EC359: A First-in-Class Small-Molecule Inhibitor for Targeting Oncogenic LIFR Signaling in Triple-Negative Breast Cancer

This article is featured in Highlights of This Issue, p. 1335

Leukemia inhibitory factor (LIF) is the most pleiotropic member of the IL6 family of cytokines (1). LIF signaling is mediated via the LIF receptor (LIFR) complex, which is comprised of LIFR and glycoprotein 130 (gp130; ref. 2). The LIFR does not have intrinsic tyrosine kinase activity. Both LIFR and gp130 constitutively associate with the JAK–Tyk family of cytoplasmic tyrosine kinases. Consequently, LIF binding to the LIFR complex activates multiple signaling pathways, including JAK/STAT, MAPK, AKT, and mTOR (2–4). LIF and LIFR are widely expressed in many solid tumors (1, 5–7) and their overexpression is often associated with poor patient prognosis (8, 9). In addition, high circulating LIF levels correlate with tumor recurrence (10).

The LIF/LIFR axis acts on multiple aspects of cancer biology to promote tumor growth, metastasis, and therapy resistance (11). LIF is a key regulator of cancer stem cells (CSC; ref. 11), plays a role in stem cell maintenance (12, 13), regulates self-renewal and pluripotency (12), and is associated with chemoresistance (10, 14). LIF functions as a growth factor to promote growth and invasion (15). Recent evidence indicates upregulated LIF–JAK–STAT3 signaling via autocrine and paracrine mechanisms in tumors (10, 16, 17). However, lack of any small-molecule inhibitors (SMI) that block LIF/LIFR signaling represents a major knowledge gap and critical barrier for advancement of LIF/LIFR–targeted cancer therapy.

Among the different subtypes of breast cancer, 60%–70% are estrogen receptor (ER) positive (ER⁺ breast cancer), and 15%–24% are triple-negative breast cancer (TNBC; ref. 18). TNBC is more aggressive, and due to the lack of targeted therapies, represents a disproportional share of the breast cancer mortality (19, 20). TNBC exhibit high propensity for metastasis, with some subtypes such as claudin-low that are highly enriched for CSCs, and frequently exhibits therapy resistance (19, 20). In breast cancer cells, LIF/LIFR signaling activates multiple signaling pathways including STAT3, AKT, and mTOR pathways and contributes to activation of mTOR downstream targets such as p70S6K and 4EBP1 (4). LIF/LIFR signaling promote tumor progression of both ER⁺ breast cancer and TNBC cells (21–23). In addition, LIF mRNA levels were elevated in invasive breast carcinomas compared with the normal breast tissues (24). Overexpression of LIF is significantly associated with a poorer relapse-free survival in patients with breast cancer (4).

In this study, we report the development of a novel LIFR inhibitor EC359 that selectively binds LIFR and blocks binding of ligands attenuating LIFR oncogenic signaling. Using molecular modeling, in vitro, and in vivo assays, we demonstrated that EC359 interacts with LIFR and inhibits cell viability of TNBC cells that express both LIF and LIFR. In addition, EC359 reduced the invasion and stemness of TNBC cells, and promoted apoptosis. In xenograft and patient-derived xenograft (PDX) assays, EC359 significantly reduced the tumor progression. This study represents the first report detailing the development of a first-in-class inhibitor of LIF/LIFR.

Human breast cancer cells MCF7, T47D, MDA-MB-231, BT-549, SUM-159, HCC1937, MDA-MB-468, HCC1806, and normal mammary epithelial cells (HMEC) were obtained from ATCC, were maintained as per ATCC guidelines, and used from early passages (<10 passages after thawing). All model cells utilized were free of Mycoplasma contamination and were confirmed by using the Mycoplasma PCR Detection Kit purchased from Sigma. Short tandem repeat polymorphism analysis of the cells was used to confirm the identity at University of Texas Health San Antonio (UTHSA) core facilities. CSCs isolated from TNBC cells were maintained in MammoCult medium along with the supplements according to the manufacturer's instructions (StemCell Technologies). The GAPDH, p-ERK1/2, ERK1/2, p-p70S6K, p70S6K, p-S6, S6, p-Akt(S473), Akt, p-p38 MAPK, p38 MAPK, p-mTOR(S2448), mTOR, p-STAT3(Y705), and STAT3 antibodies were purchased from Cell Signaling Technology. LIF and LIFR antibodies were purchased from Santa Cruz Biotechnology. β-Actin and all secondary antibodies were purchased from Sigma. ALDEFLUOR assay kit was obtained from StemCell Technologies. The Ki-67 antibody was purchased from Abcam. LIFR Knockout (KO) model cells were generated using Genescript CRISPR gRNA Constructs (Genescript-s64729-LIFR CRISPR guide RNA 1; Genescript-s64731-LIFR CRISPR guide RNA 2) and transfecting them into Cas9 stably expressing BT-549 cells followed by puromycin selection. EC359 and EC330 were synthesized using the detailed synthetic protocol described in the patent WO 2016/154203 A1 (Evestra Inc.). Characterization of EC330 and EC359 produced was described in the Supplementary Methods.

For Western blotting, cells were lysed in RIPA buffer (Thermo Fisher Scientific) containing protease and phosphatase inhibitors. Total cellular lysates were mixed with 4× SDS sample buffer and run on SDS-PAGE, and transferred onto nitrocellulose membranes and blots were developed using antibodies and the ECL Kit (Thermo Fisher Scientific). Avidin-biotin pull down was performed as described previously (25). Briefly, BT-549 total cellular lysates and purified LIFR was incubated with Biotin-control or Biotin-EC359 overnight and incubated with NanoLink Streptavidin Magnetic Beads (Solulink) for 1 hour at room temperature. The binding of EC359 to LIFR was confirmed by Western blotting. Intensity of signaling bands in Western blots were quantitated using ImageJ program (NIH, Bethesda, MD).

The effect of EC359 on cell invasion of TNBC cells was determined by using the Corning BioCoat Growth Factor Reduced Matrigel Invasion Chamber assay. MDA-MB-231 and BT-549 cells were treated with vehicle or EC359 (25 nmol/L) for 22 hours and invaded cells in all the treatment conditions were determined according to the manufacturer's protocols.

CSCs from MDA-MB-231 and BT-549 cells were sorted using established stem cell marker ALDH using the ALDEFLUOR kit and flow cytometry. CSCs were cultured in MammoCult medium with the supplements as per manufacturer's instructions. The effect of EC359 on self-renewal of CSCs was determined by ELDA. Briefly, CSCs were seeded in decreasing numbers (100, 50, 20, 10, 5, and 1 cells/well) in 96 well ultra-low attachment plates and treated with vehicle or EC359. After 10 days, the number of wells containing spheres per each plating density was recorded and stem cell frequency between control and treatment groups was calculated using ELDA analysis software (http://bioinf.wehi.edu.au/software/elda/).

The effect of EC359 on cell viability of TNBC cells was assessed by using MTT assay as described previously (25). TNBC cells were seeded in 96-well plates (1 × 10³ cells/well) and after overnight incubation cells were treated with varying concentrations of EC359 for 5 days. To test the effect of EC359 on the viability of CSCs and non CSCs, CellTiter-Glo assays were performed (Promega). Briefly, cells were seeded in 96-well, flat, clear-bottom, opaque-wall microplates and treated with vehicle or EC359 for 3 days. The total ATP content as an estimate of total number of viable cells was measured on an automatic Fluoroskan Luminometer. For clonogenic survival assays, cells were seeded in triplicates in 6-well plates (500 cells/well), after overnight incubation cells were treated with vehicle or EC359 for 5 days and after 2 weeks, colonies that contain ≥ 50 cells were counted and used in the analysis. The effect of EC359 on apoptosis was measured by Annexin V/PI staining and Caspase-3/7 activity assay as described previously (25, 26). Briefly, MDA-MB-231 and BT-549 cells were seeded in 96-well plates (2 × 10³/well), after overnight incubation cells were treated with vehicle or EC359 (20 nmol/L) for 72 hours. After treatment, equal amount of caspase-3/7 substrate containing solution was added to the media, and luciferase activity was measured using luminometer according to the manufacturer's protocol (Promega).

Reverse transcription (RT) reactions were performed by using SuperScript III First Strand kit (Invitrogen), according to the manufacturer's protocol. Real-time PCR was done using SybrGreen on an Illumina Real-Time PCR system. Primer sequences were included in the Supplementary Table S1.

Binding profiles of EC359 to LIF/LIFR were evaluated using surface plasmon resonance (SPR). Recombinant human LIF was purchased from R&D Systems (catalog no. 7734-LF-500) and human LIFR-Fc was purchased from Speed Biosystems (catalog no. YCP1132). Sensor chips were purchased from ForteBio (www.fortbio.com). Detailed SPR protocol was provided in the Supplementary Methods.

A serial dilution of the ligand (EC359) was prepared in a way to match the final buffer conditions in the reaction mix (10 mmol/L HEPES pH 7.4, 150 mmol/L NaCl, 3 mmol/L EDTA, 0.005 % Tween-20, 10% DMSO). The highest concentration of ligand was 2.00 μmol/L and the lowest 61.0 pmol/L. Five microliters of each dilution step was mixed with 5 μL of the fluorescent molecule. The final reaction mixture, which was loaded in capillaries, contained a respective amount of ligand (max. conc: 1.00 μmol/L; min. conc: 30.5 pmol/L) and constant 5 nmol/L fluorescent molecule (protein target LIFR-labeled fluorescent dye- NHS chemistry). Thermophoretic movement of fluorescently labeled protein with EC359 was performed using on a Monolith NT.115 Pico at 25°C, with 7% LED power and 60% Laser power (Nanotemper Technologies).

The atomic level interactions of EC359 against human LIFR (hLIFR) were studied by molecular modeling. The existing structural information of LIFR was utilized for the studies. The partial structure of human LIFR (hLIFR; domains D1–D5; PDB ID: 3E0G) and the structure of human LIF (hLIF) in complex with the partial murine LIFR (mLIFR; domains D1–D5) have been reported in the Protein Data Bank (PDB ID: 2Q7N; refs. 27, 28). As a preliminary step, the sequence and structural similarities of both of these LIFRs were deduced. Furthermore, the three-dimensional structure of hLIF–hLIFR complex was constructed from hLIF-mLIFR by replacing mLIFR. The complex was energy minimized to avoid the residue clashes between the hLIFR and hLIF. From the minimized complex, the hLIFR was again separated and prepared for the docking studies. Because there was no information available on the ligand-binding sites, the whole receptor was probed using Sitemap (Schrödinger) to detect possible binding sites (29). Two steps of molecular docking were performed such as standard precision (SP) and induced fit (IFD) on the identified binding sites. The purpose of SP docking was to detect the binding strength and orientations of ligand at respective binding sites. On the basis of the docking scores, the sites were ranked. Later, an appropriate ligand pose was selected and flexible docking (IFD) was performed by allowing flexibility to the surrounding amino acids (around 6 Å from the center of the ligand). On the basis of the MM-GBSA (30) score and visual inspection an appropriate pose was selected and subjected to molecular dynamics simulation (MDS) to estimate the residence time of the ligand over a period of 25 nanoseconds. The detailed description of methods used in the study was included in the Supplementary Methods and Supplementary Figs. S2 and S3.

For STAT3-luc assays, MDA-MB-231 and BT-549 cells were stably transduced with STAT3-firefly luciferase reporter lentivirus purchased from Cellomic Technology. STAT3-luc reporter–expressing cells were serum starved for 24 hours, pretreated with EC359 for 1 hour, and then stimulated with LIF or other indicated ligands for 24 hours. Cells were lysed in passive lysis buffer, and the luciferase activity was measured by using the dual-luciferase reporter assay system (Promega) using luminometer.

All animal experiments were performed after obtaining University of Texas Health San Antonio (UTHSA) Institutional Animal Care and Use Committee (IACUC) approval, and all the methods were carried out in accordance with IACUC guidelines. MDA-MB-231 cells (2 × 10⁶) were mixed with equal volume of growth factor–reduced Matrigel and implanted in the mammary fat pads of 8-week-old female athymic nude mice as described previously (31). After tumor establishment, and achievement of measurable size, mice were randomized into control and treatment groups (n = 8 tumors per group). Control group received vehicle (hydroxymethylcellulose) and the treatment group received EC359 (5 mg/kg/day) 3 days per week subcutaneously. All mice were monitored daily for adverse toxic effects. Tumor growth was measured with a caliper at 3–4 day intervals, and volume was calculated using a modified ellipsoidal formula: tumor volume = 1/2(L × W²), where L is the longitudinal diameter and W is the transverse diameter. At the end of the experiment, mice were euthanized, and tumors were excised, and processed for histologic and biochemical studies.

The TNBC tissue was obtained from a deidentified surgical specimen (F0) just after surgery from a patient with invasive ductal carcinoma (pT3 pN2a pM) via UTHSA PDX Core. The tumor tissues were divided in to three parts; the first part was snap frozen and stored in liquid nitrogen, the second part was fixed in 10% buffered formalin and processed for histologic characterization, and the third part was placed in ice-cold PBS, cut into small pieces (3–5mm³), and engrafted into mammary fat pad of NCI SCID/NCr mice. PDX tumor was confirmed negative for ER, PR, HER2 by the Pathology core. Tumors from early passages were dissected into small pieces and implanted into the flanks of SCID mice. The mice were then randomized when they reached tumor volume of approximately 150 mm³ into control or treatment groups (n = 6 tumors per group). The control group received vehicle (hydroxymethylcellulose) and the treatment group received EC359 (10 mg/kg/day) 3 days per week subcutaneously. At the end of the treatment, tumors were excised and processed for histologic studies, protein, and RNA analysis.

TNBC tissues were collected from discarded surgical samples from UT Southwestern Medical Center (UTSW, Dallas, TX) patients for research purposes after obtaining the written informed consent and in accordance with institutional review board–approved protocol (STU-032011–187). All the studies were conducted in accordance with the Declaration of Helsinki. Tissues were processed and excised into small pieces and cultured on gelatin sponges for 24 hours in medium containing 10% FBS as described previously (25). Tissues were treated with vehicle or EC359 in culture medium for 72 hours and fixed in 10% buffered formalin at 4°C overnight and subsequently processed into paraffin blocks. Sections were then processed for IHC analysis for Ki-67.

IHC analysis was performed as described previously (25). Briefly, sections were blocked with normal goat serum (Vector Laboratories) followed by incubation overnight with Ki-67 (1:100) primary antibody and subsequent secondary antibody incubation for 30 minutes at room temperature. Immunoreactivity was visualized by using the DAB substrate and counterstained with hematoxylin (Vector Laboratories). Percent of Ki-67–positive proliferating cells was calculated in five randomly selected microscopic fields.

A pharmacokinetic study of EC359 was conducted in both mice and rats following intravenous and oral administration of the compound (GVK Bio). Intravenous formulation (5 mg/kg) was prepared as described: A required volume 0.1 mL of DMSO stock (20 mg/mL) was taken in an Eppendorf tube then 0.100 mL of DMSO was added and vortexed, then sonicated, followed by addition of 1.800 mL 10% Solutol in PBS, vortexed, and probe sonicated approximately 1–2 minutes to make a final solution of 1 mg/mL concentration. t_1/2, AUC_0-last, AUC_0-inf, AUC_extra, CL, Vd, MRT_0-last, and RSQ were measured using LC/MS-MS. For oral dosing (10 mg/kg) volume, 0.200 mL of DMSO stock (20 mg/mL) was taken in an Eppendorf tube then 1.800 mL 10% Solutol in PBS was added, vortexed, and probe sonicated approximately 1–2 minutes to make a solution of 2 mg/mL concentration. C_max, T_max, AUC_0-last, AUC_0-inf, AUC_extra, F%, MRT_0-last, and RSQ were measured. GR antagonism assays were performed using SelectScreen Biochemical Nuclear Receptor Profiling Service (Invitrogen).

All the statistical analyses were carried out using GraphPad Prism 6 software (GraphPad Software). A Student t test was used to assess statistical differences between control and EC359-treated groups. All the data represented in bar graphs are shown as mean ± SE. A P less than 0.05 was considered significant.

All data supporting the conclusions are included in the paper and/or in the Supplementary Materials.

We initially synthesized several compounds to target LIFR signaling using rationalized design based on crystal structure of LIF/LIFR. Within this series of compounds, one compound (EC330) showed higher potency (Fig. 1A, left). In cell viability assays (CellTiter Glo luminescent assay) using cancer cells, EC330 inhibited growth at approximately IC₅₀ ∼ 50 nmol/L (Fig. 1B). The reported X-ray crystallographic studies of LIF suggested a four α-helix bundle topology with a compact core predominantly composed of hydrophobic residues contributed by the four α-helices (32). Initial structure–activity relationship studies in our laboratory have shown the following structural features are necessary for the LIF inhibitory action: (i) difluro-acetylenic function at the 17-alpha position and (ii) 4′-substituition at the 11-phenyl ring. Because the EC330 has a steroidal backbone, we investigated the binding of EC330 to steroid receptors such as glucocorticoid receptor (GR). EC330 showed some antagonism to GR (79.8 nmol/L), which may elicit unwanted toxicity. Therefore, we pursued medicinal chemistry modifications, which retained its potency on LIFR, while reduced steroidal–receptor interactions. Additional SPR studies and subsequent synthetic efforts resulted in the development of EC359 (Fig. 1A, right). To examine whether optimization of EC359 retained its activity on par with the initial lead compound EC330, we conducted several studies. Receptor-binding studies revealed EC359 has more desirable characteristics than EC330 including lack of affinity to GR (Fig. 1C). In cell viability assays, EC359 showed significant inhibitory activity on par with EC330 in BT-549 model cells (Fig. 1B).

To test whether EC359 directly bind to LIFR complex, binding profiles of EC359 to LIF/LIFR were evaluated using SPR. Two sets of studies were performed: (i) to verify the integrity of recombinant proteins, the interaction between LIFR and LIF was studied; (ii) small-molecule binding to LIF/LIFR by either immobilizing LIFR or LIF onto a sensor chip was tested. Results from the first set of studies confirmed the integrity of recombinant LIF and LIFR; LIF bound to immobilized LIFR-Fc with a binding constant of 7 μmol/L (Supplementary Fig. S1A). In the second set of studies, results showed EC359 binding to LIFR, but not LIF. Furthermore, EC359 bound to LIFR in a dose-dependent manner with K_d = 81 μmol/L (Supplementary Fig. S1B). The results confirmed that EC359 is a specific inhibitor of LIF/LIFR complex.

Ligand binding to the immobilized receptor in SPR technique will be insensitive to ligand-induced structural changes, and thus the measured affinity by SPR may obscure true (in vivo equivalent) affinity of the drug. Hence, we conducted an orthogonal assay, namely MST, where the receptor is not immobilized to verify EC359 binding to the receptor complex. MST is a powerful technique to quantify biomolecular interactions. By combining the precision of fluorescence detection with the variability and sensitivity of thermophoresis, MST provides a flexible, robust, and fast way to dissect molecular interactions (33, 34). MST analysis confirmed direct interaction of EC359 with LIFR with an estimated K_d of 10.2 nmol/L (Fig. 1D). To further demonstrate that EC359 directly interacts with LIFR, we generated biotinylated EC359. Biotin addition did not affected EC359 biological activity (Supplementary Fig. S1C). Using biotin-EC359, we examined whether it interacts with LIFR. Purified LIFR protein or BT-549 cellular lysate was incubated with biotin-EC359 and its ability to interact with LIFR was determined using avidin pull-down assay followed by Western blotting. Results elucidated that EC359 interacts with LIFR (Supplementary Fig. S1D and S1E).

The site predictions revealed five potential binding sites on the hLIFR (Supplementary Fig. S4) in which two sites (2 and 3) are close to the LIF-binding interface (Supplementary Fig. S4). The SP docking was performed on all the five sites and docking scores were deduced (Fig. 1EA). The docking scores toward different binding sites range from -5.8 to -1.6 kcal/mol. It was also observed that EC359 has exhibited more promising scores toward site-3 compared with other sites. The binding poses obtained from the docking were superimposed to the hLIF–hLIFR complex to see the potential clashes between the ligand and LIF. As expected, the binding poses at site-3 are making steric clashes with residues of LIF (Supplementary Fig. S5). Because the SP docking is a rigid docking method, the ligand-induced conformational changes were also studied using IFD by applying flexibility to the surrounding residues. Using standard protocol, the side chains were optimized and 28 poses were generated. The binding energies of all the 28 poses range from -80 kcal/mol to -33 kcal/mol (Supplementary Fig. S6). It was observed that all the generated ligand poses are potentially making steric clashes with hLIF. One of the top scored poses (binding energy = -77 kcal/mol) was critically analyzed for the detailed atomic level interactions (Fig. 1EB). In the selected pose, ligand-induced conformational changes were observed for the loops close to the LIF-binding region (Supplementary Fig. S7). The ligand EC359 was found to sandwich between two loops at the N-terminal of D4 domain by orienting the difluro-acetylenic group to the bulk solvent. The keto group of the EC359 was found to be involved in two hydrogen bonds with the side chain of T308 and the backbone of T316. Similarly, the hydroxyl group of the ligand was also found to mediate a hydrogen bond with the sidechain of E340. Moreover, van der Waals contacts with the surrounding residues were also found to contribute to the ligand binding. It was observed that EC359 binding to hLIFR would prevent hLIF binding due to steric clashes (Fig. 1EC). As a final step, the snapshots obtained from the MD simulation were superimposed with the initial pose and RMSD was calculated for protein and ligand separately. The snapshots were analyzed and found that the structural distortions are affected mainly to the loops (connecting separate domains in the LIFR) and the terminal regions. At the same time the ligand is found to remain bound at the binding site even after 25 nanoseconds of MD simulation. The protein ligand contacts over 25 nanoseconds of MD simulation are shown in (Fig. 1ED; Supplementary Fig. S8).

We then conducted pharmacokinetic analysis of EC359 using various established tests (Supplementary Fig. S9). Results from intravenous dosing studies using 5 mg/kg in rats indicated a mean C0 of 74669.11 ng/mL, t_1/2(hours) of 3.86 hours, AUC_0-last (ng·hour/mL) of 15,544.36, and AUC_0-inf (ng·hour/mL) of 15,573.91. Results from oral dosing studies using 10 mg/kg in rats indicated a mean C_max (ng/mL) of 919.50, T_max (h) of 2.67, AUC_0-last (ng·hour/mL) of 3,792.26, and AUC_0-inf (ng·hour/mL) of 3,876.82. Ames test confirmed that EC359 did not induce an evident (significant) >2-fold increase in the revertant counts at the doses tested (dose related), in the tester strains both with and without metabolic activation according to the evaluation criteria mentioned in OECD guideline no.471. Hence, the compound EC359 is considered nonmutagenic with Salmonella typhimurium strains TA98, TA100, TA1535, TA1537, and E.coli combo, both with and without metabolic activation. In hERG cardiotoxicity screening, up to 30 μmol/L concentration of EC359 did not show 50% inhibition, hence no liability. Among CYP enzymes, EC359 inhibits 2D6, therefore, caution is warranted in concurrent administration of drugs that inhibits 2D6 such as Prozac. Metabolic stability and plasma stability are moderate in human with high plasma protein binding. Good bioavailability (pharmacokinetics) was observed in both intraperitoneal and oral dosing (Supplementary Fig. S9). Collectively, the data from these studies indicate EC359 has specific on-target activity (pharmacodynamics) and suggests EC359 as a druggable candidate for further development.

We first examined the expression of LIF and LIFR in cells that represent various subtypes of TNBC (BT-549, SUM-159, MDA-MB-231, HCC1937, MDA-MB-468, and HCC1806), ER⁺ breast cancer (MCF7 and T47D) as well as normal mammary epithelial cells (HMEC). We found that five of the six TNBC cells expressed high levels of LIF and LIFR when compared with ER⁺ breast cancer cells and normal cells (Fig. 2A and B). Next, we examined the efficacy of EC359 on cell viability of TNBC and ER⁺ breast cancer cells. Treatment with EC359 resulted in a significant dose-dependent reduction in the cell viability of TNBC cells (IC₅₀ = 10–50 nmol/L) and their inhibition is well correlated with LIF and LIFR expression levels (Fig. 2C). Interestingly ER⁺ breast cancer cells that express low levels of LIF and LIFR exhibited low sensitivity to EC359 treatment (IC₅₀ > 1,000 nmol/L) when compared with TNBC cells (Fig. 2D). To further confirm the target specificity of EC359, we generated doxycycline-inducible LIFR-KO cells using Cas9 stably expressing TNBC cells. Results indicated a reduction of LIFR expression in BT-549 models contributed to the resistance of the EC359-mediated decrease in cell viability (Fig. 2E). Collectively, these data suggest that EC359 activity depends on presence of functional LIF/LIFR signaling axis in cells.

We next examined the efficacy of the EC359 on the survival of TNBC cells. In clonogenic survival assays, EC359 significantly reduced the colony-forming ability of MDA-MB-231 and SUM-159 cells (Fig. 2F). Given the important role of the LIF axis in the invasiveness of cancer cells, we examined the effect of EC359 in reducing the invasion of TNBC cells. Matrigel invasion assays demonstrated that EC359 significantly reduced the invasion potential of MDA-MB-231 and BT-549 cells (Fig. 2G). Furthermore, we examined whether EC359 induced apoptosis in TNBC cells using caspase 3/7 activity assay and Annexin V staining assay. EC359 treatment significantly increased caspase-3/7 activity (Fig. 2H) and Annexin V–positive cells (Fig. 2I) in both MDA-MB-231 and BT-549 cells. Collectively, these results suggest that EC359 exhibits significant inhibitory activity on invasion and promotes apoptosis of TNBC cells.

LIF/LIFR activates multiple signaling pathways including JAK/STAT3, MAPK, AKT, and mTOR; all of which are implicated in TNBC progression. To confirm the inhibitory effect of EC359 on LIF/LIFR–mediated STAT3 activation, BT-549 cells that stably express STAT3-Luc reporter were pretreated with vehicle or EC359 followed by stimulation with LIF. As expected, LIF treatment significantly increased the STAT3 reporter activity and this activation was inhibited by EC359 treatment (Fig. 3A). Because our modeling studies predicted EC359 interaction with the ligand-binding interface of LIFR, we examined whether EC359 also blocks signaling by other LIFR ligands such as Oncostatin M (OSM), Ciliary Neurotrophic Factor (CNTF), and Cardiotrophin 1 (CTF1). Results showed that EC359 blocked the OSM, CNTF, and CTF1-mediated STAT3 activity in BT-549 cells (Fig. 3A). We also confirmed that EC359 has the ability to block LIF-, OSM-, CNTF-, and CTF1-mediated STAT3 activation using MDA-MB-231 cells stably expressing STAT3-Luc reporter (Fig. 3B). In qRT-PCR assays using BT-549 cells, EC359 treatment significantly reduced the expression of several known STAT3 target genes (Fig. 3C).

To further confirm the effect of EC359 on LIF/LIFR downstream signaling pathways, MDA-MB-231 and BT-549 cells were pretreated with vehicle or EC359 and subsequently stimulated with LIF. STAT3 activation was examined using Western blotting. EC359 treatment substantially reduced the LIF activation of STAT3 in both BT-549 and MDA-MB-231 cells (Fig. 4A). EC359 also reduced the STAT3 activation by OSM and CNTF (Fig. 4B and C). In addition, EC359 treatment substantially decreased the phosphorylation of AKT, mTOR, S6, and ERK1/2 in MDA-MB-231 and BT-549 cells (Fig. 4D and E). EC359 treatment also increased the phosphorylation of proapoptotic p38MAPK in BT-549 cells (Fig. 4E). We confirmed whether alteration in downstream signaling seen upon EC359 treatment such as STAT3 occurs in the cell line that has a doxycycline-inducible deletion of the LIFR. Results showed that KO of LIFR significantly reduced the STAT3 activation (Supplementary Fig. S10A). Furthermore, stimulation of LIFR KO cells with LIF did not activate STAT3 in this model. However, EC359 is able to block LIF-mediated STAT3 activation in LIFR-expressing control cells. These results confirm that the downstream effects seen in EC359 are due to its effects on LIFR and that STAT3 is a downstream effector of LIFR in TNBC cells (Supplementary Fig. S10A). These results suggest that EC359 acts as a LIFR inhibitor and attenuates LIF and other LIFR ligand–mediated signaling in TNBC cells.

The LIF/LIFR axis plays a vital role in stemness (6, 12). To test the effect of EC359 on stemness, CSCs were isolated from MDA-MB-231 and BT-549 using ALDH⁺ flow cytometry sorting. EC359 treatment substantially decreased the phosphorylation of AKT, mTOR, p70S6K, and increased phosphorylation of proapoptotic p38MAPK in CSCs (Fig. 4F). Western blot analysis showed that ALDH⁺ (CSCs) and ALDH⁻ (non-CSCs) cells have similar levels of LIFR (Fig. 4G). Furthermore, in cell viability assays, EC359 similarly inhibited both ALDH⁺ and ALDH⁻ cells (Fig. 4H). To further study the effect of EC359 on the self-renewal ability of CSCs, extreme limiting dilution assays (ELDA) were performed. Results showed that EC359 significantly reduced the self-renewal of CSCs compared with control (Supplementary Fig. S10B). Furthermore, pretreatment of TNBC cells with EC359 significantly reduced the abundance of ALDH⁺ cells (Fig. 4I).

To test the efficacy of EC359 on in vivo tumor progression, we established MDA-MB-231 xenograft tumors in the mammary fat pad of nude mice. Mice were randomized to vehicle (hydroxymethylcellulose) and EC359 (5 mg/kg/day via subcutaneous injection) 3 days/week. EC359 treatment significantly reduced the tumor progression compared with vehicle (Fig. 5A). The body weights of mice in the vehicle and EC359-treated groups remained unchanged (Fig. 5B) confirming the low toxicity of EC359. Moreover, EC359-treated tumors exhibited fewer proliferating cells (Ki-67–positive cells) compared with vehicle-treated tumors (Fig. 5C). In addition, qRT-PCR analysis confirmed significant decrease in the activation of STAT3 target genes in EC359-treated tumors compared with vehicle (Fig. 5D). Western blot analysis confirmed that xenograft tumors express LIFR and LIF (Fig. 5E). Furthermore, EC359 treatment substantially reduced the phosphorylation of STAT3, ERK1/2, and AKT in tumors compared with vehicle-treated tumors (Fig. 5E). Collectively, these results suggest that EC359 has potent antitumor activity on TNBC in preclinical models.

We tested the utility of EC359 using an ex vivo culture model of primary breast tumors, which allowed for the evaluation of drugs on human tumors while maintaining their native tissue architecture (Fig. 6A). Briefly, surgically extirpated deidentified TNBC tissues were cut into small pieces and placed on gelatin sponge soaked in the culture medium and grown for a short term in the presence of vehicle or EC359 (Fig. 6A). Treatment of TNBC explants with EC359 substantially decreased their proliferation (Ki-67 positivity) compared with vehicle-treated tumors (Fig. 6B and C). Next, we tested the effect of EC359 on PDX tumor growth in vivo. EC359 treatment significantly reduced the tumor progression compared with the vehicle-treated control group (Fig. 6D) and did not affect the body weight (Fig. 6E). EC359-treated PDX tumors exhibited fewer proliferating cells compared with vehicle-treated tumors (Fig. 6F). qRT-PCR analysis confirmed a significant decrease in the activation of STAT3 target genes in EC359-treated mice (Fig. 6G). Western blot analysis confirmed that PDX tumors express LIFR and LIF (Fig. 6H). Furthermore, EC359 treatment substantially reduced the phosphorylation of mTOR, S6, and AKT in tumors compared with vehicle-treated tumors (Fig. 6H). These results suggest that EC359 has therapeutic activity in primary patient-derived TNBC explants and PDX tumors.

LIF is the most pleiotropic member of the IL6 family of cytokines (4) that signals via the LIFR (5). Recent evidence suggested tumors exhibit upregulated LIF/LIFR signaling via autocrine and paracrine mechanisms (1, 5–7). However, lack of specific inhibitors targeting the LIF/LIFR axis represents a critical barrier in the field. In this study, we rationally designed a small organic molecule, EC359 that emulates the LIF/LIFR-binding site and functions as a first-in-class LIFR inhibitor from a library of compounds. Using multiple TNBC cells, we demonstrated that EC359 decreases cell viability, invasion, and promotes apoptosis. Mechanistic studies using Western blot, reporter gene assays, and qRT-PCR confirmed significant reduction of activation of LIF/LIFR-mediated pathways. Utilizing PDX, and patient-derived explant (PDEx) models, we demonstrated the in vivo efficacy of EC359.

The molecular modeling and SPR suggests the putative binding site is at the interface of LIF and LIFR. EC359 may display longer resident time (i.e., slower k_off) in the LIF/LIFR complex as suggested by the molecular model. Reasonable (about 30 minutes) residence time (1/k_off) suggests it may display biological activity in vitro and in vivo. However, potency of EC359 may depend on the concentration of LIF. During the SPR assay, we noted that EC359 was incompletely dissolved in the running buffer. Thus, due to poor solubility of the inhibitor in running buffer (inhibitor precipitates at concentrations 25–50 μmol/L in 5% DMSO) derived kinetic constants must be considered approximate; spiking at 2 μmol/L (green color) suggest that the compound may be aggregated; microaggregation affects not only the transport property of the ligands, but also the off-rate. Thus, the weaker binding affinity derived from SPR studies is likely due to poor solubility of the compound, which may be attributed as limitation of SPR technique. Nonetheless, results from SPR show that EC359 is specific to LIFR.

Recently, the MST technique has been widely used for characterizing protein–ligand interactions. MST offers a unique advantage over conventional isothermal titration calorimetry (ITC); unlike SPR, in MST, the target is not immobilized, and ligand-binding is independent of size or physical properties of ligands. MST analysis indicated higher binding affinity (K_d) between LIF and LIFR (1.36 nmol/L) than LIFR and EC359 (10.2 nmol/L). Also, the longer residence time/slower k_off in SPR demonstrates this pertinent biological effect. These values were consistent with high nanomolar potency of EC359 in vitro and in vivo. The discrepancy in the binding affinity measured between SPR and MST assays may be due to either difference in steady-state binding (MST) versus kinetic binding (SPR), or drug-induced structural changes upon binding; structural changes at the binding site, or both. Despite the differences in SPR and MST techniques, our results show that EC359 directly binds and disrupts the LIFR signaling complex.

The LIF/LIFR axis exhibits differential effects, which depended on the cell type, including stimulating or inhibiting cell proliferation, differentiation, and survival (4, 11, 24). LIFR is also reported to function as a metastasis suppressor through the Hippo–YAP pathway (35) and confer a dormancy phenotype in breast cancer cells disseminating to bone (36). It should be noted that presence of LIF is important for LIFR activation as we have determined in our SPR analysis. Hence studies using low LIF-expressing cell lines such as MCF-7 or T47D may not have an overly active LIF/LIFR signaling. However, LIFR signaling is complex as multiple ligands activate LIFR including LIF, CNTF, OSM, and CTF1. Despite the ability of LIF to activate JAK1/STAT3, PI3K/AKT, and MAPK pathways in these cell lines, differences in signaling outcome may, in part, arise from differential levels of activation of these three pathways, multiple ligands to LIFR, and differences in tumor microenvironment (TME; refs. 1, 37).

Earlier studies revealed that LIF, CTF1, and OSM share an overlapping binding site located in the Ig-like domain of LIFR and different behaviors of LIF, CTF1, and OSM can be related to the different affinity of their site for LIFR (38). Our modeling studies predicted that EC359 will interact at the LIF–LIFR binding interface and block interaction of LIF to LIFR. In agreement with published studies, our reporter assays and Western blot analyses showed that EC359 has the ability to block the signaling mediated by other cytokines (CTF1, CNTF, and OSM) that interact LIFR at LIF/LIFR interface. Blockage of LIFR by EC359 can leverage additional benefit of interfering the LIFR–JAK–STAT pathway by all known four LIFR ligands. We speculate that the unique ability of EC359 to bind the common ligand-binding site blocks multiple ligands' interactions with LIFR offers an advantage over other biologics or small molecules that can only target either of these ligands alone. This may also account the apparent differences in the activity seen by EC359 in TNBC and ER⁺ breast cancer as TNBC expresses higher levels LIFR ligands compared with ER⁺ breast cancer. In SPR studies, we found that the presence of ligand LIF further enhanced LIFR interaction with EC359 compared with LIFR alone. Because the ER⁺ breast cancer cells lack or possess low levels of LIF and LIFR, the increased fold difference in activity (sensitivity) of EC359 toward TNBC cells may reflect presence of increased ligand/receptor levels in TNBC. Furthermore, EC359 is unable to block OSM, CTF1, and CNTF, interactions with their natural receptors (OSMR/gp130, LST/gp130, CNTFR/gp130 respectively); therefore, EC359 is less likely to affect the physiologic signaling of CTF1, CNTF, and OSM. As a consequence, the issue of toxicity is less likely to occur. Accordingly, in xenograft studies, we did not observe toxicity over the course of EC359 treatment. However, future studies using formal toxicity protocols are needed to address the toxicity concerns and is beyond the current scope of this work.

Breast cancer cells often exhibit autocrine stimulation of LIF–LIFR axis. Some subtypes of TNBC such as claudin-low are highly enriched for CSC markers (39, 40). The LIF promoter is hypermethylated in normal breast epithelial cells, but extensively demethylated during breast cancer progression (5). TNBC cells have higher expression of LIF and LIFR compared with ER⁺ breast cancer cells and overexpression of LIF is significantly associated with a poorer relapse-free survival in patients with breast cancer (4). Together, these emerging findings strongly suggest that LIF signaling in TNBC may be clinically actionable and that disruption of the LIF signaling cascade has potential to block progression of subtypes of TNBC that exhibit a LIF/LIFR autocrine loop.

LIF activates multiple signaling pathways via LIFR including STAT3, MAPK, AKT, and mTOR (3, 4)—all are implicated in cancer progression. Tumors exhibit upregulated LIF–JAK–STAT3 signaling via autocrine and paracrine mechanisms (1, 5–7). LIF signaling also plays a role in crosstalk between tumor cells and fibroblasts, and mediates the proinvasive activation of stromal fibroblasts (9). LIF/LIFR signaling is implicated in modulation of multiple immune cell types present in tumor microenvironment (TME) including T-eff, T-reg, macrophages (41), and myeloid cells, which results in immune suppression (42). In our studies using TNBC model cells, we found that EC359 substantially reduced the activation of STAT3, MAPK, AKT, and mTOR; and significantly delayed tumor progression in vivo. However, our mechanistic studies are limited to EC359 effects on epithelial cells; future studies are needed to clearly examine the effect of EC359 on TME.

LIF and LIFR are overexpressed in multiple solid tumors (5, 7, 43). While LIF can act on a wide range of cell types, LIF knockout mice have revealed that many of these actions are not apparent during ordinary development (1), indicating a potential therapeutic window for LIF/LIFR axis inhibitors in addition to less toxicity in normal adult tissues. Considering the importance of the LIF/LIFR axis in cancer, humanized anti LIF antibody (MSC-1) that blocks LIF signaling is being tested in a phase I clinical trial mode to determine its safety and tolerability (ClinicalTrials.gov, NCT03490669). Given the wide deregulation of the LIF/LIFR axis in multiple tumors, the small-molecule LIFR inhibitor EC359 may have utility in treating other solid tumors including glioblastoma, ovarian cancer, colon cancer, and pancreatic cancer all of which exhibit dysregulated LIF/LIFR signaling. Our studies only examined the utility of EC359 using TNBC models. Future studies are needed to further evaluate the effects of EC359 in other cancer models and to examine any potential beneficial effects of EC359 on TME.

In summary, our data demonstrated that EC359 is a highly potent and specific LIFR inhibitor. EC359 blocked LIF/LIFR physical and functional interaction, signaling, and reduced cell viability of LIF/LIFR–expressing TNBC cells both in vitro and in vivo. EC359 represents a exciting new mechanism to modulate LIF/LIFR oncogenic functions. Because EC359 is a small, stable molecule, it is amenable for translation to clinical trials for patients with TNBC as either monotherapy or in combination with current standard of care.

G.V. Raj is a founder at EtiraRx and GaudiumRx, reports receiving a commercial research grant from Bayer, has received speakers' bureau honoraria from Bayer, Janssen, Astellas, and Pfizer, is a consultant/advisory board member for Bayer and Janssen. K.J. Nickisch has ownership interest (including stock, patents, etc.) from Evestra. No potential conflicts of interest were disclosed by the other authors.

Conception and design: S. Viswanadhapalli, G.R. Sareddy, B. Santhamma, A. Chávez-Riveros, G.V. Raj, A.J. Brenner, R.R. Tekmal, H.B. Nair, K.J. Nickisch, R.K. Vadlamudi

Development of methodology: S. Viswanadhapalli, G.R. Sareddy, B. Santhamma, A. Chávez-Riveros, M. Bajda, A.J. Brenner

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Viswanadhapalli, Y. Luo, M. Zhou, M. Li, S. Ma, R. Sonavane, U.P. Pratap, X. Li, A. Chang, A. Chávez-Riveros, X. Pan, R. Murali, M. Bajda, G.V. Raj, M.K. Rao, R.R. Tekmal

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Viswanadhapalli, Y. Luo, G.R. Sareddy, M. Li, R. Sonavane, U.P. Pratap, A. Chang, A. Chávez-Riveros, K.V. Dileep, K.Y.J. Zhang, X. Pan, M. Bajda, R.K. Vadlamudi

Writing, review, and/or revision of the manuscript: S. Viswanadhapalli, G.R. Sareddy, B. Santhamma, K.A. Altwegg, K.V. Dileep, K.Y.J. Zhang, R. Murali, M. Bajda, G.V. Raj, A.J. Brenner, M.K. Rao, H.B. Nair, K.J. Nickisch, R.K. Vadlamudi

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): G.V. Raj, V. Manthati

Study supervision: R.K. Vadlamudi

Other (conducted modeling studies): K.V. Dileep, K.Y.J. Zhang

We thank Jessica Perry (Ob/Gyn UT Health San Antonio) for proofreading of the manuscript. MST studies were performed by 2bind GmbH, Germany. K.V. Dileep thanks Japan Society for the Promotion of Science for a postdoctoral fellowship. This study was supported by the DOD BCRP grant W81XWH-18-1-0016 (to R.K. Vadlamudi and K.J. Nickisch), DOD BCRP grant W81XWH-16-0294 (to R.R. Tekmal), NCI Cancer Center Support Grant P30CA054174-17, Max and Minnie Tomerlin Voelcker Fund (to G.R. Sareddy), and NIH grant 1R01CA179120-01 (to R.K. Vadlamudi and M. Rao).

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.