Résumé
Introduction: CD22-targeted chimeric antigen receptor (CAR) T-cell therapy has shown promising efficacy in relapsed/refractory B-cell acute lymphoblastic leukemia (B-ALL), including in patients who previously failed CD19 CAR T-cell therapy. However, despite high initial response rates, approximately two-thirds of patients who achieved complete remission eventually relapse, often with leukemia exhibiting reduced CD22 surface expression. The underlying mechanisms of CD22 modulation and CAR resistance remain incompletely understood.
Methods: To identify leukemia-intrinsic mechanisms of resistance to CD22 CAR T-cells, we performed a genome-wide CRISPR/Cas9 knockout screening in the B-ALL cell line NALM6, using the human GeCKOv2 pooled guide RNA (gRNA) library. Cells were co-cultured with CD22 CAR T-cells and surviving populations were analyzed to identify enriched gRNAs. Top hits were validated using CRISPR-mediated gene knockout followed by co-culture cytotoxicity assays, high throughput flow cytometry, cytokine profiling, and transcriptomic analyses. Additional knockouts of specific glycosyltransferases were used to dissect downstream signaling mechanisms.
Results: Unexpectedly, genes directly modulating CD22 surface expression accounted for only a small fraction of the CD22 CAR-resistant populations. One of the most highly enriched gRNAs following CD22 CAR T-cell treatment was SPPL3 (Signal Peptide Peptidase Like 3), a Golgi-resident protease that regulates glycosylation by cleaving multiple glycosyltransferases.
Importantly, CRISPR-mediated SPPL3 knockout (SPPL3KO) did not significantly reduce CD22 surface detection by flow cytometry, strongly suggesting that resistance was not driven by antigen loss. Indeed, we found that CAR T-cells were able to recognize CD22 on SPPL3KO cells. Nonetheless, the absence of SPPL3 did confer resistance to CD22 CAR T-cell cytotoxicity, especially under stress conditions with low effector:target (E:T) ratios.
SPPL3 knockout leukemic cells also demonstrated impaired responsiveness to immune and inflammatory cues. Bulk RNA-seq of WT and SPPLKO NLM6 exposed to CD22 CAR T-cells revealed blunted lymphocyte activation signatures and suppression of non-canonical NFkB signaling, suggesting impaired leukemia-intrinsic activation in the absence of SPPL3.
To identify downstream mediators of SPPL3-mediated resistance, we targeted glycosyltransferases known to accumulate in SPPL3-deficient cells. Based on prior reports implicating B3GNT2 (Zhuang et al. Cell Reports 2024) and B3GNT5 (Jongsma et al. Immunity 2021) in immune escape pathways, we generated SPPL3KO NALM6 cells with additional knockout of BcGN2 or B3GNT5. Strikingly, loss of B3GNT2 fully restored CD22 CAR T-cell-mediated cytotoxicity and activation, while knockout of B3GNT5 loss partially rescued sensitivity to CD22 CAR T-cell pressure. These findings indicate that dysregulated glycosylation can underly resistance by modulating leukemic cell responsiveness to immune effectors rather than through loss of antigen recognition.
Conclusions: Our study identifies a novel antigen-independent mechanism of resistance to CD22 CAR T-cell therapy in B-ALL, driven by loss of SPPL3 and altered glycosylation. SPPL3 deficiency impairs immune sensing and leukemia activation without reducing CD22 surface expression or CAR recognition. Restoration of CAR sensitivity via glycosyltransferase editing reveals a targetable axis to overcome resistance. These findings suggest that glycosylation-driven resistance to CAR T-cell cytotoxicity could represent a generalizable barrier to efficacy, warranting broader evaluation across current and emerging CAR T-cell therapies.