HDAC inhibitors reverse YAP-driven immune resistance in NSCLC: mechanistic and translational evidence

Abstract

Background

As a key downstream effector of the Hippo signaling pathway, Yes-associated protein (YAP) has emerged as a pivotal regulator of tumor immune evasion and resistance to immune checkpoint inhibitors (ICIs). However, the relationship between YAP expression and immunotherapy outcomes in patients with non-small cell lung cancer (NSCLC), as well as its potential value as a therapeutic target, remains to be systematically investigated.

Methods

We retrospectively analyzed 141 patients with NSCLC who received immunotherapy, including 9 monoimmunotherapy and 132 immunotherapy plus chemotherapy or anti-angiogenic therapy. The expression of YAP in tumor tissues was evaluated by immunohistochemistry, and the association between YAP expression and immunotherapy resistance was assessed. Gene expression profiling following YAPand transcriptional coactivator with PDZ-binding motif (TAZ) knockdown was performed to identify potential YAP inhibitors and explore their underlying mechanisms of action. Furthermore, the therapeutic potential of a YAP inhibitor in reversing ICI resistance was evaluated in two patients with high YAP expression who had developed acquired resistance to immunotherapy.

Results

In our cohort, which predominantly received combination immunotherapy (93.62%), high YAP expression was associated with poor clinical outcomes to immunotherapy. Notably, YAP expression outperformed Programmed Cell Death Ligand 1 (PD-L1) as a predictive biomarker for treatment response. Through screening for potential YAP inhibitors, tucidinostat (TD) was identified as a compound capable of suppressing YAP activity. In two elderly patients with advanced NSCLC who had developed acquired resistance to immunotherapy, the combination of TD with ICIs successfully restored antitumor responsiveness. Mechanistic investigations further revealed that TD suppressed YAP expression via SNAI2-mediated epigenetic regulation, thereby blocking YAP-dependent tumor proliferation and immune evasion.

Conclusions

YAP is a strong predictive biomarker of response to immunotherapy in NSCLC. The combination of TD and ICIs shows promise in reversing immunotherapy resistance associated with high YAP expression, offering a potential strategy to overcome acquired resistance.

WHAT IS ALREADY KNOWN ON THIS TOPIC.

WHAT THIS STUDY ADDS

• We confirmed that YAP is a reliable predictive biomarker superior to Programmed death-ligand 1 (PD-L1) in patients with non-small cell lung cancer (NSCLC) treated with combination immunotherapy. The histone deacetylase inhibitor tucidinostat downregulates YAP expression by epigenetically activating SNAI2 and effectively reversed immune resistance in patients with high YAP expression, inducing sustained tumor regression.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Targeting YAP may enhance antitumor immune responses and supports the evaluation of tucidinostat as a strategy to overcome YAP-mediated immunotherapy resistance, potentially broadening treatment options for immune-refractory NSCLC.

Introduction

Immunotherapy, a groundbreaking cancer treatment, has revolutionized clinical approaches for numerous malignancies. The combination of immune checkpoint inhibitors (ICIs) with other therapeutic modalities enhances efficacy and prolongs disease control.¹ However, primary, adaptive, and acquired resistance remain significant challenges, limiting the universal success of immunotherapy.^{2 3} Currently, the biomarkers for predicting immunotherapy efficacy include programmed death-ligand 1 (PD-L1), tumor mutational burden (TMB), mismatch repair deficiency/microsatellite instability-high (MSI-H), and circulating tumor DNA (ctDNA).⁴ Among them, MSI-H was the first tumor-agnostic biomarker approved by the Food and Drug Administration for immunotherapy selection. However, its incidence in lung cancer is low (<5%),^{5 6} limiting clinical utility in this population. TMB serves as a surrogate for tumor immunogenicity but is subject to variability due to sequencing depth and tumor heterogeneity.⁷ ctDNA also depends heavily on detection sensitivity, and its concentration is often insufficient in early-stage or low-burden tumors, increasing the risk of false-negative results.⁸ PD-L1 is the most established biomarker in non-small cell lung cancer (NSCLC);⁹ clinical trials have shown that patients with a tumor proportion score (TPS) ≥50% can achieve objective response rates (ORR) of 33.1% to 45.2% with programmed cell death protein 1 (PD-1)/PD-L1 inhibitor monotherapy.10,12 Moreover, in chemoimmunotherapy settings, improved overall survival (OS) and progression-free survival (PFS) have been observed regardless of PD-L1 expression levels.^{13 14} Nevertheless, a significant proportion of advanced patients with NSCLC fail to benefit from PD-1/PD-L1 inhibitors, and effective strategies are lacking for those with acquired resistance.¹⁵ Therefore, precisely identifying potential responders, particularly by discovering reliable predictive biomarkers in the context of combination immunotherapy and investigating the mechanisms of immunotherapy resistance, as well as potential strategies for its reversal, are essential for improving patient outcomes.¹⁶

Yes-associated protein (YAP), a key downstream effector of the Hippo signaling pathway, regulates cell proliferation, differentiation, apoptosis, and cytoskeletal dynamics, playing a crucial role in tumorigenesis.17,19 Emerging evidence highlights its involvement in immune regulation, including macrophage polarization and T-cell function. YAP promotes Th1 and Th17 differentiation while suppressing regulatory T cells.²⁰ Within the tumor immune microenvironment, YAP upregulates PD-L1 expression, facilitating immune evasion.²¹ Additionally, it modulates inflammatory factors such as interleukin (IL)-6 and C-X-C motif chemokine ligand 5 (CXCL5), influencing immune cell recruitment and function.^{22 23} During immunotherapy, interferon (IFN)-γ activation of YAP contributes to resistance against PD-1 inhibitors, a process that can be counteracted by disrupting YAP phase separation.²⁴ Given its pivotal role in immune regulation and therapy resistance, this study evaluated the potential of YAP protein expression as a predictive biomarker for the efficacy of immunotherapy in lung cancer.

Histone deacetylases (HDACs) regulate gene transcription by modulating histone acetylation levels and are involved in various tumor-related biological processes, including cell cycle progression, apoptosis, differentiation, and immune response.^{25 26} Histone deacetylase inhibitors (HDACi) have primarily been used in the treatment of hematologic malignancies.²⁷ Tucidinostat (TD), an oral HDACi developed in China, was originally approved for the treatment of T-cell lymphomas and multiple myeloma. More recently, its clinical indications have expanded to several solid tumors such as NSCLC, colorectal cancer (CRC), breast cancer, and prostate cancer.²⁸ It has been approved for use in postmenopausal or perimenopausal women with estrogen receptor-positive and human epidermal growth factor receptor 2-negative breast cancer,²⁹ and its combination with sintilimab and bevacizumab has shown benefits in microsatellite-stable (MSS) or mismatch repair-proficient (pMMR) unresectable locally advanced or metastatic CRC,³⁰ demonstrating promising antitumor activity and clinical potential in solid tumors. However, its role in acquired resistance to immunotherapy in lung cancer remains unexplored. Therefore, this study further investigates the potential application of TD in reversing secondary immune resistance in lung cancer.

We analyzed YAP and PD-L1 expression in 141 NSCLC samples from patients treated with immunotherapy. In a cohort where 93.62% received combination immunotherapy, YAP emerged as a reliable predictive biomarker for treatment efficacy. Complementary in vitro and clinical studies demonstrated that TD effectively downregulates YAP expression, leading to the reversal of immunotherapy resistance and suppression of tumor progression. These findings establish YAP as a promising biomarker for predicting immunotherapy outcomes and suggest that TD-based combination therapy may represent a strategic approach to overcoming YAP-mediated resistance in NSCLC.

Materials and methods

Study patients

This study retrospectively enrolled 141 patients with pathologically confirmed NSCLC from Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology and Taizhou Affiliated Hospital of Nanjing University of Chinese Medicine between 2018.12 and 2024.3. Eligible patients met the following inclusion criteria: (1) availability of baseline and follow-up imaging data; (2) receipt of immunotherapy; (3) at least one measurable lesion as defined by Response Evaluation Criteria in Solid Tumors (RECIST) V.1.1; and (4) Eastern Cooperative Oncology Group performance status 0–1. The median follow-up duration was 17 months (1–49 months); no patients were excluded due to loss to follow-up or incomplete data, and all participants provided written informed consent for the use of their clinical samples. Baseline demographic and clinical characteristics were collected, including age, sex (111 patients were male and 30 were female), tumor type and stage, and history of underlying diseases, in order to provide a comprehensive description of the study population.

Tissue samples

Surgical or biopsy specimens obtained from patients were fixed in 4% paraformaldehyde at 4°C for 12 hours, then dehydrated, cleared, and embedded in paraffin blocks according to standard histopathological protocols. The resulting formalin-fixed, paraffin-embedded (FFPE) tissue blocks were stored at −20°C under low-humidity conditions to minimize nucleic acid degradation. Paraffin sections (5 µm) prepared for immunohistochemistry (IHC) were sectioned and processed within 1 week to reduce oxidation and degradation. All samples were labeled and cataloged to ensure traceability and consistency.

Immunohistochemistry

Paraffin-embedded tissue sections were first deparaffinized with xylene and then rehydrated with a range of graded ethanols. Sections were treated with 3% hydrogen peroxide for 10–15 min and 5% serum for 30 min at room temperature, after which they were incubated with the primary antibodies, YAP (Cell Signaling Technology Cat# 14074, RRID:AB_2650491, 1:100), CCL2 (ServiceBio Cat# GB11199, RRID:AB_2922435, 1:500), CD8 (ServiceBio Cat# GB12068, RRID:AB_2905512, 1:500) overnight at 4°C and with the secondary antibody for 1 hour at room temperature. Colors were developed with 3,3′-diaminobenzidine (DAB) and sections were counterstained with hematoxylin. Finally, the sections were dehydrated, washed with xylene and mounted with neutral resin. Staining results were examined and recorded under a light microscope.

YAP IHC scoring

YAP protein expression was evaluated using the immunoreactive score (IRS) method. Staining intensity was graded as 0 (negative), 1+ (weak), 2+ (moderate), and 3+ (strong); the tumor cell staining proportion was graded as 0 (0–5%), 1+ (6–25%), 2+ (26–50%), 3+ (51–75%) and 4+ (>75%). The IRS was calculated by multiplying the intensity by the proportion of tumor cell staining. The cutpoint was determined using receiver operating characteristic (ROC) curve analysis with immune efficacy as the outcome, and patients were classified into low-expression (≤6) and high-expression (>6) groups accordingly. All sections were independently assessed by two pathologists in a double-blind manner, with discrepancies resolved by a third pathologist to ensure scoring objectivity and reproducibility.

PD-L1 assessments

PD-L1 IHC was performed on either freshly obtained specimens processed according to standardized fixation protocols or well-preserved archival FFPE tissue blocks with a storage duration of less than 2 years. All staining was conducted in a central laboratory using the PD-L1 IHC 22C3 pharmDx assay on the Autostainer Link 48 platform (Agilent; Cat# SK006, RRID:AB_2889976). PD-L1 expression was quantified using the TPS, defined as the percentage of viable tumor cells exhibiting partial or complete membranous staining at any intensity. Staining results were independently evaluated in a blinded manner by two experienced pathologists, with discrepancies resolved by a third pathologist to ensure scoring consistency, objectivity, and reproducibility.

Evaluation of immunotherapy response

The patients’ immunotherapy response was evaluated every two cycles via CT scans. Efficacy was assessed based on RECIST V.1.1, including complete remission (CR), partial remission (PR) and disease stability (SD) or disease progression (PD). Patients with CR, PR, or SD lasting over 6 months were classified into the response group, while those with PD or SD lasting less than 6 months were classified into the non-response group. To minimize potential bias, the outcome assessment was performed blindly, with the assessors unaware of the patients’ biomarker status during the efficacy analysis.

Cell cultures

H1299, A549, H1975, H1703 and H2170 cells were cultured in Roswell Park Memorial Institute 1640 Medium (Gibco); HEK-293T was cultured in Dulbecco’s Modified Eagle’s Medium (Gibco). All cells were cultured with 10% fetal bovine serum (Gibco), 1% 100 U/mL penicillin and 100 mg/mL streptomycin (Biosharp, Cat. # BL505A). All cell lines were obtained from Procell, verified by short tandem repeat (STR) analysis and tested negative for Mycoplasma contamination.

Cell viability assay

Cells were seeded into 96-well plates at a density of 2,000–3,000 cells per well. After 72 hours of treatment with dimethyl sulfoxide (DMSO) or various concentrations of TD, cell viability was assessed. A total of 10 µL of Cell Counting Kit-8 (CCK-8) solution (Abbkine, Cat. # BMU106-CN, 2 mg/mL) was added to each well and incubated at 37°C for 1 hour. Absorbance (OD) was measured at 450 nm using a microplate reader.

Colony formation assay

The cells were seeded in 6-well plates at a density of 1,500 cells per well and cultured for 8–10 days. The clones were fixed with 4% paraformaldehyde and stained with crystal violet for 20 min, and then washed with water and photographed by the scanner.

Real-time quantitative PCR

Total RNA was extracted using the RNA extraction kit (Vazyme, Cat. # RC112-01) and reverse-transcribed into complementary DNA (cDNA) from 1 µg of RNA using the reverse transcription reagent (ABclonal, Cat. # RK20429). Real-time quantitative PCR (RT-qPCR) was conducted on a Bio-Rad Connect real-time PCR system with 2×SYBR Green qPCR Mix (ABclonal, Cat. # RK21203). GAPDH served as the internal control, and primer sequences are provided in online supplemental table for primers.

miRNA extraction and RT-qPCR

MicroRNA (miRNA) extraction was performed using the miRNA Extraction Kit (Yeasen, Cat. # 19 331ES50) according to the manufacturer’s instructions. The extracted RNA was reverse-transcribed into cDNA using the poly(A) tailing method (Yeasen, Cat. #11148ES50). RT-qPCR was performed on the Bio-Rad CFX Connect Real-Time PCR System using the 2×miRNA Universal SYBR qPCR Master Mix (plus-tail method) (Yeasen, Cat. #11171ES03). U6 was used as the internal control. The reverse transcription kit provided reverse primers and U6, while the sequences of the miRNA-specific forward primers and the miRNA inhibitor sequences are listed in online supplemental table for primers.

Western blotting

Cells were treated with TD for 36–48 hours. After removing the medium, cells were washed with cold phosphate-buffered saline (PBS) and lysed on ice using 1% sodium dodecyl sulfate (SDS) lysis buffer, then heated at 98°C for 20 min. The lysate was centrifuged at 14,000×g for 10 min, and the supernatant was collected for protein quantification using the BCA assay kit (Thermo Fisher Scientific, Cat. # A55865). Proteins were separated by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to polyvinylidene difluoride (PVDF) membranes (Millipore, Cat. # IPVH00010). Membranes were blocked with 5% milk at room temperature for 1 hour, incubated with primary antibodies, YAP (Cell Signaling Technology Cat# 14074, RRID:AB_2650491, 1:2,000), Phospho-YAP (Ser127) (Cell Signaling Technology Cat# 4911, RRID:AB_2218913, 1:2,000), MST1 (Cell Signaling Technology Cat# 3682, RRID:AB_2144632, 1:2,000), LATS1 (Cell Signaling Technology Cat# 3477, RRID:AB_2133513, 1:2,000), Phospho-LATS1 (Ser909) (Cell Signaling Technology Cat# 9157, RRID:AB_2133515, 1:2,000), GAPDH (ABclonal Cat# AC033, RRID:AB_2769570, 1:10,000) overnight at 4°C, and then with secondary antibodies, Goat Anti-Rabbit IgG(H+L) HRP (MultiSciences, Cat. # GAR0072, 1:10,000), Goat Anti-Mouse IgG(H+L) HRP (MultiSciences Cat# GAM0072, RRID:AB_2827834, 1:10,000) at room temperature for 1 hour. Signals were detected using ECL substrate (Millipore, Cat. # WBKLS0500).

Immunofluorescence

Cells were fixed with 4% paraformaldehyde for 10 min and permeabilized with 0.5% Triton X-100 for 15 min. Signal amplification was performed using Image-iT signal enhancer (Invitrogen, Cat. # I36933) for 30 min at room temperature. Immunostaining was conducted with primary antibodies, YAP (Cell Signaling Technology Cat# 14074, RRID:AB_2650491, 1:200), TTF-1 (ZSGB-Bio, Cat. # ZM-0270, 1:200), p40 (ZSGB-Bio, Cat. # ZM-0472, 1:200) overnight at 4°C, Alexa Fluor-conjugated secondary antibodies, Donkey Anti-Mouse IgG(H+L) Alexa Fluor 488 (Thermo Fisher Scientific Cat# A-21202, RRID:AB_141607, 1:200), Donkey Anti-Rabbit IgG(H+L) Alexa Fluor 555 (Thermo Fisher Scientific Cat# A-31572, RRID:AB_162543, 1:200) and DAPI (Sigma, Cat. # 28718–90-3) for nuclear staining. The samples were mounted, sealed with nail polish, and observed using an Olympus FV3000 laser scanning confocal microscope.

RNA interference

Cells were seeded in 6-well plates at an appropriate density and transfected 12 hours later. siRNA was mixed with Lipofectamine RNAi-MAX (Invitrogen, Cat. # 13778030) according to the manufacturer’s instructions and incubated at room temperature for 15–20 min to form complexes. The complexes were then added to the cell culture medium and gently mixed. After transfection, cells were cultured for 24–48 hours before proceeding with subsequent experimental analyses.

RNA-Seq analysis

Total RNA was extracted from treated cells and reverse-transcribed into cDNA. The cDNA was amplified by PCR to construct sequencing libraries. High-throughput sequencing was performed using the Illumina platform. The raw sequencing data were assessed for quality using FastQC, followed by trimming to remove low-quality reads and adapter sequences. Clean reads were aligned to the reference genome using HISAT2, and gene expression levels were quantified. Differentially expressed genes were identified using DESeq2. Three biological replicates were prepared for each sample.

Establishment of lung cancer organoids and drug treatment

Lung cancer organoids (LCOs) were generated following standard protocols.³¹ After 2 weeks of culture in 48-well plates, organoids were digested with TrypLE Express (Gibco, Cat. # 12605010). The resulting suspension was mixed with Matrigel (Corning, Cat. # 2139001) 1:1 and 10 µL containing 2×10³ cells was seeded into each well of a 96-well plate. Once solidified, 100 µL of organoid culture medium was added per well, and LCOs were cultured for 2–3 days. DMSO or TD was then added every 3 days for two treatments. Finally, fresh medium was replaced, and cell viability was assessed by adding 100 µL of CellTiter-Glo 3D reagent (Promega, Cat# G968A) to the medium. Mix the contents vigorously for 5 min to induce cell lysis and allow the plate to incubate at room temperature for an additional 25 min to stabilize the luminescent signal, then record luminescence. Half maximal inhibitory concentration (IC50) values were calculated using GraphPad Prism.

ChIP-qPCR

A549 cells were seeded in 10 cm dishes and treated with DMSO or TD for 48 hours. Cells were cross-linked with 1% formaldehyde at room temperature for 10–20 min, followed by the addition of 125 mM glycine to stop the crosslinking reaction, incubating for 5–10 min. Cells were then lysed in extraction buffer (10 mM Tris HCl (pH 7.5), 10 mM NaCl, and 0.5% NP40) with proteinase inhibitor cocktail (Roche). The chromatins were sheared with MNase (NEB, Cat. # M0247S) in digestion buffer (20 mM Tris-HCl (pH 7.5), 15 mM NaCl, 60 mM KCl, and 1 mM CaCl2) at 37°C for 20 min and treated with 2× stop/chromatin immunoprecipitation (ChIP) buffer (100 mM Tris HCl (pH 7.5), 280 mM NaCl, 2% Triton X-100, 0.2% sodium deoxycholate, and 20 mM EDTA) to obtain fragments with an average size of 200–1,000 bp. The samples were centrifuged at 14,000×g for 10 min at 4°C, and the supernatant was collected. Chromatin was incubated overnight at 4°C with 5 µg of Anti-Histone H3 antibody (Acetyl-K27) (Abcam Cat# ab4729, RRID:AB_2118291) or SNAI2 antibody (Santa Cruz Biotechnology, Cat# sc-166902X, RRID:AB_2857906), along with species-matched IgG controls (Rabbit IgG, ABclonal, Cat# AC005, RRID:AB_2771930; Mouse IgG, ABclonal, Cat# AC011, RRID:AB_2770414), followed by the addition of protein A/G magnetic beads (Thermo Scientific, Cat. # 88802) and incubation for 2 hours to capture the antibody-chromatin complex. The beads were washed sequentially with low salt buffer (1×stop/ChIP buffer), high salt buffer (1×stop/ChIP buffer with 0.5 M NaCl), Tris/LiCl (10 mM Tris (pH 8.0), 0.25 M LiCl, 0.5% NP40, 0.5% Na-deoxycholate, and 1 mM EDTA), and Tris/EDTA buffers (50 mM Tris (pH 8.0) and 10 mM EDTA) to remove non-specific bindings. Thereafter, elution buffer (10 mM Tris (pH 8.0), 10 mM EDTA, 1% SDS, 150 mM NaCl, and 5 mM DTT) was added to the beads and incubated overnight at 65°C to reverse cross-linking. Add RNase A (Invitrogen, Cat. # AM2270) to the sample and incubate at 37°C for 1 hour to degrade RNA. Subsequently, add proteinase K (Invitrogen, Cat. # AM2546) and incubate at 55°C for 1 hour to remove proteins, then DNA was purified using a DNA purification kit (Omega, Cat. # D6538-01) and used as a template for qPCR analysis. Calculate the enrichment relative to the input chromatin using the percentage of input method. Compare the binding of promoter regions to a negative control.

Statistical analysis

All statistical analyses were performed using GraphPad Prism, IBM SPSS Statistics, and R software. Comparisons between groups were conducted using the χ² test, two-tailed unpaired Student’s t-test, or one-way analysis of variance as appropriate. Optimal cut-off values were determined based on ROC curve analysis and median values. Survival analyses were performed using the Kaplan-Meier method, and HRs with 95% CIs were estimated using Cox proportional hazards regression models. Multivariable Cox proportional hazards regression models were applied to identify factors independently associated with PFS and OS. A multivariable logistic regression model was constructed to evaluate predictors of response to immunotherapy. Candidate independent variables included YAP expression level (high vs low), PD-L1 expression level, and baseline demographic and clinicopathological characteristics, including age, sex, histological subtype, tumor stage, line of therapy, comorbidities, and smoking and alcohol consumption status. Given the relatively small sample size, Firth’s penalized likelihood method was applied in multivariable analyses to reduce small-sample bias and mitigate the risk of separation. P value<0.05 was considered statistically significant, with *p<0.05, **p<0.01, ***p<0.001, and ****p<0.0001; ns indicates no statistical significance.

Results

Comparative analysis of YAP and PD-L1 expression in predicting response to immunotherapy

Transcriptomic analyses of GSE3141, GSE26939, and The Cancer Genome Atlas-lung adenocarcinoma (TCGA-LUAD) showed that elevated YAP messenger RNA (mRNA) expression in NSCLC tissues was associated with poorer OS in patients (figure 1A-D). To further investigate the role of YAP in immunotherapy, we established a clinical cohort of 141 patients with NSCLC, comprising 136 baseline samples and 5 samples collected after the development of treatment resistance. Baseline demographic and clinical characteristics of the study subjects in different subgroups, including age, gender, tumor type, tumor stage, treatment regimen and history of comorbidities, have been summarized (figure 1E and online supplemental table 1). We systematically assessed the expression of both YAP and PD-L1, as well as their association with clinical outcomes of immunotherapy. IHC analysis was performed on surgical or biopsy specimens obtained from patients in our cohort (figure 2, A and B). The cohort was randomly divided into a training set (n=70) and a validation set (n=71) at a 1:1 ratio, baseline characteristics were well balanced between the two cohorts (online supplemental table 2). The optimal cut-off value for YAP expression in predicting immunotherapy response was determined using ROC curve analysis in the training cohort (area under the curve (AUC)=0.835, 95% CI 0.737 to 0.932). Application of this predefined threshold to the validation cohort demonstrated consistent discriminative performance (AUC=0.765, 95% CI 0.660 to 0.871) (figure 2C). Based on this cut-off, patients were stratified into YAP-low (≤6) and YAP-high (>6) groups. In the training cohort, high YAP expression was associated with a significantly lower ORR and shorter PFS and OS (online supplemental figure 1A,B). These findings were largely reproduced in the validation cohort, where a similar trend toward lower ORR and poorer survival outcomes was observed in patients with high YAP expression, although the difference in OS did not reach statistical significance (online supplemental figure 1C,D).

Figure 1. Elevated YAP in NSCLC tissues is associated with poor prognosis. (A–C) Kaplan-Meier survival curves stratified by YAP in the GSE3141, GSE26939, and TCGA-LUAD datasets. (D) Differences in YAP expression between the response (CR and PR) and non-response (PD) groups in the TCGA-LUAD cohort. (E) Firth’s penalized logistic regression analysis of patient characteristics based on the full cohort and the baseline-only cohort. CR, complete remission; IHC, immunohistochemistry; LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma; NR, non-responders; NSCLC, non-small cell lung cancer; PD, disease progression; PD-L1, programmed death-ligand 1; PR, partial remission; R, responders; TCGA, The Cancer Genome Atlas; TPM, transcripts per million; TPS, tumor proportion score; YAP, yes-associated protein.

Figure 2. The efficiency of YAP and PD-L1 in predicting immunotherapy outcome. (A–B) Representative images of YAP and PD-L1 IHC staining (10× and 40× views). (C) ROC curves of YAP for predicting immunotherapy response in the training and validation cohorts. (D) Comparison of objective response rates to immunotherapy according to YAP and PD-L1 expression in the overall cohort (χ² test). (E) ROC curves for YAP (green), PD-L1 (red), and their combination (purple) in diagnosing our cohort of 141 patients with NSCLC. (F) Survival analysis of patients with NSCLC comparing high and low YAP expression: progression-free survival (left) and overall survival (right). (G) Survival analysis based on PD-L1 expression with a cut-off of 1: progression-free survival (left) and overall survival (right). (H) Scatter plot showing the correlation between YAP and PD-L1. AUC, area under the curve; FPR, false positive rate; IHC, immunohistochemistry; LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma; NSCLC, non-small cell lung cancer; PD-L1, programmed death-ligand 1; ROC, receiver operating characteristic; TPR, true positive rate; YAP, yes-associated protein.

Applying this cut-off to the overall cohort, patients with low YAP expression showed a significant association with improved treatment response (figure 2D). ROC analysis further confirmed the predictive value of YAP for immunotherapy response (AUC=0.802) (figure 2E). Survival analysis showed that patients with low YAP expression had significantly longer PFS and OS, indicating a more favorable clinical outcome (figure 2F). Consistently, multivariable logistic regression analysis identified the YAP score as an independent predictor of response to immunotherapy in both the overall cohort (OR=13.545, 95% CI 5.708 to 35.870, p<0.001) and the subgroup restricted to baseline samples (figure 1E). Furthermore, multivariable Cox regression analysis demonstrated that high YAP expression was associated with worse PFS (HR=2.568, 95% CI 1.699 to 3.882, p<0.001) and OS (HR=1.993, 95% CI 1.118 to 3.544, p=0.020) (online supplemental figure 1E).

In contrast, PD-L1 expression evaluated using multiple TPS thresholds (1%, 5%, 10%, 25%, and 50%) failed to consistently distinguish responders from non-responders (figure 2D and online supplemental figure 2A-E). ROC analysis further indicated that YAP outperformed PD-L1 in predicting immunotherapy response (figure 2E). Moreover, combining PD-L1 with YAP did not improve predictive accuracy. Consistently, PD-L1 expression stratified by TPS was not significantly associated with either PFS or OS (figure 2G and online supplemental figure 2A-E). Although previous studies have linked YAP to PD-L1 regulation,^{21 32} our cohort, which primarily consisted of baseline samples, showed no significant association (figure 2H).

In summary, these findings suggest that YAP may serve as a promising predictive biomarker for immunotherapy response in NSCLC, particularly in the context of combination immunotherapy, and may demonstrate improved predictive capability compared with PD-L1 in this cohort.

HDAC inhibitor-Tucidinostat suppresses transcription of YAP

Given the pivotal role of YAP as a biomarker associated with resistance to immunotherapy, we aimed to identify a clinically feasible inhibitor capable of counteracting this resistance. To this end, we conducted transcriptome-matching analysis of YAP/TAZ signatures using the CLUE platform (https://clue.io/), in order to screen for candidate compounds with potential utility in combination immunotherapy. We first established YAP and TAZ knockdown H1299 cell lines (figure 3A,B) and conducted RNA-sequencing (RNA-seq) analysis to assess gene expression changes. This revealed 965 differentially expressed genes, with 268 downregulated and 697 upregulated (online supplemental figure 3A). Heatmap analysis highlighted significant downregulation of genes associated with the Hippo signaling pathway (online supplemental figure 3B). Gene Ontology (GO) enrichment analysis showed that downregulated genes were significantly enriched in biological processes related to Hippo signaling, cytokine production in immune response, and positive regulation of immune effector processes (online supplemental figure 3C), suggesting a critical role in immune regulation of YAP/TAZ. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis further confirmed Hippo signaling as a significantly enriched pathway (online supplemental figure 3D).

Figure 3. Identification and evaluation of YAP-associated inhibitors through transcriptomic analysis. (A) RT-qPCR detected the mRNA of YAP and TAZ after transfection with siNC or siYT in NCI-H1299; siNC denotes the negative control, while siYT represents the combined knockdown of YAP and TAZ. (n=3; mean±SEM; two-tailed t-test). (B) Knockdown efficiency of YAP and TAZ was validated by western blot analysis. (C) Venn diagram showing the overlap between differentially expressed genes and YAP-target genes. (D) A list of YAP-related drugs based on CLUE connectivity map analysis. (E) Lung cancer cells were treated with DMSO or TD. The mRNA levels of YAP were detected by RT-qPCR (n=3; mean±SEM; two-tailed t-test). (F) Confocal images of different lung cancer cells stained for YAP after treatment with TD. (G) Nuclear YAP fluorescence intensity with or without TD treatment. Data are shown as mean±SEM. DMSO, dimethyl sulfoxide; mRNA, messenger RNA; RT-qPCR, real-time quantitative PCR; TAZ, transcriptional coactivator with PDZ-binding motif; TD, tucidinostat; YAP, yes-associated protein.

By comparing RNA-seq data with known YAP target genes,^{33 34} a total of 59 overlapping factors were identified, comprising 2 upregulated and 57 downregulated genes (figure 3C). The top 30 upregulated genes (including the 2 overlapping genes) and the 57 downregulated genes were selected for drug prediction analysis using the CLUE platform, which identified 45 potential YAP-targeting drugs. HDAC inhibitors emerged as the largest category (11 drugs) (figure 3D). As the drug prediction results suggest that HDAC inhibitors regulate YAP, and previous studies have shown that HDAC inhibitor TD exhibits potent antitumor effects in solid tumors.28,30 Thus, we investigated the impact of TD on YAP. Quantitative analysis of mRNA levels revealed that TD significantly suppressed YAP transcription (figure 3E). Given that YAP primarily functions as a nuclear transcriptional activator, we employed immunofluorescence (IF) to evaluate the impact of TD on YAP activity. IF staining demonstrated that TD effectively suppressed the nuclear localization and transcriptional activity of YAP (figure 3F,G). Collectively, these findings highlight the robust inhibitory effect of TD on YAP transcription.

Tucidinostat suppresses tumor proliferation through targeting YAP

We further investigated the effect of TD on YAP protein. A marked reduction in YAP protein levels was observed following TD treatment, with less suppression in H1299 (figure 4A,B). Functional assays further revealed that TD inhibited cell proliferation and colony formation in both NSCLC cell lines and patient-derived lung adenocarcinoma primary cells (LUAD-P). Interestingly, TD had the weakest inhibitory effect on H1299 at lower concentrations, which was similar to the results obtained from the protein analysis (figure 4C,D). These findings suggest that the antitumor activity of TD is likely dependent on the degree of YAP suppression. Moreover, TD also suppressed the expression of YAP and its downstream target genes in LUAD-P (online supplemental figure 4A). Given that organoids recapitulate the tissue architecture and functional characteristics of primary tumors in a three-dimensional context, they provide a powerful platform for drug sensitivity testing, disease mechanism exploration, and personalized treatment development.³⁵ To extend our findings, we established organoids from surgical specimens of patients with NSCLC (LCO). Immunostainings from representative adenocarcinoma-like organoid revealed strong nuclear localization of YAP and thyroid transcription factor 1 (TTF-1), indicating active YAP signaling within these organoids (figure 4E and online supplemental figure 4B). We treated the organoids with varying concentrations of the drug, and the results demonstrated that TD effectively inhibited their growth (figure 4F,G).

Figure 4. Tucidinostat exerts an inhibitory effect on tumor proliferation by targeting YAP. (A–B) Western blot analysis of YAP expression in NSCLC cell lines and primary cells derived from patients with LUAD (LUAD-P) following TD treatment. TD was administered at a final concentration of 10 µM. Quantification is based on three independent experiments. (C) Relative OD450 value of NSCLC cell lines and LUAD-P cells treated with TD for 72 hours. (D) Photographs of colony formations in NSCLC cell lines treated with TD or DMSO; TD was treated at concentrations of 3 µM for H1299 and A549 cells, and 1 µM for H1975, H1703 and H2170 cells, with a treatment duration of 7–10 days. (E) Representative images of immunofluorescence staining for LCOs, including YAP and TTF-1. (F) Brightfield images of LCOs plated in 96-well plates; LCOs were treated with TD at concentrations of 5 µM, 10 µM, and 20 µM for 10–15 days. (G) Viability of LCOs was measured by CellTiter-Glo after TD treatment. (H) Heatmap showing mRNA levels of YAP target genes in A549 cells DMSO versus TD (n=3). (I) The impact of 72 hours TD treatment on cell proliferation in YAP-overexpressing and control cells. (J) Changes in YAP target genes after TD treatment in YAP-overexpressing and control cells (n=3; mean±SEM; one-way ANOVA). ANOVA, analysis of variance; DMSO, dimethyl sulfoxide; IC₅₀, half maximal inhibitory concentration; LCO, lung cancer organoid; LUAD-P, patient-derived lung adenocarcinoma primary cell; mRNA, messenger RNA; NSCLC, non-small cell lung cancer; TD, tucidinostat; YAP, yes-associated protein.

To probe the mechanism of inhibition of YAP by TD, we conducted RNA-seq analysis on A549 cells treated with TD, which revealed alterations in gene expression (online supplemental figure 4C). KEGG analysis enriched these genes in the Hippo pathway (online supplemental figure 4D). Clearly, TD significantly reduced YAP and its target genes (figure 4H). Since the direct target of TD is HDACs, we sought to investigate the contribution of YAP inhibition by TD on its antitumor function. We established an exogenous YAP-overexpressing A549 cell line to analyze the effects of TD in these cells. The results showed that YAP overexpression reversed the inhibitory effects of TD on tumor cell proliferation (figure 4I). Moreover, YAP overexpression partially restored some YAP target genes that were initially downregulated by TD treatment (figure 4J).

Taken together, these findings highlight the potent antitumor activity of TD in lung cancer and further support its promising clinical relevance. TD exerts its tumor-suppressive effects primarily by inhibiting the expression of YAP and its downstream target genes.

Tucidinostat suppresses YAP transcription by upregulating SNAI2

In order to elucidate the mechanism of TD-mediated YAP inhibition, the involvement of the Hippo pathway was initially assessed. Although the results showed that MST1 and p-LATS1 were upregulated, the level of p-YAP was reduced after treatment of TD. These results suggested the YAP protein level reduction plays dominant roles as it directly led to p-YAP level reductions other than upregulations by p-LATS1 (online supplemental figure 5A). Additionally, previous studies have shown that HDAC inhibitors can suppress YAP expression by upregulating specific miRNAs.^{36 37} Based on this, we tested a series of miRNAs known to inhibit YAP.^{38 39} The results revealed that treatment with TD significantly upregulated the expression of miR-132-3p, miR-138-5p (online supplemental figure 5B). However, the combined use of these miRNA inhibitors with TD did not reverse the inhibitory effect of TD on YAP (online supplemental figure 5C,D). miRNA-mediated gene silencing relies on the assembly and activation of the RNA-induced silencing complex (RISC), in which AGO2, as the core component of RISC, plays a critical role in miRNA function.⁴⁰ Therefore, we generated two AGO2-knockdown cell lines and examined the changes in YAP expression after TD treatment. The results demonstrated that the absence of AGO2 also failed to reverse the inhibitory effect of TD on YAP, suggesting that the inhibition of YAP by TD is independent of the miRNA-AGO2 pathway (online supplemental figure 5E,F).

We further analyzed RNA-seq data and found SNAI2, a transcriptional repressor, as being upregulated by TD. Previous studies indicate that SNAI2 binds to the YAP promoter and inhibits its transcription.⁴¹ RT-qPCR validation demonstrated over 15 times upregulations in SNAI2 mRNA levels and a corresponding decrease in YAP expression following TD treatment (figure 5A). The Cq value is high in A549 without TD treatment, indicating weak expression of SNAI2 in normal condition (online supplemental figure 5G). Therefore, knockdown of SNAI2 only slightly upregulated YAP levels (figure 5B). However, co-treatment with siRNA of SNAI2 and TD reversed the YAP suppression by TD (figure 5C,D), indicating that TD treatment inhibits YAP transcription through upregulating SNAI2. As TD functions as an HDAC inhibitor, we further investigated its effect on the acetylation of histone at the promoter region of SNAI2. ChIP-qPCR analysis revealed substantial enrichment of H3K27ac at the SNAI2 promoter following TD treatment (figure 5E), suggesting an increase in chromatin accessibility and transcriptional activation at this locus. To investigate whether SNAI2 directly regulates YAP, we performed additional ChIP-qPCR assays to assess SNAI2 binding at the YAP promoter. The results demonstrated that TD treatment significantly increased SNAI2 enrichment at the YAP promoter region (figure 5F). Collectively, these findings support a model in which TD suppresses YAP expression indirectly by promoting SNAI2 transcription through enhanced histone acetylation at the SNAI2 locus, thereby facilitating SNAI2-mediated transcriptional repression of YAP (figure 5G).

Figure 5. Tudicinostat downregulates YAP through histone-mediated SNAI2 activation. (A) Alteration in YAP and SNAI2 mRNA levels following TD treatment (n=3; mean±SEM; two-tailed t-test). (B) RT-qPCR analysis of the SNAI2 knockdown efficiency and YAP expression in A549 cells (n=3; mean±SEM; one-way ANOVA). (C–D) A549 cells were transfected with siNC or siSNAI2 for 12 hours, subsequently treated with or without TD for 36 hours. RT-qPCR (n=3; mean±SEM; one-way ANOVA) and WB analysis of the YAP expression. (E) ChIP-qPCR analysis of H3K27ac binding level in the SNAI2 promoter region after TD treatment (n=3; mean±SEM; one-way ANOVA). (F) Relative SNAI2 enrichment at the YAP promoter after TD treatment measured by ChIP-qPCR (n=3; mean±SEM; one-way ANOVA). (G) The model of HDAC inhibitor TD indirectly regulates YAP expression by increasing histone accessibility at the SNAI2 promoter. ANOVA, analysis of variance; ChIP, chromatin immunoprecipitation; DMSO, dimethyl sulfoxide; HDAC, histone deacetylase; HDACi, histone deacetylase inhibitors; mRNA, messenger RNA; NC, negative control; RT-qPCR, real-time quantitative PCR; TD, tucidinostat; WB, western blot; YAP, yes-associated protein.

Tucidinostat reverses immune resistance of NSCLC in clinics by inhibiting YAP

For transcriptome sequencing results after TD treatment, we also conducted a GO-biological process (BP) analysis on all the differentially expressed genes, which revealed a significant enrichment of genes involved in biological processes closely related to immune functions (online supplemental figure 6A). A heatmap demonstrated the modulation of YAP-related immune genes by TD treatment (figure 6A). Notably, several immune-related genes, including CCL2, PD-L2, FAS, MALT1, and PLAUR, which were initially downregulated by TD, were restored on YAP overexpression (figure 6B).

Figure 6. Tucidinostat overcomes immune resistance by targeting YAP in clinical trial. (A) Heatmap of changes in mRNA levels of immunity genes regulated by YAP after TD treatment in A549 cells (n=3). (B) Variation in immune genes regulated by YAP after TD treatment in YAP overexpressing and control cells (n=3; mean±SEM; one-way ANOVA). (C) Analysis of immune cell infiltration differences between YAP high-expression and low-expression groups in TCGA NSCLC cohort. Immune infiltration levels were estimated using the CIBERSORT algorithm and differences between YAP groups were compared using the Wilcoxon rank-sum test (Mann-Whitney U test). (D) Immunohistochemical staining (10×) of YAP, CCL2, and CD8 in paired tumor samples from the same patient at initial diagnosis and after the development of resistance. (E) Quantification of immunohistochemical staining from (D), (n=3 or 5 per group; mean±SEM; two-tailed t-test). (F) Post-treatment chest CT scan of patient P43 after TD combined with immunotherapy and anti-angiogenic therapy. Axial images at a comparable level of the lesion (top) and coronal multiplanar reconstruction images acquired at the level of the maximum lesion diameter (bottom). The previously enlarged left axillary lymph node showed marked shrinkage, indicating a significant reduction in tumor burden. (G) Trend changes in lung cancer biomarkers following combined treatment with TD in patient P43. (H) Representative chest CT images of patient P142 after triple therapy, indicating stabilization of the left lung nodular lesion. ANOVA, analysis of variance; CCL2, C-C motif chemokine ligand 2; CD8, cluster of differentiation 8; DMSO,dimethyl sulfoxide; mRNA, messenger RNA; NSCLC, non-small cell lung cancer; TCGA,The Cancer Genome Atlas; TD, tucidinostat; YAP, yes-associated protein.

In addition, we analyzed the immune infiltration profiles of patients with NSCLC in the TCGA cohort based on YAP expression levels. The results showed that patients with high YAP expression exhibited significantly reduced infiltration of antitumor immune cell populations, including CD8⁺ T cells, CD4⁺ memory activated T cells, follicular helper T cells, and M1 macrophages, whereas immunosuppressive M2 macrophages were markedly enriched (figure 6C). To further validate these observations, we collected paired tumor tissue samples from three patients with NSCLC at the time of initial diagnosis and during the stage of acquired immune resistance. All three patients experienced a transition from initial response to subsequent resistance following immunotherapy. Histological analyses revealed that YAP was markedly increased in the resistant tumors, accompanied by an upregulation of the downstream chemokine CCL2. Notably, CD8⁺ T-cell infiltration within the tumor tissue was substantially reduced, whereas abundant CD8⁺ T-cell accumulation was observed in baseline samples (figure 6D,E). These findings suggest that YAP plays a critical role in the development and maintenance of immune tolerance by modulating the tumor immune microenvironment, particularly by influencing CCL2 expression and CD8⁺ T-cell infiltration.

Based on prior in vitro experiments showing that TD effectively inhibited tumor growth in primary lung cancer cells and LCOs, and its regulatory effects on YAP-associated proliferation and immune-related genes, we conducted exploratory clinical validation in two patients with NSCLC who had progressed after multiple lines of therapy and developed resistance to ICIs (online supplemental figure 6B). Both patients received a combination of TD, a PD-1 inhibitor, and an anti-angiogenic agent. Patient P43 exhibited continued progression of left axillary lymph node metastases during combined immunotherapy, anti-angiogenic therapy, and concurrent brain radiotherapy (online supplemental figure 6C). Remarkably, the lesions regressed substantially after the introduction of triple-combination therapy, along with a concomitant decline in tumor marker levels (figure 6F,G). During treatment, mild thrombocytopenia and leukopenia led to intermittent interruption of TD (August 21, 2024 to December 4, 2024). Transient radiological changes during this period were attributed to treatment interruption rather than true progression. Disease control was re-established on treatment resumption, no additional significant adverse events were observed, and the overall response was assessed as PR, with a duration of disease control of 11 months. In patient P142, treatment with the triple-combination regimen resulted in sustained disease stabilization of the left lung lesion for approximately 6 months (figure 6H). Collectively, these results indicate that the combination of TD with ICIs and anti-angiogenic therapy effectively overcomes YAP overexpression-associated immune resistance in the clinical setting, thereby providing a potentially valuable therapeutic strategy for this patient population.

Discussion

YAP, a critical downstream effector in cancer-related signaling pathways, has been widely implicated in tumor immune evasion and acquired resistance to immunotherapy.^{17 21 42 43} Our previous studies have further confirmed its essential role in mediating immune resistance.²⁴ In this study, we systematically evaluated the relationship between YAP expression and clinical outcomes of immunotherapy. Notably, we found that YAP serves as a robust predictor of immunotherapeutic efficacy, particularly in patients with NSCLC receiving combined immunotherapy. In this context, YAP outperformed the commonly used biomarker PD-L1. Importantly, YAP expression assessment does not rely on any specific genetic alteration or genomic profile and can be routinely performed through IHC, providing a practical and economical biomarker for clinical application.

Nevertheless, several limitations of this study should be acknowledged. First, this was a retrospective analysis, and the cohort was derived from only two medical centers, which may introduce selection bias and potentially limit the generalizability of the findings. Second, treatment distribution within the cohort was unbalanced, with a predominance of combination therapy and a relatively small proportion of patients receiving monotherapy. Future multicenter, prospective studies conducted in more balanced and heterogeneous patient populations are warranted to further validate the predictive value of YAP and to better define its potential role in clinical practice.

Through transcriptomic profiling, we identified TD as a potential YAP suppressor and applied it to two advanced NSCLC clinical cases characterized by YAP high and acquired immunotherapy resistance. Following treatment with the TD-based triplet regimen, the patients achieved either marked tumor regression or stable disease. These clinical observations provide preliminary evidence that incorporating TD into ICI-based regimens may represent a potential strategy to overcome YAP-mediated resistance to immunotherapy. However, the current evidence is limited to two prospectively observed cases and does not permit definitive conclusions regarding the independent contribution of TD within the triple-combination regimen, and the broader clinical application of TD remains limited by its currently approved indications. Therefore, these findings should be considered exploratory and require validation in larger, multicenter, prospective studies before any clinical application can be considered.

TD inhibits HDAC 1, 2, 3, and 10,⁴⁴ significantly increasing histone acetylation and causing epigenetic changes in multiple signaling pathways. This results in the suppression of tumor cell cycles and the induction of tumor cell apoptosis. Additionally, TD exerts important immune-regulatory effects by activating natural killer cells and cytotoxic T cell-mediated immune responses.^{45 46} Clinical studies have shown that TD, in combination with other therapies, can convert MSS/pMMR colorectal cancer from a “cold tumor” to a “hot tumor,” significantly enhancing antitumor immune responses.³⁰ In this study, TD indirectly regulated YAP activity by increasing the expression of the transcriptional repressor SNAI2 through reducing histone deacetylation at the promoter of the SNAI2 gene. Although YAP is not a direct target of TD, the forced expression of YAP can counteract the antitumor effects of TD, indicating that YAP is a crucial node of TD-mediated antitumor effects. As specific YAP-targeting drugs are currently unavailable in clinical practice, TD combined with ICIs may offer a novel therapeutic option for patients with immune resistance due to high YAP expression.

Nevertheless, 12.2% of patients with low YAP expression failed to derive clinical benefit from immunotherapy, indicating that while YAP serves as a valuable predictive biomarker, it is not the sole determinant of treatment efficacy. Other mechanisms may also contribute, such as the activation of alternative immunosuppressive signaling pathways,⁴⁷ a limited repertoire of tumor neoantigens,⁴⁸ or defects in antigen presentation⁴⁹ that impair effective immune recognition. Future research should focus on elucidating the complex molecular determinants of immunotherapy response through a multidimensional approach, integrating transcriptomic, genomic, and immunogenomic data to develop more accurate and individualized predictive models.

Data availability statement

Data are available in a public, open access repository.