LIM and SH3 protein 1 deficiency confers resistance to cytotoxic lymphocyte-mediated lysis and immunotherapy in gastric cancer by disrupting cytoskeleton dynamics Authors: Teming Zhang, Yingying Sun, Chao Chen, Jingwei Zheng, Liyun Chang, Yizhen Xiang, Qiongying Zhang, Haoliang Li, Wangkai Xie, Qiaoyi Xu, Chenchen Mao, Mingdong Lu, Zhenzhai Cai, Xian Shen, Xiangyang Xue, Zheng Han Published: 2026 ABSTRACT AbstractBackgroundA substantial amount of research has been dedicated to the mechanisms by which tumor cells evade immune system recognition and manipulate the immune microenvironment to facilitate immune escape. Recent studies have shown that viruses and tumors can protect themselves from immune cytotoxicity by remodeling the actin cytoskeleton. However, cytoskeleton-mediated immune resistance and the specific cytoskeleton-related proteins involved require further research.MethodsSingle-cell RNA sequencing was used to identify cytoskeleton-related genes associated with the response to anti-programmed cell death protein 1 (PD-1) therapy across four digestive tumors. Immunohistochemistry was used to detect LASP1 expression in gastric cancer and analyze its prognostic value for survival and anti-PD-1 response. The impact of LASP1 deficiency on tumor response to anti-PD-1 treatment and cytotoxic lymphocyte-mediated lysis was determined in vivo and in vitro. Live-cell imaging was used to compare actin cytoskeletal dynamics at the immunological synapse between LASP1-deficient and mock tumor cells. Molecular mechanisms underlying LASP1-mediated immune-resistance were dissected using co-immunoprecipitation, immunofluorescence, domain deletion complementation, and Laurdan staining.ResultsThe deficiency of LASP1 in gastric cancer affected the sensitivity of tumor cells to immunotherapy and induced cytotoxic lymphocytes exhaustion. LASP1 may act as a scaffold protein to regulate the Arp2/3 complex and remodel the cytoskeleton at the immunological synapses. LASP1 deficiency in tumor cells impairs lytic immunological synapse function by disrupting cytoskeletal dynamics-mediated cell membrane lipid organization at the immunological synapse. Finally, simvastatin combined with anti-PD-1 therapy reversed immunotherapy resistance in LASP1-deficient tumors.ConclusionsThe deficiency of LASP1 in tumors mediates immunological synapse dysfunction by affecting cytoskeletal dynamics-mediated cell membrane lipid organization, thus enabling tumors to protect themselves from immune cytotoxicity and immunotherapy.Keywords: Gastric Cancer, Immunotherapy, Escape/evasion WHAT IS ALREADY KNOWN ON THIS TOPIC WHAT THIS STUDY ADDS This study elucidated that dysfunction of the cytoskeleton-related protein LASP1 in tumors can lead to immunological synapse defects through affecting cytoskeletal dynamics-mediated cell membrane lipid organization, thus protecting tumor cells from cytotoxic lymphocyte-mediated cytotoxicity and immunotherapy. And this cytoskeletal abnormality-mediated resistance to immune cytotoxicity can be reversed by pharmacological intervention targeting lipid metabolism in combination with anti-programmed cell death protein 1 therapy. This study elucidated that dysfunction of the cytoskeleton-related protein LASP1 in tumors can lead to immunological synapse defects through affecting cytoskeletal dynamics-mediated cell membrane lipid organization, thus protecting tumor cells from cytotoxic lymphocyte-mediated cytotoxicity and immunotherapy. And this cytoskeletal abnormality-mediated resistance to immune cytotoxicity can be reversed by pharmacological intervention targeting lipid metabolism in combination with anti-programmed cell death protein 1 therapy. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY This study provides new insights into the mechanisms by which tumors protect themselves from immune attacks, and offers a novel target and strategy for predicting even improving the efficacy of tumor immunotherapy. This study provides new insights into the mechanisms by which tumors protect themselves from immune attacks, and offers a novel target and strategy for predicting even improving the efficacy of tumor immunotherapy. Introduction Gastric cancer (GC) is a major global health burden, ranking fifth in incidence and fourth in cancer-related mortality, with the highest reported rates in East Asia and Europe. Antitumor immunotherapy has made tremendous progress over the past decade and has become the standard treatment for various malignant tumors, including GC.1 2 Although immunotherapy has fundamentally altered the treatment regimen for GC, its clinical response rate remains low, failing to be beneficial for many patients. In a recent clinical trial, only approximately 15% of patients responded to programmed death-ligand 1 (PD-(L)1) monotherapy.3 Multiple endogenous immune evasion and adaptive drug resistance mechanisms in tumors limit the application of immunotherapy in solid tumors, especially GC.4 5 Therefore, urgent research is needed to elucidate the molecular mechanisms of immunotherapy resistance and develop predictive biomarkers and therapeutic strategies to overcome it. The principal effector cells in the antitumor response are cytotoxic lymphocytes (CLs), primarily CD8+ cytotoxic T lymphocytes (CTLs), and natural killer cells (NK cells). Although they rely on different activation and recognition mechanisms, both NK cells and CTLs must establish a similar intercellular connection with tumor cells during cytotoxic lysis, known as the immunological synapse (IS).6 7 Tumors can achieve cytoprotection through the pathway of IS defects to evade immune killing.8 The formation, maintenance, and function of ISs are highly dependent on rearrangement of the actin cytoskeleton between effector and target cells.9 10 Several studies have elucidated the critical role of effector cell cytoskeletal reorganization in IS. However, the role of the target cell cytoskeleton in CL-mediated lysis has gradually attracted attention over the past decade.11 12 Breast cancer cells can inhibit NK cell-mediated death by regulating actin reorganization near the IS during NK cell recognition and binding.13 However, studies focusing on tumor actin cytoskeleton-related proteins that mediate immune evasion are limited. Human cytomegalovirus inhibits actin polymerization mediated by the WAVE2 complex through UL135 protein expression, leading to host cell actin cytoskeleton disassembly, thereby suppressing IS formation and evading immune cytotoxicity.14 Similarly, the Nef gene of HIV type 1 interacts with the host cell’s non-receptor tyrosine kinases, Hck and Lyn, affecting the cytoskeleton and mediating HIV-1 immune evasion.15 This led us to hypothesize that tumors share a mechanism of immune evasion with these viruses by using the key role of cytoskeleton-related proteins in cytoskeletal regulation. In this study, we identified the expression of the cytoskeleton-related protein LASP1 (LIM and SH3 domain protein 1) in four types of digestive system tumor samples (stomach adenocarcinoma, colon adenocarcinoma, esophageal squamous cell carcinoma (ESCC), and liver hepatocellular carcinoma) from patients receiving anti-programmed cell death protein 1 (PD-1) immunotherapy, which may be associated with the response to immunotherapy. Further correlation analysis of the infiltrated CLs in GC revealed that patients with low tumor LASP1 expression did not benefit from CL infiltration. In vivo experiments confirmed that LASP1 knockout in tumor cells induced CD8+ T-cell exhaustion and significantly inhibited the tumor response to anti-PD-1 immunotherapy. LASP1 acts as a scaffold protein to remodel cytoskeletal and cell membrane lipids by regulating the Arp2/3 complex at immunological synapses. Finally, simvastatin combined with anti-PD-1 therapy reversed immunotherapy resistance in LASP1-deficient tumors. Our study further reveals the significant role of cytoskeletal changes on the target cell side in tumor immune evasion and provides a novel biomarker and strategy for GC immunotherapy. Materials and methods Patients and specimens Formalin-fixed, paraffin-embedded samples were retrospectively collected from 26 patients with advanced GC who underwent surgical resection and were treated with an anti-PD-1 antibody between September 2021 and November 2022 at The First Affiliated Hospital of Wenzhou Medical University. Responses to anti-PD-1 antibodies were assessed using Response Evaluation Criteria in Solid Tumors (V.1.1) and determined at the 6-month time point after treatment initiation. Cohort 1, from The Second Affiliated Hospital of Wenzhou Medical University (Wenzhou, China), which included 529 GC tissue samples and 57 paired adjacent normal tissue samples (at least 10 cm from the negative margin), was used to detect LASP1 expression. Cohort 2 from The First Affiliated Hospital of Wenzhou Medical University (Wenzhou, China), which included 107 GC tissue samples, was used to detect exhausted CD8+ T cells and LASP1 expression. Patients in cohorts 1 and 2 underwent surgical resection for GC between December 2006 and July 2011, and between March and November 2016, respectively. Follow-up information for the patients in cohort 1 was retrospectively collected. Tissues were fixed in formalin, embedded in paraffin, and confirmed by histopathological analysis. The tissue microarray (TMA) was constructed as described previously.16 Blood samples were obtained from healthy adult volunteers after obtaining written informed consent. Quantitative PCR 15 pairs of GC tumor tissues and adjacent normal tissues were collected and preserved in the RNAlater reagent (Invitrogen). RNA was extracted using TRIzol (Invitrogen) and reverse transcribed to complementary DNA (cDNA) (TOYOBO LIFE SCIENCE, #FSQ-101) according to the manufacturer’s instructions. The transcriptional levels of LASP1 (FP: 5′‐CGAGAAGAAGCCCTACTGCAA‐3′; RP: 5′‐CTGCCACTACGCTGAAACCT‐3′) and GAPDH (FP: 5′‐CAGGGCTGCTTTTAACTCTGGTAA‐3′; RP: 5′‐GGGTGGAATCATATTGGAACATGT‐3′) were detected by real-time fluorescence quantitative PCR according to the manual (QuantiNova SYBR Green PCR Kit, QIAGEN). Immunohistochemistry Immunohistochemistry was performed using TMA to detect the expression of LASP1 (#sc-374059, Santa Cruz Biotechnology), CD56 (#ab187274, Abcam), and CD8 (#IR623, Dako). These methods have been described previously.16 A multiplex immunohistochemistry kit (Panovue) was used to detect PD-1+ TIM-3+ exhausted CD8+ T cells. Briefly, after dewaxing and hydration, slides were treated with 3% H2O2 to inactivate endogenous peroxidases. After blocking and antigen retrieval, the slides were incubated for 30 min with primary antibodies (anti-human CD8, #70306, CST; anti-human TIM-3, #45208, CST; anti-human PD-1, #86163, CST) and then incubated with Horseradish Peroxidase (HRP)-labeled secondary antibodies for 10 min. The staining was repeated sequentially for each primary antibody. After the final round of staining, the TMA was stained with 4',6-diamidino-2-phenylindole (DAPI) and coverslipped with anti-fluorescence quenching mounting medium. The histochemical score for LASP1 was determined by two pathologists blinded to the clinical data. Patients were then divided into two groups for each target protein (the cut-off value was determined by the median), and the results of expression level (low or high) depending on H-scores from both pathologists showed over 95% consistency. The counts of CD57+ or CD8+ infiltrating cells and PD-1+ TIM-3+ exhausted CD8+ T cells per unit area (cells/mm2) were calculated using QuPath software V.0.5.0.17 Bioinformatics analysis The single-cell RNA sequencing (scRNA-seq) data and clinical information for anti-PD-1 treatment cohorts of four digestive system tumors samples (GC, colorectal cancer (CRC), ESCC and hepatocellular carcinoma (HCC)) were obtained from the European Nucleotide Archives database (PRJEB40416, stomach adenocarcinoma),18 Mendeley data (DOI: 10.17632/skrx2fz79n.1, liver hepatocellular carcinoma)19 and the Gene Expression Omnibus (GEO) database (GSE203115, esophageal carcinoma; GSE236581, colon adenocarcinoma).20 21 A larger scRNA-seq dataset containing 28 primary gastric tumor tissues was obtained for a more comprehensive analysis of the immune environment (GSE183904).22 The gene set for actin cytoskeleton-related proteins was obtained from MSigDB (CYTOSKELETON: GO:0005856). LASP1 expression data for different organs and tumors were obtained from Proteinatlas (www.proteinatlas.org) and GEPIA (gepia.cancer-pku.cn). CD8+ T cell and NK cell infiltration in the Omnibus (Omnibus (GEO) dataset GSE84437 was quantified using single-sample Gene Set Enrichment Analysis. RNA sequencing (RNA-seq) of mock and LASP1-KO SGC-7901 was conducted by Novogene (Tianjin, China). DESeq software was used to calculate the differential gene transcripts between different samples. The R package ggplot2 was used for Gene Set Enrichment Analysis. Cell lines and cell culture SGC-7901 cells were cultured in Dulbecco’s Modified Eagle’s Medium (Gibco) containing 10% fetal bovine serum (Gibco); HGC-27 and MFC cell lines were cultivated in RPMI-1640 (Gibco) containing 10% fetal bovine serum. All cell lines were obtained from The Cell Bank of the Chinese Academy of Sciences and were Mycoplasma-free. Peripheral blood mononuclear cells (PBMC) were isolated using Ficoll-Paque (Cytiva) and cultivated in X-VIVO 15 (Lonza). The cells were cultured at 37°C in a humidified 5% CO2 incubator. As for lipid manipulation, 15 mg/mL 7-ketocholesterol (HY-113342, MCE) or cholesterol (HY-N0322, MCE) stock solution was diluted in a 50 mg/mL methyl-β-cyclodextrin (HY-101461, MCE) solution at 80°C to a final concentration of 1.5 mg/mL. A volume of 15 µL of the lipid solution was then added to 1 mL of cell medium containing at 37°C for 30 min. Knockout and overexpression of LASP1 LASP1-HA, LASP1-ΔNEB-HA, LASP1-ΔSH3-HA cDNA was inserted into the pcDNA3.1 vector and transfected into cells using lipo2000 (Invitrogen) to overexpress LASP1. As for the overexpression of mouse LASP1, mLASP1-HA, mLASP1-ΔNEB-HA, mLASP1-ΔSH3-HA cDNA was inserted into the pLVX vector and was packaged into lentivirus for transduction of MFC cells. The empty vector was used as the negative control. The single guide RNA sequence for CRISPR-Cas9 to knock out LASP1 (human LASP1: 5′‐CTGAACATGAAGAACTACAA‐3′; mouse LASP1: TCCCAAGCAGTCCTTCACCA) was inserted into plasmid pSpCas9 (BB)−2A-Puro (PX459), a gift from Feng Zhang (Addgene plasmid #48139) to construct LASP1 knockout cell line. An empty vector was used as the mock control. Animal experiments All animal experiments were performed in accordance with The National Institutes of Health Guide for the Care and Use of Laboratory Animals with the approval of The Experimental Animal Center of the First Affiliated Hospital of Wenzhou Medical University (No. WYYY-AEC-2024-212). 6-week-old male 615 mice were obtained from the Blood Institute of the Chinese Academy of Medical Sciences (Tianjin, China). Mice were randomly divided into two groups and subcutaneously or orthotopically injected with LASP1-KO or control MFC cells (106 cells in 100 µL phosphate-buffered saline or 50 µL matrigel). After 9 days, when the tumors grew to 100 mm3 on average, the tumor-bearing mice in each group were randomly divided into indicated groups and intraperitoneally injected with anti-PD-1 antibody (3 mg/kg, BE0273, Bio X Cell) or control IgG (BE0089, Bio X Cell) every 3 days. As for the combination therapy with statin and anti-PD-1, simvastatin (10 mg/kg, HY-17502, MCE) or vehicle was administered daily via oral gavage in mice. Subcutaneous tumor volume was monitored every 3 days using a digital caliper according to the following formula: tumor volume (mm3)=0.5×length×width2. The mice were euthanized when the tumor volume reached 2,000 mm3 and the experiment was terminated 18 days after the injection of tumor cells. Tumors were isolated, weighed, and the tumor immune microenvironment was analyzed using flow cytometry by time-of-flight (CyTOF). Cytometry by time-of-flight Tumor tissues were digested into single-cell suspensions using a tumor dissociation kit (Miltenyi Biotec). Antibody labeling with the indicated metal tag was performed using the MaxPAR Antibody Labeling kit (Fluidigm). The conjugated antibodies were titrated to determine optimal concentrations prior to use. CyTOF staining panels are shown in online supplemental table 2. Cells were stained with 250 nM cisplatin (Fluidigm) for 5 min on ice to exclude dead cells and then incubated in Fc receptor blocking solution before staining with a surface antibody cocktail. The cells were washed and fixed overnight in an intercalation solution (MaxPar Fix and Perm Buffer containing 250 nM 191/193Ir, Fluidigm). After fixation, cells were treated with Perm buffer (eBioscience) and stained with an intracellular antibody cocktail for 30 min on ice. Cells were washed, resuspended in deionized water, added to 20% EQ beads (Fluidigm), and analyzed using a mass cytometer (Helios, Fluidigm). Data for each sample were debarcoded from raw data using a doublet-filtering scheme with unique mass-tagged barcodes. Each .fcs file generated from different batches was normalized using the bead normalization method. Data were manually gated using the FlowJo software to exclude debris, dead cells, and doublets, leaving live single immune cells. The X-shift clustering algorithm was applied to all cells to partition them into distinct phenotypes based on marker expression levels. The cell type of each cluster was annotated according to its marker expression pattern on a heat map of cluster versus marker. Statistical analysis was performed to determine the frequency of the annotated cell population using the t-test. Western blot Radio Immunoprecipitation Assay (RIPA) containing phenyl methane sulfonyl fluoride (PMSF) was used to extract total cell proteins. Protein samples were subjected to 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred onto polyvinylidene fluoride membranes. The membrane was blocked and incubated with either anti-LASP1 (sc-374059, Santa), anti-GAPDH (2118, CST), anti-ARPC2 (15058–1 AP, Proteintech), anti-ARP2 (HA721348, HUABIO), anti-ARP3 (ET7107-97, HUABIO), anti-β-actin (66009–1 Ig, Proteintech) overnight at 4°C. After washing, the secondary antibody was added, and the blots were washed before visualization using enhanced chemiluminescence detection reagents (Thermo Fisher Scientific). In vitro killing assay PBMCs were activated using T-Activator CD3/CD28 (11 161D, Gibco) for 24 hours before co-incubation with PKH67 (HY-D1421, MCE) stained tumor cells at an Effect:Target (E:T) ratio of 10:1. After 16 hours of co-incubation, all cells were collected, washed, and stained using the Annexin V-APC/7-AAD Apoptosis Kit (AT105, Multi Sciences). The samples were analyzed using a FACSCanto II Flow Cytometer (BD Biosciences). FITC+ tumor cells were gated, and apoptosis was analyzed using FlowJo V.10.8 software. Flow cytometry PBMCs were co-incubated with tumors at an E:T ratio of 2:1 for 3 days to induce exhausted CD8+ T cells in vitro, as described previously.23 After 3 days of incubation, all cells were collected, washed, and stained with antibodies against CD8 (301006, BioLegend), PD-1 (329906, BioLegend), LAG-3 (369330, BioLegend), TIM-3 (17–3109-42, eBioscience), and 7-AAD (420404, BioLegend). On the third day, following a 4-hour treatment of cells with Brefeldin A (420601, BioLegend), the functional activity of CD8 T cells was assessed by measuring interferon (IFN)-γ production using anti-IFN-γ antibody (202527, BioLegend). Also on the third day, the proliferative capacity of CD8+T cells was evaluated by detecting MKI67 (Ki-67) expression using an anti-Ki-67 antibody (350513, BioLegend). For T-cell detection in murine tumors, single-cell was stained with antibodies against CD45 (103114, BioLegend), CD8 (100706, BioLegend), PD-1 (135206, BioLegend), TIM-3 (119706, BioLegend), LAG-3 (125240, BioLegend), GZMB (372207, BioLegend), Ki-67 (652405, BioLegend) and 7-AAD (420404; BioLegend). A degranulation assay was performed to detect CD107A in the cells. Briefly, the CD107A antibody (328607, BioLegend) and Brefeldin A (420601, BioLegend) were added to PBMCs and tumors 2 hours after co-incubation. 4 hours later, all cells were collected, washed, and stained with antibodies against CD3 (300426; BioLegend), CD8, CD56 (318310; BioLegend), and 7-AAD. Cytotoxicity was analyzed by measuring the level of granzyme B (GZMB) in the tumor cells. PKH67-labeled tumors were co-incubated with PBMCs for 6 hours, collected, washed, and stained with GZMB antibody. The PBMC activation assay was performed by detecting both CD69 and CD137 on CD8+ T cells and CD69 on NK cells. PBMCs without pre-activation were co-incubated with tumors at an E:T ratio of 10:1 for 24 hours and stained with CD3, CD8, CD137 (309810, BioLegend), CD69 (388604, BioLegend), CD56 (318304, BioLegend), and 7-AAD. The active perforin-1 protein (APB317Hu02, Cloud-Clone Corp) was used to assess the sensitivity of tumor cells to perforin-1. Specifically, active PFR-1 protein was added to the complete culture medium of tumor cells to a final concentration of 1 µg/mL. After a 30-minute incubation at 37°C, 7-Aminoactinomycin D (7-AAD) was used to detect the integrity of the cell membrane. All samples were analyzed on a FACSCanto II (BD Biosciences), and the data were analyzed using FlowJo V.10.8 software. Conjugate formation assay CD8+ T and NK cells were isolated from PBMCs using the MojoSort Human CD8+ T Cell Isolation Kit (480129, BioLegend) and the Dynabeads Untouched Human NK Cell Isolation Kit (11349D, Invitrogen), followed by staining with CFDA-SE (HY-D0938, MCE). The fluorescence-labeled effector and target cells were used to perform the conjugate formation assay as previously described.24 Briefly, a single-cell suspension of PKH26 (HY-D1451, MCE)-labeled tumor cells and CFDA-SE-labeled effector cells were mixed at an E:T ratio of 2:1 and incubated at 37°C for 30 min. The relative proportions of double-positive effector-target cell conjugations in each tube were determined by two-color flow cytometry using a FACSCanto II (BD Biosciences) and analyzed using FlowJo V.10.8 software. Immunofluorescence and live-cell imaging For immunofluorescence staining, the effector-target cell conjugates were gently mounted on PLL–coated coverslips, fixed, and permeabilized. Cells were then incubated with antibodies (anti-CD2, HA722484, HUABIO; anti-perforin-1, HA723797, HUABIO; anti-LASP1, sc-374059, Santa; anti-ARP2, HA721348, HUABIO), followed by incubation with secondary antibodies (Alexa Fluor 488 conjugate (A-11008, Invitrogen), Alexa Fluor 647 conjugate (A-21235, Invitrogen)), and phalloidin Alexa Fluor 568 conjugate (A-12380, Invitrogen). Finally, the glass coverslips were mounted with mountant (#P36966, Invitrogen), and digital images were visualized under a fluorescence microscope (Axio Observer 5, Zeiss) and a laser scanning confocal microscope (STELLARIS 5, Leica). The synaptic locations of CD2 and perforin-1 were determined when the fluorescence density on the synaptic side exceeded 70% that of the entire cell. 10 fields containing at least 30 conjugates were obtained, and the percentage of synaptic locations was analyzed. SGC-7901 cells were transfected with the pHBLV-LifeAct-mCherry lentivirus (HanBio Technology, Shanghai, China) for live-cell imaging. Regarding the observation of ligand polarization at immunological synapses, tumor cells are labeled with antibodies (APC anti-HLA-ABC, 311410, BioLegend; APC anti-human PD-L1, 329708, BioLegend; APC anti-human ICAM-1, 322712, BioLegend) at 37°C for 30 min before the co-incubation and live-cell imaging. PKH67-labeled effector cell were co-incubated with target cells at an E:T ratio of 2:1 for 30 min. Images of the conjugates were acquired using a laser-scanning confocal microscope (STELLARIS 5, Leica or ImageXpress Micro Confocal, Molecular Devices). Actin quantification has been described previously.13 The fluorescence intensity of LifeAct-mCherry was measured using Leica Application Suite X 3.7 (Leica). Laurdan staining and imaging were performed as described previously.25 Tumor cells were incubated with 10 µM of the lipid packing sensors Laurdan (HY-D0080, MCE) for 60 min at 37°C before being co-incubated with effector cells. Two separate imaging channels (emission wavelength of 415–460 and 470–530) were used to measure the lipid ordering of membranes. Generalized polarization (GP) values were calculated using the Fiji software script (https://github.com/WIS-MICC-CellObservatory/LaurdanInLiveCellImaging/).26 Atomic force microscope Atomic force microscope (AFM) images were captured using a Nanowizard V (JPK-Bruker) with a 5 µm diameter spherical AFM probe (SAA-SPH-5UM, Bruker) in force mapping mode (stiffness mapping). The AFM images were recorded at room temperature, and the maximum force applied to image the samples was 1 nN. The data were processed with JPK Data Processing software using the Hertz/Sneddon model with a 0.5 Poisson ratio as a fit parameter. ARPC2 knockdown The siRNA oligos for ARPC2 knockdown (siARPC2#1-1:5′‐GCACAGUGUUUAAGGAUGAUU‐3′; siARPC2#2-1:5′‐CCAUGUAUGUUGAGUCUAAUU‐3′) were synthesized. The siRNAs were transfected into GC cells using Lipofectamine 2000 (11668019, Invitrogen) at 20 nM final concentration. Co-immunoprecipitation For exogenous expression, cells were lysed for 10 min on ice with NP40 buffer (P0013F, Beyotime) containing a protease inhibitor cocktail (S8830, Sigma‐Aldrich) 24 hours after transfection. After centrifugation at 12,000 g for 10 min at 4°C, the supernatant was mixed with anti‐HA magnetic beads (88837, Thermo Scientific) and incubated at room temperature for 30 min with mixing. The protein was eluted using lane marker non-reducing sample buffer (39001, Thermo Scientific) at 95°C for 10 min. A total of 50 mM 1,4-Dithiothreitol (DTT) was added after the beads were removed. Statistics All statistical analyses were performed using IBM SPSS Statistics V.17.0 and R software (V.4.2.2). A Kaplan-Meier analysis was performed. Univariate analysis was conducted using the Cox proportional hazards regression and logistic regression models. Statistical significance between the groups was determined using a two-tailed unpaired or paired Student’s t-test. Data are presented as the mean±SEM. Statistical significance was set at p<0.05. Result LASP1 has been identified as a cytoskeleton-related protein associated with tumor immunotherapy The confounding effects of different cell types within tissues in traditional bulk RNA-seq can be avoided by using scRNA-seq. Therefore, to explore the genes related to the immunotherapy response in tumor epithelial cells, we collected four types of digestive system tumor samples (GC, CRC, ESCC, and HCC) from patients receiving anti-PD-1 immunotherapy and performed single-cell transcriptome sequencing (figure 1A). We identified the epithelial cells of tumor tissues and compared the differentially expressed genes between the epithelial cells from responders and non-responders to immunotherapy (p<0.05, |log2 fold change|≥0.4) (figure 1B and C). Finally, we obtained 18 common differentially expressed genes (DEGs) in all four datasets (figure 1C, online supplemental table 1). Among the DEGs, we identified two cytoskeleton-related genes, LASP1 and VIM (figure 1D). The downregulation of VIM in responders was consistent with previous findings that tumors can evade NK cell immune surveillance via VIM-mediated cytoskeleton remodeling,27 while the upregulation of LASP1 in responders indicates that it might also play a crucial role in the efficacy of immunotherapy. The different expression levels of LASP1 in the epithelial cells of the four tumor types are shown (figure 1E). Furthermore, the expression of LASP1 in epithelial cells was markedly diminished following anti-PD-1 therapy compared with pretreatment levels (online supplemental figure 1A,B), indicating that its deficiency may play a role in the development of acquired resistance to immunotherapy. LASP1 has been identified as a cytoskeleton-related protein associated with tumor immunotherapy. (A) Different cell types from four different kinds of tumor were projected onto the UMAP plot. (B) Epithelials from responder and non-responder to immunotherapy were projected onto the UMAP plot. (C) Differently expressed gene in epithelials from responder and non-responder were analyzed. (D) The common differentially expressed genes were intersected with cytoskeleton-related genes. (E) Different expression level of LASP1 in epithelials from responder and non-responder were analyzed. (F) The expression level of LASP1 in different organs was analyzed. (G) Different expression level of LASP1 in different kinds of tumor tissues and paired normal tissues. All values are presented as the mean±SEM. ***p<0.001. COAD, colon adenocarcinoma; CRG, cytoskeleton-related gene; ESCC, esophageal squamous cell carcinoma; FC, fold change; LIHC, liver hepatocellular carcinoma; NK, natural killer; STAD, stomach adenocarcinoma; TPM, Transcripts Per Million; UMAP, uniform manifold approximation and projection. We analyzed the expression of LASP1 in all normal organs and found that LASP1 was highly expressed in the gastrointestinal organs, particularly in the stomach (figure 1F). Additionally, we analyzed the transcriptome sequencing data of 33 tumors in the The Cancer Genome Atlas (TCGA) database and found that compared with normal tissues, LASP1 expression was significantly higher in 9 types of tumors, including all 4 gastrointestinal tumors and HCC (figure 1G). Moreover, immunohistochemistry and Reverse Transcription Quantitative Polymerase Chain Reaction (RT-qPCR) analyses both demonstrate significantly increased LASP1 expression in tumor tissues compared with paired adjacent non-tumor tissues (online supplemental figure 1C–E). These results indicate that LASP1 plays an important role in physiological processes, tumorigenesis, and tumor development in the gastrointestinal tissues. The deficiency of LASP1 prevents patients from benefiting from cytotoxic lymphocyte infiltration and affects the efficacy of anti-PD-1 immunotherapy Given the important role of LASP1 in physiological processes, tumorigenesis, and tumor development in gastrointestinal tissues, we examined LASP1 expression in tumor tissues of 492 patients with GC and analyzed its relationship with patient prognosis. We found that LASP1 expression was not directly related to patient prognosis in our cohort or in external datasets (TCGA and GSE84437) (figure 2A–C). The deficiency of LASP1 prevents patients from benefiting from cytotoxic lymphocyte infiltration and affects the efficacy of anti-PD-1 immunotherapy. (A–C) Kaplan-Meier analysis of overall survival in patients was performed based on the LASP1 expression level using own cohort (A), TCGA cohort (B) and GSE84437 cohort (C). (D,E) Univariate Cox analysis of overall survival in patients based on the stratified LASP1 expression level or CL infiltration using own cohort (D) and GSE84437 cohort (E). (F,G) Different expression levels of LASP1 in gastric cancer tissue were detected by IHC and analyzed according to immunotherapy response. (H) The progression-free survival of gastric cancer patients with different LASP1 expression levels was analyzed. (I) Subcutaneous and orthotopic tumors in the indicated groups were photographed. (J) Growth curves of subcutaneous tumors in indicated groups of mice. (K,L) Subcutaneous and orthotopic tumors were weighed and their weights were calculated. All values are presented as the mean±SEM. *p<0.05; **p<0.01; ***p<0.001. CL, cytotoxic lymphocyte; CR, complete response; IHC, Immunohistochemistry; NK, natural killer; PD, progressive disease; PD-1, programmed cell death protein 1; PR, partial response; SD, stable disease; TGCA, The Cancer Genome Atlas. Because CLs (CD8+ T cells and NK cells) play an important role in antitumor immunity and immunotherapy, we also examined the infiltration levels of CD8+ and CD57+ cells in these tissues (online supplemental figure 2A). Patients with higher CL infiltration had a significantly better prognosis (CD8: HR=0.63 (0.48–0.82), p=0.001; CD57: HR=0.77 (0.59–1.01), p=0.055; figure 2D). Interestingly, stratified analysis of patients with GC by CL infiltration levels revealed that in patients with GC with different levels of CL infiltration, the relationship between LASP1 expression and prognosis was the opposite. In patients with high CL infiltration, those with high LASP1 expression had a significantly better prognosis (CD8+ cell high: HR=0.70 (0.48–1.01), p=0.056; CD57+ cell high: HR=0.68 (0.47–0.98), p=0.039; figure 2D). Conversely, in patients with low CL infiltration, high LASP1 expression in the tumor was a risk factor (CD8+ cell high: HR=1.55 (1.03–2.32), p=0.035; CD57+ cell high: HR=1.50 (1.01–2.24), p=0.045; figure 2D). Results from an external dataset (GSE84437) further validated our findings (figure 2E). The impact of LASP1 on prognosis in a CL-dependent manner led us to hypothesize that LASP1 may also affect the efficacy of immunotherapy. We measured LASP1 expression in tumor tissues of 26 patients with advanced GC treated with anti-PD-1 immunotherapy (figure 2F). The results showed that the response to anti-PD-1 antibody was significantly associated with high LASP1 expression in tumor cells (PD vs SD: p<0.001; SD vs PR/CR: p=0.002, figure 2G). Further analysis revealed that patients with high LASP1 expression in tumor cells showed improved progression-free survival on receiving anti-PD-1 therapy (median PFS: LASP1 low vs LASP1 high=3 months vs 8 months) (p=0.03, HR=0.39 (0.16–0.95); figure 2H). The LASP1-knockout MFC cell line was constructed (online supplemental figure 2B) and subcutaneously and orthotopically injected into syngeneic 615 mice, with mock-treated MFC used as a control (figure 2I). Anti-PD-1 immunotherapy lost tumor-suppressive effect in the LASP1-KO group (figure 2J–L). The deficiency of LASP1 reprograms the immune environment To further investigate the effect of LASP1 deficiency on the efficacy of anti-PD-1 immunotherapy, a comprehensive CyTOF panel was used to profile the immune microenvironment of mock and LASP1-KO mouse subcutaneous tumors (online supplemental table 2). Using R software, unsupervised clustering analysis of CD45+ immune cells identified 27 clusters and was finally classified into 8 immune cell types (figure 3A and B). Overall, compared with the mock group, LASP1-KO tumors showed significantly increased infiltration of T cells and dendritic cells (figure 3C). Further classification of lymphocyte subsets revealed that in LASP1-KO tumors, the proportions of CD8+ exhausted T cells (C05 and C06) and immunosuppressive CD4+ regulatory T cells (Tregs) (C08) were significantly increased, accompanied by increased infiltration of CD4+Tef (C07) (figure 3D, online supplemental table 3). The analysis of CD8+ T cells in subcutaneous tumors demonstrated an increased infiltration of exhausted T cells in LASP1-KO tumors (online supplemental figure 3A,B), which was associated with decreased levels of GZMB and Ki-67 (online supplemental figure 3C–E). The deficiency of LASP1 reprograms the immune environment. (A) Immune cells from subcutaneously transplanted tumors in the mock and LASP1-KO groups were analyzed using CyTOF (n=3). Cell populations were clustered, defined manually, projected onto the t-SNE space, and assigned specific colors. (B) Heatmap showing differential marker expression in CD45+ TIL clusters identified by analysis of CyTOF data. (C) The frequency of T cells, B cells, NK cells, macrophages, DCs, neutrophils, eosinophils, and other cells in total cells were analyzed. (D) The proportion of different lymphocyte subsets in total cells was analyzed based on differential marker expression. (E–G) Subtypes of CD4+T cells (E), CD8+ T cells (F), and NK cells (G) were projected onto the UMAP plot. (H) Patients were ranked according to the different expression levels of LASP1 in epithelial cells. (I) Portions of different subtypes of CD4+T cells, CD8+ T cells, and NK cells were analyzed according to LASP1 levels. (J–L) The intensity and proportion of exhausted CD8+ T cells in gastric cancer tissue were detected and analyzed. All values are presented as the mean±SEM. *p<0.05; **p<0.01. CyTOF, cytometry by time-of-flight; DC, dendritic cell; KO, knock out; NK, natural killer; TILs, tumor Infiltrating Lymphocytes; t-SNE, t-distributed stochastic neighbor embedding; UMAP, uniform manifold approximation and projection. Next, we verified the effect of LASP1 on the immune microenvironment of tissues from patients with GC. We performed preliminary clustering and grouping of 145,331 cells obtained from single-cell sequencing of 28 primary GC tissue samples, yielding 26,301 epithelial cells and 51,871 NK/T cells (online supplemental figure 3F,G). We manually annotated NK/T cells into CD8+ T, CD4+ T, and NK cells and further annotated these three subgroups (figure 3E–G, online supplemental figure 3H,I). The marker genes of these subgroups are shown in online supplemental figure 3J–L. Subsequently, we divided the samples into LASP1-high and LASP1-low groups based on LASP1 expression levels in epithelial cells (figure 3H). We found that the proportions of CD8+ exhausted T cells and immunosuppressive CD4+ Treg cells were significantly higher in the LASP1-low group than in the LASP1-high group (figure 3I), whereas the proportions of effective CD8+ Teff and CD4+ Tem were higher in the LASP1-high group than in the LASP1-low group. As for NK cells, CD56dim CD16hi NK cells, which exhibited a higher expression of exhaustion and effect markers, also had a higher proportion in the LASP1-low group (figure 3I). The correlation between LASP1 expression and CD8+ exhausted T cells in 107 GC tissues validated our previous findings (figure 3J–L). The deficiency of LASP1 in tumor cells enables them to resist cytotoxic lymphocyte-mediated lysis while effectively activating immune cells in vitro Because LASP1 deletion in tumor cells significantly inhibits the efficacy of immunotherapy in vivo, we tested the effect of LASP1 expression on the sensitivity of GC cells to CL-mediated lysis in vitro. LASP1 was knocked out in two GC cell lines, HGC-27 and SGC-7901, using Crispr-Cas9 technology, and simultaneously overexpressed (or complemented) LASP1 in wild-type (or LASP1-KO) cells (figure 4A). The results showed that after co-culturing with PBMCs for 16 hours, the apoptosis of LASP1-KO GC cells was significantly lower than that of mock GC cells (figure 4B–D). This difference in apoptosis disappeared when effector cells were separated from the target cells using a Transwell chamber, suggesting that this difference depended on direct contact between the effector cells and the target cells, rather than being mediated by cytokines (online supplemental figure 4A–C). Conversely, overexpression or complementation of LASP1 in wild-type (figure 4E) or LASP1-KO (figure 4F) GC cells increased or restored their sensitivity to CL-mediated lysis, respectively. We detected the cytotoxic molecule GZMB in target cells and found that LASP1-KO reduced GZMB accumulation after 6 hours co-culture with PBMCs (figure 4G–I). Interestingly, although the accumulation of GZMB in target cells was reduced, the degranulation of CL (CD8+ T cells and NK cells) in PBMCs co-cultured with LASP1-KO target cells for 6 hours was significantly elevated (figure 4J–L). Next, the detection of CD69 and CD137, two markers of CD8+ T-cell activation, after 24-hour co-cultured with tumor cells suggested that LASP1-KO target cells did not inhibit but more effectively activated, CD8+ T cells than mock cells (figure 4M–O). Upregulation of CD69 was also observed in NK cells (online supplemental figure D–F). The phenotype of cytotoxic evasion and excessive CL activation in LASP1-KO GC cells ultimately led to significant CD8+ T cells exhaustion (figure 4P–R), less IFN-γ production and lower Ki-67 (online supplemental figure 4G–I) in a prolonged (3 days) PBMC-tumor co-culture model, which could be reversed when target cells and effector cells were separated in a Transwell chamber (online supplemental figure 4J–L). This result indicates that LASP1 deficiency-induced T-cell exhaustion is dependent on direct cell-to-cell contact rather than mediation by cytokines secreted from tumor cells. The deficiency of LASP1 in tumor cells enables the tumor cells to resist cytotoxic lymphocyte-mediated lysis while effectively activating immune cells in vitro. (A) LASP1 knockout HGC-27 and SGC-7901 cell lines were constructed using the Crispr-Cas9. LASP1 was overexpressed or complemented in wild-type or LASP1-KO HGC-27 and SGC-7901 cell lines. (B–D) After co-culture with PBMC for 16 hours, the apoptosis of cancer cells in the mock group and the LASP1-KO group was analyzed by flow cytometry (n=3). (E,F) After co-culture with PBMC for 16 hours, the apoptosis of wild-type (E) or LASP1-KO (F) cancer cells with overexpressed or compensated LASP1 was analyzed by flow cytometry (n=3). (G–I) The accumulation of GZMB in mock or LASP1-KO cancer cells co-cultured with PBMC for 6 hours was analyzed by flow cytometry (n=3). (J–L) The degranulation levels of CD8+ T cells and NK cells after 6-hour co-culture with mock or LASP1-KO cancer cells were analyzed by flow cytometry (n=3). (M–O) The activation levels of CD8+ T cells after 24-hour co-culture with mock or LASP1-KO cancer cells were analyzed by flow cytometry (n=3). (P–R) The exhaustion levels of CD8+ T cells after 72-hour co-culture with mock or LASP1-KO cancer cells were analyzed by flow cytometry (n=3). All values are presented as the mean±SEM. *p<0.05; **p<0.01. GZMB, granzyme B; KO, knock out; LAG-3, lymphocyte-activation gene 3; MFI, mean fluorescence intensity; NC, negative control; OE, over expression; PBMC, peripheral blood mononuclear cell; TIM-3, T-cell immunoglobulin and mucin-domain containing molecule 3; WT, wild type. The deficiency of LASP1 in tumor cells leads to dysfunction of the cytoskeleton during synaptic formation To investigate the mechanism by which LASP1-deficient GC cells evade immune killing, we performed RNA-seq on mock and LASP1-KO SGC-7901 cells. Enrichment analysis of the DEGs suggested changes in cytoskeletal regulation and focal adhesions in LASP1-KO cells (figure 5A). We also compared the expression levels of secretory factors (TGFB1, IL-6, and CSF1), which may exert an influence on CL, but found no differences (online supplemental figure 5A). In addition, flow cytometry analysis revealed that the deficiency of LASP1 did not alter the overall expression levels of the inhibitory receptor (PD-L1) and HLA-ABC (online supplemental figure 5B–E). Therefore, we hypothesized that LASP1 deficiency might affect adhesion between the tumor and effector cells. We used 5(6)-Carboxyfluorescein diacetate succinimidyl ester (CFSE) -labeled effector cells and PKH26-labeled target cells to examine the ability of LASP1-KO or mock GC cells to form conjugates with effector cells. The results showed that LASP1-KO did not affect the binding ability of tumor cells to CD8+ T and NK cells (figure 5B and C). Immunofluorescence detection of the localization of the adhesion molecule CD2 on the cell side of the IS demonstrated that LASP1 deficiency in tumor cells did not affect the binding of CD8+ T cells (figure 5D and F) or NK cells (figure 5G, online supplemental figure 5F) to tumor cells. However, LASP1 deficiency in tumor cells significantly reduced the localization of the cytotoxic molecule perforin-1 at the IS in CD8+ T cells (figure 5E and F) and NK cells (figure 5G, online supplemental figure 5G). In addition, LASP1 deficiency in tumor cells did not alter the duration of IS induced by cytotoxic lymphocytes (online supplemental figure 5H–J). To clearly observe the dynamic cytoskeletal changes in tumor cells during IS formation, we constructed Lifeact-labeled SGC-7901 cells and performed live cell imaging. We found that after forming an IS with CD8+ T cells (online supplemental video 1) or NK cells (online supplemental video 3), F-actin at the IS of mock tumor cells started to disassemble and reorganize at approximately 20–30 min. In contrast, tumor cells lacking LASP1 did not exhibit dynamic cytoskeletal changes during IS formation in immune cells (online supplemental videos 2,4). A quantitative analysis of F-actin enrichment at the immunological synapses on the tumor side was performed after 30 min of co-culturing with CL to further substantiate this observed phenomenon (figure 5H–K, online supplemental figure 5K). The absence of LASP1 in tumor cells leads to dysfunction of the cytoskeleton during synaptic formation. (A) The differentially expressed genes in mock and LASP1-KO SGC-7901 cells were identified through RNA-seq analysis, and the pathway enrichment analysis of these differentially expressed genes was conducted. (B,C) The conjugate formation experiment was performed to evaluate the adhesion of mock or LASP1-KO cancer cells to effector cells (CD8+ T cells and NK cells) (n=3). (D–G) The synaptic localization of adhesion molecules CD2 (D) and effect molecules perforin-1 (E) of CD8+ T cells (F) or NK cells (G) that form conjugates with mock or LASP1-KO cancer cells was detected and analyzed by immunofluorescence (n=10). (H–K) Lifeact labeling live cell imaging was used to analyze the reorganization of the cytoskeleton at the synaptic junctions of cancer cells after conjugates with CD8+ T cells (J) or NK cells (K) (n=10). All values are presented as the mean±SEM. ***p<0.001. CFSE, 5(6)-Carboxyfluorescein diacetate succinimidyl ester; KO, knock outl; NK, natural killer; ns, no significance; RNA-seq, RNA sequencing. LASP1 acts as a scaffold protein to regulate Arp2/3 complex and remodel cytoskeletal at immunological synapses It has been well established that LASP1 interacts directly with the Arp2/3 complex, thus regulating the cytoskeleton.28 29 To further demonstrate the role of this dynamic LASP1-mediated cytoskeletal change in tumor cells during the immune killing process, we knocked down the key subunit, ARPC2, of the Arp2/3 complex (figure 6A). In vitro killing experiments showed that ARPC2 knockdown significantly increased the resistance of mock-transfected SGC-7901 cells to CL-mediated lysis. However, this phenomenon was not observed in LASP1-KO SGC-7901 cells (figure 6B). This proved our hypothesis that LASP1 may mediate dynamic cytoskeletal changes through the Arp2/3 complex, thereby playing a role in the immune-killing process. We constructed LASP1 deletion mutants lacking distinct functional domains (ΔLIM, ΔNEB, and ΔSH3) to investigate the functional domain responsible for the interaction between LASP1 and the Arp2/3-actin complex (figure 6C). However, we failed to express the LASP1 mutant with LIM domain deletion, possibly due to protein instability. Co-Immunoprecipitation(Co-IP) experiments showed that NEB domains were essential for the interaction between LASP1 and the Arp2/3-actin complex (figure 6D). Co-localization detection of LASP1 and ARP2 by immunofluorescence further confirmed this finding (figure 6E). Next, we performed endogenous-level complementation assays using LASP1 mutants with the deletion of different functional domains to evaluate their functions (online supplemental figure 6A). We found that rescue with wild-type LASP1 alone, but not LASP1-ΔNEB or ΔSH3, was able to enhance tumor cell apoptosis mediated by CLs (figure 6F) and GZMB accumulation (figure 6G) and reduce the induction of CD8+ T-cell exhaustion (figure 6H). Consistently, only the rescue of wild-type LASP1 restored the dynamic changes in tumor cells at the ISs, regardless of the presence of CD8+ T cells or NK cells (figure 6I and J; online supplemental figure 6B). Therefore, we suggest that LASP1 acts as a scaffold protein connecting the Arp2/3-actin complex via its neublin domain with other cytoskeleton-regulating proteins via the SH3 domain, thereby regulating the Arp2/3 complex and mediating dynamic cytoskeletal changes. To further substantiate the functional domain deletion mutants of LASP1 in vivo, mouse LASP1 mutants were reintroduced at levels approximating endogenous expression into mouse LASP1-KO MFC cells (online supplemental figure 6C). In alignment with in vitro results, only the reintroduction of wild-type mouse LASP1 reinstated anti-PD-1 resistance (figure 6K–M), facilitated the accumulation of exhausted T cells (figure 6N), and restored the diminished levels of GZMB and Ki-67 in CD8+ T cells (online supplemental figure 6D,E) within subcutaneous tumors originating from LASP1-KO MFC cells. LASP1 acts as a scaffold protein to regulate Arp2/3 complex and remodel cytoskeletal at immunological synapses. (A) siRNA was used to knock down the expression of ARPC2. (B) After co-culture with PBMC for 16 hours, the apoptosis of ARPC2-knockdown cancer cells in the mock group and the LASP1-KO group was analyzed by flow cytometry (n=3). (C) LASP1 mutants with the deletion of different functional domains were constructed. (D) Co-immunoprecipitation experiments were conducted to identify the specific functional domain of LASP1 responsible for its interaction with the ARP2/3-actin complex. (E) Co-localization between LASP1 mutants and ARP2 was detected by immunofluorescence. (F–H) Apoptosis mediated by CLs (F), GZMB accumulation (G), and the induction of CD8+ T-cell exhaustion (H) in cancer cells with different LASP1 mutant expression was analyzed using flow cytometry (n=3). (I,J) Lifeact-labeled live cell imaging was used to analyze the reorganization of the cytoskeleton at the synaptic junctions of cancer cells with different LASP1 mutants expressing after coagulating with CD8+ T cells (I) or NK cells (J) (n=10). (K) Subcutaneous tumor in the indicated groups were photographed. (L) Subcutaneous tumor were weighed and their weights were calculated (n=5). (M) Growth curves of subcutaneous tumors in indicated groups of mice (n=5). (N) Proportion of PD-1+TIM-3+LAG-3+CD8+ T cell in subcutaneous tumors was analyzed using flow cytometry (n=5). All values are presented as the mean±SEM. *p<0.05;***p<0.001. CL, cytotoxic lymphocyte; GZMB, granzyme B; KO, knock out; LAG-3, lymphocyte-activation gene 3; NK, natural killer; ns, no significance; PBMC, peripheral blood mononuclear cell; PD-1, programmed cell death protein 1; TIM-3, T-cell immunoglobulin and mucin-domain containing molecule 3. LASP1-mediated cytoskeletal remodeling at the immunological synapse drives cell lipid membrane alterations Breast cancer cells MB-231 have been reported to generate densely packed postsynaptic lipid membranes by accumulating F-actin at the immunological synapse, thereby escaping cytotoxicity.25 We propose that cytoskeletal reorganization mediated by LASP1 induces modifications in cellular lipid membrane composition and dynamics. We used a cell-permeable packing sensor (Laurdan) to define lipid packing at the IS, which was represented by GP values. We found that in LASP1-complement cells, lipid packing at the cell membrane was significantly reduced alongside the clearance of actin at the immunological synapse, while this remodeling of actin and the lipid membrane was not observed in LASP1-deficient tumor cells (figure 7A–C, online supplemental figure 7A). The loss of ARP2/3 complex function or the deletion of the cytoskeleton-regulatory domain in LASP1 led to compromised dynamics of actin and plasma membrane lipid packing, suggesting that LASP1-mediated cytoskeletal modulation directly regulates these dynamics (online supplemental figure 7B,C). We used 7-ketocholesterol (7-Kc), an oxidized cholesterol that inhibits lateral diffusion and increases the stiffness of the cellular lipid membrane to block cytoskeleton-driven lipid membrane alteration.30 Compared with tumor cells treated with cholesterol, tumor cells treated with 7-Kc showed remodeling of the cytoskeleton, but maintained intact lipid membranes at the IS, regardless of conjugation with CD8+ T cells (figure 7D and E) or NK cells (figure 7F, online supplemental figure 7D). The correlation between the remodel of the cytoskeleton (F-actin intensity ratio) and lipids (ΔGP value) was disrupted by 7-Kc treatment (figure 7G). When co-culturing was performed with cholesterol or 7-Kc-treated tumor cells for 6 hours, 7-Kc treatment significantly reduced the accumulation of GZMB in LASP1-complement tumor cells; however, this effect was not observed in LASP1-deficient tumor cells (figure 7H). Given that changes in the cytoskeleton can alter the stiffness of tumor cells, thereby affecting their sensitivity to perforin-1,31 we investigated the effects of LASP1 deficiency on tumor cell stiffness and sensitivity to perforin-1. The results showed that LASP1 did not influence the stiffness of tumor cells (online supplemental figure 7E) or their susceptibility to free active perforin-1-mediated cytotoxicity (online supplemental figure 7F,G). Recent studies have shown that the cytoskeleton of tumor cells can drive the localization of the inhibitory ligand (HLA-ABC) to immunological synapses, thereby affecting the activity of NK cells.32 However, the deficiency of LASP1 did not affect the inhibitory ligand (HLA-ABC, PD-L1) or adhesion molecules (ICAM-1) localization at the immunological synapse (online supplemental figure H–J). Overall, this results suggests that LASP1 mediates the sensitivity to CL-mediated lysis not by affecting the overall stiffness of the cell and perforin-1 sensitivity, but by orchestrating dynamic cytoskeletal and lipid membrane remodeling within the immunological synapse. LASP1-mediated cytoskeletal remodeling at the immunological synapse drives cell membrane lipid alterations. (A–C) Laurdan staining was used to analyze the packing of lipid membrane at the synaptic junctions of cancer cells after conjugates with CD8+ T cells (B) or NK cells (C) (n=10). (D–F) Laurdan staining was used to analyze the packing of lipid membrane at the synaptic junctions of cholesterol or 7-Kc-treated cancer cells after conjugates with CD8+ T cells (E) or NK cells (F) (n=10). (G) The correlation between the remodeling of the cytoskeleton (F-actin intensity ratio) and lipid (ΔGP value) was analyzed in cholesterol or 7-Kc-treated cancer cells. (H) The accumulation of GZMB in cholesterol or 7-Kc-treated cancer cells co-cultured with PBMC for 6 hours was analyzed by flow cytometry (n=3). (I) Subcutaneous tumor with indicated treatment were photographed. (J) Growth curves of subcutaneous tumors with indicated treatment (n=5). (K) Subcutaneous tumor were weighed and their weights were calculated (n=5). All values are presented as the mean±SEM. ***p<0.001. 7-Kc, 7-ketocholesterol; GP, generalized polarization; GZMB, granzyme B; KO, knock out; NK, natural killer; ns, no significance; PBMC, peripheral blood mononuclear cell; PD-1, programmed cell death protein 1. Immune resistance in LASP1-deficient tumors is primarily attributed to the presence of a plasma membrane lipid barrier. It was hypothesized that pharmacological targeting of lipid metabolism, in conjunction with anti-PD-1 therapy, could potentially reverse this immune-resistant phenotype. Statins, which are clinically recognized inhibitors of cholesterol metabolism, have been widely documented to enhance antitumor immunity through various mechanisms.33 34 In LASP1-KO, tumor models, combination treatment with simvastatin and anti-PD-1 antibody demonstrated significantly superior efficacy compared with anti-PD-1 monotherapy (figure 7I–K). Discussion The fundamental purpose of immunotherapy is to enhance or retarget immune attacks on tumor cells to suppress or even eliminate the tumor.35 36 Previous studies have often focused on whether a positive immune microenvironment can be generated by immunotherapy without paying attention to the tumor cell response to CL-mediated lysis. Although many studies have been dedicated to predicting patient prognosis and responses to immunotherapy by evaluating the tumor’s immune environment,37 most GCs still show insufficient sensitivity to immunotherapy with immune checkpoint inhibitors. In this study, we found that LASP1 expression in tumor cells is a key factor in determining whether patients benefit from “hot tumors” or immunotherapy. LASP1 is a structural scaffold protein that is widely expressed in various cells, especially in digestive tract epithelial and immune cells, and plays an important role in physiology, pathology, and tumor development.38 LASP1 is involved in various physiological processes, such as gastric acid secretion and the development of neural synapse.39 40 Under pathological conditions, LASP1 promotes motility of arthritic fibroblast-like synoviocytes by interacting with the Arp2/3 complex.28 A large body of research suggests that LASP1 interacts with N-WASP to activate the Arp2/3 complex, thereby promoting tumor proliferation and migration.29 41 Research has indicated that myocardin-related transcription factors can enhance migration and metastatic invasion while making cancer cells more susceptible to the immune system by modulating the cytoskeleton.42 Similarly, we found that LASP1 acts as a scaffold protein to regulate the Arp2/3-actin complex and remodel the cytoskeleton through the nebulin repeat domain and SH3 domain during the lytic IS formation. The dual roles of LASP1 in promoting tumor invasion and mediating tumor immune killing also explain why the relationship between tumor LASP1 expression and patient prognosis is immune environmental dependent. Under the low immune pressure of “cold tumors,” LASP1 promotes tumor progression by enhancing tumor proliferation and migration, whereas under the high immune pressure of “hot tumors,” LASP1 deficiency in tumor cells enables them to evade immune cytotoxicity. This dual-role phenomenon may be widespread among cytoskeleton-related proteins and warrants further investigation. Cross-talk between epithelial-mesenchymal plasticity and tumor immune evasion has been recognized.43 Cytoskeleton-associated proteins may undergo regulation during epithelial/mesenchymal state transition. In lung cancer metastases, transforming growth factor-β mediates a non-canonical Epithelial-mesenchymal transition (EMT) phenotype via inducing actin depolymerization, facilitating immune evasion.44 In addition, aberrant skeleton-associated proteins may contribute to acquired resistance to immunotherapy. Currently, only a limited body of research has provided clear elucidation of the mechanisms by which dysregulation of the actin cytoskeleton contributes to immune evasion and resistance to immunotherapy. In soft tumor cells, reduced non-muscle myosin heavy-chain nine prevents perforin from drilling through the membrane, thereby mediating immune tolerance.31 Another study has shown that tumor cells can drive inhibitory ligands to localize to immunological synapses by accumulating the actin at IS, thus suppressing NK cell activity.32 CD8+ T and NK cells have been found to protect themselves from perforin-mediated lysis through densely packed lipid membranes.25 45 Actin accumulation at the synapses of immune cells promotes lipid raft aggregation.46 Similarly, the breast cancer cell line MDA-MB-231 can evade immunity by accumulating actin and generating densely packed postsynaptic lipid membranes at the IS on recognition and binding by NK cells.13 25 Interestingly, in this study, the disassembly and reorganization of actin and lipid membranes driven by LASP1 at the lytic IS were observed in CL-susceptible GC cells. In contrast, in LASP1-deficient tumor cells, this dynamic restructuring of F-actin and lipid membranes disappears, leading to dysfunction of the lytic IS. The pharmacological intervention targeting cholesterol metabolism through simvastatin, which consequently reduces lipid membrane integrity in tumor cells, significantly enhances the efficacy of anti-PD-1 therapy in LASP1-deficient tumors. This finding offers a mechanistic rationale and translational guidance for the combination of statins with immunotherapy in tumors characterized by cytoskeletal deficiency-mediated immune tolerance. The formation of the IS is an essential biological process for CL-mediated lysis and involves multiple components such as binding and adhesion of receptors, downstream signal transduction, and cytoskeleton polymerization.47 The formation of the IS not only mediates the stable binding of CLs and tumor cells and provides direction for the polarization of effector cells, but also ultimately mediates the directional release of effect molecules, such as granzymes and perforin.48 49 In our study, LASP1-deficient tumor cells form the IS with CLs, effectively activate CL, and promote their degranulation while maintaining low levels of the cytotoxic molecule GZMB within themselves through compromised remodeling of the cytoskeletal and lipid membranes at the IS. This suggests that tumor cells lacking LASP1 can be recognized by effector cells and can transmit activation signals. A similar phenomenon of cytoskeletal abnormalities mediating tumor immune evasion while promoting CL degranulation has also been observed in human non-small cell lung carcinoma.50 51 Although actin accumulation in target cells at immunological synapses can reduce GZMB accumulation in target cells and target cell death, it still effectively induces NK cell MTOC polarization and degranulation.13 This evidence suggests that cell skeleton abnormality-mediated cytotoxic resistance does not affect the activation of the CL by target cells and eventually leads to exhaustion of the CL, thus exacerbating resistance to immunotherapy.52 53 While LASP1 deficiency does not affect the polarization of inhibitory ligands such as HLA-ABC or PD-L1, nor adhesion molecules like ICAM-1, cytoskeletal abnormalities may still lead to immune cell exhaustion. This could occur through the mislocalization of other membrane molecules or via alternative pathways, including disrupted mechanotransduction signaling. The molecular mechanisms responsible for this phenomenon are not yet fully understood and require further investigation. In this study, disruption of the Arp2/3 complex by knockdown ARPC2 in susceptible GC cells resulted in the acquisition of a phenotype resistant to CL-mediated lysis. Mechanistically, we found that the interaction between LASP1 and the Arp2/3-actin complex depended on the nebulin repeat domain. Although LASP1 lacking the SH3 domain retained its interaction with the Arp2/3 complex, we found that neither LASP1 lacking the nebulin repeat domain nor the SH3 domain was able to regulate cytoskeletal dynamics. LASP1 has not been reported to directly promote actin polymerization or depolymerization. Instead, it is commonly regarded as a scaffold protein, that is, linked to other proteins (Zyxin, Talin, Dynamin, etc) to form complexes with the SH3 domain.54 Therefore, we propose that LASP1 integrates upstream signals by forming a cytoskeleton-regulating protein-LASP1-Arp2/3-actin complex, thereby regulating the activation or inhibition of the Arp2/3 complex and polymerization or depolymerization of actin. However, the inhibition of Arp2/3 complex-dependent actin polymerization by blocking CDC42 or N-WASP can reverse the immune resistance of tumor cells in immunoresistant breast cancer or lymphocytic leukemia.13 55 This emphasizes the importance of well-orchestrated spatiotemporal rearrangements of cytoskeletal dynamics mediated by cytoskeleton-related proteins on the target cell side during lytic IS formation. Therefore, further research is needed to elucidate these cytoskeletal dynamic changes and the role played by cytoskeleton-related proteins in this process. These efforts may provide effective targets for predicting or even intervening in the efficacy of immunotherapy in the future. Supplementary material Acknowledgements This study was supported by the Zhejiang Provincial Natural Science Foundation of China (Grant No. LQ23H160017), National Key Research and Development Program of China (Grant No. 2024YFA0918601), National Natural Science Foundation of China (NSFC) (Grant No. 32571083), and Major Scientific and Technological Innovation Project of Wenzhou Science and Technology Bureau (Grant No. ZY2022015), and Clinical Medicine Plus X Scholar Project of the Second Affiliated Hospital of Wenzhou Medical University, Zhejiang Key Laboratory of Intelligent Cancer Biomarker Discovery and Translation (Grant No. 2024ZY01003). Footnotes Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Ethics approval: All the retrospective analyzes based on FFPE tissues were approved by the Ethics Committee of The Second Affiliated Hospital of Wenzhou Medical University (No. 2021-K-06-02). This study is a retrospective analysis based on specimens already collected from patients and does not require informed consent from the patients. Data availability statement All data relevant to the study are included in the article or uploaded as supplementary information. References Associated Data Supplementary Materials Data Availability Statement All data relevant to the study are included in the article or uploaded as supplementary information.