3 recruitment to the PDR5 promoter was constitutive and independent of the cell cycle. In fact, there was no difference in the amount of Pdr1-3 activator bound to the PDR5 promoter as the cell cycle progressed (Figure 7B). We also performed ChIP analysis on the recruitment of wild-type Myc-tagged Pdr1 activator to the PDR5 promoter and the results were virtually identical to the Pdr1-3 recruitment data presented in Figure 7B (data not shown). Therefore, the recruitment of Pdr1 and Pdr1-3 activators to the PDR5 promoter was not responsible for the fluctuating levels of PDR5 mRNA in the cell cycle.
We then analyzed Pdr1 and Pdr1-3 recruitment to the PDR5 promoter in the presence of an elm1Δ allele in both PDR1 and pdr1-3 strains (Figure 7C). In the elm1Δ strains, in which the recruitment of Pdr1-3 onto the PDR5 promoter was slightly higher than that of Pdr1 (Figure 7C), the results are virtually indistinguishable from our previous observation of Pdr1-3/Pdr1 recruitment in the presence of wild-type ELM1 (Gao et al. 2004). We conclude, therefore, that the recruitment of Pdr1 is not a rate-limiting step accounting for decreased PDR5 transcription in the elm1Δ strains. Collectively, the data presented in Figure 7 indicate that cell-cycle-dependent PDR5 transcription is regulated at steps that are independent of Pdr1 recruitment.
Altered nucleosome structure at the PDR5 promoter region in elm1Δ strains:
Our previous micrococcal nuclease (MNase) mapping of PDR5 nucleosome structure in PDR1 and pdr1-3 strains demonstrated that changes in PDR5 transcription levels were associated with alterations in PDR5 nucleosome structure even though Pdr1 was constitutively bound (Gao et al. 2004; Milgrom et al. 2005). This is consistent with the notion that regulation of chromatin structure plays a major role in gene activation (reviewed by Bernstein and Allis 2005; Boeger et al. 2005). As the elm1Δ allele drastically reduced PDR5 transcription (Figures 4D and 5B), we next examined the nucleosome structure at the PDR5 promoter region and compared pdr1-3 and pdr1-3 elm1Δ strains under either noninduced or CYH-induced conditions (Figure 8A). In both pdr1-3 and pdr1-3 elm1Δ strains, CYH induction did not significantly change the pattern or intensities of bands corresponding to MNase hypersensitive sites throughout the PDR5 promoter region (Figure 8A, compare − and + CYH). It is worth noting that, unlike PDR5 coding sequences, the absence of typical well-positioned nucleosomes in the PDR5 UAS region may reflect the GC-rich nature of PDR5 PDREs sequences (Gao et al. 2004). In contrast, the pdr1-3 elm1Δ strain showed significant reductions in the intensity of several bands (marked with asterisks) clustered in the region from −700 to −900, extending to ∼−1100 (Figure 8A, labels on the right) relative to the transcription start site (+1). The loss of other bands marked from −900 to −1100 in the pdr1-3 elm1Δ strain may reflect restoration of positioned nucleosome structure in this region due to reduced PDR5 transcription. These nucleosomal alterations were unexpected because this region is located well upstream of the known Pdr1-binding sites (PDREs) and TATA box, and there were no significant differences within the PDREs or TATA region for the pdr1-3 and pdr1-3 elm1Δ strains (Figure 8A). To test the generality of these observations, we performed a set of MNase mapping experiments comparing PDR1 with PDR1 elm1Δ strains and found consistent results (data not shown).
We then investigated whether PDR1 and/or PDR3 are required for regulation of nucleosome structure upstream of the PDREs. We examined the latter by using a set of isogenic strains: wild type, pdr1Δ, pdr3Δ, and pdr1Δ pdr3Δ (WCS265–WCS268, Table 1). As expected, a strong band located close to the TATA box was observed in the wild-type strain (Figure 8B; Gao et al. 2004). Neither pdr1Δ nor pdr3Δ alone significantly changed the PDR5 promoter region (including TATA box and PDREs), consistent with overlapping roles of Pdr1 and Pdr3 as transcriptional activators (Wolfger et al. 1997). Interestingly, a pdr1Δ pdr3Δ double deletion resulted in changes not only in the PDR5 promoter (Figure 8B, #) and coding sequences (bracketed in Figure 8B) as expected, but also in the region upstream of the PDREs, including sequences from −700 to −900 (Figure 8B, asterisks). The increased site-specific MNase digestion presumably reflects less dynamic nucleosome structure in the absence of PDR1 and PDR3 (Figure 8B). These data indicate that Pdr1 or Pdr3 is required for regulation of the PDR5 nucleosome structure upstream of the PDREs, corresponding to the region where the elm1Δ strain exhibited altered nucleosome structure. However, Myc-tagged Pdr1 was not recruited to the −700 to −900 region of PDR5 promoter (data not shown). This raises the possibility that PDREs occupied by Pdr1 and Pdr3 propagate altered nucleosome structure from their binding sites to sequences farther upstream. Such a long-distance effect on nucleosomal structure resulting from changes in the interactions of transcription factors and DNA has been reported previously (Fleming and Pennings 2001).
We then analyzed whether sequences upstream of the PDREs affect PDR5 transcription. We engineered a wild-type strain with the sequences from ∼ −700 to ∼ −1100 replaced by TRP1 (diagrammed in Figure 8C, the resulting strain named pdr5 promoter Δ400 strain). PDR5 transcription in the pdr5 promoter Δ400 strain was significantly reduced compared to its parental strain PDR1 under both noninduced and induced conditions. As expected, the pdr5 promoter Δ400 strain is hypersensitive to CYH (Figure 8C). Not surprisingly, a reduced mRNA level of adjacent open reading frame YOR152C was detected in the pdr5 promoter Δ400 strain; however, deletion of YOR152C in PDR1 or pdr1-3 strains did not change their resistance to drugs (data not shown). Taken together, the data presented in Figure 8 identify sequences important for PDR5 transcription located considerably upstream of the PDREs. Significantly, the nucleosome structure in this region requires ELM1 as well as PDR1/PDR3.
We then tested whether Elm1 directly regulates PDR5 transcription. We analyzed the recruitment of Myc-tagged Elm1 onto the PDR5 promoter and upstream region (up to and including the promoter of adjacent ORF YPR152C) in the PDR1 and pdr1-3 strains. In Figure 8C, the promoter region of ORF YPR152C is depicted as iYOR152-0 (Horak et al. 2002; Harbison et al. 2004). No Elm1-Myc recruitment was detected on the promoter of either PDR5 or YOR152C (data not shown). This result is not surprising since Elm1 has been shown to be located primarily, if not exclusively, at the bud neck between mother and daughter cells (Huh et al. 2003; Thomas et al. 2003). It is worth noting that the Myc-tagged Elm1 is functional in these strains on the basis of their normal cell morphology and growth. Therefore, the results suggest an indirect mechanism for Elm1 affecting PDR5 transcription.
In this study, we investigated the regulation of multidrug resistance involving the transcription of PDR5. We show that in a pdr1-3 strain, increased PDR5 transcription (Gao et al. 2004) correlates with increased levels of Pdr5 protein (Figure 1). These results are consistent with microarray analysis, revealing PDR5 as the major target in a pdr1-3 strain (DeRisi et al. 2000). These results also underscore the notion that transcriptional upregulation is the predominant mechanism for development of yeast drug resistance. This is distinct from the development of mammalian drug resistance in which mechanisms unrelated to transcription, such as gene amplification, play a crucial role in addition to transcriptional regulation (Gottesman et al. 1995).
Employing the anticancer drug doxorubicin as a substrate, we demonstrated that cellular doxorubicin elimination follows zero-order kinetics (Figure 2D and Table 2), indicative of catalysis; that is, its rate depends on the concentration of the catalyst (ABC transporters in this case) and not on the reactant (cellular doxorubicin concentration) (Tinoco et al. 1985). These findings are similar to the elimination of viblastine via P-glycoprotein, in which the results fit to Michaelis–Menten kinetics, V = Vmax[S]/K + [S]. When the concentration of substrates like vinblastine or doxorubicin (S) exceeds K and the rate (V) approaches Vmax, the efflux catalyzed by P-glycoprotein or Pdr5 will show zero-order kinetics and the number of transporters becomes limiting (Ambudkar et al. 1997). Such kinetics ensures efficient and almost complete cellular detoxification.
We demonstrated a striking inverse correlation between PDR5 mRNA levels and cellular doxorubicin accumulation during cell cycle progression (Figure 3). In these experiments, cellular doxorubicin at any given time point reflected the collective activities of many drug transporters. However, it became apparent that Pdr5 was the most important cell-cycle-regulated transporter, at least for doxorubicin and CYH. It is worth noting that, with equivalent treatments, doxorubicin content in the pdr1-3 cells during the G1 phase (2.0 pmol/107 cells, Figure 3B) was higher than that in the unsynchronized PDR1 cells (∼1.0 pmol/107 cells, Figure 2C, 15 min). Therefore, G1-synchronized pdr1-3 cells exhibited even greater sensitivity than unsynchronized PDR1 cells with respect to detoxification of doxorubicin. The generality of this observation for different drugs and to what degree cell-cycle-dependent drug sensitivity occurs in pathogenic yeast and cancer cells remain to be thoroughly investigated.
We searched for the most appropriate substrate to measure Pdr5 efflux in the various mutants. Ideally, an inert substrate is preferable. Fluconazole, an inhibitor of ergosterol biosynthesis (Kontoyiannis et al. 1999), was separated on HPLC and detected by absorbance. However, the detection sensitivity was too low (∼5 nmol). Doxorubicin, on the other hand, was detected by fluorescence; its lowest detection limit (with a signal-to-noise ratio >3) was <5 pmol.
The doxorubicin-induced DNA damage results in cell cycle arrest at the G2/M phase (Siu et al. 1999). This process is mediated by inhibiting dephosphorylation of the p34cdc2 kinase (mammalian homolog of the S. cerevisiae Cdc28) and by inducing cyclin B1 accumulation (Ling et al. 1996). Since doxorubicin effect of the cell cycle is expected to be similar in all strains studied, the loss of doxorubicin efflux in pdr1-3 cla4Δ, pdr1-3 elm1Δ, and pdr1-3 cdc29-C127Y strains (Table 2) reflects primarily the effects of the genetic mutations on G2/M transition.
We provided several lines of evidence that establish a genetic connection between PDR5 and ELM1 or ELM1-related genes required for proper mitotic progression. First, cells harboring elm1Δ exhibit G2/M delay, reduction of PDR5 transcription, and loss of CYH resistance (Figures 4 and 5). Second, epistasis analysis indicates that although the elm1Δ mutation does not increase CYH susceptibility of the pdr1-3 pdr5Δ strain, the pdr1-3 pdr5Δ elm1Δ strain becomes more sensitive to CYH than the pdr1-3 elm1Δ strain (Figure 4D). These data suggest that ELM1 functions as an upstream regulator of the PDR5-mediated CYH resistance. Third, mutations of genes causing defective mitotic progression decrease PDR5 transcription and CYH resistance (Figures 5 and 6). Fourth, doxorubicin efflux in the pdr1-3 elm1Δ, pdr1-3 cla4Δ, and pdr1-3 cdc28-C127Y strains is virtually undetectable (Table 2). The drug efflux rates in these strains are ∼2 orders of magnitude lower than that in the pdr1-3 strain (Table 2). These data provide a quantitative estimate of the functional consequences of downregulating drug transporter genes in these mutants. The expression of PDR5 and perhaps other cell-cycle-regulated transporter genes is severely diminished by these mutations. This finding is consistent with the genomewide analysis showing that the expression of several ABC transporter genes (including TPO1, TPO2, TPO3, and TPO4) peaks during the M phase, while the transcription of other well-established transporter genes peaks at other stages of the cell cycle (e.g., SNQ2 peaks at G2 and FLR1 at S phase) (Spellman et al. 1998). Fifth, elm1Δ leads to detectable alteration in nucleosome structure upstream of known PDREs at the PDR5 promoter region (Figure 8). The alteration of PDR5 nucleosome structure by elm1Δ appears independent of the recruitment of Pdr1 to the PDREs (Figure 7). Identification of a new region whose nucleosome structure undergoes conformational changes is consistent with the initial characterization of the PDR5 promoter, which indicated that sequences of ∼1.1 kb upstream of the PDR5 translation start site were required for maximal PDR5 expression (Katzmann et al. 1996).
We note that the biochemical connection between ELM1 and PDR5 transcription is most likely via an indirect mechanism. This conclusion is based on the fact that Elm1 is a key cellular regulatory kinase and predominately located at the bud neck and is not detectable on the PDR5 promoter by ChIP. The mechanism by which Elm1 regulates the nucleosome structure of the PDR5 promoter, therefore, remains to be determined. The fact that both Pdr1 and Pdr3 are phosphorylated (Mamnun et al. 2002) raises the possibility that phosphorylation of Pdr1/Pdr3 by either Elm1 or its related kinases may affect the activity of Pdr1/Pdr3. In this regard, differential phosphorylation has been proposed to regulate the activity of Gal4 (Mylin et al. 1990), the founding member of the Cys6-Zn(II) DNA-binding transcription factor family to which Pdr1 and Pdr3 belong (Poch 1997). It is interesting to note that the sequences SPVR (amino acids 942–945) and SPLK (amino acids 890–893) of Pdr1 and Pdr3, respectively, are located within the C-terminal transcription activation domains. These are consensus target sequences (S/T-P-X-K/R) for Cdc28 kinase. It is tempting to speculate that Cdc28 could participate in transcriptional regulation of PDR5 by modulating the activities of constitutively bound Pdr1/Pdr3 activators. Elm1, then, may be linked to Cdc28 activity via regulating Swe1 phosphorylation state during mitosis (Sreenivasan and Kellogg 1999). Consistent with this possibility, a pdr1-3 elm1Δ swe1Δ strain partially restores a transcriptional defect of PDR5 observed in the pdr1-3 elm1Δ strain (Figure 5B and data not shown). A potential requisite tie among Elm1 (or Cdc28), Pdr1/Pdr3 phosphorylation states, and PDR5 transcription level remains to be investigated.
The nucleosome structure affected by elm1Δ is located ∼900 bp upstream of the PDR5 transcription start site (134 nucleotides upstream of translation start codon ATG). This region encompasses the predicted promoter of the open reading frame YOR152C, which is oriented in the opposite direction of PDR5 (Figure 8C) and encodes a putative membrane-bound protein of unknown function (Terashima et al. 2002). Interestingly, microarray studies indicated that YOR152C mRNA level was significantly upregulated in the pdr1-3 strain (DeRisi et al. 2000). However, deletion of YOR152C did not affect drug resistance of PDR1 or pdr1-3 strains (our unpublished data). Further characterization of YOR152C and the divergent promoter between the PDR5 and YOR152C open reading frames should elucidate molecular aspects of their transcriptional coregulation.
The connection between cell cycle progression and drug transporter gene expression reported here is not unprecedented in other eukaryotes. For instance, in addition to regulation at DNA and mRNA levels, expression of P-glycoprotein is regulated at the protein level. The turnover of P-glycoprotein in multidrug resistant ovarian cells was shown to be cell cycle dependent (Zhang and Ling 2000). Furthermore, it was reported that colon cancer cells overexpressing P-glycoprotein showed a reduced sensitivity to doxorubicin during G2/M (Toffoli et al. 1996). It is therefore suggestive that prior synchronization of drug-resistant cells to the cell cycle stage in which the expression of drug transporters is the lowest (such as G1 for PDR5) could improve the efficacy of treatments of fungal infections and cancers.
The transcription factors that are required for cell-cycle-dependent PDR5 transcription remain to be explored. Factors known to be recruited to the PDR5 promoter (e.g., Pdr1, Pdr3, SAGA, Mediator, and SWI/SNF complexes) are potential candidates (Gao et al. 2004). Another possibility is the transcription factor Tos4, a known substrate of Cdc28 (Ubersax et al. 2003), which regulates cell cycle progression (Iyer et al. 2001) and binds to the intergenic region iYOR152C-0 (Horak et al. 2002). Moreover, the transcription factor Sok2 also binds to iYOR152C-0 (Harbison et al. 2004) and negatively regulates pseudohyphal differentiation/elongated morphology (Pan and Heitman 2000). Interestingly, the iYOR152C-0 region overlaps with the −700 to −1100 region that requires Elm1 and Pdr1/Pdr3 for its proper nucleosome structure (Figure 8).
One intriguing question raised by our studies is why a drug transporter gene such as PDR5 is specifically expressed during mitosis. It is possible that drug transporter genes regulated by cell-cycle progression may reflect physiological functions of these transporters in addition to their roles as drug transporters (Schmitt and Tampe 2002; Jungwirth and Kuchler 2006). It has been shown that steroids, important components of the cell membrane, are physiological substrates of Pdr5 (Kolaczkowski et al. 1996). Moreover, transport of phosphatidylethanolamine is shown to be controlled by the transcription regulators PDR1 and PDR3 (Kean et al. 1997). As buds grow, biosynthesis and transportation of cell membrane components increase. Drug transporters thus may facilitate proper localization of steroids and other molecules in the newly formed daughter cell membranes. Consistent with this notion, the mammalian P-glycoprotein has been shown to transport, or “flip”, short-chain lipids between the leaflets of the cell membrane (Romsicki and Sharom 2001). Interestingly, the connection among lipid metabolism, drug resistance, and cellular morphogenesis was also demonstrated by the functional analysis of the sphingolipid biosynthetic gene CaIPT1 of C. albicans, showing its involvement in both multidrug resistance and cellular morphogenesis (Prasad et al. 2005).
In conclusion, we provide genetic, kinetic, and molecular evidence that ELM1 and functionally related kinase genes are required for multidrug resistance in S. cerevisiae. The mechanism for this regulation may include alteration of the nucleosome structure upstream of the PDR5 PDREs. The proposed mechanism is valid for both pdr1-3 and PDR1.
The authors are grateful to Alan Myers and Daniel Lew for the cdc28 mutants; Kyung Lee for the septin mutant; and Karl Kuchler, Andre Goffeau, Julius Subik, and Martin Raymond for the pdr1-3 and pdr3-2 strains. We thank Bob West and Patty Kane for comments on the manuscript. Constructive input from members of the Yeast Data Club at State the University of New York (SUNY) Upstate Medical University and Syracuse University throughout this project is appreciated. A Faculty Development Fund and a Hendrick's Fund provided by SUNY Upstate Medical University to W.-C.W.S. and a grant from the Paige's Butterfly Run to A.-K.S. supported this work.
↵1 These authors contributed equally to this work.
Communicating editor: A. P. Mitchell
- Received February 24, 2006.
- Accepted May 30, 2006.
- Copyright © 2006 by the Genetics Society of America