The Intermediate Filament Protein Keratin 8 Is a Novel Cytoplasmic Substrate for c-Jun N-terminal Kinase*

Keratins 8 (K8) and 18 are the primary intermediate filaments of simple epithelia. Phosphorylation of keratins at specific sites affects their organization, assembly dynamics, and their interaction with signaling molecules. A number of keratinin vitro and in vivo phosphorylation sites have been identified. One example is K8 Ser-73, which has been implicated as an important phosphorylation site during mitosis, cell stress, and apoptosis. We show that K8 is strongly phosphorylated on Ser-73 upon stimulation of the pro-apoptotic cytokine receptor Fas/CD95/Apo-1 in HT-29 cells. Kinase assays showed that c-Jun N-terminal kinase (JNK) was also activated with activation kinetics corresponding to that of K8 phosphorylation. Furthermore, K8 was also phosphorylated on Ser-73 by JNK in vitro, yielding similar phosphopeptide maps as thein vivo phosphorylated material. In addition, co-immunoprecipitation studies revealed that part of JNK is associated with K8 in vivo, correlating with decreased ability of JNK to phosphorylate the endogenous c-Jun. Taken together, K8 is a new cytoplasmic target for JNK in Fas receptor-mediated signaling. The functional significance of this phosphorylation could relate to regulation of JNK signaling and/or regulation of keratin dynamics.

sue-specific; for example, keratins 8 and 18 (K8/18) are preferentially expressed in simple epithelia. Phosphorylation of keratins at specific sites affects their functional and assembly state and has been suggested to play a role in cell signaling (1). Phosphorylation of the N-terminal Ser-33 on K18 enables a cell cycle-dependent interaction of K8/18 IFs with members of the 14-3-3 protein family, promotes depolymerization of K8/18 in vitro (2), and plays a role in the intracellular distribution of K8/18 polymers (3). K8 is also phosphorylated upon activation of the epidermal growth factor receptor (4) and by pro-urokinase stimulation (5). Several studies have shown that K8/18 phosphorylation is elevated upon cell stress (6,7). Drug-induced hepatotoxic stress, induced by feeding mice with a griseofulvin-supplemented diet, results in a dramatic K8/18 hyperphosphorylation (7). Drug-induced apoptosis in epithelial cells leads to phosphorylation of K18 on Ser-52 and a marked reorganization of K8/18. Pervanadate-mediated tyrosine phosphorylation of K8 and -19 in a p38 kinase-dependent pathway has been reported recently (8). In cultured HT-29 cells, hyperphosphorylation of K8 at Ser-73 was associated with apoptosis induced by anisomycin or etoposide. The kinase regulating this site has not been identified, but members of the proline-directed mitogen-activated protein kinase (MAPK) family have been suggested as candidates (9). Because Ser-73 is phosphorylated upon various stress conditions, it could be a putative target for the stress-activated MAPK family members.
The MAPK family is composed of the prototype kinase, the mitogen-activated MAPK/ERK, and the stress-activated protein kinases, c-Jun N-terminal kinase (JNK) and p38 kinase. Both JNK and p38 kinase are involved in cellular responses to physical stresses, inflammatory cytokines, and apoptosis (10 -13), whereas ERK is activated by growth factors and other mitogens. Among the most specific and physiological activators of JNK and p38 kinase are the death receptors, including the Fas receptor (FasR) and tumor necrosis factor receptor (13). The FasR, belonging to the TNF receptor family, is a physiological activator of apoptosis in a wide range of cell types. A number of reports have shown that JNK is an integral part of FasR-mediated signaling, both when cells are sensitive and insensitive to FasR-mediated apoptosis (14,15). JNK can be elicited through many of the following mechanisms in response to FasR activation: through the Daxx protein (16); by caspasedependent activation of MEKK1, a JNK kinase kinase (10,(17)(18)(19); and by augmented levels of ceramide (20). In some reports and cell models (10,15,21), activation of JNK has been suggested to be independent of apoptosis, whereas in some other models, FasR-mediated apoptosis seems to require JNK activity.
JNK was first identified as a kinase specific for phosphorylation of the transcription factor c-Jun (22). Since then, a number of other substrates, predominantly transcriptional factors, have been established. These include Elk-1, ATF, p53, DPC4, NFAT4, and Sap-1 (23,24). However, very little is known about the possible cytoplasmic targets for JNK. In this respect, the neuronal intermediate filament neurofilament-H (NF-H) has been implicated as a JNK substrate, with phosphorylation at repeated Lys-Ser(P)-Pro-X-Glu motifs within the C-terminal domain (25).
In the this study, we wanted to elucidate the possible role of JNK in regulating K8/18 phosphorylation under physiological conditions, not involving unspecific stress. Therefore, we used stimulation of the FasR in HT-29 cells as a model, because in these cells FasR results in strong activation of JNK, without leading to apoptosis. By using this model, we observed that K8 is phosphorylated by JNK both in vitro and in vivo. K8 phosphorylation in vivo was observed after stress-mediated activation of JNK and following physiological JNK activation by stimulation of the FasR. The in vivo phosphorylation pattern of K8 follows the activation kinetics of JNK. Furthermore, JNK and K8/18 are associated with each other in vivo, as determined by co-immunoprecipitation studies. These results show that JNK participates in the regulation of K8/18 polymers. The association of JNK with K8/18 polymers may have a role in regulating JNK signaling and in adaptation to both environmental and physiological stresses.
In Vitro Phosphorylation of K8/18 -JNK was immunoprecipitated as described in the kinase activity assay from control and 9-h FasRstimulated HT-29 cells. K8/18 protein preparations were used as a substrate and were boiled for 3 min to inactivate any endogenous kinase activity. The reaction was carried out at 37°C for 30 min in kinase reaction buffer (10 mM Tris, pH 7.4, 150 mM NaCl, 10 mM MgCl 2 , 25 M ATP, 0.5 mM DTT) in the presence or absence of 2 Ci of [␥-32 P]ATP. This was followed by the addition of 3ϫ Laemmli buffer to terminate the reaction. Phosphoproteins were then separated on 10% SDS-PAGE. Incorporation of phosphate groups into the substrate was quantified by using a MicroComputer Imaging Device (M4, Imaging Research Inc., Canada). For immunoblot analysis of phosphorylated substrates, the radioactive [␥-32 P]ATP was omitted.
Phosphopeptide Mapping and Manual Edman Degradation-Keratins were separated on 10% SDS-PAGE, fixed in 50% methanol, dried, and autoradiographed. The corresponding K8 bands were cut out from the gels and digested twice (9 and 3 h) with trypsin (T-8642, Sigma, 10 g/ml in 50 mM ammonium bicarbonate) at 37°C. The digested peptides were washed with sterile water and dried with a speed vacuum. The digested samples were separated on microcrystalline-cellulose TLC plates (Merck) in two dimensions. The first dimension utilized electrophoresis and the second utilized chromatography, followed by autoradiography. For mixing experiments, equal counts of in vivo and in vitro material was mixed and subjected for separation. To determine further the phosphorylation site of the phosphopeptide, manual Edman degradation was performed as described (52).

K8 Is Phosphorylated on Ser-73 Upon FasR Stimulation, and This Phosphorylation Occurs within the Soluble Keratin Pool-
Keratins have been shown to be hyperphosphorylated under various conditions, including heat stress, virus infection, and stress-induced apoptosis. To investigate whether this is true for death receptor-mediated signaling, we stimulated the FasR and followed the time course of K8 phosphorylation in HT-29 cells, which are insensitive to FasR-mediated apoptosis but can be sensitized to undergo apoptosis by CHX (14). HT-29 cells were incubated with anti-FasR antibody alone or in combination with CHX at the indicated time points (Fig. 1A). A specific monoclonal antibody, LJ4, recognizing the phosphorylated Ser-73 on K8 was used, because this site is phosphorylated during heat shock and chemically induced apoptosis in HT-29 cells (9). K8 became hyperphosphorylated 3 h after FasR ligation, reaching a plateau of ϳ7-fold activation at 9 h, then remaining elevated up to 24 h after FasR stimulation. This was not due to increased synthesis of K8, as the protein levels remained constant (Fig. 1A, lower panel). Combined treatment of anti-FasR and CHX for 6 h resulted in similar K8 phosphorylation, indicating that K8 phosphorylation upon Fas crosslinking occurs regardless if cells are undergoing apoptosis or not and indicating that there is no additive effect by the fact that the apoptotic machinery is triggered. Similarly, we observed a strong K8 phosphorylation when we subjected HT-29 cells to unspecific stress by exposure to UV light (compare 1st and 2nd lanes in Fig. 1A).
Phosphorylation of K8 and 18 is often accompanied by an increase in keratin solubility (2,26,27). To test for this possibility, we prepared cell extracts from HT-29 cells and separated the pellet, containing insoluble keratins, and the supernatant, containing soluble keratins (see "Experimental Procedures"). As demonstrated in Fig. 1B, no increase of K8-Ser-73 phosphorylation after FasR stimulation was detected in the insoluble keratin pool, even when cells were stimulated with UV light. On the contrary, hyperphosphorylation of K8 on Ser-73 upon FasR stimulation was observed in the pool of soluble keratins (Fig. 1A). This was further confirmed by an additional centrifugation of the soluble keratin pool at 200,000 ϫ g for 1 h. In this assay, all long and short keratin filaments will be pelleted, whereas the truly soluble subunits stay in the supernatant. As shown in Fig. 1D, hyperphosphorylation of K8 on Ser-73 in response to FasR stimulation occurred preferentially in the soluble pool (3rd and 4th lanes), whereas this phosphorylation remained nearly invariable in the insoluble fraction. These results clearly demonstrated that the depolymerized subunits are the target for phosphorylation, and therefore, we have employed the soluble keratin pool for further investigations.
Activation of JNK Induced by FasR Stimulation-JNK has been shown to be activated by FasR stimulation in many cell lines, both when cells are insensitive or sensitive to FasRmediated apoptosis. (14,28). Therefore, JNK was an obvious candidate kinase for the FasR-induced K8 phosphorylation. We followed the activation kinetics of JNK after stimulation of the FasR in HT-29 cells. Parallel samples were taken from treatments described above ( Fig. 2A, upper panel). Consistent with the results on K8 phosphorylation, both combined treatments of anti-FasR and CHX and UV light were able to activate JNK. The amount of JNK was confirmed by immunoblotting by using anti-JNK antibody that recognizes both the 46-and 54-kDa isoforms. A histogram of relative K8 phosphorylation and JNK activation (Fig. 2B) showed closely corresponding kinetics and levels of induction (Fig. 1C). Thus, it seemed plausible that JNK would be regulating the phosphorylation of K8 in response to FasR stimulation.
Phosphorylation of K8 on Ser-73 by JNK in Vitro-If the hypothesis of JNK being a regulator of K8 is correct, JNK should also be able to phosphorylate K8 in vitro. We verified this by immunoprecipitating JNK from control cells and from cells 9 h after FasR stimulation. K8/18 isolated from HT-29 cells was used as a substrate for immunoprecipitated JNK. The autoradiograph in Fig. 3A (upper panel) shows an elevated phosphorylation of both K8 and K18 by Fas-activated JNK; however, the phosphorylation is significantly more pronounced on K8. Next, we examined whether this phosphorylation included K8 Ser-73. For that purpose, the preparation of isolated K8/18 was phosphorylated by immunoprecipitated JNK in the absence of [␥-32 P]ATP and subjected to immunoblot analysis using the LJ4 antibody (Fig. 3A, middle panel). We could HT-29 cells were incubated with anti-FasR antibody in the presence or absence of CHX or were exposed to UV light for 5 min. The cells were harvested at the indicated time points, lysed with RIPA buffer, and centrifuged 3,000 ϫ g for 5 min as described under "Experimental Procedures." Soluble K8 (A), insoluble K8 (B), and whole cell extract (C) were resolved on 10% SDS-PAGE followed by immunoblot analysis with the LJ4 antibody specific for the phosphorylated epitope of K8-Ser-73. Confirmation of equal amounts of protein loading was performed by Coomassie Blue gel staining (lower panels). D, FasR ligation induced K8-Ser-73 hyperphosphorylation, which occurs in the soluble pool. FasR-stimulated (9 h with anti-FasR antibody) and nonstimulated HT-29 cells were extracted with RIPA buffer and centrifuged (200,000 ϫ g for 1 h). Soluble proteins (supernatant) and those insoluble in RIPA buffer (pellet) were resuspended in Laemmli sample buffer. Equal amount of proteins were separated on 10% SDS-PAGE and probed with LJ4 antibody, followed by reblotting with anti-K8/18 antibodies. Immunoblot analysis of the K8-Ser-73 phosphorylation level of the soluble and insoluble keratin 8 pools is shown (upper panel), and equal protein loading was confirmed by Western blot analysis using anti-K8/18 antibodies (lower panel). Lane C, control. clearly see that the in vitro phosphorylation of K8 by JNK occurred on Ser-73, and this phosphorylation was markedly elevated upon FasR stimulation. Equal loading of K8/18 was confirmed by reblotting the same nitrocellulose membrane with anti-K8/18 antibodies (Fig. 3A, lower panel).
To further substantiate that K8 is an in vivo JNK target, tryptic phosphopeptide maps of K8 immunoprecipitated from 32 P in vivo labeled HT-29 cells and of K8 phosphorylated by JNK in vitro were compared. Seven peptides (peptides 2-8) could be found in the maps of control cells and in those obtained from anti-FasR antibody stimulated cells (Fig. 3B), with enhanced activities primarily on peptides 3-8 (using peptide 2 as a reference). Peptides 9 and 10, which were absent in the control, appeared after FasR stimulation. Peptides 3 and 5-10 co-migrated with in vitro JNK phosphorylated K8. To identify which of these peptide(s) corresponded to Ser-73, manual Edman degradation was performed. Due to partial digestion, several peptides can potentially correspond to Ser-73. Ser-73 is located at position of 27 on its specific tryptic peptide (not shown), which makes it inaccessible for cleavage because Edman degradation is inefficient after 10 cycles. In the present study, 7 cycles of manual Edman degradation were done for each peptide. The peptides showing inducible phosphorylation, both in vivo and by JNK in vitro, and no phospholabel release within the 7 cycles was considered as phospho-Ser-73-containing peptides. Only peptides 3 and 10 met these criteria (Table  I). Peptide 2, which was not inducibly phosphorylated and showed very little phosphorylation by JNK in vitro, was considered to be Ser-431. This assignment was based on its sequence context (i.e. inaccessible to manual Edman degradation, not shown) and by the use of anti-phospho-Ser-431 K8 antibody (3), which showed that the phosphorylation of this site is not elevated by increased JNK activity in vivo or by JNK in vitro (data not shown). Peptide 6, a major inducible site in vivo, corresponds to Ser-23 of the tryptic peptide 23 SYTSGPGSR. Hence, our peptide mapping data support the assignment of K8 Ser-73 as a major in vivo phosphorylation site for JNK upon FasR stimulation.
Fas-stimulated Phosphorylation of K8 Does Not Involve Other MAPKs and Is Caspase-independent-All members of the MAPK family are involved in FasR-mediated signaling (29,30). This led us to examine whether p38 and ERK could play a role in K8 hyperphosphorylation upon FasR stimulation. The p38 kinase activity was analyzed by a kinase assay using ATF-2 as a substrate. No significant increases of the constitutive p38 kinase activity or the protein expression were observed upon FasR stimulation (Fig. 4A, upper panel), indicating that p38 kinase was not activated upon FasR stimulation. The lower panel confirmed equal protein loading.
When ERK activity was measured, no major increase was observed, as shown in the kinase assay with myelin basic protein (MBP) as a substrate (Fig. 4B, upper panel). The expression level of ERK protein appeared also to be unchanged as shown by immunoblotting (Fig. 4B, lower panel). The constitutive ERK activation in HT-29 cells is high because the type 2A phosphatase inhibitor cl-A failed to enhance the activity of ERK in these cells. To ensure that this observation was not due to a nonfunctional assay, we employed Jurkat T-cells, and we showed that cl-A activates ERK in this cell line (28) (Fig. 4C). Taken together, these results suggest that JNK is preferentially activated upon FasR stimulation in HT-29 cells and could be a potential kinase responsible for the K8 phosphorylation on Ser-73.
In vitro experiments were performed using ERK and p38 kinase immunoprecipitates that were obtained from control and Fas-stimulated cells. Immunoblot analysis with LJ4 antibody revealed that immunoprecipitated ERK or p38 kinase was able to phosphorylate K8 on Ser-73 in vitro (Fig. 4, D and E). However, this phosphorylation was not elevated in response to FasR activation. The lack of a role for ERK and p38 kinase in K8-Ser-73 phosphorylation in FasR-mediated signaling was further supported by additional experiments using specific inhibitors of both kinases. HT-29 cells were pretreated with PD98059, a specific inhibitor of the upstream regulator of ERK1/2, MKK1 (31), or with the p38 kinase inhibitor SB203580 (32), for 30 min, followed by 9 h of incubation with anti-FasR antibody. Phosphorylation of K8 on Ser-73 was analyzed by immunoblotting. As shown in Fig. 4, F and G, neither PD98059, nor SB203580 could inhibit K8-Ser-73 hyperphosphorylation after FasR stimulation. We conclude that the observed hyperphosphorylation of K8 on Ser-73 seems to be associated directly with JNK activation and occurs independently of ERK and p38 kinase.
In Vivo Interaction of JNK and Keratin 8 in HT-29 Cells-As JNK usually shows a high affinity for its substrates, we wanted to clarify whether this would apply for K8 as well. JNK was immunoprecipitated from control and Fas-treated HT-29 cells and then immunoblotted using anti-K8 antibody. As shown in Fig. 6A (upper panel), K8 was co-immunoprecipitated with JNK. The interaction was specific as no JNK was precipitated along with control rabbit IgG. The degree of association did not vary in response to FasR and UV stimulation. Correspondingly, when JNK immunoprecipitates were blotted with LJ4 antibody, pK8-Ser73 was detected among the co-immunoprecipitated K8 (Fig. 6A, middle panel), and the amount of coimmunoprecipitated pK8-Ser73 increased following FasR stimulation, as well as UV irradiation in HT-29 cells. The lower panel of Fig. 6A shows the immunoprecipitated JNK. This binding was not mediated by the heat shock protein 70, which associates with K8 (35), because heat shock protein 70 was not co-immunoprecipitated with JNK (data not shown). These results indicated that JNK not only phosphorylated K8 but also  Fig. 3B), immobilized on membrane discs, and subjected to seven Edman degradation cycles. The table shows the cycle number at which the peak radioactivity was obtained for each respective peptide and whether there was radioactivity remaining on the disc after completion of seven cycles (D). Peptide no.
Edman degradation, label released on cycle number or remained on D (disc) Total no. cycles 2 D 7 3 D 7 4 No data 7 5 2,3 and D 7 6 1 and D 7 7 2,3 and D 7 8 3 and D 7 9 3 and D 7 10 D 7 FIG. 4. p38 kinase and ERK are not involved in the phosphorylation of K8 in FasR-mediated signaling. HT-29 cells were treated as outlined in Fig. 1. A, activation of p38 kinase was measured by immunocomplex kinase assay with ATF-2 as a substrate (upper panel). The amount of nonphosphorylated p38 was determined by immunoblots with an anti-p38 antibody (lower panel). B, activation of ERK was measured by immunocomplex kinase assay with MBP as a substrate (upper panel). The loading of ERK2 was determined by immunoblots with an anti-ERK2 antibody (lower panel). C, for a positive ERK kinase assay, Jurkat cells were treated with cl-A (20 nM) for 30 min. In vitro phosphorylation of K8 by p38 kinase (D) and ERK (E), respectively, was determined as described in the legend of Fig. 2. Upper panels are Western blots by LJ4 antibody, and lower panels show equal protein loading. F, HT-29 cells were incubated with anti-FasR antibody for 9 h in the presence or absence of p38 kinase inhibitor SB203580 (SB) (20 M)  interacted with it. To test for the amount of JNK associated with K8, we examined immunodepletion of JNK by using LJ4 and anti-K8 antibodies. Interestingly, K8 was able to sequester a substantial amount of the 54-kDa isoform of JNK (Fig. 6B). JNK was immunoprecipitated twice with LJ4 and anti-K8 antibodies and blotted against anti-JNK. A reduced amount of 54-kDa JNK was left in the supernatant after the second immunoprecipitation. This observation led us to test whether this sequestration of JNK to K8 would correlate with a reduced ability to phosphorylate endogenous c-Jun. Interestingly, immunoblotting with a specific monoclonal c-Jun antibody showed that the endogenous c-Jun was not clearly shifted in response to FasR stimulation, as compared with a functional antibody control where the HaCaT human keratinocytes were treated with anisomycin (5 g/ml) for 20 min (Fig. 6C). However, the possibility that targeting of JNK to K8 could regulate the degree of c-Jun activation requires a more detailed analysis beyond the scope of the present study. DISCUSSION Our studies show that K8 Ser-73 is phosphorylated in response to FasR stimulation in HT-29 cells. This phosphorylation is likely to be regulated by JNK, as demonstrated by both in vitro and in vivo studies. In the context we studied, ERK and p38 kinase did not appear to play a role, as reflected by the failure to detect increases in their activities in response to FasR stimulation. Furthermore, K8/18 associate with the 54-kDa JNK and could thereby participate in regulating its functions.
K8 as a Substrate for JNK-Although JNK has been considered to be mainly a transcription factor kinase, some cytoplasmic cytoskeletal substrates have also been reported. A strong correlation between hyperphosphorylation of a neuronal IF protein, neurofilament high molecular weight subunit (NF-H), and activation of JNK3 has been demonstrated, thereby suggesting that the NF-H C-terminal domain was phosphorylated by JNK3 (25). Also, the neuronal microtubule-associated protein, tau, is phosphorylated in vitro by JNK at sites that are hyperphosphorylated in tau from patients with Alzheimer's disease, as compared with fetal or adult tau (36). More recently, another possible JNK-specific tau phosphorylation site has been postulated (37). Despite these implications, there are no data to firmly establish JNK-mediated phosphorylation of a cytoskeletal target in vivo. Using the specific LJ4 antibody and specific phosphopeptide mapping, we have shown that upon FasR stimulation K8 is hyperphosphorylated on Ser-73 in a JNK-dependent manner.
Previous results (9,38,39) have shown that the kinase responsible for the phosphorylation of Ser-73 could involve a kinase from the MAPK family, due to the Ser-Pro motif of K8-Ser-73 phosphoacceptor site, which could serve as an in vitro target for proline-directed MAPK/ERK. When comparing the amino acid sequence surrounding the Ser-63 and Ser-73 phosphoacceptor sites in the c-Jun transcription factor with that surrounding Ser-73 in K8 (Table II), essential similarities were observed. Hence, the Ser-73 phosphoepitope on K8 is a likely phosphorylation site for JNK. However, despite phosphoacceptor sequence similarities, specific and effective phosphorylation by JNK requires a specific docking site as shown for the JNK substrates, c-Jun and ATF-2 (39 -41). In our study, we demonstrated that K8 and JNK co-immunoprecipitated with high affinity. In searching for possible docking sites in K8,  The sequence surrounding Ser-73 of K8 (9) was compared with the known c-Jun phosphoacceptor (Ser-63 and Ser-73) sequences (39). Similar Ser-Pro motifs were observed in all three sequences.

K8-Ser-73(P)
Leu-Leu-Ser 73 -Pro-Leu-Val c-Jun-Ser-73(P) Leu-Ala-Ser 73 -Pro-Glu-Leu c-Jun-Ser-63(P) Leu-Leu-Thr-Ser 63 -Pro-Asp-Val we compared the docking site sequences from a number of JNK substrates with the sequences of K8 and K18, and we found that there are two short sequences in K8 and one in K18 that resemble the docking sequences of these JNK substrates (Table  III). Thus, K8 could either bind to JNK directly by itself or indirectly via polymerization with K18 that interacts with JNK, as described for JunD (38). However, this interaction does not seem to go through heat shock protein 70, which has been shown in complex with K8 (35). Furthermore, K8 tends to associate more efficiently with the 54-kDa isoform of JNK, indicating substrate specificities of JNK isoforms for K8. Recently, a cytoplasmic scaffold protein JIP1, which binds selectively to JNK and to the upstream activating kinases, but not to other related MAPK including p38 and ERK, has been discovered (23). Overexpression of JIP1 causes cytoplasmic retention of JNK and suppresses the effects of JNK on apoptosis (42). In our study, we did not observe any significant elevation of endogenous c-Jun phosphorylation upon FasR stimulation in HT-29 cells, although JNK was highly activated following the same treatment. This raises the possibility that in some tumor cell lines, such as HT-29 cells, K8/18 may act as a JNK sequestering complex, thereby weakening JNK from performing some of its cytosolic or nuclear tasks. However, the possibility that the observed association of JNK with K8 would affect JNK targeting needs to be examined in closer detail before any solid conclusions can be drawn.
In agreement with previous reports (8,9), we have shown that K8-Ser-73 is a site that can be phosphorylated by ERK and p38 kinases in vitro. However, phosphorylation of this K8 site by ERK and p38 kinase was not enhanced, and the activities of these two kinases were not elevated in response to FasR stimulation. Thus, the observed K8 phosphorylation does not seem to be ERK-or p38 kinase-mediated but is JNK-dependent. However, K8 Ser-73 can also be a selective in vivo p38 kinase substrate under conditions that activate this kinase (53).
The Possible Biological Functions of K8-Ser-73 Phosphorylation Mediated by JNK-The significance of keratin phosphorylation has been extensively studied (1,43). One common feature of keratin phosphorylation is the increase in their solubility (43). In the present study, a basal level of phosphorylation on K8-Ser-73 in control and stimulated HT-29 cells was discerned in the nonsoluble keratin pool, whereas the increase in Ser-73 hyperphosphorylation was observed almost exclusively in the pool of disassembled keratin IFs. This enhanced phosphorylation and solubility is likely to be of importance in regulating keratin turnover, assembly, and dynamics.
The execution of apoptosis requires cleavage of a number of proteins, such as keratins. In a human endometrial adenocarcinoma cell line SNG-M, K18, but not K8, was degraded by caspases during apoptosis (44). Similarly, in HT-29 cells, K8 was protected from cleavage when phosphorylated on Ser-73, whereas K18 was degraded regardless of its phosphorylation state (45). Furthermore, our results showed that the K8 hyper-phosphorylation and JNK activation upon FasR stimulation were both apoptosis-and caspase-independent. Recently, keratin-dependent, JNK-mediated epithelial resistance to TNFmediated apoptosis has been reported. K8 and K18 bind to the cytoplasmic domains of TNF receptor 2 and modulate the TNFdependent activation of JNK and the NFB transcription factor. Cells with the truncated form of K18 and disrupted K8/18 filaments responded to TNF by increasing the level and the duration of JNK activity. Thus, epithelial cells with decreasing amounts of K8/18 display a higher cellular sensitivity to killing by TNF (46). Taken together, it is possible that upon FasR stimulation, JNK-mediated caspase-independent hyperphosphorylation of K8 on Ser-73 would be involved in protecting the keratin network from cleavage by caspases in the FasR-resistant HT-29 cell line.
One important feature of JNK in regulation of various transcription factors, such as c-Jun, JunB, ATF-2, or tumor suppressor p53, is to target them for ubiquitination, and the interaction between JNK and these proteins appears to be essential for the targeting. Disruption of these bindings results in prolonged half-lives of these transcription factors (47,48). Under stress conditions, phosphorylation of these proteins is enhanced by activated JNK, and the ubiquitination is thereby repressed, so that these transcription factors can function to cope with stress (49). A recent report has also illustrated that phosphorylation of K8 on Ser-431/-73 and of K18 on Ser-52 protects them from ubiquitination (50). Therefore, JNK may be involved in regulating K8 ubiquitination by interacting with the keratin filaments and phosphorylating K8, and this regulation may be important for cells to withstand various stresses.
In conclusion, this study illustrates the existence of a novel cytoskeletal substrate for JNK in epithelial cells. The soluble pool of K8 is hyperphosphorylated on Ser-73 by JNK upon FasR stimulation in a caspase-and apoptosis-independent manner in HT-29 cells. Although the role for this phosphorylation is still not clear, the JNK/K8 interaction could be important in both controlling K8/18 dynamics and turnover as well as in the control of JNK signaling.