G-CSF is often used to hasten neutrophil recovery following allogeneic bone marrow transplantation (BMT), but the clinical and immunological consequences invoked remain unclear. We examined this in murine models and found that administration of both standard G-CSF and pegylated G-CSF early after BMT significantly increased graft-versus-host disease (GVHD). This effect was seen in the B6 → B6D2F1, BALB/c → B6 and C3H.SW → B6 systems of GVHD to either MHC or multiple minor histocompatibility antigens. This effect was dependent on total body irradiation (TBI) rendering host dendritic cells (DC) responsive to G-CSF by up-regulating their expression of the G-CSF receptor as determined by real-time PCR. This induction of G-CSFR expression was not seen following busulfan (Bu), cyclophosphamide (Cy) or fludarabine. The enhanced GVHD was present when G-CSF was administered to both WT and G-CSFR−/− donors but not G-CSFR−/− recipients, confirming that host signalling was critical for this effect. G-CSF administration after BMT had no effect on inflammatory cytokine generation but enhanced in vivo CTL activity after BMT when administered to WT but not G-CSFR−/−, CD1d−/−, IFNgR−/− or CD40−/− recipients. Furthermore, donor iNKT cell activation was absent in CD11c Diptheria Toxin Receptor recipient transgenic mice depleted of dendritic cells (DC) by diphtheria toxin and treated with G-CSF. Thus, stimulation of host DC by G-CSF subsequently unleashed a cascade of events characterized by CD1d dependent donor iNKT cell activation, IFNg secretion and CD40-dependent amplification of donor CTL function during the effector phase of GVHD. Critically, the detrimental effects of G-CSF on GVHD were present when administered early following TBI conditioning and at a time when residual host APC were still present (day +1), but had no effect when administered at day +8 when host DC were not detectable by phenotypic or functional analysis. This is consistent with the inefficient cross presentation of host Ag within MHC class I by donor DC after BMT. In addition, the administration of G-CSF after Bu/Cy conditioning had no effect, perhaps explaining the conflicting and somewhat controversial clinical studies from the large European and North American BMT registries since TBI conditioning predominated only in the positive European study. These data have major implications for the use of G-CSF in disease states where NKT cell activation may have important effects on outcome and suggest a guide to the safe use of G-CSF after allogeneic BMT.

Disclosures: No relevant conflicts of interest to declare.

Author notes

Corresponding author