�����ƽ��

Cell surface proteins offer significant cancer therapeutic potential attributable to their accessible membrane localization and central roles in cellular signaling, yet their promise remains largely untapped due to technical challenges inherent to profiling them. Here, we employ N-glycoproteomics to analyze 85 patient-derived xenografts (PDXs), constructing Glyco PDXplorer—an in vivo pan-cancer atlas of cancer-derived surface proteins. We develop a target discovery pipeline to prioritize proteins with favorable expression profiles for immunotherapeutic targeting and validate FAT2 as a squamous-cancer-enriched surface protein minimally detected in normal tissue. Functional studies reveal that FAT2 is essential for head and neck squamous cancer (HNSC) cell growth and adhesion through regulation of surface architecture and integrin-PI3K signaling. Chimeric antigen receptor (CAR)-T cells targeting FAT2 demonstrate anti-tumor activity. This work lays the foundation for developing FAT2-targeted therapies and represents a pivotal platform to inform therapeutic target discovery across cancers. Graphical abstract Highlights•Pan-cancer landscape of cancer-derived cell surface proteins detected in vivo•Discovery pipeline to prioritize proteins as immunotherapy target candidates•Validation of FAT2 as an SCC surface protein with minimal normal tissue expression•FAT2 CAR-T cells demonstrate anti-tumor activity in pre-clinical models Govindarajan et al. leverage N-glycoproteomics and PDX models to decode the in vivo cancer cell surfaceome and establish Glyco PDXplorer—a target discovery platform. The identification and validation of FAT2 as a previously undescribed, actionable antigen demonstrates the utility of Glyco PDXplorer for uncovering therapeutic vulnerabilities.

Hsp90 is a molecular chaperone that is often overexpressed across multiple cancer types and has a potential value as a prognostic marker as well as a therapeutic target. Given the high interest in Hsp90 therapies, positron emission tomography or PET imaging of Hsp90 can be a valuable tool for patient selection. The limitations of the previously developed Hsp90 tracers prompted us to evaluate the recently developed brain-permeable [ 11 C]HSP990 PET probe to advance the development of Hsp90-targeted therapeutics. Given the brain accumulation of [ 11 C]HSP990 probe, application for glioblastoma imaging of this tracer is of particular interest. In vitro [ 11 C]HSP990 binding was assessed in breast cancer and glioma cell lines including patient-derived cells using Hsp90 inhibitors and RNA interference knockdown of Hsp90 isoforms. Saturation binding studies were conducted on these cells and tumour tissue homogenates, and autoradiography was performed on tissue sections. Ex vivo biodistribution and in vivo dynamic µPET/CT studies were performed in healthy mice and tumour-bearing mice, including immunocompromised subcutaneous human U87 and MDA-MB-231models and immunocompetent intracranial murine NS/CT-2A models at baseline and following a pre-treatment with Hsp90 inhibitors. High Hsp90-specific tracer uptake was observed in breast cancer and glioma cells, with Hsp90β inhibition resulting in the most substantial reduction in uptake. In vivo uptake was high in U87 tumours but low in MDA-MB-231, presumably due to the differences in Hsp90 expression in tumour tissue versus cultured cells. Differences in maximum binding capacity or B max across cell and tissue types support this hypothesis, especially given that the affinity measured as dissociation constant K d remained similar across all tissue types. Despite high NS/CT-2A tumour uptake in vitro, no contrast between the healthy brain tissue and the NS/CT-2A glioma was observed in vivo due to the high uptake by the healthy brain. [ 11 C]HSP990 is a promising tracer for identifying Hsp90-overexpressing tumours and may hold potential for patient stratification, prognosis, and therapy monitoring of novel Hsp90 therapeutics. High healthy brain uptake of this tracer precluded the differentiation of the tumour in the intracranial NS/CT-2A tumour model, therefore [ 11 C]HSP990 might not be a suitable tracer for the glioblastoma imaging. Tracer with a longer half-life might be needed to compare the washout of the tracer from the brain and the tumour tissue over several hours to identify a suitable imaging window. The online version contains supplementary material available at 10.1186/s41181-025-00386-z.

Histone deacetylase inhibitors (HDACis) are promising anti-cancer agents but remain underexplored in glioblastoma (GBM). This study evaluated the effects of three HDACis—CAY10603, vorinostat (SAHA), and valproic acid (VPA)—on human GBM cell lines (U87, MGG8) with immortalized human astrocytes (IHAs) as healthy controls. HDACis were tested alone or in combination with temozolomide (TMZ), the standard chemotherapy for GBM, in both 2D (monolayer) and 3D (neurosphere) cultures. Additionally, co-culture of GBM cells with macrophages (M0, biochemically differentiated from THP-1 human monocytes) was used to examine the impact of HDACis on cancer-immune interactions. Results demonstrated that all three HDACis significantly reduced cell viability and synergistically enhanced the effect of TMZ. CAY10603 and SAHA induced early apoptosis and upregulated caspase 3 (CASP3) expression, whereas VPA primarily induced late apoptosis and necrosis in GBM cultures. VPA induced both G0/G1 and G2/M cell cycle arrest, while SAHA and CAY10603 only induced G2/M arrest. mRNA expression analysis following HDACi treatment in U87 neurospheres revealed that HDACis inhibited expression of markers for epithelial-to-mesenchymal transition (EMT), proliferation, and stemness pathways. In U87-M0 co-cultures, we observed significant upregulation of stemness markers and the pro-inflammatory cytokine TNF-α following CAY10603 and VPA treatments. In contrast, TMZ monotherapy upregulated the expression of the immunosuppressive cytokine TGF-\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\beta\:$$\end{document}. These findings suggest that HDAC inhibition—including the novel small molecule CAY10603—sensitizes GBM to temozolomide and confers potent anti-tumor effects that combat GBM (e.g., reducing proliferation, EMT, stemness). Among the HDAC inhibitors tested, CAY10603 exhibited the most potent anti-tumor effect in 3D neurosphere and macrophage co-culture models, significantly enhancing apoptosis and disrupting pro-tumorigenic and anti-inflammatory signaling in GBM. Our in vitro findings —e.g., with 3D neurospheres that better mimic physiological tumor growth than 2D monolayers—warrant future in vivo testing of HDACis alone or in combination with chemotherapy.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-20749-5.