A Novel Inhibitor of STAT3 Homodimerization Selectively Suppresses STAT3 Activity and Malignant Transformation

STAT3 is an important regulator of many biological processes including proliferation, survival, inflammation, and immune responses (1, 2). STAT3 mediates these processes by responding to ligands such as growth factors [i.e., EGF, platelet-derived growth factor (PDGF), etc.] and cytokines [i.e., interleukin (IL)-6, IFN-γ, etc.], which activate STAT3 to translocate to the nucleus and regulate the expression of a number of genes (1, 2). For example, the binding of EGF to its receptor results in tyrosine phosphorylation of the EGF receptor (EGFR) and subsequent recruitment of STAT3 through the binding of the STAT3-SH2 domain to phospho-tyrosines 1068 and 1086 on the receptors (3). Similarly, nonreceptor tyrosine kinases such as Janus-activated kinase (JAK)2 and Src, which are part of nontyrosine kinase receptor complexes (i.e., IL-6 receptor complex) phosphorylate a specific tyrosine on STAT3, which in turn induces STAT3–STAT3 dimerization through 2 reciprocal phosphotyrosine–SH2–binding interactions (4). The activated STAT3 dimers translocate to the nucleus where they bind to specific DNA sequences on the promoters of the genes regulated by STAT3 (1, 2). Under physiological normal conditions, this STAT3 activation is rapid (within 2 minutes of ligand stimulation) and transient (lost within a few hours due to dephosphorylation). In contrast, STAT3 is found persistently tyrosine phosphorylated and constitutively activated in the majority of cancers including pancreatic, breast, lung, prostrate, ovarian, colon, gastric, and head and neck cancers as well as melanoma, leukemia, multiple myeloma, and lymphoma (4). Constitutively activated STAT3 is believed to contribute to malignant transformation at several levels (5). These include uncontrolled proliferation through activation of several cell-cycle regulators, such as cyclin D1 and c-Myc as well as evasion of apoptosis by inducing the expression of several antiapoptotic proteins, such as Bcl-xL, Bcl-2, Mcl-1, and survivin. STAT3 also activates the expression of proteins involved in other hallmarks of cancer, such as invasion and metastasis [i.e., expression of matrix metalloproteinase (MMP)] and angiogenesis (i.e., expression of VEGF; refs. 2, 4). The fact that STAT3 regulates the expression of a number of genes involved in oncogenesis makes it an attractive and promising target for cancer therapy (6, 7). Validation of STAT3 as a target for cancer drug discovery comes from several lines of evidence. For example, a genetically engineered mutant of STAT3 (STAT3-C) that forms a constitutively dimerized STAT3 through disulfide bonds is oncogenic (8). On the other hand, a dominant-negative variant of STAT3, STAT3β, blocks tumor growth by inhibiting STAT3 in tumors where STAT3 is constitutively activated (9, 10). Several approaches have been proposed to suppress constitutive activation of STAT3. These include those inhibiting STAT3 tyrosine phosphorylation (i.e., inhibition of JAK2 or Src), STAT3 recruitment to the receptor and dimerization (i.e., phosphotyrosine peptide mimics that bind the SH2 domain of STAT3), STAT3 nuclear translocation and STAT3–DNA–binding and transcriptional activation (6, 7). We have focused our efforts on identifying small molecules capable of disrupting the phosphotyrosine–SH2–binding interactions of STAT3 as potential dimerization inhibitors. Here, we report on a novel STAT3–STAT3 dimerization inhibitor that selectively inactivates STAT3 and suppresses STAT3-dependent malignant transformation.

Human breast cancer (MDA-MB-468, MDA-MB-231, and MDA-MB-453), lung cancer (A549, H358, and H460) cells, human nontumorigenic epithelial cells (MCF10A), and human embryonic kidney cells (HEK293) were obtained from the American Type Culture Collection. These cell lines have not been authenticated. HEK293 cell lines with stable transfection of HA-STAT3 and FLAG-STAT3 were generated as described later. Cells were grown in Dulbecco's Modified Eagle's Medium (DMEM), RPMI-1640, and DMEM/F-12 containing 10% heat-inactivated FBS. MCF10A was cultured in DMEM/F12, supplemented with 5% horse serum (Invitrogen), hydrocortisone (0.5 μg/mL), mouse EGF (20 ng/mL), insulin (10 μg/mL), cholera toxin (100 ng/mL; Sigma). Primary antibodies against pY705STAT3, pAKT, AKT, pErk1/2, Erk, MMP-9, and cyclin D1 were purchased from Cell Signaling Technology. Primary antibodies against STAT3, Bcl-xL, survivin, hemagglutinin (HA; anti-mouse), and HA (anti-rabbit) were purchased from Santa Cruz Biotech. Primary antibody against FLAG was purchased from Sigma.

FLAG-Stat3 plasmid was obtained from Addgene. HA-Stat3 DNA was amplified from FLAG-Stat3 using PCR with HA-pcDNA3 as a vector as described for RhoB by our laboratory (11). The primers used for PCR were STAT3F-BamH1 cgcggatccgccaccATGGCTCAGTGGAACCAGCTG and STAT3R-EcoR1 ccggaattcTCACATGGGGGAGGTAGCACA. The PCR product was digested with BamH1 and EcoR1, and cloned into HA-pcDNA3 vector, further confirmed by sequencing. Ratio (1:1) of pFLAG-STAT3 and pHA-STAT3 plasmid DNAs were cotransfected into HEK293 cells, and stable G418-resistant (800 μg/mL) clones were selected. Transfections were carried out with LipofectAmine Plus (Invitrogen), according to the manufacturer's protocol.

Nuclear extract preparation was carried out as previously described (12). The STAT3–DNA–binding filter plate assay was conducted following the manual of the Filter Plate Assay Kit (Signosis), as described previously for NF-κB (13). The transcription factor (TF) binding buffer was mixed with the STAT3 probe (biotin-labeled STAT3–DNA–binding sequence) and nuclear extract and incubated at 16°C for 30 minutes to form the STAT3–DNA complex. The STAT3–DNA complex was then separated from free probe by using a filter plate. After several steps of binding and washing, bound STAT3 probe is retained on the filter in the corresponding well of filter plate and the free DNA probe is removed. The bound prelabelled STAT3 probe was then eluted from the filter plate by centrifuge with elution buffer. Eluted probes were then hybridized into a 96-well hybridization plate for quantitative analysis. The captured STAT3 probe was further detected by conjugation with streptavidin–horseradish peroxidase. The chemiluminescence of each well was detected by 2104 EnVision Multilabel Reader of PerkinElmer (PekinElmer) within 5 minutes after mixture with substrates.

Fluorescence polarization (FP) assay was conducted on the basis of fluorescence signal differences between free and STAT3-bound fluorescently labeled peptide as described by Schust and Berg (14, 15). Briefly, all reactions contained 10 nmol/L of the fluorescent peptide 5-FAM–G(pTyr) LPQTV-CONH₂ (Genscript) and 100 nmol/L GST-tagged, full-length human STAT3 protein (SignalChem) in 96-well half-area black plates (Corning). For evaluating compounds, STAT3 protein was incubated with various concentrations of S3I-1757 or S3I-1756 at room temperature for 60 minutes in the assay buffer (50 mmol/L NaCl, 10 mmol/L HEPES, pH 7.5, 1 mmol/L EDTA, 0.01% Triton-X100, and 2 mmol/L dithiothreitol). Fluorescent peptide was added at a final concentration of 10 nmol/L and incubated for 30 minutes at room temperature following which the fluorescence polarization measurements were examined by 2104 EnVision Multilabel Reader (PekinElmer) using FITC FP Dual module with excitation filter of FITC FP 480 and emission filter of FITC FP P-pol535 and S-pol535. The Z′ value was derived per the equation Z′ = 1 − (3SD_bound + 3SD_free)/(mP_bound − mP_free), in which mP is the average of fluorescence polarization.

MDA-MB-468 cells were plated into a 12-well plate with 4 × 10⁵ cells per well. The cells were transiently transfected with pLucSRE, pLucTKS3, or STAT3-C with β-gal and then were treated with vehicle, S3I-1756, or S3I-1757 for 48 hours. Then cytosolic extracts of equal total protein were prepared from S3I-1757–treated or -untreated and analyzed for luciferase activity using a TD-20/20 luminometer (Turner Designs) described by us previously (16).

HEK293/FLAG-Stat3/HA-Stat3 cells were cultured (4,000 cells per well) with G418 400 μg/mL in a 8-well chamber slide. The cells were treated with S3I-1757 or S3I-1756 for 1, 2, or 4 hours. The cells were then rinsed with PBS, fixed with ice-cold methanol for 15 minutes, washed 3 times in PBS, permeabilized with 0.2% Triton X-100 for 15 minutes, and further washed 3 to 4 times with PBS. Cells were then blocked in 1% bovine serum albumin (BSA) for 30 minutes and incubated with anti-HA (Santa Cruz) or anti-FLAG (Sigma) primary antibody at 1:100 dilution at 4°C overnight. Subsequently, cells were rinsed 4 to 5 times in PBS, incubated with Alexa Fluor secondary antibody (Invitrogen) for 1 hour at room temperature in the dark. Cells were then washed 5 times with PBS, covered with cover slides with VECTASHIELD-mounting medium containing 4′,6-diamidino-2-phenylindole (DAPI; Vector Laboratories, Inc.), and examined immediately under Zeiss Upright Fluorescence Microscope (Zeiss).

MDA-MB-468 cells were plated at 4,000 cells per well in a 8-chamber slide. The cells were treated the following day with vehicle, S3I-1756, or S3I-1757 for 2, 4, or 18 hours. Cells were fixed, washed, and permeabilized as described earlier. Specimens were then blocked in 1% BSA for 30 minutes and incubated with anti-P-STAT3 (Cell Signaling) antibody at 1:50 dilution at 4°C overnight. Subsequently, cells were rinsed 4 to 5 times in PBS and incubated with Alexa Fluor secondary antibody for 1 hour at room temperature in the dark. Specimens were then washed 5 times with PBS, covered with cover slides with VECTASHIELD-mounting medium containing DAPI (Vector Laboratories, Inc.), and examined immediately under Zeiss Upright Fluorescence Microscope (Zeiss).

MTT assay was conducted exactly as described by our laboratory (17) to determine the effects of S3I-1757 on cell proliferation. Briefly, cells were plated in a 96-well tissue culture plate (2,000 cells per well) and incubated for 12 hours. After incubation, the cells were treated with vehicle, S3I-1757 or S3I-1756 for 48 hours. After incubation, freshly prepared MTT (3 mg/mL) in 1× PBS was added to each well and incubated for 3 hours and the plate was read at 570 nm.

Cells were cultured at 500 cells per well in a 12-well plate with regular growth medium. Cells were treated by vehicle and 1757 at 50, 100, and 200 μmol/L on the following day. And cells were allowed to grow for 2 to 3 weeks until the colonies were visible. A total of 3 mg/mL MTT in PBS Buffer (Invitrogen) was used to stain the colonies for 4 hours.

A549, MDA-MB-231, and H460 cells were seeded at 6 × 10⁵ cells per dish into 60-mm plate and allowed to grow overnight. Wounds were made the following day by scratching the cells with pipette tips (1–10 μL). Cells were treated with vehicle, S3I-1756, or S3I-1757 and allowed to migrate into the scratched area for 16 hours in regular growth medium. The migration of cells was visualized at ×10 magnification using a Leica Microscope at time 0 (right before the drug was added) and 16 hours after vehicle, S3I-1756, or S3I-1757 treatments.

Invasion assay was conducted in BD BioCoat Matrigel Invasion Chamber in 24-well plates. A549, MDA-MB-468, MDA-MB-231, and H460 cells were seeded at 25,000 cells per insert in the top chamber over the Matrigel. The bottom chamber contains 20% FBS as the “chemoattractant.” Vehicle, S3I-1756, or S3I-1757 was added the following day. The cells were incubated for 48 hours, after which the cells in the top chamber were carefully removed and the filter membranes containing the invaded cells on the outside of the filter were fixed with methanol, stained with crystal violet, and photographed.

Soft agar colony formation assays were conducted in a 12-well plate as described previously by us (17). Briefly, cells were seeded at 2,000 cells per well in regular growth media containing 0.3% agar (Sigma) and S3I-1757 was added the following day. Colonies were allowed to grow for 3 to 4 weeks, and quantified by staining with 1 mg/mL MTT (Sigma) overnight.

Briefly, the GLIDE docking software (available from Schrödinger, Inc.) was used to dock small molecule 3-dimensional structures from National Cancer Institute (NCI) Plated Set to the ApY*LK site derived from the X-ray crystal structure of the STAT3 dimer bound to DNA (18). Schrödinger's Maestro 9.1 was used as the primary graphical user interface. Schrödinger's LigPrep 2.41 was used to prepare molecules for docking and Schrödinger's Protein Preparation workflow was used in the preparation of the protein structure. Schrödinger's GLIDE 5.6 was used for the generation of grid files and docking. Initially, structures were subjected to docking with GLIDE SP and then the structures from each SP job were subjected to GLIDE XP docking. Generally, at least 3 poses were saved for each run for visual inspection. PyMol (Schrodinger, Inc.) was used for graphical presentation of the results in the figures.

MDA-MB-468 cells were treated with vehicle [dimethyl sulfoxide (DMSO)], S3I-1757 at 50, 100, and 200 μmol/L, and S3I-1756 at 200 μmol/L for 18 hours. Cells were then lysed with radioimmunoprecipitation assay (RIPA) buffer [20 mmol/L Tris–HCl (pH 7.4), 5 mmol/L EDTA, 10 mmol/L Na₄P₂O₇, 100 mmol/L NaF, 2 mmol/L Na₃VO₄, 1% NP-40, 1 mmol/L phenylmethylsulfonylfluoride (PMSF), and 10 mg/mL aprotinin]. The cell lysates were denatured by boiling with 5× SDS-PAGE sample buffer for 5 minutes and run on SDS-PAGE gel. The proteins were then transferred to membranes that were blocked with 5% nonfat milk in TBS with 0.1% Tween-20 (TBST) buffer for 30 minutes at room temperature and incubated with primary antibodies (pY705STAT3, pAKT, AKT, pERK1/2, MMP-9, and cyclin D1) 4°C overnight at dilution of 1:1,000 in 3% BSA, followed by washing and incubation with secondary antibody at dilution of 1:1,000 in 5% nonmilk TBST buffer for 1 hour at room temperature. The membranes were then washed with 1× PBS buffer for 10 minutes for 3 times and developed with enhanced chemiluminescence (ECL) Kit (PerkinElmer).

HEK293/FLAG-STAT3/HA-STAT3 cells were treated for 4 hours with vehicle, S3I-1757, S3I-1756, Ac-G{pTYR}LPQTV-AAVLLPVLLAAP-NH₂, or Ac-GYLPQTV-AAVLLPVLLAAP-NH₂ and then lysed in 20 mmol/L Tris–HCl (pH 7.4), 5 mmol/L EDTA, 10 mmol/L Na₄P₂O₇, 100 mmol/L NaF, 2 mmol/L Na₃VO₄, 1% NP-40, 1 mmol/L PMSF, and 10 mg/mL aprotinin. For EGF stimulation group, HEK293/FLAG-STAT3/HA-STAT3 cells were treated by 100 ng/mL EGF for 30 minutes before making cell lysate. Protein A or G agarose (EMD Millipore) was washed twice with PBS and restored to 50% slurry with PBS. Five hundred micrograms of lysate were precleared by mixture of protein A and protein G agarose for 1 hour at 4°C and then remove protein A and G agarose by centrifuge at 13,000 rpm for 3 minutes. And then 500 μg of lysate was immunoprecipitated with 50 ng of HA antibody overnight at 4°C on shaker and then capture the immunocomplex by adding 100 μL protein A and G agarose/sepharose bead slurry for 1 hour at 4°C. Samples were washed 5 times with lysis buffer and then boiled in 5× SDS-PAGE sample buffer and run on SDS-PAGE gel. Protein was transferred to nitrocellulose membrane and then blotted as described earlier for HA, FLAG, P-STAT3, EGFR, and STAT3.

In an effort to identify STAT3–STAT3 dimerization disruptors, we have undertaken a major chemistry effort based on structure–activity relationship (SAR) studies using our previously published S3I-201 compound (16) as a starting point. These SAR studies identified several STAT3 inhibitors (19). In the present article, we focused on S3I-1757 and its closely related analog S3I-1756 (Fig. 1A). We first determined the potency of these molecules to disrupt the binding of STAT3 to fluorescein-labeled GpYLPQTV phosphotyrosine peptide by fluorescence polarization assays as described in Materials and Methods [the GpYLPQTV phosphotyrosine peptide corresponds to amino acids 903–909 from the gp-130 subunit of the IL-6 receptor and is known to bind the STAT3-SH2 domain (20, 21)]. Figure 1A shows that S3I-1757 inhibits the binding of STAT3 to the phosphotyrosine peptide in a concentration-dependent manner with an IC₅₀ value of 13.5 ± 0.5 μmol/L. In contrast, the closely related analog S3I-1756 had little effects with concentrations as high as 400 μmol/L (Fig. 1A). These data show that replacing the cyclohexyl group in S3I-1757 with a methoxy group as in S3I-1756 resulted in great (more than 29-fold) loss of potency to disrupt STAT3 phosphotyrosine-peptide binding.

To determine the potential mode of binding of S3I-1757 to the STAT3-SH2 domain, we carried out docking studies as described in Materials and Methods. Figure 1B shows a surface rendering of the SH2 domain of STAT3 with the active analog, S3I-1757 (yellow), and the inactive analog, S3I-1756 (green), bound on the basis of GLIDE XP docking. A ligand–protein interaction diagram is also shown in Fig. 1B. Figure 1B shows that S3I-1757 occupies the phosphotyrosine-binding site of STAT3. Furthermore, the salicylic acid group of S3I-1757 interacts with SH2 domain amino acids (i.e., Arg-609 and Lys-591) known to interact with P-Y-705 of STAT3 (18). The experimentally observed greater potency of S3I-1757 compared with S3I-1756 is most likely due to a number of factors as illustrated in Fig. 1B: (i) S3I-1757 forms 4 hydrogen bonds (H-bonds between Arg-609 and Glu-612) in the phospho-tyrosine–binding pocket, whereas S3I-1756 only forms 3 (H-bonds between Val-637, Arg-609, and Lys 591); (ii) S3I-1757 forms 2 excellent cation–π interactions with Lys 591 (with an N to phenyl ring centroid distance of 3.9 Å for the salicylate phenyl group and an N to phenyl ring centroid distance of 3.5 Å for the phenyl group to which the cyclohexyl substituent is attached); (iii) S3I-1756 forms 2 weaker cation–π interactions with Lys 591 (with an N to phenyl ring centroid distance of 4.6 Å for the salicylate phenyl group and an N to phenyl ring centroid 4.5 Å for the phenyl group to which the phenoxy group is attached); (iv) the surface rendering of the protein in Fig. 1B shows that the cyclohexyl substituent of S3I-1757 is buried in a pocket with a rather negative electrostatic potential, whereas the phenoxy substituent of S3I-1756 is similarly buried, although not as deeply and the π electron system of the phenoxy substituent would not interact favorably with this negatively charged environment; and (v) the methoxy-phenyl substituent of S3I-1756 is mostly solvent exposed and forms almost no interactions with the protein.

The data from the fluorescence polarization assays of Fig. 1A coupled with the molecular modeling results of Fig. 1B suggest that S3I-1757 inhibits STAT3 dimerization. However, the fluorescence polarization assay only measured the ability of S3I-1757 to displace the 7 amino acid fluorescein-labeled GpYLPQTV phosphotyrosine peptide from the SH2 domain of STAT3 protein in vitro. Therefore, we next determined if the intracellular reciprocal binding of 2 full-length STAT3 proteins is inhibited, as the ultimate goal is to develop S3I-1757 as a small molecule capable of disrupting STAT3 dimerization and subsequently inactivating constitutively activated STAT3. To this end, we developed an assay that measures directly STAT3–STAT3 dimerization in intact cells by cloning HA-tagged STAT3 and FLAG-tagged STAT3, generating HEK293 cells that stably coexpress HA-STAT3 and FLAG-STAT3 and using these cells for coimmunoprecipitation and colocalization experiments as described in Materials and Methods. Figure 2A shows that FLAG-STAT3 coimmunoprecipitated with HA-STAT3 in HEK293 cells that coexpress FLAG-STAT3 and HA-STAT3 but not in empty vector-transfected HEK293 cells (Vector). We validated this assay by treating the FLAG-STAT3/HA-STAT3 HEK293 cells with tyrosine phosphorylated or nonphosphorylated GYLPQTV peptide fused to a membrane-translocating sequence to allow cell uptake. Figure 2A shows that, in vehicle-treated cells (V), FLAG-STAT3 coimmunoprecipitated with HA-STAT3, and that treatment with phosphorylated GpYLPQTV (P-Pep) but not nonphosphorylated GYLPQTV (Pep) inhibited FLAG-STAT3 from coimmunoprecipitating with HA-STAT3. To determine if the compounds can inhibit the binding of HA-STAT3 to FLAG-STAT3, HA-STAT3/FLAG-STAT3/HEK293 cells were treated with vehicle (V), S3I-1757, or S3I-1756 and the cells were processed for immunoprecipitation with HA antibodies and immunoblotted with FLAG antibodies as described in Materials and Methods. Figure 2A shows that in vehicle-treated cells, FLAG-STAT3 coimmunoprecipitated with HA-STAT3, and that treatment with S3I-1757 inhibited the binding of FLAG-STAT3 to HA-STAT3 in a concentration-dependent manner. In contrast, S3I-1756 did not inhibit this binding with concentrations as high as 200 μmol/L (Fig. 2A). We next determined if S3I-1757 could disrupt the binding of HA-STAT3 to the EGFR, as the same STAT3-SH2 domain that is used for STAT3–STAT3 dimerization is also used to bind EGFR on p-Tyr-1068 and p-Tyr-1086 on EGFR (see Introduction). Figure 2B shows that stimulation of HA-STAT3/FLAG-STAT3/HEK293 cells with EGF increased the levels of EGFR and P-STAT3 that coimmunoprecipitated with HA-STAT3, and that treatment of these cells with S3I-1757 inhibited these interactions. These results suggest that S3I-1757 is capable of inhibiting STAT3–STAT3 dimerization and STAT3 binding to EGFR in intact cells. Next, HA-STAT3/FLAG-STAT3/HEK293 cells were treated with vehicle, S3I-1757, or S3I-1756 and the cells were processed for immunofluorescence staining for FLAG-STAT3 (red) and HA-STAT3 (green) and analyzed by confocal microscopy as described in Materials and Methods. As shown in Fig. 2C, cells treated with vehicle harbored strong yellow color suggesting colocalization of FLAG-STAT3 (red) and HA-STAT3 (green). In contrast, cells treated with S3I-1757, but not S3I-1756, showed progressively less yellow color over time, suggesting that the dimerization of FLAG-STAT3 and HA-STAT3 was disrupted. Therefore, using methods that investigated STAT3 intracellular dimerization with tagged STAT3 proteins, we showed that STAT3–STAT3 protein–protein binding in intact cells was disrupted with a small molecule designed to disrupt phosphotyrosine binding to STAT3-SH2 domain.

For STAT3 to regulate the expression of its target genes, it needs to translocate from the cytosol to the nucleus, a process that requires STAT3 tyrosine phosphorylation and subsequent STAT3–STAT3 dimerization (1, 2). The fact that S3I-1757 was able to inhibit EGFR–STAT3 binding and STAT3–STAT3 protein–protein binding (Fig. 2) suggests that it would also inhibit the levels of tyrosine phosphorylated STAT3 and its nuclear accumulation. To determine if this is the case, MDA-MB-468 breast cancer cells, which harbor persistently Y-705 phosphorylated STAT3, were treated with either vehicle, S3I-1757, or S3I-1756 for 2 and 4 hours and then were subjected to immunofluorescence staining by a specific p-Tyr-705-STAT3 primary antibody and Alexa Fluor 594 secondary antibody in medium containing DAPI to stain the nuclei as described in Materials and Methods. Figure 3A shows that in vehicle-treated cells, P-STAT3 was localized predominantly in the nucleus. In contrast, the levels of p-STAT3 in the nucleus were dramatically decreased in S3I-1757–treated cells particularly after 4 hours of treatment. To determine whether this inhibition is sustained over a longer period of time, we treated MDA-MB-468 cells with various concentrations of S3I-1757 for 18 hours. Figure 3B shows that S3I-1757 inhibited P-STAT3 nuclear accumulation in a concentration-dependent manner starting at 50 μmol/L. In contrast, S3I-1756 had little effect with concentrations as high as 200 μmol/L.

The ability of S3I-1757 to inhibit STAT3–EGFR binding, tyrosine phosphorylation, STAT3–STAT3 dimer formation, and nuclear accumulation would be predicted to result in blocking STAT3–DNA–binding activity. To evaluate this possibility, we treated MDA-MB-468 and A-549 cells with vehicle, S3I-1757, or S3I-1756 for 1, 2, or 4 hours and collected nuclear extracts for STAT3–DNA–binding activity using a STAT3 filter plate assay as described in Materials and Methods. Figure 3C shows that nuclear extracts from vehicle-treated cells contained activated STAT3 capable of binding the biotin-labeled STAT3–DNA–binding probe. In contrast, the nuclear extracts from S3I-1757–treated cells contained less activated STAT3 capable of binding the STAT3–DNA–binding probe with the least amount found after 4 hours of treatment. Figure 3C also shows that S3I-1756 did not decrease the amount of STAT3 capable of binding DNA.

We next evaluated the ability of S3I-1757 to inhibit STAT3-dependent transcriptional activation using a luciferase reporter assay. To this end, MDA-MB-468 cells were transiently cotransfected with a STAT3-responsive promoter-firefly luciferase reporter (pLucTKS3) and β-gal reporter used to normalize the transfection efficiency. To determine the selectivity of S3I-1757 to suppress STAT3-dependent over STAT3-independent transcriptional activation, MDA-MB-468 cells were also cotransfected with SRE promoter-renilla luciferase reporter (pLucSRE) and β-gal reporter. Figure 4A shows that, compared with mock transfected cells, cells transfected with STAT3-responsive reporter (pLucTKS3) had increased luciferase activity in the absence of drug treatment. In contrast, less luciferase activity was observed when the cells were treated with S3I-1757 but not S3I-1756. S3I-1757 inhibited STAT3-dependent but not STAT3-independent transcriptional activity as shown by the minimal effect it had on SRE-driven luciferase activity (Fig. 4A). We next used a constitutively dimerized mutant form of STAT3, STAT3-C, to further show the STAT3-dependence of the inhibition with S3I-1757. STAT3-C spontaneously dimerizes via disulfide bonds in the absence of tyrosine phosphorylation (see Introduction), and is therefore not predicted to be inhibited by a small molecule that is designed to mimic phosphotyrosine binding. Figure 4A shows that transfection of MDA-MB-468 cells with STAT3-C increased the transcriptional activity of the STAT3- but not the SRE-responsive promoter and rescued from the S3I-1757 inhibition.

Figures 1–4A showed that S3I-1757 inhibits STAT3 dimerization, accumulation of nuclear P-Y705-STAT3, STAT3–DNA—binding, and transcriptional activity. We next determined if the ability of STAT3 to regulate the expression of its target genes is affected by S3I-1757. First, we confirmed that S3I-1757 inhibits the phosphorylation of STAT3-Y705 by Western blotting and determined whether this is selective. To this end, MDA-MB-468 cells were treated with vehicle, S3I-1756, or increasing concentration of S3I-1757 and processed for Western blotting as described in Materials and Methods. Figure 4B shows that S3I-1757, but not S3I-1756, inhibited the phosphorylation of STAT3-Y705 in a concentration-dependent manner starting at 50 μmol/L. This inhibition was selective for STAT3 over Akt and Erk1/2 phosphorylation. Figure 4B also shows that S3I-1757, but not S3I-1756, inhibited the expression of STAT3 target genes, such as the antiapoptotic proteins Bcl-xL and survivin, the cell-cycle protein cyclin D1, and the prometastatic protein MMP-9.

The data presented so far in Figs. 1–4 showed that S3I-1757 effectively blocks STAT3 dimerization and constitutive activation and suppresses its ability to persistently upregulate the expression of genes known to mediate hallmarks of malignant transformation. We, therefore, investigated next whether S3I-1757 can suppress uncontrolled anchorage-dependent and -independent tumor cell growth, migration, and invasion. To this end, we used several cell lines, some with persistently Y705 phosphorylated STAT3 and others without (Fig. 5A). Figure 5A shows that S3I-1757, but not S3I-1756, inhibited in a dose-dependent manner anchorage-dependent proliferation by MTT only in cells that harbor (MDA-MB-468, MDA-MB-231, H358, and A549) but not in those that do not harbor (H460, MDA-MB-453, HEK293, and MCF10A) persistently activated STAT3. Similar results were obtained with colony formation on plastic (Fig. 5B). The effects of S3I-1757 on cancer cell anchorage-independent growth was next determined by soft-agar assay as described in Materials and Methods. The results show that S3I-1757 significantly inhibited the anchorage-independent growth of cancer cells with constitutively active STAT3, such as MDA-MB-468, but had little effect on anchorage-independent growth of H460 cells, which has low activated STAT3 level (Fig. 6).

The ability of S3I-1757 to inhibit selectively the migration of cancer cells that depend on STAT3 over those that do not was next evaluated by a wound healing assay. The cancer cells were cultured with serum-starved medium for 8 hours before scratching the cells with a pipette tip and treating with increasing concentrations of S3I-1757 for 16 hours. Figure 7A shows that in the absence of drug, the cells migrated within 16 hours to fill the scratched area. S3I-1757, but not S3I-1756, treatment prevented this migration in cells with persistently activated STAT3 (MDA-MB-231 and A-549). In contrast, the migration of H460 (with low levels of P-STAT3) was minimally affected by the same treatment condition. Finally, the ability of S3I-1757 to inhibit selectively invasion was determined as described in Materials and Methods. Figure 7B shows that S3I-1757, but not S3I-1756, inhibited invasion in MDA-MB-468, MDA-MB-231, and A-549 but not in H-460 cells.

The fact that S3I-1757 but not its inactive analog S3I-1756 inhibits malignant transformation selectively in cells that harbor hyperactivated STAT3 suggested that S3I-1757 mediates its effects by inhibiting STAT3. To give further support to this suggestion, we determined whether the effects of S3I-1757 can be rescued by STAT3-C, a genetically engineered mutant of STAT3 that forms a constitutively dimerized STAT3 through disulfide bonds in the absence of tyrosine phosphorylation (8). To this end, we transfected MDA-MB-468 cells with STAT3-C, then treated the cells with S3I-1757, and processed the cells for Western blotting, cell growth, migration, and invasion. Supplementary Fig. S1A shows that expression of STAT3-C increased the levels of BclxL, cyclin D1, and survivin. In contrast, S3I-1757 inhibited the expression levels of these proteins and induced activation of caspase-3 and PARP cleavage. Furthermore, Supplementary Fig. S1A also shows that ectopic expression of STAT3-C inhibited the ability of S3I-1757 to downregulate the expression of BclxL, cyclin D1, and survivin and to induce apoptosis. Similarly, Supplementary Fig. S1B shows that STAT3-C inhibited S3I-1757 from inhibiting the proliferation/survival of MDA-MB-468 cells as determined by MTT assays. We next investigated the effects of STAT3-C on S3I-1757 effects on migration and invasion. Supplementary Fig. S2A and S2B show that S3I-1757 inhibited the migration and invasion of MDA-MB-468 cells transfected with vector DNA. In contrast, the ability of S3I-1757 to inhibit migration and invasion was partially rescued in cells transfected with STAT3-C.

The first demonstration that STAT3 is involved in malignant transformation (12) was reported in 1995 only a year after its discovery (22). Less than 6 years later, peptides and peptide mimics of the phospho-tyrosine peptide PpYLKTK that bind STAT3 SH2 domain were shown to inhibit STAT3 dimerization in vitro and STAT3 activity in intact cells (23). Yet, 11 years later, there are no small molecule STAT3 dimerization inhibitors in clinical trials. One of the major reasons for this is that STAT3–STAT3 dimerization is a protein–protein interaction that involves a large surface area, which is difficult to target with drug-like small molecules. The second reason, which is even more challenging, is that the negatively charged phospho-tyrosine, which is required for binding to the SH2 domain is difficult to mimic with moieties that can be easily taken up by cells. Nevertheless, because of the critical role of STAT3 in oncogenesis, several groups have put major efforts toward developing STAT3 dimerization inhibitors based on phospho-peptide mimics as novel anticancer drugs (6, 7). For example, Mandal, et al. (24) have succeeded at obtaining cell permeable peptidomimetics of pYLPQ in which pY was replaced by phosphocinnamide derivatives to improve peptidase resistance and used the pivaloyloxymethyl prodrug strategy to improve cellular uptake, which lead to potent inhibition of STAT3 activity in whole cells. Similarly, Chen, et al. (25) have also succeeded at designing a conformationally constrained pYLPQTV peptidomimetic with a long hydrocarbon chain to improve cell permeability. Although these are outstanding achievements, there still remain physicochemical challenges about the use of phospho-tyrosine peptidomimetics in vivo (7). Therefore, nonpeptidic small molecules capable of disrupting STAT3–STAT3 dimerization is an attractive alternative approach to inhibiting STAT3 directly. We have used structure-based virtual screening and identified S3I-201, a salicylic acid sulfonamide-based compound, which inhibits STAT3 dimerization in vitro and STAT3 activity in whole cells (16). Zhang, et al. (26, 27) have subsequently reported on salicylic acid sulfonamide S3I-201 analogs with improved potency. Our recent chemistry efforts resulted in a series of novel nonsulfonamide-containing salicylic acid–based compounds (19). In this article, we have shown that one of these, S3I-1757, inhibited STAT3 dimerization in vitro and in whole cells, STAT3 tyrosine phosphorylation, nuclear accumulation, transcriptional activity, and expression of STAT3-regulated genes as well as anchorage-dependent and -independent growth, migration, and invasion.

The ability of S3I-1757 to displace fluorescein-GpYLPQTV in the fluorescence polarization assay in vitro suggested that S3I-1757 binds the SH2 domain of STAT3 at the phospho-tyrosine-705–binding site. Our molecular modeling studies give further support to this suggestion. Indeed, our molecular modeling indicated that S3I-1757 makes several contacts with Arg-609 and Lys-591, 2 critical amino acids in the SH2 domain that are known to bind phospho-Tyr-705 of PpYLKTK of STAT3 as well as phospho-Tyr-904 of GpYLPQTV of the gp-130 subunit of the IL-6 receptor. In whole cells, S3I-1757 disrupted the binding of HA-STAT3 to FLAG-STAT3 as shown by both coimmunoprecipitation and colocalization, consistent with the in vitro fluorescence polarization and molecular modeling results. Taken together, these results strongly suggest that S3I-1757 is a STAT3–STAT3 dimerization inhibitor. Furthermore, STAT3 is also known to associate with the EGFR through binding of the STAT3-SH2 domain to phosphotyrosines 1068 and 1086 on EGFR (3), and S3I-1757 inhibited the binding of STAT3 to EGFR. It is, therefore, not surprising that S3I-1757 inhibited STAT3 tyrosine phosphorylation. This suggests that the ability of S3I-1757 to inhibit nuclear translocation, DNA binding, and transcriptional activation may be due to its ability to directly disrupt STAT3–STAT3 dimerization as well as inhibition of STAT3–EGFR binding and subsequent suppression of STAT3 tyrosine phosphorylation, which would also lead to preventing STAT3 dimerization. The fact that STAT3-C, a genetically engineered mutant of STAT3 that forms a constitutively dimerized STAT3 through disulfide bonds in the absence of tyrosine phosphorylation (8) was able to rescue from S3I-1757 inhibition of transcriptional activity further solidifies the suggestion that S3I-1757 is a STAT3 dimerization inhibitor.

S3I-1757 inhibited anchorage-dependent proliferation/survival and colony formation as well as anchorage-independent soft agar growth, migration, and invasion, consistent with its ability to suppress the expression of genes that are known to drive these hallmarks of cancer, such as cyclin D1, BclxL, survivin, and MMP-9. The fact that S3I-1757 did not inhibit other signal transduction pathways such as those leading to hyperactivated P-Akt and P-Erk suggest that S3I-1757 induces these effects through inhibition of STAT3. Further support for this suggestion comes from the fact that S3I-1757 inhibited anchorage-dependent and -independent tumor cell growth, migration, and invasion selectively in human cancer cells that dependent on STAT3 over those that do not. S3I-1756, a closely related structural analog of S3I-1757 that does not inhibit STAT3–STAT3 dimerization, STAT3 tyrosine phosphorylation, DNA binding, and transcriptional activation, was not able to inhibit anchorage-dependent and -independent tumor cell growth, migration, and invasion. Finally, the fact that STAT3-C, was able to rescue from S3I-1757 induction of apoptosis and inhibition of gene expression, tumor cell growth, migration, and invasion strongly supports the suggestion that S3I-1757 mediates its effects through inhibition of STAT3.

In summary, we have identified a nonsulfonamide-containing salicylic analog of S3I-201, S3I-1757, which directly inhibits STAT3–STAT3 dimerization as well as STAT3–EGFR binding leading to inhibition of STAT3 tyrosine phosphorylation and nuclear translocation. Using several approaches, we have also shown that S3I-1757 is able to inhibit malignant transformation and that this is due to its ability to inhibit STAT3 function.

No potential conflicts of interest were disclosed.

Conception and design: X. Zhang, M.K. Urlam, H.R. Lawrence, N.J. Lawrence, S.M. Sebti

Development of methodology: X. Zhang, R. Pireddu, H. Yang, M.K. Urlam, H.R. Lawrence, N.J. Lawrence

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Sun, R. Pireddu

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): X. Zhang, R. Pireddu, M.K. Urlam, W.C. Guida, N.J. Lawrence, S.M. Sebti

Writing, review, and/or revision of the manuscript: X. Zhang, M.K. Urlam, W.C. Guida, N.J. Lawrence, S.M. Sebti

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.K. Urlam, S.M. Sebti

Study supervision: H.R. Lawrence, N.J. Lawrence, S.M. Sebti

The authors thank the Moffitt Chemical Biology Core and the Analytic Microscopy Core for their outstanding technical expertise. The authors also thank Dr. Bromberg for providing us with the STAT3-C DNA construct.

This work was partially supported by NCI grant CA140681 (to S.M. Sebti and N.J. Lawrence).

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.