Acute Myeloid Leukemia Cell-Intrinsic PD-1 Functions Promoting Leukemia Development By Recruiting SHP2

Acute myeloid leukemia (AML) is a devastating hematopoietic malignancy. With current therapies, only approximately 30% of patients achieve long-term survival. Therefore, novel, more active and less toxic treatments are urgently needed. Programmed death-1 (PD-1) is a cell surface receptor that functions as a T cell checkpoint and plays a central role in regulating T cell exhaustion. Binding of PD-1 to its ligand, programmed death-ligand 1 (PD-L1), activates downstream signaling pathways and inhibits T cell activation. Abnormally high PD-L1 expression on tumor cells and antigen-presenting cells in the tumor microenvironment mediates tumor immune escape, and PD-1/PD-L1 immune checkpoint blockade has showed promising results in cancer patients. Recently, PD-1 expressed on melanoma cells was also shown to play a pivotal role in tumor growth. To date, in AML, the function of PD-1 has been mainly studied in the host T cells, while little is known regarding the role of PD-1 in AML cells. Herein, we examined the level and role of PD-1 in AML cells using AML murine model and patient samples.

We used MLL ^PTD/WT/Flt3 ^ITD/ITD knock-in mouse in B6 background, a well characterized AML model, to study the expression and function of PD-1 in AML. Flow cytometric analysis of LSK (Lin ^-Sca-l1 ⁺c-kit ⁺) cells from the bone marrow (BM) of wild-type (WT, n=5) and AML (n=10) mice showed that 20.9%-61.9% of AML LSKs versus (vs) <5.0% of normal LSKs are PD-1 positive (P< 0.0001). Western blot and Q-RT-PCR analysis confirmed higher levels of PD-1 in AML LSKs than in normal LSKs. The PD-1 levels on LSKs increased over time and associated with disease progression. Similar results were obtained in AML patients, showing PD-1 ⁺ in 0.2%-14.6% of AML CD34 ⁺ cells vs < 1.0% of normal CD34 ⁺ cells (P< 0.05). Data analysis based on TCGA showed that higher PD-1 levels are associated with shorter survival in AML patients (P=0.0125).

To assess the functional role of PD-1 in AML, we sorted PD-1 ⁺ and PD-1 ^- fractions fromAML LSKs and observed a lower frequency of quiescent cells (G0, 16.60% vs 44.87%, P< 0.05) and a higher cell growth rate in the PD-1 ⁺ vs PD-1 ^- cells. Further in vivo study showed that PD-1 ⁺ AML LSKs (CD45.2) generated higher white blood cell (WBC) counts (P< 0.0001), higher AML engraftment (P< 0.0001) and a shorter survival (median survival 57.5 vs >75 days, P< 0.001) in recipient mice (CD45.1) compared with PD-1 ^- AML LSKs. Similar results were observed in human samples. Compared to PD-1 ^- CD34 ⁺ cells, PD-1 ⁺ CD34 ⁺ cells are less quiescent and more proliferative (P< 0.01). PD-1 ⁺ AML blasts had higher engraftment rate (13.18% vs 2.68%, p=0.0002) and shorter survival (median survival: 27 vs 45 days, P= 0.0008) in NSGS mice than PD-1 ^- AML blasts.

To evaluate if these in vivo differences observed in PD-1 ⁺ vs PD-1 ^- AML LSKs were mediated by interactions between PD-1 ⁺ AML and T cells, PD-1 ⁺ and PD-1 ^-LSKs from AML mouse were transplanted into T-, B- cell-deficient NSG mice. Recipient mice transplanted with PD-1 ⁺ AML LSKs had higher WBC counts (P< 0.01), higher AML engraftment (P< 0.0001) and a shorter survival (median survival: 76 vs >130 days, P< 0.0001) than recipients with PD-1 ^- AML LSKs, suggesting that these differences were T cell-independent.

Next, we examined whether blocking PD-1 could affect leukemic cell growth. We sorted LSK cells from AML mice and performed colony forming cell (CFC) assay in the presence of anti-PD1 antibody or isotype antibody. Blocking PD-1 with anti-PD-1 antibody significantly suppressed CFC and cell growth in vitro but did not induce apoptosis compared to isotype control antibody. To explore the molecular mechanism by which PD-1 contributes to AML growth, we then sorted PD-1 ⁺ and PD-1 ^- LSKs from AML mice for molecular studies. Western blot assays revealed higher levels of SHP-2 and phosphorylated (p) -ERK in PD-1 ⁺ vs PD-1 ^- AML LSKs. We validated these results in primary human AML cells by immunofluorescence staining. Confocal microscopy of PD-1 ⁺ and PD-1 ^- human AML CD34 ⁺cells demonstrated that PD-1 localized at the cell membrane and in the cytoplasm and p-ERK was markedly enhanced in the PD-1 ⁺ CD34 ⁺ cells.

In conclusion, we showed here that a subpopulation of murine and human AML blasts expresses high levels of PD-1 which mediated disease initiation and growth through activation of the MAPK/ERK signaling pathways. PD-1 blocking antibody reversed these activities and might contribute to the clinical efficacy of anti-PD-1 therapy in AML.