Activation of the Janus Kinase/Signal Transducer and Activator of Transcription Pathway by Osmotic Shock*

Numerous cytokines, growth, and differentiation factors elicit their intracellular responses via Janus tyrosine kinases (Jaks) and transcription factors of the STAT (signal transducer and activator of transcription) family. Additionally, environmental stress (UV light, heat, aniso-osmolarity, and radicals) has recently been shown to activate intracellular signaling cascades such as the stress-activated protein kinases and nuclear factor-κB. In this study, we demonstrate that in different cell lines a particular stress, namely hyperosmolarity, results in tyrosine phosphorylation of the Janus kinases Jak1, Jak2, and Tyk2 and in the activation of STAT1 and/or STAT3. Both transcription factors are phosphorylated at a specific tyrosine residue and translocation to the nucleus was demonstrated by the use of a STAT3/green fluorescent protein fusion protein. A prominent role for Jak1 in the activation of STATs by hypertonicity was demonstrated by the use of Jak-deficient cell lines. Stress-activated STAT1 and STAT3 transactivate a reporter gene containing the acute-phase response element of the rat α2-macroglobulin promoter. Experiments using a diffusible solute suggest that not the increase in intracellular osmolarity but the resultant cell shrinkage is the trigger for Jak/STAT activation.

Stimulation of Cells-Cells grown in a 100-mm dish to 80% confluence were stimulated with sorbitol, sodium chloride, urea, or IL-6 (Ϯ sgp80) as indicated in the figure legends. Nuclear extracts were prepared as described by Andrews and Faller (35). The protein concentration was determined with a Bio-Rad protein assay.
Immunoprecipitation-Cells were washed twice with PBS and solubilized in 1 ml of lysis buffer (1% Brij 96, 20 mM Tris/HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA) for 30 min at 4°C. Nuclei were removed by centrifugation for 10 min at 12.000 ϫ g. The supernatants were incubated with 1 g of anti-Jak1 (polyclonal), anti-Jak2 (polyclonal), anti-Tyk2 (monoclonal), or anti-gp130 (B-T12; monoclonal) antibodies, respectively, for a minimum of 8 h at 4°C. After antibody incubation, the lysates were treated with protein A-Sepharose (5 mg/ml in lysis buffer) for 2 h at 4°C. When monoclonal antibodies were used, rabbit antimouse IgG was bound to the protein A-Sepharose beads. After centrifugation of the immunocomplexes, the Sepharose beads were washed four times with wash buffer (0.1% Brij 96, 20 mM Tris/HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA). The samples were boiled in gel electrophoresis sample buffer, and the precipitated proteins were separated on a SDSpolyacrylamide (10%) gel.
Transient Transfections-HepG2 cells were grown on 60-mm dishes to 30% confluence and transfected in Dulbecco's modified Eagle's me-dium supplemented with 10% fetal calf serum, using standard calcium phosphate precipitates, with 7.5 g of reporter construct plasmid DNA and 2.5 g of internal control plasmid DNA pCH110 (Amersham Pharmacia Biotech). Cells were incubated with precipitate for 18 h, washed twice with PBS, and media changed for additional 6 h. Cells were then stimulated for 3 h and protein extracts were prepared for CAT and ␤-galactosidase assays as described (38).
Construction of the STAT3/Green Fluorescent Protein (GFP) Chimera-Green fluorescent protein cDNA was introduced into the pSBC-2 mammalian expression vector via the EcoRI/HindIII restriction sites. An oligonucleotide (5Ј-GGC CGC GGA GAT CTG GC-3Ј) containing the BglII restriction site was cloned into the NotI/EcoRI sites of pSBC-2-GFP expression vector. A BglII site was introduced at the stop-codon of STAT3 cDNA via polymerase chain reaction. The STAT3 cDNA was subcloned into the pSBC-2-GFP expression vector via NotI/BglII restriction sites resulting in a fusion protein of STAT3 with GFP at its C terminus (pSBC-2-STAT3-GFP).
HeLa Cell Transfections and Fluorescence Studies-HeLa cells were transfected using the DNA/calcium phosphate precipitation method. Approximately 10 5 cells were seeded on coverslips and cultured for 48 h. Cells were washed twice with PBS and fixed with 2% paraformaldehyde, and coverslips were mounted on slides with Mowiol TM 4 -88 (Calbiochem Corp., La Jolla, CA). Cells were analyzed using a Nikon Eclipse E600 TM fluorescence microscope at a magnification of ϫ400.

Activation of STAT Factors by Osmotic Shock in Different
Cell Lines-In order to study whether hypertonicity (osmotic shock) results in an activation of the Jak/STAT signaling pathway, human hepatoma HepG2 cells were serum-deprived for 18 h and subsequently incubated with serum-free medium containing 600 mM sorbitol for 15 min. Such a treatment was recently shown to activate the JNK cascade in HeLa cells (18). Nuclear extracts were incubated with a 32 P-labeled oligonucleotide specific for STAT1 and STAT3 (37,39) and analyzed in an EMSA. Upon sorbitol treatment a prominent band was detected in the gel shift assay that migrated at a position typical for the STAT3 homodimer (Fig. 1A, left panel). When the same experiment was repeated in the SV40-transformed simian kidney cell line COS-7, a prominent band migrating somewhat faster was observed (Fig. 1A, middle panel). This band represents the STAT1 homodimer, which is also the major STAT form activated in these cells upon stimulation with IL-6 and the soluble form of the IL-6 receptor (sgp80) (40). A rat mesangial cell line (RMC) showed two bands after osmotic shock, a STAT1 homodimer and a STAT1/STAT3 heterodimer (Fig. 1A, right panel). These results suggest that hypertonicity leads to a rapid activation of STAT factors independent of the cell type used. However, the extent of stimulation is different A, HepG2, COS-7, and RMC cells were treated with 600 mM sorbitol in the respective culture medium for 15 min. Control cells received fresh medium. Cells were harvested and nuclear extracts were prepared as described under "Experimental Procedures." 5 g of nuclear proteins were mixed with a 32 P-labeled oligonucleotide (mutated SIE probe of the c-fos promoter 5Ј-GAT CCG GGA GGG ATT TAC GGG GAA ATG CTG-3Ј), and EM-SAs were performed. The DNA-protein complexes formed were separated from the free probe by electrophoresis on a native 4.5% polyacrylamide gel. The positions of comigrating STAT1 homodimer, STAT1/STAT3 heterodimer, and STAT3 homodimer from IL-6 stimulated HepG2 cells are indicated by the arrows. B, COS-7 cells were stimulated with either 300 mM NaCl, 600 mM sorbitol, or 100 units/ml IL-6 plus 0.5 g/ml sgp80 for 15 min. EMSAs were performed as above.
between these cell lines being most prominent in COS-7 cells. Therefore, these cells were used for the majority of the experiments. To confirm that the observed STAT activation is due to the hypertonic shock and not a specific result of sorbitol treatment, we treated COS-7 cells with medium containing 300 mM NaCl, which corresponds to a hyperosmolarity of 600 mosM. As shown in Fig. 1B, in this case STAT1 is also activated after 15 min, although to a lesser extent than after treatment with IL-6/sgp80 or sorbitol.
The dose dependence of the STAT activation by hypertonicity in COS-7 cells is shown in Fig. 2A. At 200 mM sorbitol, a faint signal was observed, which became prominent at 400 mM and reached a maximum at 600 mM ( Fig. 2A). A typical time course demonstrates that the STAT activation was maximal at 30 min but was still detectable at 180 min ( Fig. 2B). When a timecourse experiment was performed in HepG2 cells, it became apparent that in these cells both STAT3 and STAT1 were activated, however with different kinetics (Fig. 2C). In the RMC cells, a similar time course of STAT factor activation could be shown (Fig. 2D). We also tested STAT5a after transient overexpression in COS-7 cells. However, this STAT factor was not activated by osmotic stress, whereas after stimulation with prolactin via cotransfected prolactin receptor a typical band shift was detected (data not shown).
Osmotic Shock Induces Nuclear Translocation of a STAT3/ Green Fluorescent Protein Chimera-In order to study whether activation of STATs via osmotic shock also results in their nuclear translocation, we transiently expressed a STAT3/green fluorescent protein chimera in HeLa cells. In control cells, a uniform staining of the cytoplasm and nucleus was observed upon fluorescence microscopy (Fig. 3A). In contrast, in HeLa cells that were stimulated with IL-6 and sgp80 or treated with hyperosmotic medium for 30 min, an exclusive staining of the nuclei was detected (Fig. 3, B and C). Thus, hyperosmolarity leads to a recruitment of STAT3 to the nucleus.
Osmotic Shock Leads to the Tyrosine Phosphorylation of STATs-Activation of STATs usually requires phosphorylation of specific tyrosine residues (Tyr-701 in STAT1 and Tyr-705 in STAT3), which then mediate the dimerization of STATs via neighboring Src homology 2 domains (39,41). In order to demonstrate that osmotic shock induces the tyrosine phosphorylation of STAT1, COS-7 cells were treated with 600 mM sorbitol for different times and nuclear extracts were prepared and analyzed by 10% SDS-PAGE. Proteins were transferred to a PVDF membrane and incubated with an antiserum specific for Tyr-701-phosphorylated STAT1. After osmotic shock, a protein migrating at a position corresponding to STAT1 was detected (Fig. 4A). The extent of phosphorylation correlates well with the DNA binding capacity shown in Fig. 2B. In HepG2 cells, a similar phosphorylation pattern was seen for STAT1 as well as for STAT3 (Fig. 4B). Note that the phosphorylation of STAT3 precedes that of STAT1, which is also reflected in the EMSA (Fig. 2C).
Jak Kinases Are Involved in Signaling after Hypertonic Shock-The Jak1 kinase was demonstrated to be essential for the Jak/STAT signaling via the IL-6 signal transducer gp130 (33). To find out whether osmotic shock leads to an activation of Jak1, cell lysates from COS-7 cells treated with 600 mM sorbitol for 15 min were immunoprecipitated with a Jak1-specific antiserum. Western blotting was performed with an phosphotyrosine-specific antibody and after stripping with an Jak1 antibody. Hypertonicity resulted in an increased Jak1 phosphorylation, which was almost as prominent as after stimulation with IL-6/sgp80 (Fig. 5A). Interestingly, after IL-6/sgp80 treatment, an additional band migrating at about 150 kDa was coimmunoprecipitated. This band most likely corresponds to the phosphorylated signal transducer gp130 to which Jak1 is constitutively bound (data not shown). Phosphorylation of gp130, however, was hardly detectable after osmotic shock. When gp130 was directly immunoprecipitated from sorbitol-or IL-6-treated HepG2 cells, the phosphotyrosine antiserum only detected gp130 in cells that were stimulated with IL-6 (Fig. 5B). These data suggest that activation of the Jak/STAT pathway by osmotic shock does not require phosphorylation of gp130. In order to assess whether also other Jak kinases are phosphorylated upon hypertonic treatment, COS-7 and HepG2 cells were treated as above and immunoprecipitations were performed with antisera against Jak2 and Tyk2. Both kinases showed an increased tyrosine phosphorylation within 10 or 20 min after addition of sorbitol, although the Jak2 signal in COS-7 cells was very faint (Fig. 6).
To analyze whether the activation of Jaks is a prerequisite for STAT activation by osmotic shock, we made use of the Jak1-deficient cell line, U4A (33). Parental fibrosarcoma cells 2fTGH, U4A cells, and U4A cells in which the Jak1 kinase has been stably reintroduced were treated with sorbitol for 15 min, and nuclear extracts were analyzed in an EMSA (Fig. 7A). In 2fTGH cells, a prominent STAT1 activation was detected that was markedly inhibited in Jak1-deficient cells. This was even more clearly seen when IL-6/sgp80 was used as a stimulus. Reconstitution of Jak1 in U4A cells resulted in a cell line that upon stimulation with IL-6/sgp80 or sorbitol displayed an en- hanced STAT activation when compared with the parental cells (Fig. 7). These findings suggest that Jak1 is involved in STAT activation after osmotic shock. However, it does not seem to be as essential as after IL-6 stimulation (33), inasmuch as in Jak1-deficient cells a residual activation by sorbitol remained.
We also analyzed ␥2A and U1A cells, which lack Jak2 and Tyk2, respectively, and compared them with clones in which the respective kinase had been reconstituted (Fig. 7B) (33). In contrast to the Jak1-deficient cell line, ␥2A and U1A cells displayed a normal responsiveness to osmotic shock, suggesting that Jak2 and Tyk2 are not crucial for STAT activation after hypertonicity.
Activation by Osmotic Shock of a Reporter Construct Con-taining STAT Binding Sites-In order to study whether activation of STATs by osmotic shock also results in an increased gene activation, we made use of a reporter construct p7xcore/ tk/CAT in which an element containing seven identical STAT binding sites (the proximal core element of the ␣ 2 -macroglobulin promoter) was cloned in front of the basal thymidine kinase promoter followed by the chloramphenicol acetyltransferase gene. A vector only containing the thymidine kinase promoter (ptk/CAT) was used as a control. These constructs were transiently expressed in HepG2 cells, and cells were treated with IL-6 or 600 mM sorbitol for 3 h after which reporter activity was determined (Fig. 8). IL-6 stimulation of cells that were transfected with the p7xcore/tk/CAT construct resulted in an 4.6-fold increased CAT activity when compared with ptk/ CAT cells (Fig. 8). Osmotic shock resulted in a decrease to about 25% of the basal reporter activity in ptk/CAT cells (data not shown). However, in osmotically shocked p7xcore/tk/CAT cells, the reporter activity was decreased to a much lesser extent resulting in a relative induction of about 2.4-fold (Fig. 8).
Thus, hyperosmotically activated STATs are capable of transactivating a respective reporter gene. Hypertonicity-induced Cell Shrinkage Is Crucial for STAT Activation-In order to assess whether hyperosmolarity by itself or the resultant cell shrinkage is the trigger for Jak/STAT activation, we increased the hypertonicity of the medium with 600 mM urea, which is known to be a rapidly permeating solute that does not affect cell volume (42). When nuclear extracts from these cells were analyzed in an EMSA, no STAT activation was detected (Fig. 9A). However, when these cells were "super-induced" with either IL-6/sgp80 or increasing concentrations of sorbitol again, a prominent STAT activation was seen (Fig. 9B). From these results, we conclude that hypertonicity itself is not sufficient for activation of the Jak/STAT pathway and that urea does not inhibit STAT activation.

DISCUSSION
In this study, we demonstrate that hyperosmotic shock activates the Jak/STAT pathway. This pathway is known to mediate the signaling of numerous cytokines, growth, and differentiation factors (19,30). Thus, we and others have recently shown that IL-6 activates three Jak tyrosine kinases, namely Jak1, Jak2, and Tyk2, which then phosphorylate the IL-6 signal transducer gp130 and the latent cytoplasmic transcription factors STAT1 and STAT3 (43)(44)(45). Activated STAT factors homo-and heterodimerize via an Src homology 2 domain/phosphotyrosine interaction, are transported to the nucleus, and FIG. 6. Tyrosine phosphorylation of Jak2 and Tyk2 after osmotic shock. COS-7 (A) and HepG2 (B) cells were treated with 600 mM sorbitol in the respective medium. Cell lysates were prepared and immunoprecipitations with either anti-Jak2 or anti-Tyk2 antibodies were performed as described under "Experimental Procedures." Precipitated proteins were separated by 10% SDS-PAGE, blotted onto a PVDF membrane, and analyzed by immunodetection with a specific anti-phosphotyrosine antibody (A  and B, upper panel). Blots were stripped and reprobed with either anti-Jak2 or anti-Tyk2 antibodies for verification of equal loading (A and B, lower panel) as indicated.
In COS-7 cells, hypertonic treatment led to an activation of STAT1. In contrast, in HepG2 and in RMC, both STAT1 and STAT3 were tyrosine phosphorylated. This preference for certain STATs could be explained either by different expression levels of STAT factors or by unknown additional regulatory mechanisms (e.g. specific phosphatases) in these cells. Activation of STATs by osmotic shock is not a general phenomenon since STAT5a was not activated even after expression in COS-7 cells. In both COS-7 and HepG2 cells, all three endogenous Jak kinases, Jak1, Jak2, and Tyk2, were tyrosine-phosphorylated after hypertonic stress, although to a different extent. Using cell lines deficient in one of the three kinases, we could demonstrate that Jak1 is involved in STAT activation upon stress, since in Jak1-negative U4A cells the STAT activation was largely reduced. Restoration of Jak1 in these cells by overexpression also restored and even enhanced the STAT1 signal. This was not observed in Jak2-deficient ␥2A and Tyk2-deficient U1A cells, suggesting either that these kinases do not play an important role or that they can replace each other. Jak1 was also found to be crucial for activation of STAT1 and STAT3 after IL-6 and interferon-␥ (33,49).
Stress-activated STATs could stimulate gene transcription of reporter gene constructs containing well characterized STAT1/3 binding sites as enhancers. When compared with the activity of the basal thymidine kinase promoter, both IL-6 and osmotic stress resulted in induction of reporter gene activity. However, it should be pointed out that under hypertonicity the basal CAT activity was strongly reduced. This could be due to a general suppression of transcription and/or to an increased degradation of the chloramphenicol acetyltransferase. An increased protein catabolism after cell shrinkage was observed in several cell systems (8).
Dimerization of surface receptors is currently believed to be the trigger for activation of associated Jaks. However, we have no evidence for an involvement of the signal transducer gp130 in the hyperosmotic signal cascade since tyrosine phosphorylation of gp130 was absent after osmotic shock. Thus, stress activation of the Jak/STAT is either triggered by a mechanism located downstream of gp130 or other cytokine/growth factor receptors are involved. Recently, it was shown that hyperosmolarity and UV light lead to a clustering and phosphorylation of the receptors for EGF, IL-1, and tumor necrosis factor ␣ (18). The EGF receptor has been described to be an activator of STAT1 and STAT3 (20,39,50), and EGF stimulates the tyrosine-phosphorylation of Jak1, a function that is dependent on the intrinsic kinase activity of the EGF receptor (41). However, phosphorylation of STATs by EGF does not require Jak1 (51), and therefore the EGF receptor is probably not involved in stress activation of STATs.
It is currently unknown whether the initial trigger for activation of intracellular signal cascades is the hyperosmolarity itself or the resulting cell shrinkage. When COS-7 cells were treated with 600 mM urea, no STAT activation was observed. Since urea readily diffuses into the cell, it does not lead to gross changes in cell volume (42), although a slight shrinkage of liver parenchymal cells due to opening of K ϩ channels in the plasma membrane was observed (52). Thus, raising the intracellular osmolarity is obviously not sufficient for signaling. Additionally, the pathways leading to activation of JNKs and p38 were insensitive to hyper-osmotic urea (53). However, addition of a non-permeating solute (600 mM sorbitol) or the IL-6/sgp80 complex to cells equilibrated in 600 mM urea still resulted in a STAT activation, suggesting that the cell shrinkage itself is the trigger for Jak/STAT activation. In human neutrophils, osmotic cell shrinkage induces tyrosine phosphorylation of several proteins and activation of the Na ϩ /H ϩ exchanger (42). In Chinese FIG. 8. Hypertonically activated STATs transactivate a reporter construct containing STAT binding sites. HepG2 cells were transfected with the ptk/CAT and the p7xcore/tk/CAT construct, respectively. After 24 h, cells were treated with 100 units/ml IL-6 or 600 mM sorbitol for 3 h and processed for reporter assays. CAT activities were normalized to ␤-galactosidase activities, and the ratio p7xcore/tk/CAT: ptk/CAT was calculated. The data presented are the mean of six values obtained from three independent experiments. The bars indicate the standard deviation. hamster ovary cells, an osmotically induced decrease in cell volume resulted in the tyrosine phosphorylation of three proteins of about 42, 85, and 120 kDa (54). The identity of the two larger proteins was not determined; however, it is conceivable that they may represent a STAT factor and a Jak kinase, respectively.
The physiological relevance of the activation of the Jak/STAT pathway by osmotic stress is currently unknown. Jaks and STATs are involved in the regulation of many important cellular processes such as growth and differentiation, e.g. STAT3 is involved in the regulation of several immediate early genes (jun-B, tis11) (46,47), acute phase protein genes (36,48) and the tissue inhibitor of metalloproteinases-1 gene (55). Whether this transcriptional activation triggers a program that is necessary for the adaptation of the cell to the osmotic stress must be studied in the future.