|Home | About | Journals | Submit | Contact Us | Français|
The receptor tyrosine kinases (RTKs) RET, MET, and RON all carry the Metp+1loop→Thr point mutation (i.e., 2B mutation), leading to the formation of tumors with high metastatic potential. Utilizing a novel antibody array, we identified constitutive phosphorylation of STAT3 in cells expressing the 2B mutation but not wild-type RET. MET or RON with the 2B mutation also constitutively phosphorylated STAT3. Members of the EPH, the only group of wild-type RTK that carry Thrp+1loop residue, are often expressed unexpectedly in different types of cancers. Ectopic expression of wild-type but not Thrp+1loop→Met substituted EPH family members constitutively phosphorylated STAT3. In both RTKMetp+1loop with 2B mutation and wild-type EPH members the Thrp+1loop residue is required for constitutive kinase autophosphorylation and STAT3 recruitment. In multiple endocrine neoplasia 2B (MEN-2B) patients expressing RETM918T, nuclear enrichment of STAT3 and elevated expression of CXCR4 was detected in metastatic thyroid C-cell carcinoma in the liver. In breast adenocarcinoma cell lines expressing multiple EPH members, STAT3 constitutively bound to the promoters of MUC1, MUC4, and MUC5B genes. Inhibiting STAT3 expression resulted in reduced expression of these metastasis-related genes and inhibited mobility. These findings provide insight into Thrp+1loop residue in RTK autophosphorylation and constitutive activation of STAT3 in metastatic cancer cells.
Growth factor receptors are transmembrane protein tyrosine kinases playing critical roles in biological processes, including proliferation, survival, and migration. The core region of the kinase domain, comprising the catalytic loop, the activation loop, and the p+1 loop, is well conserved in receptor tyrosine kinases (RTKs) of different species. RTKs with different types of mutations within these three loops have been associated with developmental disorders and cancer. The point mutation (ATG→ACG), resulting in replacement of methionine with threonine within the p+1 loop, is associated with aggressive tumors. RET is an RTK that can activate a variety of signaling pathways, including the RAS/ERK, PI3K/AKT, and phospholipase Cγ pathways and plays an important role in neuron survival or differentiation (11). RET with a Metp+1loop→Thr substitution (RETM918T) is associated with the multiple endocrine neoplasia 2B type (MEN-2B) syndrome; this substitution is defined as the 2B mutation (11). In MEN-2B patients, the tumors derived from thyroid C cells are often more aggressive than C-cell tumors that develop in MEN-2A patients who carry mutations in the extracellular domain of RET (11, 22). Similarly, the 2B mutation in HGF receptor MET (MetM1268T) has been identified in metastatic renal carcinomas (10, 24). Introduction of the 2B mutation in other RTKs, such as RON and epidermal growth factor receptor (EGFR), caused transformation of NIH 3T3 cells with high metastatic potential (20, 23). Although the 2B mutation enhanced kinase activity and such a mutation has been suspected as a gain-of-function mutation (21, 29, 35), the role of the Thrp+1loop residue in RTK catalytic activity in recruiting specific substrate(s) responsible for the metastatic phenotype has not been clarified.
The importance of autophosphorylation at conserved tyrosine residues within the activation loop on kinase activity, as well as on substrate recruitment, has been well established over recent years. The p+1 loop represents a small motif, residing immediately downstream of the activation loop. It has been implicated to play a role in recognizing the residues next to tyrosine to be phosphorylated in the substrate (36). However, the precise role of the p+1 loop on the catalytic activity resulting in RTK autophosphorylation and substrate selection remains largely unknown. To identify signaling factor(s) preferentially activated by RTK carrying the 2B mutation, a novel antibody array technology was used. We show here that the oncogenic STAT (1), STAT3, was constitutively activated by different RTKMetp+1loop 2B mutations. The ephrin type receptor (EPH) and ligand ephrin system has been implicated in the regulation of many critical events during developmental patterning processes, including axonal guidance, cell adhesion, and cell migration. Wild-type EPHs are RTK that contain the Thrp+1loop residue (25). As predicted, wild-type EPH members activated STAT3 in the absence of their ligands, whereas the Thrp+1loop→Met substitution severely impaired this effect. We provide evidence that the Thrp+1loop residue plays a critical role in kinase tyrosine autophosphorylation and subsequent STAT3 recruitment in a ligand-independent manner. Moreover, STAT3 constitutive activation is associated with expression of the CXCR4 chemokine receptor and multiple mucin isoforms. Temporary depletion of STAT3 by small interfering RNA (siRNA) transiently inhibited expression of these metastasis-related genes and was shown to be invasive in a Matrigel assay.
Three hundred antibodies (Santa Cruz Biotechnology, Inc.) were spotted on nitrocellulose membranes (2.5 by 5 cm) manually at a dose of 40 ng/spot. The antibody-printed nitrocellulose membranes were incubated with 3% bovine serum albumin at room temperature for 2 h prior to incubation for 3 h at room temperature with whole extracts prepared from NIH 3T3 cells (5 × 107) stably transfected with wild-type RET or RET-2B. After three washes with Tris-buffered saline plus Tween, the arrays were blotted with horseradish peroxidase (HRP) conjugated anti-phosphotyrosine (pY20) antibody for 3 h, followed by enhanced chemiluminescence (ECL) analysis.
Cell invasion was assayed by using a Boyden chamber assay. In brief, polycarbonate membranes (8.0-μm pore size) were coated with 5% Matrigel in the upper compartment of Transwell culture chambers. Then, 250-μl portions of cells (5 × 105) were suspended in serum-free Dulbecco modified Eagle medium (DMEM) were placed in the upper compartment, and the lower compartment of the chamber was immediately filled with 500 μl of DMEM supplemented with 1% fetal bovine serum. After 16 h of incubation, the membranes were fixed with methanol and stained with hematoxylin and eosin (H&E). Cells located on the upper surface of the filter were completely removed by wiping the filter with a moist cotton swab; cells that had invaded the Matrigel and migrated through the membrane to the lower surface were counted by using a light microscope. Each assay was repeated at least three times.
Chromatin preparation and chromatin immunoprecipitation (ChIP) experiments were performed in accordance with the protocol from Upstate Biotechnology. A single-step PCR was used to amplify the MUC1, MUC4, and MUC5B promoters. The PCR conditions were optimized so that amplification was within the linear range for each primer pair. The following primers were used: MUC1-promoter, f-primer (5′-AGAGCAACGGGTGTATCGG-3′) and r-primer (5′-GCAGTGTGAGGAGCAGACG-3′); MUC4-promoter, f-primer (5′-AGAGCAACGGGTGTATCGG-3′) and r-primer (5′-GCAGTGTGAGGAGCAGACG-3′); and MUC5B-promoter, f-primer (5′-GCTTTGCCATCTAGGACGG-3′) and r-primer (5′-CCACGTGTGTTTGCTCTCG-3′). The amplicons were detected by staining with ethidium bromide on a 2% agarose gel.
A double-stranded siRNA oligonucleotide against STAT3 (5′-AACAUCUGCCUAGAUCGGCUAdTdT-3′ and 3′-dTdTGUAGACGGAUCUAGCCGAU-5′) was provided by Dharmacon Research, Inc. Lipofectamine 2000 (Invitrogen, Carlsbad, Calif.) was used as the transfection reagent according to the manufacturer's directions with 150 nmol of siRNA per well in a six-well dish. A scrambled siRNA was used as the control. siRNA transfected cells were incubated for 48 h in DMEM supplemented with 10% fetal bovine serum.
To characterize proteins constitutively phosphorylated by RTK with the 2B mutation, we manufactured antibody arrays by spotting 300 antibodies against different signaling molecules on nitrocellullose membranes (31, 33). Protein extracts were prepared from NIH 3T3 cells transfected with wild-type RET or RETM918T and incubated with the antibody arrays, followed by anti-phosphotyrosine (pY20-HRP)/ECL immunoblot analysis. Although some signaling proteins, such as ERK1/2 and AKT, were detected in both samples, STAT3, SHP-1, and HDAC2 were only detected in RET-2B transfectants (Fig. (Fig.1A).1A). The results obtained with this antibody array were confirmed by immunoprecipitation combined with Western blotting (Fig. (Fig.1B).1B). We next examined whether other RTKs with the 2B mutation also constitutively activated STAT3. In 293T cells, constitutive phosphorylation of STAT3 was induced by transfection of MET-2B, RON-2B, or KIT-2B, but not by wild-type RTK constructs (Fig. 1C and D). STAT3 activation upon EGF treatment of cells overexpressing wild-type EGFR has been widely reported (33). Here we show that in the absence of EGF, STAT3 phosphorylation was detected in 293T cells when STAT3 was cotransfected with EGFR containing either the 2B mutation or the inhibitory C-tail truncated (EGFR-ΔY974) mutation (34), but not with wild-type EGFR (Fig. (Fig.1E).1E). EGFR-2B protein level in 293T transfectants required for STAT3 constitutive phosphorylation was lower than the endogenous EGFR level in A431 cells in which STAT3 was activated by EGF treatment (Fig. (Fig.1E1E).
The RTKMetp+1loop with the 2B mutation can constitutively phosphorylate STAT3, prompting the question of whether a wild-type RTKThrp+1loop can trigger constitutive STAT3 phosphorylation. Among 85 protein tyrosine kinase genes identified from the human genome database, 54 are of the receptor type as defined by encoding a protein with a predicted transmembrane domain. These can be divided into two groups, RTKMetp+1loop and RTKThrp+1loop, according to the presence of methionine or threonine within the p+1 loop of the tyrosine kinase domain (Fig. (Fig.2A).2A). Since leucine and isoleucine are hydrophobic residues, only three RTKs bearing Leup+1loop or Ilep+1loop (AXL, MER, and TYRO3) belong to RTKMetp+1loop group. The EPH family has 14 members constituting the only group of RTK bearing the Thrp+1loop residue except EphB6, a widely reported catalytic inactive form, in which Thrp+1loop residue is evolutionarily replaced by alanine (4). Different EPH family members, including EphA1, EphA5, EphB2, EphB3, and EphB4, were chosen to test for STAT3 activation. Constitutive phosphorylation of STAT3 was immediately detected in 293T cells transiently transfected with each of these EPH members (Fig. (Fig.2B).2B). When the Thrp+1loop→Met substitution was introduced into EphA5 and EphB2, STAT3 activation was largely abolished (Fig. (Fig.2B).2B). STAT3 phosphorylation was not affected after either cotransfection of wild-type EphB2 with its ligand EphrinB1 (Fig. (Fig.2C)2C) or by adding exogenous clustered Fc-Ephrin-B1 (data not shown).
The JAK family members are cytokine receptor-associated tyrosine kinases, responsible for STAT activation upon cytokine treatment. All JAK family members carry a tyrosine residue within the p+1 loop. JAK1 with Tyrp+1loop→Thr, Tyrp+1loop→Ser, and Tyrp+1loop→Met substitutions were constructed, and all of the mutants were compared to wild-type JAK1 for STAT3 activation. In 293T cells, alpha-interferon-stimulation-dependent STAT3 phosphorylation became constitutive after transient transfection of JAK1WT, JAK1Yp+1loopT, and JAK1Yp+1loopS constructs (Fig. (Fig.2D).2D). Among these three forms, JAK1Yp+1loopT exhibited the strongest activity in STAT3 phosphorylation. In contrast, JAK1 with Tyrp+1loop→Met substitution completely abolished its ability to phosphorylate STAT3 (Fig. (Fig.2D).2D). Therefore, the Thrp+1loop residue in protein tyrosine kinases plays an essential role in constitutive phosphorylation of STAT3.
To further characterize the role of Thrp+1loop in kinase activity, we assayed for tyrosine autophosphorylation of RTK. In 293T cells, RET tyrosine autophosphorylation was only detected in cells transfected with RET-2B but not with wild-type RET, RET-2A, and RET-HS constructs (Fig. (Fig.3A).3A). RET-2A carries a Cys634→Arg mutation in the extracellular domain and causes MEN-2A, a less aggressive carcinoma predisposition syndrome than MEN-2B, whereas RET-HS with the Arg897→Glu mutation is responsible for Hirschsprung disease, a gut motility defect caused by an absence of ganglion cells in the nerve plexuses of the lower digestive tract (9, 18). Similarly, EGFR autophosphorylation was switched from EGF-dependent into EGF-independent with introduction of the 2B mutation or with the negative C-tail truncation (Fig. (Fig.3B).3B). Under the same conditions, different wild-type EPH family members all displayed constitutive tyrosine autophosphorylation following transfection in 293T cells (Fig. (Fig.3C).3C). However, the Thrp+1loop→Met substitution markedly reduced tyrosine autophosphorylation in both EphA5 and EphB2 (Fig. (Fig.3C3C).
Although STAT3 and EGFR-2B became constitutively phosphorylated in 293T cells transiently transfected with EGFR-2B, ERK1/2 phosphorylation was still EGF-stimulation dependent (Fig. (Fig.3D,3D, left panel). Unlike EGFR, EPH transient transfection led ligand-independent ERK1/2 activation in 293T cells (Fig. (Fig.3D,3D, right panel). However, it is noteworthy that the effect of Thrp+1loop→Met substitution on ERK1/2 activation was not as strong as that on STAT3 activation (Fig. (Fig.2B).2B). AKT was constitutively phosphorylated in 293T cells and transient transfection of EGFR or EPH did not change its phosphorylation pattern (Fig. (Fig.3D).3D). Taken together, these results indicate that the presence of Thr residue within the p+1 loop of RTK is more critical for STAT activation.
The 2B mutation was found to change the substrate specificity of RET from an RTK to a cytoplasmic tyrosine kinase by using peptides with defined sequences as substrates (20, 27). We compared the ability of mutant RTK-2B with wild-type RTK to recruit STAT3 in vivo. Although RET-2A was previously reported to activate STAT3 (28), among the four forms of RET tested here, RET-2B was most efficient in recruiting STAT3 proteins, a finding consistent with strong tyrosine autophosphorylation associated with this mutation (Fig. (Fig.4A).4A). As for EGFR, constitutive STAT3 association was detected with both EGFR-2B and EGFR-ΔY974 but not with wild-type EGFR (Fig. (Fig.4B).4B). In contrast, wild-type EphA5 or EphB2, but not Thrp+1loop→Met substituted constructs were coimmunoprecipitated with STAT3 in 293T transfectants (Fig. (Fig.4C).4C). Therefore, the Thrp+1loop residue is required for an RTK to become constitutively autophosphorylated, which in turn is required for a constitutive complex formation between RTK and STAT3, regardless of the absence or presence of the ligand.
Tyrosine phosphorylated STAT translocates into the nucleus where it binds to DNA and regulates transcription. In a STAT3-dependent luciferase activity assay, wild-type RTKs including MET, RON, and KIT barely induced STAT3-dependent transcription (Fig. (Fig.5A).5A). When the 2B mutation was introduced, all of these RTKs became highly active in STAT3-dependent transcription induction (Fig. (Fig.5A).5A). When EphA5 and EphB2 were tested, wild-type but not Thrp+1loop→Met mutated forms could efficiently induce STAT3-dependent transcription in the absence of their ligands (Fig. (Fig.5A).5A). We next examined EGFR. Consistent with these EGFR autophosphorylation and STAT3 activation patterns, STAT3-activated transcription switched from EGF-stimulation dependent to constitutive when an EGFR was introduced with the 2B mutation (Fig. (Fig.5B).5B). Tyrosine mutated STAT3 (Y705F) dominant negatively blocked EGFR-2B activity in STAT3-dependent transcription (Fig. (Fig.5B).5B). To evaluate the dose effect on STAT3 luciferase activity, different doses of wild-type RET and RET-2B were transfected in 293T cells. As can be seen in Fig. Fig.5C,5C, significantly higher STAT3 transcriptional activation was induced in a dose-dependent manner with RET-2B compared to wild-type RET. Together, these findings further support the model that a Thrp+1loop type RTK is independent of ligand binding for its catalytic activity.
To further explore the consequence of STAT3 activation in cancer cells expressing a RTK with the 2B mutation, fine-needle aspiration biopsies were obtained from MEN-2B patients in whom the ret gene mutation (RETM918T) was confirmed by PCR analysis. Tumors obtained from liver, lymph nodes, and thyroid glands from MEN-2B patients were stained with anti-STAT3 antibody or anti-calcitonin antibody. Positive nuclear staining for STAT3 was found in tumor metastases in the liver and lymph nodes (Fig. (Fig.6A).6A). We then examined the correlation between EPH expression and STAT3 activation in different breast cancer cell lines. EPH family members EphA1, EphA3, EphA5, and EphB4 were all expressed in several breast cancer cell lines, including MCF-7, MB-468, and T47D (Fig. (Fig.6B,6B, right panel). In these breast cancer cell lines, STAT3 was constitutively phosphorylated in contrast to A431 cells in which STAT3 was phosphorylated only upon EGF treatment (Fig. (Fig.6B,6B, left panel). Although not constitutively phosphorylated, mitogen-activated protein kinases ERK1/2 were activated in response to EGF treatment in all of the three breast cancer cells (Fig. (Fig.6C).6C). Moreover, none of the specific inhibitors of the JAK, EGFR, or Src alone was efficient to block STAT3 phosphorylation in MB-468 or MCF-7 cells (Fig. (Fig.6D),6D), further suggesting that multiple rather than single tyrosine kinases are responsible for STAT3 constitutive phosphorylation in these breast cancer cell lines.
Activated STAT3 has been previously suggested to play a role in unregulated cell proliferation, resistance to apoptosis, and metastasis in breast tumor cells (12, 19, 38). However, suppression of STAT3 expression in up to 90% of the cells with siRNA did not show any apparent effect on proliferation or apoptosis of these three cell lines (data not shown). Using cDNA microarrays to probe for differential gene expression, we identified several genes, including CXCR4, and mucins involved in metastasis that were upregulated in the cells expressing RET-2B compared to the cells expressing wild-type RET (data not shown). We then searched for promoters containing STAT3-interacting element (SIE) by using the TFSEARCH computer program. Putative SIE sequences have been identified in the promoter of CXCR4 chemokine receptor, as well as the immediate 5′ upstream region of various mucin genes such as MUC1 (2), MUC4, and MUC5B (Fig. (Fig.7A).7A). Although CXCR4 plays a critical role in guiding motility, invasion, and cell survival at specific metastatic sites, mucins are high-molecular-weight glycoproteins that can promote adenocarcinoma cell invasion, as well as metastasis, and modulate the immune recognition of cancer cells (3, 18, 30). EGF-activated STAT3 in A431 cells bound to each of these SIE sequences in CXCR4, MUC1 (2), MUC4, and MUC5B promoters as revealed in an electrophoretic mobility shift assay (Fig. (Fig.7B).7B). To determine whether STAT3 directly regulates CXCR4 promoter activity, we performed a transient-transfection assay in A431 cells with a CXCR4-Luc reporter. Two reporters, one with the region from −1050 to −100 [CXCR4(−1050)-Luc] and the other with the region from −970 to −100 [CXCR4(−970)-Luc] of the CXCR4 promoter, which lacks the SIE motif (TTCTCCGAA−1002), were used. Reporter activity was measured in the absence or presence of EGF. EGF-induced CXCR4-Luc reporter activity with CXCR4(−1050)-Luc was apparently attenuated with CXCR4(−970)-Luc, which lacks STAT3 binding site (Fig. (Fig.7C),7C), indicating that STAT3 is required for a full activation of CXCR4 gene.
To confirm whether constitutively phosphorylated STAT3 in these breast cancer cell lines was associated with these promoters, we performed ChIP assays with isolated nuclei. Although the CXCR4 promoter was not detected, all three mucin-gene promoters were detected in anti-STAT3 immunoprecipitates obtained from these three breast cancer cell lines, in contrast to A431 cells, in which complex formation between STAT3 and CXCR4 or mucin gene promoters was EGF dependent (Fig. (Fig.7D).7D). Although MUC1 and MUC4 proteins were detected in all of the three breast cancer cell lines, MUC5B was detected as a large and a small form presumably generated by protease digestion (32) in MB-468 cells or as a small form in T47D cells. Western blotting analysis showed that protein levels of MUC1, MUC4, and MUC5B were all reduced upon treatment with STAT3 siRNA for 72 h, suggesting that these mucin genes are most likely under STAT3 regulation in these breast cancer cell lines (Fig. (Fig.7E).7E). No expression of these mucin isoforms was detected by immunohistochemistry of medullary thyroid carcinoma in this MEN2B patient (data not shown). However, in the metastasis in the liver, a tumor embolus in a blood vessel and metastatic implants in the liver were strongly positive for CXCR4 (Fig. (Fig.7F),7F), a finding consistent with STAT3 staining pattern illustrated in Fig. Fig.6A6A.
Since increased cell motility and invasion are correlated with increased metastatic potential, we investigated the invasive ability of these human breast carcinoma cell lines in a transwell chamber assay. MCF-7, MB-468, and T47D all showed 45 to 50% reduction in migration following down regulation of STAT3 with siRNA (Fig. (Fig.8A).8A). NIH 3T3 cells transfected with RET-2B or wild-type EphB2 constructs showed 50% more invasion than cells transfected with wild-type RET or EphB2T703M (Fig. 8B and C). Cotransfection of STAT3Y705F restrained this effect of RET-2B and EphB2T703M on cell invasion (Fig. (Fig.8B).8B). Together, these results indicate that constitutive phosphorylation of STAT3 can activate expression of various genes involved in invasion and metastasis in different types of tumor cells.
Our results demonstrate that the Thrp+1loop residue is critical for tyrosine autophosphorylation of RTKs, leading to constitutive phosphorylation of STAT3. For RTKMetp+1loop, the binding of a ligand triggers dimerization that promotes autophosphorylation and changes the activation loop from a locked conformation into an open one (7). Autophosphorylation of conserved Tyr within the activation loop then plays a critical role in substrate recruitment and downstream activation (8). For RET-2B, however, dimerization is not necessary for tyrosine autophosphorylation and substrate activation to take place (11). In EGFR, as well as many other receptors, the C-terminal tail functions to negatively regulate its intrinsic kinase activity, and C-tail-truncated EGFR becomes constitutively active (34). Thus, for RTKMetp+1loop, ligand binding and subsequent receptor dimerization are absolutely required steps for releasing this block from the activation loop. As for the RTKMetp+1loop in the 2B mutation or wild-type EPH family members, the Thrp+1loop residue, most likely posttranslationally modified, keeps the RTK in an unlocked conformation.
The p+1 loop was so named for its role in contacting the residue immediately COOH terminal to the phosphorylated tyrosine (the p+1 position of the residue) in the substrate (8). EGFR with the 2B mutation was associated with a decrease in the selectivity of the kinase for Phe and an increase in the selectivity for acidic residues (Glu or Asp) at the p+1 position compared to wild-type EGFR (20). The motif of pTyr705 residue in STAT3 is “YLK,” which does not obey this rule. If the side chain of Arg409 site inserts into a pocket near Thr429 site of the p+1 loop in Src, as suggested by X-ray crystallography (36), Thr429 site is most likely phosphorylated in order to form a salt bridge with residue Arg409. The substitution of the Arg residue at the same position of the activation loop of RET (Arg897→Glu) in Hirschsprung disease is a loss-of-function mutation. However, in this case, the Metp+1loop site may not form a bridge with Arg897. The results from previous studies and the present study clearly show that RET tyrosine autophosphorylation and STAT3 activation were not detected in RET-HS, providing additional evidence that the p+1 loop forms a specific conformation with the Arg residue in the activation loop required for Tyr autophosphorylation and catalytic activity of the kinase (35, 36).
The RTKMetp+1loop-2B is a gain-of-function mutation that is associated with activation of STAT3, SHP-1, and HDAC2 (Fig. (Fig.1);1); however, this 2B mutation is rare. The large EPH family represents a naturally occurring “2B” form of RTK. Although EPH receptors play roles during vertebrate cranial development and neural crest cell migration from hindbrain segments to specific branchial arches, many EPH members have been overexpressed in cancer cells with high metastatic potential (14, 39). Our results strongly indicate that ligand binding and receptor dimerization are apparently not required for EPH receptor autophosphorylation and subsequent substrate recruitment and activation. This conclusion is supported by the crystallographic analysis of EphB2 revealing that the overall structure of the ligand-binding domain of EphB2 in the complex is similar to that of the unbound EphB2 (6). It has been reported that wild-type EPH transformed cells, whereas the stimulation with its ligand actually reversed the oncogenic phenotype (17). Thus, ligand-free activation of specific signaling pathways is sufficient for EPH to fulfill its specific functions in vivo under conditions when the ligand is not available.
Studies of STAT3 conditional knockout mice indicate that STAT3 plays a role in migration rather than in proliferation of keratinocytes (13). In the ovary of Drosophila, the JAK-STAT pathway is required to convert the stationary epithelial cell into a migratory or invasive cell (26). Recently, dominant-negative STAT3 was shown to block human endothelial cell migration (37). However, downregulation of STAT3 protein by 90% did not completely block invasion (Fig. (Fig.5),5), suggesting that STAT3 is not the sole factor responsible for metastasis. Chemokine receptor CXCR4 and mucin isoforms have been implicated in metastasis in different types of cancer, but the mechanisms involved may be quite different. SIE (TTCxxxGAA) sequence presents within the region from −500 to −100 of all three mucin (MUC1, MUC4, and MUC5B) promoters or within the region −1010 to −1000 of the CXCR4 promoter. All of these SIE sequences bound to STAT3 in vitro. Constitutively activated STAT3 preferentially bound to the mucin promoters in the breast cancer cells, whereas CXCR4 stained positively in C-cell carcinoma. Hence, STAT3 may differentially activate these genes in breast cancer cell lines and neuroendocrine C-cell carcinoma. It provides additional evidence that STAT3 requires organ- or cell-specific factors for differential regulation of these genes. Extracellular signal-activated NF-κB was reported to bind to the promoters of CXCR4, MUC1, and MUC2 genes (5, 15, 16). Thus, constitutively activated STAT3 and NF-κB may have a synergistic effect in the regulation of these genes for constant expression. Taken together, our findings indicate that targeting the p+1 loop of RTK may provide a novel therapeutic intervention to curb metastasis in both adenocarcinoma and neuroendocrine cancer cells.
We thank S. M. Jhiang for wild-type and mutant RET constructs, P. Accornero and K. Furge for mouse Met constructs, A. Danilkovitch-Miagkova for human RON construct, R. Arceci for mouse KIT construct, and E. B. Pasquale, B. Wang, and R. Zhou for various EPH and ephrin constructs. Patient samples were obtained from Rhode Island Hospital and Mayo Clinic Foundation.
This study was partially supported by NIH RO1 grant (CA82549) to Y.E.C. and NIH COBRE grant (RR-15578) to Brown University.