CD200 and PD1-L1 in AML Are Associated with Expanded PD-1+ Late Differentiated CD8+ T Cells and a Decreased CD4:CD8 Ratio: a New Link Between Distinct Immunosuppressive Pathways

Long-term remission for acute myeloid leukemia (AML) is still not achieved for the majority of patients and consequently there is a need for new treatments to consolidate current therapy. A promising approach is to augment the anti-tumor immune response in these patients; however most cancers do not activate immune effector cells because they express immunosuppressive ligands. Previously we showed that CD200 overexpression on AML blasts suppresses memory CD4+ and CD8+ T cell effector function through engagement with CD200 receptor (CD200R) on these cells. Blocking CD200:CD200R, however, only partially restored T cell activity, suggesting that alternative immunosuppressive mechanisms were involved. Recently, promising clinical outcomes have been reported for melanoma and non-small cell lung cancer using humanized antibodies targeting another immunosuppressive receptor, PD-1, and we therefore investigated whether this could be contributing to the immunosuppression of T cell effector responses in CD200hi AML patients.

Initially, we investigated whether CD200 and the immunosuppressive ligand for PD-1, PD1-L1, were co-expressed in AML blasts at diagnosis. Affymetrix gene expression data from 158 AML blasts showed that AML patients in the upper quartile for CD200 expression (CD200hi) had 10-fold higher levels of PD1-L1 expression compared to CD200lo (lower quartile) patients. Analysis of CD200 and PD1-L1 protein expression on AML blast cells confirmed this association at the protein level (r2 = 0.49; p<0.01). The co-expression of CD200 and PD1-L1 on patient AML blast cells, suggested that they cooperated in immunosuppression. In support of this, we found that the CD200 and PD1-L1 cognate co-receptors (CD200R and PD-1 respectively) were present on CD4+ and CD8+ T cells from AML patients. Further characterization of PD-1+ T cells showed that the mean frequency of PD-1+ early differentiated T cells (CD57- CD28+) was increased for CD200hi AML patients CD4+ (19% ± 3 vs 13% ± 3; p<0.05) and CD8+ T cells (21% ± 3 vs 11% ± 2; p<0.05). We also found that the mean frequency of late differentiated CD8+ T cells that have poor anti-tumor function (CD57+ CD28- PD-1+) was almost twice that for CD200hi patients compared with CD200lo (38% ± 6 vs 21% ± 9 respectively; p<0.05). Expansion of these cells was also associated with a decreased CD4:CD8 ratio in these patients (2.1 ± 0.5 vs 3.7 ± 1 for CD200hi and CD200lo respectively; p<0.01). . These findings show for the first time a link between CD200 expression level on AML blast cells and the frequency of PD-1+ late differentiated CD8+ T cells. To directly test whether engagement of CD200 with CD200R was capable of mediating PD-1 up-regulation on CD8+ T cells, we co-cultured a CD8+ CD200R+ T cell clone (7E7) either with K562 cells stably overexpressing CD200 or K562 empty vector controls (negative for CD200). Co-culture with CD200+ cells, significantly increased the frequency of PD-1+ T cells (26% ± 3 vs 17% ± 4; p<0.05) and this was antagonized by CD200 blocking antibody (26% ± 3 vs 21% ± 3; p<0.01). These data show that CD200:CD200R interaction has the capacity to increase the frequency of PD-1+ CD8+ T cells. To model the functional implications of this, we created a series of K562 lines expressing CD200, PD1-L1 or both molecules in combination and analyzed the effect of T cell activation (via TNFα production). We found that both CD200 and PD1-L1 induced a similar (>50%) reduction in the frequency of activated 7E7 T cells; however, when both CD200 and PD1-L1 were co-expressed, T cell activation was almost ablated (~90% reduction; p<0.01). Moreover, the strength of the TNFα response was also reduced in co-culture assays where either CD200 or PD1-L1 were present, indicating a direct effect at the level of CD8+ T cell function (2.8 ± 0.5 vs 1.7 ± 0.5; p<0.05). These data demonstrate that CD200:CD200R and PD1-L1:PD-1 engagement on T cells can act in tandem to augment immunosuppression of CD8+ T cells.

In summary, we show for the first time that the immunosuppressive molecules, CD200 and PD1-L1 appear to be co-regulated on AML blasts and that these can act in combination to profoundly suppress T cell activation. Further, we show that CD200:CD200R engagement induces PD-1+ CD8+ T cells. Taken together we propose a novel CD200/PD1-L1 immunotherapeutic synapse in AML which should be targeted by combining CD200:CD200R and PD1-L1:PD-1 blockade in immunotherapy of AML.