Members of the Ewing sarcoma family of tumors (ESFT) contain tumor-associated translocations that give rise to oncogenic transcription factors, most commonly EWS/FLI1. EWS/FLI1 plays a dominant role in tumor progression by modulating the expression of hundreds of target genes. Here, the impact of EWS/FLI1 inhibition, by RNAi-mediated knockdown, on cellular signaling was investigated using mass spectrometry–based phosphoproteomics to quantify global changes in phosphorylation. This unbiased approach identified hundreds of unique phosphopeptides enriched in processes such as regulation of cell cycle and cytoskeleton organization. In particular, phosphotyrosine profiling revealed a large upregulation of STAT3 phosphorylation upon EWS/FLI1 knockdown. However, single-cell analysis demonstrated that this was not a cell-autonomous effect of EWS/FLI1 deficiency, but rather a signaling effect occurring in cells in which knockdown does not occur. Conditioned media from knockdown cells were sufficient to induce STAT3 phosphorylation in control cells, verifying the presence of a soluble factor that can activate STAT3. Cytokine analysis and ligand/receptor inhibition experiments determined that this activation occurred, in part, through an IL6-dependent mechanism. Taken together, the data support a model in which EWS/FLI1 deficiency results in the secretion of soluble factors, such as IL6, which activate STAT signaling in bystander cells that maintain EWS/FLI1 expression. Furthermore, these soluble factors were shown to protect against apoptosis.
Implications: EWS/FLI1 inhibition results in a novel adaptive response and suggests that targeting the IL6/STAT3 signaling pathway may increase the efficacy of ESFT therapies. Mol Cancer Res; 12(12); 1740–54. ©2014 AACR.
Advancements in the understanding of the molecular mechanisms of oncogenesis have led to the development of targeted therapeutics. For example, activating mutations in kinases such as EGFR in lung cancer or B-RAF in melanoma have been inhibited by specific small molecules to increase therapeutic efficacy. However, striking initial responses are rarely sustained due to innate and acquired resistance mechanisms (1, 2). In the case of melanoma, initial suppression of the MAPK pathway by B-RAF inhibitors is followed by reactivation that occurs through relief of a negative feedback loop (3). In other systems, activation of redundant pathways can occur through cell-autonomous mechanisms or be mediated by stromal secretion of growth factors into the tumor microenvironment (4). These adaptive responses by tumor cells to evade the effects of targeted therapeutics present a challenge to single-agent therapy.
Targeted therapy has also been used for the treatment of the Ewing sarcoma family of tumors (ESFT). As opposed to activating kinase mutations, ESFT pathogenesis is primarily driven by what appears to be an aberrant transcription factor generated by a chromosomal translocation. In most tumors, this translocation fuses the EWS gene to the ETS transcription factor FLI1 (5). The fusion protein EWS/FLI1 retains domains that facilitate interaction with transcriptional regulators and DNA binding, which provides the ability to alter gene expression (6). EWS/FLI1 is capable of oncogenic transformation, and maintenance of expression is required for ESFT cell growth, indicating a dominant role in tumorigenesis (6, 7).
Because EWS/FLI1 presents an ideal therapeutic target, several strategies have been used to identify a compound that inhibits its function. Initial small-molecule screens identified compounds that inhibited EWS/FLI1 modulation of gene expression, including cytarabine (8), mithramycin (9), and midostaurin (10). Other screens have been used to find molecules that bind to EWS/FLI1 or disrupt its ability to bind DNA. YK-4-279, a derivative of a compound found to bind to EWS/FLI1, was demonstrated to decrease EWS/FLI1 activity by blocking its interaction with the transcriptional coactivator RNA helicase A (11). In addition, low concentrations of actinomycin D were found to selectively inhibit EWS/FLI1 binding to DNA (12). Trabectidin, evaluated based on its ability to inhibit a similar fusion in myxoid liposarcoma, was also shown to inhibit EWS/FLI1 activity and induce apoptosis in ESFT cell lines (13).
Unfortunately, the in vitro efficacy of these compounds thus far has not translated to the clinic. Phase II trials of cytarabine and trabectidin did not demonstrate potent single-agent activity, and stable disease was observed in only a minority of patients (14, 15). Modest single-agent activity was also observed with other targeted therapeutics evaluated in ESFT, including drugs directed against the insulin-like growth factor receptor. These low clinical response rates highlight the adaptive responses of ESFT when exposed to single-agent therapy. As additional molecularly targeted compounds are being evaluated in clinical trials, increased understanding of ESFT cellular signaling is needed to address mechanisms of drug resistance and optimize therapeutic efficacy. Therefore, we chose to investigate changes in protein phosphorylation upon inhibition of EWS/FLI1 in ESFT. We used shRNA-mediated knockdown as a model of EWS/FLI1 inhibition because reduction of expression encapsulates the multiple mechanisms used by various small molecules. Mass spectrometry (MS)–based phosphoproteomics was used to quantitate global changes in phosphorylation levels after EWS/FLI1 knockdown. Our results revealed a paracrine signaling mechanism that induces cytokine secretion in EWS/FLI1-targeted cells and subsequent STAT3 activation in bystander cells. This novel adaptive response suggests that combination therapy with STAT3 inhibitors may increase the efficacy of targeted therapeutics in ESFT.
Materials and Methods
ESFT cell lines (RDES, TC-174, SK-N-MC, SKES, A4573, A673, and 6647) were cultured in Iscove's modified Dulbecco's medium (IMDM) containing 10% FBS. ESFT cell lines were either purchased from the ATCC or were a gift from Timothy J. Triche, MD, PhD at the Saban Research Institute, Children's Hospital Los Angeles. Cell lines from the ATCC undergo authentication via morphology check by microscopy, growth curve analysis, isoenzymology, short tandem repeat analysis, and mycoplasm detection. All cell lines underwent the following authentication process at UCLA: mycoplasm detection, morphology check and documentation with microscopy and digital photography, growth curve analysis, mitochondrial DNA analysis (in which the cell line identity is confirmed by mitochondrial DNA comparative analysis of the highly variable regions I/II modified Cambridge sequence), and extensive characterization including analysis for the EWS translocation and potential mutations (PTEN, PI3K, CDKN2A) by RT-PCR. 293T cells used for virus production were cultured in DMEM containing 10% FCS and supplemented with l-glutamine (2 mmol/L) and penicillin–streptomycin (50 IU/mL and 50 μg/mL, respectively).
EWS/FLI1 818 and EF4 shRNA constructs were cloned into the CSCG lentiviral vector as previously described (16, 17). The dominant-negative STAT3 construct, in which tyrosine 705 is mutated to phenylalanine, was cloned from pRc/CMV STAT3 Y705F Flag (Addgene plasmid 8709; ref. 18) into the SRα-MSV-TK Neo retroviral vector (19). Lentiviral and retroviral stocks were generated as previously described (17).
IGF1 was provided by Pinchas Cohen (UCLA). Stattic (STAT3 Inhibitor V) was obtained from Santa Cruz Biotechnology. Human recombinant IL6, GM-CSF, and CXCL1 were obtained from R&D Systems. Doxorubicin HCl was obtained from Shandong Tianyu Fine Chemical Co., Ltd. NVP-AEW541 was obtained from Cayman Chemical.
Quantitative real-time PCR
RNA was harvested using the RNeasy Mini Kit (Qiagen) or PureLink RNA Mini Kit (Invitrogen). cDNA was synthesized from approximately 2 μg of RNA using the SuperScript III First-Strand Synthesis System (Invitrogen). For real-time PCR, a 1:10 dilution of cDNA was combined with forward and reverse primers and master mix containing SYBR green, Taq, and dNTPs (Applied Biosystems). Reactions were run at 95°C for 10 minutes, followed by 40 cycles at 95°C for 10 seconds, 60°C for 30 seconds, and 72°C for 20 seconds on a DNA Engine Opticon 2 Real-Time Cycle (MJ Research/Bio-rad). Results were analyzed with Opticon Monitor software (MJ Research/Bio-Rad). Primers used to quantify cellular transcript levels are as follows: GAPDH: 5′-ATGTTCGTCATGGGTGTGAA-3′ and 5′-CCAGGGGTGCTAAGCAGTT-3; EWS/FLI1: 5′-GCCAAGCTCCAAGTCAATATAGC-3′ and 5′-GAGGCCAGAATTCATGTTATTGC-3′; and IL6: 5′-AGCCACTCACCTCTTCAGAACGAA-3′ and 5′-AGTGCCTCTTTGCTGCTTTCACAC-3′. EWS/FLI1 primers were originally described by Tirode and colleagues (20). IL6 primers were originally described by Inda and colleagues (21).
Cells were incubated for approximately 1 hour on ice in lysis buffer (50 mmol/L Tris pH 7.6, 0.5% NP-40, 10% glycerol, 30 mmol/L NaCl, 1 mmol/L EDTA) supplemented with Complete Mini EDTA-free protease inhibitor cocktail (Roche), 1 mmol/L Na3VO4, and 1 mmol/L NaF. Lysates were combined with 6X protein sample buffer (0.35 mol/L Tris pH 6.8, 10% SDS, 30% glycerol, 0.6 mol/L DTT, 0.012% bromophenol blue) and boiled for 5 to 10 minutes before loading on an 8% or 4% to 15% gradient polyacrylamide gel. The primary antibodies used for these studies were rabbit anti-phospho-STAT3 (Tyr705), rabbit anti-phospho-STAT3 (Ser727), mouse anti-STAT3, rabbit anti-gp130, and rabbit anti-cleaved PARP from Cell Signaling Technology; mouse anti-FLAG M2 and mouse anti–β-actin from Sigma; mouse anti-FLI1 from BD Biosciences; and mouse anti-phosphotyrosine [clone 4G10; horseradish peroxidase (HRP) conjugate] from Millipore. Secondary antibodies conjugated to HRP were sheep anti-mouse IgG from GE Healthcare; bovine anti-goat IgG and goat anti-rabbit IgG from Santa Cruz Biotechnology. Secondary antibodies conjugated to infrared dyes were IRDye 800CW goat anti-mouse IgG and IRDye 680RD goat anti-rabbit IgG from LI-COR Biosciences. Fluorescent westerns were imaged using the Odyssey Infrared Imaging System (LI-COR Biosciences). Signals were quantified by measuring the integrated intensity values of each band using Odyssey software (LI-COR Biosciences).
Cells were incubated for approximately 1 hour on ice in lysis buffer supplemented with protease inhibitor and 1 mmol/L Na3VO4. For serine/threonine enrichment, 1 mmol/L NaF was added to the lysis buffer. Lysates were centrifuged at 1,000 × g for 5 minutes and supernatant was saved. Four volumes of ice-cold (−20°C) acetone were added, and mixture was vortexed and incubated at −20°C for 1 to 2 hours. Precipitated proteins were pelleted by centrifuging at 6,000 × g for 15 minutes at 0°C. The pellet was washed once with 10 mL of ice-cold acetone to remove any residual NP-40, then resuspended in 8 mol/L urea, 50 mmol/L Tris pH 7.5, and 1 mmol/L Na3VO4 (and 1 mmol/L NaF for phosphoserine/threonine enrichment) by incubating overnight at 4°C with rotation. Phosphotyrosine peptides were enriched by immunoprecipitation with a pan-specific antiphosphotyrosine antibody (clone 4G10; Millipore) from 25 to 33 mg of total protein as previously described (22, 23). Phosphoserine/threonine peptides were purified from 9 to 10 mg of total protein by a combination of strong cation exchange chromatography and titanium dioxide (TiO2) enrichment as previously described (24), except that peptides were concentrated and desalted using ZipTip C18-based solid phase extraction (twice).
Mass spectrometry and phosphopeptide quantitation
MS was performed using a quantitative, label-free approach that has been demonstrated to show high concordance in quantitation and standard error to a label-based approach [stable isotope labeling by amino acids in cell culture (SILAC); ref. 22]. Phosphorylated peptides were analyzed by LC/MS-MS with an Eksigent autosampler coupled with a Nano2DLC pump (Eksigent) and LTQ-Orbitrap (Thermo Fisher Scientific) as previously described (25). MS/MS fragmentation spectra were searched with SEQUEST (Version v.27, rev. 12; Thermo Fisher Scientific) against a database containing the combined human-mouse International Protein Index (IPI) protein database (downloaded December 2006 from ftp.ebi.ac.uk) for peptides enriched for phosphotyrosine or against a human IPI database (version 3.71) for peptides enriched for phosphoserine/threonine. Search parameters were as previously described (25), except that dynamic modifications also included phosphorylated serine and threonine.
To identify phosphopeptide peaks sequenced in some samples but not others, the chromatogram elution profiles were aligned using a dynamic time warping algorithm (26). Further explanation of this protocol can be found in the supporting information of Zimman and colleagues (24) and Rubbi and colleagues (22). Relative amounts of the same phosphopeptide across samples run together were determined using custom software to integrate the area under the unfragmented (MS1) monoisotopic peptide peak (23, 24). All peaks corresponding to phosphopeptides were visually inspected and manually corrected if necessary.
The number of unique phosphorylation sites identified in our experiments was determined by collapsing multiple phosphopeptide ions representing the same phosphorylation site. Phosphosites with multiple detections (e.g., different ion charge state, modification) were quantified by summing the MS1 integration values for each phosphopeptide ion. In addition, for the phosphoserine/threonine analysis, phosphosites that were detected in multiple strong cation exchange fractions were quantified by summing the MS1 integration values for each fraction. The residue numbers listed for phosphosites correspond to the indicated IPI accession number.
Cell viability and growth assays
The numbers of viable cells were determined indirectly by MTT assay. Cells were seeded in 96-well plates, with each cell type or treatment condition performed in triplicate, and incubated overnight. After drug treatment or growth period, 10 μL of 5 mg/mL MTT (3-(4,5-dimethylthiazolyl-2)-2, 5-diphenyltetrazolium bromide) in PBS was added to cells and allowed to incubate for 2 to 4 hours at 37°C. Cells were then lysed with 100 μL of 15% SDS in 15 mmol/L HCl and incubated overnight at room temperature in the dark. Plate absorbance was read at 595 nm using a Bio-rad microplate reader. Percent viability was calculated by normalizing absorbance values to those from cells grown in media without drug after background subtraction. IC50 values were calculated by fitting dose–response curves to a four-parameter, variable slope sigmoid dose–response model (Prism Software; GraphPad). Synergistic, additive, or antagonistic effects of Stattic combination treatment were determined based on combination indices and isobologram plots generated with CompuSyn software (ComboSyn, Inc.) using the method of Chou and Talalay (27). Relative growth was calculated by normalizing absorbance values to those from day 0 after background subtraction.
Cells were grown on 4- or 8-well chamber slides, then fixed in 3.7% formaldehyde in PBS for 15 minutes at room temperature, and permeabilized with 100% methanol for 10 minutes at −20°C. After blocking for 1 hour with Protein Block (Dako) diluted 1:10 in PBS, cells were incubated with primary antibody overnight at 4°C and secondary antibody for 1 hour at room temperature. The primary antibodies used were rabbit anti–phospho-STAT3 Tyr705 and mouse anti-STAT3 from Cell Signaling Technology and anti-STAT3 from Abcam. The rabbit and mouse secondary antibodies used were conjugated to Alexa Fluor 594 (Invitrogen). After antibody incubation, cover slips were mounted with medium containing DAPI (VECTASHIELD Mounting Medium with DAPI; Vector Laboratories). Slides were analyzed by fluorescent microscopy with a Zeiss AxioImager microscope (Carl Zeiss).
Cells were fixed in 1.5% formaldehyde for 10 minutes at room temperature and then permeabilized with 100% ice-cold methanol for 20 minutes at 4°C. Cells were then washed twice with staining media (0.5% BSA in PBS, pH 7.4) and incubated with primary antibody for 1 hour at room temperature. After incubation, cells were again washed twice with staining media, then resuspended in PBS, and analyzed using a Becton Dickinson–modified FACScan analytic flow cytometer. Live cell events (10,000) were recorded after gating cells using forward scatter and side scatter to remove debris and dead cells. Median fluorescence intensity (MFI) was used to quantify changes in phospho-STAT3 signal. The primary antibody used for this analysis was rabbit anti–phospho-STAT3 (Tyr705) XP Alexa Fluor 647 conjugate from Cell Signaling Technology. Flow cytometry data were analyzed using FlowJo (Tree Star, Inc.).
ESFT cells were grown in 10% FBS in IMDM serum for 72 hours in 10-cm plates, between 2 and 5 days after lentiviral transduction. Alternatively, ESFT cells were transferred to serum-free IMDM or IMDM containing 1% serum 4 days after transduction and cells were grown for 48 hours. Conditioned media were centrifuged at 2,000 rpm for 5 minutes in a swinging bucket rotor to pellet any cell debris.
Cytokine array, ELISA, and neutralizing antibodies
The RayBio Human Cytokine Antibody Array C-Series 2000 Kit (RayBiotech, Inc.) was used according to the manufacturer's instructions. The concentration of IL6 in conditioned media was quantified using a human IL-6 Quantikine ELISA Kit (R&D Systems). Phospho-STAT3 levels in ESFT cells expressing EWS/FLI1 were quantified using a PathScan Phospho-STAT3 (Tyr705) Sandwich ELISA kit (Cell Signaling Technology). IL6- and gp130-neutralizing antibodies were obtained from R&D Systems.
Phosphoproteomic profiling identifies phosphopeptides modulated by EWS/FLI1
Genes modulated by EWS/FLI1 include members of signal transduction pathways, such as insulin-like growth factor binding protein 3 (28), the mitotic kinases Aurora A and B (29), and caveolin-1 (30). Therefore, we hypothesized that EWS/FLI1 inhibition could lead to changes in the activity of critical signaling components in ESFT. Phosphotyrosine immunoblot analysis showed an overall decrease in protein phosphorylation upon EWS/FLI1 knockdown (Supplementary Fig. S1A), indicating the fusion protein plays a role in cellular signaling. To identify modulated phosphoproteins in an unbiased fashion, we applied a quantitative, label-free MS-based approach (23, 26). The ESFT cell line A673 was transduced with an EWS/FLI1 shRNA construct or empty vector control. Phosphotyrosine or phosphoserine/threonine enrichment followed by tandem MS was used to quantitate relative phosphopeptide levels. Global changes in phosphorylation levels upon EWS/FLI1 knockdown were calculated by determining the phosphopeptide ratio between EWS/FLI1 shRNA and control cells.
Analysis of serine/threonine phosphopeptides detected 571 unique phosphopeptides corresponding to 336 proteins, and phosphotyrosine profiling identified 16 phosphopeptides (Supplementary Tables S1–S5). The phosphoserine/threonine dataset was filtered for peptides that were modulated upon EWS/FLI1 knockdown. This generated a list of 210 phosphopeptides, 86 of which showed an increase in phosphorylation and 124 of which displayed a decrease (Fig. 1A). Because change in phosphorylation could be due to change in total protein level, we compared this list with known genes that are regulated by EWS/FLI1 (31). Only 23 of 210 phosphopeptides were associated with genes that are modulated by EWS/FLI1 (Supplementary Fig. S1B and S1C), suggesting the majority of phosphopeptide modulation is not solely due to EWS/FLI1 transcriptional regulation. DAVID (32, 33) was used to determine pathways and biologic processes that were enriched in response to EWS/FLI1 inhibition (Fig. 1B and C; Supplementary Tables S6–S9). Phosphopeptides whose levels were increased after EWS/FLI1 knockdown were associated with adhesion and cytoskeletal organization (Fig. 1B). This agrees with a recent study that demonstrated that ESFT cells display increased adhesion and migration upon EWS/FLI1 knockdown (34). The phosphopeptides that displayed a decrease in phosphorylation were mainly associated with cell-cycle regulation (Fig. 1B and C). Because A673 cells with diminished EWS/FLI1 expression proliferate at a reduced rate, this likely contributes to the enrichment of cell-cycle–associated terms observed after EWS/FLI1 knockdown.
Phosphotyrosine and phosphoserine/threonine peptides were rank ordered based on the sum of the log fold change in phosphopeptide levels between EWS/FLI1 knockdown and control samples (Figs. 1A and 2A). The most downregulated phosphoprotein was IRS2 (insulin receptor substrate 2; Fig. 1A), an adapter protein that transmits signals from insulin and insulin-like growth factor receptors. This result is consistent with previous studies that have shown that EWS/FLI1 modulates components of the IGF1 pathway (28, 35). We also observed a large decrease in phosphorylation of PRKCB (protein kinase C beta; Fig. 1A), which was recently described to be overexpressed in ESFT. PRKCB has been demonstrated to be a direct target of EWS/FLI1, and inhibition of the protein reduces ESFT growth in vitro and in vivo (36). Phosphotyrosine-based rank ordering revealed the most differentially regulated phosphopeptide corresponded to an increase in phosphorylation of STAT3 at tyrosine 705 (Fig. 2A), with an average fold change of 11.5 between phosphopeptide levels in EWS/FLI1 knockdown and control cells. STAT3 phosphorylation was confirmed with a phospho-specific antibody (Fig. 2B). Quantitative immunoblot analysis detected low levels of phospho-STAT3 in control cells that increased an average of 15.7-fold upon EWS/FLI1 knockdown (Fig. 2C). This response was also observed with the use of a second shRNA construct (EF4; Fig. 2B).
Phospho-STAT3 upregulation primarily occurs in a subset of cells untransduced by lentiviral shRNA
To measure STAT3 activation as a result of phosphorylation, STAT3 immunofluorescence was performed to visualize localization before and after EWS/FLI1 knockdown in A673 cells. Phosphorylation at tyrosine 705 allows STAT3 to dimerize, then translocate to the nucleus where it acts as a transcription factor (37). STAT3 immunostaining showed largely nuclear signals in both knockdown and control cells, with slightly more intense cytoplasmic staining in the control cells (Fig. 3A). A similar pattern was observed with a distinct STAT3 antibody and at higher magnification (Supplementary Fig. S2). The nuclear localization suggests that active STAT3 signaling also occurs in cells that express EWS/FLI1.
The similarity in STAT3 staining between control and EWS/FLI1 knockdown cells led us to perform phospho-specific staining (Fig. 3B). A673 control cells displayed a low level of phospho-STAT3, whereas a subset of cells in the knockdown sample showed prominent staining. In general, this upregulation trend was expected based our immunoblot results. However, the subset of EWS/FLI1 knockdown cells that displayed high levels of phospho-STAT3 showed almost no overlap with the GFP-positive population marking cells transduced with the lentiviral shRNA. These data suggest a paracrine mechanism in which cells with successful EWS/FLI1 knockdown cause the activation of STAT signaling in cells that maintain EWS/FLI1 expression.
To quantitate phospho-STAT3 levels in ESFT control and EWS/FLI1 shRNA GFP-positive and -negative populations, we performed phospho-specific flow cytometry. ESFT cells transduced with empty vector control or EWS/FLI1 shRNA were divided into two populations based on GFP fluorescence intensity, and phospho-STAT3 levels were measured through the use of a fluorochrome-conjugated phospho-specific antibody. EWS/FLI1 transcript levels in EWS/FLI1 shRNA GFP-negative populations were similar to those of control cells, whereas levels in GFP-positive cells were reduced by approximately 85% (Supplementary Fig. S3D). When comparing all A673 control and EWS/FLI1 shRNA cells, those transduced with EWS/FLI1 shRNA displayed an increase in MFI of 3.17 (Supplementary Fig. S3A). However, when this comparison was performed on GFP-negative and -positive populations, the GFP-negative cells showed a nearly 4-fold increase in MFI for cells transduced with EWS/FLI1 shRNA, whereas GFP-positive cells showed only a 1.9-fold increase (Fig. 3C). Similar effects were observed using A4573 cells (Supplementary Fig. S3B). Although the magnitude of the fold change was smaller in A4573 cells, it was reproducible and statistically significant (Supplementary Fig. S3C). These results support the concept that the upregulation of phospho-STAT3 after EWS/FLI1 knockdown occurs primarily in a population untransduced by the lentiviral shRNA and is thus uninfluenced cell autonomously by EWS/FLI1 knockdown.
Soluble factors secreted upon EWS/FLI1 knockdown are sufficient to induce STAT3 phosphorylation
The evidence that phospho-STAT3 upregulation and EWS/FLI1 knockdown occur in separate subsets of cells suggested that these populations could be communicating with each other either though direct cell-to-cell contact or through secretion of soluble factors. To test for the presence of soluble factors, conditioned media from ESFT cells transduced with either vector control or EWS/FLI1 shRNA (818) were added to control cells, and STAT3 phosphorylation was assayed by immunoblot. Conditioned media from EWS/FLI1 shRNA but not control cells were able to stimulate phospho-STAT3 (Fig. 4A and B). This response occurs quickly and is maintained for at least 24 hours (Fig. 4B). STAT3 phosphorylation was also induced with conditioned media from cells transduced with a second EWS/FLI1 shRNA construct (EF4) and with serum-free conditioned media (Fig. 4A). This indicates that a soluble factor secreted upon EWS/FLI1 knockdown is responsible for STAT3 activation.
Two of the four ESFT cell lines assayed for the ability of conditioned media from EWS/FLI1 shRNA–transduced cells to stimulate phospho-STAT3, TC-174, and SK-N-MC displayed only a weak response. To determine if this is due to a lack of secreted factor in the conditioned media or expression of the appropriate receptor on the cell surface, we added A673 EWS/FLI1 shRNA conditioned media to these cells (Fig. 4B). Both cells lines showed a large increase in phospho-STAT3 upon conditioned media exposure, indicating they express the appropriate receptors, but do not secrete as much of the soluble factor upon EWS/FLI1 knockdown as the A673 cells. This may be due to a combination of lower viral transduction rates when compared with the A673 cells (Supplementary Fig. S4) and that ESFT cell lines other than A673 undergo growth arrest after EWS/FLI1 knockdown (16).
To determine which soluble factor(s) were responsible for the increase in phospho-STAT3, we used an antibody array to simultaneously measure 174 cytokines and growth factors in serum-free conditioned media from A673 cells transduced with empty vector control or EWS/FLI1 shRNA. A few cytokines displayed a dramatic increase in signal intensity, whereas the majority of the factors showed little or no change upon EWS/FLI1 knockdown (Fig. 4C). The status of all 174 cytokines is included in Supplementary Table S10. In particular, IL6, GM-CSF, and CXCL1 (GRO-α) were present in much higher levels in the EWS/FLI1 knockdown conditioned media compared with that of control (Fig. 4C and D). To test which of these factors is able to activate STAT3, purified, recombinant proteins were added to ESFT cells and phospho-STAT3 levels were compared with cells treated with conditioned media from EWS/FLI1 shRNA (818)–transduced cells. Only IL6 was able to stimulate STAT3 phosphorylation, though not to the level of the conditioned media (Fig. 4E). Because IL6 is a known activator of STAT3, we chose to further investigate its role in ESFT signaling.
STAT3 phosphorylation is induced primarily through an IL6-dependent mechanism
ELISA analysis was performed to validate the results of the cytokine array and quantitate the levels of IL6 secreted by ESFT cells. Conditioned media from A4573 and A673 EWS/FLI1 knockdown cells contained elevated levels of IL6 (Fig. 5A). Quantitation of IL6 transcript levels also demonstrated an increase in IL6 RNA upon EWS/FLI1 knockdown (Fig. 5B).
To determine if IL6 is necessary for STAT3 activation, we inhibited either the ligand or its receptor gp130. First, an IL6 neutralization antibody was added to EWS/FLI1 knockdown (818) conditioned media and subsequently immunoprecipitated. Immunodepletion of IL6 was confirmed by ELISA (Fig. 5C and D), and the media were added to untransduced cells. Resulting phospho-STAT3 levels were compared with cells treated with control or EWS/FLI1 knockdown conditioned media (Fig. 5E and F). Especially in A673 cells, removing IL6 prevents phosphorylation of STAT3. Conditioned media in which IL6 is only partially immunodepleted (Fig. 5D) retain the ability to stimulate STAT3 phosphorylation (Fig. 5F). An analogous experiment was performed in which ESFT cells were preincubated with a gp130 neutralization antibody. EWS/FLI1 knockdown conditioned media were added to these cells as well as those that were not pretreated with the antibody. Evaluation of STAT3 phosphorylation revealed that blocking gp130 also inhibited upregulation of phospho-STAT3 (Fig. 5G). These results provide evidence that STAT3 is being activated through an IL6-dependent mechanism (Fig. 5I).
We next used phospho-specific immunoblot analysis to examine the activity of STAT3 and potential upstream kinases upon EWS/FLI1 knockdown. Although we did not observe a difference in STAT3 phosphorylation at serine 727 between control and unsorted EWS/FLI1 knockdown cells, examination of sorted populations revealed that GFP-negative cells possess increased levels of phospho-S727 compared with GFP-positive cells with reduced EWS/FLI1 expression (Fig. 5H). This demonstrates that paracrine activation of STAT3 results in increased phosphorylation at both tyrosine 705 and serine 727. Examination of GFP-positive and -negative populations of cells transduced with EWS/FLI1 shRNA also showed that both JAK2 and SRC family kinases (SFK) displayed reduced phosphorylation at sites within the activation loop of the kinase domain after EWS/FLI1 knockdown (Fig. 5H). GFP-positive cells also displayed a modest decrease in gp130 levels, which may contribute to the decrease in kinase JAK2 and SFK activity. Furthermore, GFP-positive cells with reduced EWS/FLI1 expression appear to possess less total STAT3 than GFP-negative cells. This decrease in total protein and diminished activity of upstream kinases could contribute to the decreased paracrine STAT3 activation observed in cells with reduced levels of EWS/FLI1. Further experiments are warranted to fully elucidate this mechanism.
STAT3 plays a complex role in ESFT growth and survival
Because activation of STAT3 promotes tumorigenesis through upregulation of cell survival and proliferation factors (38), we sought to investigate its effects on ESFT cell growth. The paracrine activation of STAT3 that occurs upon EWS/FLI1 knockdown suggests that these cells might display increased proliferation rates or sensitivity to STAT3 inhibition. In addition, the observed nuclear localization and basal phosphorylation of STAT3 indicate a possible dependence on STAT3 signaling in cells that maintain EWS/FLI1 expression. To investigate the role of STAT3 in each of these populations, we used a small-molecule inhibitor, Stattic (39), and dominant-negative construct (18) to inhibit STAT3 phosphorylation and measured subsequent effects on ESFT cell proliferation.
STAT3 phosphorylation at tyrosine 705 was validated by ELISA in control ESFT cells in which paracrine STAT3 activation has not been induced by EWS/FLI1 knockdown (Supplementary Fig. S5B). Stattic treatment revealed these cells are sensitive to STAT3 inhibition, with an IC50 of approximately 2 μmol/L (Supplementary Fig. S5C). Increasing concentrations of Stattic were demonstrated to inhibit STAT3 phosphorylation in EWS/FLI1 knockdown and control cells (Supplementary Fig. S5A and S5B) compared with DMSO-treated controls. Stattic also inhibited the proliferation of ESFT cells regardless of EWS/FLI1 expression, though greater inhibitory effects were observed in EWS/FLI1 knockdown cells treated with Stattic (Fig. 6A). EWS/FLI1 knockdown reduced ESFT cell proliferation to degrees that varied based on knockdown efficiency. A larger growth inhibitory effect was observed in A673 cells due to more potent reduction of EWS/FLI1 expression (Supplementary Fig. S4A and S4B). Additional growth assays using a dominant-negative construct also demonstrated that STAT3 inhibition diminishes ESFT cell growth (Fig. 6B). Furthermore, dominant-negative STAT3 hinders EWS/FLI1-mediated STAT3 phosphorylation (Supplementary Fig. S5D), and combined inhibition of STAT3 and EWS/FLI1 has an increased effect compared with targeting EWS/FLI1 alone (Fig. 6B).
Given the role of STAT3 in ESFT growth, we next asked if the paracrine activation of STAT3 that occurs upon EWS/FLI1 knockdown results in increased cellular proliferation. In some instances, ESFT cells treated with conditioned media derived from cells transduced with EWS/FLI1 shRNA displayed increased growth compared with cells treated with conditioned media from control cells. However, these results were not consistent across various anchorage-dependent and -independent assays (data not shown). As a result, we focused on the role of factors secreted upon EWS/FLI1 knockdown to promote cell survival.
ESFT cell lines were treated with conditioned media containing reduced serum to induce apoptosis. After 24 hours of treatment, we observed an increased amount of cleaved PARP compared with untreated controls. ESFT cells treated with conditioned media derived from EWS/FLI1 knockdown cells displayed significantly less PARP cleavage than those treated with conditioned media from control cells (Fig. 6C and D). This effect is mediated in part by IL6. Adding IL6 to control conditioned media reduced PARP cleavage, and immunodepleting IL6 from knockdown conditioned media increased PARP cleavage (Fig. 6E). These results demonstrate that soluble factors secreted upon EWS/FLI1 knockdown confer protection against apoptosis.
The paracrine STAT3 activation that occurs upon EWS/FLI1 knockdown implies targeting this pathway could sensitize ESFT cells to EWS/FLI1-directed therapy. Therefore, we evaluated the effects of combining a STAT3 inhibitor with cytotoxic agents that inhibit EWS/FLI1 function. When Stattic was combined with either cytarabine or mithramycin, mostly additive effects were observed. Synergy was only observed at the highest dose levels tested (data not shown). However, because cytarabine disrupts DNA synthesis by acting as a nucleoside analog and mithramycin inhibits RNA synthesis by binding to the minor groove of DNA, neither of these agents specifically targets EWS/FLI1. Additional targets hit by these drugs may obscure the effects of Stattic inhibition.
Because specific small-molecule inhibitors of EWS/FLI1 are not available, we explored the effects of combining STAT3 inhibition with other ESFT therapies. We first tested if IL6-mediated paracrine STAT3 activation occurs as a response to stresses other than EWS/FLI1 knockdown. Treatment of ESFT cells with the chemotherapeutic agent doxorubicin resulted in a dose-dependent increase in IL6 secretion. However, the levels of IL6 in conditioned media from doxorubicin-treated ESFT cells were 2 to 3 orders of magnitude less than those observed for media from EWS/FLI1 knockdown cells. In addition, these conditioned media were able to increase STAT3 phosphorylation in ESFT cells, but not to the extent of knockdown conditioned media (data not shown). Although less robust than the effects observed upon EWS/FLI1 knockdown, doxorubicin-induced paracrine STAT3 activation provides a rationale for combining STAT3 inhibition with therapeutics other than those that inhibit EWS/FLI1. Therefore, we evaluated the effects of combining Stattic with conventional and targeted therapies used for the treatment of ESFT.
When Stattic was combined with the IGF1R small-molecular inhibitor NVP-AEW541, synergy was observed in both A673 and A4573 cells (Fig. 6F and G, Supplementary Fig. S6A and S6B). In A673 cells, combining Stattic with doxorubicin displayed synergistic effects (Fig. 6H and I). In A4573 cells, while the effective doses of the combination treatment lied below the linear additive isoboles (Supplementary Fig. S6D), the combination indices for four of five dose levels were approximately 1, indicating additivity (Supplementary Fig. S6C). These data indicate that STAT3 inhibition could increase the efficacy of ESFT therapies.
Less focus has been placed on the role of signal transduction in ESFT because tumor progression is primarily driven by EWS/FLI1-mediated regulation of gene expression. However, low response rates for clinical testing of targeted therapeutics in ESFT emphasize the necessity for better understanding of cellular signaling. Our studies aimed to generate a global, unbiased view of changes in cellular signaling upon EWS/FLI1 inhibition to gain further insight into potential mechanisms of drug resistance. Our results included novel phosphoproteins modulated by EWS/FLI1 as well as the elucidation of a paracrine signaling pathway. Tyrosine phosphoprofiling revealed STAT3 phosphorylation to be upregulated upon EWS/FLI1 knockdown. Single-cell analysis demonstrated that this does not occur through direct regulation, but through a paracrine mechanism mediated in strong part by IL6 secretion. STAT3 inhibition reduced ESFT cell growth alone or in combination with EWS/FLI1 knockdown and enhanced the effects of chemotherapeutics and targeted agents in ESFT. Furthermore, IL6-containing conditioned media from EWS/FLI1 knockdown cells were demonstrated to have antiapoptotic effects.
STAT3 is persistently activated in multiple malignancies and promotes tumorigenesis by upregulating cellular proliferation and survival factors as well as those that promote immunosuppression (40). This activation can occur through IL6 secretion by tumor cells or stromal cells within the tumor microenvironment. In ESFT, STAT3 is phosphorylated in approximately 50% of tumor samples in addition to multiple cell lines (41, 42). Previous studies have also demonstrated the role of STAT3 in ESFT proliferation. Treatment with a specific STAT3 inhibitor reduced the growth of ESFT cell lines in vitro (41). In addition, targeting JAK1/2 blocked both endogenous and IL6-mediated STAT3 activation in ESFT and inhibited cell growth in vitro and in vivo (43). Our own independent assessment with a distinct STAT3 inhibitor and dominant-negative construct corroborates these results. Our work further expands upon the role of IL6/JAK/STAT3 signaling in ESFT by characterizing the induction of STAT3 activity that occurs upon EWS/FLI1 inhibition. We also demonstrated the benefits of combining a STAT3 inhibitor with other agents.
Elevated IL6 levels in the tumor microenvironment have been shown to promote tumor cell proliferation and induce drug resistance. Lower drug efficacy due to cytokine secretion has been observed in HER2-positive breast cancer, where trastuzumab resistance is mediated by IL6 secretion that leads to the expansion of a stem cell subpopulation (44). In lung cancer, IL6 production by stromal fibroblasts or tumor cells harboring EGFR mutations led to STAT3 activation and resistance to the irreversible EGFR inhibitor afatinib (45). In addition, paracrine IL6 production protected both neuroblastoma and osteosarcoma cells from drug-induced apoptosis and increased the proliferation and migration of osteosarcoma cells (46, 47). Because IL6 is secreted upon EWS/FLI1 knockdown, we hypothesized that soluble factors could also play a role in ESFT pathogenesis. The antiapoptotic effects of IL6 containing conditioned media and synergistic effects from combining a STAT3 inhibitor with existing ESFT therapeutics that we observed indicate secreted IL6 may also promote drug resistance in ESFT. In addition, analysis of serum levels of patients with bone tumors including Ewing sarcoma demonstrated significantly elevated IL6 levels, which correlated with poor overall survival (48). This study supports our data that factors secreted in the tumor microenvironment enhance tumor cell survival. While our initial observation for a role of STAT3 signaling in ESFT cell survival involved a paracrine signaling event between unaffected and EWS/FLI1 knockdown cells, our cotreatment synergy results demonstrate that STAT3 signaling does play a complex survival role in EWS/FLI1-expressing cells. Further interactions between tumor cells and their microenvironment, such as stromal cell secretion of STAT3-inducing ligands, will also need to be characterized.
Although we have demonstrated that increased STAT3 phosphorylation that occurs upon EWS/FLI1 knockdown is mediated in strong part by IL6 secretion, more work is needed to fully elucidate this mechanism. Our data indicate that IL6 is the predominant factor, but blocking IL6 or gp130 did not completely abrogate induction of STAT3 phosphorylation. This argues that other secreted factors also contribute to STAT3 activation. Our cytokine array results revealed multiple growth factors and cytokines that were upregulated upon EWS/FLI1 knockdown, including IL8, GM-CSF, and CXCL1. Ewing sarcoma patient serum also contained additional elevated cytokines such as IL8, IL1ra, and M-CSF (48), suggesting that a combination of soluble factors cooperates with IL6 to mediate its effect. In addition, it is unclear how IL6 production is increased upon EWS/FLI1 knockdown. IL6 is one of several proinflammatory cytokines whose expression is mediated by the transcription factor NF-κB (49). If EWS/FLI1 represses NF-κB, release of this inhibition upon EWS/FLI1 knockdown is one possible explanation for an increase in IL6 levels. Furthermore, STAT3 activation and IL6 production can be propagated by a feedforward loop, so a small initial increase in IL6 may result in a larger, sustained response (40).
In summary, our investigation uncovered a novel paracrine signaling pathway that expanded upon the role of STAT3 signaling in ESFT pathogenesis. This provides a rationale for combining inhibitors of this pathway with other agents to enhance the efficacy of ESFT therapies. Several agents targeting components of the IL6/JAK/STAT3 pathway have been evaluated in the clinical setting, including JAK inhibitors and monoclonal antibodies that block IL6 or the IL6 receptor (50). In addition, dasatinib, which inhibits tyrosine kinases including SRC, is currently being evaluated in phase I/II trials for sarcoma both as a single agent and in combination with other therapies (ClincialTrials.gov). These studies, taken together with our results, suggest that the use of additional agents directed against members of the IL6/JAK/STAT3 pathway could improve clinical responses in ESFT.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: J.L. Anderson, C.T. Denny
Development of methodology: T.G. Graeber
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J.L. Anderson, B. Titz, R. Akiyama, A. Park, W.D. Tap, T.G. Graeber
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J.L. Anderson, E. Komisopoulou, T.G. Graeber
Writing, review, and/or revision of the manuscript: J.L. Anderson, B. Titz, W.D. Tap, T.G. Graeber, C.T. Denny
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Park
Study supervision: T.G. Graeber, C.T. Denny
Flow cytometry was performed in the UCLA Jonsson Comprehensive Cancer Center (JCCC) and Center for AIDS Research Flow Cytometry Core Facility that is supported by NIH awards CA-16042 and AI-28697, the JCCC, the UCLA AIDS Institute, and the David Geffen School of Medicine at UCLA. The authors thank Matteo Pellegrini (UCLA) for helpful discussions and providing bioinformatics assistance.
J.L. Anderson received support from the Ruth L. Kirschstein National Research Service Award GM07185 and a UCLA Graduate Division Dissertation Year Fellowship. This work was also supported by NIH grant CA087771 (to C.T. Denny). T.G. Graeber is the recipient of a Research Scholar Award from the American Cancer Society (RSG-12-257-01-TBE) and an Established Investigator Award from the Melanoma Research Alliance (20120279). T.G. Graeber and C.T. Denny are supported by NIH/National Center for Advancing Translational Science (NCATS) UCLA CTSI grant number UL1TR000124.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.