AI-assisted single-cell transcriptomics defines immunosuppressive ecosystem states and CD274-centered communication networks in myelodysplastic syndrome
- 1Department of Hematology, Yueyang Hospital of Integrated Traditional Chinese and Western Medicine, Shanghai University of Traditional Chinese Medicine, Shanghai, China
- 2Department of Hematology, Shanghai Municipal Hospital of Traditional Chinese Medicine, Shanghai, China
Abstract
BackgroundDefining the transcriptional ecosystem states that govern immune dysfunction in myelodysplastic syndromes (MDS) is critical for identifying mechanistically grounded therapeutic targets in this heterogeneous hematopoietic malignancy. The MDS bone marrow immune ecosystem undergoes profound disease-associated remodeling that shifts from immune-mediated hematopoietic suppression in early disease to a globally immunosuppressive state that enables malignant clone expansion, yet the ecosystem-state architecture underlying this transition remains incompletely characterized.MethodsWe deployed an AI-assisted single-cell transcriptomic analytical pipeline—integrating unsupervised Leiden machine learning clustering, multi-modal dimensionality reduction (UMAP, t-SNE, PCA), pathway module scoring, and ligand-receptor inference—on public bone marrow mononuclear cell datasets (GSE241417 as primary MDS scRNA-seq cohort; GSE120221 as external healthy bone marrow reference) to define immune ecosystem states across 11 transcriptionally distinct cell populations. Disease-associated T/NK-cell ecosystem state remodeling, immunosuppressive cytokine and checkpoint programs, and CD274-centered intercellular communication networks were systematically characterized. Key immune regulatory genes (IL10, TGFB1, CD274, FOXP3) were validated by RT-PCR in MDS-derived cell lines, and CD274 perturbation experiments were performed to investigate its regulatory role in shaping immunosuppressive ecosystem programs.ResultsEleven transcriptionally distinct immune ecosystem states were resolved within the MDS bone marrow microenvironment, with disease-associated redistribution toward cytotoxic and exhaustion-like T/NK programs. Pathway analysis revealed enrichment of T cell activation, interferon response, and cytokine signaling across ecosystem states. CD274-high cells were associated with elevated TGFB, checkpoint, and immunoregulatory module scores, while computationally inferred CD274-PDCD1 communication coexisted with TGFB1-TGFBR, LGALS9-HAVCR2, and IL10-related axes, suggesting a multi-layered immunosuppressive network. RT-PCR confirmed increased IL10, TGFB1, and CD274 expression in both MDS-derived cell lines; FOXP3 was significantly increased in SKM-1 but not significantly changed in MUTZ-1. CD274 perturbation was associated with marked IL-10 changes in SKM-1, whereas knockdown produced little apparent IL-10 reduction in MUTZ-1; FOXP3 responses were also cell-line dependent.ConclusionThis study defines the MDS bone marrow immune ecosystem at single-cell resolution using AI-assisted transcriptomic analysis, identifying CD274 as a molecule associated with immunosuppressive ecosystem programs. The findings support a hypothesis-generating model in which CD274-associated signaling may coexist with cytokine and checkpoint communication networks in MDS and may inform future ecosystem-focused therapeutic studies. Further validation using primary patient-derived immune subsets and protein-level assays is required.
Related
- Microbial communication networks governing the ecosystem dynamics of Salmonella in poultry environments: ecological interactions, colonization resistance, and emerging control strategies
- Machine learning–assisted performance prediction of graphene–silicon twin-port band-notched wideband antenna for THz 6G communication systems
- Endoplasmic reticulum remodeling in oocyte cytoplasmic maturation: calcium signaling, organelle communication, and clinical implications for assisted reproduction
- Multicentric study underlining the interest of adding CD5, CD7 and CD56 expression assessment to the flow cytometric Ogata score in myelodysplastic syndromes and myelodysplastic/myeloproliferative neoplasms.
- Extramedullary Hematopoiesis Showing Lymphadenopathy in Myelodysplastic Syndrome without Myelofibrosis
BibTeX
@article{shao2026aiassisted,
title = {{AI-assisted single-cell transcriptomics defines immunosuppressive ecosystem states and CD274-centered communication networks in myelodysplastic syndrome}},
author = {Yacong Shao and Wenwei Zhu and Kai Wang and Changgui Li and Jiekai Li and Hailin Chen and Yongming Zhou},
journal = {Frontiers in Cell and Developmental Biology},
year = {2026},
month = oct,
doi = {10.3389/fcell.2026.1897026},
url = {https://doi.org/10.3389/fcell.2026.1897026},
note = {CC BY},
}
RIS
TY - JOUR
TI - AI-assisted single-cell transcriptomics defines immunosuppressive ecosystem states and CD274-centered communication networks in myelodysplastic syndrome
AU - Yacong Shao
AU - Wenwei Zhu
AU - Kai Wang
AU - Changgui Li
AU - Jiekai Li
AU - Hailin Chen
AU - Yongming Zhou
PY - 2026
DA - 2026/10/30
JO - Frontiers in Cell and Developmental Biology
DO - 10.3389/fcell.2026.1897026
UR - https://doi.org/10.3389/fcell.2026.1897026
AB - BackgroundDefining the transcriptional ecosystem states that govern immune dysfunction in myelodysplastic syndromes (MDS) is critical for identifying mechanistically grounded therapeutic targets in this heterogeneous hematopoietic malignancy. The MDS bone marrow immune ecosystem undergoes profound disease-associated remodeling that shifts from immune-mediated hematopoietic suppression in early disease to a globally immunosuppressive state that enables malignant clone expansion, yet the ecosystem-state architecture underlying this transition remains incompletely characterized.MethodsWe deployed an AI-assisted single-cell transcriptomic analytical pipeline—integrating unsupervised Leiden machine learning clustering, multi-modal dimensionality reduction (UMAP, t-SNE, PCA), pathway module scoring, and ligand-receptor inference—on public bone marrow mononuclear cell datasets (GSE241417 as primary MDS scRNA-seq cohort; GSE120221 as external healthy bone marrow reference) to define immune ecosystem states across 11 transcriptionally distinct cell populations. Disease-associated T/NK-cell ecosystem state remodeling, immunosuppressive cytokine and checkpoint programs, and CD274-centered intercellular communication networks were systematically characterized. Key immune regulatory genes (IL10, TGFB1, CD274, FOXP3) were validated by RT-PCR in MDS-derived cell lines, and CD274 perturbation experiments were performed to investigate its regulatory role in shaping immunosuppressive ecosystem programs.ResultsEleven transcriptionally distinct immune ecosystem states were resolved within the MDS bone marrow microenvironment, with disease-associated redistribution toward cytotoxic and exhaustion-like T/NK programs. Pathway analysis revealed enrichment of T cell activation, interferon response, and cytokine signaling across ecosystem states. CD274-high cells were associated with elevated TGFB, checkpoint, and immunoregulatory module scores, while computationally inferred CD274-PDCD1 communication coexisted with TGFB1-TGFBR, LGALS9-HAVCR2, and IL10-related axes, suggesting a multi-layered immunosuppressive network. RT-PCR confirmed increased IL10, TGFB1, and CD274 expression in both MDS-derived cell lines; FOXP3 was significantly increased in SKM-1 but not significantly changed in MUTZ-1. CD274 perturbation was associated with marked IL-10 changes in SKM-1, whereas knockdown produced little apparent IL-10 reduction in MUTZ-1; FOXP3 responses were also cell-line dependent.ConclusionThis study defines the MDS bone marrow immune ecosystem at single-cell resolution using AI-assisted transcriptomic analysis, identifying CD274 as a molecule associated with immunosuppressive ecosystem programs. The findings support a hypothesis-generating model in which CD274-associated signaling may coexist with cytokine and checkpoint communication networks in MDS and may inform future ecosystem-focused therapeutic studies. Further validation using primary patient-derived immune subsets and protein-level assays is required.
SN - 2296-634X
ER -