Signalling pathways regulating the dephosphorylation of Ser729 in the hydrophobic domain of protein kinase Cepsilon upon cell passage.

We have recently demonstrated that in quiescent fibroblasts protein kinase C (PKC) epsilon(95) is phosphorylated at Ser(729), Ser(703), and Thr(566) and that upon passage of quiescent cells phosphorylation at Ser(729) is lost, giving rise to PKCepsilon(87). Ser(729) may be rephosphorylated later, suggesting cycling between PKCepsilon(87) and PKCepsilon(95). Here we show that the dephosphorylation at Ser(729) is insensitive to okadaic acid, calyculin, ascomycin C, and cyclosporin A, suggesting that dephosphorylation at this site is not mediated through protein phosphatases 1, 2A or 2B. We demonstrate that this dephosphorylation at Ser(729) requires serum and cell readhesion and is sensitive to rapamycin, PD98059, chelerythrine, and Ro-31-8220. These results suggest that the phosphorylation status of Ser(729) in the hydrophobic domain at Ser(729) is regulated independently of the phosphorylation status of other sites in PKCepsilon, by a mTOR-sensitive phosphatase. The mitogen-activated protein kinase pathway and PKC are also implicated in regulating the dephosphorylation at Ser(729).

The PKC 1 family of related phospholipid-dependent serine/ threonine kinases are involved in the control of many cellular processes, including cell growth and differentiation (1,2). To date 11 PKC isoforms have been identified: the conventional PKCs (␣, ␤ I , ␤ II , and ␥), which are regulated by calcium and diacylglycerol, the novel PKCs (␦, ⑀, and ), which are calcium independent but dependent upon diacylglycerol, and the atypical PKCs (, , and ), which are both diacylglycerol-and calcium-independent. Another isoform, PKC, is known as protein kinase D (3).
Recent work from many groups has highlighted the importance of phosphorylation in the regulation of PKC activity (4 -7). Initially phospholipid-dependent kinase 1 phosphorylates a conserved site on the lip of the catalytic region that corresponds to Thr 566 in PKC⑀ (8,9). Phosphorylation at this site is important for the enzymatic activity of PKC (4). Two further phosphorylation sites have been identified in the Cterminal region of the enzyme at the turn and hydrophobic motifs (4 -6). Phosphorylation at the turn motif (Ser 703 in PKC⑀) is believed to be mediated through autophosphorylation (5). There is some debate as to whether the hydrophobic site (Ser 729 in PKC⑀) becomes phosphorylated as a result of autophosphorylation or by a separate kinase. Phosphorylation at this hydrophobic site may be modulated by PKC and appears to be sensitive to rapamycin (10,11). Although phosphorylation at these two C-terminal sites is not essential for the catalytic activity of PKC, they seem to regulate the stability of the enzyme (12)(13)(14)(15)(16).
PKC⑀ is the only isoform that has oncogenic potential (17,18) that may be mediated through its interaction with Raf 1 kinase (19,20). PKC⑀ is also unique in having actin and Golgi-binding domains (21)(22)(23)(24)(25). PKC⑀ from fibroblasts migrates on SDS-PAGE as two distinct forms, with molecular sizes of 95 and 87 kDa (PKC⑀ 95 and PKC⑀ 87 ) that differ in their intracellular localization. In quiescent cells the PKC⑀ 95 form predominates, whereas after passage PKC⑀ 87 becomes the major species. We have recently reported that these forms differ in their phosphorylation status at Ser 729 (26). Thr 566 and Ser 703 are phosphorylated in both these forms of PKC⑀, and the protein has complete N and C termini (26). The formation of PKC⑀ 87 upon cell passage is not due to new protein synthesis. We have therefore suggested that a phosphatase responsible for dephosphorylation of Ser 729 is activated upon cell passage. Removal of the Ser 729 phosphate from the hydrophobic domain may reduce the stability of the enzyme, rendering it more accessible to phosphatase attack and potentially to degradation (27). Alternatively, the change in localization of PKC⑀ on passage may make it accessible to a Ser 729 phosphatase. Therefore regulation of phosphorylation at Ser 729 may prove to be yet another level of control for PKC.
The regulation of how PKC becomes dephosphorylated has not been well studied. Most work has been carried out using TPA to induce activation that leads to dephosphorylation and degradation. Our system involving cell passage allows us to examine the control of the putative Ser 729 -specific phosphatase rather than to look at the complete dephosphorylation of the protein. Our results suggest that dephosphorylation of PKC⑀ is a two-stage process. A specific phosphatase removes the phosphate at Ser 729 followed by either the removal of other phosphate groups or rephosphorylation at Ser 729 and recycling of the enzyme to the 95-kDa form.
In this study we have examined how the dephosphorylation of PKC⑀ at Ser 729 in 3T3 and 3T6 fibroblasts is regulated. We present evidence that the dephosphorylation of Ser 729 is not mediated by protein phosphatases 1, 2A, and 2B but is dependent upon serum and cell readhesion, and that MAPK and mTOR (mammalian target of rapamycin) pathways are involved. PKC is also important in regulating the phosphorylation status of Ser 729 .

Materials
Cells were from the European Collection of Animal Cell Cultures (Porton Down, UK) and the William Dunn Cell Bank (Oxford, UK). Cell culture plastic was from Nunc (Life Technologies, Inc.). All other cell culture reagents were from Life Technologies, Inc. except serum, which was from PAA Laboratories (Linz, Austria). All other chemicals were from Sigma, and chemical inhibitors were from Calbiochem (Nottingham, UK) unless otherwise stated. Nitrocellulose membrane (Hybond C) was from Amersham Pharmacia Biotech. Dried milk powder was Marvel (Premier Beverages, Stafford, UK). BCA reagents were from Pierce.
The polyclonal PKC⑀ and PKC␦ antibodies used for Western blotting and immunoprecipitations were generated to the C-terminal peptide sequence by Professor N. Groome (Oxford Brookes University, UK) as previously described (26). Peroxidase-conjugated secondary antibodies were from Sigma.

Methods
Tissue Culture-3T3 and 3T6 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum in a humidified incubator at 5% CO 2 as described previously (26). Cells were passaged by rinsing with Tris saline, followed by release with Tris trypsin. Cells were resuspended in fresh medium as above, plated into fresh tissue culture flasks, and allowed to settle for 15 min before harvesting unless otherwise stated.
Inhibitor Studies-Cells were treated with inhibitors at concentrations described in the text. Inhibitors were dissolved in Me 2 SO unless otherwise stated and effects compared with Me 2 SO control treated cells. Solvent concentrations were never greater than 0.01%. Quiescent cells were treated for 30 min before passage, and fresh inhibitor was added after passage. PKC⑀ and PKC␦ translocation inhibitor and activator peptides coupled to anttenepaedia carrier protein were generously supplied by Professor D. Mochly-Rosen (Stanford, CA).
Plasmids and Cell Transfection-Hemagglutinin-tagged pCH3␦ RD was generously donated by Professor S. Jaken (University of Vermont). 3T3 and 3T6 cells were transiently transfected with Superfect (Qiagen) and were analyzed after 48 h, were quiescent, or were passaged. Transfection was monitored by Western blotting for the hemagglutinin tag. PKC␦ antisense was created through cloning the full-length PKC␦ into the pZeo SV vector (Invitrogen). Stable transfections were selected using Zeocin (Invitrogen), and individual colonies were expanded and analyzed by Western blotting. In all cases results were compared with mock transfected cells.
[ 35 S]Methionine Incorporation-Cells were incubated in methioninefree Dulbecco's modified Eagle's medium (Life Technologies, Inc.) for 1 h. 3.7 Mbq of [ 35 S]Met/Cys (Promix; Amersham Pharmacia Biotech) was then added for 2 h. Cells were then passaged into fresh medium and flasks, and 3.7 Mbq [ 35 S]Met/Cys was retained in the medium. Cells were harvested at different time points, and PKC⑀ were immunoprecipitated as described above. Samples were resolved by SDS-PAGE, and gels were fixed in isopropanol/acetic acid/H 2 0 (25/20/65, v/v) for 1 h and then incubated for 15 min in Amplify (Amersham Pharmacia Biotech) before drying and exposing to x-ray film. All data shown are typical of at least three independent experiments.

PKC⑀ 87 Is Formation upon Cell Passage Is Not the Result of
New Protein Synthesis-In quiescent cells PKC⑀ 95 predominates, whereas 15 min after passage into serum with readhesion PKC⑀ 87 becomes the major form (Fig. 1A). We have recently shown that PKC⑀ 95 and PKC⑀ 87 differ in their phosphorylation at Ser 729 and that the formation of PKC⑀ 87 is most probably the result of dephosphorylation of PKC⑀ 95 at Ser 729 (26). It is likely that the apparent increase in total PKC⑀ protein ( Fig. 1A) is the result of increased immunoreactivity of our antibody with PKC⑀ 87 compared with PKC⑀ 95 since the polyclonal PKC⑀ antibody used in these studies is raised against the C-terminal region of PKC⑀ (-NQEEFKGFSYF-GEDLMP), which includes Ser 729 . This observation applies to other Western blots. This is confirmed by [ 35 S]Met/Cys incorporation studies that reveal that no new PKC⑀ synthesis occurs within 15 min of passage although synthesis is detected 48 -72 h after passage (Fig. 1B). Western blots show that PKC⑀ 95 reappears and becomes the predominant PKC form within 1 h after cell passage (Fig. 1A). Because there is no synthesis of PKC⑀ over this time period (Fig. 1B), this suggests that PKC⑀ 95 detected within 1 h of passage is derived through the rephosphorylation of PKC⑀ 87 .
Serum and Readhesion Are Required for the Formation of PKC⑀ 87 upon Cell Passage-When quiescent cells are passaged into serum-free medium and allowed to adhere for 15 min, no formation of PKC⑀ 87 is observed (data not shown). However, when cells are passaged into increasing concentrations of serum, PKC⑀ 87 is formed in increasing amounts ( Fig. 2A). This result shows that some factor(s) in serum is/are necessary for the formation of PKC⑀ 87 upon cell passage. Fibroblasts at 70% confluency contain both PKC⑀ 95 and PKC⑀ 87 (Fig. 2B). When these cells are serum-starved for 24 h, to promote entry into G 0 (28), PKC⑀ 95 becomes the predominant form (Fig. 2B), as is observed in cells grown to confluency and quiescence ( Fig. 2A). Readdition of serum to serum-starved cells in G 0 does not promote the formation of PKC⑀ 87 , although passage of these cells does (Fig. 2B). This is also the case if fresh serum is added to quiescent fibroblasts (results not shown). These findings suggest that serum is necessary but not sufficient for the formation of PKC⑀ 87 upon cell passage.
Readhesion is also necessary for the formation of PKC⑀ 87 upon passage. This is shown by the finding that when quiescent fibroblasts are passaged in serum containing medium onto poly-HEME-coated plastic, or shaken to prevent adhesion, no formation of PKC⑀ 87 is observed (Fig. 2C). These results emphasize the need for both serum factors and readhesion in PKC⑀ 87 formation. We have confirmed through 3-[4,5-dimeth- ylthiazol-2-yl]-2,5-diphenyltetrazolium bromide analysis that 3T3 and 3T6 fibroblasts require both serum and readhesion to exit G 0 and proliferate (data not shown). It should be noted that trypsinization of fibroblasts does not stimulate PKC⑀ 87 formation (data not shown).
Because re-entry into the cell cycle from G 0 is necessary but not sufficient for PKC⑀ 87 formation (Fig. 2D), it is clear that other factors are also important in regulating the dephosphorylation of Ser 729 . One such factor may be the disruption of cell-substratum interactions and resulting changes in the cytoskeleton that occur upon cell passage and readhesion. Indeed, if the actin cytoskeleton is disrupted through cytochalasin D treatment of quiescent cells, PKC⑀ 87 is formed (Fig. 2D). The microtubule disrupting drug nocodazole does not produce the same effect (Fig. 2D).
PP1, PP2A, and PP2B Are Not Important in PKC⑀ 87 Formation upon Passage-Treatment of cells with okadaic acid (OA) and calyculin (PP1 and PP2A inhibitors (29)) or cyclosporin A and ascomycin (inhibitors of PP2B (29)) did not inhibit the formation of PKC⑀ 87 upon passage (Fig. 3) (ascomycin C data not shown). This suggests that these protein phosphatases are not involved in catalyzing the removal of the Ser 729 phosphate and that an OA-insensitive phosphatase is mediating this dephosphorylation. We observed that OA increased PKC⑀ 87 formation upon cell passage. It is possible then that inhibition of PP1 and/or PP2A increases the activity of the Ser 729 phosphatase. Alternatively, the inhibitors may be preventing further dephosphorylation of PKC⑀ 87 at Ser 703 and Thr 566 , thus preserving PKC⑀ 87 . In fact we occasionally detect a faster migrating PKC⑀ band (PKC⑀ 84 ); formation of PKC⑀ 84 is inhibited by OA, supporting the latter idea. Hansra et al. (30) have shown that TPA-induced dephosphorylation of PKC␣ could be only partially inhibited by OA and have speculated that another protein phosphatase, insensitive to OA, may be involved. These workers have also described the importance of vesicle transport in the dephosphorylation of PKC␣, showing that incubation at 18°C inhibited dephosphorylation (31). We find that incubation at 18°C to inhibit vesicle transport had no effect on the production of PKC⑀ 87 (Fig. 3). In fact, similarly to OA, it increased the formation of PKC⑀ 87 .
PKC⑀ 87 Production upon Passage Is Increased by Inhibiting Calpain, PI 3-Kinase, and PKC␦-Treatment of fibroblasts with the calpain and proteasome inhibitor ALLN (32) increased the level of PKC⑀ 87 produced upon cell passage (Fig. 4A). This may implicate calpain in the activation of a putative Ser 729 phosphatase, or, probably more likely, this reflects an inhibition of the degradation of PKC⑀ 87 . Treatment of cells with the PI 3-kinase inhibitors LY294002 and wortmannin (Refs. 33 and 34; wortmannin data not shown) also caused an increase in formation of PKC⑀ 87 upon cell passage (Fig. 4B). PI 3-kinase has already been demonstrated to play a role in PKC phosphorylation through its activation of phospholipid-dependent kinase 1 (7,8). Our findings also implicate PI 3-kinase in the control of PKC⑀ dephosphorylation and degradation.
Transfection of fibroblasts with full-length PKC␦ antisense reduced PKC␦ expression and increased PKC⑀ 87 formation upon cell passage (Fig. 4C). Expression of the regulatory domain of PKC␦ has been shown to be a specific inhibitor of PKC␦ (35,36). Fibroblasts transfected with a construct containing this domain showed increased formation of PKC⑀ 87 upon cell passage (Fig. 4D). PKC␦ translocation inhibitor and activator peptides were also used to modulate PKC␦ in fibroblasts (37,38). A PKC␦ translocation inhibitor peptide increased PKC⑀ 87 formation upon passage, whereas control and activator peptides had no effect (Fig. 4E). Peptide modulators of PKC⑀ had no effect on the formation of PKC⑀ 87 upon cell passage. Conformation of the specificity and activity of these peptides was confirmed through analysis of PKC isoform localization to membrane and cytosol fractions (data not shown).
The Production of PKC⑀ 87 upon Passage Involves MAPK, mTOR, and PKC-The production of PKC⑀ 87 upon passage is inhibited passaging cells in the presence of PD98059, a MEK inhibitor (Ref. 39 and Fig. 5A), rapamycin, an inhibitor of mTOR (Ref. 40 and Fig. 5B), and by chelerythrine or Ro-31-  Fig. 5B; Ro-31-8220 data not shown). Such findings suggest a role for the MAPK pathway, mTOR, and PKC in the control of PKC⑀ 87 formation on cell passage. Inhibitors of protein kinase A, p38, and tyrosine kinases had no effect on PKC⑀ 87 formation upon passage, suggesting that these pathways are not important in this process. The inhibitor data are summarized in Table I. TPA Stimulates PKC⑀ 87 Production in Quiescent Cells-TPA treatment of cells activates conventional and novel PKC isoforms, causing their dephosphorylation and subsequent degra-dation (30,31). When quiescent fibroblasts are treated with 250 nM TPA for 15 min, formation of PKC⑀ 87 is detected, although this is at a reduced level compared with cell passage (Fig. 5C). This TPA effect can be blocked with the PKC inhibitor chelerythrine and does not occur with the inactive 4␣phorbol (Fig. 5C). As with PKC⑀ 87 formation upon cell passage, this effect can be inhibited by preincubation with rapamycin and PD98059 (data not shown), suggesting that the MAPK and mTOR pathways are also important here. DISCUSSION We have previously shown that when fibroblasts are passaged there is a change in the phosphorylation status of PKC⑀ and that this is associated with a change in the intracellular localization of the protein. PKC⑀ 95 , predominant in quiescent cells, has a perinuclear localization and is phosphorylated at Thr 566 , Ser 703 , and Ser 729 . PKC⑀ 87 has a cytosolic distribution and is phosphorylated at Thr 566 and Ser 703 (26). We have suggested that on cell passage a PKC⑀ Ser 729 phosphatase may be activated. The hydrophobic (Ser 729 in PKC⑀) site is conserved in most PKCs and in the ACG group of kinases (10). It is therefore essential to understand how phosphorylation and dephosphorylation in the hydrophobic domain is controlled. Here we have examined some factors that are important in the control of Ser 729 phosphorylation status.
We have shown that the removal of the phosphate group at Ser 729 from PKC⑀ 95 requires both serum and adhesion. Fibroblasts do not proliferate in the absence of serum or when prevented from adhering (28), and it is therefore likely that PKC⑀ 87 formation is dependent upon re-entry of quiescent or serum-starved cells into the cell cycle. However, we have shown that re-entry into the cell cycle alone is not sufficient for this process because if cells are serum-starved and then restimulated with serum there is no formation of PKC⑀ 87 . It is possible then that PKC⑀ 87 formation is mediated through both disruption of cell-substratum interactions and readhesion of the cells to the substratum in the presence of serum. Our finding that cytochalasin D, the microfilament-disrupting drug, stimulates the formation of PKC⑀ 87 in quiescent cells supports this view. Readherence of cells after passage involves reorganization of the cytoskeleton. PKC⑀ has an actin-binding domain (21, 22), and we therefore speculate that PKC⑀ bound to actin may be activated by microfilament disruption, thereby either stimulating phosphatase activity or allowing access to a putative Ser 729 phosphatase. Alternatively, disruption of the actin cytoskeleton through passage or cytochalasin D treatment could alter the localization of PKC⑀, making it susceptible to attack by a Ser 729 phosphatase.
Our inhibitor studies clearly show that the formation of PKC⑀ 87 upon passage is not dependent upon the more well defined protein phosphatases, PP1, PP2A, and PP2C. However, we find that the formation of PKC⑀ 87 upon cell passage is inhibited by rapamycin, indicating involvement of mTOR. mTOR has recently been implicated in the control of phosphorylation of the hydrophobic site in PKC⑀ and PKC␦ (10,11). These authors speculate that this is mediated through the activation via mTOR of a phosphatase specific for the hydrophobic site rather than through regulation of a kinase. Our data presented here support this view and also suggest that this mTOR-controlled phosphatase is not PP1, PP2A, or PP2B.
Interestingly, OA treatment and also treatment with the calpain and proteasome inhibitor ALLN, increased the level of PKC⑀ 87 detected upon cell passage. The most likely explanation for these findings is that these inhibitors are blocking further dephosphorylation and degradation of PKC⑀ 87 , thereby increasing PKC⑀ 87 levels. It seems that PKC⑀ 95 dephosphorylation occurs in two stages; firstly the removal of the phosphate at Ser 729 by a specific phosphatase, followed by dephosphoryl-ation at Thr 566 and Ser 703 by an OA-sensitive phosphatase. It has been shown that dephosphorylated forms of PKC are more susceptible to subsequent proteolytic degradation (13)(14)(15). PI 3-kinase inhibitors also increased PKC⑀ 87 formation upon passage. This may be through inhibition of a kinase that rephosphorylates PKC⑀ Ser 729 , possibly a complex including PKC (9 -11, 43 (4) or through the activity of another kinase (10,11).
Inhibitors of PKC␦ activity and translocation also increased PKC⑀ 87 detected upon passage suggesting that PKC␦ either inhibits PKC⑀ dephosphorylation or promotes PKC⑀ 87 dephosphorylation and degradation. Inhibition of PKC⑀ 95 dephosphorylation may be mediated through inhibiting a Ser 729 phosphatase or through inhibiting the activation of PKC⑀, hence its accessibility to phosphatase attack. There is evidence in the literature of a yin-yang relationship between PKC⑀ and PKC␦. For example, PKC⑀ is growth promotory, whereas PKC␦ inhibits cell growth (17). PKC⑀ acts to prevent cells from cardiac   (44,45). The role of PKC␦ in the regulation of PKC⑀ dephosphorylation in fibroblasts needs to be further investigated. PKC⑀ 87 formation upon passage is inhibited by PD98059, a MEK inhibitor. This implicates the MAPK pathway in the control of PKC⑀ Ser 729 dephosphorylation. Because serum is required for this process, it is likely that the MAPK pathway is downstream of any extracellular signals in the medium that lead to cell proliferation. As already stated, it seems that signals that stimulate cell proliferation are essential for PKC⑀ 87 formation upon cell passage.
PKC⑀ phosphorylation at Ser 729 is PKC-sensitive because we have demonstrated that PKC⑀ 87 production upon passage can be inhibited by the PKC inhibitors chelerythrine and Ro 31-8220. Also, PKC⑀ 87 production can be stimulated in quiescent cells through the addition of 250 nM TPA for 15 min. This suggests a role for another isoform of PKC in the regulation of PKC⑀ dephosphorylation. Alternatively, PKC⑀ may regulate the activity of a Ser 729 phosphatase. PKC inhibitors would therefore inhibit this phosphatase activity through inhibition of PKC⑀.
Our working model for PKC⑀ regulation on passage is summarized in Fig. 6. We have observed similar results in 3T6 fibroblasts and C6 glioma cells suggesting that our findings are not cell type-specific. Our data suggest that at least a two-stage process is involved in regulating dephosphorylation of PKC⑀ 95 . Passage of cells or TPA treatment of quiescent cells causes dephosphorylation at Ser 729 . It is probable that PKC⑀ is activated upon passage and readhesion. PKC⑀ has been shown to be important in the spreading of HeLa and other cells (46,47) where PKC⑀ is activated upon cell matrix contact and is required for actin polymerization. The loss of a phosphate at Ser 729 is sensitive to re-entry into the cell cycle from G 0 and readhesion of cells; it is also dependent upon the MAPK pathway. It has been reported that, once in an activated conformation, PKC becomes susceptible to phosphatase attack (27) and that loss of the C-terminal phosphorylations may regulate translocation of the kinase to the cytoplasm after activation (48). The loss of a phosphate at Ser 729 is not mediated through PP1, PP2A, or PP2B. PKC⑀ 87 could then be rephosphorylated at Ser 729 and recycled back to PKC⑀ 95 . This may involve a membrane-associated kinase complex including PKC. (43) Alternatively PKC⑀ 87 may be further dephosphorylated at sites Ser 703 and Thr 566 , perhaps mediated by an OA-sensitive phosphatase. Complete dephosphorylation increases the instability of PKC and targets the protein for degradation (13,14,15). It is therefore likely that once fully dephosphorylated PKC⑀ is sensitive to degradation via calpain and the proteasome.