Constitutive ERK1/2 activation in esophagogastric rib bone marrow micrometastatic cells is MEK-independent.

In this study, we examined the mitogen-activated protein kinase (MAPK) cascade in micrometastatic cell lines generated from rib bone marrow (RBM) of patients undergoing resection of esophagogastric malignancies. The molecular mechanism(s) involved in esophagogastric MAPK activation have not previously been investigated. Constitutive activation of both ERK1 and -2 isoforms was evident in each of the five RBM cell lines. Elk-1, a transcription factor activated by the ERK1/2 pathway was also found to be constitutively activated. Cell lines generated from metastases of involved lymph nodes (OC2) and ascites (OC1) of patients with esophageal cancer do not display, however, hyperphosphorylation of ERK1/2. Constitutive RBM ERK1/2 activation is protein kinase C and phosphatidylinositol 3-kinase dependent. Surprisingly, constitutive ERK1/2 activation is MEK-independent. Pharmacological inhibition of MEK with two specific inhibitors, PD 98059 and U0126, were both ineffective in blocking ERK activation. Similarly, the use of a dominant negative MEK mutant was without effect. Interestingly, experiments overexpressing two different dominant negative Pak1 mutants significantly reduced RBM ERK1/2 activation, albeit not to the same extent for all cell lines. We also examined the role of three different phosphatases, PAC1, MKP-1, and -2. While RBM ERK1/2 activation was found to be PAC1- and MKP-2-independent, surprisingly, MKP-1 was down-regulated in all five RBM cell lines. In conclusion, we provide evidence for the first time for a MEK-independent constitutive ERK1/2 activation pathway in esophagogastric RBM cell lines. These findings have important implications for drug treatment strategies which currently target MEK in other forms of cancer.

Esophageal cancer is an aggressive tumor which responds poorly to treatment and has a poor prognosis (1,2). Approximately half of patients diagnosed with localized esophageal cancer die of metastatic disease within the first 2 years following tumor resection. We have succeeded in developing a number of esophagogastric cell lines from the rib bone marrow of patients (3). We have previously shown that these cells are viable, proliferate, and grow independently in tissue culture and form malignant tumors in athymic nude mice (4). These metastatic cells are representative of the disseminated progenitors of secondary tumors and are the appropriate targets for treatment.
The precise molecular events leading to the acquisition of the metastatic phenotype remain largely unknown. Members of the Ras superfamily of small GTP-binding proteins have been implicated in tumor progression and are found to be activated in 20 to 30% of tumors (5). In its active GTP-bound state, Ras activates the serine threonine kinase Raf (6). Raf upon activation in turn phosphorylates the dual specific kinase MEK 1 (also known as MAP kinase kinase or MAPKK) which in turn phosphorylates the MAP kinases ERK1 and -2 (for extracellular regulated kinases 1 and 2). Although MEK is the only known kinase directly downstream of Raf, there are several lines of evidence suggesting that Raf can also activate other effectors (7,8). In vitro MEK1 is phosphorylated by ERKs, Cdk2, and the p21-activated kinase Pak1 (9,10). The Pak kinases Pak1, -2 and -3, are a family of protein kinases that are regulated by GTP-bound Rac and Cdc42 and are candidates for effectors that mediate both actin and JNK signaling (for review, see Ref. 11).
The MAPKs are activated by the reversible dual threonine and tyosine phosphorylation of a conserved T-X-Y motif (12). The reversible nature of MAPK phosphorylation suggests that phosphatases play a key role in regulating MAPK activity (for review, see Ref. 13). MAPK phosphatase-1 and -2 (MKP-1, -2) as well as PAC1, a phosphatase which shares a highly homologous C-terminal catalytic domain with MKP-1 (14), are known to inactivate ERKs (15) and possibly JNKs/stress-activated protein kinases (12,16). While PAC1 is expressed predominantly in hematopoietic cells, MKP-1 and -2 are more widely expressed and are induced by growth factors and genotoxic and environmental stresses (17,18).
acterized. In this study, we undertook to characterize the MAPK signaling pathway in micrometastatic cell lines generated from rib bone marrow of patients undergoing resection of esophagogastric malignancies, which until now had remained unexplored. Using complementary genetic and pharmacological approaches, we now demonstrate that in all five esophagogastric rib bone marrow (RBM) cell lines examined, both ERK1 and -2 isoforms are activated in a MEK-and Raf-1-independent manner. In contrast, however, ERK1/2 activation was found to be PKC-, PI3K-and Pak1-dependent. Moreover, we did not detect PAC1 expression, but found that MKP-1 but not MKP-2 was down-regulated. In addition, the activation of ERK1/2 correlated with increased activation of the downstream transcription factor Elk-1.

Cells
Generation of the esophagogastric metastatic rib bone marrow cell lines (lines 1-5) used throughout this study have previously been described in detail by our group, and were obtained from patients undergoing resection of esophagogastric cancers (4). Briefly, RBM1 and -5 cell lines are cells cultured from two different patients with adenocarcinoma of the gastroesophageal junction. RBM2 and -4 cell lines are cells cultured from two different patients with adenocarcinoma of the esophagus and finally, the RBM3 cell line are cells cultured from a patient with squamous carcinoma of the esophagus. OC1 and OC2 are squamous cell lines established from malignant ascites and a lymph node, respectively, of two patients with esophageal cancer. Finally, OC3 are adenocarcinoma cells established from a metastatic lymph node of a Barrett's esophageal lesion. We used normal human bone marrow stromal (HBMS) cells throughout this study as control. CACO-2, HT-29, and SW480 cells were also employed for comparative purposes in some experiments throughout this study.
Plasmids cDNA expression plasmids utilizing the cytomegalovirus promoter to express Myc-tagged Pak1, Pak1 R299 , Pak1 L83,L86 , and Pak1 L83,L86,R299 were the generous gift of Dr. J. Field (University of Pennsylvania School of Medicine). The Pak constructs have been previously described in detail (24 -26). Briefly, Pak1 is wild type Pak, Pak1 L83,L86 is a hyperactive Pak, Pak1 R299 lacks kinase activity and Pak1 L83,L86,R299 is a mutant that lacks kinase activity and also fails to bind either Rac or Cdc42 (26). The dominant-negative form of MEK1 was a generous gift of Drs. O. A. Coso and J. S. Gutkind (National Institutes of Health).

Transient Transfection of Cells
To investigate constitutive Elk-1 transcriptional activation, we employed the PathDetect trans-reporting system as previously described but with minor modifications (27). Plasmid DNAs were transiently transfected into RBM(1-5), OC1, -2, and -3 using LipofectAMINE TM 2000 Reagent (Life Technologies) according to the manufacturer's instructions. For all experiments, 1 g of Renilla luciferase vector pRL-SV40 (Promega Corp.) was included which provides constitutive luciferase expression in transfected cells and serves as an internal control to normalize transfection efficiency. Cells were co-transfected with luciferase reporter (pFR-Luc) and Renilla plasmids, along with the GAL-4c-Elk-1 expression plasmid or pFC-dbd plasmid as a negative control. Transfections were carried out using a total of 5 g of DNA (3 g of GAL-4-c-Elk-1 plus 1 g of pFR-Luc plus 1 g of Renilla or 3 g of pFC-dbd plus 1 g of pFR-Luc plus 1 g of Renilla). Forty-eight hours following transfection, cells were washed twice in phosphate-buffered saline and refed with DMEM supplemented with antibiotics and 0.5% FCS for a further 48 h. Both firefly-and Renilla-luciferase activities were measured using a commercial Dual Luicferase TM Reporter Assay Kit (Promega) following the manufacturer's instructions.

Stable Transfection of Cells
To establish stable cell lines expressing Pak1, Pak1 R299 , Pak1 L83,L86,R299 , Pak1 L83,L86 , and DN MEK, each construct was cotransfected with pCDNA3 into RBM(1-5) cell lines, OC1, -2, and -3. Transfections were carried out using LipofectAMINE TM 2000 Reagent and a total of 10 g of DNA (5 g of each test plasmid plus 5 g of pCDNA3) according to the manufacturer's instructions. Following 48 -72 h after the addition of DNA, cells were washed twice in phosphate-buffered saline, and split 1:3 into 100-mm diameter dishes. The transfected cells were selected in growth medium containing 600 g/ml Geneticin (G418; Life Technologies Inc., Grand Island, NY). After 4 -8 weeks, individual cell colonies were transferred for clone expansion and maintained in culture medium supplemented with 600 g/ml G418. Protein expression levels were determined by Western blot analysis of G418-selected cell foci using anti-Myc tag mAb 9E10 (Calbiochem) or an anti-MEK antibody (New England Biolabs) for Pak1 and MEK transfections, respectively. Blots were visualized with the procedure outlined in the enhanced chemiluminescence kit (Amersham Pharmacia Biotech).

Cell Proliferation Assays
MTT Assay-RBM (1)(2)(3)(4)(5) cell lines were seeded onto 96-well plates at 5 ϫ 10 3 cells per well in growth medium and grown overnight prior to the initiation of any experimental treatments. Following the indicated treatments, MTT (5 g/ml) was added to each well of a 96-well plate and the reduction of MTT was assayed to calculate cell numbers as previously described (30).
[ 3 H]Thymidine Incorporation-Cells were seeded sparsely in 6-well plates, and grown in DMEM supplemented with 10% FCS and antibiotics for 24 h. The cells were then starved by culturing in the same medium without serum for 48 h. The medium was then removed, and cells were refed with medium containing 10% FCS and 0.5 Ci/well [ 3 H]thymidine (Amersham Pharmacia Biotech). The cells were incubated for 48 h after which [ 3 H]thymidine incorporation was performed as previously described (31).
Cytotoxicity Assay-For experiments using various kinase inhibitors, cytotoxicity was measured by the MTT cell viability assay described above.

Statistical Analysis
Results are expressed as mean Ϯ S.E. Statistical comparisons were made by using analysis of variance with subsequent application of Student's t test, as appropriate. (1)(2)(3)(4)(5)-To determine whether ERKs are activated in esophagogastric RBM we evaluated the phosphorylation of ERK1/2 using phosphospecific antibodies as an index of ERK1/2 activation (27). All five of the RBM cell lines when cultured in 0.5% FCS, had substantially higher levels of phosphorylated ERK1/2 when compared with HBMS cells, HT-29, CACO-2, SW480, and OC1, -2, and -3 (Fig. 1B). ERK1/2 was defined as constitutively activated in RBM(1-5) as cells grown in the presence of serum (S) or treated with PMA for 5 min did not demonstrate a further increase in phosphorylation (Fig. 1B). Results are shown for RBM1 but similar results were observed for RBM(2-5) (data not shown). Since previous reports have linked constitutive ERK activation with ERK hyperexpression (20,22), we examined ERK expression by immunoblot analysis. We failed to see any significant difference in ERK protein expression when RBM(1-5) cell lines were compared with HBMS cells, HT-29, CACO-2, SW480 and OC1, -2, and -3 (Fig. 1A).

Constitutive Activation of ERK1/2 in RBM
To verify that the observed phosphorylated ERK1/2 in RBM(1-5) was active and capable of in vitro phosphorylation of cellular proteins, we used immunoprecipitated ERK from equal volumes of protein extracted from RBM (1)(2)(3)(4)(5). HBMS cells and cells grown in the presence of 10% FCS were employed for comparison purposes in in vitro kinase assays with myelin basic protein as a substrate. A representative experiment is shown in Fig. 1C which corroborates Western blot data analysis (Fig. 1B), again demonstrating a marked increase in the activation state of ERK in RBM cells compared with normal HBMS cells. No further increase in ERK activity was observed when RBM1 cells were grown in the presence of 10% FCS (Fig.  1C). Similar results were obtained for RBM(2-5) (data not shown).
To address the question of whether other members of the MAPK cascade could also be constitutively activated, we examined the phosphorylation of both JNK and p38. We failed to see either JNK or p38 phosphorylation in each of the five cell lines examined (Fig. 1, D and E). Treatment of the RBM1 cell line with UV for 5 min is shown as a positive control (Fig. 1, D and E).
Constitutive ERK1/2 Activation is Raf-and MEK-independent in RBM(1-5)-Since both PD98059 and U0126, two specific inhibitors of MEK (35,36), failed to alter the activation of ERK1/2 in RBM(1-5) (Fig. 2, A and C), we turned our attention to MEK. We examined the expression, phosphorylation, and activation of MEK in RBM(1-5). The phosphospecific MEK1/2 antibodies (New England Biolabs) only recognize the phosphorylated forms of MEK1/2 but do not distinguish between the two isoforms. As expected, serum-starved HBMS cells, as well as OC1, did not exhibit MEK1/2 phosphorylation (Fig. 3C). Similar results were found for OC2, -3, CACO-2, HT-29, and SW480 (data not shown). However, MEK1/2 was not found to be phosphorylated in serum-starved RBM(1-5) (Fig. 3B), despite the existence of ERK activation. When RBM1, HBMS, and OC1 cells were cultured in DMEM supplemented with 10% FCS (S) or treated with PMA (100 nM) for 5 min, MEK1/2 was phosphorylated (Fig. 3, B and C). Results are shown for RBM1, but similar observations were noted for RBM(2-5) (data not shown). Since a previous report has linked MEK hyperexpression with ERK constitutive activation (21), we performed immunoblot analysis to determine the expression level of MEK protein in RBM(1-5) versus HBMS, OC1, -2, and -3, using antibodies that recognize total MEK1/2. Unlike previous reports (21), we failed to see any significant differences in MEK1/2 expression for all cells analyzed (Fig. 3A). To reconfirm the absence of MEK activation in RBM(1-5) we also performed a MEK kinase assay using kinase inactive GST-ERK-1 (K71A, Upstate Biotechnology Inc.) as substrate (28). Consistent with our immunoblot results, we failed to observe an increase in MEK kinase activity (Fig. 3, D and E). Results are shown for RBM1, but were similar for RBM(2-5) (data not shown). Cells cultured in the presence of serum (S) or treatment with PMA (100 nM) for 5 min, however, showed a marked increase in MEK activity, as assessed by GST-ERK1 phosphorylation (Fig. 3, D and E). Thus, RBM(1-5) have a functional MEK as well as ERK.
Since Raf lies directly upstream of MEK, we also investi-gated whether Raf is activated or not in RBM cell lines. For each of the five cell lines examined we failed to see either Raf phosphorylation (Fig. 3F) or activation (Fig. 3G). However, cells cultured in the presence of serum or treatment with PMA (100 nM) for 5 min showed a marked increase in Raf-1 phosphorylation (Fig. 3F) as well as Raf-1 activity as assessed by GST-MEK-1 phosphorylation (Fig. 3G) as previously described (38). Thus, RBM(1-5) have a functional Raf-1 and MEK as well as ERK. (1)(2)(3)(4)(5), cells were stably transfected with a DN MEK mutant. In agreement with our results using the two specific MEK inhibitors, we failed to see any change in ERK1/2 phosphorylation in RBM(1-5) when cells overexpressed DN MEK versus nontransfected cells (Fig. 4, A  and B). The DN MEK plasmid utilized was not HA-tagged. We attempted to verify the expression of the mutant by Western blot analysis but failed to see a significant difference in the expression levels between transfected and nontransfected cells (data not shown). However, the DN MEK mutant was capable of inhibiting ERK activation by PMA (100 nM, 5 min) in OC1, -2, and -3 cells (Fig. 4C). To further confirm that cells were transfected, we analyzed the level of MEK activation by Western blot analysis. In all five RBM cell lines examined (Fig. 4D) as well as OC1, -2, and -3 cells (Fig. 4E), we observed a significant reduction in MEK1/2 phosphorylation to basal levels in cells that stably expressed the DN MEK mutant versus the nontransfected cells when cells were stimulated with PMA (100 nM, 5 min).

Dominant Negative MEK Does Not Disrupt ERK1/2 Constitutive Activation in RBM(1-5)-To further confirm that ERK activation is MEK-independent in RBM
Pak1 Contributes to the Constitutive Activation of ERK1/2 in FIG. 3. Absence of MEK1/2 and Raf-1 activation in esophagogastric micrometastatic rib bone marrow cell lines. Immunblot analysis as described under "Materials and Methods" was performed using anti-MEK1/2 antibodies that react with total MEK1/2 (A), phospho-anti-MEK1/2 antibodies that react only with the phosphorylated forms of MEK1/2 (B and C), and Raf-1 (C-12) antibody that reacts with both phosphorylated and non-phosphorylated Raf-1 (F and G). Aliquots of cell extracts were loaded on a 7.5% SDS-PAGE for F and G and all other gels were 10%. Twenty g of protein was loaded per lane (A-D, F, and G) and are representative of three different blots. Kinase activity as described under "Materials and Methods" was monitored either by autoradiography (D) or by liquid scintillation counting (E). E, results are expressed as mean Ϯ S.E. from three independent experiments. (1)(2)(3)(4)(5), we sought to identify other kinases which may lie upstream in the MAPK signaling pathway of these cells. Since we observed that the two PI3K inhibitors LY294002 and wortmannin both abolished ERK1/2 activation in RBM(1-5) (Fig. 2, A and C), we turned our attention to Pak1, a kinase which has previously been shown to be activated by PI3K (25). Pak has been shown to activate JNK (39), as well as ERK (40). We developed stable RBM (1)(2)(3)(4)(5) cell lines that expressed the various Pak mutants, confirmed by Western blot analysis of the Myc-tagged plasmids (Fig. 5A). Pak1 is wild type Pak, Pak1 R299 lacks kinase activity, Pak1 L83,L86,R299 is a mutant that lacks kinase activity and also fails to bind either Rac or Cdc42, and Pak1 L83,L86 is a hyperactive Pak. The two kinase-deficient Pak mutants are both dominant negative forms of Pak. Pak1 is seen as a 65-kDa band. Therefore, we tested the effects of the different Pak1 mutants versus empty vector on the constitutive activation of ERK1/2 in RBM (1)(2)(3)(4)(5). We observed that wild type Pak1 did not affect ERK1/2 activation (Fig. 5B). When we tested the two kinase mutants Pak1 R299 and Pak1 L83,L86,R299 , however, we consistently found that both reduced ERK1/2 phosphorylation in all of the RBM cell lines, albeit not to the same extent. We observed that the two Pak1 mutants continuously had less of an effect on RBM4 and -5 ERK activation in comparison to RBM1-3 (Fig. 5, C and D). A reduction in phosphorylated ERK in cells overexpressing the dominant negative Pak mutants has previously been described by Tang et al. (26). Interestingly, cells overexpressing the hyperactive Pak1 L83,L86 showed a further increase in ERK1/2 phosphorylation (Fig. 5, E and F). This increase in ERK1/2 activation was MEK-independent but PI3K dependent (Fig. 5, E and F). Again, results are shown for RBM1 but were similar for RBM(2-5) (data not shown). We also examined whether cells expressing the different Pak1 mutants could alter the levels of ERK and MEK protein expression. In all five cell lines expressing the four different Pak1 mutants, we failed to see any changes in protein expression (data not shown). Since Pak has been shown to activate JNK (39), we also examined whether the hyperactive Pak1 mutant, Pak1 L83,L86 , which brought about a further increase in ERK activation, could increase JNK activation in RBM (1)(2)(3)(4)(5). We failed, however, to see up-regulation of JNK in RBM(1-5) (Fig. 5G) RBM (1)(2)(3)(4)(5) Demonstrate Constitutive Transcriptional Activation of Elk-1-To further evaluate the level of activation in RBM(1-5), we investigated whether the observed upstream constitutive activation of ERK1/2 may influence gene expression. RBM (1)(2)(3)(4)(5) transiently transfected with GAL-c-Elk-1 demonstrated an elevated level of Elk-1 activation when compared with HBMS, OC1, -2, and -3 (Fig. 6A). To investigate the upsteam kinases involved in the Elk-1 activation, cells were treated with the PKC inhibitor GF 109203X (5 M). GF 109203X brought about a significant reduction in the activation of Elk-1 (Fig. 6B). Similarly, the PI3K inhibitors LY294002 (20 M) and wortmannin (100 nM), reduced the levels of Elk-1 activation (Fig. 6B). In contrast, however, both PD 98059 (20 M) and U0126 (20 M), failed to abrogate the levels of Elk-1 activation, thereby implying a MEK-independent Elk-1 upregulation in RBM(1-5) (Fig. 6B).

RBM(1-5)-Since both MEK inhibitors or DN MEK failed to alter the activation of ERK1/2 in RBM
Since we observed that the two kinase-deficient Pak1 mutants, Pak1 L83,L86,R299 and Pak1 R299 were capable of reducing the levels of ERK1/2 activation in RBM(1-5) (Fig. 5, C and D), we decided to investigate if they could also alter the levels of constitutive Elk-1 activation. Stable cell lines overexpressing each of the Pak1 mutants were transiently transfected with GAL-c-Elk-1 plus pFR-Luc plus Renilla as described above. Stable cell lines overexpressing wild type Pak1 (Pak1) did not display altered levels of Elk-1 activation in all five RBM cell lines examined (Fig. 6C). Both kinase-deficient Pak mutants, Pak1 R299 and Pak1 L83,L86,R299 , however, were efficacious in significantly reducing the levels of Elk-1 activation in RBM1 and -2 but not in RBM3, -4, and -5 (Fig. 6C). The hyperactive mutant, Pak1 L83,L86 , significantly increased the level of Elk-1 activation in RBM2 and -3 but not in RBM1, -4, and -5. Thus, while Pak1 seems to play a definitive role in ERK1/2 activation in RBM(1-5), its role in Elk-1 activation is more complex. Pak1 appears to influence Elk-1 activation in RBM1, -2 and -3, but is ineffective in RBM4 and -5, suggesting that there may be other kinases operating in these two cell lines lying outside of the Pak pathway (39,40) which influence the level of activation of Elk-1. These results are consistent with those for ERK activation where Pak1 L83,L86,R299 and Pak1 R299 were less effective in RBM4 and -5 ERK down-regulation, thereby indicating that these two cell lines may be differentially regulated compared with RBM1-3.
RBM (1)(2)(3)(4)(5) Cell Proliferation is PKC-and PI 3-Kinase-dependent But MEK-independent-The MAPK pathway is essential in cellular growth and differentiation (44). Thus, we examined the role of the constitutive activation of ERK1/2 on RBM(1-5) cell growth. To investigate the role of MEK in RBM cell proliferation, cells were treated with PD 98059 (20 M). This MEK inhibitor failed to significantly reduce the cell number (using the MTT assay) or DNA synthesis (using [ 3 H]thymidine incorporation) (Fig. 8). PD 98059 was effective, however, in significantly reducing HT-29 cell growth (data not shown). We also investigated the role of PKC, and PI3K in RBM(1-5) cell proliferation. PKC inhibition with GF 109203X (5 M) or pretreatment with PMA (100 nM) for 48 h which down-regulates PKC (27) significantly reduced cell growth (Fig. 8). Similar results were observed when cells were treated with the PI3K inhibitors, LY294002 (20 M) and wortmannin (100 nM). Wortmannin is unstable in solution and PI3K levels remain suppressed for only 9 -12 h after drug addition (45), thus cells were washed and pretreated a second time with wortmannin (100 nM) for another 12 h. Thus, ERK1/2 constitutive activation in RBM (1)(2)(3)(4)(5) correlates with cellular proliferation. DISCUSSION The present results have identified for the first time that the mitogen-activated protein kinases ERK1 and -2 as well as the transcription factor Elk-1 are constitutively activated in esophagogastric RBM cell lines. To our surprise, the observed activation of ERKs was not accompanied by MEK activation. Until now, the activated status of ERK has been shown to be predominantly accompanied by MEK activation (33,34). One previous report has documented the importance in Swiss 3T3 fibroblasts of MEK activation for the initiation of MAPK activation by PDGF but also emphasize that prolonged MAPK activation by PDGF is MEK-independent (46). Other reports documenting constitutive ERK activation have shown it to be synonymous with hyperexpression of either MEK or ERK (21,22,34). In contrast to all of these reports, however, we have failed to make any correlation between constitutive ERK activation and ERK or MEK hyperexpression or indeed with MEK activation. It may be possible that esophagogastric RBM cells may display elevated levels of receptor tyrosine kinases, or intermediate molecules which may explain the constitutive activation of ERKs observed in these cells. These questions are currently being addressed in our laboratory. Indeed, it has recently been shown that growth factor receptor tyrosine kinases are overexpressed in esophageal squamous cell carcinomas relative to normal tissue (47).
Although the specific role of ERKs in esophagogastric RBM cells is not known, it is reasonable to speculate that chronic activation of the ERK signaling cascade may influence the local metastatic environment to favor progression of esophagogastric metastatic cancer. Indeed, constitutive ERK activation has previously been shown to confer a tumorigenic and metastatic potential (48) as well as regulate cell motility and matrix metalloproteinase expression (49). Our data demonstrate that selective inhibition of the upstream kinase MEK does not affect esophagogastric RBM cell growth. This is in contrast to recent studies whereby inhibition of MEK attenuated the in vivo invasiveness of head and neck squamous cell carcinoma (50) and suppressed the growth of colon tumors in vivo (51). Pharmacological inhibition of PI3K and PKC, however, both proved efficacious in significantly reducing esophagogastric RBM cell growth as well as ablating ERK activation (52). Our studies revealed that while PI3K seems to be necessary for ERK activation and esophagogastric metastatic cell growth, it does not require Raf activation as Raf was found not to be activated in these cell lines. Furthermore, upon examination of the K-ras gene in RBM cells, we did not detect the existence of point mutations .
2 This is in agreement with a previous report demonstrating the absence of K-ras gene mutations in a series of cell lines established from esophageal cancer (2).
The involvement of PI3K lead us to investigate the role of the serine-threonine kinase Pak (p65 Pak ) a known effector of PI3K (25). The kinase-deficient Paks (Pak1 R299 and Pak1 L83,L86,R299 ) both significantly reduced the phosphorylation levels of ERK1 and -2. Interestingly, we observed that hyperactive Pak (Pak1 L83,L86 ) could further enhance the activation state of ERKs in esophagogastric RBM cells. This activation by Pak was found to be sensitive to PI3K inhibition but not MEK inhibition. Although Pak is not usually associated with ERK activation, a recent report has shown that Pak can cooperate with Raf to activate ERK in a cross-cascade activation (26). It remains to be investigated whether Pak is interacting with Raf within RBM (1)(2)(3)(4)(5). However, the absence of MEK involvement in Pak1 signaling in RBM(1-5) ERK activation reveals a novel signaling pathway in these cells. Although Pak-induced JNK activation has also been previously demonstrated (26), our study shows neither JNK nor p38 were involved in the signaling cascade in RBM (1)(2)(3)(4)(5).
Abundant data indicate that constitutive activation of the ERK cascade increases the expression of PAC1, a member of the MAPK phosphatase family (14,42) providing a pivotal role for this phosphatase in feedback inactivation of the stimulated ERK signaling pathway. PAC1 exhibits stringent substrate specificity for ERK and constitutive PAC1 expression inactivates ERK. While PAC1 expression is associated predominantly with hematopoeitic cells (14), we have now shown that it is also present in CACO-2, HT-29, SW480, OC1, and -3 cells and can be up-regulated in the presence of serum. We failed, however, to detect PAC1 gene expression in RBM cell lines, thus ruling out the possibility of a role for this phosphatase in the observed ERK constitutive activation. Two other protein phosphatases, MKP-1 and MKP-2 (17,18), may further explain the constitutive activation of ERKs in RBM (1)(2)(3)(4)(5). Previous reports have shown that both MKP-1 and -2 expression is up-regulated upon ERK activation (16,43). Moreover, up-regulation of normal MKP-1 mRNA and protein has been implicated in clinical specimens of a group of early stage carcinomas and in various stages of breast and prostate carcinoma (54,55). MKP-1 may also act as a tumor suppressor (54) as overexpressed MKP-1 has been shown to down-regulate ras-dependent mitogenic signals (53). In contrast to these reports, however, we have observed a significant decrease in MKP-1 expression in RBM(1-5) while MKP-2 expression remained unaltered. Although the significance of MKP-1 down-regula-2 O. P. Barry and S. Aarons, unpublished observation. tion and its mechanistic basis require additional investigation, we may speculate that an abnormal down-regulation mechanism of ERKs may exist in RBM(1-5) involving MKP-1. In conclusion, although the complexity of the signaling cascade which converges on ERK seems daunting, delineation of this pathway may uncover new cellular targets for the pharmacological manipulation of esophagogastric metastatic RBM cells.