Purpose:

Leukemia inhibitory factor (LIF) is a multifunctional cytokine with numerous reported roles in cancer and is thought to drive tumor development and progression. Characterization of LIF and clinical-stage LIF inhibitors would increase our understanding of LIF as a therapeutic target.

Experimental Design:

We first tested the association of LIF expression with transcript signatures representing multiple processes regulating tumor development and progression. Next, we developed MSC-1, a high-affinity therapeutic antibody that potently inhibits LIF signaling and tested it in immune competent animal models of cancer.

Results:

LIF was associated with signatures of tumor-associated macrophages (TAM) across 7,769 tumor samples spanning 22 solid tumor indications. In human tumors, LIF receptor was highly expressed within the macrophage compartment and LIF treatment drove macrophages to acquire immunosuppressive capacity. MSC-1 potently inhibited LIF signaling by binding an epitope that overlaps with the gp130 receptor binding site on LIF. MSC-1 showed monotherapy efficacy in vivo and drove TAMs to acquire antitumor and proinflammatory function in syngeneic colon cancer mouse models. Combining MSC-1 with anti-PD1 leads to strong antitumor response and a long-term tumor-free survival in a significant proportion of treated mice.

Conclusions:

Overall, our findings highlight LIF as a therapeutic target for cancer immunotherapy.

Translational Relevance

The potent therapeutic effects observed with immune oncology agents only occur in a minority of patients. Hence, there is an urgent need to find additional therapeutic targets that afford deep immunologic responses broadly across patients with cancer. We demonstrate that neutralization of leukemia inhibitory factor (LIF) represents a compelling approach to enable robust antitumor immunity.

Leukemia inhibitory factor (LIF) is a multifunctional member of the IL6 family of cytokines that includes IL6, IL11, oncostatin M, ciliary neutrotrophic factor, cardiotrophin-1, and cardiotrophin-like cytokine. LIF signals through the heterodimeric glycoprotein 130 (gp130)/LIF receptor (LIFR) complex at the cell surface (1–3). Activation of the receptor complex induces downstream signaling pathways including JAK/STAT3, PI3K/AKT/mTOR, and MAPK signaling (4). LIF is reported to have broad biological functions with roles in reproduction (5), embryonic stem cells (6), bone remodeling (7), the hypothalamo-pituitary-adrenal axis (8), and neural development (9). The diverse functions of LIF are clearly demonstrated by its pleiotropic effects in reproduction, where LIF is reported to regulate endometrial receptivity and interaction with the embryo (10), decidualization of the endometrial stroma (11), blastocyst invasion and development (12), as well as immune modulation of the implantation site and uterine leukocyte infiltration (13).

LIF also has complex functions in cancer, where it is widely thought to promote the development and progression of numerous solid tumor types and commonly correlates with poor prognosis. Although dysregulation of the LIF pathway has been described in glioblastoma (14, 15), rhabdomyosarcoma (16), pancreatic (17–19), nasopharyngeal (20), breast (21), and colon cancer (22), the specific role of LIF in tumors remains unclear. At a cellular level, LIF is reported to regulate the self-renewal of cancer stem cells (14, 18), activation of cancer-associated fibroblasts (CAF; ref. 23), suppressive functions of myeloid populations (24), tumor cell proliferation (17), and resistance to chemotherapy (18, 22), radiotherapy (20), and immune therapy. Overall, given the diverse functions ascribed to LIF, inhibition of LIF has the potential to impact multiple cancer mechanisms and thus represents a compelling and novel therapeutic approach.

In this study, we carefully examined the role of LIF in cancer. Gene expression association studies and in vitro modeling strongly supported a role for LIF driving polarization of immunosuppressive macrophages in human tumors. We developed a humanized high-affinity, function-blocking antibody against LIF (MSC-1), which improved tumor control in multiple syngeneic mouse models, and showed synergistic antitumor effects with a checkpoint inhibitor. Consistent with human gene expression association studies, mechanistic analyses of MSC-1–treated tumors revealed broad immunologic reprogramming of the tumor microenvironment (TME) with evident polarization of macrophages toward a proinflammatory phenotype. Overall, these data provide a compelling rationale to develop and test LIF inhibitors for cancer therapy. MSC-1 was recently evaluated in a phase I dose-escalation study to test its safety and preliminary antitumor activity (NCT03490669) and further development is planned.

Genomic analysis

All The Cancer Genome Atlas (TCGA) data were downloaded as the standardized datasets hosted and maintained by the BROAD GDAC Firehose (http://gdac.broadinstitute.org/) using the TCGA2STAT R package (25). These datasets comprised 7,769 samples distributed across 22 solid tumor indications (Supplementary Table S1), and the RSEM data were log2 transformed for all downstream analyses. Signature scores were computed by standardizing gene expression values to mean = 0 and SD = 1 and calculating the mean standardized expression across all genes comprising the relevant signature as described previously (26). RNA sequencing (RNA-seq) of LIF-treated human macrophages was completed in collaboration with the Princess Margaret genomics centre (www.pmgenomics.com) using the Illumina NextSeq500 (Illumina). Reads were mapped to GRCh38, transcripts were assembled using Stringtie, and finally gene-level FPKM were computed, as described previously (27). Gene set enrichment analysis (GSEA) was completed as described previously (28) and enrichment maps were visualized in Cytoscape (v3.2.1; ref. 29).

LIFR expression analysis on primary tumor cells

Dissociated tumor cells (Conversant Bio) were thawed and incubated with LIVE/DEAD Fixable violet stain (Thermo Fisher Scientific) prior to incubation in FcR block (Miltenyi Biotec) and subsequent staining using the following antibody conjugates: APC/Cy7-CD3, BV650-CD45, FITC-CD14, AF700-CD16, PE/Cy7-CD11b, BV510-CD163, PerCP/Cy5.5-CD206 (BioLegend), APC-CD68 (Miltenyi Biotec), and PE-LIFR (Clone H3, Laboratory of Gene Expression and Cancer, Vall d'Hebron Institute of Oncology). Cells were analyzed using a Beckman Coulter Cytoflex and data analysis was performed using FlowJo v10.

In vitro human macrophage experiments

Human macrophages were differentiated in vitro from peripheral blood CD14+ monocytes (Stemcell Technologies) for 7 days in 50 ng/mL CSF1 (Thermo Fisher Scientific) at 1 × 106 cells/mL, then dissociated with Accutase (StemPro) and replated in 50 ng/mL MCSF with or without 20 nmol/L recombinant human LIF (rhLIF; ACROBiosystems) at 1.5 × 105 cells/mL and cultured for 3 days. Flow cytometry was performed as above using the following antibody conjugates: PE/Cy7-CD206 (BioLegend), APC-CD68, and FITC-CD163 (Miltenyi Biotec). For T-cell suppression assays, 96-well assay plates were coated with 2 μg/mL Ultra-LEAF anti-CD3 (clone UCHT1, BioLegend) or PBS alone for 2 hours at 37°C. In vitro differentiated macrophages were harvested by Accutase treatment followed by washing with complete media (RPMI1640 with 10% heat-inactivated FBS and pen/strep). Allogeneic peripheral blood mononuclear cells (PBMC; Stemcell Technologies) were thawed and labeled with CellTrace Violet (Thermo Fisher Scientific) according to the manufacturer's protocol. PBMCs and macrophages were combined at a ratio of 12:1 in the coated 96-well plate and incubated for 72 hours prior to LIVE/DEAD Fixable Red staining (Thermo Fisher Scientific) and analysis by flow cytometry using the following antibody conjugates: FITC-CD3, PerCP/Cy5.5-CD4, and APC/Cy7-CD8. IFNγ secretion of PBMC and/or macrophage cocultures was quantified by analysis of the cell supernatants using the Human IFNg UPLEX Assay [meso scale discovery (MSD)] according to the manufacturer's instructions. For protein expression of pSTAT-3 by Western blot analysis, human macrophages were treated with or without 20 ng/mL of recombinant human LIF (Millipore, # LIF1010) alone or together with 10 μg/mL MSC-1 for 15 minutes. After that, cells were collected with complete RIPA lysis buffer to proceed with Western blot analysis. Human antibodies used were: pSTAT3 (Cell Signaling Technology, #9145) and β-Actin (Sigma, #A3854).

Protein expression and purification

The genes for the human LIF construct (residues 1–180) and LIFR (residues 45–533), containing a C-terminal TEV protease cleavage site followed by a tandem His10x tag and gp130 (residues 123–323) containing a C-terminal tandem His6x tag, were codon optimized (GeneArt) for expression in human cells and cloned into the pHLsec vector (30). Plasmids encoding LIF, LIFR, and gp130 to be used in binding studies were transiently transfected and expressed in HEK 293F cells and purified using Ni-NTA affinity chromatography followed by gel-filtration chromatography (Superdex 200 Increase) in 20 mmol/L Tris pH 8.0 and 150 mmol/L NaCl. For structural studies, LIF was transiently expressed in HEK 293S (Gnt I−/−) cells and purified as described above. Recombinant MSC-1 Fab was transiently expressed in HEK 293F cells and purified using KappaSelect affinity chromatography, followed by cation exchange chromatography (MonoS). The co-complex used for crystallization studies was prepared from purified MSC-1 Fab and LIF mixed at a 1:2.5 molar ratio and incubated at room temperature for 30 minutes followed by TEV cleavage and deglycosylation using EndoH. Gel filtration chromatography was subsequently used to purify the co-complex.

For the crystallization of unliganded MSC1, Fab fragment was purified from papain-digested IgG by affinity purification using Protein A. The flow through which contains the Fab was recovered, and further purified by cation exchange chromatography (MonoS) and gel filtration chromatography.

IHC was performed as follows. Briefly, slides were deparaffinized and hydrated, then incubated with pH 6 or pH 9 antigen retrieval solution (Dako), 10 minutes with 10% peroxidase (H2O2) solution and then with blocking solution (3% BSA) for 1 hour at room temperature. As a detection system, EnVision FLEX + (DAKO) was used according to the manufacturer's instructions, followed by counterstaining with hematoxylin, dehydration, and mounting (DPX). Quantification of pSTAT3, Ki67, and cleaved caspase3 (CC3) was performed with QuPath software, counting the total number of cells of three different fields per mouse, 3 mice/group, and calculating the percentage of positive cells. Data in graphs are presented as mean ± SEM. Quantification of LIF was performed using H-score method (3× percentage of strong staining + 2× percentage of moderate staining + percentage of weak staining), giving a range of 0–300. IHC antibodies: human/mouse LIF (Atlas; HPA018844; 1:200), murine pSTAT3 (Cell Signaling Technology; 9131; 1:50), murine Ki67 (AbCam; ab15580; 1:200), and murine CC3 (Cell Signaling Technology; 9661; 1:500).

Tumor-derived macrophages and PBMCs coculture experiment

Total blood of Balb/c mice was processed using Ficoll-Paque Premium (GE Healthcare) to obtain the PBMCs. A total of 1 × 106 PBMCs per well were cultured in a 24-well plate [previously coated with αCD3 (BioLegend, # 100340) or PBS for 2 hours] in presence of 20 ng/mL IL2 (R&D, # 402-ML-020). A total of 24 hours later and before the coculture, PBMCs were stained using carboxyfluorescein diacetate succinimidyl ester (Thermo Fisher Scientific, # C34554).

CD11b+cells were isolated from CT26 tumors of IgG or MSC-1–treated mice using CD11b magnetic beads (Miltenyi Biotech, # 130-049-601), following manufacturer instructions. Purity of the isolation was checked by flow cytometer. A total of 1 × 105 CD11b+ cells were plated in a 24-well plate and allowed to seed for 2 hours.

Coculture was performed at a ratio of 10:1 (PBMCs:CD11b+ cells), incubating them for 48 hours. T-cell activity was analyzed by flow cytometry and ELISA. Murine antibodies used were: CD3 (17-0032-82; 1:40) from eBioscience, CD45 (BD-550994; 1:200), CD8 (BD-557654; 1:40), and CD25 (555432; 1:20) from BD Biosciences. Samples were previously incubated with LIVE/DEAD fixable yellow dead stain kit (L34959; 1:2,000; Thermo Fisher Scientific). For acquisition, Navios (Beckman Coulter) was used and data were analyzed with Flow Jo software.

For the quantitative determination of IFNγ and CXCL9 protein levels secreted to the media, Mouse Duo-Set ELISA kits (R&D, # DY485-05 and # DY492) were used, following manufacturer's specifications.

Crystallization trials and structure determination

The LIF–MSC-1 co-complex was concentrated to 20 mg/mL and set up for crystallization trials, using commercial sparse matrix screens (JCSG Top96 and Mycrolytic MCSG). Crystals formed at 4°C in a condition containing 19% [volume for volume (v/v)] isopropanol, 19% (w/v) PEG 4000, 5% (v/v) glycerol, 0.095 mol/L sodium citrate, pH 5.6. A crystal from this condition diffracted to a resolution of 3.1 Å at the 08ID-1 beamline at the Canadian Light Source.

For the crystallization of unliganded MSC-1 Fab, the protein was concentrated to 25 mg/mL and set up for crystallization trials using two commercial sparse matrix screens (JCSG Top96 and Mycrolytic MCSG). The protein crystallized in five different conditions: (i) 0.17 mol/L ammonium acetate, 0.085 mol/L sodium citrate HCl, pH 5.6, 25.5% (w/v) PEG 4000, 15% (v/v) glycerol; (ii) 1.0 mol/L lithium chloride, 20% (w/v) PEG 6000, 0.1 mol/L MES, pH 6.0; (iii) 0.6 mol/L sodium chloride, 0.1 mol/L MES NaOH, pH 6.5, 20% (w/v) PEG 4000; (iv) 17% (w/v) PEG 4000, 15% (v/v) glycerol, 8.5% (v/v) isopropanol, 0.085 mol/L HEPES, pH 7.5; and (v) 0.16 mol/L magnesium chloride, 0.08 mol/L Tris HCl, pH 8.5, 24% (w/v) PEG 4000, 20% (v/v) glycerol. Crystals diffracted to resolutions ranging from 1.8 to 2.0 Å at the 23-ID-D beamline at the Argonne Photon Source. Data were processed and scaled using XDS (31). Structures were determined by molecular replacement using Phaser (32) with human LIF (PDB:2Q7N) and a Fab from our internal database as search models. Interactions of model building and refinement were performed using Coot (33) and phenix.refine (34) until the structures converged to an acceptable Rwork and Rfree. All softwares were accessed through SBGrid (35) and statistics are reported in Supplementary Table S5.

Isothermal titration calorimetry binding studies

Isothermal titration calorimetry (ITC) experiments were performed using an Auto-ITC200 (Malvern Instruments). For competition binding studies with gp130, a precomplex mixture of LIF and MSC-1 at a concentration of 50 μmol/L and 100 μmol/L, respectively, was injected into 5 μmol/L of gp130. For competition binding studies with LIFR, a precomplex mixture of LIF and MSC-1 at a concentration of 50 μmol/L was injected into 5 μmol/L of LIFR. Heat of binding was monitored for a total of 25 injections with an injection volume of 1.5 μL each and 180 seconds spacing between injections at 25°C.

Cell lines

CT26, U251, and HCC1954 cells were obtained from the ATCC, were verified by short tandem repeat analysis, and passaged less than 6 months prior to being used to complete experiments. MC38 cells were obtained from the NCI. All lines were routinely tested for Mycoplasma.

Thiazolyl blue tetrazolium bromide (MTT) assay

A total of 2 × 104 cells were plated in 24-well plates and the day after they have been treated with 3 μmol/L of MSC-1 or control IgG1. MTT assay was performed at 24, 48, and 72 hours after treatment as described below. Medium was aspirated, cells were washed twice with PBS, and then incubated with 0.5 mg/mL of MTT solution (Abcam, ab146345) at 37°C for 5–10 minutes. MTT solution was removed and formazan crystals resuspended in 200 μL of DMSO per each well and incubated for 15 minutes at room temperature by shaking. A total of 100 μL of solution was passed in 96-well plate and absorbance at 565 nm was read at the spectrophotometer Infinite M200Pro (Tecan).

Syngeneic mouse models

Animal work was performed at the University Health Network Animal Resources Centre according to the guidelines of the University of Toronto Animal Care Committee under AUP# 5565.2. All animal work was performed by technical staff of the University Health Network Animal Resources Centre who were blinded to the nature of all administered treatments. Female C57Bl/6 and Balb/c mice were purchased from The Jackson Laboratory. For all experiments MSC-1 (Northern Biologics) and RMP1-14 (BioXCell) was administered twice weekly by intraperitoneal injection at 15 mg/kg or 10 mg/kg, respectively. Tumor measurements and mouse weights were collected twice per week, and volumes were calculated using an ellipsoid formula (L × W2)/2. CT26 tumors were seeded by inoculating 5.0 × 104 cells intradermally into the flanks of 6-week-old female Balb/C mice and MC38 tumors were seeded by inoculating 5.0 × 105 cells intradermally into the flanks of approximately 6-week-old female C57Bl/6 mice. Tumor rechallenges were completed in the opposite flank at a 10X cell dose on day 122. Mice were euthanized when a single largest tumor dimension exceeded 15 mm or showed grade 4 ulceration.

Tumor digestion and cell isolation

CT26 and MC38 tumors were harvested on the indicated days after cell implantation. Tumors were finely minced using a scalpel blade and digested in either Collagenase IV (1 mg/mL; StemCell Technologies) for 3 hours at 37°C or by using the mouse tumor dissociation kit (Miltenyi Biotec) according to the manufacturer's suggested protocol for 1 hour at 37°C. Digested material was filtered through a 40-μm strainer, washed once with RPMI1640, and resuspended in either RPMI1640 supplemented with 10% heat-inactivated FBS or Flow buffer (PBS/2% FBS/2 mmol/L EDTA) depending on downstream applications.

Immunophenotyping and monitoring of tumor-specific CD8+ tumor-infiltrating lymphocyte responses by flow cytometry

Cells isolated from mouse tumors were stained with LIVE/DEAD fixable violet dead cell stain kit (Life Technologies) for 20 minutes at 4°C for dead cell exclusion. Samples were washed twice and then preincubated in mouse FcR blocking reagent (Miltenyi Biotec) for 20 minutes at 4°C to inhibit nonspecific antibody binding, followed by cell staining with the following fluorochrome-conjugated antibodies at predetermined dilutions for 20 minutes at 4°C (purchased from BioLegend, clones in parentheses): anti-I-A/I-E-AF-488 (M5/114.15.2), anti-CD11b-PerCP/Cy5.5 (M1/70), anti-Ly6C-PE/Cy7 (HK1.4), anti-F4/80-AF-647 (BM8), anti-Ly6G-AF-700 (1A8), anti-CD45-APC/Fire 750 (30-F11), anti-CD206-BV650 (C068C2), CD3-AF-488 (17A2), and CD8-PerCP/Cy5.5 (53-6.7). For direct ex vivo analysis of CD8+ tumor-infiltrating lymphocyte (TIL) function based on IFNγ production, freshly isolated tumor single-cell suspensions were stimulated with peptides (Genscript) corresponding to either the H2-Ld–restricted gp70 (423–431) MuLV epitope (AH1; the immunodominant epitope expressed by CT26 tumors) or the H2-Kb–restricted p15E (604–611) MuLV epitope expressed by MC38 tumors in the presence of Golgi Plug (3 μL/mL) and anti-CD28 (1 μg/mL, LEAF purified antibody, BioLegend) for 5 hours at 37°C. Following stimulation, cells were stained as above. Fixation/permeabilization was performed using the eBioscience Intracellular Fix & Perm Buffer set according to the manufacturer's suggested protocol and stained intracellularly with IFNγ-PE (XMG1.2). Stained samples were acquired on a CytoFLEX flow cytometer (Beckman Coulter) and analyzed using FlowJo (TreeStar).

qRT-PCR

CT26 tumors from IgG1 control- and MSC-1–treated animals were harvested on the indicated day after cell implantation. Tumors were homogenized in TRIzol (Life Technologies). RNA was extracted using RNeasy Mini Kit (Qiagen) according to manufacturer's instructions and further purified using RNase-Free DNase set (Qiagen). cDNA was prepared using iScript Advanced cDNA Synthesis Kit for qRT-PCR (Bio-Rad) according to manufacturer's instructions. qPCR was performed using SsoAdvanced Universal SYBR Green Supermix (Bio-Rad) and PrimePCR Custom Plate with mouse probes (Bio-Rad) on CFX384 (Bio-Rad). mRNA levels were normalized to Hprt, Tbp, and Tfrc and reported as fold change relative to control-treated tumors.

LIF ELISA

Plasma was collected at indicated timepoints by mandibular bleeds without anesthesia, and at study endpoint by cardiac puncture under Ketamine (Ketolar50) anesthesia. Plasma was collected in K2EDTA-coated tubes, preserved, and frozen until used (−80°C). LIF ELISA (Mouse LIF ELISA Kit, R&D, catalog no. MLF00) was performed according to manufacture protocol with two modifications. The standard was generated by using recombinant mouse LIF (ACROBiosystems, catalog no. LIF-M5227) spiked into pooled mouse plasma (Lamoire Biological Laboratories, catalog no. 7314307) at a concentration of 18 nmol/L with constant levels of MSC-1 (67 μmol/L). A 2-fold dilution series was generated to obtain a standard curve. Standards and plasma samples were incubated in MSD lysis buffer for 10 minutes before they were transferred to the ELISA plate. The SpectraMax i3 plate reader (Molecular Devices) was used to determine optical density of samples. All data were analyzed using Prism (GraphPad). A linear standard curve was calculated by linear regression. Only values within the linear range were considered for concentration calculations.

Cell proliferation

CT26 cells were plated at low density in clear bottom 96-well plate in growth media at 3,500 cells/well. Plates were maintained overnight in a 37°C/5% CO2 incubator and either rLIF of MSC-1 was added at the indicated concentration. All treatments were performed in triplicates. With the exception of mitomycin C, cells were grown in the presence of treatment throughout the experiment. Mitomycin C (in complete media) treatment was for 1 hour and washed off, and cells were thereafter maintained in complete media without mitomycin C for the entire duration of the experiment. The plates were placed into the IncuCyte Zoom contained within a 37°C/5% CO2 incubator. Wells were individually scanned every 4 hours for 3 days using the built-in camera with 4× magnification. Change in percentage of confluence over time was used as the readout for proliferation rate. Culture confluence was quantified using the Essen Bioscience IncuCyte Zoom software.

Data availability

RNA-seq data of LIF-treated human macrophages have been deposited into Gene Expression Omnibus (GSE211936). At the time of print publication, only processed data (gene-level FPKM) is available in GEO because the raw data sequencing files are on failed hardware; the authors are working to retrieve this data and will post the recovered data as soon as it is available. Please contact the corresponding authors with any questions and with additional data requests.

LIF is associated with an immunosuppressive TME

Given the broad and pleiotropic functions ascribed to LIF in cancer, we first sought to assess LIF expression patterns widely in human cancer samples with the aim of gaining further insights into the potential of LIF as a cancer target and to provide additional perspective on relevant LIF mechanisms within and across tumor types. We evaluated LIF transcript levels across 7,769 patients representing 22 solid tumor indications profiled as part of TCGA initiative (Supplementary Table S1). LIF transcript was variably expressed across and within tumor indications with all indications having at least a subset of tumors falling in each LIF expression quartile defined globally across the 7,769 patient samples (Fig. 1A). Similarly, we performed LIF IHC in tissue microarrays (TMA) from colon and ovarian cancer and observed a heterogenous expression of LIF protein with some tumors expressing extremely high levels of LIF and others not expressing LIF (Fig. 1B). To gain insight into the biological processes associated with LIF transcript expression, we completed GSEA using LIF expression as a phenotypic variable in the colon, ovarian, and bladder cancer cohorts (Supplementary Fig. S1A–S1C). These indications were selected on the basis of having a wide distribution in LIF expression levels. In these three cancer cohorts, LIF transcript expression was strongly associated with immunologically related gene sets including chemokine/cytokine/IL signaling, as well as innate and adaptive immunity, suggesting a role of LIF in regulating the immunologic microenvironment of tumors. In support of this, transcriptomic signatures (Supplementary Table S2) representing general lymphoid and myeloid cell types were also significantly associated with LIF transcript across essentially all solid tumor indications, with stronger associations being observed to the general myeloid signature (Fig. 1C and D). Intriguingly, LIF was initially discovered as regulator of myeloid differentiation in murine leukemias (36), which is consistent with our observation that LIF transcript is coexpressed with a myeloid cell signature in human tumors.

Figure 1.

LIF is associated with a macrophage-mediated immunosuppressive microenvironment in human cancer. A, Distribution of global LIF expression quartiles across 22 solid tumor indications. B, LIF IHC was performed in TMAs from human colon and ovarian cancer. The degree of staining was scored using H-score method. Representative images from the TMAs are shown. Scale bar, 50 μm. C and D, Regression and correlation plots with coefficients (Pearson) for LIF expression and expression signatures representing general lymphoid and myeloid cells (see Supplementary Table S1 for signature descriptions) across the 22 solid tumor indications. E, Correlation plots and coefficients for LIF expression and the Core_M2_Signature across the 22 solid tumor indications. F, −log(P value) calculated for the Pearson correlations between LIF and the Core_M2_Signature across the indications (P = 0.01 indicated by hashed line). G, Comparison of the Pearson r values for LIF and the Core_M2_Signature and an M1 signature (Azad_2018_M1) across the 22 solid tumor indications (t test). H–J, Flow cytometry analysis of LIFR expression on primary cells from ovarian cancer tumors or lung adenocarcinoma tumors. Cells were gated as tumor cells (CD45), TILs (CD45+ CD3+), or TAMs (CD45+ CD11b+ CD68+). LIFR MFI was normalized to respective FMO controls. K, Patient survival in the pan TCGA cohort stratified into high (top quartile), intermediate (middle quartiles), and low (bottom quartile, log-rank test). L, ROC curve analysis of the IMvigor210 dataset for LIF expression in patients responding (partial/complete responders) and nonresponding (stable or progressive disease) to immune checkpoint inhibitors. M, Patient survival in the IMvigor210 dataset from immune checkpoint inhibitor studies (anti-PDL1) stratified into high, intermediate, and low LIF levels. *, P < 0.05; **, P < 0.01; ***, P < 0.001. FMO, fluorescence minus one; MFI, mean fluorescence intensity.

Figure 1.

LIF is associated with a macrophage-mediated immunosuppressive microenvironment in human cancer. A, Distribution of global LIF expression quartiles across 22 solid tumor indications. B, LIF IHC was performed in TMAs from human colon and ovarian cancer. The degree of staining was scored using H-score method. Representative images from the TMAs are shown. Scale bar, 50 μm. C and D, Regression and correlation plots with coefficients (Pearson) for LIF expression and expression signatures representing general lymphoid and myeloid cells (see Supplementary Table S1 for signature descriptions) across the 22 solid tumor indications. E, Correlation plots and coefficients for LIF expression and the Core_M2_Signature across the 22 solid tumor indications. F, −log(P value) calculated for the Pearson correlations between LIF and the Core_M2_Signature across the indications (P = 0.01 indicated by hashed line). G, Comparison of the Pearson r values for LIF and the Core_M2_Signature and an M1 signature (Azad_2018_M1) across the 22 solid tumor indications (t test). H–J, Flow cytometry analysis of LIFR expression on primary cells from ovarian cancer tumors or lung adenocarcinoma tumors. Cells were gated as tumor cells (CD45), TILs (CD45+ CD3+), or TAMs (CD45+ CD11b+ CD68+). LIFR MFI was normalized to respective FMO controls. K, Patient survival in the pan TCGA cohort stratified into high (top quartile), intermediate (middle quartiles), and low (bottom quartile, log-rank test). L, ROC curve analysis of the IMvigor210 dataset for LIF expression in patients responding (partial/complete responders) and nonresponding (stable or progressive disease) to immune checkpoint inhibitors. M, Patient survival in the IMvigor210 dataset from immune checkpoint inhibitor studies (anti-PDL1) stratified into high, intermediate, and low LIF levels. *, P < 0.05; **, P < 0.01; ***, P < 0.001. FMO, fluorescence minus one; MFI, mean fluorescence intensity.

Close modal

Tumor-associated macrophages (TAM) comprise a major fraction of tumor-infiltrating myeloid cells within the TME and are generally thought to promote immunosuppression. Accordingly, we probed the relationship between LIF transcript expression and infiltration of TAMs by assessing the correlation between LIF transcript and a panel of suppressive macrophage and/or myeloid signatures curated from publicly available databases, publications, and internal sources (Supplementary Table S2). We further defined two additional TAM signatures, which we called Master_M2_Signature and Core_M2_Signature, that comprised all the unique genes present in the suppressive myeloid signature panel (Master) and the 30 genes present in at least two of the panel signatures (Core). In nearly every indication, LIF transcript expression significantly correlated with the individual signatures comprising this panel across the 22 solid tumor indications (Fig. 1E and F; Supplementary Fig. S2A–S2I). The strongest correlations across indications were observed for the Core- and Master_M2_Signatures. In contrast, LIF transcript expression was significantly less correlated with a signature representing immunostimulatory or M1 macrophages (Fig. 1G), suggesting that the association observed for LIF transcript and macrophage signatures was specific for immunosuppressive macrophages. To further confirm our observations, we examined the relationship between LIF transcript expression and signatures representing other common immunologic cell types and processes. The Core_M2_Signature consistently showed stronger correlation with LIF transcript when compared with additional signatures (Supplementary Table S3) representing macrophages, neutrophils, dendritic cells, B cells, and T cells, as well as T-cell processes (Checkpoints and Cytolysis), and generic tumor cell processes (proliferation and DNA repair; Supplementary Fig. S3A–S3K; Supplementary Table S2). Indeed, these results are supported by a recent report demonstrating the association between LIF and TAMs across a wide range of solid tumor types (15). Overall, these results demonstrate that LIF transcript expression is both robustly and specifically associated with gene signatures representing immunosuppressive macrophages.

We next investigated LIFR protein expression in human tumors by performing flow cytometry on single-cell suspensions prepared from human primary tumor samples. LIFR was consistently expressed highly on TAMs (CD45+/CD11b+/CD68+) but not tumor epithelial cells (CD45) or (CD45+/CD3+) present in dissociated tumor cells harvested from four human ovarian tumors (Fig. 1H and I). Similarly, dissociated tumor cells from two of four human lung tumors also showed specific expression of LIFR on TAMs relative to tumor cells and TILs (Fig. 1J). Hence, among cell types commonly found within the TME, TAMs generally express the highest levels of LIFR and likely represent a LIF-responsive cell population in human tumors.

Finally, we explored the relationship between LIF transcript expression and patient outcome in the pan-cancer TCGA cohort described above, as well as in a recently described cohort of patients treated with anti-PDL1 therapy (37). Subdividing patients within the pan-TCGA cohort based on global expression quartiles for LIF transcript (Q1: High, Q2–Q3: Int, Q4: Low), revealed that high and low levels of LIF transcript were significantly associated with worse and better overall survival, respectively (Fig. 1K; P < 0.0001). A similar result was observed when analyzing the colon, ovarian, and bladder cancer cohorts (Supplementary Fig. S4). These results suggested that high LIF transcript is associated with aggressive cancer. Given that tumor-infiltrating myeloid cells, including TAMs, have previously been described as drivers of resistance to checkpoint therapy (38), we tested whether LIF transcript levels were associated with response to anti-PDL1 and patient survival using published data from the IMvigor210 trial of atezolizumab in metastatic urothelial cancer. LIF transcript expression was a significant predictor of resistance (progressive/stable disease compared with partial/complete responses; Fig. 1L; AUC: 0.62, P = 0.0023). Furthermore, subdivision of patients based on quartiles revealed that high LIF transcript–expressing patients experienced worse survival than those whose tumors expressed low LIF transcript (Fig. 1M; P = 0.02). Taken together, these data indicate a strong association between LIF expression and TAMs, as well as link LIF expression with poor patient outcomes and resistance to checkpoint therapy.

LIF promotes immunosuppressive polarization of human macrophages

The above findings prompted us to test whether LIF treatment affected the phenotype and functional properties of human macrophages. Indeed, LIF-treated monocytes have previously been reported to inhibit CD4 T-cell proliferation and function (24). Human macrophages were derived by culturing CD14+ monocytes isolated from PBMC preparations in the presence of CSF1 for 7 days and macrophage differentiation was confirmed by monitoring CD68 expression. Although monocytes did not initially express detectable levels of LIFR, LIFR was induced in a time-dependent fashion during differentiation, with fully differentiated macrophages expressing the highest levels of LIFR, similar to our observations on TAMs (Supplementary Fig. S5A). Across three independent donors, LIF treatment of monocyte-derived macrophages markedly increased the expression of suppressive macrophage markers, including CD206, CD163, and CCL22 (Fig. 2A; Supplementary Fig. S5B–S5D). We next tested whether LIF-treated macrophages could suppress T-cell function. Across five donors, LIF-treated macrophages were potent suppressors of both CD8 (Fig. 2B) and CD4 T-cell proliferation (Supplementary Fig. S5E), relative to control-treated macrophages. Consistent with these findings, LIF-treated macrophages also suppressed IFNγ secretion to a greater extent than observed by control-treated macrophages (Fig. 2B). Overall, these data demonstrate the capacity of LIF to induce immunosuppressive phenotype and function in human macrophages.

Figure 2.

LIF drives polarization of immunosuppressive human macrophages. A, Surface expression of CD206 and CD163 are elevated on LIF-treated in vitro differentiated macrophages (left). Data from macrophages derived from three different monocyte donors are shown (right). MFI was normalized to respective FMO controls. B, Macrophage suppression of T-cell proliferation and IFNγ secretion is enhanced by LIF pretreatment. Representative flow cytometry histograms of gated CD8 T cells shows a lack of proliferation in the absence of CD3 activation (control) compared with CD3 activation (PBMCs). The bottom panels show that the suppression of CD3 activated T-cell proliferation by in vitro differentiated macrophages (MΦ + PBMCs) is enhanced by pretreatment of these macrophages with LIF (LIF MΦ + PBMCs). Data from macrophages derived from five different monocyte donors are shown (top right). The respective cell supernatants were analyzed for IFNγ secretion (bottom right). C, pSTAT3 Western blot analysis of human macrophages treated with the specified treatments for 15 minutes. D, Left, SOCS3 transcript levels in human macrophages treated ± LIF. Right, Transcript levels for a panel of suppressive macrophages in human macrophages treated ± LIF (n = 3 donors). E, Normalized enrichment scores calculated using the GSEA algorithm for immune suppressive/stimulator macrophage signatures in human macrophages treated ± LIF. F, The top three REACTOME gene sets showing positive enrichment in LIF-treated human macrophages with LIF. Heatmap depicts expression of LE genes driving enrichment of the indicated gene set. *, P < 0.05; **, P < 0.01; ***, P < 0.001. FMO, fluorescence minus one; MFI, mean fluorescence intensity.

Figure 2.

LIF drives polarization of immunosuppressive human macrophages. A, Surface expression of CD206 and CD163 are elevated on LIF-treated in vitro differentiated macrophages (left). Data from macrophages derived from three different monocyte donors are shown (right). MFI was normalized to respective FMO controls. B, Macrophage suppression of T-cell proliferation and IFNγ secretion is enhanced by LIF pretreatment. Representative flow cytometry histograms of gated CD8 T cells shows a lack of proliferation in the absence of CD3 activation (control) compared with CD3 activation (PBMCs). The bottom panels show that the suppression of CD3 activated T-cell proliferation by in vitro differentiated macrophages (MΦ + PBMCs) is enhanced by pretreatment of these macrophages with LIF (LIF MΦ + PBMCs). Data from macrophages derived from five different monocyte donors are shown (top right). The respective cell supernatants were analyzed for IFNγ secretion (bottom right). C, pSTAT3 Western blot analysis of human macrophages treated with the specified treatments for 15 minutes. D, Left, SOCS3 transcript levels in human macrophages treated ± LIF. Right, Transcript levels for a panel of suppressive macrophages in human macrophages treated ± LIF (n = 3 donors). E, Normalized enrichment scores calculated using the GSEA algorithm for immune suppressive/stimulator macrophage signatures in human macrophages treated ± LIF. F, The top three REACTOME gene sets showing positive enrichment in LIF-treated human macrophages with LIF. Heatmap depicts expression of LE genes driving enrichment of the indicated gene set. *, P < 0.05; **, P < 0.01; ***, P < 0.001. FMO, fluorescence minus one; MFI, mean fluorescence intensity.

Close modal

To fully characterize LIF-induced changes in macrophages, we completed global gene expression profiling of LIF-treated macrophages through RNA-seq analysis in three human donors. As expected, LIF treatment increased pSTAT3 levels and the expression of SOCS3, a well-known target of LIF-induced pSTAT3 signaling, in addition to multiple prototypic suppressive macrophage genes, consistent with our previous data (Fig. 2C and D). GSEA using the suppressive macrophage signatures from the TAM signature panel, described above, showed uniform positive enrichment in the LIF-treated macrophages whereas the immunostimulatory macrophage signature was negatively enriched (Fig. 2E), together suggesting that LIF drives global changes broadly associated with immunosuppressive effects. Further GSEA, using an un-biased collection of gene sets curated from the Reactome pathway database, revealed association of LIF treatment with additional biological processes including chemokine signaling, extracellular matrix (ECM) modification, and complement signaling (Fig. 2F, left; Supplementary Table S3). Interestingly, each one of these pathways has previously been linked with immunosuppressive macrophages (39–41). To understand the clinical relevance of the LIF-induced changes in macrophage gene expression to human cancer, we completed leading edge (LE) analysis to identify the most potently LIF-regulated genes associated with each of these processes (Fig. 2F, heatmaps; Supplementary Table S4). Across the pan TCGA cohort, each of the three LE gene signatures was significantly associated with poor overall patient survival, similar to that seen with LIF transcript expression (Supplementary Fig. S6A–S6C). Furthermore, LIF transcripts were also significantly correlated with the LE gene signatures across the 22 solid tumor indications (Supplementary Fig. S6D–S6F), suggesting that the LIF-induced patterns of gene expression observed in macrophages are recapitulated, at least in part, broadly across human tumors.

Identification and development of MSC-1, a high-affinity and potent LIF inhibitor

Although several previous reports have described important roles for LIF in cancer biology, high-quality LIF inhibitors with clinical potential have not been reported. Thus, we sought to evaluate a mAb that neutralized LIF signaling and was suitable for preclinical development and clinical testing. Briefly, this antibody was generated as follows: rats were immunized with human LIF to generate hybridomas, which were subsequently screened for LIF binding, function blocking capacity, and cross-reactivity to murine and cynomolgus LIF. MSC-1, a humanized mAb (IgG1) was developed through humanization of the lead rat antibody (IgG2a/κ). MSC-1 bound with high affinity to human (74 pmol/L) and mouse (27 pmol/L) LIF, as assessed by surface plasmon resonance (Fig. 3A), and potently inhibited LIF-induced pSTAT3 signaling as determined by electrochemilluminiscence assays (MSD) in U251 (Supplementary Fig. S7A) and HCC1954 cells in the subnanomolar range (Fig. 3B).

Figure 3.

MSC-1 is a potent inhibitor of LIF signaling through blockade of the LIF:gp130 interaction. A, Left, Surface plasmon resonance binding kinetics with fitted curves (Langmuir 1:1) for MSC-1 and recombinant human LIF. Right, Surface plasmon resonance binding kinetics with fitted curves (Langmuir 1:1) for MSC-1 and recombinant mouse LIF. B, Relative levels of pSTAT3 measured in MSC-1–treated HCC1954 tumor cells. C, Crystal structure of MSC1 (green) in complex with LIF (gray) with the MSC1 epitope colored in dark gray. D, Surface electrostatic rendering of the MSC1 paratope, highlighting the polar nature of the interaction between LIF and MSC1. E, Overlay of the MSC-1 (green): LIF crystal structure on the LIF (gray): gp130 (blue; PDB ID:1PVH; ref. 53), and LIF:LIFR (pink; PDB:2Q7N; ref. 54) crystal structures showing the overlapping interaction site between gp130 and MSC-1. F, Binding thermodynamics of LIF interactions with receptors. Left, gp130 binding to LIF in the presence of excess MSC-1. Right, LIFR binding to LIF in the presence of excess MSC-1. Top panels are raw data and bottom panels are binding enthalpies.

Figure 3.

MSC-1 is a potent inhibitor of LIF signaling through blockade of the LIF:gp130 interaction. A, Left, Surface plasmon resonance binding kinetics with fitted curves (Langmuir 1:1) for MSC-1 and recombinant human LIF. Right, Surface plasmon resonance binding kinetics with fitted curves (Langmuir 1:1) for MSC-1 and recombinant mouse LIF. B, Relative levels of pSTAT3 measured in MSC-1–treated HCC1954 tumor cells. C, Crystal structure of MSC1 (green) in complex with LIF (gray) with the MSC1 epitope colored in dark gray. D, Surface electrostatic rendering of the MSC1 paratope, highlighting the polar nature of the interaction between LIF and MSC1. E, Overlay of the MSC-1 (green): LIF crystal structure on the LIF (gray): gp130 (blue; PDB ID:1PVH; ref. 53), and LIF:LIFR (pink; PDB:2Q7N; ref. 54) crystal structures showing the overlapping interaction site between gp130 and MSC-1. F, Binding thermodynamics of LIF interactions with receptors. Left, gp130 binding to LIF in the presence of excess MSC-1. Right, LIFR binding to LIF in the presence of excess MSC-1. Top panels are raw data and bottom panels are binding enthalpies.

Close modal

To gain insight into the molecular mechanism by which MSC-1 inhibited LIF signaling, we determined the crystal structure of LIF in complex with the MSC-1 Fab at 3.1 Å resolution (Supplementary Table S5). MSC-1 interacts with LIF residues at its N-terminus, as well as on helices A and C thereby forming a discontinuous and conformational epitope (Fig. 3C). The electrostatic surface of the variable region of MSC-1 is predominantly electronegative and the charge complementarity between LIF and MSC-1 allows for strong polar interactions in addition to van der Waals contact (Fig. 3D; Supplementary Fig. S7B; Supplementary Table S6). The structure also revealed the stabilizing role of an unpaired, buried cysteine in the HCDR3 that makes several van der Waals interactions with the neighboring HCDR3 residues in the complex structure, as well as in five unliganded crystal structures of the MSC-1 Fab determined under different pH (5.6–8.5) conditions (Supplementary Fig. S7C–S7F; Supplementary Table S5). Point mutations at this position with all 19 other amino acids revealed that MSC1 with Cys94 was among the highest-affinity binders to LIF, observed from kinetics studies (Supplementary Fig. S7G). Thus, we conclude that despite Cys94 not being involved in direct interactions with LIF, it participates through its hydrophobic character in the efficient MCS1 engagement of LIF by conferring sufficient conformational plasticity to the antibody HCDR3. The MSC-1 paratope is thus largely preconfigured for binding LIF, which likely accounts for its high LIF affinity.

The MSC-1 epitope largely overlaps with the gp130 binding site but not with the LIFR binding site (Fig. 3E; Supplementary Fig. S7H), suggesting that MSC-1 acts as an inhibitor of LIF:gp130 binding. To corroborate these structural findings, we performed ITC experiments to test LIF:gp130 and LIF:LIFR binding in the presence of MSC-1. When complexed with MSC-1, LIF showed no capacity to bind gp130 but retained <5 nmol/L affinity for LIFR, supporting that MSC-1 inhibits LIF signaling through blockade of the LIF:gp130 interaction (Fig. 3F; Supplementary Fig. S7I and S7J) and interacts independent of LIFR binding. Overall, these data validate MSC-1 as a high-affinity and potent LIF inhibitor, which exerts its molecular mechanism of inhibition through blocking the LIF:gp130 interaction and is a suitable candidate for further development.

MSC-1 inhibits tumor growth in syngeneic mouse models of cancer

Previous studies have highlighted LIF as an attractive therapeutic target in cancer, although experiments employing high-quality therapeutic agents with clinical potential, such as MSC-1, have not been reported. To test our hypothesis that LIF drives tumor growth and promotes immunosuppression via modulation of TAM effects, we treated syngeneic mouse tumor models with MSC-1. We selected the CT26 and MC38 mouse colon cancer cell lines to generate tumor models as they expressed LIF protein in vitro and in tumors (Supplementary Fig. S8A and S8B), with higher LIF levels observed in CT26, and also represent established preclinical models for studying novel immune oncology therapies (42). Monotherapy MSC-1 treatment, administered at a dose of 15 mg/kg by intraperitoneal injection twice weekly, led to tumor growth inhibition in both models (Fig. 4A). CT26 tumors generally showed increased sensitivity to MSC-1 compared with MC38 tumors, suggesting that LIF protein levels may be a useful biomarker to determine responsiveness to MSC-1. MSC-1 treatment was well tolerated with no deleterious impact on body weight or overall health status (Supplementary Fig. S8C and S8D). Recombinant LIF and MSC-1 had no effect on CT26 and MC38 in vitro cell viability and proliferation, suggesting that MSC-1 activity is potentially driven through effects on the TME rather than by direct effects on tumor cells (Supplementary Fig. S8E–S8G).

Figure 4.

MSC-1 inhibits tumor growth and reprograms the TME to promote immune-stimulatory macrophages. A, Left, Tumor size in control IgG and MSC-1 monotherapy treated [15 mg/kg, twice per week (2QW)] mice bearing CT26 tumors. Right, tumor size in control IgG and MSC-1 monotherapy treated (15 mg/kg, 2QW) mice bearing MC38 tumors. B, Peripheral LIF levels in CT26 and MC38 tumor-bearing mice after treatment with either Control IgG1 or MSC-1. C, Representative images and quantification of Ki67, CC3, and pSTAT3 IHC stainings in CT26 mouse models treated with 15 mg/kg MSC-1 or control IgG. D, Left, Representative flow cytometric analysis of TAMs (CD11b+ Ly6Clow F4/80+) in CT26 tumors at endpoint (14 days after treatment initiation) and expression of MHC II [M1 marker (I-A/I-E)] and CD206 (M2 marker) in TAMs (IgG1 control group n = 9; MSC-1 group n = 8). Right, Quantification of pan myeloid cell population (CD11b+ in CD45+ cells), TAM (CD11b+ Ly6Clow F4/80+ in CD45+ cells), and ratio of M1/M2 in TAM population. Heatmap shows percent changes in M1 (MHC II+ CD206 TAMs) and M2 (MHC II CD206+ TAMs) cell populations in MSC-1–treated tumors compared with IgG1 control-treated tumors (n = 8). E, Transcript level of M1 and M2 genes in CT26 tumors at endpoint determined by qRT-PCR of whole tumor lysate (n = 8). Data are represented as the fold change relative to mean of IgG1 control-treated tumors (Hprt, Tbp, Tfrc). F, Schematic representation of tumor-derived macrophages isolated from CT26 tumor-bearing mice and cocultured with PBMCs isolated from non-tumor mice. G, Tumor-derived macrophage regulation of T-cell activation and proliferation, analyzed by percentage of CD25 expression on CD8 T-cell population and CSFE trace, respectively. Representative flow cytometry plots are shown. H, IFNγ and CXCL9 secretion of tumor-derived macrophages and PBMCs coculture, analyzed by ELISA. Data are presented as mean ± SEM. *, P < 0.07; **, P < 0.01.

Figure 4.

MSC-1 inhibits tumor growth and reprograms the TME to promote immune-stimulatory macrophages. A, Left, Tumor size in control IgG and MSC-1 monotherapy treated [15 mg/kg, twice per week (2QW)] mice bearing CT26 tumors. Right, tumor size in control IgG and MSC-1 monotherapy treated (15 mg/kg, 2QW) mice bearing MC38 tumors. B, Peripheral LIF levels in CT26 and MC38 tumor-bearing mice after treatment with either Control IgG1 or MSC-1. C, Representative images and quantification of Ki67, CC3, and pSTAT3 IHC stainings in CT26 mouse models treated with 15 mg/kg MSC-1 or control IgG. D, Left, Representative flow cytometric analysis of TAMs (CD11b+ Ly6Clow F4/80+) in CT26 tumors at endpoint (14 days after treatment initiation) and expression of MHC II [M1 marker (I-A/I-E)] and CD206 (M2 marker) in TAMs (IgG1 control group n = 9; MSC-1 group n = 8). Right, Quantification of pan myeloid cell population (CD11b+ in CD45+ cells), TAM (CD11b+ Ly6Clow F4/80+ in CD45+ cells), and ratio of M1/M2 in TAM population. Heatmap shows percent changes in M1 (MHC II+ CD206 TAMs) and M2 (MHC II CD206+ TAMs) cell populations in MSC-1–treated tumors compared with IgG1 control-treated tumors (n = 8). E, Transcript level of M1 and M2 genes in CT26 tumors at endpoint determined by qRT-PCR of whole tumor lysate (n = 8). Data are represented as the fold change relative to mean of IgG1 control-treated tumors (Hprt, Tbp, Tfrc). F, Schematic representation of tumor-derived macrophages isolated from CT26 tumor-bearing mice and cocultured with PBMCs isolated from non-tumor mice. G, Tumor-derived macrophage regulation of T-cell activation and proliferation, analyzed by percentage of CD25 expression on CD8 T-cell population and CSFE trace, respectively. Representative flow cytometry plots are shown. H, IFNγ and CXCL9 secretion of tumor-derived macrophages and PBMCs coculture, analyzed by ELISA. Data are presented as mean ± SEM. *, P < 0.07; **, P < 0.01.

Close modal

Cytokine stabilization has been used successfully as a biomarker for target engagement for anticytokine antibody therapies (43). Such assays take advantage of the relatively short half-lives of cytokines and the capacity to measure the accumulation of the cytokine antibody complex after dosing as an indication of target binding and saturation. Accordingly, we tested the capacity of MSC-1 to stabilize peripheral LIF and found that whereas circulating LIF levels were undetectable in control IgG1-treated mice, high LIF levels were detected in the plasma of MSC-1–treated mice both in tumor-bearing and non–tumor-bearing mice, confirming MSC-1–LIF engagement in vivo (Fig. 4B; Supplementary Fig. S8H).

We decided to analyze the tumors from the MSC-1–treated and untreated CT26 models. MSC-1 induced a marked decrease in pSTAT3 levels showing that in this model LIF was the main cytokine inducing the JAK-STAT3 pathway (Fig. 4C). Moreover, no major effect of MSC-1 on proliferation (measured by Ki67 levels) was observed, although MSC-1 promoted tumor cell apoptosis as measured by CC3 levels (Fig. 4C). This indicated that the antitumor response to MSC-1 was not mediated by an effect in cell proliferation and suggested that it could be the result of an immune response.

To examine whether MSC-1 impacted infiltrating myeloid cells in treated tumors, we analyzed total myeloid and macrophage frequency as well as macrophage phenotypes in MSC-1–treated CT26 and MC38 tumors. Within the CT26 model, MSC-1 treatment led to no consistent differences in the overall frequency of total myeloid (CD45+/CD11b+) or total TAM populations (CD45+/CD11b+/Ly6Clow/F4/80+; Fig. 4D, top). However, MSC-1 treatment reprogrammed macrophages from an immunosuppressive protumoral M2-like state (CD206high/MHC-IIlow) to an immunostimulatory antitumoral M1-like state (CD206low/MHC-IIhigh) and drove a significant increase in the ratio of M1-like to M2-like TAMs (Fig. 4D, bottom). Hence, MSC-1 treatment shifted the overall distribution of TAM phenotypes to favor immunostimulatory over immunosuppressive macrophage phenotypes. We further tested for MSC-1–induced macrophage reprogramming using transcriptional analysis in treated tumors. Similar to our observation in LIF-treated human macrophages, MSC-1 treatment induced downregulation of the M2 markers Cd206 and Cd163 in CT26 tumors (Fig. 4E, top). Moreover, MSC-1 treatment induced increases in expression of immune-stimulatory M1 markers including Cxcl9, Cxcl10, and Pd-l1 (Fig. 4E, bottom).

To assess the functional effect of MSC-1 on TAMs in an in vivo setting, we isolated TAMs from treated and untreated CT26 tumors and cocultured them in the presence of mouse leukocytes (Fig. 4F). We observed that TAMs obtained from MSC-1–treated mice induced CD8+ T-cell activation (measured by CD25 expression) and proliferation (Fig. 4G). Moreover, higher levels of IFNγ and CXCL9 were present in cocultures of TAMs obtained from MSC-1–treated mice (Fig. 4H). These results showed that the MSC-1–mediated blockade of LIF in vivo induced the activation and proliferation of CD8 T cells demonstrating the functional effect of MSC-1 on TAMs.

We extended these findings by also examining TAM phenotypes in MSC-1–treated MC38 tumors. Treated tumors showed no difference in the overall frequency of total myeloid or TAM populations; however, MSC-1 induced an increase in both the proportion of TAMs expressing MHC-II as well as the overall expression level of MHC-II in the TAM compartment, consistent with M1-like skewing (Supplementary Fig. S8I).

Together, these data demonstrate that MSC-1 inhibits tumor growth in two independent preclinical tumor models. Our analysis demonstrates that MSC-1 impacted TAM phenotypes, but not overall numbers of TAMs or total myeloid cells, suggesting that the observed efficacy occurs, at least in part, through the reprogramming of TAMs to favor antitumor immunity.

MSC-1 and anti-PD1 show combinatorial efficacy in CT26 and MC38

Recent studies have described clear roles for TAMs limiting effective antitumor immunity and several myeloid targeting therapies have shown promising combinatorial effects with checkpoint blockade (38, 44). We therefore tested MSC-1 in combination with anti-PD1 (RMP1-14) in both the CT26 and MC38 models. Whereas we did not observe durable survival benefit with MSC-1 monotherapy (Supplementary Fig. S9A and S9B) and only rarely with anti-PD1 therapy (Fig. 5A), combination of MSC-1 and anti-PD1 resulted in long-term survival benefit in approximately 40% and 30% of treated CT26 and MC38 tumor-bearing mice, respectively (Fig. 5A). The combination therapy was well tolerated, with no overall effect observed on body weights (Supplementary Fig. S9B). Importantly, long-term tumor-free CT26 survivors were resistant to tumor reimplantation, consistent with acquisition of long-lasting adaptive immunologic memory (Supplementary Fig. S9C).

Figure 5.

MSC-1 and anti-PD1 combination provide durable survival benefit and promote effective antitumor immunity. A, Left, Kaplan–Meier survival plots for anti-PD1 (RMP1-14) and MSC-1/anti-PD1–treated mice bearing CT26 (anti-PD1 started on day 10) tumors. Right, Kaplan–Meier survival plots for anti-PD1 (RMP1–14) and MSC-1/anti-PD1–treated mice bearing MC38 (anti-PD1 started on day 7) tumors. B, Left, Frequency of CD8+ TIL of total CD45+ immune infiltrate in CT26 tumors 17 days after treatment initiation (n = 7/group). Right, Frequency of IFNγ+ CD8+ TIL following ex vivo stimulation with peptide corresponding to the H2-Ld–restricted gp70 (a.a.423–431; AH1) epitope from MuLV expressed by CT26 tumors (n = 7/group). C, Left, Frequency of CD8+ TIL of total CD45+ immune infiltrate in MC38 tumors (17 days after treatment initiation; n = 6–7/group). Right, Frequency of IFNγ+ CD8+ TIL following ex vivo stimulation with peptide corresponding to the H2-Kb–restricted p15e (a.a.604–611) epitope from the MuLV expressed by MC38 tumors (n = 6–7/group). *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 5.

MSC-1 and anti-PD1 combination provide durable survival benefit and promote effective antitumor immunity. A, Left, Kaplan–Meier survival plots for anti-PD1 (RMP1-14) and MSC-1/anti-PD1–treated mice bearing CT26 (anti-PD1 started on day 10) tumors. Right, Kaplan–Meier survival plots for anti-PD1 (RMP1–14) and MSC-1/anti-PD1–treated mice bearing MC38 (anti-PD1 started on day 7) tumors. B, Left, Frequency of CD8+ TIL of total CD45+ immune infiltrate in CT26 tumors 17 days after treatment initiation (n = 7/group). Right, Frequency of IFNγ+ CD8+ TIL following ex vivo stimulation with peptide corresponding to the H2-Ld–restricted gp70 (a.a.423–431; AH1) epitope from MuLV expressed by CT26 tumors (n = 7/group). C, Left, Frequency of CD8+ TIL of total CD45+ immune infiltrate in MC38 tumors (17 days after treatment initiation; n = 6–7/group). Right, Frequency of IFNγ+ CD8+ TIL following ex vivo stimulation with peptide corresponding to the H2-Kb–restricted p15e (a.a.604–611) epitope from the MuLV expressed by MC38 tumors (n = 6–7/group). *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Close modal

Given that the therapeutic effects of checkpoint blockade are largely dependent on T cells, we examined TIL across the treatment groups. We did not observe differences in CD4 TIL frequencies in either model across the monotherapy and combination therapy treatment groups (Supplementary Fig. S9D). In contrast, combination therapy significantly increased CD8 TIL frequency in both CT26 and MC38 tumors, with only nominal effects observed with either monotherapy treatment (Fig. 5B and C, left). Overall, this suggests that the increased efficacy observed with combination therapy was potentially driven by increased CD8 TIL.

To examine whether CD8 TIL were functionally different on a per cell basis in tumors harvested from mice treated with either anti-PD1 or the MSC-1 and anti-PD1 combination, we examined their capacity to produce IFNγ in response to direct ex vivo stimulation with peptides corresponding to either CT26 or MC38 CD8 T-cell epitopes. CD8 TIL from anti-PD1 or MSC-1 and anti-PD1 combination-treated CT26 and MC38 tumors showed no difference in function when stimulated in this fashion (Fig. 5B and C, right). Taken together, our data support that the observed combination efficacy is driven by improved antitumor immunity mediated by CD8 TILs.

Because its original cloning and description as an inhibitor of proliferation in murine leukemia cells (36, 45), LIF has emerged as a multifunctional cytokine regulating both physiologic and pathophysiologic processes (2). In solid tumors, numerous and diverse functions have been ascribed to LIF, which together substantiate LIF as a promoting factor of tumor development and progression. We performed a comprehensive assessment of LIF expression across large numbers of human tumors that span multiple cancer indications. We took advantage of the publicly available TCGA resource as well as GSEA to broadly and deeply probe the mechanisms associated with LIF in human cancer. We observed that LIF transcript was consistently coexpressed with numerous gene expression signatures representing immunosuppressive TAMs. In contrast, strong coexpression was not observed for signatures representing immunostimulatory macrophages, or other common tumor-infiltrating immune cell types, demonstrating that the relationship between LIF transcript and suppressive TAMs is both specific and robust. We also found that within human tumors TAMs are uniquely poised to respond to LIF, as across tumor cells TIL and TAMs, only TAMs expressed detectable levels of LIFR. Hence, through this initially broad and unbiased approach, we concluded that the role of LIF in tumor biology, occurs at least in part, through the regulation of TAMs, which served as the focus for our subsequent experiments. Importantly, these results are consistent with a recent report describing an association of LIF expression with a TAM signature across multiple solid tumor indications as well as a role for LIF-regulating CXCL9 expression in TAMs (15).

To understand the role of LIF in macrophage biology, we treated human monocyte-derived macrophages with LIF and assessed macrophage phenotype, function, and changes in global gene expression. While the initial starting monocyte populations did not express detectable LIFR, it was strongly induced during macrophage differentiation, suggesting that high levels of surface expressed LIFR is acquired during macrophage maturation. Macrophages treated with LIF showed upregulation of common immune-suppressive markers and demonstrated increased capacity to inhibit T cells. RNA-seq analysis identified additional pathways stimulated by LIF including chemokine signaling, ECM matrix organization, and complement signaling. Gene signatures representing these pathways were significantly correlated with LIF transcript across TCGA dataset and predicted poor patient outcome, consistent with these pathways being mechanistically linked with LIF and aggressive disease biology in human tumors. Notably, chemokine signaling (46), ECM reorganization (41), and completement signaling (40), have all been described as biological processes subverted by TAMs in the TME to prevent effective antitumor immunity. These data suggested that LIF may contribute to the suppressive nature of TAMs, and drive tumor growth through regulation of these processes.

Given these data and the substantial body of scientific literature supporting LIF as a tumor-promoting cytokine, we developed and assessed a monoclonal therapeutic antibody targeting LIF named MSC-1. MSC-1 is a high affinity (<100 pmol/L) potent inhibitor of LIF signaling, which we determined to inhibit LIF signaling through blockade of the LIF:gp130 interaction. MSC-1 treatment showed tumor growth inhibition in CT26 and MC38 syngeneic tumor models and stabilized circulating LIF levels. Hence, peripheral LIF stabilization may represent a useful target engagement biomarker to monitor target saturation in clinical studies with MSC-1. Monotherapy efficacy effects were stronger in the CT26 model compared with MC38. CT26 tumors expressed higher levels of LIF and comprised a higher frequency of infiltrating TAMs, supporting the hypothesis that either tumor LIF levels or TAM infiltration, or both, may be associated with increased responsiveness to MSC-1. Careful analysis of the TME in tumors harvested from MSC-1–treated mice revealed a change in the phenotype and function of the TAM compartment to favor an immunostimulatory status. However, no effects were observed on the overall abundance of either total myeloid cells or TAMs. This is in contrast to other TAM-targeting approaches that reduce or deplete macrophages from tumors. Importantly, this could represent a significant advantage for MSC-1 over TAM-depleting therapies, as reports indicate immunostimulatory macrophages are important contributors to tumor rejection (47, 48).

Our observations that a significantly increased proportion of MSC-1 and anti-PD1 combination-treated mice experienced durable tumor-free survival is consistent with reports that TAMs can impede tumor responsiveness to checkpoint inhibitors (38, 49). Analysis of the TME, using flow cytometry revealed that combination treatment expanded the CD8 TIL compartment, including increased numbers of cytolytic, proliferative, and antigen-experienced CD8 TIL, suggesting the overall efficacy was dependent on CD8 TIL. The relevance of these effects to human patients is bolstered by our observation that LIF expression levels predict response to anti-PDL1 in metastatic bladder cancer, where response is linked to the strength of preexisting CD8 T-cell immunity (50).

Our data demonstrate a role of LIF in regulating the biology of tumor-infiltrating myeloid cells. However, this study does not preclude LIF contributing to other aspects of tumorigenesis. For example, recent reports describe roles for LIF at generating inflammatory CAFs in pancreatic tumors which drive progression (51) and/or chemotherapy resistance through effects on tumor cell differentiation (18, 19). This is particularly intriguing given that across the 22 solid tumor indications profiled by TCGA and analyzed in this study, pancreatic tumors expressed the highest levels of LIF. LIF is also reported to have diverse functions across a variety of cell types known to be important for tumor growth, including tumor cells (17), cancer stem cells (14, 18, 19), CAFs (23), myeloid cells (24), and regulatory T cells (52). Hence it is clear that the role of LIF in cancer is both multifactorial and complex, and likely extends beyond effects on TAMs and the antitumor immune response.

Overall, there is a compelling rationale to develop and test LIF inhibitors for cancer therapy. To this end, we recently completed a phase I dose escalation to carefully evaluate the safety, mechanisms of action, and preliminary antitumor activity of MSC-1 (NCT03490669). Characterization of samples collected as part of this trial will determine the effect of MSC-1 on TAMs and CD8 T cells and expand understanding of the human mechanism of action for MSC-1.

R.M. Hallett reports being an employee of Northern Biologics during the conduct of the study; in addition, R.M. Hallett has a patent for Use of MSC-1 for the treatment of cancer issued. S. Raman reports a patent for US11390670B2 issued. O. Egorova reports personal fees from Northern Biologics during the conduct of the study, as well as personal fees from Northern Biologics outside the submitted work. A.J.R. McGray reports being an employee of Northern Biologics at the time these studies were completed and is a shareholder in the Northern Mosaic Limited Partnership. E. Lau reports other support from Northern Biologics Inc. during the conduct of the study. D. Maetzel reports personal fees from Northern Biologics outside the submitted work. J. Fransson reports a patent for Antibodies against LIF issued and a patent for Combination of LIF and PD-1 Axis Inhibitors for Use in Treating Cancer issued. I. Huber-Ruano reports a patent for WO2019243898 pending and a patent for WO2017089614 pending. J. Anido reports a patent for WO/2011/124566 pending, a patent for WO/2019/243898 pending, a patent for WO/2019/197903 pending, a patent for WO/2019/220204 pending, a patent for WO/2019/243900 pending, and a patent for WO/2018/115960 pending. J.-P. Julien reports grants from MITACS during the conduct of the study; in addition, J.-P. Julien has a patent for 15/880,906 issued. J. Seoane reports grants from Northern Biologics and Mosaic Biologics during the conduct of the study, as well as grants from Roche/Glycart and F. Hoffmann-La Roche outside the submitted work. In addition, J. Seoane has a patent for LIF-related patents licensed to Mosaic Biomedicals, a patent for WO/2011/124566 pending, a patent for WO/2019/243898 pending, a patent for WO/2019/197903 pending, a patent for WO/2019/220204 pending, a patent for WO/2019/243900 pending, and a patent for WO/2018/115960 pending. No disclosures were reported by the other authors.

R.M. Hallett: Conceptualization, formal analysis, supervision, writing–original draft, writing–review and editing. E. Bonfill-Teixidor: Methodology. R. Iurlaro: Methodology. A. Arias: Methodology. S. Raman: Methodology, writing–original draft. P. Bayliss: Methodology. O. Egorova: Methodology, writing–original draft. A. Neva-Alejo: Methodology. A.R. McGray: Methodology, writing–original draft. E. Lau: Methodology, writing–original draft. A. Bosch: Methodology. M. Beilschmidt: Methodology. D. Maetzel: Methodology. J. Fransson: Conceptualization. I. Huber-Ruano: Conceptualization. J. Anido: Conceptualization. J.-P. Julien: Conceptualization. P. Giblin: Conceptualization. J. Seoane: Conceptualization supervision, writing–review and editing.

The study was undertaken with the support of the Fundación Asociación Española contra el Cáncer (AECC; GCTRA16015SEOA), Ramón Areces Foundation (CIVP19A5943), BBVA (CAIMI), the ISCIII-FIS (PI19/00318). Part of this work was funded by Mitacs Accelerate grant no. IT06689 (J.-P. Julien and J. Fransson). This work was undertaken, in part, thanks to funding from the CIFAR Azrieli Global Scholar program (J.-P. Jullien) and the Canada Research Chairs program (950-231604; J.-P. Julien). X-ray diffraction experiments were performed using beamline 08ID-1 at the Canadian Light Source, which is supported by the Canada Foundation for Innovation, Natural Sciences and Engineering Research Council of Canada, the University of Saskatchewan, the Government of Saskatchewan, Western Economic Diversification Canada, the National Research Council Canada, and the Canadian Institutes of Health Research (CIHR). X-ray diffraction experiments were also performed at GM/CA @ APS, which has been funded in whole or in part with Federal funds from the NCI (ACB-12002) and the National Institute of General Medical Sciences (AGM-12006). The Eiger 16M detector was funded by an NIH–Office of Research Infrastructure Programs, High-End Instrumentation Grant (1S10OD012289-01A1). This research used resources of the Advanced Photon Source, a U.S. Department of Energy (DOE) Office of Science User Facility operated for the DOE Office of Science by Argonne National Laboratory under contract no. DE-AC02-06CH11357. The authors are grateful to the Structural & Biophysical Core Facility (The Hospital for Sick Children) for access to the ITC instrument. Mass cytometry was performed in The SickKids-UHN Flow and Mass Cytometry Facility, with funding from the Ontario Institute for Cancer Research, McEwen Centre for Regenerative Medicine and the Canada Foundation for Innovation. The authors would like to thank Ms. Asna Choudhry for her assistance with completing the flow cytometry assessment of MC38 tumors.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.

Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).

1.
Nicola
NA
,
Babon
JJ
.
Leukemia inhibitory factor (LIF)
.
Cytokine Growth Factor Rev
2015
;
26
:
533
44
.
2.
Yue
X
,
Wu
L
,
Hu
W
.
The regulation of leukemia inhibitory factor
.
Cancer Cell Microenviron
2015
;
2
:
e877
.
3.
Taupin
JL
,
Pitard
V
,
Dechanet
J
,
Miossec
V
,
Gualde
N
,
Moreau
JF
, et al
.
Leukemia inhibitory factor: part of a large ingathering family
.
Int Rev Immunol
1998
;
16
:
397
426
.
4.
Onishi
K
,
Zandstra
PW
.
LIF signaling in stem cells and development
.
Development
2015
;
142
:
2230
6
.
5.
Stewart
CL
,
Kaspar
P
,
Brunet
LJ
,
Bhatt
H
,
Gadi
I
,
Kontgen
F
, et al
.
Blastocyst implantation depends on maternal expression of leukaemia inhibitory factor
.
Nature
1992
;
359
:
76
9
.
6.
Williams
RL
,
Hilton
DJ
,
Pease
S
,
Willson
TA
,
Stewart
CL
,
Gearing
DP
, et al
.
Myeloid leukaemia inhibitory factor maintains the developmental potential of embryonic stem cells
.
Nature
1988
;
336
:
684
7
.
7.
Metcalf
D
,
Gearing
DP
.
Fatal syndrome in mice engrafted with cells producing high levels of the leukemia inhibitory factor
.
Proc Natl Acad Sci U S A
1989
;
86
:
5948
52
.
8.
Chesnokova
V
,
Auernhammer
CJ
,
Melmed
S
.
Murine leukemia inhibitory factor gene disruption attenuates the hypothalamo-pituitary-adrenal axis stress response
.
Endocrinology
1998
;
139
:
2209
16
.
9.
Rao
MS
,
Sun
Y
,
Escary
JL
,
Perreau
J
,
Tresser
S
,
Patterson
PH
, et al
.
Leukemia inhibitory factor mediates an injury response but not a target-directed developmental transmitter switch in sympathetic neurons
.
Neuron
1993
;
11
:
1175
85
.
10.
Cheng
JG
,
Chen
JR
,
Hernandez
L
,
Alvord
WG
,
Stewart
CL
.
Dual control of LIF expression and LIF receptor function regulate Stat3 activation at the onset of uterine receptivity and embryo implantation
.
Proc Natl Acad Sci U S A
2001
;
98
:
8680
5
.
11.
Chen
JR
,
Cheng
JG
,
Shatzer
T
,
Sewell
L
,
Hernandez
L
,
Stewart
CL
, et al
.
Leukemia inhibitory factor can substitute for nidatory estrogen and is essential to inducing a receptive uterus for implantation but is not essential for subsequent embryogenesis
.
Endocrinology
2000
;
141
:
4365
72
.
12.
Rosario
GX
,
Hondo
E
,
Jeong
JW
,
Mutalif
R
,
Ye
X
,
Yee
LX
, et al
.
The LIF-mediated molecular signature regulating murine embryo implantation
.
Biol Reprod
2014
;
91
:
66
.
13.
Schofield
G
,
Kimber
SJ
.
Leukocyte subpopulations in the uteri of leukemia inhibitory factor knockout mice during early pregnancy
.
Biol Reprod
2005
;
72
:
872
8
.
14.
Penuelas
S
,
Anido
J
,
Prieto-Sanchez
RM
,
Folch
G
,
Barba
I
,
Cuartas
I
, et al
.
TGF-beta increases glioma-initiating cell self-renewal through the induction of LIF in human glioblastoma
.
Cancer Cell
2009
;
15
:
315
27
.
15.
Pascual-Garcia
M
,
Bonfill-Teixidor
E
,
Planas-Rigol
E
,
Rubio-Perez
C
,
Iurlaro
R
,
Arias
A
, et al
.
LIF regulates CXCL9 in tumor-associated macrophages and prevents CD8(+) T cell tumor-infiltration impairing anti-PD1 therapy
.
Nat Commun
2019
;
10
:
2416
.
16.
Wysoczynski
M
,
Miekus
K
,
Jankowski
K
,
Wanzeck
J
,
Bertolone
S
,
Janowska-Wieczorek
A
, et al
.
Leukemia inhibitory factor: a newly identified metastatic factor in rhabdomyosarcomas
.
Cancer Res
2007
;
67
:
2131
40
.
17.
Kamohara
H
,
Ogawa
M
,
Ishiko
T
,
Sakamoto
K
,
Baba
H
.
Leukemia inhibitory factor functions as a growth factor in pancreas carcinoma cells: involvement of regulation of LIF and its receptor expression
.
Int J Oncol
2007
;
30
:
977
83
.
18.
Shi
Y
,
Gao
W
,
Lytle
NK
,
Huang
P
,
Yuan
X
,
Dann
AM
, et al
.
Targeting LIF-mediated paracrine interaction for pancreatic cancer therapy and monitoring
.
Nature
2019
;
569
:
131
5
.
19.
Wang
MT
,
Fer
N
,
Galeas
J
,
Collisson
EA
,
Kim
SE
,
Sharib
J
, et al
.
Blockade of leukemia inhibitory factor as a therapeutic approach to KRAS driven pancreatic cancer
.
Nat Commun
2019
;
10
:
3055
.
20.
Liu
SC
,
Tsang
NM
,
Chiang
WC
,
Chang
KP
,
Hsueh
C
,
Liang
Y
, et al
.
Leukemia inhibitory factor promotes nasopharyngeal carcinoma progression and radioresistance
.
J Clin Invest
2013
;
123
:
5269
83
.
21.
Li
X
,
Yang
Q
,
Yu
H
,
Wu
L
,
Zhao
Y
,
Zhang
C
, et al
.
LIF promotes tumorigenesis and metastasis of breast cancer through the AKT-mTOR pathway
.
Oncotarget
2014
;
5
:
788
801
.
22.
Yu
H
,
Yue
X
,
Zhao
Y
,
Li
X
,
Wu
L
,
Zhang
C
, et al
.
LIF negatively regulates tumour-suppressor p53 through Stat3/ID1/MDM2 in colorectal cancers
.
Nat Commun
2014
;
5
:
5218
.
23.
Albrengues
J
,
Bourget
I
,
Pons
C
,
Butet
V
,
Hofman
P
,
Tartare-Deckert
S
, et al
.
LIF mediates proinvasive activation of stromal fibroblasts in cancer
.
Cell Rep
2014
;
7
:
1664
78
.
24.
Duluc
D
,
Delneste
Y
,
Tan
F
,
Moles
MP
,
Grimaud
L
,
Lenoir
J
, et al
.
Tumor-associated leukemia inhibitory factor and IL-6 skew monocyte differentiation into tumor-associated macrophage-like cells
.
Blood
2007
;
110
:
4319
30
.
25.
Wan
YW
,
Allen
GI
,
Liu
Z
.
TCGA2STAT: simple TCGA data access for integrated statistical analysis in R
.
Bioinformatics
2016
;
32
:
952
4
.
26.
Hallett
RM
,
Dvorkin-Gheva
A
,
Bane
A
,
Hassell
JA
.
A gene signature for predicting outcome in patients with basal-like breast cancer
.
Sci Rep
2012
;
2
:
227
.
27.
Pertea
M
,
Kim
D
,
Pertea
GM
,
Leek
JT
,
Salzberg
SL
.
Transcript-level expression analysis of RNA-seq experiments with HISAT, StringTie and Ballgown
.
Nat Protoc
2016
;
11
:
1650
67
.
28.
Subramanian
A
,
Tamayo
P
,
Mootha
VK
,
Mukherjee
S
,
Ebert
BL
,
Gillette
MA
, et al
.
Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles
.
Proc Natl Acad Sci U S A
2005
;
102
:
15545
50
.
29.
Shannon
P
,
Markiel
A
,
Ozier
O
,
Baliga
NS
,
Wang
JT
,
Ramage
D
, et al
.
Cytoscape: a software environment for integrated models of biomolecular interaction networks
.
Genome Res
2003
;
13
:
2498
504
.
30.
Aricescu
AR
,
Lu
W
,
Jones
EY
.
A time- and cost-efficient system for high-level protein production in mammalian cells
.
Acta Crystallogr D Biol Crystallogr
2006
;
62
:
1243
50
.
31.
Kabsch
W
.
XDS
.
Acta Crystallogr D Biol Crystallogr
2010
;
66
:
125
32
.
32.
McCoy
AJ
,
Grosse-Kunstleve
RW
,
Adams
PD
,
Winn
MD
,
Storoni
LC
,
Read
RJ
, et al
.
Phaser crystallographic software
.
J Appl Crystallogr
2007
;
40
:
658
74
.
33.
Emsley
P
,
Lohkamp
B
,
Scott
WG
,
Cowtan
K
.
Features and development of coot
.
Acta Crystallogr D Biol Crystallogr
2010
;
66
:
486
501
.
34.
Adams
PD
,
Afonine
PV
,
Bunkoczi
G
,
Chen
VB
,
Davis
IW
,
Echols
N
, et al
.
PHENIX: a comprehensive Python-based system for macromolecular structure solution
.
Acta Crystallogr D Biol Crystallogr
2010
;
66
:
213
21
.
35.
Morin
A
,
Eisenbraun
B
,
Key
J
,
Sanschagrin
PC
,
Timony
MA
,
Ottaviano
M
, et al
.
Collaboration gets the most out of software
.
Elife
2013
;
2
:
e01456
.
36.
Gough
NM
,
Gearing
DP
,
King
JA
,
Willson
TA
,
Hilton
DJ
,
Nicola
NA
, et al
.
Molecular cloning and expression of the human homologue of the murine gene encoding myeloid leukemia-inhibitory factor
.
Proc Natl Acad Sci U S A
1988
;
85
:
2623
7
.
37.
Mariathasan
S
,
Turley
SJ
,
Nickles
D
,
Castiglioni
A
,
Yuen
K
,
Wang
Y
, et al
.
TGFbeta attenuates tumour response to PD-L1 blockade by contributing to exclusion of T cells
.
Nature
2018
;
554
:
544
8
.
38.
De Henau
O
,
Rausch
M
,
Winkler
D
,
Campesato
LF
,
Liu
C
,
Cymerman
DH
, et al
.
Overcoming resistance to checkpoint blockade therapy by targeting PI3Kgamma in myeloid cells
.
Nature
2016
;
539
:
443
7
.
39.
Kitamura
T
,
Qian
BZ
,
Soong
D
,
Cassetta
L
,
Noy
R
,
Sugano
G
, et al
.
CCL2-induced chemokine cascade promotes breast cancer metastasis by enhancing retention of metastasis-associated macrophages
.
J Exp Med
2015
;
212
:
1043
59
.
40.
Medler
TR
,
Murugan
D
,
Horton
W
,
Kumar
S
,
Cotechini
T
,
Forsyth
AM
, et al
.
Complement C5a fosters squamous carcinogenesis and limits T cell response to chemotherapy
.
Cancer Cell
2018
;
34
:
561
78
.
41.
Zhu
Y
,
Herndon
JM
,
Sojka
DK
,
Kim
KW
,
Knolhoff
BL
,
Zuo
C
, et al
.
Tissue-resident macrophages in pancreatic ductal adenocarcinoma originate from embryonic hematopoiesis and promote tumor progression
.
Immunity
2017
;
47
:
323
38
.
42.
Mosely
SI
,
Prime
JE
,
Sainson
RC
,
Koopmann
JO
,
Wang
DY
,
Greenawalt
DM
, et al
.
Rational selection of syngeneic preclinical tumor models for immunotherapeutic drug discovery
.
Cancer Immunol Res
2017
;
5
:
29
41
.
43.
Roskos
LK
,
Schneider
A
,
Vainshtein
I
,
Schwickart
M
,
Lee
R
,
Lu
H
, et al
.
PK-PD modeling of protein drugs: implications in assay development
.
Bioanalysis
2011
;
3
:
659
75
.
44.
Zhu
Y
,
Knolhoff
BL
,
Meyer
MA
,
Nywening
TM
,
West
BL
,
Luo
J
, et al
.
CSF1/CSF1R blockade reprograms tumor-infiltrating macrophages and improves response to T-cell checkpoint immunotherapy in pancreatic cancer models
.
Cancer Res
2014
;
74
:
5057
69
.
45.
Gearing
DP
,
Gough
NM
,
King
JA
,
Hilton
DJ
,
Nicola
NA
,
Simpson
RJ
, et al
.
Molecular cloning and expression of cDNA encoding a murine myeloid leukaemia inhibitory factor (LIF)
.
EMBO J
1987
;
6
:
3995
4002
.
46.
Nagarsheth
N
,
Wicha
MS
,
Zou
W
.
Chemokines in the cancer microenvironment and their relevance in cancer immunotherapy
.
Nat Rev Immunol
2017
;
17
:
559
72
.
47.
Gubin
MM
,
Esaulova
E
,
Ward
JP
,
Malkova
ON
,
Runci
D
,
Wong
P
, et al
.
High-dimensional analysis delineates myeloid and lymphoid compartment remodeling during successful immune-checkpoint cancer therapy
.
Cell
2018
;
175
:
1014
30
.
48.
Gordon
SR
,
Maute
RL
,
Dulken
BW
,
Hutter
G
,
George
BM
,
McCracken
MN
, et al
.
PD-1 expression by tumour-associated macrophages inhibits phagocytosis and tumour immunity
.
Nature
2017
;
545
:
495
9
.
49.
Binnewies
M
,
Roberts
EW
,
Kersten
K
,
Chan
V
,
Fearon
DF
,
Merad
M
, et al
.
Understanding the tumor immune microenvironment (TIME) for effective therapy
.
Nat Med
2018
;
24
:
541
50
.
50.
Ayers
M
,
Lunceford
J
,
Nebozhyn
M
,
Murphy
E
,
Loboda
A
,
Kaufman
DR
, et al
.
IFN-gamma-related mRNA profile predicts clinical response to PD-1 blockade
.
J Clin Invest
2017
;
127
:
2930
40
.
51.
Biffi
G
,
Oni
TE
,
Spielman
B
,
Hao
Y
,
Elyada
E
,
Park
Y
, et al
.
IL1-induced JAK/STAT signaling is antagonized by TGFbeta to shape CAF heterogeneity in pancreatic ductal adenocarcinoma
.
Cancer Discov
2018
;
9
:
282
301
.
52.
Gao
W
,
Thompson
L
,
Zhou
Q
,
Putheti
P
,
Fahmy
TM
,
Strom
TB
, et al
.
Treg versus Th17 lymphocyte lineages are cross-regulated by LIF versus IL-6
.
Cell Cycle
2009
;
8
:
1444
50
.
53.
Boulanger
MJ
,
Bankovich
AJ
,
Kortemme
T
,
Baker
D
,
Garcia
KC
.
Convergent mechanisms for recognition of divergent cytokines by the shared signaling receptor gp130
.
Mol Cell
2003
;
12
:
577
89
.
54.
Huyton
T
,
Zhang
JG
,
Luo
CS
,
Lou
MZ
,
Hilton
DJ
,
Nicola
NA
, et al
.
An unusual cytokine:Ig-domain interaction revealed in the crystal structure of leukemia inhibitory factor (LIF) in complex with the LIF receptor
.
Proc Natl Acad Sci U S A
2007
;
104
:
12737
42
.

Supplementary data