Disrupting usp14-mediated PARP1 dynamics reinstates mic-a/b-driven antigen-independent CD8⁺ t-cell killing in glioma
Data files
Feb 12, 2026 version files 846.25 MB
Abstract
Antigen loss is a major mechanism of resistance to immunotherapy. MIC-A/B are stress-inducible ligands expressed by tumour cells that activate NKG2D on cytotoxic immune cells and mediate NKG2D-dependent, antigen-independent tumour cell killing, yet the mechanisms underlying their reduced expression in glioma remain unclear. Using single-cell RNA sequencing and spatial transcriptomics, we investigated ectopic MIC-A/B in mouse glioma and identified USP14 as a key regulator through deubiquitinase screening. Proteomic, coimmunoprecipitation, chromatin immunoprecipitation, immunofluorescence and ubiquitination assays characterized the interactions between USP14, PARP1 and NFIL3, while an intracranial tumour model combined USP14 inhibition and immunotherapy to evaluate effects on tumourigenesis and antitumour immunity. We found that MIC-A/B increased CD8⁺ T-cell infiltration and reversed exhaustion, and that USP14 stabilized PARP1 via K63-linked deubiquitination at lysine 653, reducing NFIL3 binding to the MIC-A/B promoter through poly(ADP-ribosyl)ation. Inhibition of USP14 activated CD8⁺ T cells in a MIC-A/B–NKG2D-dependent, antigen-independent manner, and synergized with PD1 blockade to prolong survival and enhance antitumour immunity. Clinical glioma specimens showed USP14 overexpression was correlated with PARP1 and dysfunctional CD8⁺ T-cell infiltration. Collectively, these results demonstrate that USP14 inhibition restores MIC-A/B-mediated CD8⁺ T-cell activation, reverses immune exhaustion and represents a promising strategy to enhance glioma immunotherapy.
Dataset DOI: 10.5061/dryad.vq83bk47n
This dataset contains processed data to reproduce key transcriptomic analyses testing the hypothesis that restoring stress-ligand signaling (MIC-A/B–NKG2D axis) and/or inhibiting the deubiquitinase USP14 (IU1) can reshape the immune microenvironment in the GL261 mouse intracranial glioma model, consistent with enhanced cytotoxic immune programs. Data are provided as per-sample normalized expression matrices to facilitate downstream reuse.
Description of the data and file structure
The submission comprises three coordinated modules:
(1) CD45⁺ scRNA-seq from MICAB overexpression (MICAB-OE) and matched control tumors (per-sample normalized matrices), enabling immune-cell clustering/annotation and comparison of CD8⁺ T-cell state distributions (e.g., effector vs exhausted programs);
(2) spatial transcriptomics from MICAB-OE and control tumors, including per-sample normalized spot-level matrices and an integrated/combined matrix, supporting analyses that relate condition-associated transcriptional programs to tumor architecture and inflammatory niches;
(3) CD45⁺ scRNA-seq from IU1-treated and matched vehicle/control tumors (per-sample normalized matrices), enabling assessment of treatment-associated changes in immune composition, inferred intercellular communication, and CD8⁺ T-cell activation/exhaustion programs.
This submission contains three compressed archives (.tar), each bundling processed/normalized expression matrices derived from the GL261 mouse intracranial glioma model. Within each archive, data are organized as standard 10x-style matrix bundles for each sample, consisting of three files: barcodes.tsv.gz (cell/spot identifiers), features.tsv.gz (gene/feature identifiers), and matrix.mtx.gz (sparse gene-by-cell/spot matrix in Matrix Market format). Group membership is encoded by the folder structure inside each archive.
Files and variables
File: scRNAseq_matrix_of_MICAB_overexpression_tumours.tar
Description: This archive contains scRNA-seq matrices for CD45+ tumor-infiltrating leukocytes from MICAB overexpression (MICAB) tumors and matched controls (CTRL). It includes 8 sample folders in total: 4 CTRL sample folders and 4 MICAB sample folders. Each sample folder contains barcodes.tsv.gz, features.tsv.gz, and matrix.mtx.gz, representing the normalized expression matrix for that individual biological sample (one tumor from one mouse). In these scRNA-seq bundles, barcodes correspond to single cells and features correspond to genes.
File: Spatial_transcriptome_integrated_data_and_matrix_of_MICABoe_tumours.tar
Description: This archive contains spatial transcriptomics (spot-level) matrices from MICAB overexpression versus control tumors, plus an integrated object. It includes two matrix folders: Ctrl_bc_matrix (control) and MIC-AB_bc_matrix (MICAB overexpression). Each folder contains barcodes.tsv.gz, features.tsv.gz, and matrix.mtx.gz, representing normalized spot-level expression values. In addition, the archive includes brain.merge.rdata, which stores integrated/merged data across sections/samples (including slice-level information) to facilitate cross-sample spatial analyses. For spatial data, barcodes correspond to spatial capture spots (not single cells).
File: scRNAseq_matrix_of_IU1_treated_tumours.tar
Description: This archive contains scRNA-seq matrices for CD45+ tumor-infiltrating leukocytes from tumors treated with the USP14 inhibitor IU1 and matched controls. It includes 6 sample folders in total: 3 DMSO control sample folders and 3 IU1-treated sample folders. As in Archive 1, each sample folder contains barcodes.tsv.gz, features.tsv.gz, and matrix.mtx.gz, representing the normalized expression matrix for that individual biological sample.
Abbreviations used. MICAB: MICA/B overexpression group. MICAB-OE: MICA/B overexpression (used synonymously with MICAB where applicable). CTRL: control group for the MICAB perturbation. IU1: treatment with the USP14 inhibitor IU1. DMSO: vehicle control group for IU1 treatment. scRNA-seq: single-cell RNA sequencing (here, CD45+ tumor-infiltrating leukocytes). Spatial transcriptomics (ST): spot-level gene expression profiling from tumor tissue sections. bc_matrix: barcode matrix bundle in standard 10x format (barcodes.tsv.gz, features.tsv.gz, matrix.mtx.gz). barcodes.tsv.gz: list of cell/spot identifiers. features.tsv.gz: list of gene/feature identifiers. matrix.mtx.gz: sparse expression matrix in Matrix Market format. brain.merge.rdata: an R data object containing integrated/merged spatial data and slice-level information derived from the spatial matrices.
Access information
Links to other publicly accessible locations of the data:
- Not applicable. All processed/normalized matrices are available in this Dryad submission.
Data was derived from the following sources:
- Not applicable. Matrices were generated from the authors’ sequencing experiments and processed using standard pipelines as described in the associated manuscript.
Code/Software
All analyses were performed in R (version 4.4.1). The scRNA-seq and spatial transcriptomics data are provided primarily in standard 10x-style Matrix Market format (barcodes.tsv.gz, features.tsv.gz, matrix.mtx.gz), which can be imported into R and analyzed using Seurat-based workflows. The integrated spatial object (brain.merge.rdata) can be loaded directly in R for inspection and downstream analysis, depending on the object type stored in the file.
Key R packages used for data processing, integration, visualization, and downstream analyses include: Seurat, SeuratObject, SeuratData, tidyverse (including dplyr, tibble, tidyr, readr, stringr, ggplot2, tidyselect), harmony, glmGamPoi, DoubletFinder, zellkonverter, reshape2, RColorBrewer, scRNAtoolVis, COSG, ComplexHeatmap, cols4all, patchwork, jsonlite, hdf5r, clusterProfiler, org.Mm.eg.db, and enrichplot, as well as trajectory-related packages monocle2 and monocle3. These packages support typical single-cell/spatial workflows such as quality control, normalization, batch correction/integration, clustering and cell-state annotation, differential expression, pathway enrichment, visualization, and trajectory inference.