Hyperosmotic Stress Stimulates Promoter Activity and Regulates Cellular Utilization of the Serum- and Glucocorticoid-inducible Protein Kinase (Sgk) by a p38 MAPK-dependent Pathway*

We have established that the serum- and glucocorticoid-inducible protein kinase (Sgk) is a new component of the hyperosmotic stress response. Treatment of NMuMg mammary epithelial cells with the organic osmolyte, sorbitol, caused the stable accumulation of Sgk transcripts and protein after an approximately 4-h lag. Transient transfection of a series of sgk-CAT reporter plasmids containing either 5′ deletions or continuous 6-base pair substitutions identified a hyperosmotic stress-regulated element that is GC-rich and is necessary for the sorbitol stimulation ofsgk gene promoter activity. Gel shift analysis identified four major DNA-protein complexes in the hyperosmotic stress-regulated element that, by competition with excess consensus wild type and mutant oligonucleotides and by antibody supershifts, contains the Sp1 transcription factor. Several lines of evidence suggest that the p38 MAPK signaling pathway mediates the hyperosmotic stress stimulation of sgk gene expression. Treatment with pharmacological inhibitors of p38 MAPK or with a dominant negative form of MKK3, an upstream regulator of p38 MAPK, significantly reduced or ablated the sorbitol induction ofsgk promoter activity or protein production. Using anin vitro peptide transphosphorylation assay, sorbitol treatment activates either endogenous or exogenous Sgk that is localized to the cytoplasmic compartment. Thus, we propose that the stimulated expression of enzymatically active Sgk after sorbitol treatment is a newly defined component of the p38 MAPK-mediated response to hyperosmotic stress.

intracellular protein kinase cascades that are generally conserved between metazoans and mammals, culminating in the transcriptional control of specific sets of target genes. The best characterized osmoregulatory signaling pathways have been described in Saccharomyces cerevisiae, in which solute imbalances in the environment are detected by two different receptors, the histidine kinase receptor Sln1 and the four-pass transmembrane receptor Sho1p. The activation of kinase cascades by these receptors leads to the phosphorylation and activation of the HOG1 1 proline-directed protein kinase (4,5). The HOG1 gene is necessary for the maintenance of osmotic gradients in hyperosmolar environments and allows the proliferation of yeast under osmotic stress (6). HOG1 has been shown to phosphorylate various members of the Msn family of transcription factors, resulting in their binding to and transactivation of stress-responsive elements located in the upstream promoter regions of osmotically regulated genes (7)(8)(9)(10). One class of osmoregulatory genes targeted by HOG1 encode proteins that have osmoprotective functions, such as the glycerol-3-phosphate dehydrogenase gene that is responsible for the accumulation of the yeast osmolyte, glycerol (10,11).
In mammalian cells, however, much less is known about the signal transduction pathways that control gene transcription in response to environmental insults. The mammalian homologue of the yeast HOG1 is p38 MAPK, which is a member of the mitogen-activated protein kinase (MAPK) family of serine/ threonine protein kinases, has been shown to play a role in transducing stress signals in a variety of cell systems (1,(12)(13)(14). Activation of p38 MAPK results in the phosphorylation of a variety of targets, a subset of which regulates transcription of stimulus-specific genes (15). For example, the p38 MAPK phosphorylation of the ATF2, CHOP, Elk1, MEF2C, and SAP1 transcription factors has been shown to induce their DNA binding and transactivation activity (16 -19). A number of protein kinases (such as MAPKAP kinase-2, MAPKAP kinase-3, Mnk-1, PRAK, and RSK-B) have also been shown to be targets of p38 MAPK. These studies suggest the existence of additional hitherto unidentified downstream targets of p38 MAPK signaling that may be responsible for mediating transcription of mammalian osmoregulatory genes (20 -25). Although, in some cases, p38 MAPK-initiated phosphorylation events can trigger programmed cell death (26,27), the functional connections between the immediate targets of p38 MAPK, its downstream effectors, and the cellular response to hyperosmotic stress in mammalian cells have yet to be elucidated.
Conceivably, important downstream targets of the p38 MAPK stress cascade may themselves be cell-signaling molecules that help trigger and mediate the selectivity of the stress response to environmental cues. We have reported the isolation and characterization of a serine/threonine serum-and glucocorticoid-inducible protein kinase gene, sgk, by subtractive cloning from a mammary tumor cell cDNA library that is under acute transcriptional control by both serum and glucocorticoids (28,29). sgk is approximately 45-55% homologous to Akt/PKB, protein kinase A, protein kinase C, and the rat p70S6K/ p85S6K. In serum-treated cells, Sgk is phosphorylated and shuttles between the nucleus and cytoplasm in a cell cycle-dependent manner (30). We have recently shown that Sgk enzymatic activity, phosphorylation and subcellular localization is regulated by PI 3-kinase signaling through PDK1, suggesting a possible role for Sgk in a cell survival pathway (31). Moreover, the physiological stress hormones, glucocorticoids, stimulate sgk promoter activity through a glucocorticoid response element, and sgk is a transcriptional target of the p53 tumor suppressor gene, a known target of genotoxic stress (28,32). Consistent with a role for Sgk in the cellular stress response, sgk transcripts have been shown to be elevated in response to ischemic injury of the brain, changes in cell volume, and inflammatory disease and in a screen for transcripts involved in wound repair in fibroblasts (33)(34)(35)(36). Therefore, to determine whether transcription of the sgk gene is a component of the stress response in mammalian cells, the hyperosmolar regulation of Sgk was examined in mammary epithelial cells. We demonstrate that an increase in extracellular solute levels stimulates sgk promoter activity and gene expression through the p38 MAPK-dependent pathway.

EXPERIMENTAL PROCEDURES
Cells and Materials-NMuMg nontransformed mouse mammary epithelial cells were originally derived from normal glandular tissue of an adult NAMRU mouse (37). Cells were regularly cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 10 g/ml insulin, 50 units/ml penicillin, and 50 g/ml streptomycin. Cells were propagated at 37°C in humidified air containing 5% CO 2 . Cells were passaged every 2-3 days. Cell culture reagents such as Dulbecco's modified Eagle's medium, fetal bovine serum, calcium-and magnesium-free phosphate-buffered saline, and trypsin/EDTA were supplied by BioWhittaker, Inc. (Walkersville, MD). Insulin and D-sorbitol were purchased from Sigma. [ 3 H]Acetyl coenzyme A (200 mCi/ mmol) and [␣-32 P]dATP were obtained from NEN Life Science Products. To induce hyperosmotic stress, cells received equal volumes of Dulbecco's modified Eagle's medium (control) or 0.3 M sorbitol (Sigma) in Dulbecco's modified Eagle's medium for the indicated time. For treatment with the LY294002 PI 3-kinase inhibitor, cells were treated with 50 M LY294002 for 16 h, half of the cell cultures received serum-free medium with LY294002, and the other half received 0.3 M sorbitol dissolved in serum-free medium containing LY294002. Cells were harvested 24 h later and lysed, and whole cell extracts were analyzed for Sgk protein expression as described below.
Plasmid Constructions-The construction of the Ϫ4.0 sgk-CAT plasmid, which contains sgk promoter sequences (Ϫ3560 to ϩ 51) cloned upstream of the chloramphenicol acetyltransferase (CAT) gene in the vector pBLCAT3, has been described previously (29). The various sgk promoter-CAT deletions down to Ϫ236 sgk-CAT were generated utilizing the Erase-a-Base system (Promega, Madison, WI) and are further described (32). The Ϫ78 sgk-CAT was generated as described (38). For the Ϫ67 wild type and mutant sgk-CAT vectors, sense and antisense oligonucleotides of the region between Ϫ67 and Ϫ35 were synthesized as per usual techniques (Microchemical Facility, Cancer Research Laboratory, University of California, Berkeley), and then 800 pmol of each strand were mixed and annealed. The wild type and mutant doublestranded oligonucleotides were then subcloned into the pCAT Basic vector (Promega) containing the region between Ϫ35 and ϩ55 of the sgk promoter. The Ϫ67 wild type and mutant vectors were confirmed by DNA sequencing.
Transfection Methods and CAT Reporter Gene Assays-NMuMg mammary epithelial cells from 60 -70% confluent cultures in 100-mm tissue culture plates were transfected by the calcium phosphate precipitation method as described previously (32). The total amount of DNA used in calcium phosphate transfections for CAT assays was held constant at 20 g, and in appropriate transfections, the total DNA was adjusted to this amount using the empty CAT vector plasmid pBLCAT3 or pCAT Basic. Following transfection for at least 4 h, cells were exposed to a 15% glycerol shock for 3 min at 37°C. Transfections were performed in triplicates and repeated at least three times. At 36 -40 h post-transfection, the cells were harvested for CAT assays, and the protein content of the cell extracts was estimated with the Bradford procedure (39). A quantitative nonchromatographic assay (40) was used to measure CAT activity in the cell extracts as detailed elsewhere (28) Isolation of RNA and Northern Blot Analysis-NMuMg cells were plated on 10-cm dishes so they were 80 -90% confluent on the day of harvest. Cells were scraped on ice using a rubber policeman, washed in 1ϫ PBS, and then pelleted. To each plate, 100 l of TSM buffer (10 mM Tris-Cl, pH 7.5, 150 mM NaCl, 2 mM EDTA, 0.5% Nonidet P-40, 1.5 mM MgCl 2 ) plus 1 mM dithiothreitol was added, and the cells were incubated for 5 min on ice. Nuclei were pelleted by spinning at 14,000 rpm in an Eppendorf centrifuge at 4°C. The supernatant was transferred to new tubes, and 100 l of digestion buffer (0.2 M Tris-Cl, pH 8.0, 25 mM EDTA, 0.3 M NaCl, 2% SDS) plus 100 g/ml proteinase K was added. The samples were incubated at 37°C for 30 min and then extracted two times with 1:1 phenol/chloroform. RNA was precipitated with ethanol overnight at Ϫ70°C, and pellets were resuspended in diethyl pyrocarbonate-treated water plus 0.1% SDS.
Approximately 10 -13 g of RNA isolated from each sample was separated on a 1.1% agarose/formaldehyde gel run at 3-4 V/cm for approximately 3 h. The RNA was transferred overnight by blotting onto MSI Magnagraph nitrocellulose. The RNA was covalently cross-linked onto the membrane using a Stratagene Stratalinker, and then transfer was confirmed by staining the blot with 0.02% methylene blue dye. The blot was prehybridized for a minimum of 2 h as described previously (28). The blot was then probed using a SspI/PstI fragment of murine sgk cDNA that was labeled by [␣-32 P]dCTP random primed labeling (Roche Molecular Biochemicals). A total of 10 ϫ 10 6 cpm/ml of probe in hybridization buffer was added to the blot and allowed to incubate at 42°C overnight. The blot was washed in 2ϫ SSC, 0.1% SDS twice for a total of 1.5 h at room temperature. The blot was then exposed to a Molecular Dynamics PhosphorImager screen for 24 h.
Gel Mobility Shift Assays-Preparation of nuclear extracts from NMuMg mammary cells was based on the method of Dignam et al. (41). The protein contents in the nuclear extracts were determined by the Bradford procedure (39). The sequences of the various oligonucleotides (sense) used for DNA binding studies and cold competition experiments are as follows (mutations indicated in boldface type): sgk Ϫ67/Ϫ35, 5Ј-GGTCCCGCCTGCCCCGCCCCCTGGAGGCTC-3Ј; -67/Ϫ35 mutant 1, 5Ј-TTCGAAGCCTGCCCCGCCCCCTGGAGGCTC-3Ј; Ϫ67/Ϫ35 mutant 2, 5Ј-GGTCCCTTCGAACCCGCCCCCTGGAGGCTC-3Ј; Ϫ67/Ϫ35 mutant 3, 5Ј-GGTCCCGCCTGCTTCGAACCCTGGAGGCTC-3Ј; Ϫ67/ Ϫ35 mutant 4, 5Ј-GGTCCCGCCTGCCCCGCCTTCGAAAGGCTC-3Ј; Ϫ67/Ϫ35 mutant 5, 5Ј-GGTCCCGCCTGCCCCGCCCCCTGGTTCGAA-3Ј. The Sp1 consensus oligonucleotide sequence is 5Ј-cgatGCCCCGC-CCCagtc-3Ј. The mutant Sp1-binding sequence is 5Ј-cgatGCCCTTTC-CCagtc-3Ј. The core CGC nucleotides have been changed to TTT, and Sp1 is unable to bind to this sequence (42). The EGR1 consensus oligonucleotide sequence is 5Ј-cgatTCGCCCCCTCtgac-3Ј. All oligonucleotides were synthesized by a model 394 synthesizer in the Cancer Research Laboratory Microchemical Facility of the University of California, Berkeley. Radiolabeling of 5Ј-ends of the appropriate oligonucleotides was carried out in the presence of equal amounts (10 pmol) of sense and antisense strands, [␣-32 P] ATP (6000 Ci/mmol; ICN Biomedicals Inc.), and T4 polynucleotide kinase (Roche Molecular Biochemicals) at 37°C for 30 min, followed by annealing of the labeled strands. The annealing procedure for generating either labeled or unlabeled DNA involved adding 0.1 M NaCl (0.1 volume) to equal amounts of sense and antisense strands, heating for 10 min at 70°C, and gradually cooling to room temperature. The free unincorporated nucleotides and singlestranded DNA were separated from the end-labeled double-stranded DNA by native polyacrylamide gel electrophoresis (8% gel). The labeled double-stranded oligonucleotide was excised and eluted in 400 l of TE buffer (10 mM Tris, 1 mM EDTA) and 40 l of 3 M sodium acetate, pH 5.0, for 3 h, followed by ethanol precipitation, rinsing in 70% ethanol, and resuspension in TE buffer. The radioactive oligonucleotide probes were stored at Ϫ70°C. The DNA binding reactions (18 l) contained nuclear extract proteins (7 g), 0.5 ng of 32 P-labeled (approximately 2 ϫ 10 6 cpm) DNA probe, poly(dI-dC) (2 g), and 6 l of 3ϫ binding buffer (21.6% glycerol, 36 mM Hepes, pH 7.5, 126 mM KCl, 9 mM MgCl 2 , 0.9% Nonidet P-40, 180 mM ZnCl 2 , 0.9 mM dithiothreitol) and were allowed to incubate at room temperature for 20 min. For competition experiments, a 200-fold excess (unless otherwise noted) of the indicated unlabeled double-stranded oligonucleotides was added prior to the addition of radiolabeled DNA probe. In some cases, the reaction mixtures were incubated at 4°C with either 5 l of specific anti-Sp1 polyclonal antibodies (a generous gift from the R. Tjian laboratory, University of California, Berkeley) or an equal volume (equivalent to 100 ng) of anti-p110 PI 3-kinase polyclonal antibody (Santa Cruz Biotechnology, Inc.) for 2 h before the addition of radiolabeled probes. The protein-DNA complexes were resolved on a 4% native polyacrylamide gel (19:1 acrylamide/bisacrylamide) in 1ϫ gel shift running buffer (25 mM Tris base, 190 mM glycine, 1.3 mM EDTA) at 4°C, 170 V. The gels were prerun for 0.5-1 h at 4°C. The protein-DNA complexes in the dried gels were visualized by autoradiography using Amersham Pharmacia Biotech Hyperfilm.
Western Blot Analysis-Soluble whole cell extracts (20 -50 g of protein) were electrophoretically resolved on 7.8% SDS-polyacrylamide gel electrophoresis, and the proteins were transferred to Nytran Plus membranes (Schleicher & Schuell). The membrane was probed with a 1:2500 dilution of affinity-purified anti-Sgk antibody in 50 mM Tris, pH 8.0, 150 mM NaCl, 0.05% Tween 20 with 1% nonfat dry milk. The goat anti-rabbit IgG horseradish peroxidase-conjugated antibody (Bio-Rad) secondary antibody was used at a dilution of 1:10,000. The Western blot was developed by using the Renaissance developing kit (NEN Life Science Products) and exposed to x-ray film.
Immunofluorescent Detection-Cells were plated onto eight-well LabTek chamber slides and then treated as described (30). The next day, the cells were treated with 0.3 M sorbitol for approximately 4 -6 h. Following washes with 1ϫ PBS, cells were fixed using 4% paraformaldehyde in PBS for 30 min at room temperature. They were then permeabilized in 0.1% Triton X-100, 0.1% sodium citrate for 2 min on ice. Affinity-purified anti-Sgk antibody diluted 1:150 in PBS was added to samples and allowed to incubate for 1 h at room temperature or overnight at 4°C. Goat anti-rabbit fluorescein isothiocyanate-conjugated secondary antibody was added at a dilution of 1:150 in PBS and allowed to incubate for 1 h at room temperature. The cells were then washed with 1ϫ PBS, and coverslips were mounted using Antifade (Molecular Probes, Inc., Eugene, OR). Results were visualized on a Nikon Optiphot fluorescence microscope. Images were captured using Adobe Photoshop 3.0.5 (Adobe Systems, Inc., Mountain View, CA) and a Sony DKC-5000 digital camera. Nonspecific fluorescence was determined by incubation with the secondary antibody alone and shown to be negligible.
Immunoprecipitation and in Vitro Kinase Assay of Sgk-HEK 293 cells were transfected as described (31) and then treated with either serum-free media or 0.3 M sorbitol for 5 min. Cells were placed on ice and extracted with lysis buffer containing 50 mM Tris-HCl, 25 mM NaF, 120 mM NaCl, 40 mM ␤-glycerol phosphate, 1% Nonidet P-40, 0.1 mM sodium orthovanadate, 1 mM benzamidine, 1 mM phenylmethylsulfonyl fluoride, and 2 nM microcystin (Calbiochem). Lysates were centrifuged for 15 min at 12,000 ϫ g, and the HA-Sgk protein was immunoprecipitated from 400 g of cell-free extracts with the anti-HA epitope 12CA5 monoclonal antibody coupled to protein A-Sepharose (Amersham Pharmacia Biotech). The immune complexes were washed once with lysis buffer containing 0.5 M NaCl and then with lysis buffer and finally with kinase buffer (50 mM Tris-HCl, 0.1% ␤-mercapoethanol). In vitro kinase assays were performed for 60 min at 30°C in a reaction volume of 50 l containing 20 l of immunoprecipitate in kinase buffer, 0.1 mM Sgktide peptide substrate (KKRNRRLSVA), 10 mM MgCl 2 , 50 M ATP, 1 mM protein kinase A inhibitor (Calbiochem), and 5 Ci of [␣-32 P]ATP (NEN Life Science Products). The kinase reaction was terminated with 0.55 mM EDTA. The 20 l of the reaction was spotted onto P-81 cation exchanger chromatography paper (Whatman), which was then washed four times in 1% phosphoric acid and dried with acetone. The amount of 32 P-labeled Sgktide was quantitated by scintillation counting. Kinase assays performed using endogenous Sgk protein were performed as above except for the following modifications. NMuMg cells treated with either serum-free media or 0.3 M sorbitol for 8 h were harvested and then lysed as described above. Lysates were centrifuged for 15 min at 15,000 ϫ g and precleared with rabbit IgG and protein A-Sepharose beads (Amersham Pharmacia Biotech) for 40 min. The Sgk protein was immunoprecipitated from precleared supernatants using affinity-purified anti-Sgk antibodies and protein A beads for 2 h at 4°C. A control set immunoprecipitations employed nonimmune serum. The Sgk-specific transphosphorylation was determined by subtracting the filter-bound radioactivity observed with the nonimmune antibodies from that observed with the Sgk-specific antibodies.

Stimulation of sgk Transcript and Protein Levels in Response
to Hyperosmotic Stress-To initially test whether sgk gene expression is targeted by the mammalian hyperosmotic stress pathway, the levels of Sgk protein and mRNA were determined in NMuMg (normal murine mammary gland) cells at various times after treatment with 0.3 M of the organic osmolyte sorbitol to induce a hyperosmotic stress. Western blot analysis of whole of cell extracts using affinity-purified anti-Sgk polyclonal antibodies (30) revealed that after an approximately 4-h time lag, sorbitol treatment induced a significant elevation in Sgk protein levels (Fig. 1A). A high level of Sgk protein was still maintained 24 h after the hyperosmotic stress. Northern blot analysis of total cellular mRNA of the same time course using a ␣-32 P-labeled murine sgk cDNA probe demonstrated a striking sorbitol stimulation in expression of the 2.2-kb sgk transcript ( Fig. 1B). Consistent with the protein data, sgk transcripts were not detected until 4 h after sorbitol treatment, and sgk mRNA levels continued to accumulate 24 h following treatment. In the absence of sorbitol, the level of Sgk protein and mRNA remained at a low basal level through out the entire time course.
To examine whether the sorbitol stimulation of sgk gene expression is a direct response or requires de novo protein synthesis, NMuMg cells were pretreated for 1 h with or without cycloheximide (10 g/ml), a protein synthesis inhibitor, and then each set of cells was incubated in the presence or absence of 0.3 M sorbitol for 24 h. In the cycloheximide-treated cells, protein synthesis was inhibited by greater than 95% without any significant effects on mRNA synthesis (data not shown). Northern blot analysis revealed that pretreatment of the cells with cycloheximide had no effect on the inducibility of the sgk transcripts in response to sorbitol (Fig. 2). Because the hyperosmotic stress induction of sgk mRNA occurred in the absence of de novo protein synthesis, we suggest that the regulation of sgk gene expression is mediated by one or more preexisting cellular components that are likely to be involved in the transcriptional control of this gene. As we have previously reported (29), exposure to cycloheximide stabilized sgk transcript turnover, which caused both the sorbitol-treated and -untreated cells to produce a higher level of sgk mRNA (Fig. 2). Our results suggest that a hyperosmotic stress-induced signaling pathway stimulates sgk gene transcription, which accounts for the increased production of Sgk protein after sorbitol treatment.
Identification of a Hyperosmotic Stress-responsive Region of the sgk Promoter-To functionally dissect the region within the sgk gene promoter that is responsible for the hyperosmotic stress regulation of sgk expression, NMuMg cells were transfected with a series of sgk-CAT reporter genes containing a series of 5Ј-progressive deletions starting at Ϫ1350 bp upstream of the sgk gene and terminating at ϩ55 bp downstream of the transcriptional initiation site of the sgk gene. Cells were then treated with or without 0.3 M sorbitol, and cell lysates were assayed for CAT activity 24 h after the hyperosmotic stress. As shown in Fig. 3, sorbitol stimulated transcriptional activity of sgk-CAT constructs containing the three largest sgk promoter regions with deletions ending at Ϫ1350 bp, Ϫ236 bp, and Ϫ78 bp. Maximum stimulation of sgk-CAT activity by hyperosmotic stress (5-8-fold) was observed with these constructs. Sorbitol inducibility of the sgk promoter was lost in deletions beyond Ϫ35 bp, indicating that the promoter fragment between nucleotides Ϫ78 and Ϫ35 bp of the upstream region of the sgk gene is a hyperosmolar stress-responsive region. In additional experiments, this hyperosmotic regulated region was further narrowed to between Ϫ67 and Ϫ35 base pairs (data not shown, and see below). These results suggest that the region between Ϫ67 and Ϫ35 of the sgk promoter contains transcription factor binding sites involved in the stimulation of sgk gene transcription in response to hyperosmotic stress.
Functional Analysis of the Hyperosmolar Stress-responsive Region of the sgk Promoter-To test whether the hyperosmolar stress-responsive region defined by the deletion analysis is sufficient to mediate the sorbitol responsiveness of the sgk gene, an oligonucleotide corresponding to the wild type Ϫ67 to Ϫ35 bp fragment was cloned into pCAT-Basic that already contained the Ϫ34 to ϩ55 region of the sgk promoter (38) and linked to the bacterial CAT gene. NMuMg cells were transiently transfected with Ϫ67/Ϫ35 wild type CAT reporter plasmid, and CAT activity was monitored in cell extracts isolated from cells treated with or without sorbitol. As shown in Fig. 4, sorbitol induced CAT expression by 3.7-fold compared with the unstressed mammary cells. Reporter gene activity in cells transfected with the minimal promoter tkCAT alone was low and unaffected by sorbitol treatment (data not shown). This result establishes that the hyperosmolar stress responsive region of the sgk promoter in transfected mammary epithelial cells is located between Ϫ67 and Ϫ35 bp upstream of the RNA start site.
Analysis of the hyperosmotic stress-regulated region of the sgk promoter using transcription factor consensus binding site algorithms such as TFSEARCH and MatInspector revealed the presence of a highly conserved GC-rich region (43,44). Therefore, to further characterize the promoter region necessary for the sorbitol induction of sgk gene expression, five mutants of this region were synthesized in which each mutant contained a different set of six base pairs that were mutated to the same sequence, TTCGAA, which is a BstBI restriction enzyme site (shown in Fig. 4). Annealed oligonucleotides (sense plus antisense) for the indicated five mutants (M1-M5) of the Ϫ67/Ϫ35 region of the sgk promoter were cloned into the same pCAT-Basic vector used to generate the wild type CAT reporter plasmid. The CAT reporter activities were monitored in transiently transfected NMuMg cells treated with or without sorbitol. As shown in Fig. 4, the M1, M2, and M5 mutants were fully inducible by hyperosmolar stress at a level comparable with the wild type sgk promoter fragment. In contrast, the M3 and M4 were unable to respond to sorbitol treatment. The M3 and M4 mutations overlapped with and ablated the GC-rich region of the sgk promoter, suggesting that the corresponding transcription factor is a target of the hyperosmotic stress response in mammary epithelial cells.
Electrophoretic Mobility Shift Assay Analysis of the Transcriptionally Responsive Region of the sgk Promoter-A competitive gel shift analysis was employed to characterize the DNA-binding protein complexes associated with the hyperosmolar stress responsive region of the sgk promoter. The doublestranded wild type Ϫ67/Ϫ35 oligonucleotide was radiolabeled and incubated with nuclear extracts that were prepared from 4-h sorbitol-treated or control NMuMg cells along with an excess of one of the mutant oligonucleotides (M1-M5). The gel shift pattern of the protein-DNA complexes was visualized by electrophoretic fractionation in a 6% SDS-polyacrylamide gel. As shown in Fig. 5, without any competing oligonucleotides, the Ϫ67/Ϫ35 region of the sgk promoter forms the same four major protein-DNA complexes in the presence of cell extracts from untreated control or sorbitol treated cells (defined as complexes A, B, C, and D). Incubation with a 200-fold excess of unlabeled M1, M2, or M5 mutant oligonucleotides, which are fully sorbitol-responsive, quantitatively competed with the wild type oligonucleotide for formation of each of the protein-DNA complexes using extracts from either the control or hyperosmolar stressed cells (Fig. 5). A similar complete competition was observed with the wild type oligonucleotide (data not shown). In contrast, mutant oligonucleotides M3 and M4, which effectively ablate the GC-rich sequence and are not sorbitol-responsive, were ineffective as competitors in this gel shift assay (Fig.  5). These results suggest that the GC-rich sequence located at Ϫ55 to Ϫ46 bp mediates the hyperosmotic stress regulation of the sgk promoter through the function of the corresponding protein-DNA complex in this region of the promoter.
Identification of the Sp1 Transcription Factor as a Component of the Protein-DNA Complex That Binds to the Hyperosmolar Stress-regulated Region of the sgk Promoter-It is well FIG. 1. Effect of 0.3 M sorbitol treatment on Sgk protein and transcript expression. NMuMg mammary epithelial cells were treated with or without 0.3 M sorbitol in serum-free media, and at the indicated times the cells were harvested. A, Western blots of electrophoretically fractionated total cell extracts were probed for the production of Sgk protein using affinity-purified Sgk-specific primary antibodies. B, total RNA was isolated and separated on a 1.1% agarose gel, transferred to a nitrocellulose blot, and hybridized with a radiolabeled murine sgk cDNA probe as described under "Experimental Procedures." As a loading control, 18 S ribosomal RNA stained with methylene blue is shown above the sgk transcript band. established that the Sp1 transcription factor binds to a GC-rich DNA site that is identical to the GCCCCGCCCC sequence located between Ϫ55 and Ϫ46 of the sgk promoter (42,45). The competitive gel shift assays were used to initially test whether the hyperosmolar stress-responsive region binds to a transcription factor with Sp1 specificity. Nuclear extracts from control and sorbitol-treated cells were preincubated with a 50-, 100-, or 200-fold excess of unlabeled competitor oligonucleotides corresponding to a consensus Sp1 site. As a control, a mutant Sp1 oligonucleotide, in which the middle three base pairs of the Sp1 consensus have been changed in a manner that abrogates Sp1 (42), was used as the competitor DNA. Each of these mixtures, as well as a reaction without any added competitor, were incubated with the radiolabeled wild type oligonucleotides encoding the Ϫ67 to Ϫ35 region of the sgk promoter, and the protein-DNA complexes were visualized after electrophoretic fractionation. As shown in Fig. 6A, increasing amounts of the consensus Sp1 oligonucleotide strongly competed for formation of each of the four major protein-DNA complexes, whereas the mutant Sp1 oligonucleotide was unable to compete. In contrast, an unlabeled oligonucleotide containing the consensus 5Ј-TCGC-CCCCTC-3Ј sequence corresponding to the EGR1 transcription factor, which also binds to a GC-rich DNA site, failed to compete in the same assay (data not shown).
In a complementary approach, the presence of Sp1 in the gel-shifted complexes was tested by assessing the ability of anti-Sp1 polyclonal antibodies to alter the wild type gel shift complex. Nuclear extracts isolated from control or sorbitoltreated cells were incubated for 2 h with either Sp1 antibodies (SP1Ab) or control antibodies directed against the phosphatidylinositol 3-kinase p110 subunit (p110Ab). The resulting formation of protein-DNA complexes on the wild type Sgk Ϫ67/ Ϫ35 double-stranded oligonucleotide was visualized after electrophoretic fractionation. As also shown in Fig. 6B, the anti-Sp1 antibodies caused a specific loss of band B with a corresponding increase in a new lower molecular weight protein-DNA complex (indicated by the asterisk), whereas no reproducible changes were detected in the gel mobility of bands A, C, and D. Conceivably, the Sp1 antibodies either sequestered the available Sp1 protein in the extracts, thereby preventing it from binding to the sgk wild type gel shift probe to form band B, or prevented the binding of other transcription factors that are normally tethered to the Sp1-containing gel shift complex. By comparison, incubation with the control anti-p110 antibody had no effect on the gel-shifted protein complexes. Incubation of the nuclear extracts with higher concentrations of the anti-Sp1 antibody did result in a more pronounced loss of band B, although there was no noticeable effect on the presence of any For the mutagenic fine mapping of the hyperosmotic stress-regulated region, a series of six base pair substitutions of TTCGAA were sequentially placed throughout the Ϫ67 to Ϫ35 bp region of the sgk promoter. These five mutant oligonucleotides are designated M1-M5. CAT reporter plasmids containing the indicated wild type and mutant oligonucleotides were transfected into NMuMg cells, incubated for 24 h with (ϩ) or without (Ϫ) 0.3 M sorbitol, and the resulting CAT activity and determination of CATspecific activity was assayed as described in the legend to Fig. 3. The numbers display the -fold induction of CAT-specific activity observed with each reporter plasmid. Each set of assays was performed in triplicate, and the reported values are the mean and S.D. from three independent sets of experiments. of the other protein-DNA complexes (data not shown). These results, along with the functional studies described above, implicate Sp1 as a target or a member of a multiprotein complex that is involved in hyperosmotic stress regulation of the sgk promoter.
Hyperosmotic Stress Regulation of sgk Transcript and Protein Expression Is Mediated by the p38 MAPK Pathway-It is well documented that p38 MAPK is phosphorylated and activated by the dual specificity MKK3/6 kinases at a TGY motif in response to various forms of stress, including hyperosmotic stress (6,16,46). To test whether p38 MAPK is activated in NMuMg mammary cells in response to sorbitol, a Western blot of extracts from cells treated at various times with or without 0.3 M sorbitol was probed with anti-phospho-p38 MAPK antibodies. The phosphorylated form of p38 MAPK was detected in the mammary epithelial cells between 1 and 2 h after hyperosmotic stress and then decreased to near basal levels by 4 h poststress (Fig. 7A). Interestingly, p38 MAPK phosphorylation occurs before the induction of Sgk protein expression, suggesting a causal relationship between these cellular processes. To directly determine whether the hyperosmotic stress stimulation of Sgk protein expression requires p38 MAPK activity, NMuMg cells were treated with the p38 MAPK pharmacological inhibitors SB203580 (20 M) or SB202190 (10 M) for 30 min prior to the addition of either serum-free media (untreated control) or 0.3 M sorbitol for 6 h. Western blot analysis of Sgk protein levels revealed a striking reduction in the amount of Sgk protein induced in hyperosmolar stressed cells treated with either p38 MAPK inhibitor (Fig. 7B).
The p38 MAPK pathway targets and phosphorylates several transcription factors in response to hyperosmotic stress (16), and the attenuation of Sgk protein levels in the presence of p38 pharmacological inhibitors suggested a functional link between p38 MAPK function and the stimulation of sgk promoter activity. To determine whether the p38 MAPK cascade is involved in the hyperosmotic stress stimulation of the sgk promoter, NMuMg cells transfected with the Ϫ1.3 sgk-CAT reporter plasmid were pretreated with or without the p38 MAPK inhibitor SB202190 and then incubated in the presence or absence 0.3 M sorbitol to induce the hyperosmolar stress. In the absence of SB202190, the Ϫ1.3 sgk-CAT construct was induced on average 4-fold in response to sorbitol treatment (Fig. 7C), and in cells pretreated with SB202190 the hyperosmolar stress induction of sgk promoter was abrogated. There are several p38 MAPK gene family members, and based on the selectivity of the SB203580 and SB202190 pharmacological inhibitors, our results suggest that the p38-␣-or p38-␤-mediated signaling pathway (47) Fig. 4. The radiolabeled wild type Ϫ67/Ϫ35 probe (see Fig. 4) was added to each reaction mixture for 20 min, and the resulting protein-DNA complexes were electrophoretically resolved in low ionic strength native 6% polyacrylamide gels. The arrows show the migration of the four major protein-DNA complexes (A, B, C, and D) that were observed under these conditions.
FIG. 6. Evidence for the presence of Sp1 activity in the gel-shifted complex formed with the hyperosmotic stress-regulated region of the sgk promoter. Cytoplasmic extracts (7 g) were prepared from NMuMg mammary cells that had been treated for 4 h with or without 0.3 M sorbitol in serum-free media and then preincubated for 20 min on ice with either 50-, 100-, or 200-fold excess of unlabeled Sp1 consensus oligonucleotide (SP1) or of a mutant Sp1 oligonucleotide (MutSP1). The radiolabeled wild type Ϫ67/Ϫ35 probe (see Fig. 4) was added to each reaction mixture for 20 min, and the resulting protein-DNA complexes were electrophoretically resolved in low ionic strength native 6% polyacrylamide gels. B, isolated cell extracts were incubated for 2 h on ice with either polyclonal anti-Sp1 antibodies or as a negative control with polyclonal anti-p110 antibodies. The radiolabeled wild type Ϫ67/Ϫ35 probe (see Fig. 4) was added to each reaction mixture for 20 min, and the resulting protein-DNA complexes were electrophoretically resolved in low ionic strength native 6% polyacrylamide gels. The arrows show the migration of the four major protein-DNA complexes (A-D), and the asterisks show the migration on the new protein-DNA complex that is formed only in the presence of anti-Sp1 antibodies. tributes to the hyperosmotic stress stimulation of sgk promoter activity and protein production.
One of the upstream kinases that activates p38 MAPK in response to cellular stress is MKK3 (16,46). As a complementary approach to demonstrate the involvement of the p38 MAPK pathway in the hyperosmolar stress stimulation of the sgk promoter, cells were cotransfected with the Ϫ236 sgk-CAT reporter construct and with expression plasmids for either the wild type MKK3 kinase (MKK3) or the kinase-dead form of MKK3 (MKK3AL) that has been shown to act as a dominant negative inhibitor of the p38 MAPK pathway (48). As shown in Fig. 8, co-transfection with the wild type MKK3 leads to a high level of sorbitol-stimulated reporter activity, whereas the cotransfection of the dominant negative acting MKK3AL plasmid abolished the sorbitol induction of sgk promoter activity. These results show that the hyperosmotic stress stimulation of sgk promoter activity requires the MKK3-mediated activation of p38 MAPK kinase. Because treatment with either of the p38 MAPK pharmacological inhibitors fails to completely dampen the hyperosmotic stress stimulation of sgk promoter activity, MKK3 may have one or more as yet unidentified targets in addition to the p38-␣ or p38-␤ MAPK isoforms that play a role in the sgk transcriptional response.
Hyperosmotic Stress Induces a Cytoplasmically Localized and Enzymatically Active Sgk Protein-We have previously established that the nuclear/cytoplasmic subcellular distribution of Sgk can be regulated in a stimulus-dependent manner (30). For example, in serum-treated cells Sgk shuttles between the nucleus and cytoplasm in synchrony with the cell cycle, whereas in glucocorticoid-treated cells Sgk is exclusively localized to the cytoplasmic compartment. Therefore, to characterize the effects of hyperosmotic stress on the subcellular distribution of Sgk, NMuMg cells were grown to confluency and treated with 0.3 M sorbitol for 4 h, and the fixed cells were permeabilized and stained with anti-Sgk polyclonal antibody followed by fluorescein isothiocyanate-conjugated goat antirabbit secondary antibody. Indirect immunofluorescence revealed that hyperosmotic stress stimulates the production of a cytoplasmic localized Sgk (Fig. 9).
We have recently established that Sgk protein kinase activity is activated in response to serum growth factors through a phosphatidylinositol 3-kinase-mediated pathway (31). Furthermore, other groups have shown that sorbitol inhibits other PI 3-kinase downstream targets, such as Akt/PKB and p70 S6 kinase (49 -51). Interestingly, this inhibition is not due to the inhibition of PI 3-kinase or PDK1 activity but instead through the activity of the PP2A phosphatase. Therefore, to determine whether the sorbitol induced the slower migrating phosphorylated form of Sgk as a result of signaling through PI 3-kinase, cells were pretreated with or without the pharmacological inhibitor for PI 3-kinase, LY294002. This drug renders the PI 3-kinase molecule inactive by preventing ATP binding (52). As shown in Fig. 10B, in the absence of LY294002, a Sgk protein doublet (49 and 51 kDa) is induced in response to hyperosmotic stress. Our previous work has demonstrated that these two protein species represent hyperphosphorylated and hypophosphorylated forms of Sgk (30). We also previously documented that an inhibition of PI 3-kinase activity prevented the appearance of the hyperphosphorylated form of Sgk (31). Consistent with these results, LY294002 ablated production of the slower migrating (and hyperphosphorylated) form of Sgk (Fig. 10B), suggesting that the Sgk protein, which is produced after sorbitol treatment, is targeted by the PI 3-kinase signaling pathway.
To biochemically assess whether hyperosmotic stress had any effect on endogenous Sgk protein kinase activity, NMuMg cells were treated with or without sorbitol for 8 h, and the isolated cell extracts were incubated with affinity-purified anti-Sgk antibody or nonimmune antibodies as a negative control. As a positive control for this assay, Sgk was immunoprecipitated from serum-treated and -untreated cells. The Sgktide peptide (KKRNRRLSVA) was utilized as a substrate in an in vitro transphosphorylation assay, and the Sgk-specific activity was determined by quantitating the level of 32 P-labeled Sgktide in the anti-Sgk immunoprecipitated compared with those from nonimmune antibodies. As shown in Fig. 10A, the endogenous Sgk immunoprecipitated expressed in hyperosmotically stressed cells is an active kinase with a transphosphorylation activity comparable with that observed in serum-treated cells.
As a complementary approach, the effects of hyperosmotic stress on Sgk enzymatic activity in the absence of the transcriptional stimulation were examined by expression of an exogenous sgk gene driven by the constitutive RSV mammalian expression vector. To assess the activity of exogenous wild type

FIG. 7. Inhibitors of the p38 MAPK suppresses the hyperosmotic stress stimulation of Sgk protein and promoter activity.
A, NMuMg cells were treated with (ϩ) or without (Ϫ) 0.3 M sorbitol in serum-free media and at the indicated time points. Cell extracts were electrophoretically fractionated in SDS-polyacrylamide gels. Western blots were probed with phospho-p38 MAPK-specific antibodies to analyze the production of active p38 MAPK. B, cells were treated with the p38 MAPK pharmacological inhibitors SB203580 (20 M) or SB202190 (10 M) for 30 min, and then the cells were treated with (ϩ) or without (Ϫ) 0.3 M sorbitol in serum-free media for 6 h. Total cell extracts were electrophoretically fractionated in SDS-polyacrylamide gels and Western blots probed for the production of Sgk as described under "Experimental Procedures." C, cells were transfected with the Ϫ1.3 sgk-CAT reporter plasmid, as described in the legend to Fig. 3. 24 h posttransfection, the cells were pretreated for 30 min with or without SB202190 (10 M). The cells were then treated with or without 0.3 M sorbitol in serum-free media for 24 h, harvested, and lysed, and CAT activity was assayed by quantification of the conversion of [ 3 H]acetyl-CoA into [ 3 H]acetyl-chloramphenicol by the two-phase fluor diffusion assay described under "Experimental Procedures." The resulting CAT activities were normalized to protein levels to determine the CATspecific activity, and the -fold induction was calculated as a ratio of the CAT-specific activity observed in sorbitol-treated cells to that observed in control cells. Three independent assays (data points) were performed in triplicate, and the bar graphs show the mean from these experiments.
Sgk protein, human embryo kidney 293 cells (HEK 293) were transfected with the pCDNA3 HA-sgk expression vector and then treated with either isosmotic or hyperosmotic media. The immunoprecipitated HA-Sgk protein, using antibodies directed against the HA epitope tag, was assayed for transphosphorylation activity using the Sgktide substrate. As shown in Fig. 11, hyperosmotic stress stimulated the kinase activity of exogenously expressed Sgk protein approximately 3-fold over isosmotic treatment. After the peak production of active Sgk within the first 5-15 min after sorbitol treatment, the level of Sgk transphosphorylation activity returned to basal levels during the next 60 min. Although activation of Sgk is transient, this result suggests that hyperosmotic stress regulates Sgk expression at the transcriptional level and also posttranslationally regulates Sgk enzymatic activity.

DISCUSSION
The evolutionarily conserved ability of an organism to adapt to extreme changes in nutrient levels, temperature, or osmolarity is crucial to the survival of unicellular prokaryotes as well as the eukaryotic cells that constitute mammalian tissues. Several signaling cascades have been shown to play a necessary role in the detection and response to environmental stress. One such evolutionarily conserved pathway employs the activation of the p38 MAPK/HOG1 family of protein kinases (14), which can be triggered by osmotic imbalances, UV light radiation, heat shock, DNA-damaging agents, FAS-induced cell death, and exposure to the protein synthesis inhibitor anisomycin (46,48,53). Several transcription factors and protein kinases have been identified as cellular targets of p38 MAPK in mammalian cells; however, relatively little is known about their respective gene targets and protein substrates. We and others have shown that an important cellular feature of sgk is its acute transcriptional control in rodent and human cell lines (28, 29, 32, 33, 38, 54 -57), and our study shows for the first time that the transcriptional regulation of sgk can be directly linked to a specific stress signaling pathway involving p38 MAPK in mammalian cells.
As shown in Fig. 12, we propose that exposure of mouse NMuMg mammary epithelial cells to hyperosmotic medium activates the p38 MAPK cascade, which in turn stimulates sgk transcription by targeting a GC-rich hyperosmotic stress-regulated element in the sgk promoter that is recognized by the Sp1 transcription factor. Activation of this pathway induces an increase in the production of an active Sgk protein kinase that maintains its dependence on the PI 3-kinase pathway for its phosphorylation and activity. Several lines of evidence demonstrated a role for p38 MAPK in the stimulation of sgk transcription. The robust stimulation of sgk transcription and protein production can be disrupted by treatment with p38 MAPKspecific pharmacological inhibitors or by exogenous expression of dominant negative MKK3, the immediate upstream regulator of p38 MAPK. Moreover, ectopic expression of wild type MKK3 further stimulated the magnitude of the sorbitol-induced sgk promoter activity. Our results also suggest that the activated p38 MAPK cascade either directly or indirectly targets a Sp1 transcription factor protein complex on the sgk promoter to stimulate sgk transcription (Fig. 12). Sp1 is a Cys 2 His 2 zinc finger-binding protein that has been shown to bind specifically to the GC-rich GCCCCGCCCC sequence (42, 45) that we defined as the hyperosmotic stress-regulated ele- The subcellular distribution of Sgk was examined by indirect immunofluorescence microscopy using affinitypurified rabbit polyclonal antibodies to Sgk followed by fluorescein isothiocyanate-conjugated goat anti-rabbit secondary antibodies. No immunofluorescence staining is observed when preimmune antiserum is used as the primary antibodies (data not shown). ment in the sgk promoter. Our studies are generally consistent with previous observations in the yeast S. cerevisiae in which the Msn family of transcription factors are targeted by the yeast p38 MAPK homologue HOG1 and bind to a CCCCT DNA sequence that is similar to the Sp1 binding core sequence (7,8,10). In earlier studies, Sp1 was viewed as a component of the basal transcriptional machinery. However, recent studies have shown that Sp1 transactivation activity is also subject to regulation by several different extracellular cues including oxidative and shear stress (58,59). Our results now implicate the Sp1 transcription factor as a component in the hyperosmotic stress response that leads to the stimulated expression of sgk.
A deletion analysis initially identified a hyperosmotic stressresponsive region between Ϫ67 and Ϫ35 bp of the sgk promoter that accounts for the striking sorbitol stimulation of sgk gene expression. Fine mapping of this 32-bp region defined a GC-CCCGCCCC sequence at Ϫ54 to Ϫ44 bp as being necessary for the hyperosmotic shock induction. Competitive electrophoretic mobility shift assays and mutational analyses demonstrated the presence of Sp1 in the transcription factor complex that interacts with the hyperosmotic stress-responsive region and thereby implicates the Sp1 transcriptional regulator as a necessary factor for this response. For example, inclusion of Sp1specific antibodies in the gel shift reaction caused a specific loss of one of the predominant protein-DNA complex bands with a corresponding increase in a new lower molecular weight protein-DNA complex. Also, the consensus Sp1 oligonucleotide strongly competed for formation of each of the four major protein-DNA complexes observed with the Ϫ67 to Ϫ35 hyperosmotic stress-regulated region of the sgk promoter. In contrast, a mutant Sp1 oligonucleotide, which is incapable of binding Sp1, and a wild type consensus oligonucleotide corresponding to the EGR1 transcription factor, which also binds to a GC-rich DNA site, were unable to compete in this assay. It is likely that the Sp1 transcription factor functions in the hyperosmotic stress response through an interaction with another unidentified nuclear factor, because transfection of a reporter plasmid containing the consensus Sp1 binding site with or without an expression plasmid for Sp1 (60) failed to be sorbitol-responsive (data not shown). Consistent with this concept, we observed multiple protein-DNA complexes in an electrophoretic mobility gel shift assay using an oligonucleotide corresponding to the Ϫ67 to Ϫ35 hyperosmotic stress-responsive region. We are currently examining whether one or more of the protein factors in these DNA-protein complexes may be transcriptional targets of the hyperosmotic stress cascade that interact with Sp1.
The four major protein-DNA complexes that form on the hyperosmotic stress-regulated element in the sgk promoter remained unaffected by sorbitol treatment. There are several precedents for promoters being activated through a p38 MAPK cascade that do not show obvious changes in protein-DNA complexes. For example, the lipopolysaccharide induction of a NF-B consensus binding site reporter construct was inhibited by the p38 MAPK pharmacological inhibitor SB203580 and by co-expression of a dominant negative p38 MAPK under conditions that did not change the observed gel-shifted DNA-protein complexes (61). We speculate that sgk transcription may be stimulated by the recruitment of specific coactivators or altered interactions with components of the basal transcriptional machinery that can potentially be phosphorylated by p38 MAPK in response to hyperosmotic stress. For example, the TFIID component of the basal transcription complex has been shown FIG. 10. Effect of hyperosmotic stress on the phosphorylation state and enzymatic activity of endogenous Sgk protein. A, affinity-purified anti-Sgk antibodies (Sgk) or nonimmune antibodies (NI) were added to total cell extracts from cells either treated with (ϩ) or without (Ϫ) 0.3 M sorbitol for 8 h. As a positive control for Sgk enzymatic activity, Sgk was immunoprecipitated from 48-h serum-starved cells that had been incubated with (ϩ) or without (Ϫ) 10% serum for 24 h. Transphosphorylation activity in the immunoprecipitated protein was monitored by quantifying the formation of 32 P-labeled Sgktide peptide in each reaction. Sgk-specific transphosphorylation activity was determined by subtracting the filter-bound radioactivity observed with the nonimmune antibodies from that observed with the anti-Sgk antibodies. B, NMuMg cells were pretreated with either 50 mM LY294002 or Me 2 SO carrier for 16 h and then incubated with (ϩ) or without (Ϫ) 0.3 M sorbitol in serum-free media. After 8 h, cells were harvested, and Western blots of electrophoretically fractionated total cell extracts were probed for the expression of Sgk protein.
FIG. 11. Effect of hyperosmotic stress on the enzymatic activity of exogenous Sgk protein. HEK 293 cells overexpressing HA epitope-tagged Sgk (HA-Sgk) were treated with either isosmotic (Iso) or hyperosmotic (0.3 M sorbitol) media for the indicated times. HA-Sgk was immunoprecipitated from each cell extract using anti-HA-specific antibodies and assayed for kinase activity using Sgktide as the substrate. The anti-HA-Sgk immunoblot analysis is shown in the lower panel.
to be phosphorylated by p38 MAPK and also to interact with Sp1 in other cell systems (61,62). Conceivably, other cellular components that are known to be phosphorylated by p38 MAPK, such as transcription factors and protein kinases (16 -18, 20 -22, 25, 63, 64), could potentially interact with Sp1 in a p38 MAPK-dependent manner and play a role in the hyperosmotic stress response.
We propose that the Sgk mediates certain cellular responses to osmolar imbalances, because after hyperosmotic stress of mammary epithelial cells, Sgk protein levels are robustly induced and maintained at a high level throughout the time course of sorbitol treatment. As predicted by the promoter studies, pretreatment of cells with the p38 MAPK inhibitors SB203580 and SB202190 resulted in the significant attenuation of Sgk protein levels. Although one other study has reported that anisosmotic changes regulate sgk transcript levels (34), our study is the first to define the sgk promoter elements that mediate the hyperosmotic stress stimulation of transcripts, to uncover a connection between Sgk and the p38 MAPK stress signaling cascade, and to demonstrate an increase in the production of endogenous Sgk protein and Sgkspecific serine/threonine protein kinase activity. In response to hyperosmotic stress, the Sgk protein is localized throughout the cytoplasm that presumably includes the cytoplasmic surface of the plasma membrane. One potential role of Sgk in mediating the cellular hyperosmotic stress response may be to regulate cell volume through activation of the epithelial sodium channel (55,56). Our results suggest that the enzymatic activity of Sgk is important for this response, and we are attempting to define the membrane interactions between endogenous Sgk and the epithelial sodium channel that are necessary for channel activity that probably account for the increased channel activity that is observed in response to hyperosmotic stress (65).
Following treatment of cells with serum and/or growth factors, PI 3-kinase activity is induced, which initiates a PDK-1 signaling pathway (66). These events are responsible for the activation of the downstream kinases Akt/PKB, p70 S6 kinase, and Sgk (31, 67-70). A well documented consequence of the activation of Akt/PKB is the phosphorylation of the proapoptotic protein Bad, which regulates the cell survival response (71)(72)(73). Both Sgk and Akt/PKB are downstream targets of the PI 3-kinase-dependent activation of PDK1. Treatment of 3T3L1 preadipocytes with sorbitol has been shown to inhibit Akt/PKB as well as p70 S6 kinase activity (49 -51). It has been demonstrated that PI 3-kinase and PDK1 are active under this hyperosmolar condition, and consequently the decrease in the cellular pool of phospho-Akt/PKB is attributable to the PP2A phosphatase (49,50). We observed that sorbitol induced a phosphorylated and kinase-active form of endogenous Sgk that is sensitive to the LY294002 PI 3-kinase inhibitor. It is therefore possible that in the mammary epithelial cells used in our study, the activity of the PP2A phosphatase remains low enough to maintain the activity of endogenous Sgk. In this regard, ectopic expression of sgk in the human embryo kidney 293 cells (HEK 293) only transiently produced an active Sgk enzyme, suggesting that the constitutive PP2A phosphatase activity plays a counter regulatory role on Sgk in these cells. We have previously shown that a lack of phosphorylation of Sgk at both the critical Thr 256 and Ser 422 residues results in an inactive Sgk kinase (31). Therefore, as a alternative explanation, it is possible that in the presence of sorbitol, Sgk is only phosphorylated on one of these residues and as a result cannot sustain its activity in the HEK 293 cells.
Several groups have reported that perturbations of the PI 3-kinase signaling pathway can affect insulin-dependent glucose uptake by the GLUT4 transporter and other insulin responses (74 -78). Conceivably, Sgk, which is expressed in the kidney (55,56) and expressed at high levels in the kidney glomeruli and distal tubules, 2 may be involved in the hyperosmotic response to high plasma glucose levels in which p38 MAPK activity has been shown to be induced and Akt/PKB activity inhibited (49, 50, 79 -81). Moreover, one other report suggests that Sgk may also play a role in the nephropathy that is associated with the diabetic disease state (82). Further stud-2 E. Lee and G. L. Firestone, unpublished results.

FIG. 12.
Model summarizing the regulation of Sgk by hyperosmotic stress. We propose that hyperosmotic stress stimulates sgk promoter activity through the activation of p38 MAPK. The activation of sgk transcription can be inhibited by the overexpression of a dominant negative kinase-dead form of MKK3 (dnMKK3) and by the p38 pharmacological inhibitors (SB202190 or SB203580). In addition to ATF2, CHOP, Elk-1, and other known substrates, the activated p38 MAPK pathway targets a GC-rich hyperosmotic-regulated element located at Ϫ55 to Ϫ46 base pairs in the sgk promoter that includes the Sp1 transcription factor and most likely at least one other nuclear target. Hyperosmotic stress leads to the production of a phosphorylated and enzymatically active form of Sgk that resides in the cytoplasm. Sgk is enzymatically activated through phosphorylation by the PI 3-kinase-PDK1 signaling pathway, which we have previously shown to regulate Sgk. We propose that the transcriptionally induced and active Sgk plays a key role in the cellular response to hyperosmotic stress.
ies are in progress to identify the physiological substrates of Sgk under high osmolarity conditions, which will add greatly to the understanding of the role of Sgk in cellular stress signaling pathways.