Phosphorylation of protein-tyrosine phosphatase PTP-1B on identical sites suggests activation of a common signaling pathway during mitosis and stress response in mammalian cells.

PTP-1B is a widely expressed non-transmembrane tyrosine-specific phosphatase. Previous studies indicated that, at mitosis, PTP-1B undergoes phosphorylation on two sites, 352Ser-Pro-Leu-Asn and 386Ser-Pro-Ala-Lys. Although the Ser-386 site can be phosphorylated by Cyclin B/Cdc2 in vitro, the kinase for the Ser-352 site is unknown. We have found that these phosphorylation events are not unique to normal mitosis. Instead, treatment with many, but not all, stress stimuli, in particular osmotic shock and certain phosphatase and protein synthesis inhibitors, leads to phosphorylation of PTP-1B. Tryptic phosphopeptide and mutant analysis reveals that, as in mitosis, stress-induced PTP-1B phosphorylation involves both Ser-352 and Ser-386. Activation of the proline-directed kinases Erk1/2, JNKs, and p38 was neither necessary nor sufficient for stress-induced PTP-1B phosphorylation. Our data suggest the existence of a novel mitogen-activated protein kinase pathway in mammalian cells, which is activated at mitosis and in response to osmotic shock and other stresses and results in PTP-1B phosphorylation. This pathway may be similar to the recently described Spc1/Sty1 pathway in Schizosaccharomyces pombe.

provided by Dr. John Lee (SmithKline-Beecham). PTP-1B-specific rabbit antibodies have been described (3), phosphospecific antibodies against MAP kinases were from New England Biolabs, and monoclonal anti-HA antibodies were from Boehringer Mannheim.

SDS-PAGE and Immunoblotting
PTP-1B electrophoretic mobility was determined by immunoblotting of total cell lysates (10 -25 g) resolved by 10% SDS-PAGE, as described previously (3). Erk1/Erk2 activation was assessed by immunoblotting with phosphospecific antibodies using conditions recommended by the manufacturer. Protein concentrations were determined by the Bradford method (18).

Cell Stimulations and Inhibitor Studies
HeLa and 293 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% calf serum (HeLa) or 10% fetal calf serum (293), 2 mM L-glutamine, and penicillin/streptomycin. Confluent HeLa cells were treated with various stress-inducing chemicals or physical agents as described in Fig. 1A. UV irradiation was performed using a Stratalinker (Stratagene). For heat shock, plastic dishes with cell monolayers were sealed with Parafilm and transferred to a 45°C water bath for 25 min, followed by a 60-min recovery period at 37°C. For hypo-osmotic shock, cells were incubated for 40 min in growth media diluted 1:2 or 1:4 with water.
To assess the requirement for Erk1/2 or p38 in osmotic shock-induced PTP-1B phosphorylation, cells were pretreated for 30 min with MEK or p38 inhibitors at the indicated concentrations, and then exposed to 1 M sorbitol for 1 h, washed with phosphate-buffered saline (PBS), and lysed in the continuous presence of the appropriate inhibitor, as described (10,19,20).

Cell Cycle Synchronization
Nocodazole-induced Mitotic Arrest-HeLa cells were synchronized in pro-metaphase by the addition of nocodazole (40 ng/ml) to the culture media for 18 h. Round, mitotically arrested cells (30 -50% of the cells) were washed off the plate, washed twice with PBS, and either lysed or fixed for flow cytometric analysis (see below).
Synchronization at the G 1 /S Border Was Performed by Thymidine-Aphidicholin Block (21)-Mitotic cells began to appear 10 -11 h following release from the second block and, by 12-13 h post-release, constituted 25-50% of all cells.

Flow Cytometric Analysis of Cell Cycle Distribution
Cells that remained attached to plates after nocodazole treatment/ wash-off were detached by incubation for 10 min in PBS containing 2.5 mM EDTA. These cells and the mitotic wash-off cells were washed twice with ice-cold PBS, fixed in 70% absolute ethanol (Ϫ20°C), and stored at 4°C. Immediately before analysis, cells were stained with DAPI (1 g/ml) for 20 min at room temperature, washed twice with PBS, and analyzed (Becton-Dickinson flow cytometer with FACStar PLUS Software (version 2.01)).

Immunoprecipitations, Phosphoamino Acid and Phosphotryptic Analysis
Metabolic labeling with H 3 32 PO 4 and PTP-1B immunoprecipitations were performed as described (3). Nocodazole, sorbitol, or calyculin A was included in the labeling media, where indicated. Phosphoamino acid analysis was performed by two-dimensional thin layer electrophoresis (22). Tryptic phosphopeptides were resolved by thin layer electrophoresis at pH 1.9 followed by ascending chromatography, as described (4,22).

Transient Transfection Analysis
Vectors directing the expression of HA-tagged wild-type PTP-1B and its Ser-352 3 Ala mutant (23) were introduced into 293 cells by the calcium phosphate method as described (3). Forty eight hours posttransfection, cells were treated with 1 M sorbitol for 1 h and PTP-1B mobility was determined.

RESULTS AND DISCUSSION
Since both seryl residues of PTP-1B that are phosphorylated during mitosis are followed by a proline residue (4), we suspected that a proline-directed kinase(s) catalyzes their phosphorylation. Preliminary studies, together with the work of Flint et al. (4), suggested that Cyclin/Cdks were not responsible for Ser-352 phosphorylation (data not shown). Many MAP kinase family members are activated by environmental stresses (10,11,24,(27)(28)(29)(30). Therefore, we asked whether various stress stimuli also evoked mitotic-like PTP-1B phosphorylation. Phosphorylation of Ser-352 results in an approximately 2-kDa decrease in PTP-1B electrophoretic mobility (4,5). Interestingly, several treatments reported to activate JNKs and p38 (9,11,24,28,31) resulted in appearance of the slower migrating (upper) PTP-1B isoform (Fig. 1A). Particularly good induction of this isoform was achieved by subjecting cells to osmotic shock. Induction of the upper isoform was dependent on osmolarity ( Fig. 1C) and reached a maximum (approximately 50% shift) by 30 -45 min of treatment (Fig. 1D). The Ser/Thr phosphatase inhibitor calyculin A, which is a potent activator of stress-activated kinases (data not shown), also was a good inducer of the upper isoform. However, other treatments that activate stress-activated kinases induced this isoform to either a lower extent (e.g. arsenite, heat shock, anisomycin) or not at all (e.g. tunicamycin, cycloheximide, emetine, or puromycin).
In other studies (see below), we observed that nocodazole treatment activates p38 in HeLa cells. This observation, together with the above results, made it important to verify that PTP-1B phosphorylation in nocodazole-treated cells is truly a mitotic event, rather than an artifact of cellular stress produced by drug treatment. Therefore, we asked whether enrichment for mitotic HeLa cells using other synchronization methods also resulted in appearance of the upper isoform. We were unable to obtain sufficient numbers of mitotic cells for biochemical studies from randomly growing cultures using the "shakeoff" method (32). Instead, we presynchronized the culture in early S phase by thymidine/aphidicolin block followed by re-lease into the cell cycle for various periods of time and wash-off to obtain G 2 /M-enriched populations (see "Experimental Procedures"). The upper isoform of PTP-1B, characteristic of nocodazole-arrested cells (lane Noc, Fig. 1B), also appeared 14 -15 h after release from S phase block, which correlated with the appearance of round, mitotic cells. Similar results were obtained when cells were synchronized in S phase by double thymidine block and released (data not shown). These results strongly suggest that PTP-1B phosphorylation is a bona fide mitotic event.
We next confirmed that the stress-induced PTP-1B mobility shift was caused by phosphorylation, as has been reported for the mobility shift in mitotic cells (4,5). Lysates of sorbitoltreated cells were treated with various amounts of calf intestinal alkaline phosphatase ( Fig. 2A). Phosphatase treatment reversed the osmotic shock-induced decrease in PTP-1B mobility. The action of alkaline phosphatase was prevented by inclusion of phosphatase inhibitors, indicating that it was a consequence of dephosphorylation rather than protease contamination. Phosphoamino acid analysis revealed that PTP-1B from osmotically shocked cells was phosphorylated exclusively on seryl residues (Fig. 2B), as is the case for mitotic PTP-1B (4, 5). 2 We also examined whether the osmotic shock-and mitosisinduced PTP-1B mobility shifts result from phosphorylation of the same residue, Ser-352. 293 cells were transfected with plasmids expressing either HA-tagged wild type PTP-1B (HA-PTP-1B) or an HA-tagged Ser-352 3 Ala mutant of PTP-1B (HA-PTP-1BS 3 A), and, following osmotic shock, the mobility of the respective HA-tagged proteins, as well as that of endogenous PTP-1B, was analyzed by immunoblotting (Fig. 2C). As expected, osmotic shock induced a decrease in the electrophoretic mobility of both endogenous PTP-1B (compare PTP-1B and phospho-PTP-1B (pPTP-1B) and the HA-tagged wild type protein (HA-PTP-1B) and phospho-HA-PTP-1B (pHA-PTP-1B)). Mutation of Ser-352 eliminated the osmotic shock-induced mobility shift. These results strongly suggest that, as in To confirm these observations, as well as to assess whether stress induces phosphorylation of other residues in PTP-1B, we performed phosphopeptide mapping studies. Osmotic shock (sorbitol) or calyculin A treatment produced 2 major tryptic phosphopeptides (labeled 1 and 2 on Fig. 2D), which co-migrated with the 2 major phosphopeptides in mitotic PTP-1B.
These peptides have been shown previously (4) to contain Ser-386 (spot 1) and Ser-352 (spot 2). A third peptide (spot 3), largely absent in PTP-1B from mitotic cells but strongly phosphorylated in sorbitol-and calyculin A-treated cells, probably contains Ser-378, a protein kinase C recognition site (4). Thus, during mitosis and osmotic shock, PTP-1B becomes phosphorylated on both Ser-352 and Ser-386. Since Ser-352 and Ser-386 are each phosphorylated under both conditions, it is possible that they are targets of the same kinase. Randomly growing cells display a distinctly different PTP-1B phosphorylation pattern (Fig. 2D, Control), which does not include Ser-352 or Ser-386. Moreover, inhibition of seryl phosphatases with calyculin A does not enhance the phosphorylation pattern characteristic of randomly growing cells, but instead evokes new phosphorylation events (e.g. spots 1 and 2), yielding the pattern characteristic of mitotic and osmotically shocked cells.
These results suggest that calyculin A (and, by inference, osmotic shock and mitosis) promotes PTP-1B phosphorylation not simply by inhibiting a PTP-1B specific phosphatase(s) but by activating a PTP-1B-specific kinase. This idea is in agreement with the ability of calyculin A to activate p38 and JNKs (data not shown), as well as with the ability of another Ser-Thr phosphatase inhibitor, okadaic acid, to induce a similar pattern of protein phosphorylation as TNF and IL-1 (33). Moreover, in preliminary studies, we have detected an increase in a kinase activity that phosphorylates PTP-1B on Ser-352 in membranes from osmotically shocked, calyculin A-treated, or mitotically arrested cells (data not shown).
The stress-activated kinase p38 also is an unlikely candidate for the PTP-1B kinase. First, activation of p38 was insufficient to cause PTP-1B phosphorylation. We observed that p38 becomes activated in nocodazole-treated HeLa cells (Fig. 3C, top  panel). After 18 h of nocodazole treatment, about 50% of the HeLa cells became round and could be easily washed from the plates. These cells (noc. round cells in Fig. 3C) were mitotically arrested, as evinced by their 4n DNA content (Fig. 3C, bottom  panel) and high level of Cyclin B/Cdc2 H1 kinase activity (data not shown). The remaining, adherent cells were in different stages of the cell cycle, as indicated by their DNA content similar to that of randomly growing cells (Fig. 3C, bottom  panel) and their low levels of Cyclin B/Cdc2 H1 kinase activity (data not shown). However, both populations of nocodazoletreated cells had very similar levels of (elevated) p38 activity (Fig. 3C, top panel). These results suggest that nocodazole treatment causes activation of p38 in HeLa cells independently of mitotic arrest. More importantly, though, Ser-352-phosphorylated PTP-1B was present only in nocodazole-treated mitotic cells (Fig. 3C, middle panel). Therefore, activation of p38 per se is not sufficient to cause Ser-352 phosphorylation. Moreover, p38 activation is not necessary for Ser-352 phosphorylation, as treatment with the p38-specific inhibitor SB 203580 at concentrations known to inhibit p38-mediated events in vivo (10, 19, 20, 38) did not affect osmotic shock-induced phosphorylation of PTP-1B (Fig. 3B).

FIG. 2. Osmotic shock-induced PTP-1B mobility shift results from Ser-352 phosphorylation.
A, PTP-1B mobility shift is a consequence of phosphorylation. Following osmotic shock with sorbitol, HeLa lysates were treated with the indicated amounts of calf intestinal alkaline phosphatase in the absence or presence of phosphatase inhibitors (50 mM NaF and 10 mM Na 3 VO 4 ), and PTP-1B mobility was analyzed by immunoblotting. B, phosphoamino acid analysis of PTP-1B from osmotically shocked HeLa cells. Cells were labeled with 32 P-orthophosphate during osmotic shock, and phosphoamino acid analysis of immunoprecipitated PTP-1B was performed. C, mutation of Ser-352 eliminates osmotic shock-induced PTP-1B mobility shift. 293 cells were transfected with empty vector or vectors expressing either HA-tagged wild type PTP-1B or HA-tagged Ser-352 3 Ala PTP-1B. Transfected cells were subjected to osmotic shock (OS), and electrophoretic mobilities of HAtagged and endogenous PTP-1B were analyzed by immunoblotting with anti-HA or anti-PTP-1B antibodies. pPTP-1B and pHA-PTP-1B denote the phosphorylated, slower migrating forms of the corresponding molecules. Note that endogenous pPTP-1B and exogenous, unphosphorylated HA-PTP-1B comigrate. D, phosphopeptide maps of PTP-1B isolated from mitotic (Mitosis), osmotically shocked (Sorbitol), calyculin A-treated (Caly A), and control (Control) HeLa cells. PTP-1B immunoprecipitated from [ 32 P]orthophosphate-labeled control (randomly growing) cells or cells treated with nocodazole, sorbitol, or calyculin A, was subjected to two-dimensional phosphotryptic analysis. The directions of electrophoretic and chromatographic separations are indicated. Numbered spots denote specific peptides (see text for details).
It also is unlikely that a JNK is the PTP-1B kinase. The levels of activation of JNKs caused by various stress treatments did not correlate with the ability of these treatments to induce the upper isoform of PTP-1B (Fig. 3D). For example, anisomycin and cycloheximide both evoke strong activation of JNKs, but only anisomycin treatment leads to significant PTP-1B phosphorylation. Conversely, the level of PTP-1B phosphorylation is comparable in TNF and anisomycin-treated cells, but these treatments result in dramatically different levels of JNK activity. JNKs are activated by treatment with IL-1 or peroxide, yet these agents have little effect on Ser-352 phosphorylation. Finally, in mitotic cells, Ser-352 phosphorylation of PTP-1B occurs in the absence of detectable JNK activation (Fig. 3C). Therefore, activation of JNKs is neither sufficient nor necessary for PTP-1B phosphorylation.
Our data demonstrate that PTP-1B is phosphorylated on the same sites, Ser-352 and Ser-386, during both mitosis and the response to osmotic shock and other stresses. This places PTP-1B at a point of convergence of seemingly unrelated signaling pathways (Fig. 4). Although we have no direct data indicating that the same kinase phosphorylates PTP-1B under these conditions, indirect evidence suggests that this is the case. The PTP-1B phosphorylating activity from both mitotic and stress-treated cells was associated with crude membranes. Moreover, this kinase activity (from both mitotic and stresstreated cells) is inactivated by multiple non-ionic detergents; indeed, we have not been able to establish conditions for stably solubilizing this activity. In addition, these activities phospho-rylate both Ser-352 and Ser 386 in vitro (data not shown). These properties distinguish the PTP-1B kinase activity from known members of the MAPK family, as well as from multiple other serine/threonine kinases, including Cyclin/Cdks, cAMPdependent protein kinase, and protein kinase C. Studies in fission yeast (see introduction) provide a precedent for the involvement of the same MAP kinase family member, the S. pombe homolog of p38, in both osmotic stress response and control of mitotic entry (7,8). Although human p38 does not phosphorylate PTP-1B in vitro (data not shown) and its inhibition does not affect phosphorylation of PTP-1B in vivo (Fig.  3B), we favor the hypothesis that another MAP kinase analog, activated in response to both stress and mitosis, is responsible for PTP-1B phosphorylation (Fig. 4). Validation of this hypoth- esis will require identifying the kinase that phosphorylates Ser-352 of PTP-1B at mitosis and demonstrating that it is identical to the PTP-1B kinase activated in response to stress treatments. Identification of this kinase could then provide a means for uncovering the mammalian pathway analogous to that in S. pombe.
The biological consequences of PTP-1B phosphorylation for the stress response, as well as its mitotic role, remain to be defined. It is possible that PTP-1B functions analogously to Pyp1 or Pyp2 to inactivate stress-and mitosis-induced MAP kinases. However, it should be noted that when introduced into S. pombe (39), PTP-1B behaves more like Pyp3, a mitotic entry stimulator (40), than the mitotic inhibitors Pyp1 or Pyp2 (8,14,15). The effect of phosphorylation on PTP-1B enzymatic activity is controversial. Flint et al. (4) reported a 30% decrease in PTP-1B activity at mitosis, whereas Schievella et al. (5) observed a slight increase. This difference could be due to differences in PTP assays (most likely the use of different substrates). Since PTP-1B constitutes about 0.1% of total cellular protein in HeLa cells (41), even a 30% decrease in its activity might result in a significant increase in Tyr(P) content of cellular proteins. Thus, the increase in tyrosine phosphorylation in response to UV irradiation (42) and to various other stresses, as well as during mitosis, 3 may result not only from activation of various PTKs (43) but also from inactivation of PTP-1B and perhaps other PTPs. Interestingly, several proteins of similar molecular weight become phosphorylated on Tyr during mitosis and stress-response. 2 The identities of these proteins, which could be PTP-1B targets, are currently under investigation.