Glycosylated GM-CSF expands B-1b cells and B-1b plasma cells and programs them for immunosuppression

The myeloid growth factor granulocyte-macrophage colony-stimulating factor (GM-CSF) exhibits paradoxical pro- and anti-inflammatory functions, but the factors determining these divergent outcomes remain unclear. Here, we report that this functional divergence is controlled by its glycosylation. Murine recombinant fully glycosylated GM-CSF (rgGM-CSF) specifically induces immunosuppressive cell types, whereas its recombinant non-glycosylated counterpart (rngGM-CSF) promotes effector cells. Using single-cell ATAC-sequencing and flow cytometry, we show that rgGM-CSF has a previously unrecognized ability to effectively expand IL-10+ LAG-3+ PD-L1+ B-1b plasma cells (PCs) with immunosuppressive properties. Although rgGM-CSF also promotes the expansion of hematopoietic stem and progenitor cells (HSPCs) and monocytic myeloid-derived suppressor cells (M-MDSCs), adoptive transfer experiments clearly demonstrate that only the rgGM-CSF-induced B-1b PCs are primarily responsible for an IL-10-dependent long-term protection in mice from experimental autoimmune-encephalomyelitis (EAE). Our data suggest that glycosylation enhances the systemic bioavailability and activity of GM-CSF and promotes the expansion of immunoregulatory cells rather than pro-inflammatory myeloid effector cells. Together, these results demonstrate that the 'dual activity' of GM-CSF is controlled by its glycosylation, resulting in opposing immune functions. These findings support a re-evaluation of human rgGM-CSF (regramostim) as a potential therapeutic strategy for immunosuppression in transplantation and autoimmune diseases. Overall design: Single-cell ATAC-sequencing (scATAC-seq) was performed to characterize chromatin accessibility changes in splenic hematopoietic stem and progenitor cells (HSPCs) and associated immune cell populations following in vivo administration of glycosylated GM-CSF. C57BL/6 mice were injected intraperitoneally once daily for 10 days with recombinant fully glycosylated GM-CSF, while untreated mice served as controls. On day 11, spleens were harvested and single-cell suspensions were prepared. HSPCs were enriched by fluorescence-activated cell sorting as lineage-negative (Ter119?, B220?, CD4?, CD8?), CD11b?, c-Kit? cells. To enable biological replication within a single sequencing run, cells from one wild-type and one luciferase-transgenic mouse per condition were pooled at a 1:1 ratio prior to library preparation. The luciferase transgene sequence was used as a genetic barcode to assign individual cells back to their respective animals during downstream bioinformatic analysis.