The Transcriptional Response of Yeast to Saline Stress*

Adaptation to changes in extracellular salinity is a critical event for cell survival. Genome-wide DNA chip analysis has been used to analyze the transcriptional response of yeast cells to saline stress. About 7% of the genes encoded in the yeast genome are induced more than 5-fold after a mild and brief saline shock (0.4m NaCl, 10 min). Interestingly, most responsive genes showed a very transient expression pattern, as mRNA levels dramatically declined after 20 min in the presence of stress. A quite similar set of genes increased expression in cells subjected to higher saline concentrations (0.8 m NaCl), although in this case the response was delayed. Therefore, our data show that cells respond to saline stress by inducing the expression of a very large number of genes and suggest that stress adaptation requires regulation of many cellular aspects. The transcriptional induction of most genes that are strongly responsive to salt stress was highly or fully dependent on the presence of the stress-activated mitogen-activated protein kinase Hog1, indicating that the Hog1-mediated signaling pathway plays a key role in global gene regulation under saline stress conditions.

Yeast cells have been considered an excellent model for the study of the mechanisms underlying tolerance to saline stress, particularly because it has been shown that fungi and higher plants not only have similar ion transport systems at their plasma membranes (1), but they also share similar cation detoxification mechanisms (2) and, most probably, signal transduction pathways (3,4).
Exposure of yeast cells to saline stress implies both exposure to specific cation toxicity and to osmotic stress. Certain ions such as Na ϩ or Li ϩ are toxic to cells due to their ability to inhibit specific metabolic pathways, probably through inhibition of specific targets. This has been shown to be the case for the yeast Hal2 protein and certain RNA-processing enzymes (5,6). Therefore, regulation of intracellular ion contents represents an important response to ion stress. For instance, expo-sure to sodium increases the expression of the ENA1/PMR2A gene, encoding a P-type ATPase responsible for Na ϩ and Li ϩ ion efflux (7,8).
Increases in extracellular osmolarity results in a transient induction of the expression of stress protective genes. A major outcome from this response is the accumulation of intracellular glycerol, which relies on the activation of the Hog1 1 (high osmolarity glycerol response) MAP kinase pathway (9,10). MAP kinases play a key role in regulation of stress responses in many organisms from mammals to yeast (11). Hog1 MAP kinase is essential for the survival of yeast in high osmolarity environments (9) and is activated under osmotic-stress conditions by two independent osmo-sensors, a two-component system and the transmembrane protein Sho1 (12)(13)(14). These sensing mechanisms activate a kinase cascade that involves the Ssk2, Ssk22, and Ste11 MAPK kinase kinases (13,15), the Pbs2 MAPK kinase, and finally, the Hog1 MAPK. Once phosphorylated, the Hog1 MAPK is translocated into the nucleus, where it induces diverse stress responses. It is worth noting that both phosphorylation and nuclear localization of Hog1 are very rapid and transient (13,16).
The mechanism of gene regulation through activated Hog1 is still unknown because transcription factors under the control of this MAP kinase are not well characterized. Several candidates, however, have been described. These are the transcription factors Msn1, Msn2, and Msn4 (17,18), the bZIP-type protein Sko1 (19), and Hot1 (18). Although the requirement for the Hog1 kinase has been demonstrated for the osmotic upregulation of a number of genes, an exhaustive list of the genes required for osmo-stress adaptation is far from complete.
It is important to understand that responses to ion stress require the activity of several pathways and that a single gene can receive different inputs (20). For instance, the expression of the ENA1 ATPase is regulated by both a calcium signaling pathway, which involves the protein phosphatase 2B (calcineurin), and the HOG signaling pathway (21). We felt that better understanding of the yeast response to saline stress could be achieved from the use of DNA microarrays (reviewed in Ref. 22) to perform a genome-wide analysis of the transcriptional response under this type of stress. This technology has been used to accurately analyze whole genome expression in several organisms, including yeast (23)(24)(25)(26). Because a number of examples suggested that different conditions (such as time of exposure to salt) might result in a different pattern of expression, we decided to test two different NaCl concentrations at two different time points. In addition, the role of the Hog1 MAP kinase has been investigated by testing the transcriptional response of a hog1 mutant strain under stress.

Generation of Open Reading Frame (ORFs) DNA for Deposition in
Microarrays-Saccharomyces cerevisiae genes were reamplified from 6,035 polymerase chain reaction-amplified full-length ORFs provided by Research Genetics (Huntsville, AL). Each of these originally amplified ORFs of S. cerevisiae contained the sequence 5Ј-GGAATTCCAGCT-GACCACC immediately 5Ј of the start ATG, thereby making it possible to re-amplify each ORF with the common primer YAMP1 (5Ј-GCAGTCGTGGAATTCCAGCTGACCA) and the appropriate gene-specific 3Ј primer, which was also provided by Research Genetics. Amplified ORFs were purified by using Qiagen (Valencia, CA) 96-well polymerase chain reaction clean-up kits. This amplification and purification process resulted in the production of deposition targets for approximately 85% of the S. cerevisiae ORFs.
Production of Microarrays, Hybridization, and Scanning-The yeast gene targets were arrayed in a 12-tip format using a quill-type pin by an Intelligent Automation Systems (Cambridge, MA) microarrayer. The microarrayer tip delivered approximately 4 nl per spot on silylated aldehyde-coated glass slides (CEL Associates, Houston, TX). Yeast microarrays were hybridized for 4 h under cover slides with a Cy3-dCTP (Amersham Pharmacia Biotech)-labeled cDNA probe. After hybridization, the labeled microarrays were washed and dried. The fluorescently labeled microarrays were then subsequently scanned using a confocal laser ScanArray 3000 (General Scanning Inc.) system. Data was collected using ImaGene software (BioDiscovery Inc.).
RNA Preparation-Cells were grown in YEPD (10 g/l yeast extract, 20 g/l peptone, and 20 g/l dextrose) at 28°C to an optical density of 0.7 and then subjected to a brief saline shock. RNA was obtained from untreated cells or cells treated with 0.4 M NaCl or 0.8 M NaCl for 10 min and 20 min, respectively. Cells were centrifuged for 5 min at 7000 rpm, and total RNA was extracted by using hot phenol and glass beads as described (27). Poly(A) ϩ RNA was purified from total RNA with an mRNA isolation kit (Roche Molecular Biochemicals) based on biotinlabeled oligo(dT) probes and streptavidin bound to magnetic particles. Purification was carried out following the manufacturer's instructions. Purified poly(A) ϩ RNA appeared undegraded on an agarose gel. Ribosomal bands were substantially reduced, and the final poly(A) ϩ RNA yield was approximately 1.5% of the total RNA. Before cDNA synthesis, RNAs were tested by Northern blot using specific probes.
Probe Preparation and Labeling-Fluorescently labeled cDNA was prepared from poly (A) ϩ RNA by oligo dT-primed polymerization using Superscript II reverse transcriptase (LTI, Gaithersburg, Md). The pool of nucleotides in the labeling reaction was 0.5 mM dGTP, dATP, and dTTP and 0.04 mM dCTP and Cy3-dCTP (Amersham Pharmacia Biotech) at 0.5 mM. Probes were purified using Qiagen (Valencia, CA), Qiaquick polymerase chain reaction purifications kits.

RESULTS AND DISCUSSION
Transcriptional gene induction has been recognized as an important mechanism for adaptation of yeast cells to saline stress. We have performed a DNA microarray analysis of the global transcriptional response of the yeast genome under NaCl stress. Because of the very large amount of information generated, only the most relevant aspects will be presented in this report, although the whole set of data generated from this work can be retrieved from the Universidad Autonoma de Barcelona web site. Our results revealed that expression of a quite large number of genes increased after a short exposure (10 min) to a relatively low concentration of NaCl (0.4 M). As shown in Table I, the mRNA level of more than 1300 genes increased at least 3-fold under these conditions. In some cases the increase was rather dramatic, with a ratio of expression between stressed and non-stressed conditions higher than 10 (these genes are denoted as very highly induced genes). Interestingly, exposure of the cells for a longer period (20 min) reduced the number of responsive genes by a factor of about 10-fold. When cells were exposed for 10 min to a higher NaCl concentration (0.8 M), the number of genes with at least a 3-fold increase in its mRNA level was much lower (about 400). However, the number of responsive genes doubled when cells were stressed for 20 min. This figures clearly indicate that, at the genome level, time of exposure to NaCl and the concentration of salt used markedly affect the results obtained.
One of the most interesting aspects derived from our data is that at 0.4 M NaCl, a concentration of salt widely used to test responses to osmotic shock, the transcriptional response is rather early and largely transient. As shown in Table II, about 93% of the induced genes shows an early response. We consider here early response genes whose signal increases at least 3-fold after 10 min of stress and is at least 80% of the signal at 20 min. Among them, about 85% display a transient response, defined as the case in which the level of expression at 10 min at least doubles the level of expression at 20 min. This implies that many transcriptional responses, when tested at the mRNA level, could be missed if cells are exposed for too long to stress. This situation is reflected in Fig. 1. Genes were listed on the basis of their increase in expression under a stress of 0.4 M NaCl for 10 min. The top 50 genes were selected, and their induction levels at 0.4 M and 0.8 M NaCl (after exposure of the cells for 10 and 20 min) were plotted. As can be observed, many genes that showed a strong increase in expression after 10 min at 0.4 M sharply decreased after 20 min. Fig. 1 also presents another interesting fact. Most of the genes that are highly induced after 10 min at 0.4 M NaCl remain almost silent when cells are challenged with 0.8 M NaCl (Fig. 1, left panel). However, in many cases, a strong response is triggered after 20 min of exposure to high NaCl concentration ( Fig. 1, right panel). This observation indicates that exposure to a high concentration of salt results in a delayed transcriptional response. A similar conclusion has been recently reached regarding the kinetics of induction of the gene GPD1, a key player in the response to osmotic stress in yeast (28). Our data indicates that this delayed response affects most genes that are induced at high salinity, as it can be deduced from the observation that a majority of these genes (about 62%) showed a response that can be defined as moderately late or late, whereas in cells treated with 0.4 M NaCl this behavior is observed in less than 7% of the TABLE I Global overview of changes in expression of yeast genes after exposure of cells to NaCl Data obtained from genomic chips was analyzed, and genes showing an increase in expression higher than 3-fold after saline stress were grouped within three classes: very highly induced (VHI), highly induced (HI), and moderately induced (MI). Figures in parentheses denote the percentage of each class with respect of the total number of scored genes.
Level of induction induced genes (Table II).
In conclusion, evaluation of transcriptional responses after saline stress must take into account the timing of the response. For instance, a recent survey (29) of the transcriptional response of 250 novel yeast genes from chromosome XIV after exposure of cells to 0.7 M NaCl for 1 h revealed only five regulated genes (YNL274c, YNL195c, YNL194c, YNL066c and YNL051w). Our screening also revealed three of them (YNL274c, YNL195c, and YNL066c) as increased at least 3-fold, but in addition, 27 more genes were identified as regulated after a 20-min exposure to 0.8 M NaCl. In light of our data, it seems reasonable to assume that the observed difference is the result of the fact that, after 1 h of exposure, most of the genomic transcriptional response vanished.
Quite a few genes ranking in the very highly induced and highly induced categories after exposure to 0.4 M NaCl for 10 min do not have a recognized functional role, and for many of them, the response described here represents the first clue on functional data available. In other cases, increases in the mRNA level after salt stress had been previously documented, indicating that the data generated by the genomic chip analysis was reliable and can be safely compared with reported data. As far as we could detect, the CTT1 gene was the only one not showing the expected behavior, since no induction was detected from the microarray hybridization experiments (despite the strong induction observed in our preliminary Northern blot experiments). Subsequent analysis showed that this specific target was among the few that failed to amplify or purify, and therefore, no signal could be expected. Table III presents a list of highly induced genes after exposure to 0.4 M NaCl for 10 min that can be ascribed to known functional families. Genes encoding proteins involved in carbohydrate metabolism are well represented. As mentioned above, synthesis of glycerol is a major response of yeast cells to osmotic stress (see Ref. 30 for a review). In addition to the well characterized genes GPD1 and GPD2, responsible for the synthesis of glycerol, we have detected a strong increase in expression in a number of genes encoding plasma membrane sugar transporters, such as STL1, HXT10, HXT7, and HXT5. Of particular interest is STL1, a gene encoding a putative hexose transporter still to be functionally characterized (31). This gene shows the strongest induction after 10 min at 0.4 M NaCl, whereas after 20 min its mRNA level has declined to reach almost basal levels. Genes encoding glucose-phosphorylating enzymes, such as GLK1 and the related ORF YDR516c, as well as HXK2 were also strongly induced.
A close correlation between trehalose content and stress tolerance, including osmotic stress (32,33), has been established in many cases, although both circumstances can be uncoupled (34). Our results indicated that expression of trehalose-metabolizing enzymes as Tsl1, Tps1, and Tps3 (encoding subunits of the trehalose 6-phosphate synthase-phosphatase complex) was increased, as previously documented (32,34), although under our conditions, Tps2, the remaining member of the complex, was only slightly affected. The neutral trehalase encoded by NTH1, which contains three STRE (stress-responsive element) sequences in its promoter, was also greatly induced, despite that it was previously reported as insensitive to NaCl exposure (35).
Another interesting set of genes that quickly respond to a mild NaCl stress are those related to glycogen metabolism. It has been shown that GSY2 (encoding the major isoform of glycogen synthase) and GPH1 (encoding glycogen phosphorylase) are induced under saline stress (32). We show here that this can be considered within the frame of a more general effect, since virtually every gene involved in the synthesis or degradation of glycogen shows a positive response under saline stress. We include in this group genes such as UGP1, encoding UDP-glucose pyrophosphorylase (the enzyme that forms UDPglucose, the direct substrate for glycogen synthase), the less prominent isoform of glycogen synthase (GSY1), the branching enzyme GLC3, and even the self-glucosylating initiator of glycogen synthesis, GLG1. Genes encoding components known to modulate glycogen synthase activity can be also included, as the type 1 protein phosphatase Glc7 (36) and its glycogenspecific regulatory subunit Gac1 (previously reported as inducible by saline stress in Ref. 32) as well as the type 2A phosphatase Pph21 (37). A remarkable feature is that, in contrast with the observation that the induction of a large number of genes after short term exposure to 0.4 M NaCl is substantially dependent of the presence of the HOG1 gene (see below), the induction of this set of genes appears to be largely independent of the HOG1-mediated signaling pathway. The general response of genes encoding glycogen-metabolizing enzymes is very intriguing, because although salt stress results in a mild glycogen accumulation (32), the role of this polysaccharide in stress protection is still obscure.
Exposure of cells to 0.4 M for 10 min also results in the induction of a rather large number of components of the protein biosynthesis machinery. This includes 31 genes encoding ribosomal proteins (both cytosolic and mitochondrial), a figure that represents about one-fourth of the 137 genes encoding ribosomal proteins found in yeast (38). It is worth noting that transcriptional regulation of the expression of ribosomal proteins has been recognized as an important mechanisms to control ribosome assembly and function. Although a general pattern  cannot be established, in many cases the induction of the ribosomal genes is largely or fully independent of the presence of the HOG1 gene. In addition to specific ribosomal genes, response to saline stress involves increases in the level of several aminoacyl-tRNA synthetases and a number of translation initiation factor mRNAs, including TIF1 and TIF2, PRT1, TIF11, HYP2, and CDC95.
Saline stress results in transcriptional induction of a number of genes encoding proteins related to ion homeostasis. These include the well characterized gene ENA1/PMR2A, encoding a P-type, salt-inducible ATPase, which is the major determinant for Na ϩ efflux (7,8). Interestingly, a remarkable induction is observed for several components of the vacuolar H ϩ -ATPase complex, such as VMA6, VPH1, VMA7, VMA5, and TFP1. The role of this complex is pivotal in creating the electrochemical gradient of protons required for sequestration of sodium into the vacuole (39). In plant cells, tonoplasts are fundamental for ion compartmentation in the vacuole. It is remarkable that, in TABLE III Classification into functional families of ORFs whose transcripts are induced more than 5-fold (VHI plus HI) after 10 min of treatment with 0.4 M NaCl The asterisks denote the existence of previously reported experimental data, proving that the transcription of the specific gene was induced by saline stress. In some cases, the intensity of the hybridization signal was out of the linear range of the detection system, resulting in an increase in expression probably higher than the one calculated. This is denoted by a Ͼ symbol. The total number of ORFs within the VHI and HI classes was 461. The complete set of data for all the conditions tested will be accessible through the Universitat Autonoma de Barcelona  tobacco cells, an increase in the mRNA levels for the 70-kDa catalytic subunit of the tonoplast H ϩ -ATPase has been reported after a short term exposure to NaCl (40). Finally, a number of genes encoding proteins required for signal transduction are also induced. Interestingly, these include a catalytic (SRA3) and regulatory (SRA1) subunit of the cAMP-dependent protein kinase as well as cytosolic adenylate kinase. This is remarkable because a role for cAMP signaling in general stress response as well as in regulation of sodium extrusion (20) has been postulated. Known components of signal transduction pathways related to stress response appears also induced. Examples are the MSN2 transcriptional activator (17,41), the HOT1 transcription factor (18), the HOG1 MAP kinase (42), and the RCK2 kinase (which appears to be a direct substrate for the Hog1 MAPK) (47), as well as the BMH1 gene, encoding a protein of the 14-3-3 family essential for Ras/MAPK cascade signaling during pseudohyphal development (43).
Transcriptional response to stress is often mediated through the STRE system (core consensus AG 4 or C 4 T). A recently conducted computer search for STRE-containing promoters in the yeast genome revealed 69 candidate genes for stress-induced response (44). The expression of several of them was tested for a variety of stresses, although the response of the majority (54 genes) remained still uncharacterized from an experimental point of view. We have analyzed the expression levels of this subset of genes and found that 27 of them (50%) increase their expression at least 3-fold in wild type cells under at least one of the conditions tested. In fact, an interesting aspect derived from the availability of data on the transcriptional response of the yeast genome to a defined set of conditions is that this data allows for computer analysis in search of motifs present in the promoters of responsive genes that could be candidates for mediating such response. Although a detailed analysis remains to be done, we have used the method described by van Helden et al. (45) to detect over-represented oligonucleotides within the promoter regions of salt-responsive genes. When this kind of analysis was performed by selecting the region comprising nucleotides Ϫ600 to Ϫ1 for the 100 genes that showed a stronger induction after exposure to 0.4 M NaCl for 10 min, the sequences AG 4 and C 4 T appeared heavily overrepresented. An equivalent set of promoters dissected from genes that did not show any response to salt stress failed to produce this (or any other) significant consensus pattern. Induction of a number of genes after osmotic chock has been documented to be mediated at least in part by the Hog1 signal transduction pathway (42). Here we have analyzed the contribution of the Hog1 pathway by comparing the level of induction of responsive genes in wild type cells and in a strain lacking the MAP kinase gene. It is worth noting that, as far as we know, this is the first report on the effect of the absence of a stressactivated MAP kinase on the transcriptional response at the global level. Analysis of the data indicated that in many cases the response in hog1 cells was different from that observed in wild type cells. However, from a quantitative point of view, the changes ranged from a weak effect to a virtual loss of the response. This indicates that the Hog1 pathway is certainly involved in the transcriptional response of most salt-responsive genes, but in many cases, it is not the unique relevant signaling pathway. The existence of different signaling pathways controlling the osmotic stress response has been documented in some detail in a few cases, including the ENA1 (20) and the GPD1 (28) genes.
As shown in Table IV, the level of dependence of Hog1 differs upon the stress conditions. Interestingly, in a large number of cases (75.5%), the response after 20 min of exposure to 0.8 M NaCl was fully or strongly dependent on the presence of Hog1, whereas in cells exposed to 0.4 M for 10 min, other pathways appears to be highly relevant. Therefore, our data confirm and expand a previous report based in two-dimensional analysis of 35 S-labeled proteins after 20 -40 min of exposure of the cells to 0.7 M NaCl (46). Akhtar et al. (46) report that 29 proteins were strongly induced under these conditions (although only seven were actually identified). In many cases, induction was largely or fully dependent on the existence of the PBS2 gene and, therefore, of activation of Hog1. A relationship between the intensity of the response and the involvement of the Hog1 kinase can be drawn (Fig. 2) from the observation that genes considered very highly induced are more dependent on Hog1 than genes only moderately induced. In any case, our data indicate that there are more than 200 yeast genes whose response to saline stress relies almost exclusively on the activation of the HOG pathway, thus proving a key role for an stress-activated MAP kinase in the generation of a wide range of transcriptional responses. A most exciting challenge will be to define at the molecular level the connection between the activation of this kinase and each specific transcriptional response.
In conclusion, in this report we show that exposure of yeast cells to saline stress results in a substantial transcriptional response. A relevant outcome of our global analysis is that the magnitude of the changes is defined in many cases by both the timing of the response and the intensity of the stress. A clear example of this can be found in Fig. 3, in which the responses of a number of genes identified in our global microarray analysis and representative of different functional families, have been tested by conventional Northern blot analysis. We can observe, for instance, the response of gene STL1, which is extremely intense after both mild and severe saline stress. Interestingly, the response is early and very transient in the first case and late under the second condition, although always fully dependent of the presence of the Hog1 kinase. In contrast, the ORF YDR057w, encoding a protein of unknown function, shows a late response under both stress conditions, which is essentially independent of the Hog1-mediated signal transduction pathway.

TABLE IV
Dependence of gene induction upon saline stress on the presence of the Hog1 MAP kinase Genes induced at least 3-fold under the indicated conditions were grouped into four categories according to the degree of induction observed in the absence of the HOG1 gene: fully dependent (less than 15% of the induction observed in wild type cells), strongly dependent (from 15% to 50%), weakly dependent (50 to 77%), and non-dependent (more than 77%). A total number of 1351 (0.4 M, 10 min) and 728 (0.8 M, 20 min) genes were scored.   5) or hog1⌬ cells (lanes 6 to 8) were grown in yeast extract/peptone/dextrose medium and treated with NaCl. Lanes 1 and 6, no NaCl was added; lanes 2 and 7, 0.4 M NaCl was added for 10 min; lane 3, 0.4 M NaCl was added for 20 min; lane 4, 0.8 M NaCl was added for 10 min; lanes 5 and 8, 0.8 M NaCl was added for 20 min. Twenty micrograms of total RNA/lane was loaded in the gels. Filters were probed sequentially with the indicated probes. The signal of the ribosomal protein RPL28A, whose expression was unmodified by the stress treatments, is shown here as a reference for equivalent RNA loading and transfer.  Table IV, were grouped on the basis of the intensity of their response to saline stress as described in Table I. Open bars denote fully or strongly Hog1-dependent genes, hatched bars indicate weakly dependent genes, and crossed bars account for Hog1-independent genes.