Hydroxamic acid and benzoic acid-based Stat3 inhibitors suppress human glioma and breast cancer phenotypes in vitro and in vivo

Abstract

STAT3 offers an attractive target for cancer therapy but small molecule inhibitors with appealing pharmacologic properties have been elusive. Here we report hydroxamic acid-based and benzoic acid-based inhibitors (SH5-07 and SH4-54, respectively) with robust bioactivity. Both inhibitors blocked STAT3 DNA binding activity in vitro and in human glioma, breast, and prostate cancer cells and in v-Src-transformed murine fibroblasts. STAT3-dependent gene transcription was blocked along with Bcl-2, Bcl-xL, Mcl-1, Cyclin D1, c-Myc and Survivin expression. Nuclear magnetic resonance analysis of STAT3-inhibitor complexes defined interactions with the SH2 and DNA binding domains of STAT3. Ectopic expression of the SH2 domain in cells was sufficient to counter the STAT3 inhibitory effects of SH4-54. Neither compound appreciably affected STAT1 or STAT5 DNA binding activities, STAT3-independent gene transcription or activation of a panel of oncogenic kinases in malignant cells. Each compound decreased the proliferation and viability of glioma, breast and prostate cancer cells and v-Src-transformed murine fibroblasts harboring constitutively active STAT3. Further, in mouse xenograft models of glioma and breast cancer, administration of SH5-07 or SH4-54 effectively inhibited tumor growth. Our results offer preclinical proof of concept for SH5-07 and SH4-54 as candidates fof further development as cancer therapeutics.

Introduction

The signal transducer and activator of transcription (STAT) proteins mediate cytokine and growth factor responses, including promoting cell growth and differentiation, and immune responses [1,2]. Ligand binding to the receptor promotes STAT activation by inducing the critical tyrosine (Tyr) phosphorylation by growth factor receptor Tyr kinases, JAKs or Src. Phosphorylation in turn drives STAT:STAT dimerization through a reciprocal phospho-Tyr-Src Homology (SH)2 domain interaction. STAT:STAT dimers translocate to the nucleus and bind specific DNA-response elements in target gene promoters to mediate transcription, thereby regulating fundamental cellular processes.

Aberrant-activation of Stat3, however occurs in malignant transformation and is implicated in glioma, breast, prostate, ovarian, and many other cancers [3-5]. Mechanisms to induce tumorigenesis and tumor progression include dysregulation of gene expression that leads to uncontrolled growth and survival of tumor cells, enhanced tumor angiogenesis, and tumor metastasis [1,3,6-8]. Stat3 activity further represses tumor immune surveillance [5,9]. Moreover, Stat3 cross-talks with nuclear factor (NF)-κB [10] and regulates mitochondrial functions to drive malignant transformation in specific contexts [11]. The Stat3 signaling pathway is considered an attractive target for the discovery of novel anticancer drugs.

On-going discovery campaigns for novel therapeutic modulators of Stat3 signaling [5,12-19] have largely focused on targeting the critical dimerization step to develop Stat3 inhibitors [5,12,13,15,19-23]. The Stat3 dimerization inhibitor, BP-1-102 emerged as an important lead compound that induced antitumor cell effects in vitro at 10-20 μM and antitumor effects in pre-clinical models of breast and non-small cell lung cancers [15]. Towards further improving the potency of the salicylic acid, BP-1-102 [15], we have synthesized and evaluated the hydroxamic acid, SH5-07 and benzoic acid, SH4-54, analogs, which show improved in vitro inhibitory activities at 1-8 μM. Structural data suggests these agents interact with the Stat3 SH2 and DNA-binding domains. Further, both agents inhibit in vivo growth of human glioma and breast cancer xenografts that harbor aberrantly-active Stat3.

Materials and Methods

Chemical synthesis of SH4-54 and SH5-07

Synthesis and detailed characterization of agents are described in Supplementary Materials, “Methods”.

Cells and reagents

Normal mouse fibroblasts (NIH3T3), counterparts transformed by v-Src (NIH3T3/v-Src) or overexpressing the human epidermal growth factor (EGF) receptor (NIH3T3/hEGFR), and the human breast (MDA-MB-231 and MCF-7), pancreatic (Panc-1) and prostate (DU145) cancer cells have all been reported [15,21,24,25] [26,27]. Stat3 null mouse embryonic fibroblast line (MEF/ST3KO) and ovarian cancer cells (A2780S) were kind gifts of Drs. Valeria Poli, University of Turin, Italy and Jin Cheng, Moffitt Cancer Center, Tampa, FL, respectively. The human glioma lines, U251MG, U373MG and U87MG (Sigma-Aldrich Corporation, St. Louis, MO), and SF-295 (Division of Cancer Treatment and Diagnosis Tumor Repository of the National Cancer Institute, Frederick, MD) were obtained from the designated sources and cultured in Roswell Park Memorial Institute medium-1640 supplemented with 1% nonessential amino acids (Corning Inc., Corning, NY) and containing 10% heat-inactivated fetal bovine serum (FBS). All other cells were grown in Dulbecco's modified Eagle's medium plus 10% heat-inactivated FBS. Except where designated, all antibodies were purchased from Cell Signaling Technologies (Danvers, MA).

Plasmids and molecular cloning

The Stat3-dependent luciferase reporter, pLucTKS3, and the Stat3-independent reporter, pLucSRE, have been previously reported [28,29]. The pLucTKS3 reporter contains seven copies of the Stat3-specific binding sequence in the C-reactive protein gene promoter driving firefly luciferase expression, while the Stat3-independent, pLucSRE reporter is driven by the serum response element (SRE) of the c-fos promoter. More details of the reporters and the Stat3 SH2 and DNA-binding domain constructs are provided in Supplementary Materials, “Methods”.

Transient transfection of expression vectors and luciferase reporter plasmids and reporter assay

Transient transfection using Lipofectamine 3000 (Life Technologies, Grand Island, NY) and luciferase assays were performed as previously reported [28,29]. Details are provided in Supplementary Materials, “Methods”.

siRNA transfection using Dharmacon SMARTpool

The ON-TARGETplus human Stat3 siRNA SMARTpool (L-003544), and the control (ON-TARGETplus Non-targeting Pool, D-001810-10-20) were both purchased from GE Dharmacon Inc (Lafayette, CO). Cells were transiently transfected with siRNA (25 nM) using Lipofactamine 3000 (Life Technologies) according to the manufacturer’s instructions. Forty-eight hours after transfection, Stat3 and its downstream genes were assayed in a pool of cells by Western blotting, and another pool of transfected cells was cultured in 96-well plates for additional 72 h and subjected to CyUANT cell proliferation assay (Life Technologies).

Nuclear extract preparation and gel shift assays

Nuclear extract preparation and DNA-binding/electrophoretic mobility shift assay (EMSA) were performed as previously described [24,29]. Details are provided in Supplementary Materials, “Methods”.

Surface plasmon resonance analysis

Studies were performed as previously reported [14,15]. Purified Stat3 (50 μg/ml) was injected onto the HisCap Sensor Chip for immobilization. Various concentrations of agents in running buffer (1X PBS, 0.5% DMSO) were passed over the chip to produce response signals. The association and dissociation rate constants were calculated using the Qdat software. The ratio of the association and dissociation rate constants was determined as the binding affinity (K_D).

Nuclear magnetic resonance (NMR) studies of Stat3-compound interactions

NMR studies were performed using the human Stat3β protein encompassing residues 127-711 in solution with agents and are described in detail in Supplementary Materials, “Methods”.

Immunoprecipitation and SDS-PAGE/Western blotting analysis

These studies were performed as previously described [29,30].

Chromatin-Immunoprecipitation (ChIP) and quantitative polymerase chain reaction (qPCR) studies

ChIP assay was performed as previously reported [31], with minor modification. Briefly, 1 × 10⁷ cells in culture were treated with 5 or 8 μM SH4-54 for 3 h and fixed with 1% formaldehyde for 7 min at room temperature. Cells were then treated with glycine (0.125 M, 5 min) at room temperature for cross-linking, washed with ice-cold PBS and lysed with ice-cold lysis buffer (10 mM Tris-HCl, pH 7.5, 10 mM NaCl, 3 mM MgCl₂, 0.5% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride) and centrifuged. Nuclear pellet was resuspended in buffer (50 mM Tris-HCl, pH 8.0, 10 mM EDTA, 1% SDS and protease inhibitors) (Roche, Indianapolis, IN) for lysis. To shear the DNA, the nuclear lysates were sonicated (Omni International, Kennesaw, GA) at 30% power for 3 pulses for 10 s intervals on ice. Samples were pre-cleared with protein A/G agarose beads (Santa Cruz) for 1 h at 4 °C, with rocking and incubated with anti-Stat3 (C20X) or anti-Stat5 (C-17) antibodies or with normal rabbit IgG overnight at 4 °C overnight, with rocking for immunoprecipitation. Immunecomplexes were collected with 20 μl protein A/G agarose bead, washed multiple times with wash buffer A (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl, pH 8.0) and two times with wash buffer B (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 500 mM NaCl, 20 mM Tris-HCl, pH 8.0) and eluted with freshly prepared elution buffer (1% SDS, 100 mM NaHCO₃). Cross-links were reversed by heating at 65 °C in the presence of NaCl followed by proteinase K treatment (20 μl, 20 mg/ml) (Pierce) overnight. The DNA was recovered and purified using ChIP spin columns (Zymo Research Corp, Irvine, CA). The purified chromatin immunoprecipitated DNA was used as a template for PCR amplification of the promoters for inducible nitric oxide synthase (iNOS), Survivin, Bcl-2 and β-casein genes. The PCR products were resolved on 2% agarose gel. ChIP results were analyzed by triplicate qPCR of the IP samples and corresponding 1% input samples using primers flanking the Stat3 binding region of each indicated gene promoter. Quantification was performed using the ΔCt method, appropriate due to the >90% PCR efficiency of each primer set. The percent input enrichment was determined by normalizing the Ct value for each sample to its corresponding 1% input Ct value. These absolute values were plotted as in the figures. Details and the PCR oligonucleotide primers are described in Supplementary Materials, “Methods”.

Cell viability assays

CyQuant cell proliferation assay to evaluate compounds was performed, as previously reported [14,15] and following the manufacturer’s (Invitrogen Corp/Life Technologies Corp, Carlsbad, CA) instructions.

Soft-agar colony formation and clonogenic survival assays

These studies were performed as previously reported [15,28,29]. Details are provided in Supplementary Materials, “Methods”.

Cell cycle profile and Annexin V binding with Flow cytometry analyses

Cells were treated with 0-8 μM agent for 24-48 h. For cell cycle profile analysis, cells were harvested and fixed with 70% ice-cold ethanol and stained with propidium iodide (PI). For apoptosis analysis, cells were collected and stained with FITC-Annexin V using Apoptosis Detection Kit (BD Biosciences, San Jose, CA). Both the DNA content of cells and the Annexin V-positive cells were analyzed by FACScan flow cytometer (BD Biosciences). Cell cycle phase distribution was analyzed using the Cell-Fit program. Data acquisition was gated to exclude cell doublets.

Wound healing assay for migration

Studies were performed as previously reported [15]. Details are provided in Supplementary Materials, “Methods”.

Mice and in vivo tumor studies

All animal experiments were performed under a protocol approved by the Institutional Animal Care and Use Committee. Four-to-five week-old female athymic nude mice were purchased from Jackson Laboratory and maintained in the institutional animal facilities approved by the American Association for Accreditation of Laboratory Animal Care. Mice were injected subcutaneously in the left flank area with U251MG cells (1 × 10⁷) in 200 μL of PBS/Matrigel matrix (1:1, BD Biosciences), or MDA-MB-231 cells (5 × 10⁶) in 100 μL of PBS. Mice with tumors of 90-150 mm³ (MDA-MB-231) or 150 mm³ (U251MG) were grouped for identical mean tumor sizes, administered 3, 5 or 6 mg/kg SH5-07 or SH4-54 via oral gavage daily or tail vein injection every 2 or 3 days, and monitored every 3-7 days. Tumor sizes were measured with calipers and converted to tumor volume, V, as follows: V=0.52 × a² × b, where a, smallest superficial diameter, and b, largest superficial diameter. For each treatment group, the tumor volumes for each set of measurements were statistically analyzed relative to the control (1% DMSO-treated) group.

Statistical analysis

Statistical analysis was performed on mean values using Prism GraphPad Software, Inc. (La Jolla, CA). The significance of differences between groups was determined by the paired t-test at p <0.05*, <0.01**, and < 0.001***.

Results

Compounds preferentially inhibit Stat3:Stat3 DNA-binding activity

SH4-54 and SH5-07 (Fig. 1A) are benzoic and hydroxamic acid analogs, respectively, of BP-1-102 [15]. Pre-incubation of NIH3T3/v-Src nuclear extracts of equal total protein containing constitutively-active Stat3 with 0-10 μM SH5-07 or SH4-54 for 30 min at room temperature, prior to incubation with the radiolabeled high-affinity sis-inducible element (hSIE) probe that binds Stat3 and Stat1 and subjecting to EMSA analysis [12,13,15] dose-dependently inhibited Stat3 activity (Fig. 1B), with IC₅₀, 3.9 ± 0.6 μM and 4.7 ± 0.5 μM, respectively. These potencies are improved over that of BP-1-102, of IC₅₀, 6.8 μM [15] (Fig. S1A). Similar studies using EGF-stimulated NIH3T3/hEGFR nuclear extracts containing active Stat1, Stat3 and Stat5 show agents preferentially inhibited Stat3:Stat3 DNA-binding activity, ahead of inhibiting Stat1:Stat3 activity, with minimal effects on Stat1:Stat1 activity (Fig. 1C). Parallel EMSA analysis using nuclear extracts from EGF-stimulated NIH3T3/hEGFR cells and the radiolabeled mammary gland factor element (MGFe) probe that binds Stat1 and Stat5 showed no inhibition of Stat1:Stat1 or Stat5:Stat5 activity (Fig. 1D). Supershift study with anti-Stat3 antibody (α-ST3) identifies Stat3-DNA complex (Fig. S1B).

Figure 1. SH5-07 and SH4-54 and effects against STAT DNA-binding activities in vitro.

(A) SH5-07 and SH4-54; and (B-D), Nuclear extract preparations from (B) NIH3T3/v-Src containing activated Stat3, or (C, D) EGF-stimulated NIH3T3/hEGFR containing activated Stat1, Stat3 and Stat5 were pre-incubated 0-20 μM SH5-07 or SH4-54 for 30 min at room temperature prior to incubation with the radiolabeled (B, C) hSIE probe that binds Stat3 and Stat1 or (D) MGFe probe that binds Stat1 and Stat5 and subjecting to EMSA. Positions of STATs:DNA complexes are labeled; control lanes (0) represent extracts treated with 0.05% DMSO. Bands corresponding to STATs:DNA complexes were quantified using ImageQuant, calculated as percent of control (cont), and are shown or plotted against concentration to derive IC₅₀ values. Data are representative of 3-4 independent determinations.

Inhibition of intracellular Stat3 activation

EMSA analysis [14,15] of nuclear extracts from human glioma (U251MG, U373MG, U87MG and SF295), breast (MDA-MB-231), and prostate cancer (DU145) cells, or NIH3T3/v-Src cells [4,5,8] and treated with 0-10 μM SH5-07 or SH4-54 for 1 h (Fig. 2A, B) shows inhibition of constitutively-active Stat3 in the U251MG and U87MG cells at 1 μM and higher, U373MG and SF295 lines at 1-3 μM and higher, and in MDA-MB-231 and DU145 cells at 3-5 μM and higher (Fig. 2A, B). These compare more favorably over the potencies BP-1-102 of 10-20 μM [15]. Immunoblots confirmed pY705Stat3 reduction by agents at 3-10 μM (Fig. 2C, D), compared to 10-20 μM activity for BP-1-102 [15]. Inhibition of Stat3 DNA-binding activity (Fig. 2B) or pY705Stat3 (Fig. 2D) occurs early (0.5-1 h) and is sustained for 3-6 h, followed by an apparent weakening at 6 h or later. By contrast, BP-1-102 inhibited pY705Stat3 at 15 μM and 24 h, with little effect at 1 h (Fig. S2A). Assuming the possibility of diminishing intracellular inhibitor levels, a time-course study for the effects of 5 μM SH5-07 with or without a second treatment at time 5 h after the first treatment for additional 1 h (6^#) and 4 h (9^#) showed a sustained pY705Stat3 inhibition for up to 9 h (Fig. 2E, lanes 5 and 6). Both inhibitors had little effect on pS727-Stat3 (Fig. S2B).

Figure 2. Agents inhibit intracellular Stat3 activation.

(A and B) EMSA analysis of Stat3 DNA-binding activity in nuclear extracts from the designated malignant cells that were treated with (A) 0-10 μM SH5-07 (i) or SH4-54 (ii) for 1 h, or (B) 5 or 8 μM SH5-07 (i) or SH4-54 (ii) for 0-24 h; (C-E) Stat3 and pY705-Stat3 immunoblotting analysis of malignant cells treated with (C) 0-10 μM SH5-07 (i) or SH4-54 (ii) for 1 h, (D) 5 or 8 μM SH5-07 (i) or SH4-54 (ii) for 0-24 h, or (E) 5 μM SH5-07 for 0-9 h, with a repeat dosing at 5 h for additional 1 h (6^#) or 4 h (9^#); and (F) NIH3T3 fibroblasts were transiently co-transfected with v-Src plasmid and the (i) Stat3-dependent (pLucTKS3) or (ii) Stat3-independent (pLucSRE) luciferase reporter and treated with 0-8 μM SH5-07 for 24 h. Luciferase reporter activity in cytosolic extracts is plotted as fold change over control. Positions of STATs:DNA complexes or proteins in gel are labeled; control lanes (−, 0) represent nuclear extracts, cytosolic extracts, or whole-cell lysates from 0.05% DMSO-treated cells. For each transfection, luciferase activity was normalized to transfection efficiency with β-galactosidase activity as an internal control. Data are representative of 3-4 independent determinations. Values, mean ± S.D., n=4, each performed in triplicate. * − <0.05.

The observation that intracellular constitutive Stat3 DNA-binding activity decreased (Fig. 2A) with no corresponding change in pY705Stat3 at 1-3 μM, except at higher, 5-10 μM (Fig. 2C, S2C), suggests that inhibition is strongest on Stat3 DNA-binding activity, compared to pY705Stat3 (Fig. 2A vs. C, and Fig. S2C). This apparent lack of correlation within the context of constitutively-active Stat3 suggests the disruption of the pre-existing Stat3:Stat3 dimers, which directly leads to lower DNA-binding activity (Fig. 1B) [12-15,32], has a non-linear relationship with the turnover of the disrupted pStat3 molecules. This likely reflects differences in kinetics and highlights the complexity of the events involved in the Stat3 activation and inactivation. We observe a similar non-agreement between the pStat3 and Stat3 DNA-binding activity changes caused by the over-expression of 1 μg exogenous Stat3 SH2 domain, which suppressed DNA-binding activity, with no corresponding pStat3 change (Fig. S2D, lane 4). The over-expression of higher, 2 or 4 μg Stat3 SH2 domain, however, caused concurrent reductions in both pStat3 and Stat3 DNA-binding activity (Fig. S2D, lanes 5 and 6), consistent with previous report that a peptide derived from the SH2 domain inhibited Stat3 signaling [33] and that the SH2 domain expressed alone is functional. Moreover, dimerization disruptors could inhibit DNA-binding activity and function, without necessarily inhibiting pY705Stat3. Nuclear extracts containing constitutively-active Stat3 and pre-incubated with 5 μM SH4-54, SH5-07 or BP-1-102 [15] had lower Stat3 activity in the cell-free EMSA analysis (Fig. S2E(i)), while the corresponding pY705Stat3 immunoblots were unchanged (Fig. S2E(ii)), presumably due to the absence of intracellular events to promote pY705Stat3 turnover. Therefore, despite being a direct measure of the effects of dimerization disruptors on Stat3 activation, DNA-binding activity inhibition may not necessarily correlate with the pY705Stat3 inhibition. Altogether, these data highlight complexities in the intracellular Stat3 signaling induction and turnover that impact the measured activities of dimerization disruptors or SH2 domain antagonists.

Luciferase reporter studies showed v-Src induces Stat3-dependent pLucTKS3 luciferase reporter activity [20,21,29] by 25 to 34-fold, which was suppressed by 3-8 μM SH5-07 (Fig. 2F(i), lanes 2 vs. 1, and lanes 3 and 4 vs. 2), while similar treatment did not inhibit v-Src-induced, Stat3-independent pLucSRE reporter activity [20,21,28,29] (Fig. 2F(ii)). Agents had no significant effects on pY1068EGFR, pY416Src, pJAK2, pShc, pErk1/2^MAPK, and pS473Akt levels (Fig.S3A-D) at concentrations that inhibit Stat3 activity (Fig. 2A-F). We note the agreement between the inhibitory potencies in both the cell-free Stat3 DNA-binding assay (IC₅₀, 3.9-4.7 μM; Fig. 1B) and the intracellular constitutively-active Stat3 studies (1-8 μM; Fig. 2A-F). These potencies are an improvement over the 10-20 μM activities for BP-1-102 [15] (Fig. S2A).

Agents bind Stat3, disrupt Stat3 association with growth factor receptor, and thereby inhibit Stat3 phosphorylation

Stat3 dimerization disruptors suppress pY705Stat3 [13-16,20,21,34-36]. STATs are recruited to the receptor phospho-Tyr motif for close proximity to Tyr kinases. We focused on Stat3 recruitment and assessed the effect of agents on EGFR:Stat3 association. In co-immunoprecipitation studies, EGF stimulation of NIH3T3/hEGFR fibroblasts induced 1.38-fold increase in Stat3:EGFR complexation (Fig. 3A, lanes 3 vs 2, Stat3), which was associated with pY705Stat3 induction (Fig. 3B, lanes 2 vs 1, pY705Stat3). Twenty-four hour prior treatment of fibroblasts with agents decreased EGF-stimulated Stat3:EGFR co-immunoprecipitation (Fig. 3A, lanes 4 and 5 vs 3, Stat3), in parallel with decreased EGF-induced pY705Stat3 (Fig. 3B, lanes 3 and 4 vs 2, pY705Stat3), without inhibiting pY1068EGFR or pERK1/2 induction (Fig. 3C). Moreover, prior SH5-07 treatment of fibroblasts had little effect on EGF-stimulated pY1068EGFR induction, or pY705Stat3, except at 24 h (Fig. S4). By contrast, 1-h prior treatment of MDA-MB-231 cells with agents inhibited pre-existing and/or interleukin-6 (IL-6)-stimulated pY705Stat3 induction (Fig. 3D, lanes 3 vs 2 and 1, and lanes 5 vs 6). We note the differences in the time-to-inhibition between the EGF-stimulated pY705Stat3 in mouse fibroblasts, IL-6-stimulated pY705Stat3 in human breast cancer cells, and pre-existing pY705Stat3 in tumor cells, which altogether indicate signal-type (constitutive, EGF- or IL-6-stimulated) and cell-type (normal mouse fibroblasts versus human tumor cells) contexts of pY705Stat3 suppression. Therefore, agents inhibit ligand-stimulated de novo Stat3 induction. Furthermore, disruption of Stat3:receptor interaction represents one of the pY705Stat3 inhibition mechanisms.

Figure 3. Compounds inhibit Stat3 binding to EGF receptor and ligand-induced pY705Stat3.

Immunoblotting analysis of (A) EGFR immunecomplexes or (B, C) whole-cell lysates from NIH3T3/hEGFR fibroblasts pre-treated or not with 10 μM SH5-07 or SH4-54 for 24 h prior to EGF stimulation (100 ng/ml, 12 min), or (D) whole-cell lysates from MDA-MB-231 cells pre-treated for 1 h prior to IL-6 stimulation (20 ng/ml, 10 min) and probing for EGFR, pY1068EGFR, pY705-Stat3, Stat3, pERK1/2, ERK1/2, or β-actin. Positions of proteins in gel are labeled; control lane (−) represents whole-cell lysates or EGFR immunecomplexes prepared from 0.05% DMSO-treated cells. Bands corresponding to proteins were quantified using ImageQuant, calculated as percent of control (cont) and shown. Data are representative of 2-3 independent determinations.

Surface plasmon resonance (SPR) analysis of Stat3:agent interactions shows SH4-54 and SH5-07 bound Stat3, with affinities (K_D) of 2.4 μM (Fig. 4A and data not shown). We performed NMR analysis of Stat3:compound interactions. 1D ¹H NMR spectra of 200 μM compounds (from DMSO stock) in aqueous solution suggested 20 μM effective concentration (Fig. S5A, and data not shown), suggesting Stat3 could not be fully saturated by the compounds in NMR studies. Given Stat3 protein concentration of 20 μM and the 20 μM maximal compound detected, the sample reached approximately 1:1 stoichiometry. The % saturation, based on K_D of 2.4 μM, is approximately 70%. Solubility issues prevented BP-1-102 analysis in a similar study. NMR chemical shift perturbation (CSP) indicates that compounds interact with Stat3 in a specific manner. The overlay of the ¹³C-¹H HMQC spectra of Stat3, free and in complex with the representative compound, SH4-54, is shown (Fig. 4B, left). All 35 peaks from Ile residues of Stat3 have been assigned (Namanja et al, manuscript submitted), providing probes for every structured domain. In addition, site-directed mutations have been made to replace Leu residues in the DNA-binding domain (DB), one at a time, in order to identify their resonances. Binding of compounds caused selective changes in line-widths at Ile residues 597, 386 and 439 and CSP of Leu411 (Fig. 4B, left). The line broadening effects observed by the Ile signals indicate that the compounds bind to both the Stat3 SH2 and DB domains (Fig. 4B, left). Residues Leu411, Ile386 and Ile439 form a hydrophobic pocket that is likely involved in compound binding. The peak from I364, which is next to the other perturbed residues in the DB domain, also showed CSP. The ribbon representation of Stat3 structure in complex with DNA and with the locations of these residues is shown (Fig. 4B, right).

Figure 4. Stat3 interacts with compounds and its activity is rescued by SH2 domain over-expression.

(A) Surface plasmon resonance analysis of SH4-54 interaction with Stat3; (B) NMR analysis of Stat3 in solution with SH4-54. (i) Overlay of the ¹H-¹³C HMQC spectra of Stat3, free (black) and SH4-54-bound (red). Residues with significant changes in either resonance line-widths or NMR chemical shifts are indicated; and (ii) the ribbon representation of Stat3:DNA complex showing the locations of the affected amino acid residues; and (C) Stat3 DNA-binding assay/EMSA analysis of nuclear extracts from U251MG cells transiently over-expressing Stat3 SH2 or DNA-binding (DB) domain and treated with 8 μM SH4-54 for 3 h. Positions of Stat3:DNA complex are labeled; control lanes (−) represent nuclear extracts treated with 0.05% DMSO. Data are representative of 2-3 independent determinations

NMR data identify a novel putative binding to the DB domain. We probed the specificity of the Stat3:compound interaction and the significance of the DB domain. The most likely non-specific event that induces CSP is the alkylation of the Cys residues that do not form disulfide bonds in the DB domain, one of which is at the DB interface. NMR analyses of 20 μM Stat3 sample with and without 100 μM NEM (Fig. S5B) showed the CSP due to alkylation is different from that due to interaction with compounds. The CSP at residues I364 and L411 upon compound interaction did not occur upon alkylation by NEM, suggesting the Stat3:compound interaction is specific. Unfortunately, the poor solubility and moderately weaker activity of BP-1-102 prevented the collection of 2D ¹H-¹³C HMQC spectra for comparison. Moreover, EMSA analysis shows transiently-expressed exogenous Stat3 SH2 domain rescued Stat3 activity from compound effects, compared to the moderate rescue by the DB domain (Fig. 4C), indicating that the expressed SH2 domain is functional (Fig. S2D) and represents the primary target site for the compounds, while the DB domain interaction only moderately contributes to the overall compound effect against Stat3. We note that the expressed SH2 domain functions as a negative regulator of Stat3 signaling (Fig. S2D, lanes 4-6). However, the exogenously expressed Stat3 SH2 domain, when concurrently present, binds compound, consistent with the previous report that the SH2 domain and its peptide, SPI, both sufficiently interacted with the dimerization disruptor, S3I-201 and other Stat3 SH2 domain-binding peptides [33]. The easier accessibility to the exogenously expressed SH2 domain favors interaction with the compound more than the occluded SH2 domain in the Stat3:Stat3 dimer.

Compounds induce antitumor cell effects against malignant cells harboring constitutively-active Stat3

In agreement with Stat3’s role in tumor cell phenotype [3-5,14,15], 72-h treatment with the compounds of cultured tumor cells harboring variable Stat3 activities (Fig. S6) inhibited viability to different degrees (Fig. 5A), with IC₅₀ of 1.0-2.7 μM for glioma U251MG and U87MG cells (Fig. 5A, middle panels), which are most sensitive. Other IC₅₀ values are 3.8-4.5 μM for breast cancer, MDA-MB-231 (231) cells (Fig. 5A, left panels), 3.8-7.4 μM for glioma, U373MG and SF295 cells (Fig. 5A, middle panels), 5.3-5.8 μM for DU145 prostate cancer cells (Fig. 5A, left panels), 4.1-9.2 μM for NIH3T3/v-Src (vSrc) (Fig. 5A, left panels), and 9.6-10.3 μM for the least sensitive human pancreatic cancer, Panc-1 cells (Fig. 5A, left panels). The variable growth inhibition likely reflects the Stat3 activation levels that show different inhibitor sensitivities (Fig. 2A-D) across the tumor cells. The 1.0-7.4 μM range that inhibits cell viability overlaps with the 1-5 μM range that inhibits Stat3 DNA-binding activity in tumor cells. Comparatively, the viability of MCF-7, A2780S, normal NIH3T3 (3T3) and Stat3-null mouse embryonic fibroblasts (MEF/S3T3) that do not harbor aberrantly-active Stat3 (Fig. S6) is only moderately inhibited, with IC₅₀ values of 8.1-10.8 μM or higher (Fig. 5A, right panel). We note the general trend that glioma and breast cancer cells harboring aberrantly-active Stat3 are more sensitive.

Figure 5. Antitumor cell effects in vitro of compounds against malignant cells harboring aberrantly-active Stat3.

(A) Cultured human tumor cells that harbor varying degrees of constitutively-active Stat3 and cells that do not (NIH3T3 (3T3), Stat3 null mouse embryonic fibroblasts (MEF/ST3KO), MCF7 and A2780S) were treated once with 0-10 μM (i) SH5-07 or (ii) SH4-54 for 72 h and assayed for cell proliferation and plotted as % cell viability against concentration, from which IC₅₀ values (inserts) were derived; (B and C) MDA-MB-231 cells were transiently-transfected with (B) FLAG-tagged Stat3 DNA-binding (DBD) or SH2 domain expressing plasmid and untreated or treated with 3 or 5 μM SH4-54 for 72 h or (C) Stat3 siRNA or scrambled siRNA for 24 h and (B(i) and C) assayed for cell proliferation, or (B(ii)) or immunoprobed for FLAG or β-actin; (D) Single-cell cultures of designated cells treated once with 0-5 μM agents and cultured until large colonies were visible, which were stained with crystal violet and imaged; (E) Plots of numbers of large visible colonies of tumor cells growing in soft-agar and treated once with 0-8 μM SH5-07; (F and G) Flow cytometry analysis of cells in culture treated with 0-8 μM SH5-07 for 24-48 h, stained with (F) propidium iodide (PI) for DNA content and cell cycle distribution, which is plotted, or (G) PI and Annexin V for apoptosis, which is represented as percent Annexin V positive cells. Control (−, 0) represents samples or whole-cell lysates prepared from scrambled siRNA-transfected or 0.05% DMSO-treated cells. Values, mean ± S.D., n=3-4. Data are representative of 3 independent determinations.

To validate the Stat3-dependency, the transient over-expression of the SH2 domain (Fig. 5B(ii)) rescued MDA-MB-231 cells from the effects of SH4-54 (Fig. 5B(i), SH2), while the DB domain over-expression only moderately rescued cells (Fig. 5B(i), DBD), consistent with the findings in Figure 4C that the SH2 domain represents the major target site for the inhibitors, and with the previous report that the Stat3 SH2 domain rescued tumor cells from BP-1-102-induced apoptosis [15]. Furthermore, siRNA knockdown of Stat3 suppressed U251MG viability (Fig. 5C).

In colony survival assay, one-time treatment with compounds at 1-3 μM dose-dependently suppressed U251MG and MDA-MB-231 colony formation (Fig. 5D, left two panels), with higher activity against U251MG cells, while minimally affecting MEF/ST3KO (Fig. 5D, right panel). SH5-07 further dose-dependently suppressed the soft-agar growth of U251MG and MDA-MB-231 cells, with a higher activity against MDA-MB-231 cells (Fig. 5E and Fig. S7A).

Flow cytometry/cell cycle profile analysis shows sub-G0 phase population increased by 15-55% for MDA-MB-231 cells treated with 5-8 μM SH5-07 for 24-48 h, with decreased cell numbers in G1, S and G2/M phases (Fig. 5F(i)), and 20% for U251MG cells treated with 8 μM SH5-07 for 48 h, associated with reduced G0/G1 phase population (Fig. 5F (ii), 8 μM), while treatment had minimal effects on MEF/ST3KO (Fig. 5F (iii)). Annexin V binding/flow cytometry analysis showed 5.8 % and 18 % apoptosis, respectively, for U251MG and MDA-MB-231 cells treated with 8 μM SH5-07 for 24 h (Fig. 5G(i) and (ii)) and minimal effects on similarly-treated MEF/ST3KO cells (Fig. 5G (iii)). Therefore, cell cycle changes and apoptosis contribute to the growth inhibition of tumor cells that harbor persistently-active Stat3. Further, 8 μM SH5-07 or SH4-54 treatment for 22 h suppressed MDA-MB-231, U251MG and DU145 cells migration into the denuded area (Fig. S7B(i) and (ii), upper panels), without affecting cell viability (Fig. S7B(i) and (ii), bottom panels).

SH5-07 inhibits the expression of known Stat3-regulated genes

Immunoblotting analysis shows reduced Bcl-2, Bcl-xL, c-Myc, Survivin, Cyclin D1 and Mcl-1 expression in response to 24 h, 5 μM SH5-07 treatment (Fig. 6A). Chromatin immunoprecipitation analysis with qPCR confirmed decreased promoter occupancy by Stat3 of the iNOS, bcl-2 and survivin genes in cells treated for 3 h with 5 or 8 μM SH4-54, and not Stat5 occupancy of β-casein gene promoter (Fig. 6B). siRNA knockdown of Stat3 suppressed iNOS, Bcl-2 and Survivin expression (Fig. 6C). Studies altogether validate the inhibition of aberrantly-active Stat3, suppression of constitutive induction of Stat3-regulated genes [3-6,8];, and the suppression of tumor phenotype by SH5-07 and SH4-54.

Figure 6. Suppression of Bcl-2, Bcl-xL, Cyclin D1, c-Myc, Mcl-1 and Survivin expression and the Stat3 occupancy of inducible nitric oxide synthase, bcl-2 and survivin promoters.

(A and C) Immunoblots of (A) Bcl-2, Bcl-xL, c-Myc, Survivin, Cyclin D1, Mcl-1, β-actin or GAPDH the designated cells treated with 0 or 5 μM SH5-07 for 6-24 h, or (C) iNOS, Bcl-2 and Survivin from MDA-MB-231 cells transiently-transfected with Stat3 siRNA or scrambled siRNA; and (B) percent enrichment, as detected by qPCR, of the DNA fragments for the indicated gene promoters from Stat3 or Stat5 immunoprecipitated DNA complex relative to input. PCR amplifications were performed using specific primers for the promoter regions. Positions of proteins in gel are shown; control (−, 0) lane represents whole-cell lysates from scrambled siRNA-transfected or 0.05% DMSO-treated cells. Bands corresponding to proteins were quantified using ImageQuant, calculated as percent of control (cont) and shown. Values, mean ± S.D., n=3-4. Data are representative of 2-3 independent determinations. * − <0.05

SH5-07 and SH4-54 inhibit growth of human breast and glioma tumor xenografts

Tail vein injection (5-6 mg/kg every 2-3 days) or oral gavage delivery (3 mg/kg daily) of SH5-07 or SH4-54 inhibited growth of 90-150 mm³ established subcutaneous mouse xenografts of human glioma (U251MG) (Fig. 7A) and breast (MDA-MB-231) (Fig. 7B) tumors that harbor aberrantly-active Stat3, associated with decreased c-Myc, Mcl-1 and Cyclin D1 expression (Fig. 7C). No significant changes in body weights (Fig. S8A), blood cell counts (Fig. S8B), or the gross anatomy of organs (Fig. S8C), or obvious signs of toxicity, such as loss of appetite, decreased activity, or lethargy were observed.

Figure 7. In vivo antitumor efficacy response against subcutaneous human glioma and breast tumor xenografts in mice and effect on Stat3-regulated genes.

Growth curves for subcutaneous human (A) glioma, U251MG or (B) breast, MDA-MB-231 tumor xenografts and the effects of (A and B, left panel) intravenous administration (5 or 6 mg/kg, every 2 or 3 days), or (B, right panel) oral gavage delivery (3 mg/kg, every day) of agetns; and (C) Mcl-1, c-Myc or Cyclin D1 immunoblots in tumor tissue lysates from 2 different control and 3 different treated mice for each inhibitor. Positions of proteins in gel are labeled; control lanes represent tissue lysates from mice treated with 0.05% DMSO. Values, mean ± S.D., n=6. * − <0.05

Discussion

The prevalence of constitutively-active Stat3 in human tumors has placed an increasing importance on the discovery of novel Stat3-inhibiting anticancer drugs [5]. The hydroxamic acid, SH5-07 and the benzoic acid, SH4-54, analogs of BP-1-102 [15] have emerged as potent and selective inhibitors, with single-digit micromolar activities against Stat3 signaling. Studies show that constitutively-active Stat3 in tumors responds variably across tumor lines to small-molecule inhibitors. The differential sensitivities of aberrantly-active Stat3 signaling suggest complexities in the Stat3 induction and decay mechanisms in tumor cells. Compared to other notable agents, including, BP-1-102 [15] and S3I-1757 [19], both SH5-07 and SH4-54 possess stronger activities.

Present studies, together with published reports [13-15,17,19,23,37], demonstrate the biological responsiveness to small-molecule Stat3 inhibitors of tumor models harboring persistently-active Stat3, including glioma, breast, prostate, and lung cancer models. Interestingly, among the Stat3-relevant tumor models, the human glioma tumor cells, U251MG and U87MG appear to be more sensitive than the glioma lines, U373MG and SF295 or the breast cancer line, MDA-MB-231, prostate cancer cells, DU145, or the v-Src-transformed mouse fibroblasts, while Panc-1 pancreatic cancer cells are least responsive. Given the importance of constitutively-active Stat3 in promoting aggressive glioma phenotype and its clinical relevance in predicting a poor clinical outcome, current studies together with others [37-40] raise the potential that small-molecule Stat3-inhibiting therapeutics, such as SH5-07 and SH4-54, could be particularly useful in anti-glioma therapy.

Aberrantly-active Stat3 dysregulates expression of critical genes that control tumor cell growth and survival, tumor angiogenesis, and tumor cell migration, invasion and tumor metastasis [15,20,21,23,25,26,41-44]. The inhibition of Stat3 signaling by SH5-07 and SH4-54 is associated with decreased Bcl-2, Bcl-xL, Cyclin D1, c-Myc, Mcl-1, and Survivin expression, and antitumor responses in human glioma and breast cancer models. Against this compelling evidence, however, studies by others showed that the phosphatase-stable, cell-permeable SH2 domain-targeting prodrug inhibitor of Stat3, PM-73G, did not inhibit cyclin D1, Bcl-2 or survivin expression or induce apoptosis [35,45] at the concentrations that inhibited pY705Stat3 in tumor cells, except at 50-fold higher concentration [35]. The failure in those studies to observe antitumor cell response to agents at concentrations that inhibit pY705Stat3 contrasts with the current study and others, and likely suggests complex mechanisms and functions of constitutive Stat3 activation in malignant progression that may differentially impact the cellular responses to Stat3 inhibitors. It further indicates there is more to learn about the relationship between the inhibition of Stat3 signaling by small-molecules and the overall biological outcome. By comparison, the suppression of the known Stat3-regulated genes and antitumor cell responses to SH5-07 or SH4-54 occurred at 1-8 μM concentrations that inhibit constitutively-active Stat3, measured by pY705Stat3, Stat3 DNA-binding activity, and Stat3-dependent luciferase reporter transcription. The inhibitory responses are validated by the SH2 domain over-expression that countered the effects of the small-molecules and by the siRNA knockdown of Stat3 that produced similar results.

Altogether, the present study identifies SH5-07 and SH4-54 as potent Stat3 inhibitors, which induce antitumor cell effects in vitro and antitumor response in vivo against human glioma and breast cancer models. The efficacy studies against subcutaneous xenografts are intended to provide proof-of-concept for the potential efficacy against glioma and breast cancer. SH5-07 and SH4-54 are therefore potential candidates for further development for clinical application, particularly for human glioma and breast cancers.

Abbreviations

STAT

signal transducer and activator of transcription

MEF/ST3KO

Stat3 knockout mouse embryonic fibroblasts

EMSA

electrophoretic mobility shift assay

MAPK

mitogen-activated protein kinase

ERK

extracellular signal-regulated kinase

GAPDH

Glyceraldehyde-3-phosphate dehydrogenase

qPCR

quantitative polymerase chain reaction

ChIP

chromatin immunoprecipitation