HSP90 inhibitor AUY922 suppresses tumor growth and modulates immune response through YAP1-TEAD pathway inhibition in gastric cancer

Abstract

Heat shock protein 90 (HSP90), a vital chaperone involved in the folding and stabilization of various cellular proteins, regulates key functions in many tumor cells. In the context of gastric adenocarcinoma (GAC), where HSP90’s role remains largely unexplored, we aimed to investigate the significance of HSP90 inhibitor, AUY922, in regulating the YAP1/TEAD pathway and its association with the tumor immune microenvironment (TME). Our results showed that AUY922 effectively inhibited GAC aggressiveness in both the invitro and invivo models, induced apoptosis, and cell-cycle arrest. Various functional assays elucidated that AUY922 potently inhibited the expression and interaction among YAP1/TEAD and HSP90, resulting in down-regulation of target functional genes. AUY922 additionally altered the tumor microenvironment (TME) into an inflamed state with increased cytokine production in T cells, including interferon gamma, granzyme B, and perforin, and inhibited M2 polarization of tumor-associated macrophages, rendering it a favorable partner for immune checkpoint inhibition. Our findings highlighted the suggestion of targeting HSP90 in GAC therapy via down-regulating YAP1/TEAD signaling. Additionally, our results suggest that AUY922’s ability to reshape the GAC TME favoring the host sets the stage for a clinical trial that combines HSP90 and checkpoint inhibition, where HSP90 could serve as a biomarker for patient selection.

1. Introduction

Heat shock protein 90 (HSP90) is a crucial chaperone that plays a central role in folding and stabilizing various cellular proteins [1]. Tumor cells often exhibit high levels of HSP90 and rely heavily on its functions, making it an attractive target for cancer therapy [1,2]. HSP90 plays a vital role in supporting various oncogenic proteins involved in cell survival, proliferation, and tumorigenesis, such as receptor tyrosine kinases, signaling kinases, transcription factors, telomerase, and apoptosis-related proteins [2–4]. In gastric cancer specifically, HSP90 stabilizes key oncogenic proteins like HER2, MET, and AKT, which are essential for cancer cell survival and proliferation [5–8]. Consequently, the inhibition of HSP90 has gained considerable attention as a potential therapeutic strategy. AUY922 is a highly effective and relatively less toxic small molecule. It inhibits the ATPase activity of HSP90 and leads to HSP90 failing to bind to client proteins [9,10]. AUY922 has been evaluated in several clinical trials, including Phase I and Phase II, demonstrating promising antitumor activity and manageable toxicity profiles in various cancers [11,12]. However, evidence of AUY922’s efficacy in influencing gastric adenocarcinoma (GAC) tumorigenesis and the immune microenvironment remains limited.

YAP signaling is vital in GAC due to its multifaceted role in promoting tumorigenesis, metastases, cancer stem cell maintenance, and radio-resistance [13–17]. YAP1 functions co-operating with the TEAD family and Hippo pathway as well as co-factor TAZ [18,19]. Dysregulation of the Hippo pathway with LATS dephosphorylation, typically characterized by YAP1 activation, contributes to tumor development and progression [18,19]. YAP1, when triggered by oxidative, mechanical stress, or other signaling and not inhibited by the Hippo pathway, translocates into the nucleus and interacts with TEAD transcription factors, thus driving the expression of target genes involved in tumor essential cell functions [18–21]. The aberrant activation of YAP1 and TEAD in GAC leads to aggressive tumor growth and metastases, emphasizing their importance as therapeutic targets in combating GAC. However, how HSP90 inhibition is associated with these YAP1 regulations remains largely unexplored, especially in GAC.

The tumor immune microenvironment (TME) plays a critical role in cancer progression and response to therapy in GAC [22,23]. The 1st line therapy for advanced-GAC can include the immune checkpoint inhibitor to manage GAC patients [22,24,25]. Immunotherapy is a pivotal treatment strategy in some GAC patients. Recently, basic studies demonstrated HSP90 has one of the core factors in shaping TME [26–28]. The combination effect of HSP90 inhibitor with immune checkpoint inhibitor was also reported in breast and colon cancer cells [29,30], but the evidence of interaction of HSP90 with the TME is still scarce. Further understanding the relationship between HSP90 and the TME is essential for developing more effective cancer immunotherapy strategies, particularly in GAC.

The aim of our research project is to decipher the significance of HSP90 inhibition in GAC, with a specific focus on the effect of HSP90 inhibition on the YAP1/TEAD pathway and their relationship with the TME. Furthermore, we explored the role of HSP90 inhibition in modulating the immune response within the TME as well as the potential synergistic effect with immune-checkpoint inhibition. There is still an urgent need to discover novel targets and corresponding therapeutics in GAC. Our findings could contribute to a better understanding of the therapeutic potential of HSP90 inhibition in GAC and may pave the way for the development of more effective treatment strategies targeting HSP90 and its associated pathways in cancer.

2. Materials and methods

2.1. Cells and reagents

We utilized several human cell-lines such as gastric adenocarcinoma cell-lines (AGS), and a metastatic gastric cancer cellline (GA0518). The AGS cell line was purchased from the American Type Culture Collection, and GA0518 was generated from a GAC patient as described previously [31,32]. We also used the murine gastric cancer cell line, KP-Luc2 cells, which were obtained from Dr. Jo Ishizawa in the Department of Leukemia of MDACC [32,33]. The resulting cells were stored at −80 °C and authenticated every 6 months before being used in subsequent experiments. The cells were cultured in DMEM or RPMI supplemented with 10 % fetal bovine serum and antibiotics and incubated at 37 °C in 5 % CO2. The cells were authenticated in the Cell Line Core Facility at The University of Texas MD Anderson Cancer Center every 6 months to ensure their identity and purity. For labeling GA0518 cells with mCherry-Luciferase, a plasmid FUW-CBRLuc-mCherry was made into lentiviruses, and the transduced cells with mCherry fluorescence were then subjected to flow cytometer sorting. KP-Luc2 cells were labeled with GFP [33]. AUY922 (Luminespib) was sourced from MedChemExpress (Cat# HY-1021). The compound was initially dissolved in dimethyl sulfoxide (DMSO) at a concentration of 10 mM, with aliquots stored at −20 °C. Fresh working solutions were prepared as required for each experiment.

2.2. Mice

All animal procedures were conducted under a peer-reviewed Institutional Animal Care and Use Committee (IACUC)-approved protocol (#00000232-RN03, and #00001488-RN02). The IACUC at MD Anderson Cancer Center (MDACC) approved all animal experiments in accordance with NIH guidelines. NOD.Cg-Prkdcscid/J (SCID) (The Jackson Laboratory, #001303), C57BL/6 (The Jackson Laboratory, #000664) mice were bred and maintained in the North MDACC Mouse Facility in accordance with institutional requirements. SCID and C57BL/6 mice aged nine to eleven weeks in both genders were used in this study. The investigators did not perform any experiments blindly.

2.3. Cell survival MTS assay

Cell proliferation was assessed using the CellTiter 96^® AQueous One Solution kit (Promega, G5421), as per the manufacturer’s guidelines. In brief, 1000 cells in 100 μl of culture medium were seeded into each well of a 96-well plate. The following day, drugs were added at varying doses, and cells were incubated at 37 °C with 5 % CO₂ for 6 days. After incubation, the medium was removed, and 120 μl of fresh medium containing 20 μl of the CellTiter 96^® AQueous One Solution Reagent was added to each well. Following a 2-h incubation at 37 °C, absorbance was measured at 490 nm.

2.4. siRNA transfection

AGS and GA0518 cells were transfected with siControl or siHSP90 siRNA sequences, as listed in the Supplementary Key Resource Table. Cells were seeded in 6-well plates at an appropriate density to reach 50–70 % confluence on the day of transfection. Transfection was performed using Lipofectamine RNAiMAX (Invitrogen) following the manufacturer’s instructions. Briefly, 25 nM of siRNA (either siControl or siHSP90) was mixed with Lipofectamine RNAiMAX in Opti-MEM Reduced Serum Medium (Gibco) and incubated for 5 min at room temperature to allow complex formation. The siRNA-lipid complexes were then added dropwise to each well containing cells in serum-free medium. After 6 h of incubation, the transfection medium was replaced with complete growth medium, and cells were incubated for 48–72 h before subsequent analyses.

2.5. Establishment of knockout cells

Additionally, the Lentiviral CRISPR/Cas9 system was used to knockout the YAP1 gene in GA0518 cells. The guide RNAs for YAP1 were designed following the principle in MIT website http://crispr.mit.edu/or German Cancer Research Center’s E-Crisp website http://www.e-crisp.org/E-CRISP/designcrispr.html. The gRNA duplexes for YAP1 knockout (see Supplementary Key Resource Table) were inserted into the V2mO plasmid (pLentiCrispr.V2.mOrange), which was also used as the empty vector control. The resultant clones were screened by the sizes of inserts and verified by sequencing. Next, the plasmids containing the correct gRNA inserts were co-transfected with pCMV.Dr8.2 and pCMV-VSV.G into HEK293T cells with ~70 % confluency using JetPrime transfection reagent. After 48 h, the lentiviral supernatants were harvested and retrieved, and after an additional 24 h, The lentiviral supernatants were centrifuged, and the clear upper solutions were used for immediate transduction or stored in a −80 °C freezer for later use. The target GA0518 cells were seeded in 6-well plates and lentiviral supernatants were added with 10 mM polybrene. The transduced cells were then selected by puromycin at a concentration determined by kill curves for 3–6 days. Surviving cells labeled with clone C and D were then subjected to single-cell cloning, and positive clones were screened by Western blotting with appropriate antibodies. The mixed cells with clone C and D were applied into proliferation assay with AUY922.

2.6. Cell proliferation

The cell proliferation was assessed using a label-free, non-invasive cellular confluence assay by Incucyte Live-Cell Imaging Systems (Sartorius, Ann Arbor, MI, USA). Cells were seeded in 96-well plates at a specific concentration ranging from 1500 to 2500 cells per well and then treated with AUY922 or the corresponding vehicle control (10 % DMSO, 5 % PEG-400, and 85 % PBS) at the specific concentration. The incubation chamber was maintained at 37 °C, and the status of cells was monitored every 12 h for 5–6 days using IncuCyte. Pictures were taken to observe the cell morphology and calculate the occupied area (% confluence) of cell images over time. All experiments were performed in triplicates or quadruplicates to ensure the reliability of the results.

2.7. Crystal violet staining

Cells (1 × 10³) were plated in 6-well plates, with the medium refreshed every 2 days. Afterward, plates were washed with 1 × PBS, fixed with 4 % paraformaldehyde for 20 min, and stained with a crystal violet solution (0.1 % crystal violet in 10 % methanol) for 20 min, followed by rinsing with tap water. Images of the stained plates were captured and analyzed using the ‘Analyze Particles’ function in ImageJ software (NIH, Bethesda, Maryland, USA). The cell-covered area was compared to control cells to calculate a relative cell area ratio.

2.8. Trans-well migration assay

A 24-well transwell plate with 8.0-mm filter inserts (Corning) was utilized. The lower chamber of the transwell plate was filled with 750 μL of RPMI medium containing 20 % FBS. 1–2 × 10⁵ target cells were seeded in the upper chamber of the transwell plate using RPMI medium supplemented with 1 % FBS. AUY922 was initially added at a specific concentration. The plate was incubated at 37 °C for 24–48 h. At the endpoint, non-migrating cells on the upper chamber filter were removed by scraping, and each filter was fixed in 10 % formalin and stained with 0.5 % crystal violet. For visual clarity in the figures, we applied different magnifications and utilized various microscopes to better illustrate the observed differences. All quantitative analysis conducted consistently at 20x magnification. Each assay was performed in triplicate. Images from the trans-well migration assay were captured and analyzed with the ‘Analyze Particles’ function in ImageJ software (NIH, Bethesda, Maryland, USA), where colony counts per well were calculated automatically. Migrated cells on the filter’s lower side were quantified by counting five randomly selected fields under a microscope at 20 × magnification. Each assay was performed in triplicate.

2.9. Tumor sphere formation assay

Target cells were seeded in triplicate on a six-well ultralow attachment plate at a density of 500–800 cells/well. The cells were incubated at 37 °C in serum-free Dulbecco’s modified Eagle’s medium/F-12 medium supplemented with epidermal growth factor (20 ng/mL), insulin (5ug/mL), hydrocortisone (0.5ug/mL), B27 supplement without vitamin A (2 %), and N2 supplement (1 %). AUY922 was initially added at a specific concentration. Cells were grown for 14–21 days of culture and Images of representative fields were taken. Tumor spheres with a diameter >100μm were counted under a microscope. The number of tumor spheres per field and sphere area were automatically calculated using the default settings of the IncuCyte S3 system (Sartorius).

2.10. Cell cycle analysis

The cells treated with AUY922 for 48 h were harvested. Subsequent procedures were performed following the commercial instructions (BD Biosciences). Briefly, the cells were washed with PBS, fixed in cold 70 % ethanol, and then incubated with propidium iodide (PI) for 15 min. Thereafter, cells were analyzed by a flow cytometer (Gallios, Beckman Coulter).

2.11. Annexin-V labeling apoptosis assay

The cells treated with AUY922 for 48 h were harvested. Subsequent procedures were performed following the instructions provided in a commercial Annexin V-FITC assay kit (BioLegend). Briefly, the cells were washed with PBS, suspended in a binding buffer, and then incubated with annexin-V FITC and PI for 10 min. The samples were then analyzed using a flow cytometer.

2.12. Western blotting

The protein was extracted from whole-cell lysates, and the protein amount in RIPA buffer was quantified using a BCA protein assay kit (Thermo Fisher) following the manufacturer’s instructions with bovine serum albumin as a standard. Equal amounts of protein were separated by 10 % polyacrylamide gel electrophoresis and transferred to a PVDF membrane using a TransBlot Turbo transfer system (BioRad). The membranes were incubated with the designated primary antibody overnight at 4 °C, followed by appropriate secondary antibodies conjugated to horseradish peroxidase. Chemiluminescence detection was performed using ECL Western Blotting detection reagent (Thermo Fisher) to visualize the protein bands. The following primary antibodies were utilized at specified dilutions: YAP1 (1:1000), TAZ (1:1000), pYAP (Tyr357) (1:2000), pYAP1 (Ser127) (1:1000), TEAD1 (1:1000), TEAD4 (1:1000), ACTB (1:100000), pLATS1 (Ser909) (1:1000), pMST1/pMST2 (1:333), pS6 (1:1000), pAKT (Ser473) (1:1000), SMAD2/3 (1:1000), HSP90a (1:5000). The applied antibodies were listed in Supplementary Key Resource Table.

2.13. Immunofluorescent staining

Immunofluorescence staining was conducted following previously described methods [34]. For antigen retrieval, Antigen Unmasking Solution (BioGenex laboratories) was used. The following primary antibodies were utilized at specified dilutions: human YAP1 (Santa Cruz Biotechnology, Cat#sc101199, 1:50), human BIRC5 (Survivin) (Cell Signaling Technology (CST), Cat#2803, 1:500), human CTGF (Santa Cruz Biotechnology, Cat#sc365970, 1:50), human Ki67 (Fisher Scientific, Cat# RM-9106-S1, 1:150), human TEAD1 (CST, Cat#12292, 1:100), human TEAD4 (Abcam, Cat#58310, 1:100), human HLA class-I ABC (Proteintech, Cat#15240-1-AP, 1:100), mouse CD8α (CST, Cat#98941, 1:100), mouse CD206 (Abcam, Cat#64693, 1:100), mouse F4/80 (CST, Cat#70076, 1:100). Subsequently, slides were mounted using VectaShield Mounting Medium with DAPI (Vector Laboratories) and visualized under a Nikon T2 confocal laser scanning microscope. YAP1 nuclear localization was quantitatively assessed using Image-J software.

2.14. RT-qPCR

Total RNA was extracted from cells or tissues using Trizol (Thermo Fisher). The extracted RNA was then reverse-transcribed using Luna-Script RT SuperMix Kit (New England Biolabs), and the resulting cDNA was used for RT-qPCR with SYBR Green Master Mix (Applied Bio-systems). The expression levels of the indicated genes were normalized to the expression of GAPDH using the 2^−ΔΔCt method. Each assay was performed in triplicate. Primer sequences were listed in Supplementary Key Resource Table.

2.15. YAP/TEAD luciferase reporter assay

YAP/TEAD transcriptional activity was assessed using the Dual Luciferase reporter system (Promega) with co-transfection of UAS-luc and Gal4-TEAD plasmids [35]. Transient co-transfection of these plasmids into AGS and GA0518 cells with a Renilla vector, was carried out. The human YAP1 overexpression vector was also co-transfected in part of the experiment arms. After 8 h of transfection, the cells were treated with AUY922 (25 nM) or vehicle control for 40 h. Subsequently, luciferase assays were performed using a luciferase assay kit (Promega). Firefly luciferase activity was normalized to Renilla luciferase activity to account for variations in transfection efficiency. The transfection experiments were independently performed at least three times, with each experiment conducted in triplicate.

2.16. Immunoprecipitation assay

We utilized the Thermo Scientific^™ Pierce^™ Classic Magnetic IP/Co-IP Kit (#88804) following the manufacturer’s protocol. Briefly, cell lysates were prepared and incubated with either Hsp90 alpha antibody (GeneTex, GTX109753, 1:100) or YAP1 antibody (Cell Signaling Technology, #14074, 1:50) for IP. The lysate-antibody mixtures were incubated for 1–2 h at room temperature (RT) or overnight at 4 °C to allow antigen-antibody complex formation. Next, Protein A/G magnetic beads were added to bind the complexes for 1 h at RT. The beads were then washed twice with IP Lysis/Wash Buffer and once with purified water to remove unbound proteins. Finally, the antigen-antibody complexes were eluted from the beads for analysis. Antibody concentrations for subsequent Western blotting were consistent with those used in standard Western blot assays, as described above (Supplementary Key Resource Table).

2.17. Chromatin immunoprecipitation (ChIP) analysis

Chromatin immunoprecipitation analysis was performed following previously described methods [36]. Briefly, chromatin was extracted from AGS and GA0518 cells treated with either vehicle control or 25 nM AUY922 for 48 h. The chromatin was then fragmented and subjected to immunoprecipitation using antibodies against YAP1 (CST, Cat#4912), and H3K27ac (Abcam, Cat#ab4729). For ChIP-qPCR, we used the primer pairs to amplify the CTGF and YAP1 promoter region, which includes the TEAD1 binding sequence: Primer of human CTGF promoter, forward: 5′-TGAACATGGAGGTGCAGATATC-3′ and reverse: 5′- TCAGCAAGGATGTGGAGGAA-3’. Primer of human YAP1 promoter, forward: 5′-CGGATTGGACCCATCGTTTG-3′ and reverse: 5′-CGGATTGGACCCATCGTTTG-3’. The TEAD1 binding site was identified through Motif-Map (https://motifmap.ics.uci.edu/), revealing one consensus motif (M0090, CATTCCa/t).

2.18. Reverse-phase protein arrays (RPPA)

RPPA analysis was conducted on GA0518 cell lysates following treatment with either vehicle control or 25 nM AUY922 for 48 h in triplicates. This analysis was performed at the RPPA core facility at the UTMD Anderson Cancer Center. Samples were serially diluted in a 2-fold manner for five dilutions and probed with a panel of unique antibodies, resulting in a total of 422 proteins being assessed. The arrays were mounted on nitrocellulose-coated slides, and relative protein levels were normalized for protein loading. Data was then determined by interpolation of each dilution curve from the standard curve, following established protocols [37]. Differentially expressed proteins were identified using the R package DESeq2. Pathway analysis was conducted using the Hallmark geneset of gene set enrichment analysis (GSEA). Visual representations, including enrichment bars, volcano plots, and heatmaps, were generated through the ClustVis and SRplot online platforms.

2.19. Single-cell RNA sequencing analysis of malignant ascites from GAC patients

Regarding our cohort of GAC patients, 20 malignant ascites samples obtained from advanced GAC patients were studied for scRNA-Seq analysis, following the methods previously reported by our group [38]. The processing of ascites for RNA-seq analysis was performed as described in previous studies. Briefly, collected malignant ascites were centrifuged at 2000 rpm for 20 min, and the cell pellets were resuspended in RPMI1640. The cells were then lysed in ACK lysis buffer, and tumor cells with 10 % DMSO were cryopreserved in liquid nitrogen for subsequent sequencing analysis. Germline DNAs were isolated using the QIAamp DNA Mini Kit (Qiagen), while total RNAs were isolated using the miRNeasy FFPE Kit (Qiagen) following the manufacturer’s instructions. Only samples that passed the sample intake quality check, with gDNA >200 ng and total RNA with an RNA integrity number (RIN) > 7, were further processed for RNA-seq. The cell types and states were determined based on Seurat-defined cell clusters. In the clustering analysis, the 50 nearest neighbors of each cell were determined based on 30 PCs to construct SNN graphs. Differentially expressed genes (DEGs) of each cluster were identified using the FindMarkers function in Seurat. DEG lists were filtered based on the following criteria: the gene should express in 25 % of cells in the more abundant group; expression fold change 1.2; and FDR adjusted p-values (p.adj) ≤0.05. In addition, mitochondrial and ribosomal genes were filtered out from the DEG lists. In parallel, bubble plots were generated for selected DEGs and a suggested set of canonical markers for immune, stromal, and epithelial cell lineages. Multiple layers of information, including the cluster distribution, cluster-specific gene expression in particularly the top 50 DEGs, and canonical cell lineage markers, were integrated and carefully reviewed to define cell types and transcriptomic states.

2.20. Online cohort of GAC patients

Bulk-RNAseq data from human gastric cancers and corresponding normal and adjacent normal tissues obtained from TCGA-STAD and GTEX were downloaded as log2(normalized counts+1) values from UCSC-Xena. Prognostic relevance in GAC patients was analyzed using the Kaplan-Meier plotter (https://kmplot.com/analysis/). YAP1 (224894_at) and BIRC5 (202094_at) probes were utilized, and a total of 875 patient samples were included to evaluate overall survival. The best cut-offs were determined using the default online setting.

2.21. Orthotopic patient-derived xenograft (PDX) mouse model

SCID mice, aged 10–12 weeks, were randomly divided into two groups. Each group was orthotopically inoculated with 0.5 × 10⁶ GA0518-mCh2 or YAP1-KO GA0518-mCh2 cells on the stomach wall inside the abdominal cavity as previously described(56). AUY922 (25 mg/kg, intraperitoneal injection [IP] once a week) or vehicle control (10%DMSO, 5%PEG-400, and 85%PBS) was administered starting 3 days post-tumor injection. Tumor burden was assessed using bioluminescence imaging. For photon quantification, D-luciferin (luciferase substrate) was injected intraperitoneally at a dose of 150ug/kg. After 10 min, the emitted photons from the converted D-luciferin were measured. After 24 days from inoculation, the mice were sacrificed. All orthotopic tumors were collected, weighed, and subjected to subsequent analyses.

2.22. Syngeneic mouse model

Eleven-week-old C57BL/6 mice were subcutaneously inoculated with 1 × 10⁶ Kp-Luc2 cells mixed with Matrigel (at a 1:1 ratio with PBS) on both flanks of each mouse. After 2 days, the mice were evenly divided into four experimental groups: vehicle control (N = 7; 10%DMSO, 5% PEG-400, and 85%PBS), AUY922 (N = 6; 25 mg/kg, IP, weekly), anti-PD-1 antibody (N = 7; 200ug, intravenous injection, twice a week), and a combination of CYD0461 and anti-PD-1 antibody (N = 6). One mouse in the AUY922 group had to be excluded due to a fighting injury that required euthanasia. Consequently, the final analysis was conducted with control (N = 7), anti-PD-1 (N = 7), AUY922 (N = 5), and AUY922 + anti-PD-1 therapy (N = 6). Tumor growth was measured twice a week using a digital caliper. After 3 weeks from the initial inoculation, the mice were sacrificed. Tumor tissue was excised, and its weight was measured. Subsequently, the tissue underwent single-cell isolation and flow cytometry, total RNA extraction for RT-qPCR, and histological analysis, respectively.

2.23. Flow cytometry analysis

Regarding mouse tissue, extracted tumors were digested with 5 mL of RPMI medium containing 2 mg/mL Collagenase-IV (Gibco) and 2 mg/mL Dispase (Gibco) at 37 °C for 30 min, followed by dissociation using the gentleMACS Dissociator (Miltenyi Biotec). Cells were filtered through 70um cell strainers and washed. These dissociated cells were then washed and stained with Live/Dead Aqua dye (1:200, Life Technologies, 405 nm) in PBS, followed by 100uL surface antibody cocktail including CD3⁻ APC-Cy7 (BioLegend, Cat#100222, 1:100), and CD8⁻ Alexa-Fluor700 (BioLegend, Cat#100730, 1:100) diluted in PBS with 0.1 % BSA at 4 °C for 30 min. The cells were washed once with FACS buffer, fixed and permeabilized with the Foxp3/Transcription Factor Staining Buffer Set (eBioscience, Cat#00-5523-00), washed twice with BD Cytofix/Cytoperm (Becton) as manufacture’s instruction, and intracellular stain was performed in Permeabilization buffer with IFNγ-PE-Cy7 (BioLegend, Cat#505825, 1:100) in the dark on ice for 30 min. After washing with FACS buffer, data was acquired using the Gallios561 flow cytometer (Beckman Coulter) and analyzed using FlowJo Ver10. Regarding in-vitro human PBMC after co-culture experiments, the suspended cells were collected directly after washing with FACS buffer, and similarly processed for flow cytometry. The antibody cocktails for staining PBMC separately included CD3⁻ APC-Cy7 (BioLegend, Cat#300318, 1:100), CD8-Alexa Fluor 700 (BD Biosciences, Cat#557945, 1:100), and intracellular staining with GranzymeB-BV421 (BD Biosciences, Cat#563389, 1:100), Perforin-PE-CF594 (BD Biosciences, Cat#563763, 1:100), and IFNγ-PE-Cy7 (BioLegend, Cat#506518, 1:100).

2.24. In-vitro T cell response of PBMCs to AUY922-treated GAC cells

Human PBMCs from GAC patient (GAC43-3993, GAC43-4085, and GAC543-4279) were isolated by density centrifugation as previously described [32]. These PBMCs were treated with 100 IU/ml human-IL-2 (R&D, Cat# 202-IL-010/CF) for 2 days. Subsequently, 1 × 10⁵ PBMCs were co-cultured with either 2.5 × 10⁴ GA0518 or 1.25 × 10⁴ AGS cells after treatment with vehicle control or AUY922 (25 nM, 48 h) in 48-well plates using RPMI-1640 medium supplemented with 10 % FBS for an additional 2 days. The co-cultured PBMCs were then subjected to flow cytometry analysis to characterize cytokine expressions including IFNγ, Granzyme-B, and perforin in CD3⁺ T-cells.

2.25. In vitro T cell response assay

Mouse bone marrow cells were isolated by flushing the tibia and femurs of C57BL/6 mice with RPMI medium supplemented with 10 % FBS. Cells were centrifuged, and red blood cells were lysed using ACK buffer for 3–5 min. The cells were then passed through a 70-μm cell strainer and resuspended in medium containing 10 ng/mL M-CSF (Peprotech). Bone marrow-derived cells were plated in untreated petri dishes (10 mL per mouse) and cultured for 7 days. Bone marrow-derived macrophages were characterized by the expression of CD11b and F4/80 markers.

Murine spleens were dissociated using a 100-μm mesh strainer. The resulting cell suspension was layered on Histopaque^®-1119 (Sigma-Aldrich) and centrifuged at 700×g for 15 min at room temperature with the brake off. Cells from the intermediate phase were collected and washed three times with PBS. Splenocytes were cultured in RPMI 1640 (Corning) supplemented with 10 % FBS, 1 % penicillin-streptomycin, and 50 μM β-mercaptoethanol. T cells were activated with 2 μg/mL anti-CD3e (BD) and 2 μg/mL anti-CD28 (BD) for 24 h, then co-cultured with macrophages and KP-Luc2 cancer cells.

Finally, KP-Luc2 cells were treated with 25 nM and 50 nM AUY922 for 48 h. Following treatment, murine macrophages, activated T cells, and tumor cells were co-cultured for an additional 48 h. T cell responses were subsequently assessed by flow cytometry.

2.26. Immune killing assay

GA0518-mCh2 cells were treated with either vehicle control or AUY922 (25 nM or 100 nM) for 2 days before co-culturing. Human PBMCs were also treated with 100 IU/ml of human-IL-2 (R&D, Cat# 202-IL-010/CF) for 2 days. Following this treatment, the medium of GA0518-mCh2 cells was replaced with new medium, and these GAC cells were seeded in 96-well plates at a density of 6000 cells/well. Next day, GAC cells were co-cultured with 2.4 × 10⁵ human PBMCs at an E: T ratio of 1:4 in RPMI-1640 with 10 % FBS. Cytotox Green Dye (Sartorius, Cat#4633), a reagent capable of detecting cell death, was added at a final concentration of 50 nM. Cell viability and cytotoxicity were monitored every 4 h for 3 days using the IncuCyte system (Sartorius). Tumor cell viability and cytotoxicity were evaluated by calculating the orange area (% confluence), and the green area divided by the sum of the green and orange areas (cytotoxicity index) in the Incucyte system respectively. All experiments were performed in quadruplicate per sample.

2.27. Statistical analysis

All statistical analyses were conducted using GraphPad Prism version 10.0.3 (217). Data were expressed as mean ± standard deviation (SD) or standard error of the mean (SEM), as indicated in the figure legends. For comparisons between two groups, an unpaired two-tailed Student’s t-test was performed. For comparisons among multiple groups, one-way analysis of variance (ANOVA) was used to assess significant differences. Kaplan-Meier survival analysis was used for survival data, and statistical significance was determined using the log-rank (Mantel-Cox) test. The significance level was set at p < 0.05 for all tests. Sample sizes for each experiment are indicated in the figure legends. Statistical tests were chosen based on the experimental design and data distribution, with assumptions checked as appropriate.

3. Results

3.1. Suppression of in-vitro GAC function by AUY922

Given that AUY922 is an HSP90 inhibitor, we employed the primary GAC cell line, AGS, the metastatic GAC cell line, GA0518, which was derived from malignant ascites of a GAC patient [31,32], we first used siRNA to suppress HSP90 expression in GA0518 and AGS cells (Fig. 1A). We then compared the cell growth of control GAC cells with siHSP90-treated GAC cells. As shown in Fig. 2B and C, cell growth was significantly inhibited in the siHSP90 group. To further evaluate the efficacy of AUY922, we conducted several in vitro functional assays with additional murine GAC cell line, KP-Luc2 [32,33]. We first checked the IC50 of GA0518, AGS, and KP-Luc2 cells, which is 2.347 nM, 15.23 nM, and 37.3 nM, respectively (Supplementary Fig. S1A). AUY922 robustly inhibited cell proliferation in a dose-dependent manner (Fig. 1D–E, and Supplementary Fig. S1B). Treatment with 20 nM AUY922 for 72 h reduced cell confluence by 30 % in GA0518 cells and by 50 % in AGS cells. (Fig. 1F). AUY922 significantly reduced the transwell migration ability in a dose-dependent manner (Fig. 1G–I, and Supplementary Figs. S1C–D). Moreover, cancer stem cell (CSC)-like properties are pivotal in driving the aggressive characteristics of cancer cells, including resistance and metastases [15,40]. AUY922 significantly reduced tumor sphere formation in both cell lines, even at very low concentrations such as 1–5 nM (Fig. 1J–L, and Supplementary Figs. S1E–F). Collectively, these findings confirmed that AUY922 robustly suppressed key tumor aggressive functions, including CSC traits, in GAC.

Fig. 1.

AUY922 strongly suppressed tumor aggressiveness in multiple in-vitro functions.

(A) Immunoblotting showing protein expression changes after transfection with siControl or siHSP90 with GA0518 and AGS cells for 48 h in AGS and GA0518 cells. (B) Cell proliferation rates of AGS and GA0518 cells transfected with siControl or siHSP90 for 72 h (N = 3). (C) Representative bright field images of GA0518 and AGS cells after transfected with siControl or siHSP90 for 72 h. (D–E) Cell proliferation rates of AGS (D) and GA0518 (E) cells treated with vehicle control and AUY922 at the indicated concentrations over 150 h (N = 4). (F) Representative images of crystal violet staining of AUY922-treated GAC cells with vehicle control and AUY922 at 10 nM and 20 nM at 72 h (N = 3). (G) Representative images of transwell migration assay of AUY922-treated GAC. (H–I) Transwell migration analysis of AGS (H) and GA0518 (I) cells treated with vehicle control and AUY922 at 2.5 nM and 10 nM (N = 3). (J) Representative images of tumor sphere formation assay of AUY922-treated GAC. (K–L) Tumor sphere formation abilities of AUY922-treated AGS (K) and GA0518 (L) cells (1 nM and 5 nM for 10 days). Representative images of tumor spheres in both cells and their quantitative analysis (N = 3 per group). Data are presented as mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001, ****P < 0.0001 unless otherwise indicated in the graph. Statistical significance compared to control was calculated using one-way ANOVA, and post-hoc Tukey’s HSD tests. N represents the number of biological replicates.

Fig. 2.

AUY922 induces cell cycle arrest and apoptosis in GAC.

(A) Typical cell cycle profile of AGS cells after vehicle or AUY922 treatment (100 nM). (B–C) Percentages of cell-cycle phases in AGS (B) and GA0518 (C) cells after vehicle control and AUY922 treatment (25 nM and 100 nM) for 48 h. (D) Representative dot charts of Annexin-V/-FITC/PI Flow cytometry analysis on GAC cells. (E–F) Percentage of apoptotic cells after AUY922 treatment (25 nM and 100 nM) in AGS (E) and GA0518 (F) cells for 48 h. Data are presented as mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001, ****P < 0.0001. Statistical significance compared to control was calculated using one-way ANOVA, and post-hoc Tukey’s HSD tests. N represents the number of biological replicates.

3.2. AUY922 induced cell cycle arrest and apoptosis in GAC

Induction of cell cycle arrest and apoptosis are important anti-proliferation mechanisms for cancer progression [41]. We investigated whether the growth inhibitory activity of AUY922 involved the control of cell cycle progression and apoptosis induction. We treated AGS and GA0518 cells with various concentrations of AUY922 for 24 h and then analyzed the cell cycle distribution by flow cytometry (Fig. 2A) The AUY922 exposure resulted in the significant accumulation of the proportion of cells in the G2/M phase in both GAC cells (Fig. 2B–C). Further flow cytometry analysis using Annexin-V showed that the proportion of apoptotic cells significantly increased in a dose-dependent manner in both AGS and GA0518 cells upon exposure to AUY922 (Fig. 2D–F). These findings collectively demonstrated that AUY922 significantly induced G2/M cell cycle arrest and apoptosis in GAC cells.

3.3. Clinical significance of the YAP1/TEAD axis and its suppression by AUY922 in GAC

YAP1 signaling is one of the core mechanisms in shaping GAC [13–17]. YAP1 cooperates with TEAD transcription factors to control downstream target genes, including BIRC5, CTGF, CYR61, and more, which play roles in regulating cell proliferation, apoptosis, and senescence [19,42]. To characterize the clinical significance of YAP1 signaling in GAC, we analyzed single-cell RNA sequencing from ascites of GAC patients [38], identifying universal HSP90AA1 expression and a pronounced presence of YAP1, TEAD family, and BIRC5, one of the representative targets of the YAP1/TEAD axis, in EPACM-positive tumor clusters (Fig. 3A and Supplementary Fig. S2A). Notably, YAP1 and BIRC5 expressions were more prominent in the cells from short-term survivors than those from long-term survivors (Fig. 3B). Similar results were observed in TCGA-STAD and GTEX cohorts, where YAP1 and BIRC5 showed elevated expression in tumor tissues compared to normal tissues (Fig. 3C). An independent online GAC patient cohort also demonstrated that higher YAP1 and BIRC5 expression was significantly associated with worse overall survival (Supplementary Figs. S2B–C). These findings underscored the crucial role of the YAP1-TEAD axis in GAC progression.

Fig. 3. Clinical significance of YAP1-TEAD axis and proteomic analysis for AUY922 efficacy on GAC.

(A) UMAP showing expression profiles of HSP90AA1, EPCAM, YAP1, and BIRC5 genes in all PC cells from single cell RNA-seq (scRNA-seq) data. (B) Violin plot showing YAP1 and BIRC5 expressions according to long-term and short-term survivors of GAC patients from scRNA-seq data. (C) YAP1 and BIRC5 gene expressions according to tumor, adjacent normal, and normal tissue in GAC patients’ samples from TCGA-STAD and GTEX cohort. Jonckheere–Terpstra test was used for trend analysis. (D) Gene set enrichment analysis (GSEA) of hallmark geneset from Reverse-phase protein arrays (RPPA) results of AUY922-treated GA0518 cells (25 nM for 48 h, N = 3) compared to vehicle control. (E) Heatmap of phosphorylated protein expressions from RPPA results. (F) Barplots of TEAD1-4 and phospho-YAP1 (Ser127) expression changes from RPPA results. (G) Immunoblotting showing protein expression changes after vehicle control or AUY922 treatments with corresponding concentration for 48 h in AGS and GA0518 cells. (H) Representative image of immunofluorescence staining of YAP1 (red) in AGS and GA0518 cells after AUY922 treatment (25 nM for 48 h, N = 3). Nuclei were counterstained with DAPI (blue). Scale bars indicate 5um. (I) Quantification of YAP1/DAPI ratio in nuclei of AGS and GA0518 cells after vehicle control or AUY922 treatment. Three replicates were analyzed. (J) Immunoblotting showing Hippo pathway protein expressions after vehicle control or AUY922 treatments with corresponding concentration for 48 h in AGS and GA0518 cells. Data are presented as mean ± SD. Statistical significance compared to control was calculated using two-sided Student’s t-tests or one-way ANOVA, followed by post-hoc Tukey’s HSD tests according to sample style. P-values were indicated in the graph. N represents the number of biological replicates.

To gain further insights into the mechanisms and targets of AUY922 against GAC, we conducted a reverse phase protein array (RPPA) analysis on GA0518 cells treated with 25 nM AUY922 and vehicle control. The analysis, highlighted by Principal Component Analysis (Supplementary Fig. S2D), revealed distinct separation between control and AUY922 treatment groups. Out of 422 investigated proteins, 79 were significantly upregulated, and 45 were downregulated compared to control-treated cells (Supplementary Fig. S2E). Pathway analysis using GSEA hallmark gene sets revealed amplified apoptosis and inflammation-related pathways, while the PI3K/AKT/mTOR pathway, cell cycling, and p53-related genes were downregulated (Fig. 3D), as previously substantiated [2,43]. These findings validate our prior phenotypes. Our investigation then delved into the phosphorylation profiles of a broader spectrum of pathway components, revealing the down-regulation of phosphoproteins, primarily associated with the PI3K/AKT/mTOR pathway (Fig. 3E). Intriguingly, TEAD family protein expressions decreased, and phospho-YAP (Ser127), associated with YAP1 inactivation through the Hippo pathway [19,44], was significantly increased (Fig. 3E–F). Immunoblots collectively confirmed that both YAP1 and phosphor-YAP1 (Tyr 357), which activates YAP1 and its translocation to the nucleus and forms a transcriptionally active complex with the TEAD family [19,45,46], demonstrated dose-dependent reductions in response to AUY922 in both GAC cells (Fig. 3G). In contrast, phospho-YAP1 (Ser127) increased in a similar dose-dependent manner. Furthermore, the expressions of TEAD family represented as TEAD1 and TEAD4 were significantly suppressed in accordance with YAP1. While TAZ expression showed a reduction to some extent, it was not as pronounced as the reductions observed in YAP1-TEAD expressions. A similar trend was observed in murine GAC cells (Supplementary Fig. S2F). Additionally, immunofluorescence analyses demonstrated significantly decreased nuclear YAP1 expression after AUY922 treatment with an increase of cytoplasmic YAP1 expressions in both GAC cells (Fig. 3H–I). We further investigated the effects of AUY922 on the Hippo pathway and the other signaling pathways through immunoblots. Surprisingly, AUY922 treatment led to an upregulation of phospho-LATS1 and phospho-MST1 activation in AGS cells, suggesting that HSP90 inhibition activated Hippo pathway signaling (Fig. 3J). In concordance with the proteomic analysis, proteins representative of the AKT-mTOR pathway were significantly decreased in a dose-dependent manner following AUY922 treatment (Supplementary Fig. S2G). These findings collectively underscored the broad influence of AUY922 on multiple protein signaling pathways, with a new highlight on the substantial suppression of the YAP1-TEAD axis in GAC.

3.4. AUY922 inhibited the interaction of YAP1 with HSP90 and TEAD1 along with subsequent transcriptional activity on target genes in GAC

We next elucidated the interaction between HSP90 inhibition and YAP1/TEAD signaling. Representative target gene expressions of the YAP1/TEAD axis collectively decreased after AUY922 treatment in GAC (Fig. 4A, and Supplementary Fig. S3A). Similarly, the protein expression of CTGF and BIRC5 also decreased in tumor cells (Fig. 4B, and Supplementary Fig. S3B). To functionally validate the suppression of this axis, we assessed the YAP1/TEAD interaction on transcriptional activity using luciferase reporter system in both AGS and GA0518 cells (Supplementary Fig. S3C) [35]. TEAD transcriptional activities significantly increased after co-transfection of YAP1, and this activity was reduced after AUY922 treatment (Fig. 4C–D), implying that AUY922 specifically affected YAP1-mediated TEAD transcriptional ability. Further, we examined the interaction between YAP1, HSP90α, and TEAD1 in AGS and GA0518 cells treated with vehicle control or AUY922. Using immunoprecipitation and immunoblotting, we confirmed specific interactions among these proteins, as no binding was observed in the IgG control, validating assay specificity. In both cell lines, YAP1 was found to interact with HSP90α and TEAD1, as shown by IP using antibodies for YAP1 and HSP90α. Upon treatment with AUY922, the interactions between YAP1, HSP90α, and TEAD1 were notably reduced, suggesting that HSP90 inhibition disrupts the YAP1 complex with HSP90α and TEAD1 (Fig. 4E–F). Interestingly, the interaction between TAZ and YAP1 was not affected by AUY922 treatment (Supplementary Fig. S3D). To further determine whether AUY922 blocks the TEAD regulation of the following target genes at the transcriptional level, we performed a ChIP-qPCR assay using CTGF and YAP1 promoter regions that contain TEAD binding sites (Supplementary Fig. 3E). After pulling down with a YAP1 or HSP90α antibody in GAC cells treated with or without AUY922, mRNA expression levels of both CTGF and YAP1 promoter regions were assessed. We confirmed that HSP90 and YAP1 prominently enriched at the TEAD binding sites of both CTGF and YAP1 promoter regions in both AGS and GA0518 cells (Fig. 4G–H), while AUY922 impaired the binding of YAP and HSP90 proteins to these binding sites in a dose-dependent manner. We also investigated the impact on open-chromatin regions using known representative proteins (H3K27ac) involved in transcription activity. AUY922 significantly reduced the occupancy of H3K27ac in the TEAD binding sites of the promoter locus (Supplementary Figs. 3F–G). To further assess whether YAP1 signaling mediates the efficacy of AUY922 in GAC, we used a Lentiviral CRISPR/Cas9 system to knock out YAP1 in GA0518 cells, resulting in a dramatic reduction in YAP1 expression (Fig. 4I). Since YAP1 is the critical activator of TEAD, YAP1 knockout (KO) effectively inhibits the YAP-TEAD pathway [42]. In a cell proliferation assay, YAP1 depletion partially attenuated the tumor-inhibitory effects of AUY922, indicating that YAP1 signaling plays a significant role in AUY922’s mechanism of action (Fig. 4J–K). Collectively, these findings confirmed that YAP1 and TEAD proteins work as one of the HSP90 clients, mediating YAP1/TEAD/HSP90 interactions and transcriptional activity of the YAP1/TEAD axis to downstream target genes in GAC. AUY922 efficiently prevented these interactions and bindings, resulting in the downregulation of target functional genes. Additionally, YAP1-KO models implied that the YAP signaling pathway was one of the core mechanisms explaining the abrogation of tumor aggressiveness in GAC.

Fig. 4. AUY922 inhibits the interaction of YAP1 with HSP90, and TEAD1 along with subsequent transcriptional activity on target genes in GAC.

(A) RT-qPCR analysis of down-stream target genes YAP1/TEAD signaling in AGS cells after vehicle control or AUY922 treatment with corresponding concentrations for 48 h (N = 3). Only significant statistical results are shown. (B) Representative image of co-immunofluorescent staining of BIRC5 (green), and CTGF (red) in the vehicle- or AUY922-treated AGS cells. Nuclei were counterstained with DAPI (blue). The scale bars indicate 25um. (C–D) Luciferase reporter assay measuring YAP1/TEAD transcriptional activity in AGS (C) and GA0518 (D) cells. The cells were treated with vehicle control or AUY922 (25 nM for 40 h after transfection; N = 3). (E–F) Immunoblots represent immunoprecipitated AGS (E) and GA0518 (F) cells after treatment of vehicle control or 25 nM AUY922 for 48 h. GAC cell lysates were immunoprecipitated (IP) with anti-YAP1 and anti-HSP90α or normal IgG antibody, and immunoblotted (IB) for YAP1, HSP90α, and TEAD1. (G–H) Quantitative ChIP-qPCR analysis of the CTGF (G) and YAP1 (H) promoter spanning the TEAD1 binding site after pulling down with YAP1, HSP90α, or IgG antibody in AGS and GA0518 cells treated with vehicle control or AUY922 (25 nM for 48 h; N = 3 per group). (I) Protein expression of YAP1 and TAZ in YAP1-KO GA0518 cells (clone#C and clone#D) compared to control. (J) Cell proliferation rates of YAP1-KO and corresponding parental control GA0518 cells under vehicle control or 10 nM AUY922 (N = 4 per group). †† indicates statistical significance (P < 0.0001) between control and AUY922 groups in YAP1-KO GA0518 cells. (K) Cell confluency ratio of 1 YAP1-KO and corresponding parental control GA0518 cells under vehicle control or 10 nM AUY922 (N = 4 per group). Overall, data are presented as mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001, ****P < 0.0001 unless otherwise indicated in the graph. Statistical significance compared to control was calculated using one-way ANOVA, and post-hoc Tukey’s HSD tests. P-values were indicated in the graph, and N represents the number of biological replicates.

3.5. In vivo efficacy of AUY922 in orthotopic patient-derived xenograft (PDX) model

In our assessment of AUY922’s efficacy in an orthotopic PDX model using GA0518 cells labeled with mCherry (GA0518-mCh2) (Fig. 5A), we observed tumor nodules in the stomach wall, all of which showed histologically as gastric tumors (Fig. 5B). Notably, the total fluorescent signals emitted by tumors after injection were significantly lower in the AUY922-treated group compared to the control (Fig. 5C–D). AUY922 resulted in a significant suppression of tumor growth in injected GA0518-mCh2 cells, as confirmed by the final tumor weight (Fig. 5E–F). Importantly, there were no significant decreases in body weight (Supplementary Fig. S4A), and no fatalities occurred in the treatment group. Ki67 expression, a marker of tumor-associated proliferation, was potently reduced in the AUY922-treated tumor tissues (Fig. 5G–H). Furthermore, in line with previous in-vitro results, both YAP1 and TEAD proteins had significant reductions within the resected tumor tissues, along with decreased BIRC5 expression (Fig. 5G–H). We also conducted the same experiment with YAP1-KO GA0518-mCh2 cells (Supplementary Fig. S4D) and found no significant difference in tumor growth between the control and AUY922 treatment groups (Supplementary Figs. S4B–D), suggesting that the tumor regression efficacy of AUY922 was reversed by YAP1 KO. This finding aligns with our in-vitro tumor proliferation result, highlighting the significance of YAP1 and its downstream signaling in GAC pathogenesis. In summary, AUY922 effectively inhibited tumor growth in the orthotopic PDX mouse models without causing severe adverse effects, and consistently suppressed YAP1-TEAD signaling in the treated tumors.

Fig. 5. In-vivo efficacy of AUY922 using orthotopic patient-derived xenograft (PDX) model.

(A) Experimental design of the orthotopic PDX Model. 0.5 × 10⁶ GA0518-mCh2 cells were orthotopically inoculated on the stomach wall in NOD/SCID mice. Vehicle control or AUY922 (25 mg/kg, weekly) treatment was initiated 3 days after tumor cell inoculation (N = 10 per group). (B) Representative image of H&E staining of resected inoculated tumor on the stomach wall. The scale bars are indicated in the figure. (C) Representative in-vivo bioluminescence images of the orthotopic PDX models on Day 17. (D) Quantification of the bioluminescence images of tumor burden in orthotopic PDX models on Day 17 (N = 10 per group). (E) Representative macro images of the resected stomach of the orthotopic PDX models at sacrifice. Tumor parts are indicated with a yellow dashed line. (F) Final weight of the extracted orthotopic tumors at the endpoint (N = 10 per group). (G) Immunofluorescent stainings of Ki67, YAP1, TEAD1, TEAD4, and BIRC5 expressions (green) in the tumor tissues according to experimental groups as Indicated. Nuclei were counterstained with DAPI (blue). The scale bars indicate 50um. (H) Quantification of the immunofluorescent Ki67, YAP1, TEAD1, TEAD4, and BIRC5 positivity. The ratios of positively stained cells per DAPI-stained cells were analyzed (N = 4 per group). Data are presented as mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001, ****P < 0.0001 unless otherwise indicated in the graph. Statistical significance compared to the control was calculated using two-sided Student’s t-tests. N represents the number of biological replicates.

3.6. Efficacy of AUY922 in syngeneic models and its synergy with anti-PD-1 immunotherapy

AUY922 exhibited modulation of the inflammation-related pathway (Fig. 3D), and previous literature hinted at the potential synergistic effects of HSP90 inhibitors. Motivated by these observations, we explored how AUY922 impacted the TME and its interaction with immunotherapy in GAC. In a syngeneic mouse experiment involving subcutaneous injection of GFP-labeled KP-Luc2 cells (Fig. 6A), both single treatments with AUY922 or PD-1 blockade led to significant tumor regression compared to the control (Fig. 6B). Remarkably, the combination of AUY922 and anti-PD-1 therapy demonstrated the most amplified effects. This synergy was further validated by the final tumor weight (Fig. 6C–D). In line with our orthotopic PDX model results, AUY922 treatment consistently reduced the expression of YAP1 and BIRC5; however, an amplified effect with immunotherapy was not observed (Fig. 6E–F). The density of Ki67-positive cells was drastically reduced across all experimental arms, indicating AUY922’s interference with cell-cycle signaling (Fig. 6E–F). Importantly, no mice exhibited mortality, and no severe adverse effects occurred during the experiment. We also evaluated immune profile changes induced by AUY922 in combination with immunotherapy. Multiple in-situ fluorescence analyses were performed to assess immune cell profiles, revealing a notable increase in the number of infiltrating CD8+T-cells in the tumor nests treated with the combination therapy (Fig. 6G–H). This increase was significantly higher than that observed with any single therapy. In contrast, the density of F4/80+ macrophages within the tumors increased in all treatment groups compared to the control, but the density of infiltrating CD206-positive cells, representing M2-like macrophages with immunosuppressive functions in the TME, was significantly reduced in all treatment groups (Fig. 6G–H). Regarding the cytokine profile, flow cytometry analysis (Supplementary Fig. S5A) of extracted tumors indicated a trend toward increased abundance of CD3⁺ and CD8⁺ Tcells in the combination group (Fig. 6I). The production of IFNγ by CD3⁺ and CD3⁺CD8⁺ Tcells also showed a similar trend (Fig. 6I). RT-qPCR analysis of tumor tissue confirmed the increased presence of inflammation markers, including tumor-killing-related cytokines and macrophage markers, in the combination treatment group compared to the control and single therapies (Supplementary Fig. S5B). In our previous orthotopic-PDX models, CD206 expression in AUY922-treated tumor tissue consistently decreased compared to the control (Supplementary Figs. S5C–D). Collectively, these findings suggested that AUY922 modulated the immune profile, collaborated favorably with immune checkpoint inhibitors, led to an inflamed TME, and enhanced immunotherapy sensitivity in the GAC syngeneic model.

Fig. 6. AUY922 enhances the tumor immune response with immune checkpoint inhibitors in the syngeneic gastric cancer model.

(A) Experimental design of the syngeneic mouse model. 1 × 10⁶ kP-Luc2 cells were subcutaneously injected into C57BL/6J mice in two spots per mouse. Vehicle control, AUY922 (25 mg/kg, weekly), anti-PD-1 antibody (200μg, twice/week), and the combination of CYD0461 and anti-PD-1 antibody were initiated 2 days after tumor cell injection (N = 5–7 per group). (B) Tumor growth of subcutaneous tumors according to the experiment arms. (C) Representative macro images of extracted subcutaneous tumors among the four groups at endpoint. (D) Final weights of the extracted subcutaneous tumors at sacrifice. (E) Immunofluorescent staining of Ki67, YAP1, and BIRC5 Expressions (green) in the tumor tissues. Nuclei were counterstained with DAPI (blue). The scale bars indicate 25um. (F) Quantification of the immunofluorescent Ki67, YAP1, and BIRC5 positivity. The ratios of positively stained cells per DAPI-stained cells were analyzed (N = 6 per group). (G) Immunofluorescent staining of immune cells (green), including CD8, CD206, and F4/80 positive cells in the tumor tissues. Nuclei were counterstained with DAPI (blue). The scale bars indicate 25um. (H) Quantification of the immune cell densities in the tumor immunofluorescent images as indicated. The average counts of positive cells per high-power field were calculated and compared (N = 6 per group). (I) Quantification of the frequencies of CD3⁺ and CD8⁺ T-cells, and IFNγ expressions from those Tcells within the extracted tumors. The variables are normalized with the processed tumor weight (N = 4 per group), as examined by flow cytometry. Data are presented as mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001, ****P < 0.0001 unless otherwise indicated in the graph. Statistical significance compared to the control was calculated using one-way ANOVA, and post-hoc Tukey’s HSD tests. N represents the number of biological replicates.

3.7. TME modulation of GAC by AUY922

To further substantiate the potential effects of AUY922 in the TME of GAC, we co-cultured GA0518 cells treated with either vehicle control or AUY922 with peripheral blood mononuclear cells (PBMCs) from GAC patients, and subsequently assessed cytokine production from Tcells using flow cytometry (Supplementary Figs. S5A and S6A). We observed more prominent Tcell responses when PBMCs were co-cultured with AUY922-treated GA0518 cells as compared to the vehicle-treated cells. Specifically, there was an increase in Granzyme-B, perforin, and IFN-gamma in Tcells following AUY922 treatment in comparison to the control group (Supplementary Fig. S6B). A similar increase in Granzyme-B production was observed when co-culturing PBMCs with AUY922-treated AGS cells, though the other cytokines, while elevated, were not statistically significant (Supplementary Fig. S6C). Considering the interaction with tumor-associated macrophages (TAM) by AUY922 treatment in the mouse models, we performed in-vitro tumor immune microenvironment experiments. We co-cultured AUY922- or vehicle-treated KP-Luc2 cells with primary murine macrophage cells and mouse splenotytes. Consistent with previous findings, cytokine production from Tcells was significantly more pronounced in the group co-cultured with AUY922-treated KP-Luc2 cells compared to the control group (Fig. 7A and B). These trends were also observed in AGS cells (Supplementary Fig. S6D). We further assessed the cytokine expression levels in primary murine macrophages using RT-qPCR, which indicated significant decreases in Ccl2 and Tgfb1 expressions after co-culture with AUY922-treated KP-Luc2 cells (Fig. 7C). These cytokines are known to be associated with TAM polarization from M1 to M2, thus promoting a pro-tumor TME [39]. Next, we specifically evaluated T-cell cytotoxic reaction on AUY922- or vehicle-treated GA0518 cells using an immune-killing assay. A 24-h AUY922 treatment did not change tumor growth (Supplementary Fig. S6E). However, when co-cultured with activated PBMCs, AUY922-treated cells inhibited cell survival under attack by immune cells and displayed greater susceptibility to cell death compared to the vehicle-treated cells (Fig. 7D–F). We further investigated adaptive immune responses via MHC class-I expression to explain this immune-killing phenotype. Collectively, the MHC class-I expressions on GAC cells increased after AUY922 treatment (Fig. 7G–H). In summary, these results confirmed that AUY922 enhanced the susceptibility of tumor cells to immune responses, while also promoting an inflamed TME characterized by an altered cytokine profile and the polarization of tumor-associated macrophages. Overall, AUY922 effectively suppressed tumor aggressiveness, induced cell cycle arrest and apoptosis, and significantly impacted various signaling pathways, notably reducing the YAP1-TEAD signaling in GAC. It also restrained tumor growth in the mouse models, boosted tumor cell susceptibility to immune responses, and promoted an inflamed tumor microenvironment. This collaboration with immune checkpoint inhibitors will enhance immunotherapy efficacy (Fig. 8).

Fig. 7. Modulation of GAC TME by AUY922.

(A–C) Murine splenocytes were treated with 2 μg/ml anti-CD3e (BD) and 2 μg/ml anti-CD28 (BD) for 24 h, followed by co-culture with macrophages and KP-Luc2 cancer cells for 48 h. KP-Luc2 were also treated with vehicle control or AUY922 (25 nM or 50 nM) for 48 h before co-culture. Quantification of IFNγ, and perforin expression in CD3⁺ T-cells after co-culture with KP-Luc2 and macrophages, as examined by flow cytometry (B, N = 3 per group). Quantification of Ccl2 and Tgfb1 expressions in macrophages after co-culture, as examined by RT-qPCR (C, N = 3 per group). (D–F) Representative images (D) of co-cultured GA0518-mCh2 cells before adding PBMCs and 3 days after PBMC addition. GA0518-mCh2 cells were treated with vehicle control or AUY922 (25 nM or 100 nM) for 48 h and seeded into 96-well plates. After 24 h, PBMCs were added and co-cultured over 3 days. Cell viability of GA0518-mCh2 cells was evaluated with the orange area (E), and cell cytotoxicity was evaluated with the green/green + orange area ratio as a cytotoxicity index (F). The cell status was monitored by the Incucyte system every 4 h over 3 days (N = 4 per group). (G) Western blot analysis showing MHC-1 expression change in AGS and GA0518 cells after vehicle control or AUY922 treatment (25 nM or 100 nM for 48 h). (H) Immunofluorescent staining of MHC class-I expressions (red) in AGS and GA0518 after vehicle control or AUY922 treatment (100 nM for 48 h). Nuclei were counterstained with DAPI (blue). Scale bars indicate 20um. Data are presented as mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001, ****P < 0.0001 unless otherwise indicated in the graph. Statistical significance compared to the control was calculated using one-way ANOVA, and post-hoc Tukey’s HSD tests. N represents the number of biological replicates.

Fig. 8. Graphical summary of the effect of HSP90 inhibition by AUY922 on GAC.

Graphical illustration outlines the mechanism of AUY922 in GAC. AUY922 inhibits the aggressive tumor characteristics of GAC by disrupting the YAP1-TEAD axis by inhibiting the interaction among HSP90, YAP1, and TEAD1, as well as subsequent TEAD transcriptional activity. Simultaneously, AUY922 promotes Hippo pathway activity through the phosphorylation of YAP1 (Ser127) and dephosphorylation of LATS1/2 and MST1/2. AUY922 additionally induces changes in the tumor immune microenvironment, shifting it toward an inflamed state, which enhances the responsiveness to immunotherapy.

4. Discussion

AUY922 demonstrated significant inhibitory effects on multiple aspects of gastric adenocarcinoma (GAC), both in vitro and in vivo, aligning well with our research aims (Figs. 1, 5–7). By targeting critical oncogenic pathways, inducing apoptosis, and promoting cell-cycle arrest, AUY922 disrupts essential processes in GAC progression (Fig. 2). The observed down-regulation of YAP1/TEAD signaling and its downstream targets emphasizes the role of this pathway in mediating the effects of HSP90 inhibition (Figs. 3–4). This study provides compelling evidence that AUY922’s modulation of the tumor immune microenvironment (TME) not only enhances immune response against tumors but also produces synergistic effects when combined with immune checkpoint inhibitors (Fig. 6). Taken together, our findings suggest that AUY922’s influence on the TME and its targeted inhibition of the YAP1/TEAD axis position it as a promising candidate for GAC treatment.

Previous research has highlighted the role of HSP90 in stabilizing proteins critical to cell survival, proliferation, and metastasis in various cancers, including GAC [41,47]. The connection between HSP90 inhibition and YAP1/TEAD signaling has emerged from studies that explore the Hippo pathway in the context of heat shock stress [41,44,47–49]. Early studies, such as those by Huntoon et al., demonstrated that HSP90 inhibition with 17-AAG leads to alterations in YAP1 signaling, suggesting that HSP90 might indirectly modulate the YAP1/TEAD axis [44]. However, recent work has shown that heat stress can independently activate YAP1 through HSP90, suggesting that YAP1 activation may depend on specific cellular contexts and stress conditions [47]. In line with these findings, our study uniquely demonstrates that HSP90 inhibition by AUY922 disrupts interactions between YAP1, HSP90α, and TEAD proteins in GAC, which in turn suppresses YAP1/TEAD signaling (Figs. 3–4). This is the first evidence of this regulatory mechanism in GAC, adding a new dimension to our understanding of YAP1/TEAD’s involvement in cancer.

The immunomodulatory effects of AUY922 observed in our study support its potential use alongside immune checkpoint inhibitors in GAC treatment (Figs. 6–7). HSP90 is known to impact several immune processes, including antigen presentation and lymphocyte activation, which are critical for an immune-responsive TME [2,30]. Previous studies have shown that HSP90 inhibition can upregulate MHC-I-related genes, such as MICA and MICB, thus enhancing natural killer (NK) cell activity and tumor surveillance [50]. Consistent with these findings, our data show increased MHC-I expression in GAC cells treated with AUY922 (Fig. 7G), improving immune cell recognition and response [27,30]. Moreover, HSP90 inhibition reduces the expression of EphA2, an immune-suppressive receptor, which enhances the cytotoxic activity of CD8⁺ T cells against tumor cells [28,51,52]. Our results confirmed a significant reduction in phosphate EphA2 expression in GA0518 cells following AUY922 treatment, suggesting improved immune cell engagement (Fig. 3E).

In addition to modulating the TME, YAP1 has been implicated in creating an immunosuppressive environment within tumors by upregulating PD-L1 and recruiting tumor-associated macrophages [53–55]. The inhibition of YAP1 and HSP90 pathways by AUY922 could thus relieve immunosuppressive effects in GAC, facilitating an inflamed and immune-responsive TME. Furthermore, HSP90 inhibition impacts other immune-related pathways, including TGF-β signaling, which we observed through the suppression of SMAD2/3 in AUY922-treated cells (Supplementary Fig. S2G) [4]. This effect on TGF-β signaling further supports the transformation of the TME into an immune-permissive state, providing additional rationale for combining AUY922 with immunotherapies in GAC treatment.

Despite these promising findings, our study has some limitations. In this study, we used GA0518 (diffuse-type GAC) and AGS (intestinal-type GAC) cell lines to represent different GAC subtypes and enhance the biological relevance of our findings. Our results suggest that AUY922’s inhibitory effects on tumor aggressiveness may extend to both GAC types. However, we acknowledge that differences between these cell lines may affect generalizability. Future in vivo validation, RPPA assays, and YAP1-KO experiments in AGS cells will further strengthen the evidence of AUY922’s effects. Likewise, our focus was primarily on HSP90 inhibition by AUY922, providing limited insights into the roles of other heat shock proteins, such as HSP70 and HSF-1, in GAC [2]. Future studies should explore how these proteins interact with the YAP1/Hippo pathway in GAC and other cancers. Moreover, in the loss-of-function experiment of YAP, minimal differences were observed between control and YAP1-KO GAC cells (Fig. 4J). This may be due to compensatory pathways, as the absence of YAP1 can activate TAZ (WWTR1), a paralog with similar functions, which helps maintain cell proliferation and migration. Studies suggest that TAZ upregulation can compensate for YAP1 loss in cancer cells [56,57]. To further investigate, we plan to specifically block the YAP1-TEAD interaction. Additionally, AUY922’s potential interactions with other signaling pathways, such as PI3K/Akt and Src, are known to influence oncogenic functions in GAC and are connected to YAP1/Hippo pathway activity [54,58–60]. While our findings highlight the benefits of HSP90 inhibition, further research is needed to elucidate these interactions comprehensively.

The translational potential of AUY922 lies in its dual role as a YAP1/TEAD pathway inhibitor and an immunomodulator. This study lays the groundwork for future clinical trials evaluating AUY922 in combination with immune checkpoint inhibitors, particularly in advanced GAC, where treatment options are limited. Previous clinical trials have demonstrated AUY922’s efficacy in various solid tumors, including gastric cancer (NCT01084330, NCT01402401, and NCT01613950), indicating its readiness for further clinical investigation in GAC. Furthermore, HSP90 could serve as a biomarker for patient selection, enabling a more targeted approach in clinical settings.

In summary, AUY922’s inhibition of the YAP1/TEAD axis and its enhancement of immune responsiveness suggest a novel therapeutic avenue in GAC treatment. Combination therapy with AUY922 and immune checkpoint inhibitors offers an exciting possibility for improving outcomes in advanced GAC, as supported by our findings and previous studies. This study underscores the need for further research into HSP90 inhibitors as enhancers of immunotherapy in difficult-to-treat cancers like GAC.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.canlet.2024.217354.