AUTHOR=Sun Tianyue , Chen Yan , Chen Ying-Xuan 

TITLE=Single-cell and bulk transcriptome analyses reveal elevated amino acid metabolism promoting tumor-directed immune evasion in colorectal cancer

JOURNAL=Frontiers in Immunology

VOLUME=Volume 16 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/immunology/articles/10.3389/fimmu.2025.1575829

DOI=10.3389/fimmu.2025.1575829

ISSN=1664-3224

ABSTRACT=IntroductionColorectal cancer (CRC), the third most common cancer worldwide, often shows limited responsiveness to immunotherapy due to its predominantly immune-excluded phenotype. Despite increasing insights into the complex tumor microenvironment (TME), the metabolic heterogeneity of CRC cells and their interactions with tumor-infiltrating immune cells remain poorly understood.MethodsWe analyzed 46,374 epithelial cells from 17 CRC patients treated with PD-1 blockade to develop an amino acid (AA) metabolism score using the AUCell algorithm. This score was applied to a separate single-cell RNA sequencing (scRNA-seq) dataset from 23 CRC patients to investigate cell-cell interactions and functions of tumor-infiltrating immune cells, revealing distinct immune TME landscapes shaped by tumor metabolism. An in vitro co-culture assay of CRC cells and CD8+ T cells was performed to validate the findings. Additionally, LASSO and Cox regression analyses were conducted to construct an AA metabolism-related risk score for predicting prognosis and drug sensitivity across multiple bulk transcriptome cohorts.ResultsThis study identified a link between elevated amino acid metabolism in CRC epithelial cells and resistance to PD-1 blockade therapy. A 31-gene AA score was developed by intersecting differentially expressed genes between responders and non-responders to PD-1 blockade with amino acid metabolism-related genes from the Molecular Signature Database (MSigDB). Using this score, 23 additional CRC samples were classified into high and low AA score groups. Comparative analysis revealed that the low AA group exhibited a more robust immune response, characterized by a greater number and stronger cell-cell interactions. Tumor-infiltrating immune cells in this group demonstrated enhanced activation and anti-tumor functions. Furthermore, CD8+ T cells showed increased Granzyme B levels when co-cultured with CRC cells in which Psat1 or Shmt2 was knocked down. Finally, a machine learning-derived risk score based on six genes was established to translate single-cell findings to bulk transcriptomes. This risk score was found to correlate with immune checkpoint expression and immune cell infiltration, with potential implications for predicting prognosis and drug sensitivity.ConclusionOur findings highlight the role of elevated epithelial amino acid metabolism in shaping an immune-suppressive microenvironment, offering insights for patient stratification and therapeutic decision-making.