|Home | About | Journals | Submit | Contact Us | Français|
Persistently activated or tyrosine-phosphorylated STAT3 (pSTAT3) is found in 50% of lung adenocarcinomas. pSTAT3 is found in primary adenocarcinomas and cell lines harboring somatic-activating mutations in the tyrosine kinase domain of EGFR. Treatment of cell lines with either an EGFR inhibitor or an src kinase inhibitor had no effect on pSTAT3 levels, whereas a pan-JAK inhibitor (P6) blocked activation of STAT3 and inhibited tumorigenesis. Cell lines expressing these persistently activated mutant EGFRs also produced high IL-6 levels, and blockade of the IL-6/gp130/JAK pathway led to a decrease in pSTAT3 levels. In addition, reduction of IL-6 levels by RNA interference led to a decrease in tumorigenesis. Introduction of persistently activated EGFR into immortalized breast epithelial cells led to tumorigenesis, IL-6 expression, and STAT3 activation, all of which could be inhibited with P6 or gp130 blockade. Furthermore, inhibition of EGFR activity in multiple cell lines partially blocked transcription of IL-6 and concurrently decreased production and release of IL-6. Finally, immunohistochemical analysis revealed a positive correlation between pSTAT3 and IL-6 positivity in primary lung adenocarcinomas. Therefore, mutant EGFR could activate the gp130/JAK/STAT3 pathway by means of IL-6 upregulation in primary human lung adenocarcinomas, making this pathway a potential target for cancer treatment.
Lung cancer is the leading cause of cancer deaths in the United States, resulting in approximately 160,000 deaths per year (1). Non–small cell lung cancer (NSCLC) accounts for more than 85% of all lung cancers. Adenocarcinomas, a major subtype of NSCLC, represent the most common histologic subtype of lung cancer in the United States and many countries worldwide. Some adenocarcinomas distinguish themselves by overexpressing wild-type EGFR and EGFR containing kinase domain–activating mutations (2–5).
EGFR is a membrane-bound receptor tyrosine kinase that belongs to the ErbB subfamily. On ligand binding, EGFRs initiate activation of a series of cellular signal transduction pathways that regulate cell proliferation and survival. The EGFR either is mutated or shows altered expression in a variety of human cancers, including carcinomas of the lung, breast, head and neck, ovary, and bladder as well as gliomas (6). Somatic-activating mutations in the tyrosine kinase domains of the EGFR are found in approximately 10% of lung adenocarcinomas (3, 4, 7). Most lung adenocarcinoma–associated somatic EGFR tyrosine kinase domain mutations fall into 1 of 2 classes: in-frame deletions in exon 19 that eliminate the conserved LREA motif (ΔEGFR) and a T-to-G base substitution in exon 21 that substitutes arginine for leucine at position 858 (L858R) (3). Patients whose tumors contain either of these 2 classes of mutations have increased sensitivity to the tyrosine kinase inhibitors (TKIs) gefitinib (ZD-1839 [ZD]) and erlotinib. In addition, lung cancer–derived cell lines harboring these kinase domain–activating mutant forms of the EGFR are sensitive to TKIs (8–10).
One of the important signaling mediators downstream of normally and abnormally activated EGFR is STAT3. STAT3 is a latent transcription factor found in the cytoplasm of cells. It is activated by tyrosine phosphorylation, leading to dimerization, nuclear translocation, DNA binding, and gene activation (11). In addition to EGFR, STAT3 can be activated by other receptor and nonreceptor tyrosine kinases, such as the IL-6/gp130 receptor family, PDGFR, src kinase, and JAKs. STAT3 is transiently activated in normal cells but constitutively activated in a wide variety of hematologic and epithelial primary tumors and tumor-derived cell lines, such as leukemias; lymphomas; head and neck, prostate, ovarian, breast, and renal cell cancers; and melanoma (12, 13). STAT3 is also persistently activated in about 50% of NSCLC primary tumors and lung cancer–derived cell lines (14–16). STAT3 is critically involved, primarily through differential gene regulation, with almost all aspects of tumorigenesis, including cell cycle progression, tumor invasion and metastasis, host immune system evasion, and tumor angiogenesis (17–19).
IL-6 belongs to the IL-6 cytokine family, which also contains oncostatin M (OSM), leukemia inhibitory factor (LIF), IL-11, and ciliary neurotropic factor (CNTF) (20). IL-6 regulates immune and inflammatory responses, but recent reports suggest that IL-6 expression is implicated in the regulation of tumor growth and metastatic spread, including lung cancers (21, 22). By combining with their ligand-specific binding subunit, IL-6 and other IL-6 family cytokines use the common receptor gp130 as the signaling subunit to activate receptor-associated tyrosine kinases JAK1 and JAK2 as well as tyrosine kinase 2 (Tyk2) (20). JAK activation leads subsequently to phosphorylation of gp130 and recruitment and phosphorylation of signaling molecules, such as STAT3, MAPK, and PI3K (13, 21, 23).
A positive correlation between elevated EGFR activity, including wild-type EGFR and kinase domain–activating mutant EGFR (referred to as mutant EGFR), and pSTAT3 has been described in many human primary tumor specimens and tumor-derived cell lines, including NSCLC as well as breast and head and neck tumors (9, 14–16, 24–27). Previous studies have demonstrated that TKI, src, or JAK2 inhibition of mutant EGFR–expressing cell lines did not lead to inhibition of STAT3 activation. Thus, the mechanisms of STAT3 phosphorylation and the phenotypic consequences of inhibiting STAT3 activity in either mutant EGFR–transfected cell lines or cancer-derived cell lines expressing mutant EGFR remain unclear (8, 14, 28).
In this study, we discovered a strong correlation between pSTAT3 and adenocarcinomas harboring somatic-activating mutations in the tyrosine kinase domain of the EGFR (ΔEGFR and L858R). We determined that STAT3 activation in mutant EGFR–expressing cell lines occurs, in part, through mutant EGFR–mediated transcriptional upregulation of the IL-6 gene, which, in turn, activates the gp130/JAK signaling pathway. Blockade of this pathway inhibits cell-cycle progression and cell growth and represses tumorigenesis.
Immunohistochemical analysis of tissue microarrays (TMAs) of primary lung adenocarcinomas (92 tumor specimens) showed that 3.3% contained high (+3), 17.4% contained moderate (+2), and 29.3% contained low (+1) levels of nuclear pSTAT3, while 50.0% had no (0) pSTAT3 (Figure (Figure1).1). Given the clear distinction between a staining score of 0 and +1, a score of +1, +2, or +3 was regarded as positive, and a score of 0 was regarded as negative. A similar distribution and grading system for pSTAT3 levels has been described by others (14–16). As can be seen in Table Table1,1, with the use of the Fisher exact test and Mann-Whitney U test for statistical analyses, there was a strong correlation between the presence of mutant EGFRs and pSTAT3 (P = 0.002). Other positive correlations existed between pSTAT3 and (a) pEGFR, pAKT, and pERK1/2 (pMAPK); (b) decreased expression of caspase-3 protein; (c) non-solid tumor; (d) small tumor size (≤ 3 cm); and (e) thyroid transcription factor 1 (TTF1) expression. A positive correlation between pSTAT3 levels and tumor size, pEGFR, decreased apoptotic bodies, and well-differentiated tumors has been described previously (14, 15). pSTAT3 positivity on immunohistochemical analysis did not correlate with Ki67 protein expression; amplification of the EGFR gene (chromogen in situ hybridization [CISH]); or patient age, sex, cigarette smoking pack year history, and survival (data not shown).
To begin exploring the mechanisms responsible for the elevated levels of pSTAT3 in mutant EGFR–expressing lung adenocarcinomas, we obtained lung adenocarcinoma–derived cell lines harboring mutant forms of EGFR, which are known to contain constitutively activated STAT3 (9, 14, 28). As can be seen in Figure Figure2,2, the 11-18 (L858R) (29), H1650 (exon 19 deletion, Δ746-750), H1975 (L858R), and H3255 (L858R, 11-fold amplification) cell lines all expressed high levels of typrosine-phosphorylated STAT3 (pSTAT3) by Western blot analysis. The exception was noted for control H460 (wild-type EGFR) cells (Figure (Figure2A).2A). The mechanism of STAT3 activation in these cell lines remains unclear, as inhibition of src and JAK2 activity did not affect pSTAT3 levels and inhibition of EGFR activity with the use of ZD reduced pSTAT3 levels in only a few of the cell lines (14, 28). We reexamined the role of these kinases in mediating STAT3 phosphorylation in the 4 cell lines described above. We recently described a novel pan-JAK inhibitor (P6) as being more specific and sensitive than AG490 in the treatment of myeloma cells (30). P6 markedly abrogated pSTAT3, while neither EGFR tyrosine kinase inhibition (ZD) nor src inhibition (dasatinib) had any effect on pSTAT3 levels (Figure (Figure2A2A and data not shown). P6 had a minimal effect on pEGFR levels, and it did not affect any of the 4 important tyrosine phosphorylation sites (i.e., tyrosine 845, 992, 1045, and 1068) on EGFR (Figure (Figure2A2A and data not shown). P6 had no effect on pMAPK levels in these cell lines, while an effect on pAKT was observed for the H1975 cell line (whose relative pAKT levels are quite low) (Figure (Figure2A).2A). Addition of EGF to the 11-18 and H3225 cells increased pEGFR and pMAPK levels, but it had no effect on pSTAT3 levels (Figure (Figure2B2B and data not shown). P6 had no effect on pEGFR levels but completely abrogated pSTAT3 levels, while ZD blocked pEGFR but did not affect pSTAT3 (Figure (Figure2B2B and data not shown). These data suggest that the EGF/EGFR pathway is independent of the JAK/STAT3 pathway.
STAT3 is known to support cellular transformation and sustain cell survival in a number of cancer-derived cell lines, including lung cancer (13, 14, 28). We subsequently determined whether P6-induced pSTAT3 inhibition had any effect on cancer cell growth. P6 treatment dramatically repressed cell proliferation of pSTAT3+ 11-18, H1650, and H1975 cells, but not of pSTAT3– H460 cells (Figure (Figure2C).2C). P6 had no effect on cell growth of H3255 cells. FACS analyses of P6-treated 11-18, H1650, and H1975 cells showed a cell-cycle arrest at the G2/M phase, while cleaved poly(ADP ribose) polymerase (PARP) levels and TUNEL assays failed to detect any apoptosis in these samples (data not shown). We hypothesize that the growth-inhibitory effects from P6 are principally due to blockade of STAT3 activity. We therefore introduced a STAT3 short hairpin RNA (shRNA) construct into the H1650 cells. This led to a partial decrease in STAT3 levels, which was associated with a 50% reduction in the proliferation of these cells in comparison with findings for scrambled control–infected cells (Figure (Figure2D). 2D).
We next tested the effect of P6 on tumorigenesis of these cell lines. Anchorage-independent soft agar assays were performed, and P6 strongly repressed colony formation of all pSTAT3-harboring cell lines 11-18, H1650, H1975, and H3255 but not of non–pSTAT3-harboring H460 cells (Figure (Figure3,3, A and B). The EGFR TKI ZD inhibited colony growth of H3255 cells, and partial inhibition was observed in H1975 cells (Figure (Figure3B).3B). To further assess the antitumorigenic effect of P6, 11-18 and H1650 cells were treated with P6 in vitro for 16 hours and subsequently injected subcutaneously into the flanks of nude mice. After 14 days, the P6-treated cells had grown more slowly compared with the control (DMSO-treated) cells, leading to tumors that were smaller in size and mass (Figure (Figure3,3, C and D).
The gp130/JAK pathway signaling has been shown to mediate STAT3 phosphorylation in many cancer-derived cell lines (22, 31–33). The IL-6 family cytokines, including IL-6, OSM, and LIF, share the same gp130-signaling receptor, leading to activation of JAK1, JAK2, and Tyk2 and subsequent phosphorylation of STAT3 (21, 23). To determine whether the IL-6/gp130 signaling pathway was involved in STAT3 activation, we treated the lung cancer–derived cell lines with either a functional blocking antibody to the gp130 receptor or an IL-6–sequestering antibody. Treatment with anti-gp130 antibody (B-R3) or anti–IL-6 antibody inhibited pSTAT3 in 11-18, H1650, and H3255 cells (Figure (Figure4A).4A). To determine whether paracrine production of IL-6 was responsible for activating the gp130 receptor in these cell lines, conditioned medium (CM) was isolated from 11-18 cells and applied to MCF-10A cells (which do not express pSTAT3) in the absence or presence of various blocking antibodies against gp130, IL-6, IL-6–specific receptor IL-6R, and other gp130 pathway cytokines, OSM and LIF. 11-18 CM induced high levels of pSTAT3, which was inhibited by gp130, IL-6 ligand, or IL-6 receptor antibodies, but not in control IgG-, anti-OSM-, or anti-LIF antibody–treated MCF-10A cells (Figure (Figure4B).4B). We examined mRNA levels for the IL-6 family of cytokines, including IL-6, OSM, LIF, IL-11, and CNTF, and only IL-6 mRNA was detected in 11-18, H3255, H1650, and H1975 cell lines (data not shown). IL-6–secreted protein levels were determined by ELISA in the CM collected from near-confluent cell cultures: 11-18, 1,800 pg/ml; H3255, 2,900 pg/ml; H1650, 7,700 pg/ml; and H1975, 7,800 pg/ml. These levels are high in comparison with the levels of normal cells or other cancer-derived cell lines, for which no more than 10–60 pg/ml are seen (M. Berishaj and J.F. Bromberg, unpublished observations). These findings demonstrate that IL-6 is being secreted from these cell lines, leading to activation of the gp130/JAK/STAT3 signaling pathway.
To determine the consequences of IL-6 blockade on growth, we introduced an IL-6 shRNA construct into the lung cancer–derived cell lines as the blocking antibodies led only to a transient inhibition of IL-6 signaling. IL-6 shRNA–expressing cell lines produced substantially less IL-6 compared with shRNA control–expressing cells as measured by ELISA of CM (Figure (Figure5A).5A). The consequences of decreased IL-6 production included inhibition of pSTAT3, as well as inhibition of growth in vitro and in vivo (Figure (Figure5,5, B–D).
ΔEGFR mediates IL-6 production, STAT3 activation, and cellular transformation of MCF-10A cells. We wanted to determine whether the expression and the activity of mutant EGFR could regulate IL-6 production. To address this question, we stably introduced and expressed ΔEGFR in immortalized, but not transformed, breast epithelial MCF-10A cells (Figure (Figure6).6). The ΔEGFR-containing cells expressed high levels of pEGFR and its downstream targets, including pSTAT3 and MAPK, compared with vector control–containing cells (Figure (Figure6A).6A). The ΔEGFR-expressing cells appeared larger, more irregular, and mesenchymal in nature compared with the “cobblestoned”-appearing control cells (data not shown). ΔEGFR-expressing MCF-10A cells grew in an anchorage-independent manner as well as in the flanks of immunocompromised mice, in contrast to control cells, demonstrating that the sufficiency of ΔEGFR in mediating tumorigenesis (Figure (Figure6,6, B and C). We next determined whether the ΔEGFR-expressing MCF-10A cells produced high levels of IL-6, leading to activation of STAT3, as was observed in lung adenocarcinoma–derived cell lines harboring mutant EGFR. The ΔEGFR–MCF-10A cells were treated with P6, gp130, and IL-6 blocking antibodies, which inhibited STAT3 activation. In contrast, EGFR inhibition (ZD) had no effect on pSTAT3 levels (Figure (Figure7A).7A). RT-PCR analysis revealed high IL-6 mRNA expression in ΔEGFR–MCF-10A cells relative to expression in control MCF-10A cells (Figure (Figure7B).7B). The CM collected from these cells stimulated STAT3 phosphorylation when applied to control MCF-10A cells, while the CM collected from vector control MCF-10A cells did not mediate STAT3 phosphorylation (Figure (Figure7C).7C). As determined by ELISA, ΔEGFR–MCF-10A cells produced high levels of IL-6 (900 pg/ml). Levels of OSM, LIF, and gp130 mRNA were also determined and were not elevated (data not shown). These findings suggest that ΔEGFR expression can upregulate IL-6 mRNA and protein expression, leading to STAT3 phosphorylation.
We demonstrated that transient inhibition of EGFR kinase activity in either cancer-derived cell lines or ΔEGFR–MCF-10A cells does not affect pSTAT3 levels (Figure (Figure2A2A and Figure Figure7A).7A). However, expression of ΔEGFR led to increased IL-6 production, resulting in phosphorylation of STAT3. How do we resolve these 2 observations? The IL-6 protein is stable (34), and we have demonstrated that IL-6 levels were between 900 and 8,000 pg/ml in CM derived from cell lines expressing mutant EGFR. As little as 50–100 pg/ml can mediate STAT3 phosphorylation in MCF-10A cells (data not shown). We hypothesized that transient inhibition of IL-6 production through EGFR kinase blockade would have no effect on pSTAT3 levels in cells bathed in high levels of IL-6-containing media. To examine the contribution of EGFR activity on IL-6 production, the existing IL-6 had to be removed. To test our hypothesis, we replaced the CM with fresh medium, washing away the existing IL-6 and treating cells with ZD, the EGFR TKI. De novo IL-6 production was inhibited in 11-18 and H1650 cells treated with ZD compared with DMSO control. In contrast, IL-6 levels from cells that did not undergo a medium change were unaffected by ZD treatment (Figure (Figure8A).8A). Of importance, pSTAT3 levels were inhibited only in cells that had undergone a medium change with ZD, in contrast with no medium change and ZD (Figure (Figure8B).8B). IL-6 production is regulated at multiple levels, including transcription, mRNA stability, posttranslational modification, and secretion. We examined the possible contribution of EGFR activity to the transcriptional control of the IL-6 gene. We first determined that treatment of 11-18 cells for 16 hours with ZD, preceded by a medium change, resulted in a decrease in IL-6 mRNA levels, as determined by semiquantitative RT-PCR and real-time PCR (Figure (Figure8C).8C). In contrast, IL-6 mRNA levels were unchanged in cells that had not undergone a medium change before the addition of ZD (data not shown). Finally, we examined the effect of ΔEGFR expression and activity on luciferase activity in an IL-6 promoter luciferase reporter assay. pGL3-IL-6-Luc contains 2,120 bp of promoter sequence with known binding sites for transcriptional factors such as C/EBPβ, CREB, AP1, and NF-κB (35). Cotransfection of ΔEGFR and this IL-6 promoter luciferase construct into NIH3T3 cells revealed an increase in luciferase activity, which was repressed by 50% with ZD (Figure (Figure8D).8D). These results support the suggestion that ΔEGFR activity can, in part, mediate transcriptional activation of the IL-6 gene and the subsequent production of IL-6 protein. The transcriptional regulation of the IL-6 gene is complex, involving multiple transcription factors, and our data suggest that other factors (possibly present in CM) can mediate its regulation. Nevertheless, our data suggest that overexpression of mutant EGFR protein may be the initiating step in the production of IL-6.
Given that the apparent mechanism of STAT3 phosphorylation is through the IL-6/gp130/JAK pathway in lung cancer–derived cell lines harboring EGFR mutations, we sought to determine whether IL-6 levels would correlate with pSTAT3 levels in primary lung cancer specimens. Immunohistochemical analysis of the previously described TMAs of primary lung adenocarcinomas revealed that 7.6% contained high (+3), 58.7% contained moderate (+2), and 28.3% contained low (+1) levels of IL-6, while 5.4% had no (0) IL-6. A previous analysis of 10 primary lung tissue samples revealed that 90% expressed high levels of IL-6, which is consistent with our observations (36). In the current work, a strong positive correlation was observed in lung specimens that were pSTAT3 positive (staining score of +1, +2, or +3) and expressing moderate-to-high levels (staining score of +2 to +3) of IL-6 (P < 0.001) (Figure (Figure9).9). In Figure Figure9,9, examples of selected tumor specimens from the same TMAs stained for IL-6 (top) and pSTAT3 (bottom) are shown. These findings strongly underscore the contribution of IL-6 in persistent activation of STAT3 in both primary lung adenocarcinomas and cancer-derived cell lines.
Identification of potential factors and regulatory pathways mediating STAT3 activation has important implications for the development and selection of molecularly targeted therapy in lung cancer. Our current findings demonstrate that the presence of pSTAT3 in lung adenocarcinoma tumor samples by immunohistochemical analysis correlates positively with expression of EGFR tyrosine kinase domain mutations. Owing to insufficient numbers of tumor specimens expressing somatic-activating mutations within the kinase domain, we were unable to determine whether pSTAT3 positivity correlated differentially with either ΔEGFR or L858R. Much of our data confirms prior observations regarding pSTAT3 and lung adenocarcinomas. Specifically, 50% of samples express variable levels of pSTAT3 by immunohistochemical analysis; a positive association exists between pSTAT3 and small tumor size, TTF1, pAKT, and pEGFR; and an inverse correlation exists between pSTAT3 and caspase-3 (14–16). No association between pSTAT3 and survival was observed. A recent description of a murine model revealed inducible expression of EGFR kinase domain–activating mutations targeted to the lung epithelium gives rise to adenocarcinomas containing pSTAT3 and pAKT, demonstrating an association between this oncogene and activated STAT3 (26). Of interest, expression of constitutively activated STAT3 targeted to the lung epithelium gives rise to de novo adenocarcinoma of the lung (C. Yan, unpublished observations).
The evidence demonstrating a direct association between EGFR and STAT3 is based primarily on work done in cell lines expressing high levels of EGFR, such as A431 and head and neck squamous cell carcinoma (HNSCC) cells (37, 38). The biochemical differences between cell lines expressing kinase domain–activating mutations within the EGFR versus wild-type EGFR include enhanced differential phosphorylation of specific tyrosine residues on the EGFR and alterations in the levels of pMAPK, pAKT, p-src, and pSTAT3 (8, 9, 28, 39, 40). However, the mechanism of STAT3 phosphorylation in either mutant EGFR–transfected cell lines or cancer-derived cell lines is unclear, as inhibition of EGFR (with ZD), src (with dasatinib), or JAK2 (with AG490) activity does not lead to inhibition of STAT3 phosphorylation in cancer-derived cell lines (8, 14, 28). Our current findings demonstrate that inhibition of JAKs with the use of a pan-JAK inhibitor (P6) completely abrogated STAT3 phosphorylation in all cancer-derived cell lines expressing mutant forms of EGFR (Figure (Figure2).2). Of importance, inhibition of neither EGFR activity (with ZD) nor src activity (with dasatinib-BMS) resulted in a change in pSTAT3 levels (Figure (Figure2A2A and data not shown). The use of a pan-JAK inhibitor, rather than an inhibitor specific for only a single JAK, is likely critical, as multiple JAKs could be implicated in mediating STAT3 phosphorylation in an EGFR-dependent manner. The JAKs have been described as being associated with EGFR, but they are not required to mediate STAT phosphorylation in response to EGF (41). P6 treatment of mutant EGFR cancer-derived cell lines did not markedly affect EGFR phosphorylation or pMAPK, which is a target of the EGFR (Figure (Figure2A).2A). Furthermore, the addition of EGF to these cell lines increases pEGFR, which was unaffected by P6 treatment (Figure (Figure2B2B and data not shown). These data suggest an uncoupling between EGFR and JAKs.
Identification of the regulatory pathways leading to STAT3 activation has clear implications for the development of targeted therapy in lung cancer (13). Of importance, we demonstrate that inhibition of JAK kinases with the use of the pan-JAK inhibitor P6 led to decreased proliferation and a G2/M cell cycle block in 11-18, H1975, and H1650 cell lines (Figure (Figure2C2C and data not shown). We demonstrate that STAT3 is likely the principal P6 target, as a partial knockdown of this protein in the H1650 cells led to a decrease in their proliferative capacity (Figure (Figure2D).2D). Inhibition of anchorage-independent growth by P6 was observed in all 4 cell lines, while ZD inhibited growth only in soft agar to comparable levels seen with P6 in H3255 cells (Figure (Figure3,3, A and B). The H1975 cell line is resistant to the growth-inhibitory actions of ZD and dasatinib (8). However, our data reveal that ZD, at a relatively higher concentration (5 μM), can inhibit the EGFR tyrosine phosphorylation and partially inhibit anchorage-independent growth of H1975 cells (Figure (Figure2A2A and Figure Figure3B).3B). This result is consistent with prior observations that ZD, at concentrations ranging from 5 to 10 μM, inhibits cell growth of H1975 (40). As P6 is insoluble, we could not administer it in vivo. Thus, subsequently, we were only able to show that P6-treated cells in vitro are less capable of growing in vivo compared with DMSO-treated control cells (Figure (Figure3,3, C and D). Nevertheless, these findings suggest that JAK inhibition may prove to be effective therapy in the treatment of cancers that depend on mutant EGFR activity, including those that may be less sensitive or resistant to TKIs.
Activation of JAKs and STAT3 has been shown to occur through binding of the IL-6 family of cytokines to the gp130 receptor in many different cancer-derived cell lines, including myeloma; HNSCC; breast and prostate cancers; cholangiocarcinoma; and NSCLC (22, 24, 31–33, 42, 43). Furthermore, high circulating levels of IL-6 have been found in patients who have metastatic breast and lung cancers (44, 45). Our current findings demonstrate that blockade of IL-6/IL-6R/gp130 signaling with the use of both sequestering antibodies and blocking antibodies inhibits STAT3 activation (Figure (Figure4A).4A). Furthermore, all cell lines produce very high levels of IL-6, which was rapidly secreted into the culture medium and could stimulate STAT3 phosphorylation in MCF-10A cells (Figure (Figure4B).4B). The degree of pSTAT3 inhibition with the blocking antibodies was not as profound as with P6 treatment. Furthermore, the blockade with both gp130 and IL-6 was only transient, lasting less than 24 hours (data not shown) (33). To examine the consequences of IL-6 blockade on growth, we introduced an IL-6 shRNA construct into the lung cancer–derived cell lines. We demonstrated that reduced expression of IL-6 led to a decrease in pSTAT3, as well as a decrease in the in vitro and in vivo growth of H1975, H1650, and 11-18 cells (Figure (Figure5).5). Despite reduced production of IL-6, there was no effect on in vitro growth in the H3255 cell line, which is consistent with the lack of growth inhibition by P6 in this cell line (data not shown). Thus, blockade of the IL-6 pathway is sufficient to inhibit the growth of these cancer-derived cell lines and demonstrates the potential therapeutic benefit of IL-6 inhibition.
Given the previously described association between the EGFR family members and the gp130 receptor, we attempted but were unable to show any physical relation between EGFR and gp130 in H3255 (which expresses the highest levels of EGFR), 11-18, H1975, and H1650 cell lines (data not shown) (42, 46, 47). To examine the relation between mutant EGFR and IL-6/gp130 receptor signaling, we introduced ΔEGFR into MCF-10A cells, and our current findings demonstrate an increased production of IL-6 mRNA and protein, with subsequent STAT3 activation and cellular transformation (Figure (Figure6).6). An immortalized breast epithelial cell line was chosen, as introduction of mutant EGFR into immortalized lung epithelial cells has resulted in an increase in pSTAT3 levels and anchorage-independent growth but no evidence of tumorigenesis (48). As observed with the lung cancer–derived cell lines, STAT3 phosphorylation depends on IL-6/gp130/JAK signaling, but not on EGFR kinase activity in the ΔEGFR MCF-10A cells (Figure (Figure7).7). These novel findings suggest that ΔEGFR can drive expression of IL-6, which ultimately is responsible for mediating gp130/JAK/STAT3 signaling.
If mutant EGFR is capable of upregulating IL-6 synthesis, why is there no effect on pSTAT3 levels after inhibition of EGFR activity (Figures (Figures22 and and6)?6)? The IL-6 protein is produced to high levels and is quite stable within the CM of these cancer-derived cell lines. Thus, decreasing its synthesis in the context of such excess would clearly have little to no consequence for STAT3 activation. To examine the contribution of mutant EGFR activity on IL-6 production and STAT3 activation, we first had to remove the IL-6–containing CM. After removal of CM, inhibition of the EGFR activity led to a partial decrease in IL-6 production and pSTAT3. This clearly demonstrates a requirement for mutant EGFR activity, but nonetheless supports the suggestion that other factors are mediating the regulation of IL-6 synthesis (Figure (Figure8,8, A and B). Some of the described transcriptional mediators of the IL-6 gene include NF-κB, AP-1, C/EBPβ, and CREB, whose expressions and activity vary as a function of cell type and ligand stimulation (35, 49–51). The EGF has been shown to activate NF-κB and AP-1 transcription factors in HNSCC-derived cell lines (52). It has been suggested that STAT3, in association with NF-κB, could bind to the IL-6 promoter through an NF-κB binding site and promote expression of the IL-6 gene (53). Of interest, IL-6 production has been shown to depend on activated STAT3 in a mouse gastric model system (54). We hypothesize that mutant EGFR leads to NF-κB activation and that STAT3 can potentiate IL-6 transcription through this critical interaction.
Our cell line data demonstrate a positive relation among mutant EGFR, IL-6, and pSTAT3. We examined lung adenocarcinoma TMAs for IL-6 expression and demonstrate a positive association between pSTAT3-positive samples (i.e., those with a staining score of +1, +2, or +3) and samples expressing moderate-to-high (+2 to +3) levels of IL-6. In summary, we demonstrate that the mechanism by which STAT3 is activated in lung adenocarcinomas is through mutant EGFR regulating expression of the IL-6 cytokine, which, in turn, activates the gp130/JAK pathway. Of importance, JAK kinase and IL-6 blockade leads to inhibition of growth and tumorigenesis, suggesting that the development of inhibitors of this pathway may be an important therapeutic target for these malignancies.
Human mutant EGFR–expressing NSCLC adenocarcinoma cell lines 11-18, H1650, H1975, and H3255 and control wild-type EGFR cell line H460 were cultured in RPMI-1640 medium with 10% FBS. The MCF-10A cells (obtained from ATCC) were cultured in 10A medium as previously described (55). The NIH3T3 cells were cultured in DMEM containing 10% FBS. To measure cell growth, 2 × 104 to 4 × 104 cells per well (2 × 103 to 4 × 103/cm2) were plated into 6-well dishes. Ten hours later, P6 (1 μM) or control DMSO was added (day 0). Cells were counted at days 1, 2, 3, 4, and 5. Each data point represented the mean value from triplicate wells. In addition, viability assays were performed with the use of calcein AM (acetoxymethylester of calcein; Molecular Probes; Invitrogen) according to the manufacturer's instructions. Cells (2.5 × 104 to 5.0 × 104) were plated in 0.35% agar in culture medium with DMSO, ZD (5 μM), or P6 (1 μM). Fresh drug was added to cells every 4 days. Colony number was assessed after 14 days by staining with MTT (0.5 ml of 5-mg/ml concentration). P6 (tetracyclic pyridone 2-tert-butyl-9-fluoro-3,6-dihydro-7H-benz[h]-imidaz[4,5-f]isoquinoline-7-one) was synthesized by in our laboratory (56). ZD was a gift from Astra Zeneca, provided by Jackie She of the Anti-Tumor Assessment Core Facility, MSKCC. Dasatinib (BMS-354825) was synthesized in our laboratory.
pBabe-puro and pBabe-EGFR-Δ747-752 were introduced into MCF-10A cells by retroviral infection and selected with puromycin (2 μg/ml), as previously described (57). IL-6 shRNA lentiviral construct and control lentiviral constructs were purchased from Open Biosystems, and infections into lung cancer–derived cell lines, followed by selection with puromycin (2 μg/ml), were performed according to the manufacturer’s instructions. Given that cell growth was inhibited by the IL-6 shRNA, proliferation assays were conducted and in vivo growth was determined within 72 hours of puromycin selection. STAT3 shRNA and control lentiviral constructs were generated as previously described (58).
pBabe-EGFR-Δ747-752 (7), a mutant human EGFR expression construct, was generated by subcloning a wild-type EGFR cDNA into the XhoI site of pBluescript. A PCR-based mutagenic deletion of amino acids 747–752 was subsequently performed using DNA polymerase Pfu and primer sets: 5′-CCCGTCGCTATCAAGGAACCGAAAGCCAACAAGG-3′, 5′-CCTTGTTGGCTTTCGGTTCCTTGATAGCGACGGG-3′. The mutated EGFR was cut out of the pBluescript with the use of XhoI and ligated to the SalI site of the retroviral construct pBabe-puro (59).
Adherent cells were washed with PBS and lysed in whole-cell lysis buffer (150 mM NaCl, 10% glycerol, 1% IGEPAL, 0.5% sodium deoxycholate, 2 mM EDTA, 1 mM NaF, 1 mM Na3VO4, 1 mM Na2MoO4, 0.1% SDS, 20 mM Tris-HCl, pH 8) with proteinase inhibitors. Western blot analysis was performed as previously described (60). Antibodies used for Western blot analysis were as follows: anti–phospho-STAT3 (Tyr705), total STAT3, phospho-MAPK (p44/42, Thr202/Tyr204), phospho-AKT (Ser473), and total AKT (Cell Signaling); anti-MAPK (ERK2), EGFR, and HRP-conjugated anti-rabbit IgG (Santa Cruz Biotechnology Inc.); anti–phospho-tyrosine (4G10) (Upstate Technologies); anti–α-tubulin (Sigma-Aldrich); and HRP-conjugated anti-mouse IgG (Jackson ImmunoResearch Laboratories Inc.). For cytokine- and gp130-blocking experiments, cells were treated with blocking antibodies (5 μg/ml) for 16 hours. Human IL-6–blocking antibodies B-E8 and 522 were obtained from Cell Science and R&D Systems. Human gp130–blocking antibody B-R3 was obtained from Cell Science. IL-6R–, LIF-, and OSM-blocking antibodies were obtained from R&D Systems.
pGL3-IL-6-Luc, an IL-6 promoter luciferase reporter construct, was kindly provided by Espen Spangenburg (University of Maryland, College Park, Maryland, USA) (35, 61). The pGL3-IL-6-Luc, a normalization construct TK-Renilla, and either control vector pBabe-puro or pBabe-EGFR-Δ747-752 were cotransfected into NIH3T3 cells with the use of Lipofectamine 2000 reagent (Invitrogen). The next day, the medium was changed, and DMSO (5 μl/well) or ZD (5 μM) was added. Twenty-four hours later, measurements of firefly and Renilla luciferase activities in cell lysates were carried out with the use of the Dual-Luciferase Reporter Assay System (Promega). Data were presented as the ratio of firefly luciferase activity to Renilla luciferase activity in each sample performed in triplicate ± SD.
Cells were grown to approximately 50%–85% confluence, and medium was aspirated. Cells were subsequently washed twice with PBS, and fresh medium was added. At various time points, CM was analyzed for IL-6 with the use of an IL-6 ELISA kit (Cell Science) according to the manufacturer’s instructions.
Total cell RNA was extracted from cells by using an RNeasy kit (QIAGEN). Two micrograms of total RNA was used for IL-6 and β-actin RT-PCR using an iScript RT-PCR kit (Bio-Rad) according to the manufacturer’s instructions. Sequences of primers for amplification of the IL-6 gene were as follows: forward primer, 5′-TAGCCGCCCCACACAGACAG-3′; reverse primer, 5′-GGCTGGCATTTGTGGTTGGG-3′. The β-actin primers were as follows: forward primer, 5′-CGTGCGTGACATTAAGGAGA-3′; reverse primer, 5′-TGATCCACATCTGCTGGAAG-3′. The same cDNA samples were also analyzed for real-time quantitative PCR using a Bio-Rad IQ5 Multicolor Real-Time PCR Detection System/SyBR.
Cells were resuspended in 0.5 ml of ice-cold 1% BSA-PBS and fixed with 70% ethanol. The fixed cells were subsequently washed 3 times with BSA-PBS, resuspended in 1 ml of BSA-PBS containing propidium iodide (5 μg/ml) and RNase A (100 μg/ml), and incubated at 37°C for 15 minutes. Single-cell suspensions were obtained by filtering cells through a cell strainer for flow cytometry analysis.
All animal work was done in accordance with a protocol approved by the MSKCC Institutional Animal Care and Use Committee. Female homozygous athymic nu/nu mice (National Cancer Institute) were used for xenografting studies. After being treated with DMSO or P6 for 16 hours, 11-18 and H1650 cells were harvested. They were subsequently mixed with an equal volume of Matrigel (BD Biosciences) and injected subcutaneously into the flanks of 45-day-old female nude mice. Both tumor size and tumor weight were determined after 14 days. Four animals were used for each experimental group. Control shRNA– and IL-6 shRNA–infected 11-18, H1650, and H1975 cell lines were harvested, and 0.5 × 106 cells were mixed with an equal volume of Matrigel and injected subcutaneously into the flanks of 45-day-old female nude mice. Both tumor size and tumor weight were determined after 21 days. Four animals were used for each experimental group. MCF-10A control pBabe and ΔEGFR-expressing cells (1 × 107 cells) were mixed with an equal volume of Matrigel and injected subcutaneously into the flanks of 45-day-old female nude mice. Both tumor size and tumor weight were determined after 21 days. Four animals were used for each experimental group.
Multitissue blocks of formalin-fixed, paraffin-embedded NSCLC tissue (containing 3 or 4 representative 0.6-mm cores) were prepared using a tissue arrayer, and immunohistochemistry was performed, as described previously (60). Antigen retrieval, with the use of citric acid (pH 6.0) at 97°C for 30 minutes, was followed by treatment with 3% hydrogen peroxide. pSTAT3 (Tyr705), pEGFR (pY1068), phospho-AKT, pERK1/2 (pMAPK), and caspase-3 antibodies (Cell Signaling) were used at 1:200, 1:100, 1:50, 1:100, and 1:300 dilutions, respectively. Anti-TTF1 mAb (Dako) was used at a 1:100 dilution. Anti-human IL-6 goat polyclonal antibody (R&D Systems) was used at a 1:100 dilution (62). Secondary reagents were from DakoCytomation EnVision+ Dual Link System-HRP (DAB+) kit or the DakoCytomation LSAB+ system-HRP kit. Counterstaining was performed with the use of hematoxylin. Scoring of the TMAs was performed by 2 independent observers, with a high correlation observed between scorers (P < 0.001) for both pSTAT3 and IL-6. For a tumor to be considered positive for either pSTAT3 or IL-6, all 4 replicates in the tissue array had to have a similar staining intensity. Otherwise, it was excluded. Statistical analyses were done using StatView (SAS Institute). The correlation between the scores of both scorers and the relation between those of pSTAT3 and IL-6 were measured by using the χ2 test.
DNA samples were extracted from frozen tissue after microdissection. For EGFR, restriction fragment alterations were also used to detect the most common point mutation in exons 19 and 21 (63). After screening was completed, the mutated cases were confirmed by sequencing. For samples with only RNA available, RT-PCR was used with the primers listed above, followed by direct sequencing. Di-deoxy sequence analysis of mutational hot spots for exons 18–22 of EGFR was performed with the use of specific primers listed as follows: EGFR-forward, 5′-CCATCGCCACTGGGATG-3′; EGFR-reverse, 5′-TCTTGACATGCTGCGGTG-3′.
This work was supported by NIH grant R01 CA87637, a Patrick M. Byrne grant, the Sussman Family Fund, the Charles and Marjorie Holloway Foundation, the Breast Cancer Alliance, and Lerner and Leigh Awards (to J.F. Bromberg) as well as New York State Department of Health grant C021338 (to S.P. Gao). pGL3-IL-6-Luc was kindly provided by Espen Spangenburg. We thank Koichi Hagiwara for the 11-18 cell line. Opinions expressed are solely those of the authors and do not necessarily reflect those of the New York State Department of Health or the State of New York.
Nonstandard abbreviations used: CM, conditioned medium; CNTF, ciliary neurotropic factor; HNSCC, head and neck squamous cell carcinoma; LIF, leukemia inhibitory factor; NSCLC, non–small cell lung cancer; OSM, oncostatin M; P6, tetracyclic pyridone 2-tert-butyl-9-fluoro-3,6-dihydro-7H-benz[h]-imidaz[4,5-f]isoquinoline-7-one; shRNA, short hairpin RNA; TKI, tyrosine kinase inhibitor; TMA, tissue microarray; TTF1, thyroid transcription factor 1; Tyk2, tyrosine kinase 2; ZD, ZD-1839 (gefitinib).
Conflict of interest: The authors have declared that no conflict of interest exists.
Citation for this article: J. Clin. Invest. 117:3846–3856 (2007). doi:10.1172/JCI31871
See the related Commentary beginning on page 3988.