Graft-versus-host disease (GVHD) limits the broader application of allogeneic bone marrow transplantation. While initial T cell activation in GVHD occurs predominantly in secondary lymphoid organs, we have consistently observed MHCII+ donor-derived APCs in histopathologic GVHD lesions in tissues such as skin and intestine, frequently adjacent to infiltrating T cells. We hypothesized that productive interactions occur between donor APCs and T cells in situ in GVHD target tissues, which propagate disease. To address this hypothesis we utilized two photon intravital microscopy to analyze interactions between fluorescently labeled donor CD8+ T cells and tissue infiltrating donor dendritic cells (DCs), within skin lesions of living mice in a murine GVHD model. Lethally irradiated 129 mice received T cell-depleted (TCD) marrow from C57BL6 (B6) mice expressing yellow fluorescent protein (YFP) driven by the CD11c promoter (B6-CD11cYFP), B6 red fluorescent protein (RFP)+ CD8+ T cells and WT (unlabeled) B6 CD4+ T cells. The skin of the ear was imaged in anesthetized, living mice, between days 21 and 28 post-transplant, using a LaVision two photon laser scanning microscope, scanning 60uM thick z-stacks every 30 seconds for 1 hour. Individual RFP+ CD8+ T cells were tracked over time throughout the 3 dimensional image. Detailed surface maps of the YFP+ dendritic cell (DC) network were generated. A distance transformation to calculate the distance of each RFP+ CD8+ T cell from the surface of the YFP+ DC network at all times was performed. Through these analyses, we observed both highly motile donor CD8+ T cells making contact with donor DCs (defined as <=2μM between RFP+ CD8+ T cell and YFP+ DC surface), and a proportion which make long term, stable contact (continuous contact between RFP+ CD8+ T cell and YFP+ DC for at least 30 minutes, during which the RFP+ CD8+ T cell speed is below 5.5μM/min). To test whether CD8+ T cell:DC interactions required cognate TCR:MHCI interactions, lethally irradiated 129 mice received a 1:1 mixture of WT B6 and MHCI-deficient B6 BM. In group 1, labeled marrow was from B6-CD11cYFP, and unlabeled marrow from B6-β2-microglobulin (β2m)−/− donors. In group 2, labeled marrow was from B6-CD11cYFP/β2m−/− mice, and unlabeled marrow from B6. In addition to these bone marrow mixtures, all mice received B6 RFP+ CD8+ T cells and unlabeled B6 CD4+ T cells. Mice were imaged as above. Long lasting contacts between donor RFP+ CD8+ T cells and YFP+ donor DCs in skin lesions were less frequent when YFP+ DCs were β2m−/−, and therefore MHCI-deficient. We have also examined whether MHCII-dependent interactions occur between CD4+ T cells and DCs in situ in skin lesions. In preliminary experiments 129 mice received B6-CD11cYFP bone marrow, B6 RFP+ CD4+ T cells, and B6 unlabeled CD8+ T cells. After 1 hour of imaging, mice received anti-MHCII antibody or isotype control and imaging was continued for 2 hours thereafter. RFP+ CD4+ T cell motility increased after anti-MHCII but not after isotype control antibody treatment. Because GVL occurs primarily in BM and spleen, targeting of tissue-infiltrating APCs could represent a unique strategy to ameliorate GVHD while preserving GVL.
No relevant conflicts of interest to declare.
Asterisk with author names denotes non-ASH members.