A human SIRPA knock-in xenograft mouse model to study human hematopoietic and cancer stem cells

Xenogeneic transplantation using immunodeficient mouse models is an established method to analyze human hematopoiesis and tumorigenesis.^1-3  Over the decades, extensive progress has been made in establishing mouse lines that are capable of efficient engraftment and growth of human hematopoietic stem cells (HSCs) or cancer stem cells. These xenotransplantation models are necessary to develop translational cancer research.⁴ 

Successful xenotransplantation can be achieved primarily through prevention of human graft rejection. To deplete T and B cells from recipient mice, the scid mutation in Prkdc^5-7  or disruption of the recombination activating gene 1 and 2 (Rag1 and Rag2)^8-10  have been introduced. In addition, to deplete NK cells or their functions, the interleukin-2 receptor common γ chain subunit (Il2rg) or B2m was disrupted.^11-13  It has also been empirically shown that mice with the nonobese diabetic (NOD) or BALB/c genetic background exhibit efficient human-to-mouse xenotransplantation of hematopoiesis.¹⁴  Accordingly, immunodeficient mice on a NOD or BALB/c background, including NOD-scid Il2rg^null,^12,13  NOD.Rag1^nullIl2rg^null (NOD-RG),¹⁵  and BALB/c.Rag1/2^nullIl2rg^null (BALB-RG)^14-16  strains, have been gold standard mouse models for xenogeneic transplantation studies.

We have reported that the strain-specific genetic determinant of human hematopoiesis engraftment was a polymorphism of the signal-regulatory protein α (Sirpa) gene.^17-19  SIRPA is a transmembrane protein that is expressed on macrophages and binds its ligand, CD47, which is ubiquitously expressed.²⁰  The binding of CD47 by SIRPA mediates inhibitory signals for macrophages to prevent autophagocytic activities: the so-called “don’t eat me” signaling.^20-22  This binding is species specific; in most mouse strains, mouse SIRPA cannot recognize human CD47, resulting in activation of mouse macrophages and rejection of grafted human tissues.^17,18 

In normal hematopoiesis, CD47-SIRPA signaling is necessary to maintain homeostasis, because homozygous Sirpa-knockout (Sirpa^−/−) mice developed anemia and thrombocytopenia, presumably as a result of enhanced blood cell clearance from the circulation²³  and/or malformation of the hematopoietic niche.²⁴  Interestingly, the mouse SIRPA immunoglobulin variable (IgV) domain is highly polymorphic, and the unique SIRPA IgV domains of NOD and BALB/c mice confer an enhanced affinity for human CD47.¹⁷  The binding affinity of recipient SIRPA to human CD47 should be 1 of the critical factors that determines the efficiency of human-to-mouse xenotransplantation. We have shown that NOD SIRPA has the strongest affinity to human CD47, whereas BALB/c SIRPA has an intermediate affinity, and C57BL/6 (B6) SIRPA has no affinity; the affinity levels appeared to correlate with reconstitution levels of human hematopoiesis after xenotransplantation.^17,25  This was confirmed by our study showing that the replacement of B6 Sirpa with NOD Sirpa in the C57BL/6.Rag2^nullIl2rg^null (BRG) mouse greatly improved the efficiency of human hematopoiesis engraftment.¹⁹  The BRG-Sirpa^NOD/NOD (BRGS^NOD) mice efficiently supported human hematopoiesis that was equal to or even better than NOD-RG mice,¹⁹  whose efficiency had been reported to be nearly equivalent to NOD.Cg-Prkdc^scidIl2rg^tm1Wjl/SzJ (NSG) mice.^14,15 

Several studies have also shown that the enforced expression of human SIRPA can enhance human hematopoiesis in immunodeficient mice. The introduction of a bacterial artificial chromosome (BAC) minigene that encompasses the entire human SIRPA gene into 129×BALB/c.Rag2^nullIl2rg^null (129.BALB-RG) mice significantly improved human hematopoiesis to a level comparable to that of NSG mice.²⁶  In another study, the extracellular domain of human SIRPA (exons 2-4) was knocked into the 129.BALB-RG strain to form chimeric human/mouse Sirpa, preserving the intracellular portion of mouse Sirpa.²⁷  129.BALB-RG mice with a heterozygous human SIRPA knock-in allele could reconstitute human hematopoiesis comparable to NSG mice; interestingly, homozygous mice exhibited a significantly lower level of human hematopoiesis. This result was interpreted that, like the phenotype of Sirpa^−/− mice,²⁴  SIRPA signal–dependent osteogenic differentiation for forming hematopoietic niches in the bone might be impaired in homozygous SIRPA knock-in mice, presumably because the chimeric SIRPA did not recognize mouse CD47.

In the present study, we generated B6 mice with a knock-in allele for full-length human SIRPA encompassing all 8 exons (B6-Sirpa^hu/hu). In vitro measurement of affinity for human CD47 demonstrated that B6-Sirpa^hu/hu macrophages had the greatest affinity, and heterozygous B6-Sirpa^hu/W and B6-Sirpa^NOD/NOD macrophages exhibited intermediate affinity. Strikingly, B6-Sirpa^hu/hu macrophages showed a moderate level of affinity for mouse CD47. We then established the BRG mouse line harboring Sirpa^hu/hu (BRGS^human). BRGS^human mice had a long lifespan and did not develop anemia or thrombocytopenia. Human hematopoietic multilineage reconstitution after transplantation of human CB was significantly improved in BRGS^human mice, which was further emphasized in the peripheral blood and spleen. In addition, BRGS^human mice showed better engraftment potential for human acute myeloid leukemia (AML) cells and human colorectal cancer (CRC) cells compared with BRGS^NOD mice. Thus, the BRGS^human mouse should be a useful xenograft model to analyze normal and malignant hematopoiesis, as well as tumorigenesis, in humans.

C57BL/6J, B6.Cg-Rag2^tm1.1Cgn/J, B6.129S4-Il2rg^tm1Wjl/J, and NSG mice were purchased from The Jackson Laboratory. We established a new human SIRPA knock-in mouse line using the targeting vector, including full-length human SIRPA coding sequence with the BGHpA sequence (supplemental Figure 1A, available on the Blood Web site), which resulted in a B6 mouse line with human SIRPA knocked into the mouse Sirpa loci. We intercrossed these strains by conventional breeding to obtain BRGS^human mice. BRGS^NOD mice were developed in our laboratory (at Department of Medicine and Biosystemic Science, Kyushu University Graduate School of Medical Sciences) as previously described.¹⁹ Rag2, Il2rg, and Sirpa genes were genotyped by direct sequencing after PCR amplification to identify the genotype of BRGS^human mice. Primer sequences are described in supplemental Table 1. All mice were bred and maintained under specific pathogen–free conditions at the Kyushu University Animal Facility. All experiments were conducted following the guidelines of the institutional animal committee of Kyushu University.

BRG and BRGS^NOD mice exhibit more tolerance for irradiation than do NOG/NSG mice, which are highly radiosensitive, presumably because of the scid mutation. To determine the irradiation dose for BRGS^human mice, 6- to 8-week-old BRGS^human mice were irradiated with 400 to 640 cGy and monitored for 8 weeks (supplemental Figure 1B). Early deaths were observed in the mouse group irradiated with >620 cGy, whereas mice irradiated with 450 to 600 cGy were still alive at the end of the 8 weeks. Thus, as preconditioning for xenotransplantation experiments, BRGS^human mice were irradiated with 600 cGy, and BRGS^NOD, BRG, and NSG mice were irradiated with 550 cGy, 670 cGy, and 220 cGy, respectively, based on previous studies.^11,13,19 

Lineage (Lin)-depleted CB cells were obtained using magnetic beads (Lineage Cell Depletion Kit; Miltenyi Biotec), and Lin⁻CD34⁺CD38⁻ CB cells were sorted using a FACSAria II or III (BD Biosciences). It has been shown that human cell chimerism after xenotransplantation is higher in female mice than in male mice.^19,28  As shown in supplemental Figure 2, this was also the case for BRGS^human mice, in terms of human cell chimerism after transplantation, for the bone marrow and the periphery. It has also been shown that, with regard to xenotransplantation using adult mice, intrafemoral injection was more efficient than IV injection.²⁹  As shown in supplemental Figure 3, intrafemoral injection was significantly superior to IV injection in BRGS^NOD mice. Based on these data, transplantation was restricted to female recipients using intrafemoral injection to exclude those variables.

To reconstitute human hematopoiesis in mice, Lin⁻CD34⁺CD38⁻ CB cells (5 × 10³ per mouse) were injected into the right femur of 6- to 8-week-old irradiated mice, as previously described.^19,28  Mice were euthanized, and samples were analyzed 10 to 12 weeks or 20 to 24 weeks after transplantation. Human CB cells were collected during normal full-term deliveries, after obtaining informed consent and in accordance with the Declaration of Helsinki, and were provided by the Japanese Red Cross Kyushu Cord Blood Bank.

To assess the reconstitution capability of human cancer cells, xenotransplantation of AML and CRC cells was performed. A total of 1 × 10⁶ CD33⁺CD45^lo or CD34⁺ leukemia cells purified from AML patients was transplanted into irradiated mice. The DLD-1 cell line was purchased from American Type Culture Collection and injected subcutaneously with Matrigel (BD Biosciences), as previously described.³⁰  For in vivo limiting dilution analysis, 40, 200, or 1000 DLD-1 cells were injected into BRGS^human and BRGS^NOD mice (n = 6 each). To establish a patient-derived CRC xenograft model, CRC cells were obtained as surgical specimens and injected subcutaneously with Matrigel into BRGS^NOD mice. Engrafted tumor cells were collected, and 1 × 10⁵ engrafted CRC cells were injected into BRGS^human and BRGS^NOD mice (n = 6 each). All animal experiments were performed in accordance with the Institutional Animal Welfare Guidelines of Kyushu University. Clinical samples were collected after written informed consent was obtained from patients at Kyushu University Hospital. The study protocol was approved by the Institutional Review Board of Kyushu University Hospital.

For the analyses of mouse T, B, and natural killer (NK) cells, mouse peripheral blood cells were stained with fluorescein isothiocyanate (FITC)-conjugated anti-CD19, phycoerythrin (PE)-conjugated anti-CD3, allophycocyanin (APC)-conjugated anti-NK1.1, and Pacific Blue–conjugated anti–Gr-1 antibodies. To confirm expression of SIRPA at the protein level, mouse macrophages were stained with FITC-conjugated anti-mouse CD11b, PE-conjugated anti-mouse SIRPA, and APC-conjugated anti-human SIRPA antibodies. Sorting of the Lin⁻CD34⁺CD38⁻ subfraction in human CB samples was accomplished by staining with FITC-conjugated anti-CD34 antibodies, BV421-conjugated anti-CD38 antibodies, and a PerCP-Cy5.5–conjugated lineage cocktail including anti-human CD3, CD4, CD8, CD10, CD11b, CD14, CD19, CD20, CD56, and CD235ab antibodies. For the analysis and sorting of human cells after xenogeneic transplantation, we used a combination of FITC-conjugated anti-human CD33 or CD34, PE-conjugated anti-human NKp46 or CD10, PE-Cy7–conjugated anti-human CD19 or CD20, APC-conjugated anti-human CD3 or IgM, APC-Cy7–conjugated anti-human CD45, Pacific Blue–conjugated anti-mouse CD45 or anti-human CD19, and PerCP-Cy5.5–conjugated anti-mouse TER-119 antibodies. Fluorochrome-conjugated antibodies used for flow cytometric analysis are listed in supplemental Table 2. The cells were analyzed and sorted with a FACSAria II or III cell sorter.

Mouse macrophages were obtained by peritoneal lavage, as described previously.^17,21  The binding of SIRPA and CD47 was assessed by flow cytometry and a peroxidase activity assay, as detailed previously.¹⁷  For the analysis using flow cytometry, mouse macrophages were stained with FITC-conjugated CD11b, PE-conjugated anti-mouse SIRPA, and biotinylated human CD47-Fc plus APC-conjugated streptavidin. The cells were analyzed with a FACSAria III cell sorter. For the peroxidase activity assay, mouse macrophages were incubated at 37°C in a 96-well plate in the presence of increasing concentrations of purified human or mouse CD47-Fc fusion protein and then incubated at 4°C with horseradish peroxidase–conjugated goat polyclonal antibody specific for the Fcγ fragment of human IgG (Jackson ImmunoResearch), followed by a peroxidase assay. Human or mouse CD47-Fc fusion protein binding was determined by absorbance at 490 nm on a microplate reader. Nonlinear regression analysis was performed to calculate dissociation constant (Kd) and maximal binding capacity (Bmax) using the KaleidaGraph analysis program.

Phagocytic activity of mouse macrophages against the human CD34⁺CD38⁻ population was evaluated in vitro, as described.¹⁷  In brief, mouse peritoneal-derived macrophages were incubated with mouse interferon-γ (100 ng/mL; R&D Systems) and lipopolysaccharide (0.3 μg/μL) and then opsonized human CB-derived HSCs were added. After 2 hours of incubation, the phagocytic index was calculated using the following formula: phagocytic index = number of ingested cells/(number of macrophages/100). At least 200 macrophages were counted per well.

The Student t test was used for single comparisons, and 1-way analysis of variance, followed by Tukey’s HSD (honestly significant difference) test was used for multiple comparisons. All statistical analyses were performed using JMP software (version 13.0; SAS Institute), and P values < .05 were considered statistically significant. The Mantel-Cox log-rank test was used to obtain survival curves.

The function of mouse and human SIRPA was tested in peritoneal macrophages from B6 mice with wild-type mouse Sirpa (Sirpa^W/W), B6 mice with homozygous NOD-type mouse Sirpa (Sirpa^NOD/NOD), and B6 mice with heterozygous or homozygous knock-in human SIRPA (Sirpa^hu/W or Sirpa^hu/hu, respectively). As shown in Figure 1A, macrophages from Sirpa^W/W mice expressed only mouse SIRPA, whereas those from Sirpa^hu/hu mice expressed human SIRPA but not mouse SIRPA, demonstrating that mouse and human SIRPA were successfully replaced in the Sirpa^hu/hu mouse.

Figure 1B shows dose-response curves for the interaction between SIRPA on macrophages and human CD47-Fc protein, as determined by a SIRPA-CD47 binding assay. Sirpa^hu/hu macrophages showed the strongest binding with human CD47-Fc protein (Bmax, 3.02), whereas Sirpa^hu/W and Sirpa^NOD/NOD macrophages had an intermediate level of binding (Bmax, 1.69 and 1.80, respectively), and Sirpa^W/W macrophages only showed nonspecific binding. The amount of CD47-Fc protein capable of binding to SIRPA expressed on macrophages from each strain was measured by flow cytometry (Figure 1C). Differences in the saturated amount of human CD47 on the surface of each type of macrophage corresponded to their affinity in the binding assay (Figure 1B), confirming that the affinity of Sirpa^hu/hu macrophages for human CD47 is superior to that of Sirpa^NOD/NOD macrophages.

To test human CD47-SIRPA signaling, we measured the phagocytic activity of macrophages purified from each mouse strain toward an opsonized human Lin⁻CD34⁺CD38⁻ HSC population (Figure 1D). Sirpa^W/W macrophages had strong phagocytic activity against the human HSC population, whereas Sirpa^NOD/NOD macrophages showed significantly reduced phagocytic activity. Phagocytosis by Sirpa^hu/hu macrophages was further significantly reduced compared with that by Sirpa^NOD/NOD macrophages, correlating inversely with their affinity in binding assays (Figure 1B).

Although Sirpa^−/− mice were reported to develop anemia or thrombocytopenia,^31,32 Sirpa^hu/hu mice did not appear to have these cytopenias (data not shown). Therefore, we tested the binding affinity for mouse CD47-Fc protein in macrophages from Sirpa^W/W, Sirpa^NOD/NOD, or Sirpa^hu/hu mice. As shown in Figure 1E, B6-Sirpa^NOD/NOD macrophages had a strong binding affinity to mouse CD47 (Bmax, 0.80 ± 0.09) that was comparable to normal B6-Sirpa^W/W mice (Bmax, 0.70 ± 0.05). Interestingly, B6-Sirpa^hu/hu macrophages bound to mouse CD47 at a moderate level (Bmax, 0.29 ± 0.02), suggesting the possibility that human SIRPA derived from a full-length human SIRPA knock-in allele can recognize mouse CD47 to prevent phagocytosis against autologous mouse cells.

The BRGS^human mouse was established by crossing the BRG mouse with the B6-Sirpa^hu/hu mouse. BRGS^human mice were born healthy and exhibited good fertility. As shown in Figure 2A, they had normal levels of hemoglobin and platelets, unlike Sirpa^−/− mice,²³  and had reduced numbers of leukocytes as a result of T, B, and NK lymphocyte depletion (Figure 2B). These phenotypes were similar to BRGS^NOD mice.¹⁹  Like NSG or BRGS^NOD mice, BRGS^human mice had a median lifespan of 71 weeks without developing lymphoma, which is the major cause of death in NOD-scid mice or NOD-Rag1^null–based strains (Figure 2C).^7,9 

Six- to 8-week-old BRG, BRGS^NOD, and BRGS^human mice were irradiated and transplanted with 5 × 10³ CD34⁺CD38⁻ human CB cells. At 10 to 12 weeks after transplantation, human CD45⁺ cells were not detectable in BRG mice (Figure 3A, left panel). BRGS^NOD and BRGS^human mice showed successful human hematopoietic reconstitution in the bone marrow, and the average frequency of human CD45⁺ cells in BRGS^human mice was significantly higher than that of BRGS^NOD mice (80.9% vs 68.9%, respectively). We then analyzed the long-term reconstitution at 20 to 24 weeks after transplantation, because maintenance of human hematopoiesis for >20 weeks is dependent upon self-renewal of human HSCs in the mouse bone marrow niche.³³  As shown in Figure 3A (right panel), BRGS^human mice maintained a high level of human cell chimerism (mean, 55.9%) in the bone marrow, exceeding the level in BRGS^NOD mice (mean, 42.9%). Then, 10⁶ human CD45⁺ cells were purified from primary BRGS^NOD or BRGS^human recipients at 10 to 12 weeks after xenotransplantation, and these cells were retransplanted into irradiated adult BRGS^NOD mice. Eight weeks after the secondary transplantation, re-engraftment of human hematopoietic cells was observed at equivalent levels in the 2 groups (Figure 3B). These data suggest that the BRGS^human and BRGS^NOD microenvironments are able to support self-renewal of human HSCs.

Next, we performed a limiting dilution assay to quantitate the frequency of repopulating cells by injecting graded doses of CD34⁺CD38⁻ CB cells (10¹, 10², or 10³ cells per injection) (Figure 3C). The frequency of detectable repopulating cells in the bone marrow was significantly higher in BRGS^human mice (1 per 32 injected cells) vs BRGS^NOD mice (1 per 141 injected cells).

Figure 4A shows the lineage analysis of human CD45⁺ cells reconstituted in the bone marrow after injection of 5 × 10³ CD34⁺CD38⁻ CB cells. The percentage of CD33⁺ myeloid cells in BRGS^human mice was 22% and 26% at 10 to 12 weeks and 20 to 24 weeks after transplantation, respectively, whereas the percentage was only 9% and 13%, respectively, in BRGS^NOD mice, demonstrating that myeloid reconstitution was significantly better in BRGS^human mice (Figure 4B, top panels). As a result, the percentages of B cells were lower in BRGS^human mice (Figure 4B, middle panels). The majority of reconstituted B cells were CD10⁺CD19⁺CD20⁻ immature B cells (supplemental Figure 4). T-cell reconstitution in the bone marrow appeared to be more efficient in BRGS^human mice (6%) than in BRGS^NOD mice (1%) at 20 to 24 weeks, although the difference was not statistically significant.

Reconstitution of human hematopoietic cells in the periphery is generally low in xenotransplantation models.^34,35 Figure 5 shows the reconstitution in the blood and the spleen in BRGS^NOD and BRGS^human mice. Compared with BRGS^NOD mice, peripheral reconstitution of human CD45⁺ cells was dramatically improved in BRGS^human mice at 10 to 12 weeks after transplantation, and it was further improved at 20 to 24 weeks, reaching up to 22% and 42% in the blood and the spleen, respectively. Of note, the percentages of CD33⁺ myeloid cells within CD45⁺ cells also progressively increased in the periphery up to 32% and 17% in the blood and the spleen, respectively, at 20 to 24 weeks after transplantation.

We wanted to test whether the BRGS^human mouse showed efficient engraftment of cancer cells. We transplanted 10⁶ CD33⁺CD45RA^lo AML cells (patients 1 and 2) or 10⁶ CD34⁺ AML cells (patient 3) purified from the bone marrow into irradiated BRGS^NOD and BRGS^human mice and serially evaluated chimerism in the blood after xenotransplantation. Clinical characteristics of patients are shown in supplemental Table 3. As shown in Figure 6, BRGS^human recipient mice showed accelerated leukemia reconstitution, and human CD45⁺CD33⁺ leukemia cell chimerisms in the blood were significantly higher than those in BRGS^NOD recipients 10 to 16 weeks after transplantation in all 3 experiments. The chimerism of AML cells in the bone marrow was also higher in BRGS^human recipients at the final analysis. In the analysis of patient 3, we also transplanted AML cells into NSG mice in parallel with BRGS^NOD and BRGS^human mice. The AML cell chimerism appeared to be higher in BRGS^human mice than in NSG mice in the blood and the bone marrow, although the difference was not statistically significant.

The DLD-1 cell line was used to test xenotransplantation efficiency for human CRC. We transplanted 10³ DLD-1 cells subcutaneously with Matrigel, as previously described.³⁰  As shown in Figure 7A, DLD-1 cells successfully engrafted, but the tumor was always bigger in BRGS^human recipient mice vs BRGS^NOD recipient mice. A limiting dilution analysis for tumor engraftment showed that the frequency of detectable tumor-initiating DLD-1 cells was 2.4-fold higher in BRGS^human recipient mice (Figure 7B).

We then tested xenotransplantation efficiency for primary human CRC cells. Primary tumor was obtained from patients with CRC, surgical specimens were injected into BRGS^NOD mice subcutaneously, and engrafted tumor cells were collected. A total of 10⁵ engrafted CRC cells was subcutaneously injected into BRGS^human or BRGS^NOD mice, and secondary tumorigenesis of colon cancer cells was monitored for 10 weeks. Tumor formation was facilitated to a more significant degree in BRGS^human mice compared with BRGS^NOD mice (Figure 7C-D). These data strongly suggest that the introduction of human SIRPA into immunodeficient mice can provide a microenvironment suitable for engraftment and growth of human malignant hematopoiesis, as well as for CRC development.

In establishing xenotransplantation models, deletion of lymphocytes has been achieved by introduction of the scid mutation in the Prkdc gene or disrupting key lymphoid genes, such as Rag and Il2rg. Another important factor for xenotransplantation efficiency is “macrophage tolerance” driven by the CD47-SIRPA axis, and the NOD-type SIRPA has a polymorphism that is best suited for xenotransplantation models, because it can recognize human CD47 with the highest affinity among a variety of mouse strains.¹⁷  In this study, by directly comparing the xenotransplantation efficiency of BRGS^human mice with BRGS^NOD mice, whose efficiency was at least equal to NOD-RG mice, we showed that homozygous replacement of NOD-type mouse Sirpa with human SIRPA significantly improved the engraftment efficiency of normal human hematopoiesis, as well as of malignant blood and solid cancers.

The high efficiency of xenotransplantation in BRGS^human mice should be ascribed to the fact that human SIRPA has a much higher affinity for human CD47 than does NOD-type mouse SIRPA (Figure 1B). Interestingly, human SIRPA can also recognize mouse CD47 (Figure 1E). Sirpa^hu/hu mice and BRGS^human mice did not develop anemia or thrombocytopenia, probably because the engagement of mouse CD47 with human SIRPA can induce signaling at a level sufficient to keep homeostasis within the bone marrow microenvironment in BRGS^human mice. As a result, the BRGS^human mouse line is stable and has a lifespan comparable to BRGS^NOD mice (Figure 2C).

Moreover, the capabilities to support human hematopoiesis in BRGS^human mice might be better than those in previous mouse lines expressing human SIRPA. The 129.BALB-RG mouse with a transgenic BAC human SIRPA showed engraftment levels similar to the NSG mouse that has the NOD-type mouse SIRPA in the bone marrow, the blood, or the spleen.²⁶  It might be difficult to guarantee physiologically faithful expression of the BAC transgenic human SIRPA; transgenic expression of human SIRPA depends upon human SIRPA regulatory elements that are regulated by mouse intracellular components. Signaling downstream of BAC transgenic SIRPA should also be complicated in this mouse, because its macrophages coexpress endogenous mouse SIRPA.

The 129.BALB-RG strain has also been used to establish human SIRPA knock-in mice, whose knock-in construct was restricted to exons 2 to 4 that code for the extracellular IgV domain of human SIRPA.²⁷  Mouse and human SIRPA have 8 exons: exons 1 to 4 for the extracellular domain, exon 5 for the transmembrane domain, and exons 6 to 8 for the intracellular domain. The intracellular domain is mostly preserved in mice and humans. In this human SIRPA knock-in model that has human/mouse chimeric SIRPA, the xenotransplantation efficiency of heterozygous knock-in mice was almost equal to that of NSG mice.²⁷  It appears to be accurate, because the affinity of Sirpa^NOD/NOD macrophages for human CD47 was similar to that of heterozygous Sirpa^hu/W macrophages in our study (Figure 1B). However, homozygous knock-in mice showed significantly lower xenotransplantation efficiency compared with heterozygous mice, and the investigators speculated that hematopoietic niche formation was impaired by the lack of mouse SIRPA-CD47 signal-dependent bone and stromal cell differentiation, which was observed in SIRPA-deficient mice.²⁴  Because native human SIRPA can recognize mouse CD47 (Figure 1E) and BRGS^human mice that have homozygous full-length human SIRPA showed higher xenotransplantation efficiency than BRGS^NOD mice (Figures 3 and 4), it is possible that the chimeric SIRPA somehow lost its affinity for mouse CD47, which native human SIRPA has.

It is noteworthy that CD33⁺ myeloid cell development, as well as hematopoietic reconstitution in the periphery, was significantly improved in BRGS^human mice. In mouse models based on the NOD or BALB/c background, human hematopoietic reconstitution is B-cell dominant, whereas reconstitution for human myeloid lineage cells is usually limited.^12,36  Improvement of myeloid reconstitution has been reported by introducing knock-in human cytokines, including human thrombopoietin³⁷  and stem cell factor,³⁸  into NSG mice or introducing a combination of cytokines, such as macrophage colony-stimulating factor, interleukin-3, granulocyte-macrophage colony-stimulating factor, and thrombopoietin, into human BAC-SIRPA transgenic or heterozygous chimeric SIRPA knock-in mice^39-41 ; this suggests that a deficiency in cross-reactivity in mouse and human cytokines causes this skewed differentiation. Significant improvement in myeloid differentiation in BRGS^human mice without humanization of cytokines further suggests that physiologically appropriate CD47-SIRPA signaling is critical for human myeloid engraftment, presumably through normalization of macrophage functions in the bone marrow microenvironment^42,43  or through inhibition of macrophage engulfment or activation in the periphery.^21,44  Because the human cytokine(s) knocked-in and the homozygous full-length SIRPA^human knocked-in should contribute to myeloid engraftment through independent mechanisms, it might be promising to further improve engraftment by humanization of cytokines in the BRGS^human mouse model. It is also tempting to introduce c-Kit mutations^28,45  into BRGS^human mice, because BRGS^NOD mice with a Kit^Wv mutation exhibited highly efficient engraftment of human erythropoiesis and thrombopoiesis.²⁸ 

Human leukemia and a variety of solid cancer cells express CD47 at a level higher than normal tissue cells to reinforce engagement of CD47 and SIRPA, as well as to escape from innate immune surveillance by macrophages.^46-48  CD47-SIRPA signaling is also required for cancer cells to survive in the xenotransplantation setting, because human cancer cells transplanted into NOD-scid mice were eliminated by administration of anti-CD47–blocking antibodies.⁴⁸  Therefore, significant improvement of human cancer cell engraftment in our BRGS^human model (Figure 7) might be due to enhanced binding of tumor CD47 and human SIRPA that evokes strong signaling to inhibit xenogeneic tumor immunity.

In summary, we generated BRGS^human mice with a knock-in allele of the entire human SIRPA. Engraftment of human CB in BRGS^human mice was more efficient than in BRGS^NOD mice in the bone marrow and in the periphery, and myeloid reconstitution was also significantly improved. The engraftment of AML cells and CRC cells was also more efficient in BRGS^human mice than in BRGS^NOD mice. These data strongly suggest that the BRGS^human mouse should be very useful to establish future xenogeneic transplantation assays that are more realistic for the analysis of hematopoiesis, leukemogenesis, and tumorigenesis in humans.

The authors thank Yuichiro Semba, Masayasu Hayashi, Jun Odawara, and Yasuyuki Ohkawa for purification of the CD47-Fc protein and the Japanese Red Cross Kyushu Cord Blood Bank for providing CB samples.

This work was supported in part by a Grant-in-Aid for Scientific Research (B) from Japan Society for the Promotion of Science (18H02840, K.T.), a Grant-in-Aid for Scientific Research (S) from Japan Society for the Promotion of Science (16H06391, K.A.), and a grant from the Shin-nihon Foundation of Advanced Medical Treatment Research (Y.K.)

Contribution: F.J., T.Y., and K.T. coordinated the project, designed and performed the experiments, analyzed the data, and wrote the manuscript; A.Y., T.N., M.N., C.I., T.O., K.M., and Y.K. performed experiments; and K.K., T. Maeda, T. Miyamoto, E.B., and K.A. designed the experiments, reviewed the data, and edited the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Koichi Akashi, Kyushu University Graduate School of Medical Sciences, 3-1-1 Maidashi, Higashi-ku, Fukuoka 812-8582, Japan; e-mail: akashi@med.kyushu-u.ac.jp.