Transcript expression in Saccharomyces cerevisiae at high salinity.

Transcript expression of Saccharomyces cerevisiae at high salinity was determined by microarray analysis of 6144 open reading frames (ORFs). From cells grown in 1 m NaCl for 10, 30, and 90 min, changes in transcript abundance >2-fold were classified. Salinity-induced ORFs increased over time: 107 (10 min), 243 (30 min), and 354 (90 min). Up-regulated, functionally unknown ORFs increased from 17 to 149 over this period. Expression patterns were similar early, with 67% of up-regulated transcripts after 10 min identical to those at 30 min. The expression profile after 90 min revealed different up-regulated transcripts (identities of 13% and 22%, respectively). Nucleotide and amino acid metabolism exemplified the earliest responses to salinity, followed by ORFs related to intracellular transport, protein synthesis, and destination. Transcripts related to energy production were up-regulated throughout the time course with respiration-associated transcripts strongly induced at 30 min. Highly expressed at 90 min were known salinity stress-induced genes, detoxification-related responses, transporters of the major facilitator superfamily, metabolism of energy reserves, nitrogen and sulfur compounds, and lipid, fatty acid/isoprenoid biosynthesis. We chose severe stress conditions to monitor responses in essential biochemical mechanisms. In the mutant, Deltagpd1/gpd2, lacking glycerol biosynthesis, the stress response was magnified with a partially different set of up-regulated ORFs.

In yeast, many components underlying the signaling pathways that control these biochemical entities are known (4, 8 -12). Most information is available about signal transductionaltering carbohydrate metabolism, where MAPK 1 phosphorelays, exemplified by protein kinase Hog1p (High Osmolarity Glycerol), transmit osmotic changes (13,14) leading to the induction of transcription that activates downstream biochemical functions (10,(15)(16)(17)(18). Activation of Hog1p constitutes an early phase of the salinity stress response, which then seems to diverge into different pathways (18 -20). One pathway is mediated by the transcription factors Msn2p/Msn4p binding to stress response elements (STREs) and acting in signal amplification (21)(22)(23). Hot1p, another transcriptional activator, has recently been identified as a Hog1p partner in this signaling (12,20). In addition, the derepression of genes, for example through the regulated action of Sko1p, seems to add another level of complexity controlling the salinity stress response machinery (24,25). Partially interacting with HOG-based signal transduction is a pathway associated with the action of the protein phosphatase calcineurin, a mediator for many cellular responses to calcium signals (8,11,26). The calcineurin pathway responds predominantly to challenges in the ionic environment.
The rationale for focusing on transcriptional reactions of yeast to salt stress through a genome-wide expression analysis is based on our interest in plant salinity stress responses (2,6,27). Similar abiotic stress-induced gene expression programs seem to exist in yeasts and plants, including conserved signaling pathways and biochemical defense determinants (28,29). Comparative studies promise to reveal the similarities and distinctions between cell-specific and organismal components involved in tolerance acquisition. An experimental outline that describes the portion of the yeast genome required for osmotic stress tolerance will aid in delineating the conserved cellular functions of homologous elements in multicellular organisms.
Yeast microarrays provide information about the transcription of all genes. Genome-wide monitoring of transcript changes in yeast could show previously unrecognized cellular aspects of stress protection and reveal genes that represent a yeast-specific solution to survival in high sodium. Such studies with yeast have recently become available (25,30). The three times replicated time-course experiments reported here add to these analyses. One novel aspect is the description of a succession of biochemical categories that are progressively up-regulated. Early stress responses, affecting mainly nucleotide and protein biosynthetic pathways, are different from later responses, which included intracellular protein and metabolite transport activities and increased energy consumption for metabolic and ion homeostasis. Transcription after prolonged stress also exemplifies ascending functions in cell rescue, in aging (cell death) and defense-related roles, and reveals a large number of functionally unknown ORFs.
A yeast genome array, 6144 coding regions deposited on nylon filters, was used for a complete analysis of changes in transcript expression following hyper-osmotic stress. The results confirm many of the ORFs and proteins previously reported as stress-regulated (3,10,15,23) and adds a number of stress-regulated transcripts that had not been recognized before. We describe early response transcripts distinguished from those that act at later times in different functional categories that seem to maintain cell integrity. We identify a set of ϳ200 salt stress-regulated functionally unknown ORFs. Some of the unknown ORFs have orthologs in cDNA libraries from saltstressed plants. Our results complement and extend through the use of a salt stress-sensitive mutant, which is unable to synthesize glycerol, recent reports that have targeted yeast salinity stress responses though the analysis of arrayed ORFs (25,30).

MATERIALS AND METHODS
Yeast Strains, Growth, and Stress Conditions-Saccharomyces cerevisiae strain S150-2B (MAT␣ ura3-52 his3⌬ leu2-3,112 trp1-289) was grown at 30°C to mid-log phase (A 600 ϭ 1.0) in standard rich media, YPD (1% yeast extract/2% peptone/2% glucose, pH 5.5). Control (no salt) cultures were harvested immediately, and salt-stressed yeast cultures were harvested after 10, 30, and 90 min. To salt stress the cells, an equal volume of YPD containing 2 M NaCl was added directly to the yeast cultures. Cell count, optical density measurements, and streaking of cells on non-selective media during the stress experiments indicated a decline in cell numbers by ϳ25% during the 10-min time point but cell number remained constant thereafter and increased after ϳ4 h of stress. Cells were collected by centrifugation at 3000 rpm for 5 min and rapidly washed once in sterile water, and the cell pellet was frozen and stored at Ϫ70°C until RNA extraction. The experiment was repeated three times with independently grown cultures. As an additional control, yeast strain W3031A (MAT␣ leu2-3,112 ura 3-1 trp1-1 his3-11,15 ade2,1 can1-100 SUC2 GAL mal0 GPD1::URA3 GPD2::TRP1), which is defective in glycerol biosynthesis (kindly provided by Dr. S. Hohmann, Gøeteborg, Sweden), was grown to mid-log phase in rich medium and stressed with 0.5 M NaCl for 60 min (31).
Yeast Gene Filters-Yeast Gene Filters (Research Genetics Inc.) contain 6144 PCR products bound to nylon filters (available at the Research Genetics web site). The size of the PCR products ranged from 300 bp to 4 kb. The filters are missing ϳ300 PCR products from chromosome 16. Average changes in transcript abundance for each time point were calculated relative to the "no salt" control using "Pathways" software (also available at the Research Genetics web site). Approximately 30% of the ORFs whose hybridization signal varied at background levels under all experimental conditions were eliminated from the final analysis. The filters were used only once to avoid variations caused by unequal stripping of probe or DNA from the membranes. ORFs with (partially) overlapping reading frames are identified in the supplemental material. Averaging reduced the number for absolute fold induction but the spread over all hybridizations indicated that a 2-fold induction was significant.
Preparation of mRNA and Hybridizations-Total RNA was isolated from frozen cell pellets by extraction with hot acidic phenol (32). Complementary DNA was prepared using oligo(dT) primers and 5 g of total RNA labeled with [␣-33 P]dCTP and purified through a QIAquick column (Qiagen Inc.). The cDNAs (three for each time point) were hybridized to 12 individual sets of gene filters. Detailed experimental procedures for treatment of the filters can be found at the Research Genetics web site.
Microarray Analysis-Phosphorimages of the yeast microarray filters were captured with a resolution of 50 m on a Storm PhosphorImager (Molecular Dynamics Inc.) and analyzed using Pathways software, version 2.01, which provided a 16-bit imaging capability (Research Genetics Inc.). Normalization between sets of filters was based on the average of the signal intensities of all the data points on the individual filters. Comparisons for each of the experimental conditions (10, 30, and 90 min of NaCl stress) were calculated relative to the no-stress control. The relative fold changes in transcript abundance for 10, 30, and 90 min of salt stress represent the average changes in gene expression for three experiments each. We considered expression levels Ͼ2.0-fold as induced, Ͻ2.0-fold as repressed, and between Ϫ1.9and ϩ1.9-fold as unchanged. Induced ORFs were categorized based on the MIPS classification of yeast ORFs (available on the Web). Approximately 30% of the 6144 signal intensities were removed from the analysis, because they were equal to background intensities under all experimental conditions. The complete data sets can be found at the Research Genetics Web site.
Northern Blots and Densitometry-Total RNA was isolated from mid-log phase (A 600 ϭ 1.0) yeast cultures after 0, 10, 30, and 90 min of exposure to 1 M NaCl. Probes were made by PCR amplification of specific ORFs from genomic DNA and random primer-labeled with [␣-32 P]dCTP. Probes were chosen based on the microarray data. We chose seven ORFs with no change in gene expression and 32 ORFs with induced gene expression (see the Stress Genomics Web site for a list). Standard procedures for RNA blot hybridizations were followed (33). Phosphorimaging analyses of the RNA hybridizations were created on the Storm PhosphorImager, and densitometry was calculated using ImageQuaNT software (Molecular Dynamics Inc.). Statistical analyses used standard commercial software programs.

RESULTS AND DISCUSSION
Microarray Evaluation-Global gene expression during salinity stress was determined using Yeast Gene Filters (Research Genetics Inc.), assembled from PCR-amplified open reading frames (ORFs) for 95% of the yeast putative and confirmed genes. Yeast cultures were grown to mid-log phase in complete media supplied with 1 M NaCl for 0, 10, 30, and 90 min. Exploratory experiments at different time points, which have been performed only once, are not included. Compared with previous microarray experiments (25,30), the lag phase for recovery of growth in our experiments was ϳ2 times longer. Each hybridization was repeated three times. The [␣-33 P]dCTP signal intensities for all hybridizations ranged from background levels (3-65 dpm) to high intensity (ϳ72,000 dpm). The most highly expressed transcripts for both stressed and unstressed yeast represented ϳ10% of the yeast genome with an average signal intensity value of 4500 dpm, whereas 60% of the transcripts had an average signal intensity value of 169 dpm ( Fig. 1). Standard deviations, which included all hybridization were within Ϯ 1.5 S.D. for the most highly expressed transcripts (614 ORFs) and slightly lower for transcripts expressed in the medium to low range (ϳ1000 ORFs).
High salinity affected expression levels of ϳ10% of the yeast ORFs. During salt stress the number of ORFs induced increased more than 2-fold from 107 (10 min) to 354 (90 min) ( Fig.  1). After 10 and 30 min, 27 and 78 ORFs, respectively, were induced more than 3-fold, and an additional 87 (10 min) and 165 (30 min) ORFs had average changes in transcript levels between 2-and 3-fold. After 90 min of salt stress, 170 ORFs (Ͼ3-fold) and 185 ORFs (2-to 3-fold) showed increased average changes in transcript abundance. The up-regulated ORFs in major MIPS categories are shown (Fig. 2). The nature of regulated transcripts over time changed, suggesting that different functions needed to be activated at different time points. As a control, 39 ORFs were examined (Fig. 3, not all data included) by RNA blot analysis, to independently verify changes for transcripts in different abundance categories. Open reading frames were chosen with varying levels of transcript abundance in microarrays; 11 ORFs Ͼ 3.0 fold, 13 ORFs Ͼ 2.0-to 3.0-fold, 7 ORFs with no change in message levels, and 8 with decreases of more than 2-fold. ImageQuaNT software was used to determine changes in transcript levels in Northern hybridizations, which were compared with the average changes in gene expression from the microarrays. Among the selected ORFs, 36 of 39 agreed with the microarray data (3 ORFs predicted to be upregulated by a factor of less than 2 showed no change in RNA blot hybridizations) (Fig. 3). Overall, low and moderate changes in transcript abundance in the comparison between the RNA blots and the averaged microarray data differed by less than 2-fold, but large changes in abundance can differ by Ͼ10-fold, mainly attributable to low basal transcript levels (e.g. YGL037C, YGR243W, and YHR087W).
Correlation with known Yeast Stress Responses-The behavior of many transcripts in the analyses correlated with known biochemical hyper-osmotic stress responses. Glycerol, for example, an osmoprotectant known to accumulate rapidly in response to stress in yeast (31), accumulated as documented by high pressure liquid chromatography analyses (data not shown), and transcripts in the glycerol biosynthetic pathway increased. Indeed, dehydrogenases and phosphatases leading to glycerol production, GPD1/2, GPP1/2, were up-regulated at all time points during salinity stress, most strongly as time progressed (Fig. 4).
Similarly, ORFs for enzymes involved in trehalose metabolism, GLK1, PGM2, HXK1, YKL035W, TPS1, TPS2, and NTH1, were up-regulated at 90 min, but not at 10 and 30 min of salt stress. Transcripts for all enzymes of the pathway were among those most highly induced (Fig. 4). Trehalose, like glycerol, is implicated in yeast stress responses as an osmoprotectant, although trehalose does not accumulate to osmotically significant concentrations in salt-stressed bakers' yeast (34). PGM2, UGP1, TPS1, TPS2, and the regulatory factor encoded by TSL1, catalyze trehalose biosynthesis, whereas Nth1p and Ath1p (trehalases) lead to trehalose degradation and the formation of glucose (35). Completion of this cycle seems to be indicated by the up-regulated transcripts for the kinases HXK1 and GLK1 (Fig. 4). The presence of high transcript amounts for Nth1p, Hxk1p, and Glk1p may explain why the osmoprotectant trehalose does not accumulate during salt stress. A circular flux of carbon, based on the induction of all ORFs in this pathway, seems to indicate a function for trehalose in a regulatory role, for example in redox control, similar to what has been documented for the functions of the two GPD enzymes (36,58). Trehalose synthesis and degradation, in combination with glycerol production, plays a key metabolic role in the protection against high salinity. Such a conclusion, also based on gene expression changes, has recently been put forward (58). The 1,4-glucan branching enzyme involved in glycogen biosynthesis was only moderately up-regulated under our conditions. These results are similar to recently published data with the exception that the high NaCl concentration, 1 M, tended to delay up-regulation compared with what has been reported for the yeast transcriptome response in 0. The extent to which transcript increases correlate with protein amount has been verified in some studies. Apart from increases in the activity of enzymes and the phenotype of knockout mutants, two-dimensional electrophoresis of proteins and partial sequencing of up-regulated peptides indicated general proportionality between RNA and protein amounts for metabolic enzymes, and this also extended to the down-regulation of, for example, enolases (ENO1/2) (Refs. 60 -62; see Supplemental Table sIII).
Induced Gene Expression during Hyper-osmotic Stress-Global gene expression patterns were determined for ORFs induced more than 2-fold after 10, 30, and 90 min of exposure to 1 M NaCl (Fig. 2). Expression patterns for early-induced transcripts, at 10 and 30 min following stress, were similar, with 67% of the ORFs induced after 10 min being identical to those induced after 30 min. The profiles are characterized by rapid transcript increases for ORFs in protein metabolism, mainly attributable to transcripts for components of protein synthesis, protein destination, and the regulation of protein fate. Nearly half of all up-regulated transcripts (42%; 44 ORFs) originated from the categories "protein destination" (ORFs related to protein modification, transport, and targeting), "intracellular transport" (cellular import, protein trafficking, and vesicular transport), and "protein synthesis" (ribosomal proteins). These three categories represented 13% (53 ORFs) of the ORFs upregulated after 90 min. Based on a relative scale, the difference seems to indicate the significance of these ORFs for the adjustment of metabolism early during the salt stress.
ORFs for Ribosomal Proteins-Ribosomal proteins (RPO) account for the majority of induced ORFs after 10 min (17 ORFs) and 30 min (45 ORFs), and the 17 ORFs found after 10 min are included in the group of 45 ORFs found at 30 min (Table I). This is in contrast to the expression patterns of all ORFs for ribosomal proteins at 90 min of salt stress when 93 of the 176 (55%) strongly down-regulated ORFs encoded ribosomal proteins. At the late time point, only three RPO were up-regulated, all of them encoding mitochondrial RPO (MRP8, MRPS18, MET13). The number of transcripts strongly repressed after 90 min included 31 of the early-induced RPO ( Table I) (37), the 44 up-regulated cytosolic RPO suggest specific promoter and possibly also signaling events. Their up-regulation may also indicate a requirement for the presence of specific RPO under stress conditions. We have found a similar theme in the salt stress response of rice (Oryza sativa), recorded by microarrays of 1728 transcripts. The early time points of salinity stress in rice lead within 1 h to the up-regulation of a large number of rice transcripts encoding plant cytosolic ribosomal proteins (38). Our hybridizations with yeast distinguished isoforms at identity scores of ϳ90% at the 2-fold increase threshold that was chosen. Relaxing the stringency to, for example, 1.6-fold increase would have included an even larger number of Rpo transcripts. RPO up-regulation has also been observed by Posas et al. (30).
Other Up-regulated Functions-Other differences in gene expression between 10 and 30 min of salt stress were related to amino acid and nucleotide biosynthesis. At 10 min, induced ORFs in the categories amino acid (11.2%) and nucleotide metabolism (9.3%) are higher relative to 30 min (6.2% and 3.3%, respectively). Strongly up-regulated are 10 of 12 ORFs associated with amino acid biosynthesis (YAL004w, THR4, HOM2, FIG. 3. Comparison of RNA blot hybridizations and microarray expression data. RNA blots (12 are shown) for 39 ORFs were done in triplicate using RNA isolated from control (Ϫ) and salt-stressed yeast (ϩ) exposed for 90 min to 1 M NaCl. The RNA blot data (N) were generated using ImageQuaNT software and represent average changes in signals comparing stress to control. The probes were generated by PCR amplification of entire ORFs from genomic DNA and by random primer labeling with [␣-33 P]dCTP. The average microarray data (M) represent the average of three experiments (no salt versus 90-min salt stress), and standard deviation (S) was calculated using the "nonbiased" or "n Ϫ 1" method. The results for 3 of 39 ORFs did not agree (YDR387c (ITR1-like), YHR048w (unknown), YFR017c (unknown)), because expression at one of the time points was at background level and could not be measured accurately.
YER081w, PRS3, THR1, YIL074c, MET25, GLN1, and LYS21). Five of 11 are related to nucleotide metabolism: URA1, ADE1, PRS3, SHM2, and ADE13. The earliest metabolic activities, it seems, in need of up-regulation include amino acid biosynthesis and ribonucleotide metabolism. Between 10 and 30 min the transcriptional and translational machineries generate, modify, and transport proteins within the cell, and this process seems to be completed by 90 min. At that time, ORFs involved in protein synthesis have declined precipitously, and the majority of gene expression changes now target different functions, such as chaperone and detoxification increases and intracellular transport.
The category "energy supply" showed a peak of induction in the number of regulated transcripts, specifically in the subcategories respiration and the metabolism of energy reserves ( Fig.  2 and supplemental material). After 30 min of salt stress, 9.1% of the induced ORFs are components of the electron transport chain, such as cytochromes, or subunits of the ATP synthase complex, with the comparable percentages of induced ORFs, 5.6% and 4.0% after 10 and 90 min, respectively.
Functions in Cell Defense-Not surprisingly, the number of ORFs in the categories cell rescue, defense, cell death, and aging increased during the stress (Table II). These included the ORFs for glycerol and trehalose production, and five genes involved in cellular detoxification (GRX1 and TTR1 (glutaredoxin), the duplicated CUP1A/CUP1B (metallothionein), and CCP1 (cytochrome-c peroxidase precursor)). In the case of CUP1A/B, their high sequence homology led to cross-hybridization and only one of the genes may be up-regulated. YGP1 (gp37, glycoprotein secreted in response to nutrient limitation) is also induced, although its function during salt stress is not clear. It may be associated with cell wall biogenesis, because it is 47% homologous to SPS100, which is involved in cell wall formation (44).
The percentages of ORFs related to the defense response ( Fig. 2) are 10% (10 min), 5% (30 min), and 7% (90 min). The lower percentage at 30 min relative to other times can be attributed to heat shock proteins (HSPs) (Table II), which have been reported to be up-regulated in a variety of stresses such as oxidative stress (H 2 O 2 ), methyl methanesulfonate, and heat shock (45)(46)(47). Four HSPs (SSA1, HSC82, YGL128C, and YDR033W) are induced within 10 min; one HSP (HSP12) after 30 min, and nine HSPs after 90 min of exposure to salt (HSP12,  YRO2, HSP26, SSE2, HSP78, SSA4, MDJ1, HSP104, and  DDR48). Only HSP12 was induced during two time points (30 and 90 min). The groups of HSPs induced during 10 and 90 min are distinct, indicating that different sets of HSPs could have different functional targets in the early and later responses to salinity stress. Also, the majority (5 of 6) of induced genes involved in DNA repair (Table II) showed increased transcript amounts only during the later stress response (90 min). These included HSP12, FUN30, HEL1, THI4, and RNR4.
Transport Functions-During the early response to salinity stress, few induced ORFs were observed in the categories of the major facilitator superfamily (MFS) (three ORFs), nitrogen and sulfur metabolism (one ORF), and in lipid, fatty acid, and isoprenoid metabolism (six ORFs) (Fig. 2). This changed during the 90-min time point: 17 up-regulated ORFs encode MFS transport proteins, and 9 and 16 induced ORFs were involved in nitrogen and sulfur metabolism and in lipid, fatty acid, and isoprenoid metabolism, respectively (Table III). Members of the MFS are characterized as permeases with 12 transmembrane domains. Many proteins in this class act as sugar, amino acid, or multidrug transporters (49). Five of the 17 induced MFS genes have been characterized: the iron transporter, FET3; the amino acid transporters FUR4, PUT4, and BAP2; and the low affinity hexose transporter, HXT1. It remains to be seen whether the 12 unknown MFS transcripts are translated into membrane-associated proteins, and if their induction is salt-specific.
Among the ion transporters known to play a role in sodium detoxification (3,48,50,51), several transcripts were up-regulated. The Na-ATPases, ENA1-4, were induced after 30 min of salt stress. Vacuolar H ϩ -ATPase subunits were induced during all time points, and the H ϩ -ATPase PMA1 was not induced (Fig. 2, and supplemental information). It is difficult to deter-mine metabolic functions in stressed cells for the induced ORFs relative to nitrogen, sulfur, lipid, fatty acid, and isoprenoid metabolism. However, 12 ORFs related to nitrogen and sulfur metabolism were found to be up-regulated following a different stress, MMS (47), and four of these are also found during 90 min of salt stress (Table IV): NPR1 (serine/threonine kinase), and three unknown ORFs, YFL030W (similar to transaminases) and YPL135W and YOR226C (both similar to proteins involved in nitrogen fixation).
We also analyzed the regulation of stress response genes in a group of 84 ORFs, which include at least two STREs in their 5Ј-upstream regions (23). A subset of these genes is affected by the Hog1p-dependent signal transduction pathway (10,25,30).   Magnification of the Stress Response in a ⌬gdp1, ⌬gdp2 Strain-As a control to test filter reliability, hybridizations used RNA from a mutant strain, gpd1⌬ gpd2⌬ (31), which is deficient in glycerol production. The GPD1 and GPD2 transcripts were not detectable (not shown). The inability of these cells to osmotically adjust to high salinity also affected a number of other transcripts. The ⌬gpd1/2 strain, when grown for 60 min in 0.5 M NaCl, showed a significantly different transcript induction profile compared with wild type grown in 1 M NaCl (Table IV), a concentration that is lethal for this mutant. Up-regulated transcripts were mostly comparable to the 90min time point in wild type in 108 of highly induced transcripts (see the supplemental material), including most of the metabolic functions. Up-regulated (Ͼ2-fold) in the deletion mutant, but not in wild type, were another 87 transcripts (Table IV). Most conspicuous were transcripts for components of retrotransposons. Although one gene related to retrotransposons (YOR344c) was among the up-regulated ORFs in wild type, the number increased to 17 (including YOR344c) in ⌬gpd1/2. Cellular stress has been implicated as an agent that increases transposition (e.g. Ref. 63). In yeast, FUS3, a mitogen-activated protein kinase inhibits Ty1 retrotransposition (64). This tran-script is strongly down-regulated in ⌬gpd1/2 (Ϫ3.0) but unaffected by 1 M NaCl in wild type (see Supplemental Table sIV).
In addition, another 34 ORFs were induced for which no function is known (or where the putative function has not experimentally been verified). At least 11 of the ORFs strongly up-regulated in the mutant are functionally associated with mitochondria, possibly indicating a function in aging or cell death. Also, ␣-factor receptor, mating pheromone ␣-2 factor, mating factor ␣, agglutinin, and transcripts for surface proteins increased. Finally, an ORF with strong similarity to a protein kinase (Stl2p), and ORFs for two transcription factor-like proteins (YLR256w, YKL109w [HAP4]) were also newly up-regulated in the glycerol-deficient mutant. The transcriptional behavior in this mutant supports a connection between osmoregulatory and pheromone response pathways: The cross talk in signaling to regulate growth and osmotic stress defense pathways reported before (e.g. Ref. 52). We can only speculate that, in the absence of glycerol synthesis and accumulation in this strain, the hyper-osmotic condition leads to a severe stress that affects cell integrity to a much larger extent than in the wild type at even higher osmolarity.
Functionally Unknown Salinity-regulated ORFs-Unclassified, putative, and unconfirmed ORFs represent ϳ40% of the  IV Transcripts in yeast ⌬gpd1/gpd2 induced higher than in wild type Transcripts up-regulated in ⌬gdp1/2 after 60 min in 0.5 M NaCl, including the ϳ200 most highly induced ORFs, in comparison to transcripts upregulated at 10, 30, and 90 min in wild type.  2 In addition, more than 50 yeast genes for which functions are known are also up-regulated in salt-stressed plants where a similar progression of different up-regulated functions has been observed (38). 3 Several of the plant homologs identified ORFs termed hypothetical in yeast and exhibited similarities to putative serine/threoninetype protein kinases. Yeast knockout strains that eliminated these ORFs did, however, not show a clear phenotype under salt stress conditions. 3 Salinity Tolerance: A Progression of Programs?-We have catalogued changes in gene expression in yeast over time following a severe salt shock, which affects growth significantly.
Determining number and nature of induced transcripts and the changes in time of transcript categories allowed for several conclusions. The profile of the genes expressed is similar at the early times (10 and 30 min), but changes later (90 min) are considerable. The evidence points toward the initiation of protein synthesis and restructuring of protein composition as a major task for the initial period (ribosomal proteins, transcription machinery, amino acid synthesis, and utilization). Another part of the early response is the induction of chaperone-type proteins. Only early on are a few components of signaling pathways found up-regulated; we think that more of these components could be induced at even earlier time points or under less severe stress.
To place the information into context, we indicate four response types that allow for clear statements. 1) A few transcripts are rapidly and strongly up-regulated early. The most dramatically early up-regulated ORF (YDR276c) encodes PMP3/SNA1. Deletion of PMP3 confers hyper-sensitivity to sodium in mutants that lack sodium efflux systems (e.g. Pmr2p/Enap and Nha1p). A function for Pmp3p has recently been documented in the control of membrane polarization, such that its deletion leads to membrane hyper-polarization and increased influx of monovalent cations (56). A plant (Arabidopsis thaliana) PMP3 homolog complements this phenotype. PMP3 homologues, which are salt stress-dependently up-regulated in microarray analyses (57), have also been found in the halophytic plant Mesembryanthemum crystallinum (27,57) and in salt-stressed rice (38). PMP3 transcript induction in our experiments changed from an early induction (17-fold at 10 min, 14-fold at 30 min) to 2-fold induction at 90 min. 2) Ribosomal proteins and several other functions exemplify the second response type, gradual up-regulation to a peak at 30 min followed by the later repression of all RPOs with the exception of a few mitochondrial proteins. Among those up-regulated, some if them transiently, in the second phase are also components for mitochondrial functioning, glycolysis, transcription, inter-compartmental protein transport, and protein turnover.
3) The third type comprises ORFs unaffected or down-regulated early and gradually induced later. These included many known yeast salinity stress responses. ORFs in this category have also been reported as up-regulated in two previous array analyses (25,30) with few cases where the induction was close to the cutoff value of 2-fold (see supplemental materials). The increases reported in previous studies are generally higher than the fold increases found in our experiments, but all three data sets show the same trend (25,30). What might be upregulated 50-fold or 10-fold in previous reports is shown in our analyses to have an averaged induction of 20-to 5-fold (see Supplementary Table sI). In addition to these ORFs (25,30), our analyses detected other strongly up-regulated transcripts. A secreted glycoprotein (YGP1; YNL160w) for example, reported as 3-fold up-regulated (25), was 26-fold up-regulated in our experiments. ARE2 (acyl-CoA sterol acyltransferase) is strongly up-regulated as are a putative resistance protein (YOR273c), PEP4 (aspartyl protease), ERG5 (C-22 sterol desaturase), YLL028c (similarity to multidrug resistance proteins), GYP6 (GTPase-activating protein), or the CUP1A/1B metallothioneins. These have not been reported before (25,30), indicating strain differences or reflecting the different stress conditions. 4) The fourth response type, consistently up-regulated transcripts, falls in many different functional categories. Each of the four response categories includes up-regulated functionally unknown, putative, and/or questionable ORFs with the absolute numbers increasing at the 90-min time point (Fig. 2).
Among the late-induced ORFs, highly expressed transcripts identify membrane transporters, cell detoxification functions, and a number of unclassified ORFs. Plant ESTs exist with homology not only to functionally identified transcripts but also to ORFs whose functions remain to be determined. Resources that will aid in deciphering the roles of these unknown ORFs include the generation and availability of yeast deletion strains (e.g. Ref. 53), programs for protein function determination by protein-protein interactions (54) or in computation and the microarray expression data from related organisms such as other fungi, bacteria, and plants. The advantage provided by the three time points chosen in connection with a strong salt shock seems to lie in a drawn-out stress response that allowed for a separation of successively engaged response pathways. This succession identifies early, strongly up-regulated functions and intermediate and late functions. The analysis presented here does not, however, include the equally numerous down-regulated functions, which provide additional information about the effects of stress and the responses by which tolerance can be achieved. Further in-depth analyses, including additional time points, different salt stress treatments or other abiotic stress factors, and additional mutant strains will be necessary for a complete understanding of the yeast stress response. Our results extend the reported data (25,30) through the inclusion of several time points under severe stress conditions and integrates immediate cellular responses with long term metabolic and physiological events that seem to assure survival. The kinetics of this response may serve as a paradigm against which plant salinity stress responses could be scaled.  (38).