SIRPα controls CD47-dependent platelet clearance in mice and humans

Abstract

Over the last decade, more data has revealed that increased surface expression of the "don't eat me" CD47 protein on cancer cells plays a role in immune evasion and tumor progression, with CD47 blockade emerging as a new therapy in immuno-oncology. CD47 is critical in regulating cell homeostasis and clearance, as binding of CD47 to the inhibitory receptor SIRPα can prevent phagocytosis and macrophage-mediated cell clearance. The purpose of this study was to examine the role of the CD47-SIRPα signal in platelet homeostasis and clearance. Therapeutic reagents targeting the CD47-SIRPα axis are very promising for treatment of hematologic malignancies and solid tumors, but lead to transient anemia or thrombocytopenia in a subset of patients. We found that platelet homeostatic clearance is regulated through the CD47-SIRPα axis and that therapeutic blockade to disrupt this interaction in mice and in humans has a significant impact on platelet levels. Furthermore, we identified genetic variations at the SIRPA locus that impact platelet levels in humans such that higher SIRPA gene expression is associated with higher platelet levels. SIRPA expression at either end of the normal range may affect clinical outcomes of treatment with anti-CD47 therapy.

Figure 1.. CD47 blockade induces splenomegaly, splenic red pulp expansion from erythropoiesis, and increased megakaryocytes.

In vivo, CD47 blockade leads to splenomegaly with extramedullary erythropoiesis. Mice were treated with MIAP410 (clone-3 anti-CD47) for 2 weeks. (A) Spleens were removed and photographed with visible increase in size and pigment of spleens from animals who received the CD47 blockade treatment (bottom) compared to the IgG control treated animals (top). (B) Spleen sections 5uM thick were stained with anti-TER119 demonstrating expansion of erythroid cells in the red pulp of the spleen. (C) Hematoxylin and eosin (H&E) staining of interfollicular spaces of spleens demonstrates that the white pulp of the spleen is diminished in anti-CD47-treated mice, with the space now devoted to extramedullary erythropoiesis and platelet production, with significant increase in megakaryocytes evident in anti-CD47-treated mice (representative megakaryocytes shown in the boxes). (H&E) images at 200 μM resolution.

Figure 2.. Platelets are reduced in number but increases in volume upon CD47 blockade in vivo.

Complete blood counts were run after 2wks of anti-CD47 blockade in C57/B6 mice. A) Platelet count shown in thousands per cubic mL. B) Platelet related percentages shown as indicated, MPV: mean platelet volume (average platelet size), P-LCR: platelet large cell ratio, PCT: volume occupied by platelets in blood as percentage C) Reticulocyte absolute count shown per microliter. D) Reticulocyte related percentages shown as indicated, IRF: immature reticulocyte fraction, MFR: middle-fluorescence reticulocytes, LFR: low-fluorescence reticulocytes. CBC counts were run on two separate comparisons of CD47 blockade, one representative experiment is shown n=3 mice per group, significance of unpaired T-test comparisons shown as (*p ≤.05, **p ≤.01, ***p ≤.005).

Figure 3.. Variants at SIRPA are associated with platelet levels.

(A) Variation at 3’ to CD47 at BBX locus associated with platelet levels. (B) Variant rs3197744 at SIRPA locus associates with amount of platelets. (C) Examining associations across currently available GWAS results (PheWAS) over 2500 independent studies at rs3197744 at SIRPA locus shows specific association with platelet and immune cell measurements. (D) Expression profile of SIRPA across human tissues shows highest expression in the brain and in blood.

Figure 4.. Schematic representation of genetic variation affecting structure, expression and disease associations on SIRPA locus.

Single nucleotide polymorphisms are marked on top of the gene structure. Missense variants are highlighted in red as part of the gene structure and outlined yellow below the gene structure. Frameshift/insertions or deletions are highlighted with red as part of the missense variants. Haplotype from rs6132062 to rs6136377 encompassing exon 2 has the largest impact on SIRPA gene expression in blood. In addition, exon 2 contains missense variants that can be divided into two common haplotypes that are referred to as V1 and V2. V1 allele is shown first in the graph. V1 is in linkage disequilibrium (r2=1) with the SIRPA eQTLs (rs6132062 and rs6136377) having impact therefore both on expression and on coding sequence. Disease associations with SIRPA are located in the 3’UTR including the variants that associate with platelet levels.

Figure 5.. SIRPA expression and broader context of platelets in human disease.

(A) TWAS analysis of SIRPA expression with human phenotypes. X-axis shows effect on expression by Z-statistics. (B) Association of leading SIRPA and CD47 SNPs with stroke excluding subarachnoid hemorrhage (SAH) and coagulation traits that have been implicated in platelet function. (C) Analysis of causality using Mendelian randomization from platelet levels to stroke, (D) and to coronary artery disease.

Figure 6.. SIRPA and CD47 are inversely correlated, and are independently associated with platelet count in ET patients.

(A) Correlation of platelet normalized gene expression between CD47 and SIRPA for patients with essential thrombocythemia and healthy donors. Closed circles navy or light blue indicate sample type respectively of healthy donors as controls (CTRL, n=11) and patients with essential thrombocythemia (ET, n=23). Inverse correlation is indicated by a dotted curve as a guide to the eye. Platelet gene expression as assessed by normalized counts of (B) SIRPA and (C) CD47 as a function of patient platelet counts (K/uL) at the time of specimen banking and platelet isolation. Closed circles navy or light blue indicate sample type respectively of healthy donors as controls (CTRL, n=11) and patients with essential thrombocythemia (ET, n=23).