Mutations in the EGFR kinase domain mediate STAT3 activation via IL-6 production in human lung adenocarcinomas (original) (raw)
The presence of activated STAT3 correlated positively with adenocarcinomas harboring mutant EGFR (ΔEGFR and L858R). 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 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 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).
STAT3 is tyrosine phosphorylated in primary lung adenocarcinomas. Immunohistochemical analysis of 92 primary lung adenocarcinomas for tyrosine-phosphorylated STAT3 was performed; 46 were scored as 0; 27 as +1; 16 as +2; and 3 as +3, with associated percentages indicated in parentheses. Two examples of each are shown. Original magnification, ×200.
Mutant EGFR expression correlates with tyrosine-phosphorylated STAT
JAK inhibition, but not EGFR inhibition, blocks STAT3 phosphorylation, induces a G2/M cell cycle arrest, and reduces anchorage-independent growth and tumorigenesis in human lung adenocarcinoma–derived cell lines harboring mutant constitutively activated forms of EGFR. 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 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 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 2A 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 2A 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 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 2B 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 2B and data not shown). These data suggest that the EGF/EGFR pathway is independent of the JAK/STAT3 pathway.
JAK inhibition, but not EGFR inhibition, blocks STAT3 phosphorylation and induces growth arrest in human lung adenocarcinoma–derived cell lines harboring mutant EGFR. (A) Extracts from H3255, H1975, H1650, 11-18, and H460 cell lines, treated with DMSO (D), ZD (5 μM), P6 (1 μM), or BMS (5 μM) for 16 hours, were analyzed for phospho- and total EGFR, STAT3, AKT, and MAPK as well as α-tubulin as a loading control by Western blot analysis. (B) Extracts from 11-18 cells, treated with DMSO, P6 (1 μM), or ZD (5 μM) for 30 minutes and stimulated with EGF (E) (100 ng/ml) for 10 minutes, were analyzed for pEGFR, EGFR, pSTAT3, STAT3, pMAPK, and MAPK. (C) Cell proliferation of H3255, H1975, H1650, 11-18, and H460 cells treated with P6 was determined over a 5-day period. Light gray line, DMSO treatment; black line, P6 treatment. (D) Extracts obtained from scrambled control (C) and STAT3 shRNA–expressing (S3Sh-expressing) H1650 cells were analyzed for pSTAT3, STAT3, and α-tubulin (left). Proliferation of the H1650 control (light gray line) and H1650 S3Sh (black line) was evaluated by calcein AM.
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 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 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 3, A and B). The EGFR TKI ZD inhibited colony growth of H3255 cells, and partial inhibition was observed in H1975 cells (Figure 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 3, C and D).
P6 inhibits tumorigenesis of human lung adenocarcinoma cell lines in vitro and in vivo. (A) 11-18, H1650, H1975, H3255, and H460 cells (5 × 103/well) were plated in soft agar in the presence of DMSO, ZD, or P6. Colony numbers were counted after 14 days. A representative colony growth of 11-18 is shown with the indicated treatment. (B) Soft agar colony numbers of treated 11-18, H1650, H1975, H3255, and H460 cell lines are shown (mean ± SD). (C) 11-18 cells were treated with DMSO or P6 for 16 hours and injected into the flanks of nude mice. Size and weight of tumors were determined after 14 days (mean ± SD). Examples of 2 animals with representative injections are shown. (D) Weight of the tumors from DMSO- or P6-treated cells are shown (mean ± SD).
The IL-6/gp130/JAK pathway is responsible for STAT3 phosphorylation in lung adenocarcinoma–derived cell lines. 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 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 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.
The IL-6/gp130/JAK pathway mediates STAT3 phosphorylation in lung adenocarcinoma–derived cell lines. (A) Extracts isolated from H3255, 11-18, and H1650 cell lines treated with control mouse IgG, gp130-blocking mAb (B-R3), or IL-6–blocking mAb (αIL-6) after a medium change were analyzed by Western blotting for pSTAT3, STAT3, and α-tubulin as a loading control. (B) CM collected from 11-18 cells was added to MCF-10A cells in the presence of control mouse IgG, B-R3, αIL-6, IL-6R (αIL-6R), OSM (αOSM), and LIF (αLIF) blocking antibodies. Extracts isolated from these treated cells were analyzed by Western blot analysis for pSTAT3 and STAT3.
Decreased IL-6 blocks pSTAT3 and growth in vitro and in vivo of lung cancer–derived cell lines. 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 5A). The consequences of decreased IL-6 production included inhibition of pSTAT3, as well as inhibition of growth in vitro and in vivo (Figure 5, B–D).
Blockade of IL-6 signaling with IL-6 shRNA inhibits growth of cell lines. (A) IL-6 shRNA lentivirus (ShRNA) and control lentivirus (C) were introduced into H1975, H1650, and 11-18 cell lines. After 72 hours of selection with puromycin, levels of IL-6 were determined by ELISA of CM. (B) Extracts isolated from the above-described cell lines were analyzed by Western blotting for pSTAT3, STAT3, and α-tubulin as a loading control. (C) A total of 2,000 cells/cm2 were seeded, and proliferation was determined daily with the use of calcein AM. (D) H1975, 11-18, and H1650 cells, expressing either control or IL-6 shRNA (Sh), were injected into the flanks of nude mice. The tumor weight was determined after 21 days (mean ± SD) (right). An example of an animal injected with H1975 cells infected with control or IL-6 shRNA is shown (left).
Δ_EGFR 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 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 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 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 7A). RT-PCR analysis revealed high IL-6 mRNA expression in ΔEGFR–MCF-10A cells relative to expression in control MCF-10A cells (Figure 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 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.
Overexpression of ΔEGFR protein in MCF-10A cells induces persistent phosphorylation of STAT3, AKT, and MAPK as well as tumorigenesis. (A) Extracts isolated from control pBabe (pB) and ΔEGFR-expressing MCF-10A cells were analyzed by Western blot for phospho- and total EGFR, STAT3, AKT, and MAPK as well as α-tubulin as a loading control. The lanes were run on the same gel but were noncontiguous. (B) Soft agar colony formation assays for MCF-10A control pB and ΔEGFR-expressing cells are shown. Colony numbers are shown below. (C) MCF-10A control pB and ΔEGFR-expressing cells were injected into the flanks of nude mice. No tumor was detected with pB, while ΔEGFR-expressing MCF-10A cells formed tumors. Tumor weight was determined after 21 days (mean ± SD) (right). An example of an animal injected with MCF-10A–ΔEGFR–expressing cells is shown (left).
Overexpression of ΔEGFR mediates IL-6 production of MCF-10A cells. (A) Extracts isolated from ΔEGFR MCF-10A cells treated with DMSO, ZD, P6, B-R3, or αIL-6 for 16 hours were analyzed by Western blotting for pEGFR, EGFR, pSTAT3, STAT3, pAKT, pMAPK, and MAPK. The last lane was run on the same gel but was noncontiguous. (B) Human IL-6 mRNA levels from MCF-10A control (pB) and ΔEGFR MCF-10A cells were determined by RT-PCR and normalized to β_-actin_. (C) Extracts were isolated from MCF-10A cells treated with CM from control cells (pB) and ΔEGFR MCF-10A cells and analyzed by Western blotting for pSTAT3 and STAT3.
EGFR tyrosine kinase inhibition reduces de novo production of IL-6. 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 2A and Figure 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 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 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 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 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.
EGFR tyrosine kinase inhibition reduces de novo production of IL-6. (A) 11-18 and H1650 cells were treated with a medium change (MC) with ZD (MC+ZD), no MC with ZD (No MC+ZD), or MC with DMSO (MC+D). Levels of IL-6 were measured by ELISA at the indicated times after the addition of ZD (mean ± SD). (B) Shown are extracts isolated from 11-18 and H1650 cells treated as described above after 16 hours and analyzed for pSTAT3, STAT3, and α-tubulin. (C) Human IL-6 mRNA levels from 11-18 cells treated with DMSO or ZD for 16 hours after a medium change were determined by RT-PCR and normalized to β_-actin_. The same mRNA samples were analyzed by quantitative real-time PCR (QPCR), and the IL-6 mRNA levels (normalized to hypoxanthine-guanine phosphoribosyltransferase [_HPRT_]) are shown below. (D) NIH3T3 cells were cotransfected with an IL-6 reporter construct, a TK-Renilla construct (for transfection/loading control), and either a pB vector (baseline activity) or a pB-ΔEGFR expression construct. Twenty-four hours after transfection, DMSO or ZD was added, and an additional 24 hours later, cells were lysed and subjected to firefly and Renilla luciferase activity measurements. The bars show fold induction over the baseline activity (mean ± SD).
pSTAT3 levels correlate positively with IL-6 expression in primary lung adenocarcinomas. 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 9). In Figure 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.
pSTAT3 levels correlate positively with IL-6 expression in primary lung adenocarcinomas. Immunohistochemical analysis of TMAs of primary lung adenocarcinomas (92 tumor specimens) for IL-6 were scored as 0 (5/92), +1 (26/92), +2 (54/92), or +3 (7/92). Examples of selected tumor specimens from sequential sections of the same TMAs stained for IL-6 (top, left to right, 0, +1, +3, and +3) and pSTAT3 (bottom, left to right, 0, 0, +3, and +3) are shown. A positive correlation was observed in specimens that were pSTAT3 positive (+1, +2, or +3) and expressing moderate-to-high (+2 to +3) levels of IL-6 as determined by analysis with the Fisher exact test (P < 0.001).