Noncanonical role of NCOR1 as a facilitator of DNA mismatch repair and its deficiency sensitizes cancers to immune checkpoint blockade therapy [scRNA-seq]

Tumor genomic heterogeneity significantly influences the efficacy of immune checkpoint blockade (ICB) therapy. To elucidate the underlying mechanisms, we pioneered the development of an ICB response score (IRS) based on machine learning analysis of microenvironmental features. Through a pan-cancer genomic screening, we identified the nuclear receptor corepressor NCOR1 as a key regulator, the tumors with genomic deficiency of which display a tumor infiltrating lymphocyte (TIL)-inflamed microenvironment. Genetic ablation of Ncor1 delays the tumorigenesis and metastasis in syngeneic and spontaneous tumor models in immunocompetent mice. Mechanistically, NCOR1 guards DNA mismatch repair by directly interacting with MSH2 and recruiting MutSa to chromatin for mismatch recognition, independent of its transcriptional regulatory activity coupling with histone deacetylase. Ncor1 depletion in hepatocellular carcinoma and melanoma cells promotes tumor immunogenicity by increasing tumor mutation burden (TMB), neoantigen load, and TIL infiltration and activation, thereby enhancing anti-PD-1 therapy efficacy. Our findings reveal that NCOR1 loss represents a novel mechanism underlying TMB-high tumors with intact MMR components, broadening the application of ICB. Overall design: Subcutaneous xenografts derived from Hepa1-6 and B16 cells were generated in immunocompetent C57BL/6J mice. Subcutaneous xenografts were dissociated and the single-cell suspensions were obtained using the MACS Miltenyi Mouse Tumor Dissociation kit (130-096-730, Miltenyi Biotec, Germany) on gentleMACs dissociator with Octo Heaters (130-096-427, Miltenyi Biotec, Germany) according to manufacturer's instructions. Cell suspensions were passed through a 70-µm filter twice. CD45+ immune cells were then sorted from the cell suspension using flow cytometry (Cat# 35-0451-U100, TONBO Biosciences, USA), and then subjected to the BD Rhapsody Single-Cell Analysis System (BD Biosciences, USA) following the manufacturer's guidelines. The scRNA-seq libraries were constructed using BD Rhapsody WTA Amplification Kit (BD Biosciences, USA). WTA libraries and LMO libraries were pooled and sequenced on NovaSeq 6000 (Illumina, USA), and the data was processed using the BD Rhapsody WTA analysis pipeline (v1.9.1; BD Biosciences, USA). WTA reads were aligned to the mm10 reference mouse genome, allowing a maximum of 1 mismatch. All samples were merged in Seurat and high-quality cells were kept with a detection threshold of 500 - 5,000 genes and a proportion of expressed mitochondrial genes less than 10%. After quality control, a total of 9,693 cells from 3 groups of HCC samples (sgCtrl, sgNcor1-1 and sgNcor1-2), and 4,910 cells from 3 groups of melanoma samples (sgCtrl, sgNcor1-1 and sgNcor1-2) were retained for downstream analyses. Data of different samples were integrated utilizing Harmony (Korsunsky et al., 2019) to identify anchors. Then, data were log-normalized (divided by the total expression and amplified scaling factor 10,000) using the “NormalizeData” function in Seurat. For clustering, we implemented the “FindVariableFeatures” function in Seurat to select the top 2,000 highly variable genes (HVGs). A principal component analysis (PCA) matrix was calculated to reduce noise using the “RunPCA” function in Seurat with default parameters. For visualization, the dimensionality was further reduced using “RunUMAP” function in the Seurat with parameters n.neighbors = 50, min.dist = 0.1, spread = 1, and metric = euclidean. We then used the function “FindClusters” with resolution = 1.0 to perform the first-round clustering and annotated each cluster by known cell type markers.