AUTHORS
Allison Banuelos, PhD, Leyla Yilmaz2, James Chen, MD, PhD3, Irving Weissman, MD4, Jonathan M Tsai, MD, PhD5
AFFILIATIONS
1 Stanford University, Stanford, California, United States
2 Harvard Medical School, Boston, Massachusetts, United States
3 Stanford University, Stanford, California, United States
4 Stanford University, Stanford, California, United States
5 Harvard Medical School, Boston, Massachusetts, United States, Harvard Medical School, Boston, Massachusetts, United States
ABSTRACT
Therapeutic CD47 blockade induces rapid removal of CD47 from circulating red blood cells (RBCs), resulting in the acute RBC toxicity associated with anti-CD47 therapy. Although this process, termed RBC pruning, occurs without RBC destruction, the cellular mechanism responsible for selective CD47 removal remains unknown.
We first investigated the cellular and structural requirements for RBC pruning. Pruning remained intact following macrophage and neutrophil depletion and in Fcγ receptor-deficient mice, demonstrating that professional phagocytes and Fcγ receptor signaling are dispensable. Treatment with intact anti-CD47 IgG or bivalent F(ab')₂, but not monovalent Fab fragments, demonstrated that antibody bivalency, rather than the Fc domain, is required for RBC pruning. Furthermore, antibody-mediated CD47 removal required crosslinking between RBCs and CD47-expressing non-hematopoietic cells, suggesting that interactions with non-hematopoietic cells drive membrane remodeling.
To identify the initiating cell type, RBC membranes were metabolically labeled with Ac4ManNAz prior to transfusion and tracked following anti-CD47 treatment using click chemistry. Labeled membrane material was first detected predominantly on splenic endothelial cells during the first 12 hours after treatment before accumulating within splenic macrophages at 24 hours, suggesting that endothelial cells initiate membrane remodeling while macrophages subsequently clear released membrane fragments.
To directly test this model, wild-type bone marrow was transplanted into CD47-deficient recipients, restoring CD47 expression exclusively within the hematopoietic compartment while maintaining CD47 deficiency in non-hematopoietic tissues. Anti-CD47-induced RBC pruning was significantly reduced in these chimeras, demonstrating that non-hematopoietic CD47 is required for efficient membrane remodeling. Importantly, endothelial-specific deletion of CD47 (Cdh5-Cre;CD47 fl/fl) likewise significantly reduced RBC pruning, identifying endothelial CD47 as a critical mediator of this process. Finally, to determine whether antibody-mediated mechanical tethering is sufficient to drive RBC pruning, we developed a cell-free mechanical model in which RBCs were flowed over immobilized anti-CD47 under physiological shear. This minimal system recapitulated in vivo RBC pruning while preserving RBC integrity in the complete absence of endothelial or immune cells, demonstrating that antibody-mediated mechanical tethering is sufficient to drive selective RBC membrane remodeling.
Together, these studies identify splenic endothelial cells as the initiating mediator of anti-CD47-induced RBC pruning. We propose that therapeutic anti-CD47 antibodies mechanically tether circulating RBCs to endothelial CD47, where physiological shear forces drive selective removal of CD47-containing membrane without RBC destruction, followed by macrophage clearance of the released membrane fragments. These findings redefine the mechanism underlying anti-CD47-associated RBC toxicity and identify endothelial-dependent mechanical membrane remodeling as a previously unrecognized mechanism of therapeutic antibody action.
DISCLOSURES
Allison Banuelos, PhD: No relevant conflicts of interest to declare.