Leukemia Cells Evade the Innate Immune System by Expressing “Don’t Eat Me” Signals

In back-to-back papers in the journal Cell, two related studies from the laboratory of Dr. Irving Weissman at Stanford University demonstrate that CD47 is expressed on normal hematopoietic stem and progenitor cells and that CD47 expression on leukemia cells is a poor prognostic indicator. CD47 is a transmembrane protein that serves as a “don’t eat me” signal to prevent phagocytosis by endogenous macrophages, thus allowing cells to evade the innate immune system. Specifically, when a protein called Signal Regulatory Protein-α (SIRPα) on macrophages binds to CD47 on a neighboring cell, it transmits a signal that prevents phagocytosis by the macrophage, thus protecting the CD47 cell. Most cells express CD47, but expression decreases as cells undergo senescence. In this context, senescent cells undergo additional changes, such as exposure of phosphatidyl serine and calreticulin, which lead to activation of neighboring macrophages to remove the senescent cells. This occurs routinely in the reticuloendothelial system for aged neutrophils, erythrocytes, and platelets.

The investigators found that murine hematopoietic stem cells (HSC) mobilized to exit the bone marrow by G-CSF had up-regulated CD47 expression. Similarly, human CD34+ peripheral blood stem cells had higher CD47 expression than resident bone marrow CD34+ cells. To demonstrate that CD47 expression on HSCs prevents them from being destroyed, the investigators transplanted cells from mice with low or absent levels of CD47 into wild-type recipients. They found that the cells were unable to engraft, suggesting that CD47 expression is necessary for survival of transplanted HSCs. If the recipient’s endogenous macrophages were first destroyed using a drug, CD47-low cells were able to engraft, validating the hypothesis that uptake by macrophages is the normal means by which CD47-low cells are prevented from engrafting.

The investigators also tested the hypothesis that CD47 expression on leukemia cells would allow increased cell survival and correlate with long-term prognosis. They used a CD47 negative human leukemia cell line engineered to express mouse CD47 and found that expression of CD47 inhibited phagocytosis in vitro and promoted leukemic engraftment in recipient mice. When primary AML cells from 13 different patients were assessed, all but four had higher levels of surface CD47 than control CD34+ bone marrow cells. They then examined CD47 expression in 636 adult AML samples. There were no differences in CD47 expression among the FAB subtypes but those leukemias with an activating mutation in the FLT3 gene had higher CD47 levels. Mutation of FLT3 is a poor prognostic factor in AML; it is associated with both decreased response to chemotherapy and decreased long-term survival. Approximately 25 percent to 35 percent of AML patients harbor FLT3 mutations, irrespective of FAB subtype. However, the correlation between the expression level of CD47 on leukemia cells and prognosis (both time to relapse and overall survival) is not unique to AML with FLT3 mutations; there is a strong correlation even when patients with FLT3 mutations are excluded.

The investigators then tested whether pharmacologic blockade of the CD47- SIRPα interaction could lead to enhanced phagocytosis of leukemia cells and improved survival. This was demonstrated both in vitro and in vivo using tumor cell lines that expressed different levels of CD47. Because CD47 is expressed on most cell types in addition to hematopoietic cells, CD47 expression on malignant cells is likely not exclusive to leukemia. Dr. Weissman’s group recently demonstrated that CD47 is expressed on cancer stem cells in bladder cancer.¹