01
Adoptive T cell therapies show promise against hematologic malignancies, but their effectiveness against solid tumors is limited due to hostile tumor microenvironments that impair mitochondrial function and lead to T cell exhaustion. Traditional approaches to enhance T cell mitochondrial fitness often focus on single targets and fail with dysfunctional mitochondria. Recent studies highlight intercellular mitochondrial transfer via tunneling nanotubes (TNTs), which could repair damaged T cells.
Bone marrow stromal cells (BMSCs) significantly modulate T cell function and differentiation. In co-culture studies with CD8+ T cells, complex nanotube structures bridging BMSCs and T cells were observed, confirming intercellular connections. Confocal microscopy revealed that mitochondria were actively transferred from BMSCs to CD8+ T cells via these nanotubes, with the transferred mitochondria maintaining their membrane potential and increasing mitochondrial content in recipient T cells. Genetic analysis confirmed the presence of donor mtDNA in CD8+ T cells, demonstrating effective mitochondrial transfer and integration. Mito+ cells exhibited significantly higher respiration rates and spare respiratory capacity.
Gene expression analysis (bulk and single-cell RNA sequencing) identified TLN2 as upregulated in Mito+ cells, indicating its role in nanotube formation. To assess TLN2's specific role, CRISPR-Cas9 gene editing was employed to knockout TLN2 in both CD8+ T cells and BMSCs, revealing that its absence significantly impaired mitochondrial transfer rates, particularly from BMSCs.
Mitochondrial transfer enhances the respiration and spare respiratory capacity (SRC) of CD8+ T cells, providing an energetic advantage in hostile tumor microenvironments. Following adoptive transfer into tumor-bearing mice, Mito+ cells demonstrated enhanced tumor regression and prolonged survival compared to Mito− cells. And Mito+ cells showed increased engraftment and expansion within tumors. Moreover, the scRNA-seq analysis revealed that Mito+ cells retained a more robust effector profile, exhibiting lower levels of exhaustion markers compared to Mito− cells, which were predominantly terminally exhausted. The study also assessed T cell metabolism using SCENITH, demonstrating that Mito+ cells maintained an advantageous metabolic state, particularly in nutrient-limited tumor conditions.
Lastly, this technology was applied to human CD8+ T cells, which were transduced with a CD19-specific CAR and co-cultured with Mito-dsRed BMSCs. The results showed enhanced cytotoxicity against leukemia cells in vitro and improved survival in vivo. Notably, Mito+ CAR T cells maintained functionality across multiple stimulation rounds, while Mito− cells skewed to exhaustion.
DOI: 10.1016/j.cell.2024.08.029
02
Neoantigen-based
cancer immunotherapy has gained significant attention for treating various
tumors. However, accurately identifying immunogenic neoantigens remains
challenging due to the limited number of mutant peptides that elicit strong
antitumor responses, highlighting the importance of understanding TCR–pHLA
interactions. Current predictive models primarily focus on pairwise
interactions, overlooking the complex TCR-antigen-HLA landscape and often
lacking interpretability, hindering effective immune response modeling.
To address limitations, researchers proposed the physics-inspired sliding transformer (PISTE) model, which integrates physical and biological principles to enhance prediction accuracy and interpretability. The PISTE pipeline for neoantigen screening involves obtaining amino acid sequences of TCR–antigen–HLA triples, predicting binding scores using PISTE, and ranking the peptides based on their immunogenicity. Candidate peptides are synthesized and tested for immunogenicity through T cell assays. The PISTE model includes a sequence encoder and a sliding-attention module to evaluate interactions at both residue and sequence levels, and it has been benchmarked against eight state-of-the-art models. Experimental results demonstrate that PISTE significantly outperforms competitors across various datasets and evaluation metrics, providing valuable insights into TCR–pHLA interactions and enhancing our understanding of immune responses.
PISTE not only predicted sequence-level binding accurately but also generates attention scores that revealed residue-level interactions, allowing for the identification of interacting residue pairs and their distributions. Through virtual mutagenesis, PISTE assessed the impact of mutations on binding outcomes, revealing that specific mutations in both TCRs and antigens significantly affected binding scores. Notably, mutations in certain CDR3 segments and specific amino acids, such as arginine and tyrosine, had a greater impact on the predicted binding status.
To evaluate the clinical utility of PISTE, a series of immunological investigations were conducted, including T cell clonal expansion analysis, the identification of immunogenic neoantigens within tumor microenvironments, and the validation of personalized neoantigen-driven T cell responses. The study found a positive correlation between predicted TCR–pHLA binding scores and T cell clonal expansion. PISTE identified neoantigens in skin cutaneous melanoma and glioblastoma, demonstrating that neoantigens exhibited greater immunogenicity than their wild-type counterparts, with PISTE-predicted immunogenic neoantigen load correlating with patient survival rates. Additionally, PISTE facilitated the identification of personalized neoantigens in prostate cancer patients, showing a 75% response rate in T cell assays, further confirming its potential as a predictive tool for neoantigen-based immunotherapy.
DOI: 10.1038/s42256-024-00901-y
关注我们获取更多免疫学文献资讯
Editor & Reviewer: Shiyang Song
