CELL INTELLIGENCE · 肿瘤细胞治疗研究

Genome editing has progressed from a laboratory capability for targeted DNA manipulation to a clinically relevant strategy for correcting, silencing, or regulating genes implicated in human disease. In this Review, we synthesize the mechanisms, capabilities, and constraints of the principal programmable platforms-zinc-finger nucleases (ZFNs), transcription activator-like effector nucleases (TALENs), and CRISPR-Cas systems-and highlight how base editors, prime editors, and epigenetic editors expand the range of achievable outcomes beyond double-strand break-dependent repair to precise nucleotide substitutions, small insertions/deletions, and transcriptional modulation. We compare genome-editing cargo formats, including plasmid DNA, viral-vector DNA, mRNA, guide RNA, and ribonucleoprotein complexes, together with the delivery modalities used to transport them, including AAV, adenoviral and herpesviral vectors, lipid nanoparticles (LNPs), electroporation, and virus-like particles. We then consolidate key biomedical applications enabled by these technologies, spanning endogenous gene tagging, high-throughput functional variant screening, molecular recording, and the generation of genetically faithful disease models. Across oncology, respiratory, hematologic, cardiovascular, metabolic, neurodegenerative, viral, ocular, and immune disorders, genome editing is advancing both ex vivo and in vivo interventions, including engineered cellular immunotherapies, hematopoietic stem and progenitor cell editing for hemoglobinopathies, and emerging liver-directed programs for lipid and coagulation targets. Finally, we discuss priorities for broad clinical implementation: improving editing fidelity and PAM flexibility, increasing performance in non-dividing cells, enabling tissue-selective delivery to difficult organs (for example, lung and central nervous system), and addressing manufacturing scalability, long-term monitoring, and equitable global access.