Introduction of Macrophage Tolerance By Human SIRPA Knock-in Enhances Peripheral Reconstitution of Human Hematopoiesis in Xenotransplantation Model

NOD background is one of the critical determinants for the efficient human cell engraftment in xenotransplantation models. We have reported that the NOD-specific polymorphism of Sirpa allows the mouse SIRPA to bind human CD47, preventing activation of recipient macrophages to engulf human hematopoietic cells (Nat Immunol. 2007). We then developed a C57BL/6.Rag2^nullIl2rg^null (BRG) mouse line harboring the NOD-type Sirpa, named BRGS, in which the xenograft efficiency of human hematopoiesis was equal to that in NOD.Rag1^nullIl2rg^null strain (Blood 2013). However, there are several unsolved problems in the current xenograft models including BRGS: The reconstituted human hematopoiesis is B-lymphoid dominant; the engraftment levels of human erythrocytes are quite low; and the distribution of human cells to peripheral tissues is not enough. We have reported that the binding affinity of recipient SIRPA with human CD47 is a decisive factor for the efficiency of xenotransplantation (Exp. Hematol.2014). Thus, we consider that introduction of human SIRPA confers much more efficient engraftment to recipients compared to that of NOD SIRPA.

Here, we newly generated human SIRPA knock-in mice on C57BL/6 background and intercrossed these mice with BRG mice to obtain the BRGhS (C57BL/6.Rag2^nullIl2rg^nullHuman SIRPA) mice. The BRGhS mice were born healthy and displayed good fertility. In contrast to Sirpa^nullmice, the BRGhS mouse did not develop anemia and thrombocytopenia, probably because human SIRPA partially bound to mouse CD47.

We found that the BRGhS macrophages have a better affinity for human CD47 than the BRGS macrophages by protein binding assay (Kd=7.2nM vs. 11.4nM, p=0.04). Human CD47 binding to the BRGhS macrophages activated the SHP-1 phosphatase, an anti-phagocytic signal. Consistent with this, when macrophages of each strain were cultured with human HSCs, the number of HSCs engulfed by the BRGhS macrophages was much less than those by the BRGS macrophages. To quantify the reconstitution activity of human hematopoiesis in vivo, we transplanted 5 × 10³ CD34⁺CD38^- human cord blood cells intrafemorally into irradiated BRGhS or BRGS mice. At 10-12 weeks after transplantation, both strains showed successful reconstitution, and the frequency of human CD45⁺ cells in the bone marrow of BRGhS was better than that of BRGS (80.2% vs. 72.3% p=0.09). However, the reconstitution of human erythrocytes/megakaryocytes lineages is limited in both strains. The frequency of human CD45⁺ cells was maintained at least until 24 weeks after transplantation. Surprisingly, the frequency of human CD45⁺cells in the spleen and peripheral blood (PB) of BRGhS was significantly higher than that of BRGS (Spleen: 37.8% vs. 11.2% p=0.01, PB: 11.4% vs. 5.2% p<0.01). To examine the phagocytic activity of macrophages against circulating human cells in vivo, human erythrocytes were injected into both strains and their clearance was monitored. Transfused erythrocytes were rapidly eliminated from the circulation of the BRGS recipients, but they remained detectable in the BRGhS recipients.

The drastic improvement of the reconstitution in peripheral tissues and the much less elimination of infused human erythrocytes suggest that the introduction of human SIRPA strongly inhibits phagocytosis by macrophages, especially in peripheral tissues. Recently, we reported that robust erythroid reconstitution in bone marrow was obtained by introduction of Kit mutation into the BRGS strain (Stem Cell Reports, in press). Thus, combination Kit mutation with human SIRPA knock-in should be a powerful tool for establishing a more faithful humanized model with circulating mature human blood cells.