Allogeneic editing: making donor cells invisible

Allogeneic cell therapy is a deal with the immune system. A donor T cell carrying a CAR would be perfect therapy except for two facts: it sees the patient’s tissues as foreign, and the patient’s immune system sees it as foreign. Gene editing removes both objections at their molecular roots - knocking out the T-cell receptor ends graft-versus-host before it can start, and knocking out HLA markers hides the cell from host T cells. The result is a dose that can be manufactured once and given many times, which is the entire economic argument of the off-the-shelf model. The editing stack has converged on a few cuts. TRAC removes the native T-cell receptor - no graft-versus-host. B2M strips HLA class I - no CD8-mediated rejection of the dose. The CAR itself is inserted at a defined locus so expression is set by the editing design rather than random integration. What remains open is innate immunity: a cell stripped of HLA class I looks to a natural-killer cell exactly like a virus-infected cell, and NK attack on HLA-deficient allogeneic products is the field’s standing counter-question, with engineered HLA-E or CD47 expression the published counter-ideas, still in clinical testing. The editing choices carry the field’s intellectual property and its regulatory history. TALEN engineering at Cellectis produced the UCART line that reached FDA RMAT designation in June 2026 with a Phase 1 reporting complete response in relapsed B-ALL. CRISPR editing at CRISPR Therapeutics produced the first approved CRISPR medicine and an end-to-end GMP facility in Framingham - and that approved base is autologous: sickle-cell disease in December 2023, transfusion-dependent thalassemia in 2024, children from age two in 2026, with the allogeneic portfolio funded downstream of it. Natural-killer cells take a third route: they do not cause graft-versus-host, so cord-blood-derived lines need less editing and bank more easily - the Artiva model, with an FDA Fast Track in refractory rheumatoid arthritis, 741.9 million dollars raised and a cGMP site in San Diego, capacity measured in patients per year rather than doses per quarter. Banking is the quiet revolution. Once doses are inventoried in liquid nitrogen, apheresis logistics, manufacturing delays and patient deterioration before infusion - the three chronic failures of autologous therapy - disappear from the value chain. What replaces them are release testing per lot, cold-chain integrity, and review frameworks that had to invent categories for an edited, banked, donor-derived product: RMAT in the US, ATMP in Europe. One logistics note keeps the model honest: even off-the-shelf doses ride on a lymphodepleting conditioning regimen to open the niche - what disappears is the apheresis and the per-patient manufacture, not the clinical protocol. The honest limits are immune persistence and attrition: edited allogeneic cells do not last as long as autologous ones, and the banked-dose model lives or dies on durability data still being generated. The sibling pages own the rest of that map: autologous CAR-T practice, NK platforms, and the iPSC substrate technology this field increasingly borrows.