Pharmacological inhibition of LIF signaling reverses PD-L1-mediated immune suppression in gastric cancer Authors: Cristina Di Giorgio, Silvia Marchianò, Michele Biagioli, Ginevra Lachi, Carmen Massa, Benedetta Sensini, Eleonora Giannelli, Maria Rosaria Sette, Ginevra Urbani, Francesca Paniconi, Luigi Cari, Maria Chiara Monti, Valentina Sepe, Nicola Natalizi, Luigina Graziosi, Eleonora Distrutti, Annibale Donini, Angela Zampella, Stefano Fiorucci Published: 2026 ABSTRACT Highlights•LIF and PD-L1 are co-upregulated in gastric cancer tissues and immune cell subsets.•LIF triggers STAT3 signaling, driving PD-L1 transcription in gastric cancer.•The LIFR antagonist, LRI-305 reduces PD-L1 expression and epithelial–mesenchymal transition.•Blocking LIF signaling restores immune activity in the gastric tumor microenvironment.Keywords: Gastric cancer, Leukemia inhibitory factor, PD-L1, STAT3, Immune evasion, Tumor microenvironment, Checkpoint blockade Introduction Gastric cancer (GC) is a leading cause of cancer-related mortality worldwide, due to high heterogeneity, late-stage diagnosis, and poor prognosis despite multimodal therapeutic approaches [1]. In recent years, the advent of immune checkpoint inhibitors (ICIs) has transformed the therapeutic landscape of solid tumors, including GC, offering new therapeutic approaches by harnessing the host immune system to fight tumor progression [2]. Treatment regimens based on anti-programmed death-1 (PD-1) and/or anti PD-ligand (L)-1, such as nivolumab and pembrolizumab alone or in combination with chemotherapy agents, have been evaluated in patients with GCs. Results from a recent meta-analysis based on 14 study arms from 5 different clinical trials, however, suggest that favorable outcome of ICIs treatment depends on the cancer molecular signature and histopathology, since beneficial effects of ICIs on overall survival (OS), are restricted to patients with high microsatellite instability (MSI) compared to stable microsatellite status, and intestinal histological subtype(3) [4]. In contrast, progression free survival (PFS) and disease control rate were unchanged or even worsened by ICIs treatment in other molecular or histological subtypes [3], highlighting the role of matrix/cancer cell interactions in mediating the ICIs efficacy. The Leukemia Inhibitory Factor (LIF), a pleiotropic cytokine of the IL-6 family, is increasingly recognized as a modulator of tumor biology. Beyond its physiological roles in tissue regeneration, embryogenesis, and immune regulation [5], LIF expression is often upregulated in solid cancers, including GC, and mechanistic studies have provided evidence supporting a role of LIF in promoting proliferation, chemoresistance, and immune escape [6]. In target tissues LIF signals by binding to a heterodimeric receptor complex formed by the LIF receptor (LIFR) and glycoprotein 130 (gp130) [7]. Upon binding to LIFR/gp130 complex, LIF activates multiple signaling pathways including JAK/STAT3, AKT, MAPK, and mTOR(5) [8] promoting the development of an immunosuppressive tumor microenvironment (TME). In particular, LIF promotes the recruitment and functional polarization of tumor-associated macrophages (TAMs), inhibits CD8⁺ T cell infiltration, and impairs the efficacy of anti-PD-1 therapy [8]. Our previous work has shown that elevated expression of LIFR correlates with aggressive tumor features and poor patient prognosis in GC, highlighting its potential as a prognostic marker and a therapeutic target [6]. Crucially, LIF-mediated STAT3 activation has been implicated in the transcriptional upregulation of PD-L1 in breast cancers and NK/T cell lymphoma, suggesting a conserved oncogenic axis across malignancies [9,10]. In the context of GC, this pathway may contribute to tumor cell immune privilege allowing escape from cytotoxic T cell surveillance [11]. Given this background, we have designed a study to investigate the functional role of LIF in regulating PD-L1 expression and promoting immune evasion in GC. By integrating transcriptomic, immunofluorescence, and high-dimensional cytometric analyses, along with pharmacological inhibition studies using a selective LIFR antagonist LRI-305, we aim to elucidate the immunoregulatory circuitry driven by the LIF/STAT3/PD-L1 axis in GCs. In this context, t-SNE analysis identified a distinct LIF⁺/PD-L1⁺ cell population enriched within the neoplastic mucosa, spanning both myeloid and non-hematopoietic compartments, thereby providing direct evidence of a specialized immunosuppressive niche. Our findings support the notion that blockade of LIF signaling helps to restore immune responsiveness and potentiate the therapeutic efficacy of immune checkpoint blockade in GC. Material and methods Human sample Internal cohort Gastric tissue specimens were collected from a total of 61 patients diagnosed with gastric carcinoma who underwent surgical resection at the Surgery Unit of the University Hospital of Perugia (Italy). Among these, 55 patients provided matched pairs of neoplastic tissue and non-neoplastic gastric mucosa, the latter harvested from areas distant from the tumor margin to ensure the absence of microscopic neoplastic infiltration. Additionally, three tumor-only and three non-neoplastic-only samples were obtained. All patients provided written informed consent prior to surgery, and none had received neoadjuvant chemotherapy or radiotherapy. Tissue sampling was performed intraoperatively by a trained biologist. Immediately following excision, specimens were immersed in RNAlater stabilization solution, transferred to the Gastroenterology Research Laboratory, and subsequently snap-frozen at –80 °C for molecular analyses. The study protocol was approved by the Ethics Committee of Regione Umbria (CEAS; permit FI00001, protocol no 2266/2014, dated February 19, 2014), and by the Bioethics Committee of the University of Perugia (permits FIO0003, protocol no 120,687, dated January 12, 2021; and FIO0009, protocol no 357,803, dated November 14, 2024). All procedures were conducted in accordance with institutional guidelines and international ethical standards for the use of human biological material in biomedical research. Validation human cohorts The GSE66229 dataset The GSE66229 dataset, generated by the Asian Cancer Research Group (ACRG), includes gene expression profiles from 300 gastric cancer samples and 100 non-tumorous gastric tissues. The cancer genome atlas stomach adenocarcinoma (TCGA-STAD) The TCGA-STAD, cohort comprises 350 gastric cancer samples and 31 healthy tissue samples. Cell colture and experimental conditions Immortalized cell line The human gastric cancer cell line MKN45 was sourced from the Japanese Collection of Research Bioresources (Human Science Resources Bank, Osaka, Japan). Cells were cultured in RPMI 1640 medium (Sigma-Merck Life Science S.r.l., Milan, Italy), supplemented with 10% fetal bovine serum (FBS), 1% l-glutamine, and 1% penicillin/streptomycin. To ensure optimal proliferation, cells were routinely passaged and only used at early passages (fewer than 10 post-thaw). Prior to experimental procedures, cells were seeded, subjected to serum starvation for 24 h, and then stimulated for 8, 24, or 48 h depending on the experimental design. Mycoplasma contamination was excluded through routine screening using a Mycoplasma PCR Detection Kit (Sigma). Primary cell lines Peripheral blood mononuclear cells (PBMCs) were isolated from heparinized whole blood collected from healthy voluntary donors through density gradient centrifugation using Ficoll-Hypaque. The resulting cell suspension was subsequently used for the separation of macrophages and lymphocytes. Monocytes were enriched using CD14⁺ MicroBeads UltraPure, human (Miltenyi Biotec, cat. 130–118–906), while T cells were isolated employing the Naive CD4⁺ T Cell Isolation Kit II, human (Miltenyi Biotec, cat. 130–094–131). CD14+ monocytes Monocytes were differentiated into M2-polarized macrophages by culturing them for 6 days in RPMI medium supplemented with 10% fetal calf serum (FCS) and 50 ng/mL macrophage colony-stimulating factor (M-CSF). Following differentiation, macrophages were stimulated with total protein extracts derived from MKN45 cells, pre-treated with LIF (10 ng/mL), either alone or in combination with the LIFR antagonist, LRI-305 (10 μM). LIF was used at 10 ng/mL based on previous studies from our group in gastric cancer models [6]. Protein extraction was performed by co-isolation using the Direct-zol™ RNA MiniPrep Kit with Zymo-Spin™ IIC Columns (Zymo Research, Irvine, CA), following the manufacturer’s protocol for simultaneous recovery of RNA and proteins. As control, macrophages were exposed to protein extracts from untreated MKN45 cells. Co-Colture MKN45/T cells On Day 0, MKN45 cells were seeded at a density of 1.5 × 10⁶ cells/mL. After 24 h of serum starvation, cells were pre-treated with leukemia inhibitory factor (LIF, 10 ng/mL), either alone or in combination with the LIF receptor antagonist LRI305 (10 μM). On Day 4, MKN45 cells were co-cultured in direct contact with CD4⁺ T cells (1.5 × 10⁶ cells/mL) and treated with an anti-PD-L1 antibody (30 µg/mL; Leinco Technologies, St. Louis, Missouri) for 24 h. At the end of the co-culture, cells were harvested and stained with specific antibody panels for flow cytometric analysis. All cell cultures were kept in a humidified incubator at 37 °C with 5% CO₂. Results Transcriptomic profiling reveals disrupted mucosal homeostasis and immune evasion signatures in gastric cancer This study involved a comprehensive transcriptomic analysis of paired non-neoplastic and neoplastic gastric mucosa obtained from 61 GC patients, who underwent surgery at the Perugia University Hospital between 2013 and 2022 (Table 1). None of them received chemotherapy or radiation before surgery. A subcohort of 4 paired samples was selected for high-resolution t-distributed stochastic neighbor embedding (t-SNE) analysis to evaluate transcriptional heterogeneity within the tumor microenvironment (Fig. 1A). Table 1Clinicopathological characteristics of the gastric cancer patient cohort. The table summarizes patient demographics (age and sex), tumor localization, type of gastrectomy, extent of lymphadenectomy, Lauren histological classification, pathological TNM stage, overall stage, peritoneal cytology (CY), and neoadjuvant treatment status.Table 1 dummy alt textIDAgeSexLocalizationType of gastrectomyLymphadenectomyLaurenpTpNpMStageCYNeoadiuvante171FlowersubtotalD2intestinal1b00I01275MmiddlesubtotalD2intestinal310I00380MmiddlesubtotalD2intestinal33a0IV10483FlowersubtotalD2intestinal310II00583MuppertotalD1intestinal4a20IV10676MuppertotalD2intestinal4a3a0III00777MlowersubtotalD2intestinal4a10III20880FlowersubtotalD1intestinal1b00I00984MlowersubtotalD1intestinal320III001080FmiddlesubtotalD2diffuse200I001166FlowertotalD1diffuse321IV001262MmiddletotalD1intestinal100I001383MlowersubtotalD1intestinal200I001475MlowersubtotalD2intestinal4a00II001575FlowersubtotalD2intestinal210II201678FlowersubtotalD2intestinal300II201757MuppertotalD2diffuse4a21IV011872MuppertotalD2intestinal4b3b0III201976MuppersubtotalD2diffuse4a3b0III002076MlowersubtotalD2intestinal4a3a0III002167MlowersubtotalD2intestinal4a00IV102282FlowersubtotalD2intestinal33b1IV102374FmiddlesubtotalD2diffuse320III002477FmiddletotalD2intestinal33a0III002578FlowersubtotalD2intestinal33a0III002688FlowersubtotalD1diffuse33a0IV102786FuppertotalD1diffuse4b00III202878MlowersubtotalD2intestinal320III002963MlowersubtotalD2intestinal33a0III003079FlowersubtotalD2diffuse4a3b0III003176MuppertotalD2intestinal4a3a1III003249MuppertotalD1diffuse33b1IV003371MmiddlesubtotalD2intestinal320III003465MuppertotalD2intestinal4b11IV113575FuppertotalD2diffuse4a3b0III003667MmiddlesubtotalD2diffuse220IV103771MmiddlesubtotalD2diffuse4a3b0IV103853MuppertotalD2intestinal300II013970FlowersubtotalD2intestinal1b00I004068FlowersubtotalD2intestinal310II014152FmiddletotalD2diffuse310II004276MlowersubtotalD2intestinal200I004345MlowersubtotalD2diffuse310II014461MmiddletotalD2intestinal1a00I004556MuppertotalD2intestinal310II014662MlowersubtotalD2intestinal1b00I014779MmiddlesubtotalD2intestinal23a0III014876FlowersubtotalD2intestinal1b00I014979MlowersubtotalD2intestinal320III005070MuppertotalD2intestinal1a00I005174FmiddletotalD2intestinal200II015280FlowersubtotalD2intestinal1b00I005372MmiddletotalD2intestinal300II015466FlowersubtotalD2intestinal1b00I005563MmiddletotalD2diffuse33b0III015682FlowersubtotalD1intestinal310II005784FmiddletotalD2intestinal4a3a0III005863FlowersubtotalD2intestinal0000005975MlowersubtotalD2intestinal310II006064MlowersubtotalD2intestinal200I016189FmiddlesubtotalD1diffuse300II00 Clinicopathological characteristics of the gastric cancer patient cohort. The table summarizes patient demographics (age and sex), tumor localization, type of gastrectomy, extent of lymphadenectomy, Lauren histological classification, pathological TNM stage, overall stage, peritoneal cytology (CY), and neoadjuvant treatment status. Non-neoplastic and Neoplastic mucosa of GC exhibit distinct molecular profiles. A) Transcriptome analysis of non-neoplastic and neoplastic tissues was carry out in 61 GC patients, who underwent surgical resection at Perugia University Hospital. Panel B) illustrated Principal Component Analysis (PCA) plot analysis. Panel C) depicted Scatter Dot Plot in Neoplastic mucosa (y axis) and Non-neoplastic Mucosa of GC patients (x axis). In panel D) the number of differential expressed genes (DEGs) and in E) pathways modulated in neoplastic mucosa compared to non-neoplastic mucosa were represented. Panel F) illustrates the top 30 significantly upregulated and top 30 significantly downregulated genes in neoplastic mucosa compared to non-neoplastic mucosa and G) their relative fold change is represented. Unsupervised Principal Component Analysis (PCA) revealed a clear segregation between neoplastic and non-neoplastic tissues, supporting the existence of robust transcriptional reprogramming associated with GC progression (Fig. 1B). Differential gene expression analysis identified 503 significantly dysregulated genes, the majority of which were downregulated (79.72%, Fig. 1D). Functional pathway enrichment demonstrated a suppression of pathways critical for mucosal homeostasis, including PI3K-Akt signaling, cytokine-receptor interactions, NRF2 signaling, and DNA damage response via the Ataxia-Telangiectasia Mutated (ATM) pathway (Fig. 1E). These findings reflect the profound reorganization of cellular identity and signaling networks in neoplastic gastric epithelium. Among the downregulated genes, gastrokine-1 and −2 (GKN1, GKN2) and trefoil factor 2 (TFF2), were particularly notable due to their established role in epithelial barrier maintenance and mucosal repair [12,13]. Conversely, upregulated genes included CEACAM6, CLDN1, GDF15, IL17RB, and SERPINB5, all of which are implicated in immune evasion, epithelial-mesenchymal transition (EMT), and tumor aggressiveness (Fig. 1F, G) [[14], [15], [16]]. Given the prominence of immune-evasion pathways in this transcriptional profile, we next focused on specific mediators with known roles in modulating the tumour–immune interface. In particular, we investigated LIF and PD-L1, two key immunoregulatory molecules previously implicated in gastric cancer progression, to define their expression patterns and potential interplay within the tumour microenvironment (Fig. 2). LIF Overexpression correlates with PD-L1 expression, predict a poor prognosis and immune suppression in Neoplastic GC Mucosa. A) Hematoxylin and eosin (H&E), IF analysis of LIF (red) and PD-L1 (green) in non-neoplastic and neoplastic mucosa of GC patients and relative Mean Fluorescence intensity. Gene expression (Log2) of: B)LIF, C)LIFR and D)PD-L1. E) Correlation analysis between LIF and PD-L1. Kaplan Meyer analysis of F) LIF and G) LIFR. H) Immune-deconvolution analysis based on human transcriptomic analysis was performed. I) Frequencies of Immune subpopulation cells in gastric mucosa in non-neoplastic and neoplastic mucosa. (*represents statistical significance versus Non-neoplastic, p < 0.05). Integrated transcriptomic and t-SNE analysis reveals LIF-Mediated immune evasion in gastric cancer Immunofluorescence (IF) and quantitative RNA-seq analyses revealed increased Leukemia Inhibitory Factor (LIF) and PD-L1 expression in neoplastic gastric mucosa, suggesting their coordinated role in immune evasion (Fig. 2A–D). Although the correlation strength was modest, correlation analyses revealed a statistically significant positive association between LIF and PD-L1 expression levels, supporting the existence of a potentially coordinated regulatory axis (Fig. 2E). Validation through external cohorts (TCGA and ACRG) further supported the association between elevated LIF expression and poorer overall survival (Fig. 2F). In addition, Kaplan–Meier analysis of LIFR expression was included for descriptive purposes, although the observed trend did not reach statistical significance (Fig. 2G). Since other inflammatory cytokines may contribute to PD-L1 regulation, we additionally analyzed the expression of Oncostatin M (OSM) and IFNγ and their correlation with CD274/PD-L1 expression in the internal cohort as well as in the TCGA-STAD and ACRG validation datasets (Supplementary Figure 2). OSM expression showed variable modulation across cohorts and displayed weak-to-moderate positive correlations with CD274 expression. In contrast, IFNγ expression demonstrated a stronger and more consistent positive correlation with CD274, particularly in the TCGA-STAD and ACRG cohorts, in agreement with the established role of IFNγ in PD-L1 induction. Nevertheless, these findings do not exclude an independent contribution of LIF signaling, which was directly validated by our mechanistic experiments demonstrating LIF-induced STAT3 activation and PD-L1 upregulation in GC cells. Immune deconvolution analysis revealed a marked increase in immunosuppressive cellular subsets, particularly regulatory T cells (Tregs) and M2-polarized myeloid-derived suppressor cells (MDSCs), within neoplastic mucosa, reinforcing the hypothesis that LIF/PD-L1 axis contributes to immune escape [[17], [18], [19]] (Fig. 2H-I). To validate these findings, single-cell suspensions from paired neoplastic and non-neoplastic mucosa obtained from a sub-cohort of four patients were analyzed by flow cytometry, as detailed in the Supplementary Materials and Methods. High-dimensional t-SNE analysis focusing on LIF/LIFR and PD-L1/PD1 expression identified three distinct CD45⁻ LIF⁺ cell populations, all significantly enriched in neoplastic compared to paired non-neoplastic mucosa (Fig. 3A–C). t-SNE analysis of tumor bulk identifies LIF/LIFR/PD-L1/PD-1 immune and non-immune subsets enriched in gastric neoplastic mucosa. Single-cell suspensions from matched neoplastic and non-neoplastic gastric mucosa of four patients were subjected to flow cytometry. High-dimensional t-SNE analysis was performed to identify subpopulations expressing LIF, LIFR, PD-L1, and PD-1. A-E) CD45- cells A) t-SNE 2D scatter plot depicting CD45⁻ LIF⁺ cell populations. B) Histograms showing the expression levels of all analyzed markers across the three identified LIF⁺ subpopulations. C) Frequency trend of CD45⁻ LIF⁺ subpopulations in paired non-neoplastic and neoplastic gastric mucosa. D) t-SNE 2D scatter plot of PD-L1⁺ cells within the CD45⁻ LIF⁺ POP1 subpopulation. E) Frequency of PD-L1⁺ cells in POP1 between paired non-neoplastic and neoplastic gastric mucosa. F-H) Myeloid cells: F) t-SNE 2D scatter plot depicting LIFR⁺ myeloid cell populations. G) Histograms showing the expression levels of all analyzed markers across the seven identified LIFR⁺ subpopulations. H) Frequency trend of LIFR⁺ significative subpopulations in paired non-neoplastic and neoplastic gastric mucosa. I-K) T helper cells. I) t-SNE 2D scatter plot depicting PD1⁺ cell populations. J) Histograms showing the expression levels of all analyzed markers across the six identified PD1⁺ subpopulations. K) Frequency trend of PD1⁺ significative subpopulations in paired non-neoplastic and neoplastic gastric mucosa. T cytotoxic cells. I) t-SNE 2D scatter plot depicting PD1⁺ cell populations. L) Histograms showing the expression levels of all analyzed markers across the four identified PD1⁺ subpopulations. M) Frequency trend of PD1⁺ significative subpopulation in paired non-neoplastic and neoplastic gastric mucosa. (*p > 0.05) O) Schematic representation of results. Notably, one of these populations (LIF⁺ population 1) contained an increased frequency of PD-L1⁺ cells in the tumor samples (Fig. 3D-E). Parallel analysis of CD45⁺ immune populations revealed seven LIFR⁺ myeloid subsets, four of which were significantly expanded in tumor mucosa (Fig. 3E–F). Although clearly defined LIF⁺ or LIFR⁺ T cell subsets were not identified, a subpopulation of helper T cells (POP2) exhibited low levels of both LIF and LIFR expression, while two cytotoxic T cell populations (POP1 and POP2) expressed detectable LIFR expression. Notably, only PD-1⁺ helper and cytotoxic T cells were found to be increased in neoplastic tissues (Fig. 3I–N). Taken together, these findings suggest a model in which neoplastic epithelial cells secrete LIF, which in turn acts in an autocrine manner via LIFR—also expressed on their own surface—to sustain JAK1/STAT3 activation. This, in turn, reinforces a feed-forward loop maintaining LIF production [5] while simultaneously inducing PD-L1 expression. In parallel, tumor-derived LIF acts in a paracrine way on LIFR-expressing macrophages, within the tumor microenvironment, promoting PD-L1 upregulation. These PD-L1⁺ macrophages can then engage PD-1 on tumor-infiltrating T cells, driving their apoptosis and functional exhaustion. Altogether, this dual autocrine/paracrine mechanism positions LIF/LIFR signaling as a key orchestrator of immune suppression and a facilitator of immune escape in gastric cancer. STAT3 directly binds the PD-L1 promoter in response to LIF/LIFR signaling Previous findings support the hypothesis that LIF may directly regulate PD-L1 expression via JAK1/STAT3 signaling, both within in tumor cells and the surrounding immune microenvironment. To dissect the underlying molecular mechanisms, we carried out in vitro experiments using the poorly differentiated gastric cancer cell line MKN45, which endogenously express both LIF and LIFR (Fig. 4A). LIF increases the expression of PD-L1 in a JAK1/STAT3 dependent manner in a gastric cancer cell line. A) Immunofluorescence (IF) analysis of basal expression of LIFR (red) and LIF (red) on MKN45 cell line and their relative Mean Fluorescence Intensity (MFI). MNK45 cells were exposed to LIF for 24 h or left untreated. IF analysis of PD-L1 (green) in MKN45 exposed to LIF or left untreated, and the estimated Mean fluorescence intensity (MFI) of PD-L1. C-E) Representative Western blot analysis and densitometric analysis demonstrating C) p-JAK1/JAK1, D) pSTAT3-STAT3 and E) PD-L1/GAPDH ratio. F) Relative mRNA expression of PD-L1. Each value is normalized to GAPDH and is expressed relative to not treated group, which are arbitrarily set to 1. G) A schematic diagram shows the location DNA sequences of STAT3 inducibile elements (SIE) on the PD-L1 promoter. ChIP assay: cell lysates of MKN45 cells were immunoprecipitated with anti-STAT3 or control IgG (Mock IgG), and the presence of specific regions of PD-L1 promoter in the immunoprecipitate was determined by real-time PCR. H) Fold enrichment analysis and I) % of Input analysis. In each experimental set results are the mean ± SEM of 3–5 samples for group. (*=p < 0.05). To investigate the potential role of the LIF/LIFR pathway in regulating the PD-L1, MKN45 cells were exposed to 10 ng/mL LIF. This treatment robustly induced PD-L1 expression at both the transcript and protein levels, mediated by the activation of JAK1/STAT3 signaling, as evidenced by increased phosphorylation at JAK1 (Tyr1022) and STAT3 (Tyr705) (Fig. 4B-G). Promoter analysis of PD-L1 identified six STAT3 inducible elements (SIEs) upstream of the ATG at positions: −1830/−1819, −2448/−2437, −2687/−2676, −3602/−3591, −3647/−3636 and −4512-/−4501 (Fig. 4H). Chromatin immunoprecipitation (ChIP) assays validated STAT3 binding to these sites in MKN45 cells upon LIF exposure, demonstrating a direct transcriptional mechanism for PD-L1 upregulation mediated by the LIF/LIFR/STAT3 pathway (Fig. 4I-J). LIFR antagonism by LRI305 attenuates LIF-Driven tumor progression and PD-L1 expression in MKN45 cells Based on the evidence that modulation of PD-L1 by LIF/LIFR/STAT3 axis has therapeutic potential and considering the lack of clinically approved LIFR antagonists, we have developed a novel small molecule, LIFR-inhibitor, (LRI)-305 (Fig. 5A)) inspired by the repositioning of mifepristone as a LIFR antagonist(20). LRI-305 reverted LIF-Induced proliferation and EMT process in MKN45 cells. A) 2D structure of the novel LIFR antagonist, LRI305. LRI-305 inhibition activity of LIFR/LIF binding accessed by B) AlphaScreen Assay and C) Transactivation assay D) The effect on vitality of MKN45 of LRI-305 (0.1, 1, 5, 10, 20 µM) was determined using MTS assay. E) Representative Western blot analysis and densitometric analysis demonstrating : pSTAT3-STAT3 (on left) and PD-L1/GAPDH ratio (on right). F) Representative IC-FCM shows frequencies of PD-L1+cells. G) Frequencies of PD-L1+ cells; Mean fluorescence Intensity (MFI) of PD-L1+ cells and H) Relative mRNA expression of PD-L1. I-J) Cell-cycle analysis was performed by Ki-67/7-AAD staining through IC-FCM. I) Representative IC-FCM and J) frequencies of Apoptotic cells, Resting cells (G0-G1 phase) and Proliferating cells (S-G2 phase) and Mitotic phase (M phase) in NT, LIF 10 ng/mL and LIF + LRI-305 5 and 10 µM experimental groups. Relative mRNA expression of proliferation markers F)KI67 and G)CMYC. M) Representative IC-FCM shows KI-67/E-CADH staining through IC-FCM. N) Frequencies of Non Proliferating Cells; Frequencies of Proliferating and Differentiated cells; Frequencies of Proliferating and De-differentiated cells. Relative mRNA expression of Epithelial mesenchymal transition (EMT) markers: O)ECADH and P)SNAIL1.Q) A scratch wound healing assay is shown. MKN45 cell monolayers were scraped in a straight line using a p200 pipette tip, then they are left untreated or primed with LIF 10 ng/mL alone or in combination with LRI-305 5 µM and 10 µM. The wound generated was captured at 0 h and 48 h of incubation with the compounds above described. The images show cell migration at the two times point indicated and percentage of wound closure at 48 h. In each experimental set results are the mean ± SEM of 3–5 samples for group. In mRNA expression analysis each value is normalized to GAPDH (*=p < 0.05). Alpha Screen assay demonstrated that LRI-305 effectively inhibits LIF/LIFR interaction with an IC50 of 0.86 ± 0.239 µM (Fig. 5B). Using the LRI-305 we have investigated the role of LIFR inhibition on cell proliferation and viability. MKN45 were exposed to 10 ng/mL LIF, either alone or in combination with escalating concentrations of LRI-305 (0.1, 1, 5, 10, 20 µM) for 24 h. MTS assays revealed that LRI-305 reversed the LIF-induced proliferative effects with an IC50 of 0.7 ± 0.19 µM (Fig. 5C). To gain mechanistic insight into STAT3-mediated regulation of PD-L1, we conducted a transactivation assay using HepG2 cells transiently transfected with a l STAT3-inducible elements (SIE) luciferase reporter plasmid, as previously described [6]. Treatment with LIF alone or in combination with LRI-305 revealed a pronounced, dose-dependent inhibition of STAT3-mediated transcriptional activity by LRI-305, with an IC₅₀ of 2.9 ± 0.9 µM (Fig. 5C). Western Blot and cytometry analysis confirmed these findings, showing a concomitant reduction in STAT3 phosphorylation and PD-L1 expression at both the protein (Fig. 5E-G) and gene expression level (Fig. 5H). The impact of LRI-305 on cell replication was also assessed by intracellular flow cytometry (IC-FCM) in MKN45 cells stained for Ki-67 and 7-AAD. As shown in Fig. 5I-J LIF reduced apoptosis and promoted cell cycle progression into the S-G2/M phases. Co-treatment with LRI-305 (5–10 µM) induced G0-G1 arrest and increased apoptosis. At the molecular level, LRI-305 markedly reversed the LIF-induced upregulation of the proliferation markers KI67 and CMYC (Fig. 5K-L). We next investigated the effect of LRI-305 on epithelial-mesenchymal transition (EMT). Cytometry analysis revealed that LIF exposure increased the percentage of proliferating dedifferentiated cells (Ki-67⁺/E-cadherin⁺), while reducing E-cadherin expression and upregulating the EMT transcription factor SNAIL1. These changes were reversed by LRI-305, which restored epithelial identity and suppressed mesenchymal reprogramming (Fig. 5M-P) [21]. To assess the functional consequences on migratory behavior and cell motility, a scratch wound healing assays were performed. As shown in Fig. 5Q, LIF markedly enhanced MKN45 cell migration, leading to ∼80% wound closure at 48 h. LRI-305 co-treatment significantly impaired wound healing, reducing closure to ∼28% and ∼10% at 5 µM and 10 µM, respectively. To further address the biological heterogeneity of gastric cancer, selected experiments were extended to an additional gastric cancer cell line, MKN74 (Supplementary Figure 3). MKN74 cells, derived from a moderately differentiated tubular gastric adenocarcinoma, retained responsiveness to both LIF stimulation and LRI-305 treatment. Specifically, LIF exposure increased the expression of proliferation- and EMT-related markers, including KI67, SNAIL1, and PD-L1, while reducing CDH1 expression. Co-treatment with the LIFR antagonist LRI-305 significantly attenuated these effects. Overall, these findings indicate that activation of the LIF/LIFR axis and its pharmacological modulation by LRI-305 are conserved across distinct gastric cancer cellular contexts. Collectively, these findings demonstrated that LIFR antagonism by LRI-305 effectively impaired LIF-induced STAT3 signaling and its downstream oncogenic and immunosuppressive effects, including enhanced cell proliferation, EMT, migration, and PD-L1 upregulation in gastric cancer cells. The broad reversal of these malignant phenotypes highlights the therapeutic potential of LRI-305 as a targeted therapeutic agent in gastric cancer. Dual targeting of LIFR and PD-L1 enhances anti-tumor immune responses in co-culture models To investigate the role of LIF exposure on extracellular matrix remodeling and immune modulation, we conducted in vitro experiments using macrophages and T cells isolated from the buffy coats of healthy donors (Fig. 6). Combined LIFR and PD-L1 blockade reduces macrophage PD-L1 expression, limits T cell apoptosis, and decreases PD-1⁺ T cell frequency in co-culture models. A-C) M2-polarized human macrophages were exposed to protein lysates derived from MKN45 cells pre-treated for 24 h with LIF (10 ng/mL) or left untreated (NT), in the presence or absence of the LIFR antagonist LRI305. PD-L1 surface expression was assessed by flow cytometry. A) schematic experimental flow. B) Representative Flow cytometry analysis; C) shown as the percentage of PD-L1⁺ cells (D-F) In another experimental set MKN45 cells were treated with LIF alone or in combination with LRI305 (10 μM). Autologous T cells were exposed to MKN45 proteins for 24 alone or with anti–PD-L1 antibody (30 µg/mL). E) Apoptosis cell rate was evaluated by Annexin V staining, and F) % PD-1 + T cells was measured. G) % of HLA-DRA/CD38+ Activated T cells. Data represent mean ± SEM of 3–5 independent experiments. *p < 0.05. Initially, we observed that the exposure of the M2-polarized macrophages (as detailed in the Supplementary Methods) to the protein lysates derived from MKN45 cells pre-treated with LIF or vehicle (NT) significantly upregulated PD-L1 expression on the macrophage surface (Fig. 6A–C). This effect was reversed by LIFR antagonism, resulting in an approximately fourfold decrease in the proportion of PD-L1⁺ macrophages. To investigate the impact on T cells, we employed MKN45 cells and autologous T cells (Fig. 6D). MKN45 cells were pre-treated for 24 h with LIF alone, or in combination with the LIFR antagonist, LRI305. Following pre-treatment, MKN45 protein lysates were obtained and used on T cells, alone or in combination with an anti-PD-L1 antibody (30 µM) for 24 h. The effects of LIF-induced-PD-L1–signaling were assessed by IC-FCM using Annexin V staining (Fig. 6E). Our results showed that ≈ 23% T cells cultured with protein obtained from LIF-treated MKN45 cells underwent apoptosis. These findings support the hypothesis that LIF-driven PD-L1:PD-1 interactions contribute to immune evasion within the tumor microenvironment. Conversely, pre-treatment of MKN45 cells with the LIF/LIFR antagonist significantly decreased the proportion of apoptotic T cells of about In both experimental conditions the addition of an anti-PD-L1 antibody significantly reduced the percentage of apoptotic T cells ≈ 3%. Notably, the combined treatment with anti-PD-L1 and LRI305 not only further decreased T cell apoptosis but also significantly reduced the proportion of T cells expressing PD-1, suggesting a dual benefit in preventing both functional exhaustion and depletion of effector T cells (Fig. 6F). To further characterize the functional impact of dual LIFR/PD-L1 blockade on T-cell status, additional IC-FACS analyses were performed to assess the activation markers HLA-DR and CD38 (Fig. 6G). Combined anti–PD-L1/LRI-305 treatment significantly increased the frequency of double-positive HLA-DR/CD38 activated T cells compared with exposure to conditioned media derived from LIF-treated MKN45 cells alone, further supporting restoration of T-cell activation following dual blockade. Taken together, these results underscore that LIFR inhibition effectively counteracts LIF-mediated immune evasion by limiting PD-L1 upregulation in both tumor and myeloid compartments, thereby disrupting PD-L1:PD-1 inhibitory signaling. When combined with PD-L1 blockade, this strategy exerts additive effects-reducing T cell apoptosis, diminishing PD-1 expression on surviving T cells, and potentially restoring their cytotoxic capacity. These findings provide mechanistic support for the combined therapeutic targeting of LIF/LIFR and PD-L1 as a rational approach to enhance anti-tumor immunity in GC, likely through suppression of the JAK/STAT3-dependent transcriptional program driving PD-L1 expression. Discussion It is well recognized that immune escape by cancer cells is supported by the development of an immunosuppressive tumor microenvironment (TME) characterized by predominance of regulatory T cells (Tregs), M2-polarized macrophages, and myeloid-derived suppressor cells (MDSCs), due to the activation of immunosuppressive pathways of immune checkpoints (ICs) such as PD‐1/PD‐L1 and CTLA‐4, and more immunosuppressive factors like interleukin 10 (IL‐10) and transforming growth factor beta, , associated with low tumor mutational burden (TMB) [[22], [23], [24]]. These tumors are typically unresponsive or poorly responsive to immune checkpoint inhibitors (ICIs). This immunological phenotype is a hallmark of several aggressive cancers, including pancreatic ductal adenocarcinoma (PDAC) [25], prostate cancer [26], epithelial ovarian carcinoma [27], specific subtypes of renal cell carcinoma [28], and microsatellite-stable (MSS) gastric cancer [29]. These tumor types share a common barrier to effective immunotherapy—namely, a suppressive immune landscape that impairs antigen presentation, cytotoxic activity, and T cell infiltration. Reversing the immune-suppressive phenotype and promoting immune infiltration, antigenicity, and responsiveness to ICIs—is a major focus of current therapeutic strategies. In this study, we identified a putative role for LIF/LIFR/STAT3 pathway in the regulation of PD-L1 expression and immune evasion in GC, unveiling a mechanism through which neoplastic epithelial cells and tumor-associated macrophages orchestrate immunosuppressive signals within the TME. By integrating transcriptomic profiling, t-SNE analysis, based on cell cytometry, functional assays, and pharmacological inhibition, we have provided convincing evidence that STAT3 activation by LIF promotes oncogenesis and contributes to immunosuppression by regulating PD-L1 transcription, thereby contributing to GC immune escape. Consistent with this setting [[8], [9], [10]], we have shown that LIF expression is upregulated in GC tissues and displays a modest but significant positive correlation with PD-L1 expression. Furthermore, by t-SNE analysis we have identified specific cell subsets co-expressing LIF and PD-L1. These LIF⁺/PD-L1⁺ cells were detected both in the epithelial and myeloid compartments of neoplastic mucosa, supporting an immunoregulatory circuit wherein LIF signals both in an autocrine manner in cancer cells and in a paracrine manner in LIFR+ immune cells, essentially myeloid cells/macrophages, is involved in PD-L1 regulation. Together, these expression data suggest that LIF/ LIFR promotes PD-L1 transcription in cancer cells and macrophages fostering the development of a tolerogenic microenvironment in GC [8,9]. We have functionally validated these expression data by challenging MKN45 cells, a GC cell line, with LIF. The results of these experiments demonstrate that LIF directly increases PD-L1 expression in MKN45 GC cells, and that this effect involves the phosphorylation of JAK1 (Tyr1022) and STAT3 (Tyr705), two intracellular signaling pathways activated by binding of LIF to the LIF/Gp130 complex [30]. Since phosphorylation of STAT3 is essential for recruitment of this transcription factor to target genes, we performed a CHIP assay on the PD-L1 promoter, and confirmed that the direct recruitment of STAT3 to multiple sites within the PD-L1 promoter, in response to challenging MKN45 cells with LIF. These findings extend previous observations in NK/T cell lymphoma and breast cancer, where STAT3 activation was shown to directly promote the expression of immune checkpoints in cancer cells [8,9]. The findings that LIF promotes a STAT3-dependent transcription of PD-L1 further expand on the role of this cytokine in cancer development, as indicated by the identification of LIF/PD-L1 co-expressing cells in GC. To explore the translational potential of this pathway, we have generated and characterized a small-molecule antagonist of LIFR, LRI-305 inspired by the structure of mifepristone [20]. LRI-305 is LIFR antagonist that disrupts LIF–LIFR interaction at nanomolar concentrations and abrogates STAT3 activation in a concentration-dependent manner. Using this agent, we have shown (Fig. 5), that LIFR antagonism reverses STAT3 phosphorylation and PD-L1 expression induced by LIF along with remodelling of oncogenic features induced by LIF, including MKN45 proliferation, molecular signature of EMT, and migratory capacity. These results highlight the feasibility of pharmacological blockade of LIFR as a therapeutic strategy to reshape communication between cancer cells and extracellular matrix. Furthermore, the combination of LRI-305 with anti-PD-L1 antibody in co-culture assays enhanced T cell proliferation and reduced T cells apoptosis, suggesting that dual inhibition of LIFR and PD-L1 may synergistically restore anti-tumor immune activity. This aligns with prior reports showing that LIF suppresses CD8⁺ T cell infiltration and limits response to ICIs [8,18] and reinforces the notion that targeting the LIF/LIFR axis could overcome resistance to immunotherapy in non-MSI-high GC, a subset typically unresponsive to checkpoint blockade alone [3,4]. Our findings also extend on the immunosuppressive role of macrophages in GC. Exposure of M2-polarized macrophages to LIF-conditioned tumor lysates markedly upregulated PD-L1 in these cells, an effect reversed by LRI-305. This finding suggests that LIF/PD-L1 macrophages could represent a newly discovered subsets of TAMs that could be reprogrammed by cancer cells-derived LIF to express immune-inhibitory ligands, contributing to dampening T cell responses—a concept recently validated in peritoneal metastasis models of GC [[17], [31]]. Taken together, this study provides compelling mechanistic evidence that LIF/LIFR/STAT3 signalling governs PD-L1 expression and immune evasion in GC. While in vivo validation will be required to further confirm the therapeutic potential of LIFR inhibition in GC, the present study integrates multiple complementary experimental approaches, including transcriptomic analyses of human samples, high-dimensional immune profiling, functional in vitro assays, and immune co-culture systems, thereby providing robust mechanistic support for the role of the LIF/LIFR/STAT3/PD-L1 axis in tumor immune evasion. Furthermore, by demonstrating that pharmacological inhibition of LIFR by LRI-305 restores immune activation and suppresses the oncogenic phenotypes, this study supports the therapeutic targeting of LIFR in combination with ICIs as a rational strategy for improving immunotherapy efficacy in GC patients. Funding This work was partially supported by grant from the Italian Ministry of University and Research (MUR): PRIN-2022 n. 20223K7L88 and PRIN-2022-PNRR n. 20227JB3W to SF and MB. Data availability statement The data generated in this study are publicly available The transcriptomic data obtained from the experimental cohort are publicly available in Mendeley Data, Mendeley Data, V1, doi: 10.17632/j46b34szhw.1 data that support the findings of this study were ob-tained from The GSE66229 Dataset from ACRG at [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE66229], The TCGA-STAD at [https://portal.gdc.cancer.gov/]. CRediT authorship contribution statement Cristina Di Giorgio: Writing – review & editing, Writing – original draft, Supervision, Project administration, Data curation, Conceptualization. Silvia Marchianò: Formal analysis, Data curation. Michele Biagioli: Writing – review & editing, Resources, Funding acquisition, Formal analysis, Data curation. Ginevra Lachi: Methodology, Data curation. Carmen Massa: Methodology, Data curation. Benedetta Sensini: Methodology, Data curation. Eleonora Giannelli: Methodology, Data curation. Maria Rosaria Sette: Methodology, Data curation. Ginevra Urbani: Writing – review & editing. Francesca Paniconi: Methodology, Data curation. Luigi Cari: Software, Formal analysis. Maria Chiara Monti: Writing – review & editing, Supervision, Data curation. Valentina Sepe: Writing – review & editing, Supervision, Data curation. Nicola Natalizi: Formal analysis. Luigina Graziosi: Formal analysis. Eleonora Distrutti: Writing – review & editing. Annibale Donini: Writing – review & editing, Supervision. Angela Zampella: Writing – review & editing, Conceptualization. Stefano Fiorucci: Writing – review & editing, Writing – original draft, Supervision, Resources, Investigation, Funding acquisition, Conceptualization. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements None. Footnotes Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.tranon.2026.102848. Contributor Information Cristina Di Giorgio, Email: cristi.digiorgio@gmail.com. Stefano Fiorucci, Email: stefano.fiorucci@unipg.it. Supplementary materials References Associated Data Supplementary Materials Data Availability Statement The data generated in this study are publicly available The transcriptomic data obtained from the experimental cohort are publicly available in Mendeley Data, Mendeley Data, V1, doi: 10.17632/j46b34szhw.1 data that support the findings of this study were ob-tained from The GSE66229 Dataset from ACRG at [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE66229], The TCGA-STAD at [https://portal.gdc.cancer.gov/].