Global Phenotypic Analysis and Transcriptional Profiling Defines the Weak Acid Stress Response Regulon in Saccharomyces cerevisiae

Abstract

Weak organic acids such as sorbate are potent fungistatic agents used in food preservation, but their intracellular targets are poorly understood. We thus searched for potential target genes and signaling components in the yeast genome using contemporary genome-wide functional assays as well as DNA microarray profiling. Phenotypic screening of the EUROSCARF collection revealed the existence of numerous sorbate-sensitive strains. Sorbate hypersensitivity was detected in mutants of the shikimate biosynthesis pathway, strains lacking the PDR12 efflux pump or WAR1, a transcription factor mediating stress induction of PDR12. Using DNA microarrays, we also analyzed the genome-wide response to acute sorbate stress, allowing for the identification of more than 100 genes rapidly induced by weak acid stress. Moreover, a novel War1p- and Msn2p/4p-independent regulon that includes HSP30 was identified. Although induction of the majority of sorbate-induced genes required Msn2p/4p, weak acid tolerance was unaffected by a lack of Msn2p/4p. Ectopic expression of PDR12 from the GAL1-10 promoter fully restored sorbate resistance in a strain lacking War1p, demonstrating that PDR12 is the major target of War1p under sorbic acid stress. Interestingly, comparison of microarray data with results from the phenotypic screening revealed that PDR12 remained as the only gene, which is both stress inducible and required for weak acid resistance. Our results suggest that combining functional assays with transcriptome profiling allows for the identification of key components in large datasets such as those generated by global microarray analysis.

INTRODUCTION

Short-chain weak organic acids such as sorbate, benzoate, and propionate are naturally occurring compounds that prevent microbial growth. They are therefore widely used in food and beverage preservation. They enter the cell by passive diffusion and dissociate at the higher intracellular pH generating protons and acid anions that accumulate. The mechanisms underlying the antimicrobial effects of these acids are still obscure but may be attributed, at least in part, to a cytoplasmic acidification that inhibits essential metabolic functions (Krebs et al., 1983; Pearce et al., 2001). Furthermore, lipophilic compounds such as sorbate and benzoate are thought to impact membrane permeability (Holyoak et al., 1999). Several genes have been implicated in the weak acid stress response and adaptation so far. The ATP-binding cassette (ABC) efflux pump Pdr12p (Piper et al., 1998) is a major factor counteracting intracellular anion accumulation. Cells lacking Pdr12p, which is highly induced upon weak acid stress, are hypersensitive to weak acids with a chain length from C3 to at least C8 (Wolfger, H., unpublished observations; Holyoak et al., 1999; Hatzixanthis et al., 2003). Moreover, the plasma membrane H⁺-ATPase Pma1p as well as the heat shock proteins Hsp30p and Hsp26p appear modulated during weak acid stress (Holyoak et al., 1996; Piper et al., 1997; de Nobel et al., 2001).

Saccharomyces cerevisiae cells can only cope with weak acid stress by inducing Pdr12p, leading to stress adaptation (Piper et al., 1998). Therefore, genes upregulated by weak acid stress are prime candidates for important resistance factors. Indeed, cells adapted for growth on sorbate (de Nobel et al., 2001) display distinct mRNA profiles, displaying enhanced expression levels of many genes. This set of genes overlaps partly with transcriptional patterns seen in the responses to several other adverse conditions (Rep et al., 1999; Gasch et al., 2000; Causton et al., 2001). Interestingly, the spectrum of genes upregulated in sorbate-adapted cells includes many genes regulated by Msn2p and Msn4p (de Nobel et al., 2001). These are the main regulators of the general stress response, and both also appear activated by sorbic and benzoic acid stress (Görner et al., 1998).

Drastic changes of environmental contexts such as exposure to stressful conditions lead to a rapid and broad transcriptional response with up to hundreds of genes induced or repressed. DNA microarray mRNA profiling methods (DeRisi et al., 1997; Gasch et al., 2000; Young, 2000; Causton et al., 2001) allow for efficient monitoring of these dynamic changes of gene expression regulation. Interestingly, global expression profiles as a consequence of a given condition or specific for a distinct cell type represent a remarkable stable diagnostic phenotype. For example, transcriptome analysis can be used to classify tumor subtypes (Golub et al., 1999) and sometimes even enable assignment of functional roles to uncharacterized genes (Hughes et al., 2000). However, many genes whose functions are entirely unrelated to a given physiological or nonphysiological condition are usually also modulated in their expression. To distinguish functionally relevant genes from “innocent bystanders,” it is therefore necessary to combine global microarray profiling with functional assays or biochemical experiments (Shoemaker et al., 1996; Hughes et al., 2000). Functional assays at the genomic scale would be most informative, because such evidence complements microarray datasets. In this work, we used a two-pronged approach to identify and characterize components important for the molecular defense against weak organic acid stress. Thus, we combined a genome-wide phenotypic screening using the EUROSCARF deletion strain collection with global transcriptome microarray analysis.

The induction of PDR12 by weak acid stress is fast and mainly regulated at the level of transcription (Kren et al., 2003). Importantly, Pdr12p induction is specific for weak acid stress but not for other stress types and is independent of Msn2p/4p (Piper et al., 1998). A novel transcription factor of the Zn₂Cys₆ family, War1p, was recently identified as a stress regulator of PDR12 (Kren et al., 2003). Cells lacking War1p are weak acid hypersensitive, suggesting that genes induced by this factor are important under these conditions. To identify the genes dependent on War1p, we exploited genome-wide mRNA profiles to determine the global transcriptional response in weak acid-stressed cells. Our results suggest War1p, Msn2p/4p, and a third pathway as major players in the weak acid stress response. We can distinguish the regulons of War1p and Msn2p/4p, and we show that these are largely nonoverlapping. Strikingly, comparison of the results from the functional assays with microarray datasets indicated PDR12 as the only gene that is both sorbate-induced and essential for stress adaptation. Although we identified many genes whose expression levels are modulated by weak acid stress, our results nicely illustrate that only a combination of functional genomics and whole-genome expression profiling can efficiently identify key players in biological response pathways.

MATERIALS AND METHODS

Yeast Strains, Culture Conditions, and Cytotoxicity Assays

All yeast strains used in this study are listed in Table 1. Rich medium (YPD) and synthetic medium (SC), supplemented with appropriate auxotrophic components, were prepared as described elsewhere (Kaiser et al., 1994). Unless otherwise indicated, all yeast strains were grown routinely at 30°C. Tests for weak acid resistance phenotypes were performed with cells grown to the exponential growth phase and diluted. Identical volumes of cultures as well as 1:10 and 1:100 serial dilutions were spotted onto agar plates containing the indicated concentrations of weak acids. Colony growth was inspected after a 48-h incubation. Gene deletions were carried out by a PCR-based method using disruption cassettes derived from plasmids pFA6a-HIS3Mx6 and pFA6a-KanMx6 (Wach et al., 1997). YCS105 (war1Δ msn2Δ msn4Δ) was obtained by crossing the isogenic strains YAK120 (war1Δ) (Kren et al., 2003) and W303-1A msn2 msn4 (Görner et al., 1998). The correct genotypes of progeny were verified by PCR analysis using appropriate primers. The plasmid YCplac22 (Gietz and Sugino, 1988) was used to deliver the TRP1 gene into the strains W303-1A, YBB14, and YAK120 for growth assays in the presence of sorbate. WAR1 was deleted from strain FY-GAl-PDR12 carrying a PDR12 gene under control of the GAL1-10 promoter (Hatzixanthis et al., 2003), using the hphMX4 cassette system (Goldstein and McCusker, 1999), resulting in strain FY-GAL-PDR12 war1Δ.

Table 1.

Yeast strains used in this study

Strains	Genotype	Source
W303-1A	MATaura3-1 leu2-3,112 his3-11,15 trp1-1 ade2-1 can1-100	Rothstein (1983)
YBB14	MATapdr12Δ::hisG-URA3-hisG (otherwise isogenic to W303-1A)	Kren et al. (2003)
YBB24	MATahsp30Δ::HIS3 (otherwise isogenic to W303-1A)	This study
YBB25	MATapdr12Δ::hisG-URA3-hisG hsp30Δ::HIS3 (otherwise isogenic to W303-1A)	This study
YAK120	MATawar1Δ::HIS3MX6 (otherwise isogenic to W303-1A)	Kren et al. (2003)
W303 msn2 msn4	MATamsn2::TRP1 msn4::HIS3 (otherwise isogenic to W303-1A)	Görner et al. (1998)
YCS105	MATamsn2::TRP1 msn4::HIS3 war1Δ::HIS3MX6 (otherwise isogenic to W303-1A)	This study
YAK2	MATaura3::PDR12-lacZ (otherwise isogenic to W303-1A)	Kren et al. (2003)
YBB27	MATaura3::PDR12-lacZ war1Δ::HIS3MX6 (otherwise isogenic to W303-1A)	This study
FY1679-28c	MATaura3-52 leu2Δ1 his3Δ200 trplΔ63	EUROSCARF Collection
FY-GAL-PDR12	MATapdr12::GAL1-10-PDR12 (otherwise isogenic to FY1679-28c)	Hatzixanthis et al. (2003)
FY-GAL-PDR12war	MATapdr12::GAL1-10-PDR12 war1Δ::HIS3 (otherwise isogenic to FY1679-28c)	This study
BY-4741	MATaura3-Δ0 his3-Δ1 leu2-Δ0 met15-Δ0	EUROSCARF Collection
BY-Deletions	GEN::kanMX4 deletions of HSP12, HSP26, HSP104, PDR12, TPO1, YBL059W, AUT7, FUN34, YKL051W, ECM39, TPO4, RPB4, GAL11, SWI3	EUROSCARF Collection

Preparation of Cell Extracts and Immunoblotting

The preparation of cell-free extracts for immunoblotting was performed exactly as described elsewhere (Egner and Kuchler, 1996). Cell lysates equivalent to 0.5 A₆₀₀ units were resolved by SDS-PAGE in a 7.5% gel and transferred to nitrocellulose membranes. Pdr12p was visualized on immunoblots using the ECL chemiluminescence detection system (Amersham Bioscience, Piscataway, NJ) and a polyclonal anti-Pdr12p antiserum (Piper et al., 1998).

β-galactosidase Measurements and RNA Methods

Cells carrying a PDR12-LacZ reporter gene were grown in YPD medium to the midexponential growth phase. Weak acids were dissolved in water as 0.5 M potassium salt stock and added to the culture medium up to desired concentrations. Cells were harvested by centrifugation after 30 min. β-galactosidase enzyme measurements were carried out exactly as described elsewhere (Rose, 1990). All experiments were done at least in triplicate. Preparation of RNA for Northern analysis and fractionation by agarose gel electrophoresis was done exactly as described previously (Wolfger et al., 1997). DNA probes were radiolabeled with a Megaprime Labeling kit according to the manufacturer (Amersham). Signals were detected by exposure to phosphor imager screens (Molecular Dynamics, Sunnyvale, CA) and analyzed with ImageQant software (Molecular Dynamics).

Phenotypic Screening

EUROSCARF (Frankfurt, Germany) deletion strains representing haploid (viable) single deletions of 4873 yeast ORFs were grown on solid YPD medium in 96-well microtiter plates. Strains were transferred into 96-well plates containing 100 μl YPD and grown overnight at 30°C. Next, cells were spotted onto YPD, pH 4.5, plates containing 1 mM sorbic acid and YPD pH4.5 reference plates, using a sterile spotting device. Growth was inspected after a 48-h incubation and compared with the reference plate to identify sensitive strains, all of which were subjected to a second screening round to avoid spotting artifacts.

DNA Microarray Profiling Experiments

Culture conditions: Cells from overnight cultures in YPD were diluted in fresh YPD to an OD₆₀₀ of 0.1 and grown at 30°C until an OD₆₀₀ of 1-1.1 was reached. Cultures were split and potassium sorbate (Sigma St. Louis, MO) was added at a final concentration of 8 mM to one half of the culture. After 20 min, both untreated and treated cultures were harvested by centrifugation at room temperature (2 min, 4000 × g), and cells were immediately washed in ice-cold water, reharvested at 4°C, and frozen at -80°C. Total RNA was prepared by the hot phenol method and quantified by spectrophotometry at 260 nm in TE buffer exactly as described elsewhere (Kren et al., 2003). For cDNA synthesis, 20 μg of total RNA from treated and untreated cells was converted to cDNA using 200 U SuperscriptII reverse transcriptase (Invitrogen, Carlsbad, CA). Reactions included either Cy3dCTP or Cy5dCTP (Amersham Biosciences). Labeled cDNAs were pooled and RNA was hydrolyzed for 20 min in 50 mM NaOH at 65°C, neutralized with acetic acid, precipitated with an equal volume of isopropanol, washed briefly in 70% ethanol, and finally dissolved in 5 μl sterile distilled water. Glas microarrays (OCI Toronto; http://www.microarrays.ca/) containing PCR fragments of 6144 predicted S. cerevisiae ORFs spotted in duplicates were used for the expression profiling. Hybridization was done in a total volume of 60 μl in DigEasyHyb solution (Roche Diagnostics, Basel, Switzerland) with 0.1 mg/ml salmon sperm DNA (Sigma) as carrier solution at 37°C for 14-16 h. Microarrays were disassembled, washed three times in 1× SSC, and 0.1% SDS at 50°C for 15 min, followed by a 1-min wash in 1× SSC at room temperature. Slides were spun-dry for 5 min at 500 rpm in a table-top centrifuge at room temperature, scanned on an Axon4000B scanner (Axon Instruments, Foster City, CA) and analyzed using the GenePix Pro4.1 software (Axon Instruments). Microarrays and experimental protocols were obtained from the Ontario Cancer Institute (OCI Toronto; http://www.microarrays.ca/).

Computational Analysis

Microarray data were collected and analyzed with GenePix Pro4.1 (Axon Instruments). The entire raw dataset is available at http://www.at.embnet.org/molg/kuchler/repository/. Spots not found automatically by GenePix Pro4.1 because of low signal intensities were flagged absent and subsequently excluded from further analysis. Microarrays were normalized using GenePix Pro4.1 such that the sum of the Median of Ratios of unflagged values was 1. Median of Ratios of individual experiments were filtered using conventional spreadsheets. Median of Ratios of the wild-type, war1, msn2/4, and war1 msn2/4 sets were log-transformed for averaging, omitting flagged values and ORFs with less than four values. ORFs with least one average Median of Ratios value of induction to repression above two in all four datasets were further analyzed. The assignment of ORFs into clusters that are regulated by WAR1, MSN2/4, or both was based on the induction relative to the wild-type control using a cutoff value of 0.6 (see Figure 2). Furthermore, clustering was performed using the cluster3 software (Eisen et al., 1998). Uncentered correlation and complete linkage for genes and arrays, and subsequent visualization with Java TreeView revealed similar groups (see Figure 2). Detection of short common sequences in promoters was done with AlignACE version 2.2, (Roth et al., 1998), using 800 base pairs of upstream sequence, sampling in a maximum window of 30, using 10 columns, 10 expected elements, including the following input ORFs: YNR002C, YPL058C, YPL122C, YNR030W, YOL032W, and YGR046W. A logo generator (http://weblogo.berkeley.edu/logo.cgi) was used for graphical presentation of the sequence logo (Schneider and Stephens, 1990).

Figure 2.

Clustering of transcript profiles. Hierarchial clustering of induction values representing average inductions over three replicate profiles of each the wild-type (W303-1A) and the isogenic derivatives (war1Δ, msn2Δ msn4Δ, war1Δ msn2Δ msn4Δ) after 20 min of 8 mM potassium sorbate treatment. Color codes refer to different regulatory groups after selection of differently regulated genes as revealed by ratio discrimination.

RESULTS

To identify nonessential genes required for growth in the presence of sorbic acid, we used the haploid deletion strains of the EUROSCARF collection (EUROSCARF; http://www.uni-frankfurt.de/fb15/mikro/euroscarf/). Strains were grown overnight in microtiter plates in liquid YPD and spotted onto YPD plates, pH 4.5, containing 1 mM sorbic acid as well as YPD control plates lacking sorbate. After incubation for 48 h, sorbate sensitivity was scored. Eighty-six of 4827 strains tested, reproducibly showed impaired growth in the presence of sorbate (Table 2). Furthermore, this experiment has since been repeated independently with the corresponding homozygous diploid mutants from the EUROSCARF collection using higher sorbate concentrations (M. Mollapour, unpublished results), revealing a somewhat larger set of 194 genes, of which our set represents a subgroup. The results of the 1 mM sorbate screen at pH 4.5 are listed in Table 2, including the relevant gene ontology assignments (GO) listed in the Saccharomyces genome database (ftp://genome-ftp.stanford.edu/yeast/data_download/literature_curation/orf_geneontology.tab).

Table 2.

List of ORFs implicated in weak acid resistance listed by canonical name

ORF name	Gene	Pathway	Biological function
YAL016w	TPD3	Protein biosynthesis	Protein phosphatase type 2A
YBL079w	NUP170	mRNA-nucleus export	Structural molecule
YBR036c	CSG2	Calcium ion homeostasis	Molecular function unknown
YBR096w		Biological process unknown	Molecular function unknown
YBR127c	VMA2	Vacuolar acidification	Hydrogen-transporting two-sector ATPase
YBR132c	AGP2	Response to osmotic stress	Hydrogen:amino acid symporter
YBR133c	HSL7	G2/M transition of mitotic cell cycle	Protein-arginine N-methyltransferase
YBR173c	UMP1	Protein catabolism	Chaperone
YBR267w		Biological process unknown	Molecular function unknown
YBR282w	MRPL27	Protein biosynthesis	Structural constituent of ribosome
YCR081w	SRB8	Negative regulator of Pol-II transcription	RNA polymerase II transcription mediator
YDL005c	MED2	Transcription from Pol-II promoter	RNA polymerase II transcription mediator
YDR007w	TRP1	Tryptophan biosynthesis	Phosphoribosylanthranilate isomerase
YDR017c	KCS1	Stress response	Protein kinase
YDR027c	LUV1	Not yet annotated	Molecular function unknown
YDR069c	DOA4	Deubiquitination	Ubiquitin-specific protease
YDR173c	ARG82	Arginine metabolism	Inositol/phosphatidylinositol kinase
YDR364c	CDC40	Not yet annotated	Molecular function unknown
YDR484w	SAC2	Actin filament-based process	Molecular function unknown
YER017c	AFG3	Protein complex assembly	Adenosinetriphosphatase
YER044c	ERG28	Ergosterol biosynthesis	Molecular function unknown
YER083c	RMD7	Cell wall organization and biogenesis	Molecular function unknown
YER090w	TRP2	Tryptophan biosynthesis	Anthranilate synthase
YER169w	RPH1	DNA repair	Not yet annotated
YFL017w-a	SMX2	mRNA splicing	Pre-mRNA splicing factor
YGL026c	TRP5	Tryptophan biosynthesis	Tryptophan synthase
YGL070c	RPB9	Transcription from Pol-II promoter	DNA-directed RNA polymerase II
YGL148w	ARO2	Not yet annotated	Chorismate synthase
YGL173c	KEM1	35S primary transcript processing	5′-3′ Exoribonuclease
YGR063c	SPT4	DNA-dependent transcription regulation	Pol-II transcription elongation factor
YGR064w		Biological process unknown	Molecular function unknown
YGR076c	MRPL25	Protein biosynthesis	Structural constituent of ribosome
YGR104c	SRB5	Transcription from Pol-II promoter	RNA polymerase II transcription mediator
YGR112w	SHY1	Respiratory gaseous exchange	Molecular function unknown
YGR135w	PRE9	Ubiquitin-dependent protein catabolism	Proteasome endopeptidase
YGR159c	NSR1	rRNA processing	RNA binding
YGR162w	TIF4631	Translational initiation	Translation initiation factor
YGR222w	PET54	Protein biosynthesis	Molecular function unknown
YGR229c	SMI1	Cell wall organization and biogenesis	Not yet annotated
YGR240c	PFK1	Glycolysis	6-Phosphofructokinase
YGR272c		Biological process unknown	Molecular function unknown
YHR060w	VMA22	Protein complex assembly	Molecular function unknown
YIL018w	RPL2B	Protein biosynthesis	Structural constituent of ribosome
YJL063c	MRPL8	Protein biosynthesis	Structural constituent of ribosome
YJL075c		Biological process unknown	Molecular function unknown
YJL140w	RPB4	Transcription from Pol-II promoter	DNA-directed RNA polymerase II
YJL176c	SWI3	Chromatin modeling	RNA polymerase II transcription factor
YJL189w	RPL39	Protein biosynthesis	Structural constituent of ribosome
YKL169c		Biological process unknown	Molecular function unknown
YKL212w	SAC1	Dephosphorylation	Inositol/phosphatidylinositol phosphatase
YLR044c	PDC1	Ethanol fermentation	Pyruvate decarboxylase
YLR056w	ERG3	Ergosterol biosynthesis	C-5 sterol desaturase
YLR148w	PEP3	Non-selective vesicle docking	Protein binding
YLR399c	BDF1	Sporulation (sensu Saccharomyces)	Transcription factor
YLR402w		Biological process unknown	Molecular function unknown
YML008c	ERG6	Ergosterol biosynthesis	Delta(24)-sterol C-methyltransferase
YML061c	PIF1	DNA recombination	DNA helicase
YML076c	WAR1	Biological process unknown	Molecular function unknown
YMR060c	TOM37	Mitochondrial translocation	Protein transporter
YMR183c	SSO2	Non-selective vesicle fusion	t-SNARE
YMR202w	ERG2	Ergosterol biosynthesis	C-8 sterol isomerase
YMR205c	PFK2	Glycolysis	6-Phosphofructokinase
YMR231w	PEP5	Non-selective vesicle docking	Molecular function unknown
YNL084c	END3	Cytokinesis	Cytoskeletal adaptor
YNL225c	CNM67	Microtubule nucleation	Structural constituent of cytoskeleton
YNR032w	PPG1	Not yet annotated	Protein serine/threonine phosphatase
YNR037c	RSM19	Protein biosynthesis	Structural constituent of ribosome
YOL023w	IFM1	Translational initiation	Translation initiation factor
YOL051w	GAL11	Transcription from Pol-II promoter	RNA polymerase II transcription mediator
YOL086c	ADH1	Fermentation	Alcohol dehydrogenase
YOL143c	RIB4	Vitamin B2 biosynthesis	Riboflavin synthase
YOR235w		Biological process unknown	Molecular function unknown
YOR291w		Biological process unknown	Molecular function unknown
YPL042c	SSN3	Protein amino acid phosphorylation	RNA polymerase II transcription factor
YPL045w	VPS16	Protein-vacuolar targeting	Molecular function unknown
YPL058c	PDR12	Transport	Xenobiotic-transporting ATPase
YPL059w	GRX5	Response to osmotic stress	Thiol-disulfide exchange intermediate
YPL097w	MSY1	Not yet annotated	Tyrosine-tRNA ligase
YPL193w	RSA1	Ribosomal large subunit assembly	Molecular function unknown
YPL268w	PLC1	Pseudohyphal growth	1-PI-4,5-bisphosphate phosphodiesterase
YPR067w	ISA2	Iron transport	Molecular function unknown
YPR100w		Biological process unknown	Molecular function unknown

The mutant strains identified by our phenotypic analysis are involved in diverse cellular functions (Table 2). A major group belongs to energy metabolism, including genes such as phosphofructokinase alpha and beta subunits (PFK1, PFK2), as well as pyruvat decarboxylase 1 (PDC1) and alcohol dehydrogenase (ADH1). Another group appears involved in mitochondrial function, although no general effect of petite mutants was evident. Interestingly, several genes of the ergosterol biosynthesis pathway (ERG2,3,6,28) were identified, perhaps suggesting that perturbations of membrane ergosterol can lead to dramatic sorbate hypersensitivity. Interestingly, strains lacking genes involved in the aromatic amino acid biosynthesis (e.g., TRP1, TRP2, TRP5, ARO2) also scored as sorbate sensitive. However, none of them was required for, or implicated in, transcriptional PDR12 stress regulation (our unpublished results). Further, several mutants affected in RNA polymerase II (Pol-II)-dependent transcription were also identified, including SWI3, SSN6, MED2, SRB5, and GAL11, as well as RPB9 and RPB4, two subunits of Pol-II. Notably, RPB4 was previously shown to be required for heat and starvation stress tolerance (Choder, 1993). The sorbate hypersensitivity of these mutants emphasizes the importance of major transcriptional adjustments to elicit weak acid resistance and stress adaptation. Finally, weak organic acids might also cause oxidative stress (Piper, 1999; de Nobel et al., 2001), consistent with the sorbate sensitivity of mutants lacking the GRX5 mitochondrial glutaredoxin, an important determinant of oxidative stress resistance (Rodriguez-Manzaneque et al., 1999).

To identify genes directly involved in PDR12 regulation, we also screened all 86 hypersensitive strains (Table 2) present in EUROSCARF collection for Pdr12p induction using immunodetection of Pdr12p levels after sorbate challenge (Piper et al., 1998). Interestingly, war1Δ was the only mutant that completely lacked Pdr12p induction upon sorbate stress (Figure 1C). War1p is a Zn(II)₂Cys₆ transcription factor recently identified as a dedicated stress regulator of PDR12 (Kren et al., 2003). Notably, we found also a significantly reduced amount of Pdr12p protein levels in the rpb4Δ mutant strain, suggesting a transcriptional defect as a cause of its increased sorbate sensitivity (Figure 1C). In summary, our functional genetic analysis revealed a plethora of different cellular activities reflecting the pleiotropic impact weak acid stress can exert on yeast cells. However, a sensitivity phenotype of mutants does not imply that the encoded wild-type gene is required for a weak acid-specific detoxification process, but merely indicates the inability to oppose weak acid stress.

Figure 1.

(A) FUN34 is induced by sorbic acid in a War1p-dependent manner. Northern analysis of FUN34 under sorbate stress. Sorbic acid, 10 mM, was added to exponentially growing cultures of W303-1A and YAK120 (war1Δ), mRNA was prepared from aliquots taken at the indicated time points. Blots were probed for FUN34 and HSP30 mRNAs, methylene blue-stained rRNA served as a loading control. (B) Sequence logo of a putative consensus War1p DNA-binding motif. An alignment of related sequences detected by AlignACE in the promoter regions of War1p-dependent genes is displayed as a sequence logo. (C) Pdr12p levels are strongly reduced in rpb4Δ mutants. Wild-type and the indicated mutant strains were grown in YPD to the exponential growth phase, stressed with 10 mM sorbic acid and Pdr12p levels were assayed on Western blots. A cross-reacting band served as a loading control.

War1p Activates a Small Msn2p/4p-independent Regulon

S. cerevisiae can adapt to weak organic acids in a process requiring rapid adjustment of its transcriptional program. To identify potentially protective genes induced by sorbate, we determined the whole-genome transcriptional profile during acute stress conditions. Wild-type cells were cultivated to the early exponential growth phase and treated with 8 mM sorbate for 20 min. The isolated mRNA was transcribed to fluorescent-labeled cDNA by incorporation of Cy3- or Cy5-conjugated dCTP and hybridized to Saccharomyces cerevisae whole-genome microarrays. Each experiment was done at least in triplicate, revealing a set 100 representative genes that were significantly induced during short-term exposure to sorbate (Tables 3, 4, 5, 6, 7).

Table 3.

Selection of genes induced by sorbic acid: WAR1-dependent

Gene	ORF name	Biological process	Wild type	SDEV	war1 Δ	SDEV	msn2 Δ msn4 Δ	SDEV	war1 Δ msn2/4 Δ	SDEV	a	STRE	b
PDR12c	YPL058C	Transport	7.09	2.13	0.92	0.26	14.25	9.45	1.55	0.29	a
TFB2	YPL122C	Transcription initiation Pol-II	3.18	1.33	1.03	0.18	4.16	0.86
FUN34c	YNR002C	Transport	2.60	0.58	1.02	0.05					a		b
ECM39c	YNR030W	Protein amino acid glycosylation	2.24	0.31	1.27	0.29	4.39	2.46	1.34	0.20

Values are x-fold induction. STRE: STRE-like sequences detected in the promoter region (Treger et al., 1998; Moskvina et al., 1998).

^a Genes induced in sorbic acid-adapted cells (de Nobel et al., 2001).

^b Genes assigned MSN2/4-dependent by acidic stress (Causton et al., 2001).

^cPutative War1p consensus element (5′-C G G c/t T g/ct/g T A a/t-3′). The entire set of genes in each cluster are found in the supplementary material.

Table 4.

Genes induced by sorbic acid: MSN2/MSN4-dependent

Gene	ORF name	Biological process	Wild type	SDEV	war1Δ	SDEV	msn2Δ msn4Δ	SDEV	war1Δ msn2/4Δ	SDEV	a	STRE	b
SPI1	YER150W	Biological process unknown	8.19	5.77	6.16	1.43	1.60	0.37	1.42	0.74	a		b
	YHR087W	Biological process unknown	6.48	5.00	8.58	4.89	1.04	0.10	0.80	0.25			b
DDR2	YOL053C-A	Response to stress	5.23	1.46	3.71	2.91			1.16	0.28		STRE
HSP26	YBR072W	Response to stress	5.22	4.89	7.56	1.63	2.25	2.79	0.89	0.39	a	STRE	b
HXK1c	YFR053C	Fructose metabolism	4.95	3.10	5.81	3.82	1.05	0.34	0.93	0.27		STRE	b
GRE3	YHR104W	Response to stress	4.50	3.04	5.18	3.03			1.47	0.76			b
PHM7	YOL084W	Biological process unknown	4.43	3.48	2.99	2.56	1.66	0.15	1.03	0.38
CTT1	YGR088W	Response to stress	4.21	1.51	5.91	0.23			1.21	1.27	a	STRE	b
	YER067W	Biological process unknown	4.13	0.22	4.16	2.28			0.85	0.38			b
TFS1c	YLR178C	Regulation of proteolysis	4.10	1.91	2.52	1.97	1.29	0.67	1.50	0.50		STRE	b
ACH1c	YBL015W	Acetate metabolism	3.76	1.41	3.04	1.01	1.98	2.26	1.06	0.75		STRE
ALD3	YMR169C	Response to stress	3.61	3.14	2.65	3.45			0.93	0.51	a
HXT6	YDR343C	Hexose transport	3.34	0.72	5.13	2.61	0.43	0.17	0.46	0.09	a
	YOR173W	Biological process unknown	3.27	2.84	4.75	2.12			0.80	0.43		STRE	b
CYC7	YEL039C	Electron transport	2.98	1.80	3.57	2.51	1.19	0.84	1.29	0.35		STRE	b
UGP1	YKL035W	Glycosylationc	2.94	1.28	2.11	2.46	1.07	0.21	0.95	0.22
HXT7	YDR342C	Hexose transport	2.87	0.31	1.96	2.37	0.47	0.16	0.55	0.08	a
NCE103	YNL036W	Biological process unknown	2.81	2.66	2.57	2.90	1.14	0.32	1.38	0.42
SSE2	YBR169C	Protein folding	2.81	1.79	1.81	1.65			1.48	1.22			b
GPX1	YKL026C	Response to oxidative stress	2.45	2.16	1.67	2.17	1.00	0.27				STRE
OSH2	YDL019C	Steroid biosynthesis	2.42	1.25	1.45	0.37	1.21	0.13	1.08	0.28
	YFL030W	Glyoxylate cycle	2.40	1.14	2.02	0.80	0.97	0.16	1.29	0.73			b
GAT1c	YFL021W	Transcription initiation Pol-II	2.30	1.27	1.89	1.31	0.94	0.35	0.73	0.11
	YPL004C	Biological process unknown	2.27	0.66	2.54	1.42	1.35	0.37	1.24	0.26		STRE	b
MCR1	YKL150W	Response to oxidative stress	2.26	1.20	1.49	0.93			1.15	0.56	a	STRE	b
MSS11	YMR164C	Pseudohyphal growth	2.24	2.05	3.34	0.67	0.73	0.33	0.83	0.17
HXT3	YDR345C	Hexose transport	2.22	0.24	1.81	0.77	0.63	0.18	0.49	0.12
	YLR251W	Biological process unknown	2.17	0.63	1.58	0.35	1.25	0.38	1.28	0.66			b
HXT4	YHR092C	Hexose transport	2.16	0.31	2.56	0.52	1.12	1.13	1.04	0.63
PMC1	YGL006W	Calcium ion transport	2.07	0.77	1.92	0.79	1.20	0.49	1.10	0.57			b
YPS6	YIR039C	Biological process unknown	2.07	0.75	2.49	0.61			1.11	0.25
PDR15	YDR406W	Transport	2.06	0.48	1.80	0.36	0.64	0.32	0.56	0.16			b
	YGR043C	Biological process unknown	2.02	1.06	1.91	0.60	1.18	0.60	0.91	0.43		STRE
	YNR065C	Biological process unknown	2.01	0.23	1.67	0.46			1.01	0.29
TRS120	YDR407C	ER to Golgi transport	1.86	0.36	2.23	0.18			0.85	0.32

Values are x-fold induction. STRE: STRE-like sequences detected in the promoter region (Treger et al., 1998; Moskvina et al., 1998).

^a Genes induced in sorbic acid-adapted cells (de Nobel et al., 2001).

^b Genes assigned MSN2/4-dependent by acidic stress (Causton et al., 2001).

^cPutative War1p consensus element (5′-C G G c/t T g/ct/g T A a/t-3′). The entire set of genes in each cluster are found in the supplementary material.

Table 5.

Genes induced by sorbic acid: MSN2/MSN4- and WAR1-dependent

Gene	ORF name	Biological process	Wild type	SDEV	war1Δ	SDEV	msn2Δ msn4Δ	SDEV	war1Δ msn2/4Δ	SDEV	a	STRE	b
GLK1	YCL040W	Carbohydrate metabolism	3.66	1.07	5.22	2.96	2.27	1.70	1.74	0.47		STRE	b
NCE102	YPR149W	Protein secretion	3.31	0.23	3.13	0.29	2.03	0.90	1.58	0.45		STRE
	YCL042W	Biological process unknown	2.83	0.88	2.82	0.25	2.44	2.10	1.22	0.63
ALD4c	YOR374W	Ethanol metabolism	2.66	0.42	1.77	0.48	1.71	0.89	0.94	0.27	a	STRE
	YNL335W	Dubious ORF	2.46	0.78	2.47	1.06	1.55	1.47
TPS1	YBR126C	Response to stress	2.32	0.84	2.76	0.80	1.42	0.73	1.17	0.24		STRE	b
RPA135	YPR010C	Transcription Pol-I	2.30	1.69	3.16	1.52	2.30	1.13	1.08	0.52
	YNL224C	Biological process unknown	2.16	1.15	1.98	1.22	1.90	0.50	1.25	0.66
	YBR134W	Dubious ORF	2.09	0.47	1.90	0.41	1.39	0.92	0.93	0.25
LSM4	YER112W	Nuclear mRNA splicing	2.08	0.43	1.78	0.31	2.06	1.40	0.74	1.01
	YMR090W	Biological process unknown	2.06	0.55	1.73	0.57	1.51	0.80	0.88	0.39			b
TPS3	YMR261C	Response to stress	2.04	1.18	1.90	0.14	1.68	1.46	1.00	0.41		STRE	b
GPD1	YDL022W	Accumulation of glycerol	1.98	1.23	3.81	1.06	1.48	1.06	0.76	0.21			b

Values are x-fold induction. STRE: STRE-like sequences detected in the promoter region (Treger et al., 1998; Moskvina et al., 1998).

^a Genes induced in sorbic acid-adapted cells (de Nobel et al., 2001).

^b Genes assigned MSN2/4-dependent by acidic stress (Causton et al., 2001).

^cPutative War1p consensus element (5′-C G G c/t T g/c t/g T A a/t-3′). The entire set of genes in each cluster are found in the supplementary material.

Table 6.

Genes induced by sorbic acid: MSN2/MSN4- and WAR1-independent

Gene	ORF name	Biological process	Wild type	SDEV	war1Δ	SDEV	msn2Δ msn4Δ	SDEV	war1Δ msn2/4Δ	SDEV	a	STRE	b
HSP30	YCR021C	Response to stress	11.88	6.42	10.35	1.86	4.82	1.75	8.89	6.75
	YER053C	Biological process unknown	5.01	4.88	6.71	3.06	3.05	2.40	4.93	2.38			b
YGP1	YNL160W	Response to stress	4.48	0.92	5.00	1.97	4.92	2.92	6.75	2.95
YPC1	YBR183W	Ceramide metabolism	4.15	1.80	3.27	1.53	3.31	1.95	2.57	1.12		STRE	b
YRO2	YBR054W	Biological process unknown	3.97	0.52	2.67	2.08	9.63	6.61	6.55	2.63
CIT1	YNR001C	Tricarboxylic acid cycle	3.52	0.55	4.09	2.11	4.81	4.41	4.90	2.39		STRE
	YDR533C	Biological process unknown	3.07	0.55	2.31	1.77	2.04	1.12	2.39	0.79	a
	YBR053C	Biological process unknown	2.91	1.52	2.97	0.82	5.32	2.41	5.42	2.40			b
ARO10	YDR380W	Leucine catabolism	2.61	2.09	2.20	2.12	5.46	3.83	5.21	2.57
SDH2	YLL041C	Tricarboxylic acid cycle	2.46	0.89	1.58	0.50	1.93	1.09	1.58	0.81
	YMR181C	Biological process unknown	2.37	1.73	1.78	0.23	1.75	0.74	2.49	1.12		STRE
YPS1	YLR120C	Protein processing	2.17	0.75	1.89	1.00	3.00	0.85	2.85	0.38
	YNL305C	Biological process unknown	2.11	0.14	2.01	0.12	1.53	0.44	1.72	0.38
TPO2	YGR138C	Polyamine transport	2.09	0.29	1.85	0.37	6.89	3.68	9.73	1.78
YDC1	YPL087W	Response to heat	2.08	0.14	1.61	0.21	3.14	1.37	2.94	0.73			b
PDC5	YLR134W	Pyruvate metabolism	2.00	0.52	1.97	0.67	3.83	4.69	2.33	0.73
SFK1	YKL051W	Actin cytoskeleton	2.00	0.30	1.37	0.27	3.26	1.04	5.25	1.68
BDH1	YAL060W	Butanediol fermentation	1.95	0.98	3.42	1.63	2.66	2.53	1.79	0.64
TIP1	YBR067C	Cell wall organization	1.91	0.25	2.18	0.21	1.53	0.63	2.37	0.83	a
TPO4	YOR273C	Polyamine transport	1.91	1.28	1.62	0.53	1.73	0.71	2.43	1.04			b
SDH4	YDR178W	Tricarboxylic acid cycle	1.85	0.06	2.28	0.48	1.31	0.80	1.43	0.43

Values are x-fold induction. STRE: STRE-like sequences detected in the promoter region (Treger et al., 1998; Moskvina et al., 1998).

^a Genes induced in sorbic acid-adapted cells (de Nobel et al., 2001).

^b Genes assigned MSN2/4-dependent by acidic stress (Causton et al., 2001).

^c Putative War1p consensus element (5′-C G G c/t T g/c t/g T A a/t-3′). The entire set of genes in each cluster are found in the supplementary material.

Table 7.

Genes induced by sorbic acid: “No-category”

Gene	ORF name	Biological process	Wild type	SDEV	war1Δ	SDEV	msn2Δ msn4Δ	SDEV	war1Δ msn2/4Δ	SDEV	a	STRE	b
RTN2	YDL204W	Biological process unknown	5.55	2.77	3.16	2.23			1.08	0.69	a		b
CAK1	YFL029C	Protein phosphorylation	3.95	1.72	2.15	1.28	1.01	0.33	1.01	0.29
	YAL061W	Biological process unknown	3.78	2.67	2.04	0.95	1.94	0.95	0.83	0.92		STRE	b
	YDL222C	Cell wall organization	3.37	0.89	1.87	1.24	1.42	1.59	0.75	0.47
TPO1c	YLL028W	Polyamine transport	3.25	1.81	1.81	0.79	4.92	1.64	3.48	0.96
ARO9	YHR137W	Arom. amino acid metabolism	2.87	0.64			10.26	3.49	6.54	1.93
GAD1	YMR250W	Response to oxidative stress	2.80	1.12			1.85	0.93	1.06	0.70			b
RAD9	YDR217C	DNA repairc	2.65	1.24	1.55	0.66	1.29	0.69	1.46	0.67
	YOR161C	Biological process unknown	2.61	0.53	1.43	1.36	0.91	0.59	0.58	0.11			b
CBP2	YHL038C	Group I intron splicing	2.57	2.36	0.91	0.13
	YGR248W	Biological process unknown	2.50	1.58	1.81	0.97							b
	YKL036C	Dubious ORF	2.40	1.29	1.64	1.24
HOT1	YMR172W	Transcription factor	2.36	1.34	2.03	2.01
	YOR052C	Biological process unknown	2.35	1.31			1.13	0.35	1.17	0.30
YJU2	YKL095W	Biological process unknown	2.34	1.57	1.67	0.79
	YNL194C	Biological process unknown	2.23	0.72	1.63	0.40						STRE
	YNL200C	Biological process unknown	2.18	0.41			1.21	0.17	1.51	0.55			b
	YOL048C	Biological process unknown	2.18	0.47	1.36	0.43			1.72	0.55			b
OM45	YIL136W	Biological process unknown	2.16	0.98	2.48	1.51							b
	YOL032W	Biological process unknown	2.14	1.66	1.19	0.32			1.12	0.29
SSA4	YER103W	Response to stress	2.12	1.97			2.24	2.21	1.42	0.38		STRE	b
	YBR116C	Dubious ORF	1.99	0.47			3.50	3.12	1.66	0.94	a
TPO3	YPR156C	Polyamine transport	1.84	0.65			3.33	4.32	4.51	3.81
	YGR046W	Biological process unknown	1.84	1.00	1.11	0.08	2.18	2.14
	YPL247C	Biological process unknown	1.83	0.53	2.42	0.91						STRE
BAG7	YOR134W	Signal transduction	1.80	0.74	2.03	0.14			1.18	0.47		STRE
YPR157W	YPR157W	Biological process unknown	1.80	0.63	1.52	0.71			4.72	3.47

Values are x-fold induction. STRE: STRE-like sequences detected in the promoter region (Treger et al., 1998; Moskvina et al., 1998).

^a Genes induced in sorbic acid-adapted cells (de Nobel et al., 2001).

^b Genes assigned MSN2/4-dependent by acidic stress (Causton et al., 2001).

^cPutative War1p consensus element (5′-C G G c/t T g/ct/g T A a/t-3′). The entire set of genes in each cluster are found in the supplementary material.

Additional experiments were performed with propionate-treated (30 mM) cells to exclude a possible sorbate-specific response as opposed to a general weak acid response. The sets of genes induced by sorbate and propionate were very similar, except for a less pronounced but detectable induction of Msn2p/4p-dependent genes by propionate (our unpublished results). A slightly acidic medium favors the protonated form of these weak acids (pK_a 4.6-4.7) and yeast is more sensitive to weak acids at pH 4.5. However, 1 mM sorbic acid in pH 4.5 medium resulted in the same pattern of induced genes, indicating that the intracellular weak acid accumulation is causing these changes rather than the pH of the growth medium (our unpublished results). Notably, the pattern of sorbate-inducible genes during short-term stress was quite distinct from the one reported earlier, derived from sorbate-adapted cells (de Nobel et al., 2001). Interestingly, the pattern of rapidly induced genes showed a striking resemblance to the regulons modulated by the environmental stress response (Gasch et al., 2000; Causton et al., 2001).

To distinguish target genes of the general stress response pathway, which requires the transcription factors Msn2p/Msn4p (Martinez-Pastor et al., 1996; Gasch and Eisen, 2002), from the War1p-dependent weak acid-specific regulon (Kren et al., 2003), we repeated the microarray analysis in msn2Δ msn4Δ, war1Δ, and war1Δ msn2Δ msn4Δ mutants. The assignment into different clusters representing tentative regulons was done manually as well as by cluster analysis (Figure 2). We choose a cutoff value of 0.6 for the ratio of the respective induction factors between war1, msn2/4, war1 msn2/4, and the wild-type. Following this scheme, War1p-dependent genes are required to have a fold induction of <0.6 of the wild-type but >0.6 of the msn2/4 mutant (Figure 2). A permutation of this scheme yielded five gene clusters: War1p-dependent, Msn2/4p-dependent, War1p- and Msn2/4p-dependent, and War1p- and Msn2/4p-independent (HSP30) as well as “No-category.” Except for a small group of four genes including PDR12 (4%), the induction of most weak acid-inducible genes was unaffected by the absence of War1p. However, a large set with 35 genes (35%) was highly dependent upon Msn2p/4p. Using a war1Δ msn2Δ msn4Δ mutant strain (YCS105), we found 13 genes apparently coregulated by War1p and Msn2p/4p. Out of the remaining genes, 21 were assigned to a group regulated by neither War1p nor Msn2p/4p, thus forming an unexpected and as yet unknown regulon that included HSP30. Hence these results indicate the existence of a third weak acid sensing system besides War1p and Msn2p/4p.

Alternatively, we also used hierarchial clustering (Eisen et al., 1998) and found a very similar distribution of the coregulated genes in different branches (Figure 2). Genes from the unassigned group mainly clustered with the HSP30 branch, whereas genes from the War1p Msn2/4p-dependent group clustered to the Msn2/4p branch. However, War1p-dependent genes were clearly separated from the other groups (Figure 2). Northern analysis also demonstrated that induction of HSP30 mRNA by sorbic acid is independent from War1p and partly independent from Msn2p/4p (see Figure 6A). Furthermore, the expression of 27 genes could not be placed in a regulatory context (Tables 3, 4, 5, 6, 7). The 25 genes previously identified in sorbate-adapted cells (de Nobel et al., 2001) did not show a bias of distribution toward Msn2p/4p- or War1p-dependent genes. To confirm the microarray-based results, we tested several representative genes such as FUN34 and HSP30 by Northern analysis. Indeed, both genes were induced by sorbic acid stress, and as predicted, only FUN34 induction required War1p (Figure 1A).

Figure 6.

Northern analysis of sorbic acid stress. (A) Exponentially growing cells of the indicated background (W303-1A, YAK120 (war1Δ), W303 msn2 msn4 (msn2Δ msn4Δ), and YCS105 (war1Δ msn2Δ msn4Δ) were stressed for 20 min with 10 mM sorbic acid. Strains W303-1A and YAK120 carried a TRP1 gene on the centromeric plasmid YCplac22. mRNA levels of HSP12, HSP30, and PDR12 were determined by simultaneous Northern hybridization. Methylene blue staining of rRNA served as a loading control. (B) HSP30 does not modulate sorbate sensitivity of pdr12Δ cells. Strains lacking the HSP30 and PDR12 genes in all combinations were tested for growth phenotypes on sorbic acid medium pH 4.5. Cells were spotted onto YPD, pH 4.5, with and without 0.3 mM sorbic acid and plates inspected after a 48-h incubation at 30°C.

Finally, several genes were strongly and rapidly repressed by weak acid treatment (Table 8). However, the repression was independent of War1p and Msn2p/4p, indicating that these factors can only function as positive regulators. Interestingly, a number of genes encoding transporters and major facilitators were also strongly repressed (Table 8). For instance, repressed genes included those involved in cellular uptake of different nutrients and metals, such as phosphate (PHO84), zinc (ZRT1), amino acids (GAP1), purines (FCY2), sulfur amino acids (MMP1), and uridine (FUI1), but also amino acid biosynthesis genes (HIS1, ILV5) as well as genes of the ergosterol biosynthesis pathway (ERG1, ERG11).

Table 8.

Genes repressed by sorbic acid stress

Gene	ORF name	Biological process	Wild type	SDEV	war1Δ	SDEV	msn2Δ msn4Δ	SDEV	war1Δ msn2/4Δ	SDEV
PHO84	YML123C	Phosphate transport	0.12	0.06	0.26	0.64	0.08	0.01	0.04	0.02
GLY1	YEL046C	Threonine catabolism	0.18	0.05	0.23	0.13	0.57	0.08	0.60	0.14
ILV5	YLR355C	Mitoch. genome maintenance	0.20	0.10	0.35	0.80	0.38	0.02	0.47	0.11
GAP1	YKR039W	Amino acid transport	0.24		0.48	0.09	0.25	0.03	0.24	0.06
NSR1	YGR159C	rRNA processing	0.26	0.25	0.19	0.03	0.76	0.72	0.38	0.13
FUI1	YBL042C	Uridine transport	0.29	0.15	0.48	0.05	0.71	0.86	0.30	0.24
ERG3	YLR056W	Ergosterol biosynthesis	0.29	0.16	0.51	0.69	0.51	0.09	0.75	0.13
SEO1	YAL067C	Transport	0.31	0.26	0.87	0.53	2.39	1.83	0.68	0.71
SAM1	YLR180W	Methionine metabolism	0.31	0.09	0.61	0.66	0.50	0.16	0.47	0.07
ENA2	YDR039C	Sodium ion transport	0.33	0.19			1.45	1.92	0.80	0.16
RHR2	YIL053W	Response to osmotic stress	0.33	0.25	0.33	0.05	0.91	0.86	0.75	0.24
YLR419W	YLR419W	Biological process unknown	0.33	0.03	0.52	0.38	0.52	0.17
HIS1	YER055C	Histidine biosynthesis	0.34	0.18	0.43	0.12	0.70	0.39	0.57	0.09
CBF5	YLR175W	rRNA modification	0.34	0.32	0.56	0.47	0.67	0.42	0.44	0.16
EXG1	YLR300W	Cell wall organization	0.35	0.06	0.58	0.32	0.90	0.52	0.57	0.12
YNL300W	YNL300W	Biological process unknown	0.35	0.08	0.35	0.19	0.24	0.12	0.27	0.06
ZRT1	YGL255W	High-affinity zinc ion transport	0.35	0.20	0.34	0.07	0.52	0.17	0.38	0.33
HPT1	YDR399W	Purine nucleotide biosynthesis	0.35	0.23	0.28	0.27
PTR2	YKR093W	Peptide transport	0.36	0.11	0.60	0.27	0.40	0.15	0.51	0.08
YBR043C	YBR043C	Biological process unknown	0.36	0.24			1.46	0.91	0.79	0.38
SUR4	YLR372W	Sphingolipid biosynthesis	0.36	0.03	0.51	0.22	0.67	0.10	0.60	0.26
ECM1	YAL059W	Cell wall organization	0.36	0.08					0.64	0.92
YDR492W	YDR492W	Biological process unknown	0.37	0.10	0.57	0.14	1.04	0.80	0.47	0.23
ERG1	YGR175C	Ergosterol biosynthesis	0.37	0.04	0.39	0.08	0.77	0.50	0.60	0.14
CRH1	YGR189C	Biological process unknown	0.37	0.03	0.46	0.13	0.77	0.13	0.79	0.16
VTC4	YJL012C	Vacuole fusion (non-autophagic)	0.37	0.04	0.39	0.12	0.45	0.06	0.36	0.08
URA7	YBL039C	Phospholipid biosynthesisa	0.38	0.08	0.69	0.77	0.89	0.67	0.41	0.10
PDR5	YOR153W	Response to druga	0.38	0.31	0.29	0.25	0.53	0.05	0.43	0.16
YFR055W	YFR055W	Copper ion homeostasisa	0.38	0.11	0.38	0.15	0.93	0.30	0.77	0.25
CPA1	YOR303W	Arginine biosynthesis	0.38	0.08	0.45	0.08	0.49	0.28	0.41	0.13
ENA5	YDR038C	Sodium ion transport	0.38	0.21	0.97	0.95	1.27	1.07	0.50	0.11
MET6	YER091C	Methionine biosynthesis	0.38	0.16	0.49	0.12	0.80	0.73	0.65	0.15
YBR033W	YBR033W	Biological process unknown	0.38	0.32	0.93	1.34	0.50	0.15	0.72	0.45
ITR1	YDR497C	Myo-inositol transport	0.39	0.16	0.54	0.09	0.67	0.53	0.71	0.25
MEP2	YNL142W	Pseudohyphal growtha	0.39	0.20	0.64	0.28	0.64	0.51	0.31	0.10
HTA2	YBL003C	Chromatin	0.39	0.06	0.47	0.19	0.58	0.21	0.50	0.15
ERG11	YHR007C	Ergosterol biosynthesis	0.40	0.08	0.39	0.06	0.85	0.59	0.69	0.21

A cutoff ratio of 0.4 was arbitrarily chosen to visualize repressed genes. The complete list of repressed genes is found in the supplementary materials.

War1p and Msn2p/4p Act through Distinct cis-acting Motifs in Target Genes

To identify a possible common cis-acting motif in various promoters of genes present in the respective groups, we used the AlignACE program (Roth et al., 1998). This analysis indicated a shared motif (5′-CGG c/t T g/c t/g TA a/t -3′) among the strongly War1p-dependent genes. This motif appears in the following genes: Target genes containing this motif are marked by an asterisk (Tables 3, 4, 5, 6, 7) and are represented graphically as a sequence logo reflecting the relative frequency of each base in the motif (Figure 1B). Related sequences appear five times in the PDR12 promoter, overlapping with cis-acting sequences previously shown to be recognized by War1p in vivo (Kren et al., 2003). Furthermore, the distribution of STRE motifs (5′-WAGGGG-3′) was determined according to previously published analysis (Moskvina et al., 1998; Treger et al., 1998). This motif was clearly enriched in both Msn2p/4p- and War1p-Msn2p/4p-dependent groups (34 and 38%, respectively, compared with 14 and 19% in the HSP30 and No-category groups), strongly supporting our assignment. Finally, the assignment of Msn2p/4p dependency under sorbate stress showed a remarkable overlap of >40% (Tables 3, 4, 5, 6, 7), with the one of Msn2p/4p-dependent genes under acid stress conditions (Causton et al., 2001). These correlations confirmed our analysis to a large extent. However, we were unable to identify a common motif shared by most of the War1p-Msn2p/4p-independent genes.

To assess individual roles of War1p-regulated genes in weak acid resistance, we investigated growth of the respective viable EUROSCARF mutants on sorbate-containing plates and found only the pdr12Δ mutant strain to be highly sensitive, whereas all other mutants failed to show differences to the wild-type control in the plate assay (our unpublished results). Nevertheless, the functional screening and the microarray profiling essentially addressed two related issues. What are the genes required for coping with a given stress, and which genes are activated as a consequence of the subsequent response? Strikingly, the comparison of genes required for stress adaptation to genes induced by the same stress revealed only one gene that appeared exclusively in both datasets, namely PDR12. It was therefore tempting to speculate that Pdr12p is the major cellular determinant of weak acid resistance.

Pdr12p Is Necessary and Sufficient for Weak Acid Resistance

To address this question, and to analyze the contribution of Msn2p/4p versus War1p-regulated genes to stress relief, we constructed a set of isogenic strains carrying war1Δ (YAK120), pdr12Δ (YBB14), msn2Δ msn4Δ (W303 msn2 msn4) as well as war1Δ msn2Δ msn4Δ (YCS105) deletions. At 1 mM sorbate, the msn2Δ msn4Δ strain grew almost like the W303-1A wild-type control, whereas the pdr12Δ strain failed to grow (Figure 3A). The war1Δ and war1Δ msn2Δ msn4Δ cells displayed the same degree of resistance. A low level of growth was perhaps due to a very low basal Pdr12p production. Therefore, absence of Msn2p/4p did not further exacerbate the hypersensitivity of war1Δ cells. Furthermore, we were unable to demonstrate weak acid hypersensitivity in mutants lacking the HSP12, HSP26, and HSP104 genes, all of which are otherwise strongly regulated by Msn2p/4p (Martinez-Pastor et al., 1996; Amoros and Estruch, 2001; Figure 3B).

Figure 3.

War1p, but not Msn2p/4p, is required for weak acid resistance. (A) Lack of MSN2 and MSN4 fails to influence growth properties on sorbic acid. Sensitivity of W303-1A and the mutant strains YBB14 (pdr12Δ), YAK120 (war1Δ), W303 msn2 msn4 (msn2Δ msn4Δ), and YCS105 (war1Δ msn2Δ msn4Δ) was tested on YPD pH4.5 plates. Strains W303-1A, YBB14, and YAK120 carried the TRP1 gene on the centromeric plasmid YCplac22. Growth was monitored after 48 h at 30°C. (B) HSP12, HSP26, and HSP104 are not required for sorbic acid resistance. BY4741 derivatives were spotted onto sorbate-containing plates. Colony growth was recorded after 48 h.

These data as well as published results indicate that the weak acid sensitivity of cells lacking War1p is only due to its function in controlling Pdr12p induction. To test this assumption, we placed the PDR12 gene under the control of the GAL1-10 promoter, thus rendering Pdr12p inducible by galactose but not weak acids (Figure 4A). As expected, expression of Pdr12p driven by the GAL1-10 promoter was undetectable in glucose-grown cells, but was strongly induced on galactose. The absence of War1p did not influence Pdr12p levels on galactose (Figure 4A). Hence, PDR12 expression is completely independent of War1p function in this strain. The levels of Pdr12p due to galactose-triggered induction fully compensated for the absence of War1p on sorbate medium (Figure 4B) as well as on propionate (our unpublished results). Such induction restored almost wild-type levels of weak acid resistance to war1Δ cells (Figure 4B). These results demonstrate that ectopic expression levels of PDR12 are sufficient to confer weak acid stress resistance in war1Δ cells, firmly establishing PDR12 as the single most important and perhaps only relevant target gene of War1p.

Figure 4.

Pdr12p is necessary and sufficient for sorbic acid resistance. (A) Pdr12p levels as derived from a GAL1-10-PDR12 gene are independent from War1p and sorbate challenge. Cells growing on YPD and YPgalactose were exposed to sorbic acid for 20 min. Immunoblots were decorated using polyclonal anti-Pdr12p antibodies. A cross-reacting band served as a loading control. (B) Cells from logarithmically growing cultures were spotted onto YPD, pH 4.5, and YPGal, pH 4.5, with and without 1 mM sorbic acid. Plates were monitored after a 48-h incubation at 30°C.

The War1p and Msn2p/4p Regulons Are Activated by Weak Organic Acids

One plausible explanation for the different roles of Msn2p/4p and War1p could be a distinct response to different, but possibly overlapping, spectra of inducing cues. To test this idea, we used the fact that weak acid induction of HSP12 only demands Msn2p/4p, whereas PDR12 was only regulated by War1p (Figure 5A). To test the weak acid substrate spectrum for War1p eliciting activation of the PDR12 gene, we analyzed the activity of an integrated PDR12-lacZ fusion construct in response to various weak acids. We detected a rapid activation of this reporter construct by weak acids ranging from C3 to C8 chain length, which was completely absent in the war1Δ control reporter strain (Figure 5B). As also shown earlier (Hatzixanthis et al., 2003), acetate (C2) was unable to induce Pdr12p (Figure 5B).

Figure 5.

Northern analysis of HSP12 and PDR12 expression during weak acid stress. (A) Cells were grown to the logarithmic growth phase in rich medium and exposed to the indicated weak acids for 20 min. Methylene blue staining of rRNA served as a loading control. (B) Weak acid stress-induced PDR12 promoter activity is dependent on WAR1. Logarithmically growing cells were treated with the indicated concentrations of weak organic acids for 45 min. β-galactosidase activity was determined in crude extracts. Wild-type: YAK3 PDR12-lacZ, war1Δ: YBB27 PDR12-lacZ war1Δ::HisMX6c.

We then determined the mRNA levels of both HSP12 and PDR12 after addition of propionate, butyrate, and sorbate as well as hexanoic, heptanoic, octanoic acids, myristylate, and palmitate (Figure 5A). Both Msn2p/4p and War1p activate their respective target genes by organic acids with a chain length ranging from C3 to C10. Induction of HSP12 by C14 and C18 acids pointed to marked differences in the War1p- and Msn2p/4p-dependent response mechanisms. Acetic acid and dodecanoic acid failed to induce the PDR12 promoter, demonstrating that only weak acids of a certain chain length can activate War1p. Msn2p/4p activation as detected by HSP12 induction seemed to include weak acids of higher chain length as well (Figure 5A). Because Msn2p-GFP is known to accumulate rapidly in the nucleus of stressed cells, its activation was also monitored in living cells by visualization of the nuclear-cytoplasmic distribution of a Msn2-GFP fusion protein (Görner et al., 1998). Except for acetic acid, all weak acids tested caused a rapid nuclear accumulation of Msn2p in <5 min (C. Shüller, unpublished results).

Finally, and in agreement with our microarray results, HSP30 seemed activated through an as yet unknown mechanism, because we detected its mRNA induction by sorbate in a war1Δ msn2Δ msn4Δ strain (Figure 6A). Although PDR12 and HSP12 were strongly and exclusively affected by absence of War1p and Msn2p/4p, respectively, we found a partial effect of Msn2p/4p, but no effect of War1p on HSP30 expression (Figure 6A). Because war1Δ cells were sorbate sensitive and because certain HSP30 mutants display a longer lag-phase on sorbate (Piper et al., 1997), we investigated a possible synthetic phenotype of cells lacking both HSP30 and PDR12. However, strains devoid of HSP30 failed to display significant viability defects in the plate assay on media over a range of sorbic acid concentrations (Figure 6B, our unpublished results). Therefore, the war1Δ msn2Δ msn4Δ cells were not compromised in their response to sorbate.

Taken together, our data show that it is not only the inducibility of War1p and Msn2p/4p by different weak acids that determines their physiological role. Instead, their specific physiological function in stress response and adaptation is determined at the level of the relevant downstream target genes. Moreover, the identification of all key genes of the weak acid stress response regulons was only possible by combining microarray profiling with functional screening approaches.

DISCUSSION

Adjustment of the transcriptional programs enables yeast cells to rapidly tune expression patterns in response to environmental challenges. With the advent of genome-wide analysis of gene expression, transcriptome changes can be easily identified because they relate to great many different conditions (see e.g., http://genome-www4.stanford.edu/cgi-bin/SGR/publication/publicRef). A major message of these results is that only a fraction of a given genomic response is actually necessary to cope with any adverse condition. In other words, the transcriptional profile typical for a given stress situation usually includes many more genes than immediately required for stress relieve. Nonetheless, transcriptome patterns are remarkably stable phenotypes, which can be exploited in many ways. Another major achievement of the genomic age is the availability of large-scale systematic functional assays. Genome-wide knockout strain collections are available for S. cerevisiae. Similar screens are now also possible in higher eukaryotes using the siRNA methodology (Lum et al., 2003). Although resources like that allow for rapid genetic screens, they cannot completely replace classical mutagenesis approaches. In this study, we pursued both strategies to identify genes involved in the response and adaptation to the fungistatic effects of short-chain weak organic acids such as sorbic acid or propionate. Our analysis leads us to the conclusion that a combination of phenotypic screening or functional assays (i.e., functional genomics) with datasets from mRNA profiles can effectively pinpoint crucial genes operating in complex biological pathways.

Weak acids have a potent but transient fungistatic effect on yeast cells. In our search for genes important for recovery during sorbate stress, we identified many genes with very different physiological roles or metabolic functions. Perhaps the most unexpected finding was the identification of a class of mutants in the aromatic amino acid biosynthesis pathway. Closer inspection of this phenotype revealed that an inability to synthesize tryptophan increases the sorbate sensitivity (Y. Mamnun, unpublished results). Because these mutants show normal sorbate-sensing and signaling as judged by induction of Pdr12p, the effect is most likely caused by reduced tryptophan uptake (Bauer et al., 2003) through mechanisms that are unknown at present. Interestingly, the number of sorbate-sensitive genes appears proportional to the amount of sorbate used for phenotypic selection (M. Mollapour, unpublished results). The stringency of selection therefore has a strong effect on the resulting number of candidate genes implicated in susceptibility phenotypes. The main players identified by this approach include PDR12, an ABC transporter previously identified to confer weak acid resistance, probably through its ability to efflux acid anions (Piper et al., 1998), as well as WAR1. The latter is a typical Zn(II)₂Cys₆ transcription activator recently shown to be a major regulator of PDR12 (Kren et al., 2003).

In a second approach, we analyzed genome-wide changes in mRNA levels following weak acid stress (see below). A comparison of the results from both approaches used in this study recovered only one gene, PDR12. We therefore reasoned that Pdr12p is the prime player in weak acid stress response and that War1p represents its main stress regulator. We tested this idea by conditional expression of Pdr12p, thereby circumventing War1p-dependent induction of Pdr12p. As expected, placing Pdr12p under the control of the GAL1-10 promoter eliminates the requirement for War1p in weak acid response and stress adaptation. This suggests that Pdr12p is necessary and sufficient for resistance, suggesting that PDR12 is the main target gene of War1p at least in the context of weak acid resistance. Therefore, our hypothesis drawn from the genomic data are valid, and as shown here, the combination of two genome-wide assays readily identified the most important target gene.

The deletion of War1p results in weak acid hypersensitive cells (Kren et al., 2003). Hence, identification of War1p targets might allow us to identify additional genes important for weak acid resistance. Genome-wide mRNA changes in sorbate-treated cells involve the rapid induction of ∼170 genes, with a quite similar pattern of modulated genes in propionate-treated cells. The sorbate results might thus be of general applicability regarding other weak acids, at least in terms of target genes. Many sorbate-induced genes included those previously reported under the control of Msn2p and Msn4p, which govern the general stress response (Estruch, 2000; Gasch et al., 2000). Expression profiling of strains lacking War1p and/or Msn2p/4p allowed us to identify the War1p-dependent genes. More than 50% of the sorbate-induced genes require Msn2p/4p. Weak acids also trigger a rapid nuclear accumulation of Msn2p/Msn4p, probably due to inhibition of their nuclear export signals (Görner et al., 2002), followed by an increased occupancy of their cognate DNA binding sites (STREs) in the target promoters (Görner et al., 1998). As for War1p, it is unclear how and why weak acids activate Msn2p/4p. A possible trigger might be the drop of the intracellular pH as caused by weak acid dissociation. Nevertheless, our data show that both War1p and Msn2p/4p respond to a broad range of weak acids, leading to the modulation of many downstream target genes. However, differences in the response patterns (Figure 5) make it unlikely that both are activated by the same mechanism.

The experiments reported in this study address the immediate early transcriptional response, similar to the environmental stress response (ESR) dataset (Gasch et al., 2000) and the common environmental response (CER; Causton et al., 2001). Short-term ethanol stress (Alexandre et al., 2001) or salt stress (Rep et al., 1999) has been reported to also induce many Msn2p/4p-dependent genes. The situation is different when following long-term exposure to weak acids (de Nobel et al., 2001), because only 25 genes of the 167 genes described here are detected under both the acute response and the adapted response (de Nobel et al., 2001). Interestingly, these 25 genes are overrepresented in War1p- and Msn2p/4p-dependent regulons, suggesting that long-term responses are also mediated by these factors.

A strong HSP30 induction by sorbate and heat shock has been demonstrated in msn2Δ msn4Δ cells and in a hsf1Δ heat shock transcription factor mutant (Seymour and Piper, 1999). Here we provide evidence that HSP30 may be part of a larger regulon, which is independent from War1p and Msn2p/4p, thus establishing a third regulatory mechanism activated by sorbate stress. But what then is the role of this regulon in weak acid resistance? Interestingly, several genes of this regulon are involved in processes on the level of the plasma membrane. Hsp30p is the only membrane-bound heat shock protein (Piper et al., 1994); therefore, it could be involved in sensing membrane damage as exerted by weak organic acid exposure. Interestingly, we found that all four genes coding for polyamine transporters (TPO1,2,3,4) were also activated (Albertsen et al., 2003). Tpo1p may also act as an herbicide and quinidine-resistance determinant (do Valle Matta et al., 2001; Teixeira and Sa-Correia, 2002). Other genes in this group include the HSP30 homologue YRO2 (Wu et al., 2000) as well as CIT2 encoding a nonmitochondrial citrate synthase that participates in the glyoxylate cycle (Kim et al., 1986). The demonstration of a functional significance for this regulon and the identification of upstream factors is clearly a next important step.

Although the pattern of upregulated genes is characteristic for a given environmental condition, this is also true for repressed genes. During the environmental stress response, twice as many genes are repressed when compared with the induced ones (Gasch et al., 2000; Gasch and Werner-Washburne, 2002). In the case of weak acids, a number of genes encoding transporters localized to the plasma membrane are dramatically reduced. These genes are involved in diverse cellular functions such as phosphate (PHO84), sulfur amino acids (MMP1), zinc (ZRT1), amino acids (GAP1), and purine (FCY2) uptake. The physiological role of this downregulation is currently unclear, but it may indicate changes in the protein composition of the plasma membrane. Alternatively, this repression might reflect a growth-related effect, as cells are in a stasis and thus might downregulate transport processes related to secondary metabolism. Moreover, down-regulation of GPA1 might contribute to the growth defect of strains lacking genes of aromatic amino acid biosynthesis identified in our functional screen.

The results give a fairly clear picture about the genes comprized in the War1p-dependent regulon. Although the precise mechanism of War1p activation is not known, it might directly bind and thus sense the activating agents such as weak acids. War1p, which constitutively decorates the PDR12 promoter (Kren et al., 2003), could become activated through a conformation change induced by posttranslational modification such as phosphorylation. Indeed, War1p is rapidly and phosphorylated after sorbate stress (Kren et al., 2003), and mutations that prevent phosphorylation lead to nonfunctional War1p (Bauer et al., unpublished results). In the uninduced state, self-masking could prevent constitutive activity of War1p. For example, self-masking has been described for Leu3p, a transcription factor reacting to α-isopropylmalate (Wang et al., 1997, 1999), Hap1p that recognizes heme (Zhang and Guarente, 1995), and Put3p responding to proline levels (Des Etages et al., 2001). Stress-induced War1p phosphorylation and its link to PDR12 induction might be a cause or consequence of conformational changes. In any case, the kinase mediating War1p phosphorylation has escaped discovery as yet. Further, neither the genetic nor the functional screen uncovered candidate kinases, implying that they could be essential for viability or functionally redundant.

Our data illustrate that the general stress response, although induced by the same cues as the specific response through War1p, is not required for weak acid resistance. War1p and Msn2p/4p might function as upstream receivers for weak acid stress signals, but a distinct downstream physiological response is brought about by striking differences in target gene specificity. Why then do yeast cells require two partially redundant signaling mechanisms for weak organic acids? A PDR12 gene under Msn2p/4p control would presumably result in Pdr12p induction under many adverse conditions such as starvation, heat shock, or osmotic stress. One reason for a second and yet specific signaling mechanism may a need to avoid unwanted effects of higher Pdr12p levels under general environmental or metabolic stress. At any rate, our studies show that multiple mechanisms contribute to at least three distinct sorbate-inducible regulons. More importantly, our results indicate that only a combination of functional genomic analysis and global expression profiling represents an efficient way to identify key players operating in different complex biological stress response pathways.