Shrinkage-induced Protein Tyrosine Phosphorylation in Chinese Hamster Ovary Cells*

To investigate the signal transduction of osmotic stress, we examined hypertonicity-induced tyrosine phosphorylations in Chinese hamster ovary cells. Hyperosmosis elicited characteristic phosphotyrosine accumulation in at least 3 proteins (≈42, ≈85, and ≈120 kDa). The most prominent response occurred in the 85-kDa band (p85) whose phosphorylation was rapid, sustained, apparent already at mild hypertonicity (350 mosm), proportional to the extracellular osmotic concentration, and reversible. Hyperosmotic environment could not induce tyrosine phosphorylation if cell shrinkage was prevented by nystatin and appropriately composed media. Conversely, isotonic shrinkage caused strong tyrosine phosphorylation. Thus, the initial signal is a decrease in cell volume and not an increase in the intra- or extracellular osmotic concentration, or a rise in cytosolic K+ and Cl− levels. Tyrosine phosphorylation of p85 was not due to the hypertonicity-induced protein kinase C-dependent stimulation of the extracellular signal-regulated protein kinase, nor to the activation of stress-activated protein kinases. Tonicity-responsive proteins interacted with Grb2-glutathione S-transferase fusion proteins: the 120-kDa protein complexed with the SH2 and both SH3 domains, whereas p85 associated with the SH2 and the N-terminal SH3 domains of the adapter. Tyrosine phosphorylation of p85 is a sensitive indicator of reduced intracellular hydration and might signify a hitherto unrecognized, early volume-dependent signaling event.

Osmotic changes in the extracellular environment can influence a variety of vitally important cell functions (reviewed in Ref. 1). Alterations in tonicity has been reported to regulate several ion transporters (including different isoforms of the Na ϩ /H ϩ exchanger (NHE-1-4) 1 (2)(3)(4)(5), the Na ϩ /K ϩ /Cl Ϫ cotrans-porter (6), K ϩ and Cl Ϫ channels (see Ref. 7)) to modify the activity of key metabolic enzymes (1) and to affect the transcription of certain genes (8,9). Although many of these events are of homeostatic nature which serve the restoration of the near-normal cell volume (7,10), ample evidence has been accumulating that moderate volume changes themselves constitute an important signal that can be induced by and may be necessary for the action of different metabolic hormones (1). While the role of cellular hydration state as both a regulated and a regulating factor has been well established, very little is known about the mechanisms through which it is detected and about the subsequent signaling steps that convey the information to the effectors. Not even the exact parameter sensed during aniso-osmotic challenge is clear: it may be a change in the extra-and/or intracellular total osmotic activity, or an alteration in the cell volume. Furthermore, the volume change itself may exert its effect via various mechanisms. Conceivably, it may act by eliciting changes in the intracellular ionic strength or in the concentration of specific ions or other regulatory constituents. It is noteworthy that cytosolic Cl Ϫ concentration was found to modulate the operation of the Na ϩ /H ϩ exchanger (11,12) and the Na ϩ /K ϩ /Cl Ϫ cotransporter (13)(14)(15) as well as the activity of different enzymes including glycogen synthase (16) and yet unidentified protein kinases (17,18). Alternatively, the mechanical strain of various cytoskeletal components may also initiate some of the effector responses (see Ref. 7).
Whatever the initial events are, the further processing of the signal is believed to involve a set of protein phosphorylations. Seminal discoveries, first made on yeast (19 -21) and later verified in mammalian cells, have indicated that hyperosmotic stress stimulates certain extracellular signal regulated protein kinases (ERKs or mitogen-activated protein kinases) and a related group of enzymes, the stress-activated protein kinases (SAPKs or JNKs). Serine/threonine kinases which were shown to be stimulated by hyperosmolarity in mammalian cells now include ERK-1 and 2 (22)(23)(24), SAPK46 and 55 (23,(25)(26)(27), p38 (28,29), and the recently described Big MAP kinase (also known as ERK-5) (30). Interestingly in certain cell types hypotonicity also was found to activate ERK-1 and -2 (31,32). While the potential link between these serine/threonine kinases and the effector functions (e.g. membrane transporters) remained unresolved, pharmacological data indicate that protein tyrosine phosphorylations may also play a pivotal role in mediating some of the hypo-or hyperosmotic stress-induced functional responses. Tyrosine kinase inhibitors abrogated the hypotonicity-triggered ion efflux from epithelial cells (31) and prevented the hypertonicity-evoked inhibition of the apical Na ϩ /H ϩ exchanger (NHE-3) in kidney cells (33). These drugs were reported to suppress ERK activation induced by hypotonicity (32) but did not abolish the hypertonicity-elicited stimulation of this MAP kinase (22). This finding suggests that their effects may not be ascribed to the inhibition of MEKs and SEKs, the only identified members of the above cascades exhibiting tyrosine kinase activity. Despite their likely significance, the osmotic stress-induced protein tyrosine phosphorylations have never been systematically studied. Very little information is available about tonicity-regulated tyrosine kinases and their substrates are mostly unknown.
The aim of the present work was to investigate whether hypertonicity provokes characteristic tyrosine phosphorylations in cellular proteins and to establish whether the primary signal for these reactions is a change in the intracellular osmotic concentration or in the cell volume. We also addressed whether the observed tyrosine phosphorylations might be downstream of the activity of ERKs and SAPKs or may signify upstream or independent events. These experiments were performed on Chinese hamster ovary (CHO) cells expressing known isoforms of the Na ϩ /H ϩ exchanger in which the functional responses to hypertonicity (i.e. stimulation of NHE-1 and inhibition of NHE-3) have been well characterized (2,34,35).
Our results indicate that hypertonicity induces tyrosine phosphorylation in several proteins. The most prominent of these is an approximate 85-kDa band. Its phosphorylation is quick, proportional, and specific to the imposed osmotic stress, requires a decrease in the cell volume, and is not downstream to ERKs or SAPKs. The 85-kDa protein may represent an early element of a shrinkage-sensitive signaling pathway.

EXPERIMENTAL PROCEDURES
Materials-Dimethyl sulfoxide, nystatin kept in a stock solution of 400,000 units/ml (77.5 mg/ml) in dimethyl sulfoxide, phorbol 12-myristate 13-acetate (PMA), Hepes, N-methyl-D-glucammonium (NMG), nocodazole, anisomycin, proteinase inhibitors, cytochalasin B, colchicine, and tissue culture reagents were purchased from Sigma. Genistein was from Calbiochem, PD 098059 was a kind gift from Dr. Julian Downward. Anti-phosphotyrosine (clone 4G10) was obtained from Upstate Biotechnology Inc. (UBI). Anti-ERK-2 and anti-Pan-ERK were from Transduction Laboratories. Polyclonal anti-phosphoinositide 3-kinase used for Western blotting was from UBI. Since this antibody was generated against a GST fusion protein, the association of GST-Grb2 with PI 3-kinase was detected with another anti-PI-3 kinase, obtained from Sigma. Anti-phospho-p38 and anti-phospho-SAPK antibodies which specifically react with the phosphorylated forms of the respective kinases were from New England Biolabs. Protein-A Sepharose was obtained from Pierce. Peroxidase-conjugated anti-mouse IgG and the enhanced chemiluminescence protein kit were purchased from Amersham. Grb2-glutathione S-transferase (Grb2-GST) fusion proteins coupled to agarose resin were prepared as described in Ref. 36. [ 14 C]Sucrose (540 mCi/mmol ϭ 20 GBq/mmol) was from Amersham and tritiated water (0.78 mCi/ml ϭ 29 MBq/ml) was from Izotóp Intézet, Budapest.
Cell Culture-Wild type CHO K1 or CHO cells devoid of endogenous NHE and stably transfected with rat NHE-1 or NHE-3 were generous gifts from Dr. G. L. Luká cs and Dr. John Orlowski, respectively. Cells were grown in ␣-minimal essential medium containing 25 mM NaHCO 3 , supplemented with 10% fetal calf serum, penicillin, and streptomycin, under a humidified atmosphere of air/CO 2 (19:1) at 37°C. Cultures were re-established from frozen stocks and cells from passage number between 8 and 25 were used. Prior to experiments cells were incubated for at least 3 h in serum-and HCO 3 Ϫ -free RPMI 1640 at 37°C. To down-regulate protein kinase C (PKC) the culture medium was supplemented with 100 nM PMA for 18 h and PMA was also present during serum depletion. The pattern of hypertonicity-induced phosphorylations was identical in the different transfectants and in the wild type CHO cells.
Preparation of Cell Extracts-Confluent cultures grown in 6-well tissue culture plates or glass Petri dishes were preincubated for 10 min in iso-sodium medium and then subjected to various treatments as indicated. The bathing medium was then removed and cells were lysed with an ice-cold buffer containing 100 mM NaCl, 1% Triton X-100, 20 mM sodium fluoride, 1 mM Na-EGTA, 10 mM benzamidine, 10 g/ml aprotinin, 10 M leupeptin, 2 M pepstatin A, 1 mM Na 3 VO 4 , 1 mM phenylmethylsulfonyl fluoride, 30 mM Na-Hepes, pH 7.5.
Detection of Tyrosine Phosphorylation-Aliquots of total cell lysates were diluted (1:1) into 2 ϫ Laemmli sample buffer, boiled for 3 min, and subjected to electrophoresis on 10% SDS-polyacrylamide gels. The separated proteins were transferred to nitrocellulose (Hoefer Pharmacia Biotech Inc.) using a Bio-Rad Mini Protean II apparatus. To check effectiveness of transfer and similarity of protein amount, lanes were visualized by staining with Ponceau S. Blots were then blocked in Tris-buffered saline containing 2% bovine serum albumin and incubated with the anti-phosphotyrosine antibody (dilution 1:5000). The binding of the antibody was visualized by peroxidase-coupled goat antimouse IgG, using the enhanced chemiluminescence method according to the manufacturer's instructions.
Densitometry-Quantification of the bands was performed using an UltroScan XL laser densitometer (LKB, Bromma) and evaluation of data was carried out with the Gelscan® computer program. In the absorbance range used for analysis, the OD values of the bands were linearly proportional to the protein amount as verified by blotting and scanning different (1-10-fold) dilutions obtained from the same sample. Hypertonicity-induced changes are always expressed as -fold increase in the integrated OD compared with the corresponding isotonic control.
Gel Retardation Assay for Detection of ERK Activation-To follow ERK activation, samples were subjected to electrophoresis on 15% polyacrylamide gels (16 ϫ 18 cm) and transferred to nitrocellulose. The blots were blocked in Tris-buffered saline containing 2% bovine serum albumin then incubated with anti-ERK-2 or anti-Pan-ERK antibodies (500-fold dilution) in 0.5% bovine serum albumin containing Tris-buffered saline-Tween. After extensive washing, the reaction was detected with a peroxidase-coupled goat anti-mouse IgG and blots were developed using enhanced chemiluminescence.
Detection of Phosphorylation of SAPK and p38 -This was carried out by Western blotting using a phospho-specific SAPK antibody (dilution 1:1000), and a phospho-specific p38 antibody (dilution 1:500).
Detection of in Vitro Interaction of Proteins with Grb2-GST-To precipitate proteins interacting with the full-size Grb2 or with certain domains of the adapter, cells were treated and lysed as described earlier then Triton X-100-insoluble material was sedimented out. The various Grb2-GST fusion proteins (5 g/sample) coupled to agarose resin and containing either full-length Grb2 or a mutant version of it (comprising either only a functional SH2 domain, or the N-terminal SH3 domain or the C-terminal SH3 domain) were added to the supernatant. After allowing 1 h for the interaction to occur under continuous rotation at 4°C, samples were sedimented in Eppendorf centrifuge 5402 (4°C, 15,600 ϫ g). The pellet was washed 3 times with phosphate-buffered saline supplemented with 0.2% Triton X-100 and 1 mM Na 3 VO 4 , and the supernatant was removed. The pellet was suspended in sample buffer, boiled for 5 min, and subjected to Western blotting with anti-phosphotyrosine or anti-phosphoinositide 3-kinase.
Immunoprecipitation of Phosphoinositide 3-Kinase-Lysates containing equivalent amounts of proteins obtained from iso-or hypertonically-treated cells were clarified by centrifugation at 15,600 ϫ g for 5 min. Extracts were incubated overnight with 8 g of anti-phosphoinositide 3-kinase followed by a 3-h incubation with immobilized protein A at 4°C. Then beads were sedimented and washed four times with lysis buffer containing 0.25% Triton X-100 and 1 mM Na 3 VO 4 . Precipitated proteins and aliquots of the first supernatant were electrophoresed and probed with anti-phosphotyrosine. After stripping, the same blot was developed with anti-phosphoinositide 3-kinase (1:1000 dilution).
Determination of Intracellular Water Space-Cell volume was measured as the difference between the [ 14 C]sucrose permeable space and total 3 H 2 O space of the samples (37). Cells from confluent cultures were detached with trypsin-EGTA, washed, and resuspended (approximately 10 6 cells/sample) in 1 ml of the corresponding medium (iso-sodium, iso-NMG, or iso-potassium supplemented with 1 mM unlabeled sucrose).
When applied, 400 units/ml nystatin was added to the cells for 5 min. Thereafter cells were either left untreated or exposed to hypertonicity by the addition of 200 mM sucrose (from a 2 M stock) or 100 mM KCl (from a 1 M stock solution). Three minutes later (or at the indicated time) [ 14 C]sucrose (1 Ci/ml) and 3 H 2 O (10 Ci/ml) were added to the suspension and after an additional 2-min incubation the cells were pelleted (15,600 ϫ g for 1 min). After taking a 10-l aliquot for scintillation counting, the rest of the supernatant was completely removed and the pellet was dissolved in 50 l of iso-sodium containing 1% Triton X-100. Radioactivity of the whole pellet and the sample from the supernatant was determined in a toluene/Triton based mixture using a Beckman LS-250 liquid scintillation spectrometer.
Other Methods-Data are presented as representative immunoblots of at least three similar experiments or as mean Ϯ S.E. of the number of experiments indicated (n). Protein concentration was determined with Bio-Rad Protein assay using bovine serum albumin as standard.

Characterization of the Osmotic Stress-induced Tyrosine
Phosphorylation in CHO Cells-To test the effect of osmotic challenge on tyrosine phosphorylation of proteins, confluent cultures of CHO cells were exposed to iso-or hypertonic media and whole cell lysates obtained from these samples were analyzed by immunoblotting with anti-phosphotyrosine antibodies. Hyperosmolarity (imposed by the inclusion of 200 mM sucrose into the iso-sodium medium) caused an increase in the level of tyrosine phosphorylation in some protein bands with apparent molecular masses of Ϸ42, Ϸ85, and Ϸ120 kDa (Fig.  1A, lanes ISO and HYPER). The most remarkable and always reproducible phosphotyrosine accumulation appeared in the Ϸ85-kDa band. In lysates of isotonically-treated cells this region showed only a faint labeling over the background whereas a prominent signal occurred in the extracts obtained from hypertonically stimulated samples. The overall increase in the absorbance of this area was 8.6 Ϯ 1.1-fold (n ϭ 10) as measured by densitometry (Fig. 1B). The observed rise in chemiluminescence was due to protein tyrosine phosphorylation and not to an aspecific binding of the primary or secondary antibody as evidenced by the following observations. 1) No labeling was found in this region when the blots were incubated with the primary antibody in the presence of 1 mM phosphotyrosine (not shown); 2) the band was not detected either when only the secondary antibody was applied (not shown); and 3) pretreatment of the cells with the protein tyrosine kinase inhibitor genistein (50 M, 30 min) largely reduced (but did not completely abolish) the change provoked by hypertonicity (Fig. 1C). As the phosphorylation of the 85-kDa protein (referred below as p85) seemed to be far the most sensitive to increased tonicity, we decided to use this band as a semi-quantitative indicator for further characterization of the hyperosmotic response.
To establish the dependence of tyrosine phosphorylation on the extracellular osmotic concentration, we treated the cells with sodium medium supplemented with varying concentrations of sucrose. Increased tyrosine phosphorylation of p85 was occasionally observable upon addition of as low as 25 mM sucrose, whereas a clearly detectable rise consistently occurred after using 50 mM sugar (total osmolarity approximately 350 mosM). Other non-permeable osmolytes (NaCl and KCl) were equally effective in eliciting phosphorylation (see below). The signal became markedly stronger with increasing osmolarity, approaching a plateau above 600 mosM (Fig. 2, A and B). Fig. 2, C and D, reports the kinetic of the response. Using an osmotic concentration of 500 mosM, the signal was readily visible after 0.5-1 min (the earliest time points that could be reliably tested) and it increased further with time. Maximal effects were generally reached between 5 and 10 min. In most samples the signal showed a slow decay at times longer than 20 -30 min. To test the reversibility of tyrosine phosphorylation, cells were first treated with a hypertonic solution for 10 min after which isotonicity was re-established for varying times before lysing the cells. As illustrated on Fig. 2E, upon readdition of the isotonic sodium medium a quick and large decrease occurred in the phosphotyrosine content of p85: after 1 min the labeling became 80% weaker and by 2 min it dropped to the background level. To decide whether downward departures from isotonicity could also elicit the phosphorylation of p85, we reduced the osmolality of the extracellular medium by 33%. This maneuver did not cause phosphotyrosine accumulation in this band. Taken together, the tyrosine phosphorylation (and dephosphorylation) of the 85-kDa protein is a rapid, reversible, time-and osmotic concentration-dependent process, the level of which reflects the course and magnitude of changes in extracellular osmotic activity in a wide range above the isotonic values.
Is Cell Shrinkage a Prerequisite for Inducing Tyrosine Phosphorylation by Hypertonicity?-When the extracellular tonicity is increased by non-permeable osmolytes, cells undergo a proportional shrinkage. It was of interest to establish whether the trigger for the observed phosphorylation events is a change in osmotic concentration itself, or rather the consequent change in the cell volume.
To dissect the imposed changes in intra-and extracellular tonicity (and ionic strength) on the one hand from the changes in the cell volume on the other, we used two approaches. 1) We applied hypertonicity while the shrinkage was prevented and 2) we induced cell shrinkage under isoosmotic conditions. These manipulations were brought about by using appropriately composed extracellular media and the non-selective, monovalent ionophore, nystatin as detailed below.
To mimic cytoplasmic ionic concentrations the following experiments were performed in an isotonic KCl medium. The volume of the cells bathed in this iso-potassium solution was stable and identical with the value measured in the iso-sodium medium (not shown). Keeping the cells in iso-KCl medium did FIG. 1. Hyperosmolarity induces tyrosine phosphorylation of proteins in CHO cells. A, cells were preincubated in isotonic sodium medium (Iso-Na) for 10 min. The medium was then replaced either by the same iso-sodium (290 mosM) or by hyper-sodium medium (isosodium supplemented with 200 mM sucrose, 500 mosM) for an additional 10 min at 37°C. At the end of the incubation period cells were lysed with ice-cold lysis buffer and tyrosine phosphorylation in the extracts obtained from the iso-or hypertonically-treated samples was detected by Western blotting with a monoclonal anti-phosphotyrosine antibody using enhanced chemiluminescence (lanes ISO and HYPER, respectively). Upon hypertonic exposure the predominant tyrosine phosphorylation was seen in a band of approximately 85 kDa (indicated by the arrow). B, tyrosine phosphorylation of the 85-kDa band was quantified by densitometry (see "Experimental Procedures"). In each experiment the integrated OD of the corresponding band in the extract of isotonically-treated samples (taken as unity) was compared with the OD of the band obtained from lysates of hypertonically-treated samples (n ϭ 10). C, effect of genistein on the hypertonicity-induced tyrosine phosphorylation of p85. Cells were incubated for 20 min in iso-sodium medium in the presence of 50 M genistein (GENISTEIN) or the same volume of dimethyl sulfoxide (CONTROL), then the medium was changed for either iso-sodium (ISO) or hyper-sodium (HYPER) supplemented with 50 M genistein for 10 min. not induce any change in the level of tyrosine phosphorylation of the 85-kDa protein (Fig. 3A, lane 1) or in any other tonicitysensitive band. Raising the extracellular osmolarity by addition of 100 mM KCl resulted in approximately 30% cell shrinkage and in a substantial increase in the tyrosine phosphorylation of the 85-kDa band (Fig. 3A, lanes 1 and 3, and  columns 1 and 3). An entirely different pattern was obtained if this experiment was carried out on cells which had previously been permeabilized for monovalent ions with nystatin. The ionophore added to the cells induced a sizable (3.5-fold) swelling (Fig. 3A, column 2) in complete agreement with earlier observations made on HeLa cells (38). This volume increase is due to a colloid-osmotic water uptake because in these permeabilized cells the major intracellular ions are no more capable of counterbalancing the osmotic pressure exerted by the intracellular proteins (38). Under this condition of isotonic hypervolaemia the basal level of phosphorylation was identical with that observed in control cells (compare lanes 1 and 2 on Fig.  3A). As expected, raising the extracellular osmolarity by the addition of 100 mM KCl to the permeabilized cells did not induce any decrease in their volume (column 4). Importantly, this treatment failed to elicit tyrosine phosphorylation either (lane 4). It is noteworthy, that although the cell volume remained unaffected in nystatin-permeabilized cells, the addition of KCl increased both the intra-and extracellular osmotic activity, ionic strength, K ϩ , and Cl Ϫ concentrations. Despite these changes, however, no phosphorylation occurred. An alternative interpretation of this lack of the response to KCl in the ionophore-treated cells could be that nystatin unspecifically interfered with the normal operation of tyrosine kinases or other tonicity-sensitive elements. To test this assumption, control and nystatin-treated cells were exposed to a hypertonic medium in which osmolarity was elevated by sucrose (200 mM) instead of KCl. As sucrose cannot permeate through the nystatin-formed pores, in this case hypertonicity led to a pronounced shrinkage in both the control and in the drug-treated cells (Fig. 3A, columns 5 and 6). In fact the sucrose-induced shrinkage was even bigger in the presence of nystatin. In keeping with this observation, sucrose induced substantial tyrosine phosphorylation both in the absence and presence of nystatin. The intensity of this reaction was proportional to the corresponding volume changes (Fig. 3A, lanes 5 and 6).
To further substantiate that neither nystatin itself nor the transient swollen state of the cells caused by the drug were responsible for the prevention of tyrosine phosphorylation upon addition of KCl, the same experiments were repeated using an isotonic medium in which 30 mM KCl was replaced by 60 mM sucrose. In this solution the colloid-osmotic pressure of cellular proteins was mostly counterbalanced by the impermeable sugar. This is reflected by the fact that under these conditions nystatin caused only a marginal swelling (Ͻ15%). In all other respects, however, the results were identical with those reported on Fig. 3A: nystatin prevented the KCl-but not the sucrose-induced tyrosine phosphorylation (not shown).
Thus we have established that hypertonic environment without cell shrinkage is not capable of facilitating tyrosine phosphorylation of p85. Next we asked whether a decrease in cell volume in the absence of hyperosmosis is sufficient to induce the phenomenon. To test this, cells were exposed to an isotonic medium (NMG-gluconate) devoid of the major intracellular ionic constituents. This composition was chosen to generate extracellularly directed gradients which can serve as a driving force for the spontaneous efflux of ions (mostly K ϩ and Cl Ϫ ) and the accompanying water leading to cell shrinkage. Incubation of the cells for 15 min in this environment resulted in a huge loss in their water content and a strong tyrosine phosphorylation of the 85-kDa band (Fig. 3B). In this case nystatin (i.e. rising the ion permeability of the membrane) did not inhibit but rather facilitated tyrosine phosphorylation (not shown). Thus the same ionophore could prevent or promote tyrosine phosphorylation of p85 in perfect concordance with its effect on cell volume under various conditions. Taken together, we conclude that a decrease in cell volume is certainly sufficient to trigger tyrosine phosphorylation of p85. On the other hand an increase in the intra-or extracellular osmotic concentration, ionic strength, or KCl level do not seem to be necessary or required for this reaction.

Investigation of the Possible Relationship of p85 Tyrosine Phosphorylation to the Osmo-sensitive Kinase Cascades-In
Madin-Darby canine kidney cells hypertonicity was reported to turn on a kinase cascade involving the sequential activation of protein kinase C, Raf1 kinase, MAP kinase kinase (MEK), and ERK (22,24). The aim of the following experiments was to establish whether a similar signaling route operates also in CHO cells and to clarify whether the tyrosine phosphorylation of p85 might be a consequence of the activation of one or more members of this cascade. We applied different treatments known to influence the activity of PKC and the activation of the

FIG. 2. Characterization of the hypertonicity-induced tyrosine phosphorylation of p85.
A and B, dependence of tyrosine phosphorylation of p85 on extracellular osmolarity. Cells were exposed to media with the indicated osmotic concentration (iso-sodium medium supplemented with varying concentration of sucrose) and lysed after 10 min. Samples from whole cell lysates were subjected to Western blot analysis as in Fig. 1. One typical blot is shown out of 4 similar ones. Numbers below the lanes indicate osmolarity in mosM. A, quantification of the response was performed by densitometry as in Fig. 1. Data are from three to five independent experiments (B). C and D, time course of tyrosine phosphorylation of p85. Cells were exposed to hyperosmotic stress (500 mosM) for the indicated time and tyrosine phosphorylation was detected as above. The blot (C) and the densitometric analysis (D) show the results of one typical experiment out of 4 similar ones. E, reversibility of the osmotic stress-induced tyrosine phosphorylation of p85. Cells were pretreated for 10 min in hyper-sodium medium, then immediately lysed (HYPER) or the medium was removed and the cells were further incubated for 1 or 2 min in iso-sodium before lysis. classical MAP kinase pathway, and compared the effect of these manipulations on ERK stimulation and p85 phosphorylation. The most abundant MAP kinase variant in our CHO cells was ERK-2, since using a PAN-ERK antibody (raised against an epitope that is common in several MAP kinases) resulted in the predominant labeling of a 42-kDa band which was also visualized by a specific anti-ERK-2 antibody. Stimulation of ERK-2 was detected as a reduction in the electrophoretic mobility of the enzyme indicative of its phosphorylation and consequent activation (32). Protein kinase C-activating phorbol esters were shown to stimulate ERK-2 in several cell types (26,39,40). Indeed PMA induced a prominent shift in the mobility of a substantial portion of ERK-2 also in CHO cells as evidenced by the appearance of two distinct anti-ERK-2 immunoreactive bands ( To determine the potential role of PKC in the hypertonicity-related phosphorylations, cells were depleted of PKC by overnight incubation with PMA, and subsequently they were challenged either by the phorbol ester again or by hyperosmolarity. Under these conditions PMA was no more capable of inducing ERK-2 activation (lane 3, lower panel) verifying that the down-regulation of PKC was effective. Importantly, in PKC-depleted cells also, hypertonicity lost its potential to activate ERK-2 but it preserved its effect on p85 phosphorylation (lane 5, upper panel). Moreover in non-depleted cells both the PMA-and the osmotic shockpromoted ERK-2 activation but not p85 tyrosine phosphorylation was completely abolished by PD 098095, a specific inhibitor of MEK (41) (Fig. 4B).
Together these results unambiguously indicate that the hypertonicity-induced stimulation of ERK-2 is mediated by the sequential activation of protein kinase C and MAP kinase kinase also in CHO cells. However, neither PKC nor MEK or ERK-2 are involved in the hyperosmolarity-related tyrosine phosphorylation of the 85-kDa protein.
The SAPKs (also referred to as c-Jun N-terminal kinases or JNKs) constitute another branch of the MAP kinase superfamily that was shown to be activated by various cellular stresses including hyperosmolarity (42,43). The question whether these enzymes could be upstream of the phosphorylation of p85 was approached by pharmacological means and by applying different stress conditions. The protein synthesis inhibitor anisomycin was demonstrated to be the most potent activator of SAPKs in various cell types including CHO cells (26,44,45). This drug, FIG. 4. The hypertonicity-induced tyrosine phosphorylation of p85 is not dependent on the osmotic stress-promoted activation of PKC and ERK-2. The activation of ERK-2 (as detected by the mobility shift assay) and tyrosine phosphorylation of the 85-kDa band were followed in parallel. Aliquots of the sample were electrophoresed either on big-sized (16 ϫ 18 cm) 15% gels or 10% mini gels, blotted onto nitrocellulose, and probed with anti-ERK-2 (lower panels, marked as ERK-2) or anti-phosphotyrosine (upper panels marked as PY) antibodies, respectively. A, cells either untreated or depleted of PKC (Depl) were incubated in iso-sodium medium for 10 min and then challenged with the following additions: none (cont); 50 nM PMA for 5 min (PMA); 600 mM sucrose (Ϸ900 mosM) for 10 min (900). Details of the PKC depletion are described under "Experimental Procedures." B, cells were preincubated for 20 min in iso-sodium medium without (lanes 1 and 3) or with (lanes 2 and 4) the MEK inhibitor, PD098059, and then exposed to PMA (lanes 1 and 2) or hyperosmolarity (lanes 3 and 4) for 10 min. Lysates were made and processed as detailed above.

FIG. 3. Tyrosine phosphorylation is induced by cells shrinkage and not by an increase in intra-or extracellular osmolarity.
Changes in cell volume (columns, upper panels) and tyrosine phosphorylation of p85 (blots, lower panels) were followed in parallel. A, cells were kept in iso-potassium medium and treated as indicated in the table under the blots. Nystatin (400 units/ml), when present, was added to the cells 5 min prior to any further treatment. Hypertonic exposure (10 min) was achieved by the addition of either 100 mM KCl or 200 mM sucrose. For volume determinations, cells were used in suspension (Ϸ10 6 /ml) and the intracellular water space was measured by [ 14 C]sucrose and 3 H 2 O as detailed under "Experimental Procedures." Data were normalized to the average volume of untreated cells obtained in iso-potassium medium (n ϭ 3). Tyrosine phosphorylation was detected as described in the legend to Fig. 1. B, cell volume and tyrosine phophorylation were determined after a 10-min incubation either in iso-sodium or iso-NMG medium as indicated. Neither of the above treatments caused cell detachment or lysis as verified by microscopic inspection and protein determinations. however, failed to induce tyrosine phosphorylation of p85 and did not interfere with the appearance of the reaction upon exposing the cells to a hypertonic environment (Fig. 5, lanes  marked as aniso). Similarly, heat shock (warming up the cells to 42°C for 30 min), another well known inducer of SAPKs (26,46) did not elicit p85 phosphorylation and did not affect the hypertonic response (lanes marked as 42°C). Finally the potential action of oxidative stress was tested. Incubating the cells for 5 or 10 min with 200 or 400 M H 2 O 2 had either no effect or resulted in a slight (hardly detectable) increase in the level of phosphotyrosine accumulation into p85. Osmotic stimulation could always markedly increase the level of phosphorylation of p85 also in H 2 O 2 -treated cells (lanes marked as H2O2). Similar results were obtained when the xanthine/xanthine oxidase system was applied to generate superoxide anions initiating the formation of other aggressive oxygen radicals (not shown).
These findings lend credence to the conclusion that tyrosine phosphorylation of the 85-kDa protein is not downstream to the activation of SAPKs. Moreover hyperosmolarity seems to be a rather specific stimulus for p85 phosphorylation since other cellular stresses are not (or only very weak) inducers of this phenomenon.
To test the possibility that the tyrosine phosphorylation of p85 may be upstream of the phosphorylation of osmo-sensitive MAP kinases, we examined the pattern of phosphorylation of ERK-2, SAPKs (p54, p46), and p38 under conditions when osmolarity and cell volume were varied independently, exactly as described for p85 (compare Fig. 3 and Fig. 6). The phosphorylation of SAPKs and p38 were followed by phosphorylationspecific antibodies. The stimulation of ERK-2 correlated with cell shrinkage, however, the effect was clearly detactable only at strong volume reduction (Fig. 6, ERK-2, lane 6). The phosphorylation of SAPK p54 was very similar to that of p85: the trigger was the decreased volume and not the increased osmolarity (Fig. 6, SAPK, compare lanes 3 and 4) and the response was proportional to the reduction in cell volume (lanes 3, 5, and 6). SAPK p46 reacted in a similar manner but apparently it required a stronger shrinkage. In contrast, the behavior of p38 was sharply different: its phosphorylation was strongly stimulated both under iso-or hypertonic swelling (p38, lanes 2 and 4) and under hypertonic shrinkage (lanes 3, 5, and 6). In conclusion, p85 tyrosine phosphorylation might be either upstream of SAPK (and perhaps ERK) activation, or it may be a consequence of an early event necessary for the activation of these MAP kinases as well. The activation of p38 does not correlate with p85 phosphorylation.
Interaction of p85 and Other Osmo-sensitive Proteins with SH2 and SH3 Domains of Grb2-Interactions between tyrosine-phosphorylated proteins and adapter molecules (e.g. Grb2) constitute an important step in numerous signal-transducing pathways (47). Besides providing information about the signaling route itself, investigation of these interactions may help to find characteristic structural domains on the osmo-sensitive proteins, and can be useful for the future purification attempts. Therefore we tested whether hyperosmotic stress could induce a detectable change in the pattern of tyrosine-phosphorylated proteins that complex with Grb2, or certain domains of this adapter molecule. For these experiments fusion proteins were used which contained glutathione S-transferase (GST) linked to either the full-length Grb2 sequence (consisting of one SH2 domain and two SH3 domains) or to certain parts of it. Three mutant Grb2 proteins were applied: the first possessed only a functional SH2 domain (double SH3 mutant), the second had only the N-terminal SH3, whereas the third had only the Cterminal SH3 domain. Lysates of iso-or hypertonically-treated cells were incubated with the appropriate fusion protein immobilized on glutathione-agarose beads. The material precipitated by the beads was subjected to electrophoresis and blotted with anti-phosphotyrosine antibodies. As shown on Fig. 7, the whole Grb2 molecule could precipitate multiple phosphotyrosine-containing proteins (lanes Grb2). The association of these to the beads was due to the presence of Grb2 since no proteins were isolated with beads covered only with GST (not shown). Comparing the pattern obtained from iso-or hypertonically-treated samples showed that hyperosmosis induced an approximately 4-fold increase in the content of a Grb2precipitable, tyrosine-phosphorylated protein of 85-kDa. The location of this band corresponded perfectly to the osmo-sensitive band observed in whole cell lysates. Using Grb2 possessing only a functional SH2 domain gave similar results except that the difference at the level of p85 was even more pronounced and other hypertonicity-related bands also became apparent with molecular masses of Ϸ100 and Ϸ120 kDa (see lanes SH2). It is noteworthy that an increase in the signal in these regions (especially at 120 kDa) was often observed also in whole cell lysates (see e.g. Fig. 1A) but these proteins, being just below a huge, phosphotyrosine-containing, and osmotically not respon- FIG. 5. Tyrosine phosphorylation of p85 is induced by osmotic shock but not by other cellular stresses. Cells were preincubated in iso-sodium for 10 min followed by incubation in iso-sodium (I) or hypersodium (H) supplemented with 10 g/ml anisomycin (Aniso) or 200 M H 2 O 2 (H 2 O 2 ) for 10 min. Heat shock (lanes marked as 42°C) was induced by incubation of cells at 42°C in iso-sodium medium (I) for 30 min or in iso-sodium for the first 20 min and hyper-sodium for an additional 10 min (H). None of these treatments resulted in detachment of cells from the plate as controlled by microscopic inspection at the end of the incubation period. Tyrosine phosphorylation of p85 was detected as in Fig. 1. The p85 phosphorylation in iso-or hypertonically-treated control samples which were not challenged by any pretreatment was not different from the pattern observed in the presence of anisomycin, heat shock, and H 2 O 2 (not shown).
FIG. 6. Differential effect of changes in cell volume and osmolarity on the phosphorylation of ERK-2, SAPK, and p38. Cells were kept for 10 min in iso-potassium medium and then treated as indicated in the table. The same experimental conditions were used as on Fig. 3A. The resultant changes in osmolarity and/or volume are shown in the last two rows of the table. Whole cell lysates were subjected to electrophoresis, blotted onto nitrocellulose, and probed with anti-ERK-2 antibody, or with antibodies against the phosphorylated forms of SAPK or p38. Equal loading was confirmed by using the corresponding, nonphosphorylation-specific antibodies against these kinases (not shown). The upper and lower arrows on the SAPK blot indicate p54 and p46 SAPK, respectively. The nonspecific band at ϳ50 kDa is not osmotically sensitive and was present under each condition. sive band, remained partly hidden. 2 While these molecules, presumably through their phosphorylated tyrosines, were able to bind to the SH2 domains of Grb2, it is conceivable that they can interact also with other parts of the adapter in a phosphorylation-independent manner. In fact, the N-SH3 domain of Grb2 could precipitate out both the 85-kDa and the Ϸ120 kDa bands (Fig. 7, lanes N-SH3). These proteins likely formed a complex with N-SH3 Grb2 also in the lysates of isotonicallytreated cells, but, not being phosphorylated, they could not be visualized. Surprisingly, a different pattern was seen when we applied C-SH3 Grb2. This domain could still bind to the higher molecular weight band but could not precipitate out the 85-kDa protein (lanes C-SH3).
These results show that, at least in vitro, three osmotically responsive proteins can bind to Grb2. Two of these proteins can probably form a complex with Grb2 by two distinct mechanisms involving both the SH2 and SH3 domains of the adapter. Currently it cannot be decided whether this complex formation occurs by direct binding of the respective osmo-sensitive phosphoproteins to the various domains of Grb2 or through participation of other connecting proteins. Nevertheless a possible interpretation of these results is that p85 and p120 might contain proline-rich motifs through which they can associate with SH3 domains. A certain conclusion is that binding of p85 and p120 to Grb2 occurs independently: while the binding of p85 is lost if only the C-SH3 domain is intact, this alteration does not affect the association of p120.
The regulatory subunit of phosphoinositide 3-kinase (PI 3-kinase) is an 85-kDa protein which can be tyrosine phosphorylated and was shown to interact with Grb2 (48). To establish whether the osmo-sensitive p85 could be identical with PI 3-kinase, we immunoprecipitated the regulatory subunit from the lysates of iso-and hypertonically-treated cells and probed these precipitates with anti-phosphotyrosine antibodies. Absolutely no labeling occurred in the precipitates obtained from control or osmotically challenged cells. The presence of PI 3-kinase in these samples was verified by probing the precipitates with anti-PI 3-kinase (Fig. 8A, left panels). Importantly, in aliquots of total cell lysates, which contained comparable amounts of PI 3-kinase as the precipitates, the hypertonicityinduced tyrosine phosphorylation was readily detectable (Fig.   8A, right panels). This finding indicates that the absence of the phosphotyrosine signal in the precipitates could not be accounted for by the insufficient immunoprecipitation of PI 3-kinase. To further investigate whether PI 3-kinase and the osmosensitive p85 are in fact distinct, we compared the binding pattern of these proteins to the various domains of Grb-2 (compare Fig. 8B and Fig. 7). PI 3-kinase could bind to both the Nand the C-terminal SH3 domains of the adapter (in agreement with earlier reports (48)), while, as described above, p85 did not complex with the C-SH3 domain. Furthermore, the SH2 domain-containing beads, which bound p85 in lysates of hypertonically-treated cells, did not precipitate out PI 3-kinase from these samples. Taken together, these findings argue against the possibility that the osmo-sensitive p85 is the regulatory subunit of PI 3-kinase. DISCUSSION The present study demonstrates that an increase in extracellular osmotic activity induces characteristic tyrosine phosphorylations in at least three proteins (with apparent molecular masses of approximately 42, 85, and 120 kDa) of CHO cells (Fig. 1). This effect is most pronounced in an 85-kDa protein whose response is quick (detectable within 1 min), reversible, proportional to the applied osmotic concentration and occurs already at mild hypertonicity (plus 50 mosM) (Fig. 2). All these features render the tyrosine phosphorylation of p85 a sensitive and accurate indicator of the decrease in cellular hydration. We have provided evidence that the triggering factor is not an increase in extra-or intracellular osmotic concentration per se, but rather a decrease in cell volume. Cell shrinkage seems to be FIG. 7. Interaction of proteins tyrosine phosphorylated in response to osmotic shock with Grb2-GST fusion proteins. Lysates from cells which have been exposed for 10 min to iso-or hyperosmotic (600 mosM) media (marked as I and H, respectively) were centrifuged to remove Triton X-100-insoluble material and the supernatant was incubated for 1 h with agarose resin-coupled Grb2-GST proteins containing full-length Grb2 (left panel), or different mutants possessing only one functional domain of the adapter (labeled as SH2, N-SH3, and C-SH3). Beads were sedimented, washed several times, and the precipitated proteins were subjected to Western blotting with anti-phosphotyrosine antibody.
FIG. 8. Distinctive properties of the regulatory subunit of PI 3-kinase and the osmo-sensitive p85 protein. A, left panel: PI 3-kinase was immunoprecipitated from lysates of iso-or hypertonicallytreated samples and the precipitates were probed with anti-phosphotyrosine (upper lanes), and after stripping, with anti-PI 3-kinase (lower lanes). Right panel, Western blots of total cell lysates were probed with anti-phosphotyrosine or anti-PI 3-kinase. The amount of cell lysates were selected to give a comparable PI 3-kinase signal as observed in the immunoprecipitates. Samples from the precipitates and the lysates were probed with the same antibodies and developed in parallel, for the same time, on the same film. B, lysates from iso-(I) or hypertonically (H)-challenged cells were incubated with beads coupled to various domains of Grb2 and processed as described under Fig. 7. Precipitated proteins were subjected to Western blotting and probed with anti-PI 3-kinase. a prerequisite for the process, and it is equally effective, irrespectively whether it occurs under iso-or hyperosmotic conditions. Changes in Cl Ϫ and K ϩ concentration are thought to regulate kinases and were reported to affect the phosphorylation of ion channels (18) and the Na ϩ /K ϩ /Cl Ϫ cotransporter (14,15). In contrast, the shrinkage-induced rise in the cytosolic concentration of these ions is certainly not responsible for the observed tyrosine phosphorylations (Fig. 3). Pharmacological interference with tubulin assembly or with actin polymerization did not inhibit the shrinkage-dependent tyrosine phosphorylation (data not shown), suggesting that the integrity of these systems is not an absolute requirement for this response. It awaits elucidation whether the initializing volume-dependent event is the increase of a yet unidentified cytoplasmic regulator and/or the occurrence of specific protein-protein interactions due to macromolecular crowding (49).
Hyperosmolar environment was found to stimulate several subgroups of the MAP kinase superfamily including ERKs, SAPKs, and p38 (43). The tyrosine-phosphorylated 42-kDa protein is very likely to be ERK-2 since this band co-migrated with the anti-ERK-2 reactive protein and ERK-2 became indeed phosphorylated under our conditions (Figs. 4 and 6).
The molecular identity of p85, the most osmo-responsive protein, remains unresolved. Nevertheless our functional data obtained by using activators and inhibitors of various signaling pathways (Figs. 4 and 5) together with the structural information regarding the interaction of this band with different domains of Grb2 (Figs. 7 and 8) provide a basis to relate p85 to identified members of tonicity-responsive kinase cascades and to narrow down the number of potential candidates. The phosphorylation of p85 is not downstream to the activation of PKC or ERK-2, since the prevention of the hyperosmosis-induced ERK activation by PKC depletion or by inhibition of MEK leaves the tyrosine phosphorylation of p85 intact. Similarly, p85 phosphorylation is not a consequence of SAPK stimulation, because activators of these kinases (e.g. anisomycin, heat shock, or oxidative stress) fail to provoke the effect. ERK-5, another osmo-sensitive MAP kinase contains unique prolinerich regions (50,51) which might be able to associate with SH3 domains. On the other hand several findings argue against ERK-5 being the 85-kDa tyrosine-phosphorylated protein.
First, the apparent molecular mass of ERK-5 was found to be 110 kDa (30). Second, an anti-PAN-ERK antibody raised against a region of ERK-2 which has 60% identity with the corresponding portion of ERK-5, and which was shown to react with a high molecular weight MAP kinase, failed to react with any band Ͼ60 kDa in our cell lysates. Third, ERK-5 was described as a redox-sensitive kinase, whereas oxidative stress was ineffective in eliciting p85 tyrosine phosphorylation. MAP kinase kinases (i.e. MEKs, SEK1, and MKK3) which constitute the next level of these pathways and might autophosphorylate on tyrosines can also be excluded simply on the basis of their size. Finally, we have shown that p85 is probably not identical with another important signaling enzyme of similar molecular mass, the regulatory subunit of PI 3-kinase (Fig. 8).
In the light of all these considerations we suggest that the tyrosine phosphorylation of p85 likely represents a hitherto unrecognized, cell shrinkage-dependent, early event. This reaction may be one of the upstream mechanisms through which extracellular physical parameters are converted to intracellular biochemical signals.
While in the plasma membrane of yeast cells a two-component osmosensor system that regulates MAP kinases have been identified (20,21), the initial steps of the hypertonicity-related signal transduction in mammalian cells are largely unknown. Emerging evidence, however, suggests that tyrosine phospho-rylations might in fact participate in this proximal stage. During the coarse of our studies two closely related and newly discovered tyrosine kinases, ACK and PYK2, were reported to be activated and tyrosine phosphorylated by hypertonicity (52,53). Remarkably the activity of PYK2 correlated with SAPK activation (53) giving credence to the hypothesis that this tyrosine kinase might be involved in the regulation of this stress pathway. PYK2 might stimulate the small GTPase-dependent serine/threonine kinase PAKs which in turn may activate the MEKK1/SEK1/SAPK pathway (54). We believe that the 120-kDa osmo-sensitive protein of CHO cells could be an ACK/ PYK2 homologue. This assumption is supported by the corresponding molecular mass, and by the observation that binding of p120 to various domains of Grb2 is similar to the reported interaction between ACK and the adapter (52).
Finally we would like to consider the potential functional consequences of hypertonicity-provoked phosphorylations. The housekeeping Na ϩ /H ϩ exchanger isoform NHE-1 could be strongly stimulated by hypertonic exposure without the direct phosphorylation of the antiporter (55). According to a recent paper (56) ERK may be involved in the thrombin-induced activation of NHE-1 in platelets. In contrast, we found that neither the inhibition of PKC by staurosporine nor the addition of the MEK-inhibitor PD 098059 (i.e. two conditions that abolished the hypertonicity-provoked ERK activation) could prevent the shrinkage-induced stimulation of NHE-1 in CHO cells. 3,4 Our next studies are directed to test whether the shrinkage-induced tyrosine phosphorylations may play a role in the regulation of various NHE isoforms.