Transgenic Expression of Human CD47 Markedly Increases Engraftment in a Murine Model of Pig-to-Human Hematopoietic Cell Transplantation

Abstract

Mixed chimerism approaches for induction of tolerance of solid organ transplants have been applied successfully in animal models and in the clinic. However, in xenogeneic models (pig-to-primate), host macrophages participate in the rapid clearance of porcine hematopoietic progenitor cells, hindering the ability to achieve mixed chimerism. CD47 is a cell-surface molecule that interacts in a species-specific manner with SIRPα receptors on macrophages to inhibit phagocytosis and expression of human CD47 on porcine cells has been shown to inhibit phagocytosis by primate macrophages. We report here the generation of human CD47 (hCD47) transgenic GalT-KO miniature swine that express hCD47 in all blood cell lineages. The effect of hCD47 expression on xenogeneic hematopoietic engraftment was tested in an in vivo mouse model of human hematopoietic cell engraftment. High-level porcine chimerism was observed in the bone marrow of hCD47 progenitor cell recipients and smaller but readily measurable chimerism levels were observed in the peripheral blood of these recipients. In contrast, transplantation of WT progenitor cells resulted in little or no bone marrow engraftment and no detectable peripheral chimerism. These results demonstrate a substantial protective effect of hCD47 expression on engraftment and persistence of porcine cells in this model, presumably by modulation of macrophage phagocytosis.

Introduction

Xenotransplantation of pig organs offers the best near-term hope for satisfying the limitation imposed by the shortage of allogeneic solid organs. In the last two decades, considerable progress has been made in understanding the immunobiology of pig-to-nonhuman primate transplantation, thus allowing better understanding of other barriers such as molecular incompatibility, that may impede successful xenotransplant outcomes (1). Unfortunately, the immune response is considerably stronger to xenografts than it is to allografts, at the levels of both antibody and T cell immune responses (2–5). For this reason, it seems likely that the success of clinical xenografts will depend, at least in part, on finding ways of safely inducing tolerance across xenogeneic barriers rather than relying entirely on non-specific immunosuppressive agents.

Hematopoietic chimerism has been exploited as a modality for induction of tolerance of solid organ transplants in murine (6;7), porcine (8) and primate (9–12) allogeneic animal models, rodent (13) and primate (14) xenogeneic animal models, and more recently in human clinical trials (15–17). Unfortunately, previous studies have shown that porcine hematopoietic cells transplanted into pre-conditioned non-human primates are rapidly cleared from the primate circulation, even in the absence of the α-1,3-galactosyl epitopes responsible for hyperacute rejection through preformed natural antibodies (18). In vitro studies have demonstrated that human macrophages rapidly phagocytose pig erythrocytes in a gal epitope independent manner (19), suggesting that this rapid clearance may be mediated by phagocytic cells. Treatment of primates with medronate liposomes to deplete macrophages has been shown to greatly increase the level and duration of xenogeneic chimerism (20), but this treatment is toxic and incompatible with tolerance induction mechanisms relying on costimulation blockade (21).

CD47 (Integrin-associated protein, IAP) is a ubiquitously expressed 50-kDa cell surface glycoprotein that serves as a ligand for thrombospondin-1, Signal Regulatory Proteins (SIRPs) and several integrins (22). The role of CD47 inhibition of phagocytosis through Signal Regulatory Protein-α (SIRPα; CD172a, SHPS-1) expressed on macrophages has been described (23). Red blood cells lacking CD47 were found to be efficiently and rapidly cleared following transfusion in normal mice (24). Loss and alteration of CD47 structure has been implicated in the normal clearance of aged red blood cells (25). In contrast, increased expression of CD47 has been noted in both normal hematopoietic cells exiting the bone marrow and in myeloid leukemias (26). Species incompatibilities between CD47 and SIRPα have been noted (27) and may be attributable to substantial CD47 sequence divergence (28). Wang et al. (29), using immunocompetent mice, found that expression of murine CD47 on porcine cells inhibited macrophage engulfment in vitro and delayed clearance of porcine cells in vivo. They also showed that expression of human CD47 on porcine cells inhibited phagocytosis by human macrophages (30) and improved survival of these cells in mice carrying an allele of SIRPα that productively binds hCD47 (31). Together these studies suggested that decreased clearance of porcine cells in primates may be achievable by transgenic expression of primate CD47. In this study, we report the generation of human CD47 (hCD47) transgenic miniature swine on an α-1,3-galactosyltransferase (GGTA1, GalT-KO) null background (32). Furthermore, as proof of principle for pig-to-primate transplantation studies, we demonstrate that such expression on porcine hematopoietic progenitor cells results in increased engraftment and peripheral chimerism in a murine model of human hematopoietic cell engraftment.

Materials and Methods

Animals

Cloning of hCD47 transgenic animals is described below. Fibroblasts for hCD47 transfection were obtained from fetus 19120-4, obtained by caesarian section at 42 days of gestation from the mating of WT and GGTA1 null partially inbred MGH miniature swine. CHEF-NSG mice were obtained by mating NOD/scid mice transgenic for porcine cytokines IL-3, GM-CSF and SCF (33) to non-transgenic NSG mice (Jackson Labs) and genotyping for cytokine Tg/IL2Rγ null offspring. Mice were housed in a specific pathogen-free microisolator environment and were transplanted at 12–13 weeks of age. All animal work was conducted in accordance with NIH and USDA guidelines and with approval from the MGH Institutional Animal Care and Use Committee.

Vector Construction and Fibroblast Transfection/Selection

Knock-in targeting vector EF1 was based on the previously described GGTA1 knockout vector pGalGT (34) with the following modifications. An EMCV IRES was substituted for the previous Bip IRES as a translational start for G418R coding sequences, followed by the bovine growth hormone pA region. A hCD47 expression cassette, consisting of the human EF1α promoter and coding sequences for splice form 1 of the human CD47 gene (35), was inserted downstream of the BGH pA. GGTA1 heterozygous fibroblasts from fetus 19120-4 were transfected as described (34) and selected in bulk for 15 days with 400 ug/ml G418 beginning 2 days after transfection. This concentration allowed selection of cells containing the knock-in modification in a population of cells resistant to G418 at 50–100 ug/ml due to the presence of the knockout GGTA1 allele. The G418 selected population was depleted of αgal epitope bearing cells by incubation with 4 ug/ml FITC conjugated lectin BS-IB₄ (Sigma), followed by binding to anti-FITC microbeads and passage through an LD magnetic depletion column (Miltenyi Biotech). The selected and depleted population expanded an additional 7 days before freezing for use in cloning.

Pig Cloning

Chromatin transfer (CT) was performed using in vitro matured oocytes as recipient cytoplasts. Porcine fetal fibroblasts were prepared as described by (36). Briefly, the cells were trypsinized and washed in Ca/Mg-free Hank’s Balanced Salt Solution (HBSS) and permeabilized by incubation of 50,000–100,000 cells in 31.25 units Streptolysin O (SLO-Sigma, St. Louis, MO) in 100 μl for 30 minutes in a 37°C H₂O bath. Permeabilized fibroblasts were washed, pelleted and incubated in 40 μl of mitotic extract prepared from MDBK cells containing an ATP-generating system (1 mM ATP, 10 mM creatine phosphate and 25 μg/ml creatine kinase) for 30 min at 38°C. At the end of incubation, the reaction mix was diluted with 500 μl of cell culture media (DMEM with 10% FBS), pelleted and resuspended in NSCU-Hepes. A single donor cell was inserted into the perivitelline space of an enucleated oocyte. Membrane fusion between the donor cell and recipient cytoplast was induced electrically. The reconstructed embryos were chemically activated using 10 μM ionomycin for 5min and 2mM 6-dimethylaminopurine (6-DMAP) for 2–3 h. Embryos were transferred immediately after activation. Recipient animals were crossbred gilts (59–68 kg), synchronized by oral administration of 15 mg per day Matrix^™ (Merck Animal Health) mixed into the feed for 14–17 days using a scheme dependent on the stage of the estrous cycle. Follicular development was induced with a single i.m. injection of P.G.600 (Merck Animal Health) 24 h after the last Matrix^™ feeding. Seven hundred and fifty units of Chorulon® (Merck Animal Health) were administered i.m. 82 h after the P.G.600 injection. Reconstructed embryos were transferred between 22 and 24 h post Chorulon® injection. Transgenic fetuses were recovered following euthanasia 38 days following embryo transfer. All cloning experiments and post-delivery animal care was conducted in accordance with NIH and USDA guidelines and approved by the Institutional Animal Care and Committee of Minitube of America (# A4520-01).

PCR Assays

Donor cells and fetal fibroblasts from cloned fetuses were analyzed using allele specific RT-PCR assays, which also served to confirm targeting of the vector sequences to the GGTA1 locus (see Figure 1). A common forward primer (X7-76) located in GGTA1 exon 7, upstream of the vector end, was used in separate reactions with reverse primers specific for the WT (X9-365; 290 bp expected fragment length), knockout (Bip393; 318 bp) or knock-in (EMCV-QR; 349 bp) GGTA1 loci.

Figure 1. Vector design and assays for knock-in expression of human CD47 from the GGTA1 Locus.

(A) Predicted structure of the hCD47 knock-in locus. The targeting vector contains approximately 21 kb of GGTA1 homology, beginning just downstream of exon 7 and continuing approximately 6 kb downstream of the end of exon 9. The catalytic domain of galtransferase in exon 9 is disrupted by a 4.5 kb selection/expression cassette consisting of a G418 resistance gene translated from an EMCV internal ribosome entry site and a human CD47 CDS expressed from the human elongation factor 1α promoter. Expected replacement-type targeting is confirmed in genomic DNA assays with primers CDF789 (within the hCD47 CDS region) and GTR924 (downstream of the 3′ vector end). Primer X7-76, located upstream of the 5′ end of the vector) is used in conjunction with allele specific RT-PCR primers (see below) to further confirm expected gene-targeted replacement). (B) Predicted transcripts from the hCD47 knock-in (KI), GGTA1 knockout (KO) and WT alleles (exons 1–6 not shown for simplicity). The knock-in allele is predicted to generate two transcripts. One originates at the native GGTA1 promoter and is linked via an EMCV internal ribosome entry site to coding sequences for G418 resistance. The second knock-in transcript contains the hCD47 coding sequences, expressed from the EF1α promoter. Knockout and WT transcripts both originate at the native GGTA1 promoter. Knock-in, knockout and WT transcripts from the GGTA1 promoter can be distinguished by RT-PCR using common forward primer X7-76 and allele specific reverse primers EMCV-QR (predicted fragment length 349 bp) or Bip393 (318 bp). Primer X9-365 (290 bp), located downstream of the selection/expression cassettes of both the knock-in and knockout vectors, serves as a specific primer for the WT locus.

Transfection of fibroblasts heterozygous for a GGTA1 knockout and WT allele followed by G418 selection results in a heterozygous subpopulation of galtransferase null cells expressing human CD47 in which the WT locus has been replaced by KI vector sequences.

Expected replacement targeting of the WT GGTA1 allele with the hCD47 knock-in allele was confirmed in genomic DNA of cloned fetuses by LA Taq amplification (Takara) using a forward primer within the hCD47 CDS (CD789) and a reverse primer (GTR924) located downstream of targeting vector sequences. Predicted lengths of Eco RI digested products (to increase size resolution) are 3.0 and 6.8 kb.

hCD47 expression in tissues from piglet 1524 was assessed using forward (CD47QF) and reverse (CD47QR) primers from exons 5 and 6 of the human CD47 gene. See Table S1 for primer sequences.

FACS Analysis of hCD47 and αgal epitope expression

Cell surface expression of hCD47 was detected using phycoerythrin or AlexaFluor 647 conjugated mouse-anti-human CD47 clone B6H12 (BD Biosciences). αgal epitope expression was detected using FITC conjugated lectin BS-IB₄ (Sigma). Acquisition was performed on a FACSCalibur (Becton-Dickinson, Mountain View, CA, USA) and analyzed using FloJo software (Treestar, Ashland OR).

Hematopoietic progenitor cell transplantation

Peripheral blood stem and progenitor cells from hCD47 transgenic pig 18286 and a WT miniature pig were obtained at 9–11 months of age by leukapheresis following mobilization for 6 days with porcine IL-3 and porcine SCF as previously described (37). Low density cells from the leukapheresis product were isolated by gradient centrifugation over Histopaque (Sigma, St. Louis, MO, USA) diluted to p=1.070. Low density cells were enriched for the ckit⁺ population using biotinylated porcine SCF as previously described (38). The degree of enrichment was assessed by FACS using biotinylated SCF and streptavidin conjugated AlexaFluor 647 and were frozen in aliquots for later transplantation.

CHEF-NSG mice were conditioned with 2.25 Gy total body irradiation and transplanted by tail vein injection within 24 hours with 1 × 10⁶ porcine ckit⁺ isolated from WT and hCD47 transgenic swine as previously described (38). Irradiated (25 Gy) porcine peripheral blood mononuclear cells from a WT miniature pig were added as carriers, to a total of 2 × 10⁷ cells/mouse. Control groups receiving just TBI or carriers alone were included in both experiments and porcine chimerism was undetectable in any of these mice. Porcine clonogenic progenitor cell content in ckit-enriched cells was determined at the time of transplantation by plating in Methocult H4230 medium (Stem Cell Technologies, Vancouver, BC, Canada) supplemented with a pig specific cytokine mixture consisting of 2 ng/mL porcine IL-3, 25 ng/mL pig SCF, 5 ng/mL pig GM-CSF and 2 U/mL human EPO. Colonies (>100 cells) were counted after 12 to 14 days and the frequency of clonogenic (BFU-E, CFU-GM and CFU-mix) colonies determined. Based on the clonogenic frequencies determined at the time of the first transplant trial, injected cell numbers were adjusted for the second trial; the actual number of clonogenic progenitors transplanted in the second trial transplantation was determined at the time of second transplant (Table 1). Peripheral blood and bone marrow were prepared from recipient mice at sacrifice 17–21 weeks following transplantation and analyzed for porcine chimerism essentially as described (38). Cells were stained with PE-conjugated anti-mouse CD45.1 mAb clone A20 (BD Biosciences) and pan porcine specific mAb 1030H-1-19 (39) after red cell depletion with ACK lysis buffer (Lonza). Rat anti-mouse FcγR mAb clone 2.4G2 (40) was used to block nonspecific binding. 1030H-1-19 positive cells not stained by anti-mouse CD45 mAb were considered to be of porcine origin. Acquisition was performed on a FACSCalibur and propidium iodide excluding cells within the lymphocyte gate analyzed using FloJo software.

Table 1. Hematopoietic Progenitor Cell Transplantation and Recovery.

Analysis of hematopoietic progenitor cell content of porcine cells transplanted to CHEF-NSG mice and recovered from bone marrow 21 weeks later (Experiment 2). The number of clonogenic progenitor cells transplanted was estimated from the total number of ckit-enriched cells transplanted and the clonogenic rates of these cells plated at the time of transplantation. Recovery rates were determined at sacrifice and are expressed as the average rate/100,000 recovered bone marrow cells for recipients within the donor group. Cells were plated in CFU medium supplemented with porcine IL-3, porcine GM-CSF and human EPO. Mice transplanted with carrier cells only yielded a few small colonies, readily distinguished from porcine colony forming cells.

Donor	Mice	Injected cell #/mouse		Recovered (rate/105 BMC)
		Total	Clonogenic
hCD47	4	3.7E+05	22, 951	33.6
WT	3	1.8E+06	28,800	5.2
Carriers	2	NA	NA	0

Results

Strategy for expression of human CD47 in swine

Our overall strategy for expressing human CD47 sequences in pigs using a knock-in type targeting vector is diagrammed in Figure 1. In order to mimic the widespread expression pattern of the native CD47 gene, we chose the ubiquitously expressed human EF1α promoter for expression of human CD47 (hCD47) coding sequences . To minimize the chances of variegated expression of the transgene due to genomic position effects, we chose to target the integration of the CD47 expression cassette into the α-1,3-galactosyltransferase (GGTA1) locus. The GGTA1 locus is ubiquitously expressed and loss of function is desired for the purposes of xenotransplantation. Additionally, we have previously described an effective gene trap targeting vector for GGTA1, which results in high relative targeting rates under G418 selection (34).

Generation of hCD47 transgenic cells for cloning

In order to produce targeted hCD47 transgenic pigs on a GGTA1 null background directly, without the need for additional mating, we chose to start with cells from a GGTA1 heterozygous animal. Fibroblasts from GGTA1 heterozygous KO miniature swine male fetus 19120-4 were transfected with the above knock-in vector and initially selected for resistance to high concentrations of G418 (see Materials and Methods). To determine if cells containing the expected hCD47 knock-in allele of GGTA1 could be detected within this G418 resistant population, we performed RT-PCR using a forward primer located in exon 7 of the GGTA1 gene (upstream of the 5′ end of the vector) and specific reverse primers for WT, GGTA1 KO and GGTA1 CD47 KI alleles. Transcripts from an hCD47 KI allele could be readily detected within the G418 selected population (Fig. 2A, left). The G418 selected population was then subjected to a single round of depletion of cells bearing the α-gal epitope using FITC labeled lectin BS-IB₄ and anti-FITC magnetic beads. Allele-specific RT-PCR following depletion revealed that RNA expression from a WT GGTA1 allele could no longer be detected within the population (Fig. 2A, right). FACS analysis of the population confirmed cell surface loss of gal-epitopes and expression of hCD47 (Fig. 2B).

Figure 2. Enrichment for hCD47 knock-In cells by microbead depletion with IB4 lectin.

(A) RT-PCR assays of the knock-in vector transfected, G418 selected, fibroblast population before and after depletion of α-1,3-gal epitope bearing cells by microbead depletion with IB4 lectin. Assays utilized a common forward primer located beyond the 5′ end of the vector and reverse primers specific to WT (290 bp), knockout (318 bp) and knock-in (349 bp) GGTA1 alleles (see Figure 1). M: 100 bp ladder sizes 300, 400 and 500 bp. Transcripts from cells bearing a WT allele were undetectable following lectin depletion. (B) Cell surface expression of hCD47 (using a primate specific anti-CD47 antibody) and α-1,3-gal epitopes (using labeled BS-IB₄ lectin) was assessed in untransfected fibroblasts (left) and in the knock-in vector transfected population following G418 selection and IB4 depletion (right).

Production of hCD47 Knock-in Pigs

We employed two rounds of cloning using chromatin transfer technology to generate the desired transgenic piglets. In the first round, the G418 selected/gal epitope depleted cell population above was used to produce early-to-mid gestation fetuses. This allowed us to generate genetically uniform, low passage fibroblast lines and confirm hCD47 expression in these lines. Cesarean section delivery of one recipient 38 days following embryo transfer yielded 2 apparently viable fetuses (846-1 and 846-2). Allele-specific RT-PCR of fibroblast lines established from these fetuses showed that both were heterozygous for a GGTA1 hCD47 knock-in allele and a GGTA1 knockout allele (Fig. 3A). Additional genomic PCR using a primer within the hCD47 coding sequences and a primer distal to the 3′ end of the knock-in vector confirmed the expected replacement type event of the WT GGTA1 allele in both fetuses (Fig. 3B). FACS analysis confirmed the cell surface expression of hCD47 and absence of gal epitopes in these lines (Fig. 3C).

Figure 3. Analysis of cloned fetuses produced with knock-in selected donor cells.

(A) GGTA1 allele specific RT-PCR was performed on untransfected fibroblasts and fibroblasts derived from cloned fetuses 846-1 and 846-2 as described in Figure 2 and confirmed the presence of 1 knock-in and 1 knockout allele in each. M: 100 bp ladder sizes 300 and 400 bp. (B) Eco RI digests of genomic PCR products using a forward primer within the hCD47 CDS and a reverse primer located distal to the vector end (see Figure 1) in fetus 846-1 (lane 1) and 846-2 (lane 3) generated expected bands of 3.0 and 6.8 kb, confirming expected replacement targeting of the WT GGTA1 allele. Untransfected fibroblasts (lane 2) yielded no amplification product. M: 1 kb ladder. (C) Cell surface expression of α-1,3-gal epitopes and human CD47 on fibroblasts from untransfected cells (filled), fetus 846-1 (dotted line) and fetus 846-2 (solid line). Both fetuses expressed human CD47 but not α-1,3-gal.

A second round of cloning using the above fetal cell lines was used to produce live-born piglets. Chromatin transfer with fibroblasts from fetus 846-2 produced 1 weaned and healthy piglet (18286). FACS analysis of peripheral blood mononuclear cells from this pig confirmed uniform expression of hCD47 on the surface of these cells (Fig. 4A). Uniform expression was also observed on red blood cells from this piglet (Fig. 4B). Cloning with cells from fetus 846-1 produced a piglet (1524) which died of indeterminate causes 1 week after birth. Marrow harvested from the long bones of this piglet was placed into hematopoietic colony forming culture and cell surface hCD47 expression was confirmed by FACS on the resulting colonies (Fig. 4C). RT-PCR of cardiac muscle, hepatocytes and brain gray matter demonstrated hCD47 RNA expression in cells of all three germ layers (Fig. 4D). These results demonstrate that widespread expression of a hCD47 transgene, including expression in hematopoietic lineages, can be obtained by targeting into the GGTA1 locus.

Figure 4. Human CD47 expression in cloned piglets.

(A) Expression of hCD47 on peripheral blood mononuclear cells of pig 18286, cloned from 846-2 fetal fibroblasts, and a non-transgenic control. Uniform expression of hCD47 was observed in the lymphocytic and monocytic lineages and (B) on red blood cells in the cloned piglet. (C) Expression of hCD47 on bone marrow derived hematopoietic colonies from pig 1524, cloned from 846-1 fetal fibroblasts. (D) Top: hCD47 RNA expression, assessed by RT-PCR, in cardiac muscle (H), hepatocytes (L) and brain gray matter (B) in piglet 1524 and a WT control. Bot: GAPDH expression in these same samples. M: 50 bp ladder.

hCD47 promotes engraftment and survival of porcine hematopoietic progenitor cells in vivo

To test the efficacy of hCD47 expression on engraftment and survival of porcine hematopoietic cells, we transplanted progenitor enriched cells from either transgenic founder 18286 or a WT pig into NOD background mice. Mice on the NOD background express a SIRPα receptor allele which can productively bind hCD47, enabling engraftment of human CD34+ cells in mice also bearing the severe combined immunodeficiency (scid) mutation (41;42). Furthermore, transplantation of WT porcine bone marrow cells, or enriched progenitors along with irradiated PBMC carriers, into irradiated NOD/scid mice expressing porcine cytokines IL-3, SCF and GM-CSF has previously been reported to result in low level engraftment of pig progenitor cells, providing a baseline for determining the efficacy of hCD47 expression in this murine model (38).

To further isolate the role of macrophage engulfment in xenogeneic engraftment, we chose as recipients porcine cytokine transgenic NOD/scid mice also null for expression of the IL2Rγ chain (CHEF-NSG). These mice are deficient not only in T and B cells but also NK cells, leaving only recipient myeloid cells to interact with transplanted porcine cells (43;44). hCD47 and WT donor cells were obtained by leukapheresis following stem cell mobilization with porcine SCF and IL-3. To better control progenitor cell content in transplanted cell populations in the absence of a porcine anti-CD34 reagent, we enriched the leukapheresis products for cells expressing ckit (CD117); the ckit⁺ population in pigs has been shown to contain the entire population of progenitor cells as measured in colony forming unit (CFU) assays (38). Mice received ckit-enriched cells from either hCD47 or WT donors, along with irradiated pig PBMC to normalize injected cell numbers as previously described (38). A control group received irradiated carriers only. FACS analysis of progenitor enriched cells from the hCD47 transgenic pig showed uniform expression of hCD47 in this population (Fig. S1A). Porcine chimerism levels were determined by FACS analysis using an anti-pig monoclonal reagent with pan specificity in the peripheral blood and bone marrow of individual mice 17–21 weeks following transplantation (Fig. S1B and S1C).

In a first experiment, we normalized the number of porcine progenitor cells to the percentage of ckit⁺ cells in the enriched leukapheresis products and analyzed chimerism 17 weeks following transplantation. High level porcine chimerism was observed in the bone marrow of hCD47 progenitor cell recipients (ave 43%, range 28–62%) and smaller but readily measurable chimerism levels observed in the peripheral blood of these recipients (Fig. 5). A much smaller, although statistically significant, difference in the level of bone marrow chimerism was observed in the WT group vs. Carrier Only group, while no detectable chimerism was observed in the peripheral blood of these groups.

Figure 5. Porcine chimerism in the peripheral blood and bone marrow of hCD47 progenitor cell recipients.

Progenitor enriched cells from cytokine mobilized peripheral blood of hCD47 transgenic pig 18286 or a WT control were transplanted along with irradiated porcine PBMC carriers to irradiated porcine cytokine transgenic NOD scid gamma (CHEF-NSG) mice. Chimerism was analyzed using an anti-pan pig antibody 17–21 weeks later (see Supplementary Figure 1) and expressed as the percentage of pig cells in the leukocyte population. (A) Individual mice were examined for peripheral blood and bone marrow chimerism 17 weeks following transplant of cells normalized to CD117⁺ (ckit⁺) content. (B) Analysis of individual mice 21 weeks following transplant of cells normalized to hematopoietic colony forming unit content. Student’s t-test of 1-way ANOVA was used for statistical analysis (threshold p <0.05).

In a second experiment, we normalized transplanted cells to progenitor cell content based on hematopoietic colony forming units and analyzed chimerism 21 weeks later. As in the first experiment, high level bone marrow engraftment and readily detectable peripheral chimerism were observed only in the recipients of hCD47 progenitor cells (Fig. 5B). Additionally, we observed a substantially greater frequency of porcine colony forming units in the bone marrow of hCD47 progenitor cell recipients at sacrifice (Table 1).

Discussion

Attempts to establish hematopoietic chimerism in pig to primate xenogeneic models to date have been hampered by rapid clearance of porcine cells from primate circulation by host macrophages. Studies in murine models using transfected porcine tumor cells have suggested that transgenic expression of primate CD47 on pig cells may substantially enhance peripheral survival, which in turn may lead to increased engraftment. Here, we report the generation of transgenic, MHC-defined, GalT-KO miniature swine expressing human CD47 on all hematopoietic lineages examined.

We found a clear enhancement of peripheral blood and bone marrow chimerism using an hCD47 transgenic donor of hematopoietic progenitor-enriched cells in porcine cytokine transgenic NSG mice, with all transgenic recipients in both trials showing engraftment. Unlike NSG models for optimal human cell engraftment (41) that receive transplants neonatally, recipient mice in this study were 12–13 weeks of age, presumably fully mature with respect to their capacity for clearance of xenogeneic cells at the time of transplant and fully transitioned to bone marrow based hematopoiesis. In contrast to a previously published report of WT porcine progenitor transplantation in cytokine transgenic mice (38), we failed to observe statistically significant engraftment of WT cells in one of our trials, and considerably lower engraftment levels in the other. One factor contributing to this difference may be our use of mobilized peripheral cells as a donor source, as opposed to the bone marrow utilized in the previous study. Although roughly equivalent numbers of ckit-enriched cells were transplanted in both cases, the clonogenic cell rate in our mobilized ckit⁺ cells was roughly one-third of that reported in the bone marrow study. We chose to use mobilized cells for these studies as large numbers of porcine bone marrow cells are difficult to obtain via aspiration and we wished to maintain the donor for breeding and other experimental purposes. Additionally, we used cytokine transgenic NSG recipients, prepared with 225 cGy of total body irradiation, as opposed to cytokine transgenic NOD/scid recipients irradiated with 300 cGy. The combined effect of the IL2 receptor gamma chain knockout and the irradiation reduction required with NSG mice is unknown. Like the previous study, we found substantially higher levels of chimerism in the bone marrow of hCD47 transgenic recipients than in peripheral blood.

In designing the transgenic strategy, we chose to add a human CD47 expression cassette rather than attempt to replace the native CD47 sequences via targeting. Although largely normal under specific pathogen free conditions, CD47 KO mice have been found to have increased susceptibility to lethal infection (45). Although the role of hCD47 in SIRPα-mediated inhibition of macrophage phagocytosis is of primary interest here, it should be noted that CD47 has been implicated in a number of other functions in vivo (22). Of note, binding of CD47 to thrombospondin-1 has been associated with hematopoietic cell apoptosis (46), increased renal reperfusion injury (47) and increased sensitivity to radiation (48). Four evolutionarily conserved splice variants of CD47, all of which differ only in the length of the cytoplasmic tail (35), have been described. While it is not clear at this point what the significance of these differences is, we chose to express splice form 1 of hCD47. Expressed primarily in keratinocytes, the 4 amino acid cytoplasmic domain is the shortest of the variants and perhaps less likely to result in an undesired gain of function, either in the swine or following xenogeneic transplantation. Other than a congenital front limb tendon defect, common in cloned miniature swine (34) and responsive to therapy, transgenic pig 18286 displayed no abnormalities prior to sacrifice for other experiments at 18 months of age. WBC and hematocrit over this period were consistently in the normal range for our swine herd (WBC 21–25 × 10⁶ cells/ml, Hct 35–40) and detailed phenotypic analysis of the PBMC fraction found normal ranges for CD4 and CD8 T cells, B cells, monocytes and NK cells.

In an effort to avoid position effect silencing or variegation of transgene expression common in mammalian transgenics (49), we targeted integration of the hCD47 expression unit under the control of the ubiquitous EF1α promoter to the ubiquitously expressed GGTA1 locus. We observed uniform, long-term expression of hCD47 in peripheral blood lineages and hematopoietic progenitors in one pig and in hematopoietic lineages and other tissues examined in a second pig 1 week after birth.

We employed a novel selection process to obtain the desired targeted cells for as donor cells in cloning based in part on loss of function of the GGTA1 locus (see Methods). This methodology permitted the selection of the knock-in allele in the presence of the existing knockout allele without the need for the introduction of an additional selection marker. It is of interest to note that such a strategy can be successful in primary cells over a short selection period even when based on cell surface loss of a common glycosyl epitope.

In summary, this study demonstrates that hematopoietic progenitor cells from transgenic pigs expressing human CD47 have increased engraftment and survival in a murine model of human bone marrow engraftment, presumably through modulation of macrophage phagocytosis. These results should now provide a solid foundation for efficacy studies of bone marrow and mobilized peripheral blood progenitor transplantation in pig to primate models of xenogeneic tolerance and studies directed toward this goal are currently under way in our laboratory. These studies will also allow us to address questions regarding porcine hematopoietic cell differentiation and the relationship between hCD47 expression levels and macrophage inhibition in various hematopoietic lineages in a clinically relevant model.

Supplementary Material

Supp FigureS1-S2

Figure S1. (A): hCD47 expression in CD117 enriched porcine progenitor cells. Uniform expression of hCD47 was observed in the transplanted cell population obtained via leukapheresis of cytokine mobilized transgenic donor pig 18286. Representative histograms of cells stained with anti-pan pig monoclonal antibody 1030H-1-19 in leukocyte gated populations obtained from the peripheral blood (B) and bone marrow (C) of representative mice are shown. The percentage of pan-pig positive cells from this analysis comprises the data set for Figure 5.

Table S1. PCR Primers.

Table S2. Embryo Transfers. Six transfers of embryos reconstructed from cloned fetal cell lines from 846-1 and 846-2 resulted in 4 pregnancies to term and 15 live births. Other than a lower leg tendon defect in pig 18286, none displayed gross external abnormalities. Necropsies were performed on 4 piglets which died within days of birth (including piglet 1524); no internal abnormalities were observed and a definite cause of death could not be determined for any of these animals. Early deaths without definitive cause are not uncommon in cloned swine and have been reported previously in cloned miniature swine (32,34).

Abbreviations

αgal

α-1,3-galactosyltransferase epitope

ANOVA

analysis of variance

BFU-E

burst forming unit erythroid

BGH

bovine growth hormone

BS-IB₄

Bandeiraea simplicifolia isolectin B4

CFU-GM

colony forming unit granulocyte-macrophage

DMEM

Dulbecco’s modification of Eagle’s medium

EF1α

Elongation Factor 1α

EMCV

Encephalomyocarditis virus

EPO

erythropoietin

FITC

fluorescein isothiocyanate

GalT-KO

α-1,3-galactosyltransferase knock-out

GGTA1

α-1,3-galactosyltransferase gene

GM-CSF

Granulocyte-macrophage Colony Stimulating Factor

hCD47

human CD47

IL2Rγ

interleukin 2 receptor gamma chain

IL-3

Interleukin-3

IRES

internal ribosome entry site

KI

knock-in

KO

knock-out

MDBK

Madin-Darby bovine kidney cell line

NSG

NOD scid IL2 receptor gamma chain knockout

pA

poly adenylation site

SCF

Stem Cell Factor

scid

severe combined immunodeficiency

SIRP

Signal Regulatory Protein

TBI

total body irradiation