Anti-CD47 Antibodies Promote Phagocytosis and Inhibit the Growth of Human Myeloma Cells

ORIGINAL ARTICLE Anti-CD47 antibodies promote phagocytosis and inhibit the growth of human myeloma cells

D Kim1, J Wang1,3, SB Willingham1, R Martin1, G Wernig2 and IL Weissman1

Multiple myeloma is a plasma cell neoplasm residing in bone marrow. Despite advances in myeloma therapies, novel therapies are required to improve patient outcomes. CD47 is highly expressed on myeloma cells and a potential therapeutic candidate for myeloma therapies. Flow cytometric analysis of patient bone marrow cells revealed that myeloma cells overexpress CD47 when compared with non-myeloma cells in 73% of patients (27/37). CD47 expression protects cells from phagocytosis by transmitting an inhibitory signal to macrophages. Here we show that blocking CD47 with an anti-CD47 monoclonal antibody increased phagocytosis of myeloma cells in vitro. In xenotransplantation models, anti-CD47 antibodies inhibited the growth of RPMI 8226 myeloma cells and led to tumor regression (42/57 mice), implicating the eradication of myeloma-initiating cells. Moreover, anti-CD47 antibodies retarded the growth of patient myeloma cells and alleviated bone resorption in human bone-bearing mice. Irradiation of mice before myeloma cell xenotransplantation abolished the therapeutic efﬁcacy of anti-CD47 antibodies delivered 2 weeks after radiation, and coincided with a reduction of myelomonocytic cells in spleen, bone marrow and liver. These results are consistent with the hypothesis that anti-CD47 blocking antibodies inhibit myeloma growth, in part, by increasing phagocytosis of myeloma cells.

Leukemia (2012) 26, 2538–2545; doi:10.1038/leu.2012.141 Keywords: multiple myeloma; myeloma-initiating cells; CD47; xenotransplantation; therapeutic antibody

INTRODUCTION in vitro and inhibiting the growth of the various hematological and 12,13,15,16 Multiple myeloma is a clonal plasma cell neoplasm that utilizes a solid tumors in vivo. bone marrow microenvironment for survival and proliferation.1–3 A previous microarray analysis indicated that CD47 is among the Current myeloma therapies such as hematopoietic cell trans- highly expressed genes by malignant myeloma cells compared with 18 plantation and combinatorial chemotherapies are rarely curative healthy plasma cells. Here, we examined the expression of CD47 and relapse is common.4 This implies that therapy-resistant on the surface of patient myeloma cells, the therapeutic potential myeloma-initiating cells exist and that new therapeutics must be of an anti-CD47 monoclonal antibody (B6H12) in vitro and in xeno- developed to target and eradicate these myeloma-initiating cells. transplantation models. The B6H12 mouse anti-CD47 antibodies CD47 is an integrin-associated receptor protein that is expressed are of the IgG1 class, and did not induce antibody-dependent on a broad range of cell types, including red blood cells.5,6 CD47 cell-mediated cytotoxicity (ADCC) or complement-dependent interacts with signal regulatory protein-a (SIRPa), thrombospondin cytotoxicity (CDC). However, anti-CD47 monoclonal antibodies did (TSP)-1 and -2 to mediate various cellular functions.7,8 In particular, induce the phagocytosis of human myeloma cells by macrophages CD47 signaling through SIRPa inhibits the phagocytosis of CD47- in vitro and signiﬁcantly inhibited myeloma growth in human fetal expressing target cells by SIRPa-expressing macrophages. Anti- bone-free and -bearing xenotransplantation models. CD47 monoclonal antibodies inhibiting the interaction of CD47 and SIRPa promote the phagocytosis of tumor cells by macro- MATERIALS AND METHODS 9–11 phages, and also by monocytes and neutrophils. Cancer cells Preparation of patient bone marrow cells including acute myelogenous leukemias,11,12 lymphomas,13,14 and 15–17 Clinical bone marrow samples were obtained from newly diagnosed, various solid cancers express CD47 at a high level, and can relapsed or post-treatment patients with multiple myeloma from October be phagocytosed by macrophages in the presence of blocking 2007 through August 2011, according to protocols approved by the 11–17 antibodies to CD47. In those cases where the tumor-initiating Institutional Review Board of the Stanford University School of Medicine cells (or cancer stem cells) are known, for example, from acute (eprotocol no.: 9718). Mononuclear cells were isolated using Ficoll-Paque myelogenous leukemias and bladder cancers, the tumor-initiating (Pharmacia, Uppsala, Sweden). cells also upregulate CD47 expression, possibly to avoid phagocytosis. Indeed, the enhanced CD47 expression in these Human myeloma cell lines and generation of luciferase-expressing cancers has been correlated with poor prognosis.12,15,16 Blocking RPMI 8226 the CD47-SIRPa interaction with anti-CD47 monoclonal antibodies Human myeloma cell lines (RPMI 8226 and NCI H929) were purchased from has proven effective in inducing the phagocytosis of tumor cells American Type Culture Collection (Manassas, VA, USA). Human myeloma

1Institute for Stem Cell Biology and Regenerative Medicine, Stanford University School of Medicine, Stanford, CA, USA and 2Department of Pathology, Stanford University School of Medicine, Stanford, CA, USA. Correspondence: Dr D Kim or Professor IL Weissman, Institute for Stem Cell Biology and Regenerative Medicine, Stanford University School of Medicine, 265 Campus Drive, Room 3155, Stanford, CA 94305, USA. E-mail: [email protected] (DK) or [email protected] (ILW) 3Current address: Department of Biomedicine, University of Bergen, Jonas Lies Vei 915009, Bergen, Norway. Received 17 April 2012; revised 15 May 2012; accepted 16 May 2012; accepted article preview online 30 May 2012; advance online publication, 26 June 2012 Anti-myeloma effect of an anti-CD47 antibody D Kim et al 2539 cell lines, U266, KMS18 and MM1.S, were obtained from Drs Anna Scuto images. At least three independent experiments were performed for each and Richard Jove (City of Hope, Duarte, CA, USA). For in vivo biolumines- cell line and patient samples. cence imaging, RPMI 8226 cells were transduced with pCDH-CMV-MCSEF1- puro HIV-based lentiviral vector (Systems Biosciences, Mountain View, CA, 16 Analysis of anti-CD47 direct cytotoxicity, CDC and antibody- USA) encoding eGFP and luciferase as described previously. All human dependent cell-mediated cytotoxicity myeloma cell lines were cultured in RPMI 1640 medium (Invitrogen, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (FBS), Pen/ The cytotoxicity of anti-CD47 (clone B6H12) antibodies was evaluated using Strep (Invitrogen), GlutaMax (Invitrogen), sodium pyruvate (Invitrogen), XTT (sodium 2,3,-bis(2-methoxy-4-nitro-5-sulfophenyl)-5-((phenylamino)-car- non-essential amino acids (Invitrogen) and b-mercaptoethanol (Sigma, bonyl)-2H-tetrazolium inner salt; Sigma) assays. Myeloma cells were incubated St Louis, MO, USA). in RPMI 1640 complete media with or without 10 mg/ml of B6H12. After 24–48 h, XTT and PMS (N-methyl dibenzopyrazine methyl sulfate; Sigma) were Flow cytometric analysis added to cells and incubated for 4 h. Colorimetric analysis was performed with Multiskan EX plate reader (Thermo Electron Corporation, Pittsburgh, The following antibodies were used to stain human bone marrow PA, USA). mononuclear cells: CD138 (B-A38, PE, Beckman Coulter Inc., Fullerton, CDC analysis was performed using two methods. First, we performed CA, USA), CD19 (HIB19, AlexaFluor 700 or APC, BD Biosciences, San Jose, XTT assays on myeloma cells cultured with mouse serum (10% intact or CA, USA), CD38 (HIT2, FITC or AlexaFluor, BD Biosciences), CD45 (HI30, heat-inactivated serum) as described above. Second, we performed Calcein PE-cy7, BD Biosciences) and CD47 (B6H12, PE, BD Biosciences). Flow AM release assays. Here, myeloma cells were labeled with Calcein cytometric analysis was performed on a BD FACSAriaII. AM (Invitrogen) following the manufacturer’s instruction and cultured with 10% mouse sera and mouse anti-CD47 antibodies for 4 h. After Mouse and xenotransplantation models centrifugation, fluorescence in supernatant was measured with Spectra- Max M3 (Molecular Devices, Sunnyvale, CA, USA). All procedures involving animals were approved by the Stanford University For ADCC assays, we incubated Calcein AM-labeled myeloma cells School of Medicine’s Administrative Panel on Laboratory Animal with mouse peripheral blood cells at the indicated ratio and analyzed as Care Committee (protocol no.: 16886). Six- to 10-week-old recombinase- described above. activating gene 2/common cytokine receptor-g chain-deficient (RAG2 À /gc À ) mice were anesthetized with isofluorane and subcutaneously injected with 2–5 Â 105 luciferase-expressing (GFP þ ) RPMI 8226 Bioluminescent imaging myeloma cells suspended in phosphate buffered saline (PBS) containing Bioluminescent imaging was performed on an IVIS Spectrum (Caliper Life 25% Matrix Matrigel (basement membrane; BD Biosciences). For experi- Science, Hopkington, MA, USA) and quantified using Living Image 4.0 ments involving radiation, mice received 400 rad whole-body irradiation software (Caliper Life Science) as described previously.16 Mice were imaged using Faxitron Bioptics (Lincolnshire, IL, USA) 12 h before the transplanta- 18 min after luciferin injection and total flux (photons/second) values were tion of 2000 myeloma cells. calculated within a region of interest corresponding to the anatomic À À Human bone-bearing RAG2 /gc mice were generated as region of tumor engraftment when peak radiance was achieved. Mice 19–21 previously reported. Briefly, femurs, tibias and fibulas of 17- to 18- exhibiting comparable total flux values were then randomized into gestational-week-old fetuses (Advanced Bioscience Resources; Alameda, treatment groups and imaged weekly. CA, USA) were implanted subcutaneously into 6–8 week-old mice. After 4 weeks, patient bone marrow cells in PBS supplemented with 25% high protein Matrigel (BD Biosciences) were transplanted directly into human Enzyme-linked immunosorbent assay bone grafts. Levels of human immunoglobulin kappa or lambda light chains in mouse serum were determined as an indicator of myeloma cell growth using a human kappa or lambda immunoglobulin enzyme-linked immunosorbent Anti-CD47 antibody treatment of xenograft tumors assay kits according to the manufacturer’s directions (Bethyl Laboratories, Mice transplanted with luciferase-expressing RPMI 8226 were randomized Montgomery, TX, USA). Mouse sera were collected every 2–3 weeks to into treatment cohorts after confirming engraftment of myeloma cells with measure the levels of circulating human antibodies. bioluminescent imaging. Antibody treatment consisted of daily intraperitoneal injections of 500 mg of anti-CD47 monoclonal antibody (clone B6H12.2) or mouse polyclonal IgG (Innovative Research, Novi, MI, USA) Histology control in 100 ml of PBS for 4–5 weeks. Fractionated mouse bone marrow cells were identified morphologically by Mice transplanted with patient myeloma xenograft or primary myeloma Wright-Giemsa staining (Sigma-Aldrich, St Louis, MO, USA) of cytospin slide cells were injected intraperitoneally with 500 mg of anti-CD47 monoclonal preparations. Slides were evaluated by a minimum of three qualified antibody (B6H12) or PBS daily for 4 weeks after the detection of human researchers. antibody in mouse sera.

Phagocytosis assay RESULTS Macrophages were derived from bone marrow mononuclear cells of 6–8 Patient myeloma cells express high levels of CD47 week-old RAG2 À /gc À female mice. Bone marrow mononuclear cells were We examined the expression of CD47 on myeloma cells in patients isolated using Ficoll-Paque (Pharmacia) and cultured in RPMI 1640 through flow cytometric analysis. Myeloma cells in bone marrow supplemented with 10% FBS, Pen/Strep (Invitrogen), GlutaMax (Invitrogen), aspirates were identified by the high expression of CD38.22 Nearly sodium pyruvate (Invitrogen), non-essential amino acids (Invitrogen), all CD38high myeloma cells from 37 patient samples expressed b-mercaptoethanol (Sigma) and 20 ng/ml human GM-CSF (Peprotech, Rocky Hill, NJ, USA) for 5–6 days. Adherent macrophages were collected by CD47. Next, we examined the relative expression level of CD47 on gentle scraping and plated at a density of 5 Â 104 cells per well in a 24-well myeloma and non-myeloma cells from the patient bone marrow. tissue-culture plate. Human myeloma cells were fluorescently labeled with We gated bone marrow cells into four subpopulations based on 2.5 mM carboxyfluorescein succinimidyl ester (CFSE; Invitrogen)) following CD38 and CD45 expression and analyzed the mean fluorescence the manufacturer0s instructions. Macrophages were incubated in serum- of intensity of CD47. CD38high and CD38low/ À CD45high cells free medium for 2 h before adding 2 Â 105 CFSE-labeled live myeloma expressed CD47 at higher levels than CD38low/ À CD45low cells cells. Antibodies (10 mg/ml) were added and incubated for 2 h in an and CD38 À CD45 À cells of myeloma patients (see Supplementary incubator maintained at 37 1C. Macrophages were washed at least five Figure S1 for representative analyses). The relative expression times to remove non-phagocytosed myeloma cells, fixed with freshly of CD47 on CD38high myeloma cells and CD38low/ À CD45high prepared 4% paraformaldehyde (Sigma) and then imaged using Leica DMI6000 B inverted fluorescence microscope (Leica Microsystems GmbH, non-myeloma cells was further examined to determine whether Wetzlar, Germany). Composite images were generated from three to anti-CD47 antibodies could preferentially target myeloma five regions from each well, including a bright-field image and fluorescent cells compared with non-myeloma cells (Figure 1a). Myeloma high images. The phagocytic index was calculated as the frequency of phago- cells (CD38 ) from 64.9% (24/37) of patients expressed at least cytosed CFSE þ myeloma cells per 100 macrophages in the composite twofold more CD47 than CD38low/ À CD45high cells while 10.8%

& 2012 Macmillan Publishers Limited Leukemia (2012) 2538 – 2545 Anti-myeloma effect of an anti-CD47 antibody D Kim et al 2540

Figure 1. CD47 is highly expressed on myeloma cells. (a) Representative analysis of CD47 expression on MM patient bone marrow cells. (b) Relative expression of CD47 on CD38high myeloma cells and CD45highCD38low/ À non-myeloma cells from myeloma patients. (c) CD47 expression on myeloma cell lines and patient-derived myeloma xenograft cells (pt. D).

(4/37) of patients had lower expression of CD47 on CD38high cells.26 We examined whether anti-CD47 antibodies (B6H12) cells than on CD38low/ À CD45high cells (Figure 1b). Lowering affected the viability of myeloma cells. Neither 10 nor 50 mg/ml the threshold to 1.5-fold, myeloma cells from 73.0% (27/37) of of B6H12 affected the viability of RPMI 8226 and NCI H929 patients had higher expression of CD47 than non-myeloma myeloma cells after 24–48 h incubation (Supplementary Figure CD38low/ À CD45high cells. S3A). In addition, we did not observe CDC or ADCC initiated by CD47 expression was also observed on almost all cells in various B6H12 at an antibody concentration (10 mg/ml) that promoted myeloma cell lines, including RPMI 8226, NCI H929, MM1s, KMS18 phagocytosis of the same myeloma cells (Supplementary Figures and U266 (Figure 1c). The universal expression of CD47 on S3B and S3C). These data indicate that increased phagocytosis myeloma cells suggests that CD47 may be a potential therapeutic of myeloma cells by anti-CD47 antibody (B6H12) does not result as antibody target for myeloma cells, including the myeloma- a consequence of cell death caused by direct cytotoxicity, CDC, or initiating cells enriched within CD38high cells of multiple myeloma ADCC, followed by cell removal. patients.23,24

Anti-CD47 antibody, B6H12 promotes the phagocytosis Anti-CD47 antibodies inhibit the growth of myeloma cells in vivo of myeloma cells We next evaluated whether anti-CD47 antibody treatment Blocking CD47 with anti-CD47 monoclonal antibodies has enabled inhibits the growth of luciferase-expressing RPMI 8226 cells in the phagocytosis of leukemia and lymphoma cells.12,13,15 We RAG2 À /gc À mice. Two weeks after xenotransplantation, we investigated whether an anti-CD47 monoclonal antibody (B6H12) confirmed tumor formation and initiated anti-CD47 antibody promotes the phagocytosis of myeloma cells. Macrophages therapy (Figure 3a). In five independent experiments, we observed derived from mouse bone marrow cells phagocytosed human significantly inhibited tumor growth in mice treated with anti- myeloma cells at a low frequency when treated with PBS or a CD47 antibodies (Figures 3b and c). In addition to the reduced control mouse IgG. However, anti-CD47 antibodies significantly tumor burden, we observed significant tumor regression in anti- increased the phagocytosis of RPMI 8226, NCI H929, KMS18 and CD47 antibody-treated cohorts compared with mice treated with MM1.S myeloma cells (Figure 2a). As shown previously,12,13,15,25 a control mouse IgG (remission rate: 73.7% vs 40.7%) (Figure 3d). non-blocking anti-CD47 antibody, clone 2D3, enabled a minimal Tumor remission was maintained in anti-CD47 antibody-treated increase in phagocytosis (Supplementary Figure S2). Anti-CD47 mice 1 week after completion of the 5-week treatment while the antibodies also increased the phagocytosis of patient-derived regressed tumors in the control group regrew after stopping xenograft myeloma cells (Figure 2b) and patient bone marrow treatment (remission rate: 71.9% vs 18.6%) (Figure 3d). These cells consisting of 95% of myeloma cells (Figure 2c). results indicate that anti-CD47 antibody treatment inhibits the Anti-CD47 antibodies were previously reported to directly growth of RPMI 8226 cells in mice, implicating the eradication of initiate cell death upon binding to chronic lymphocytic leukemia myeloma-initiating cells or myeloma cancer stem cells.

Leukemia (2012) 2538 – 2545 & 2012 Macmillan Publishers Limited Anti-myeloma effect of an anti-CD47 antibody D Kim et al 2541

Figure 2. An anti-CD47 antibody, B6H12, promotes the phagocytosis of myeloma cells by macrophages. (a) Increased phagocytosis of various myeloma cells. Upper: green signals indicate macrophage engulfed CFSE-labeled myeloma cells, inset: a representative image of macrophages that the green myeloma cells are inside, lower: quantitation of the phagocytic index of myeloma cell lines as described in Materials and Methods. (b) Phagocytic index of a patient-derived myeloma xenograft cells treated with control mouse Ig (mIgG) or B6H12. (c) Phagocytic index of patient myeloma cells coclutured with control mouse Ig and B6H12.

Figure 3. Blocking CD47 inhibits the growth of RPMI 8226 myeloma cells. (a) Experimental scheme: 200 000–500 000 luciferase-expressing myeloma cells were subcutaneously transplanted into RAG2 À /gc À mice. After 2 weeks, 500 mg of control mouse IgG (mIgG) or B6H12 antibodies were injected daily, intraperitoneal, for 5 weeks. (b) Representative bioluminescent images of tumor growth in mice treated with control mouse IgG and B6H12 after 5 weeks of treatment. (c) Fold changes of tumor sizes at weeks 5 and 6. All tumors were combined from ﬁve independent experiments. Each independent experiment showed similar tumor growth patterns. (d) Tumor regression rates in mice at weeks 5 and 6. All 58 tumors were combined from ﬁve independent experiments. Each independent experiment showed similar regression rates.

Anti-CD47 antibodies inhibited myeloma growth in the presence fragments into immunocompromised mice and directly trans- of a human bone marrow microenvironment planted patient-derived myeloma cells into the bone grafts. After The bone marrow microenvironment supports the survival and detecting human immunoglobulin in mouse sera, we treated mice proliferation of myeloma cells.2 To determine whether anti-CD47 daily with anti-CD47 (B6H12) antibody or PBS for 4 weeks antibodies also inhibit myeloma growth in the presence of human (Figure 4a). Mice treated with the anti-CD47 antibody produced bone marrow microenvironment, we grafted human fetal bone human immunoglobulin at a lower level than mice in the control

& 2012 Macmillan Publishers Limited Leukemia (2012) 2538 – 2545 Anti-myeloma effect of an anti-CD47 antibody D Kim et al 2542

Figure 4. B6H12 inhibits the growth of myeloma cells growing in human bone grafts. (a) Experimental scheme: 500 000 myeloma xenograft cells derived from a patient myeloma sample (pt. D) were transplanted into human fetal bone grafts in RAG2 À /gc À mice. After 4 weeks, anti- CD47 antibodies (500 mg, B6H12) or PBS were injected intraperitoneal for 4 weeks daily. Human Ig levels were measured weekly. (b) Fold change of human Ig level in mice. The combined results of at least three independent experiments are shown. (c) Radiography of mice treated with B6H12 or PBS 12 weeks after myeloma cell transplantation. The left image is of a mouse that was not transplanted with myeloma cells. (d) Tumor sizes treated with control mouse Ig or anti-CD47 antibodies.

group (Figure 4b). Consistent with the myeloma growth in the PBS 500 mg of B6H12 or mouse IgG for 4 weeks (Figure 6a). Though a group, bone grafts in the PBS cohort exhibited significant bone small number of tumor cells (2000 cells) were initially trans- resorption compared with grafts in the anti-CD47 antibody- planted, anti-CD47 antibody therapy did not inhibit tumor growth treated mice (Figure 4c). The reduction of myeloma growth in the (Figure 6b). This implies that the efficacy of anti-CD47 antibody anti-CD47 antibody-treated group was further confirmed by tumor treatment requires irradiation-sensitive effectors, such as progeny volume measurements (Figure 4d). of hematopoietic cells. We therefore examined the frequency We next examined whether anti-CD47 antibodies inhibit the of myeloid-lineage cells in the bone marrow, spleen and liver of growth of primary patient myeloma cells in human bone-bearing irradiated mice, as RAG2 À /gc À mice lack mature B, T or NK cells. mice. We transplanted bone marrow cells from four myeloma We observed a significant reduction of Gr-1medMac-1 þ / À cells in patients into human bone grafts. After 4 weeks, we treated mice the bone marrow, spleen and liver of irradiated mice compared daily with PBS or 500 mg of anti-CD47 antibodies (Figure 5a). with those of control mice 10 days after irradiation (Figures 6c We have previously shown that one can monitor the growth of and d). The morphological analysis of Gr-1medMac-1 þ / À cells human myeloma cells by serially analyzing the mouse serum indicated that these cells enriched for undifferentiated blasts and for human immunoglobulin, here myeloma immunoglobulin.24 monocytic cells (Figure 6e). These cells were further characterized Human immunoglobulin was detected in the sera of PBS-treated as Ly-6C-CD31high/ þ cells and macrophage progenitor-enriching mice (6/8), although only one of seven anti-CD47 antibody-treated Ly6c þ CD31 þ /lowMac-1 þ cells (Figure 6f as described by De Bruijn mice produced human antibodies (Figure 5b). We collected bone et al.27 In addition, we observed the reduced Gr-1lowMac-1 þ marrow cells from human bone grafts 8 weeks after initiation of macrophages in the spleens of irradiated mice (Figure 6e). These antibody therapy and examined whether human myeloma cells observations implicate progeny of radiation-sensitive hemato- repopulated in human bone grafts. CD138 þ cells were observed poietic cells in the bone marrow and periphery as critical effectors in 4 of 8 mice treated with PBS. In contrast, only 1 of 7 mice of the anti-CD47 antibody-mediated therapeutic effect, in addi- treated with anti-CD47 antibodies had detectable human CD138 þ tion to the phagocytic activity of resident, radiation-resistant cells (Figure 5c). These results indicate that blockade of CD47 macrophages. with targeted monoclonal antibodies inhibits the growth of myeloma cells, even in the context of a human bone marrow microenvironment. DISCUSSION CD47 is a multifunctional surface protein expressed in a wide range of cells. In particular, CD47 negatively regulates the Irradiation abolishes the anti-myeloma effect of anti-CD47 phagocytic activity of macrophages.6,9 CD47 on target cells antibody in mice interacts with SIRPa on macrophages and transmits a ‘don’t eat To examine whether the anti-myeloma effect of anti-CD47 me’ signal through SH2 domain-containing phosphatases, SHP-1 antibodies is dependent on cells surviving 400 rad given whole and -2, and results in the inhibition of target cell body 2 weeks before the beginning antibody treatment, we phagocytosis.7,28,29 CD47 is expressed at lower levels on bone sublethally irradiated mice and transplanted luciferase-expressing marrow resident hematopoietic stem cells (HSCs) than on RPMI 8226 cells; 2 weeks later we treated the mice daily with circulating HSCs. However, when bone marrow HSCs were

Leukemia (2012) 2538 – 2545 & 2012 Macmillan Publishers Limited Anti-myeloma effect of an anti-CD47 antibody D Kim et al 2543 the regression of tumors in mice. Although regressed tumors regrew in control IgG-treated mice after stopping treatment, no tumors regrew in mice experiencing remissions following anti-CD47 antibody treatment (Figure 3), probably caused by the eradication of myeloma-initiating cells. B6H12 is one of the best-characterized anti-CD47 antibody clones that block the interaction of CD47 and SIRPa. Consistent with previous reports,12,13,15,25 B6H12 induced the phagocytosis of myeloma cells but a non-blocking anti-CD47 antibody, clone 2D3, showed minimal increase of the phagocytosis (Supplementary Figure S2). Previously, it was reported that bivalent single-chain antibodies targeting CD47 induced the apoptosis of leukemia,31 myeloma,26 monocytes and dendritic cells.32 However, B6H12 itself did not have cytotoxicity to myeloma cells (Supplementary Figure S3). Though 10 mg/ml of B6H12 did not induce CDC or ADCC, low concentrations (1–10 ng/ml) of B6H12 induced minimal levels of ADCC, as observed previously.13 Meanwhile, we found that mouse polyclonal IgG used for control treatment also induced minimal cytotoxicity to myeloma cells (Supplementary Figure S4). The anti-myeloma activity of B6H12 appeared to require phagocytic cells or their progenitors such as Gr-1medMac-1 À / þ monocytic cells, Gr-1lowMac-1 þ macrophages and a subset of dendritic cells in bone marrow, spleen and liver. The importance of macrophages for anti-CD47 antibody-mediated anti-tumor effects was reported in previous studies wherein immunocompromised mice lacking T cells, B cells and NK cells rapidly eliminated transplanted CD47-deﬁcient hematopoietic cells or human AML cells, but this elimination was blocked following the depletion of macrophages.12,33 The bone marrow microenvironment is critical for myeloma cell survival and proliferation.34–37 In addition, the bone marrow microenvironment has been reported to protect myeloma cells from chemotherapy.38,39 We have shown that patient myeloma cell growth is dependent on human bone marrow micro- Figure 5. Anti-CD47 antibody treatment inhibits the growth of environment.24 The growth of patient myeloma cells was patient myeloma cells in human bone grafts. (a) Experimental inhibited by anti-CD47 antibodies, even in the presence of a scheme: 1 Â 107 patient bone marrow cells of myeloma patients human bone microenvironment. In addition, bone resorption is (four patients) were transplanted into human fetal bone grafts in À À one of the most severe pathological events in multiple myeloma RAG2 /gc mice. After 4 weeks, 500 mg of B6H12 or PBS were patients. In our xenotransplantation models, anti-CD47 antibody injected intraperitoneal for 4 weeks daily. (b) Repopulation of therapy signiﬁcantly reduced the human bone resorption. human antibody-secreting cells in human bone grafting mice treated with PBS or anti-CD47 antibodies at week 12. (c) The reduced bone resorption in mice treated with anti-CD47 Representative results of CD138 þ human myeloma cell engraftment antibodies indicates that the antibody therapy inhibited the in human bone grafts in mice treated with PBS or anti-CD47 growth of myeloma cells without toxicity to normal human bone antibodies. marrow cells. Therefore, these results imply that the CD47 therapy may be capable of increasing the quality of life of patients by reducing bone resorption as well as inhibiting myeloma growth. mobilized out of bone marrow, they upregulated CD47.11 As a multifunctional protein, CD47 has roles in various biological Ide et al.30 demonstrated that the overexpression of CD47 on activities. In addition to acting as a ‘don’t eat me’ signal, Kukreja donor cells or treatment of recipient mice with soluble CD47 et al.40 have suggested a role of CD47 in osteoclastogenesis. CD47 enhanced the engraftment of xenogenic donor cells. This suggests on myeloma cells also interacts with TSP-1 on immature dendritic that the overexpression of CD47 may be correlated with higher cells (iDCs) and, it has been reported, induces the differentiation of resistance to phagocytosis by macrophages. Indeed, increased the iDCs into multinuclear osteoclast-like cells.40 An anti-TSP-1 CD47 expression has also been observed on various patient tumor antibody blocking the interaction of CD47 and TSP-1 abolished cells.12,13,15–17,25 In some instances, CD47 expression levels were the differentiation of iDCs to osteoclast-like cells. Therefore, highly correlated with a worse prognosis.12,13,15,16,25 blockade of CD47 with anti-CD47 antibodies may abolish the Here we have shown that in most cases (65%) CD47 expression interaction with TSP-1 and inhibit the formation of osteoclast-like on patient CD38high myeloma cells is at least twofold higher than cells and reduce bone resorption in patients. Indeed, the reduced in patient-matched healthy bone marrow cells. Consistent with bone resorption in mice treated with anti-CD47 antibodies may our observation, Zhan et al.18 demonstrated that CD47 mRNA partially result from reduced osteoclastogenesis. expression levels are highly upregulated in myeloma cells In this preclinical study, we have shown that CD47 is a compared with healthy plasma cells. In an ongoing parallel promising novel therapeutic target on myeloma cells. Blocking study, we found that myeloma-initiating ability is enriched in CD47-SIRPa interaction promoted the phagocytosis of myeloma CD138 þ and/or CD38high bone marrow cells of multiple myeloma cells and inhibited the growth of ectopically growing myeloma patients.24 Here we report that nearly 100% of CD38 þ myeloma cells and patient primary myeloma cells growing in human cells express CD47 at a high level. These results establish CD47 as bone marrow microenvironment. To improve the anti-myeloma a potential therapeutic antibody target on myeloma cells, effect, therapies that combine other anti-myeloma antibodies or including the critical myeloma-initiating cell subset. Consistent other agents with anti-CD47 antibodies are currently under with this hypothesis, anti-CD47 antibody (B6H12) treatment led to investigation.

& 2012 Macmillan Publishers Limited Leukemia (2012) 2538 – 2545 Anti-myeloma effect of an anti-CD47 antibody D Kim et al 2544

Figure 6. Radiation-sensitive cells are important for the anti-myeloma effect of B6H12 in vivo.(a) Experimental scheme: 6–8-week-old female RAG2 À /gc À mice were irradiated with 400 rad X-ray. After 12 h, 2000 RPMI 8226 cells were transplanted subcutaneously into RAG2 À /gc À mice. (b) Fold change of tumor size in mice treated with control mouse IgG and B6H12 after 5-week treatment. The graph represents the sum of three independent experiments with similar results. (c, d). Reduction of Gr-1medMac-1low/ À cells and GR-1lowMac-1 þ cells in irradiated mice. Myeloid cell composition in bone marrow, spleen and liver was examined 10 days after irradiation **Po0.01. (e) Indicated cells were fractionated and stained with Wright-Giemsa stain ( Â 40). (f) The Gr-1 þ Mac-1low/ À cells (red box) are Ly6ChighCD31 þ / À monocytes described by De Bruijn et al.27

CONFLICT OF INTEREST REFERENCES ILW et al. ﬁled US. Patent Application Serial No. 12/321,215 entitled ‘Methods for 1 Yaccoby S, Barlogie B, Epstein J. Primary myeloma cells growing in SCID-hu mice: Manipulating Phagocytosis Mediated by CD47.’ The other authors declare no conﬂict a model for studying the biology and treatment of myeloma and its manifesta- of interest. tions. Blood 1998; 92: 2908–2913. 2 Kirshner J, Thulien KJ, Martin LD, Debes Marun C, Reiman T, Belch AR et al. A unique three-dimensional model for evaluating the impact of therapy on multiple myeloma. Blood 2008; 112: 2935–2945. ACKNOWLEDGEMENTS 3 Kyle RA, Rajkumar SV. Multiple myeloma. N Engl J Med 2004; 351: 1860–1873. We thank Dr Bruno C Medeiros for providing patient specimens; Libuse Jerabek, 4 Rajkumar SV. Treatment of relapsed or refractory multiple myeloma. In: Kyle RA, Theresa Storm and Adriane Mosley for laboratory and mouse management; Drs Basow DS (eds) UpToDate. UpToDate: Waltham, MA, 2010. Ingrid Ibarra and Yasuo Mori for their help with interpretation of bone marrow cell 5 Mawby WJ, Holmes CH, Anstee DJ, Spring FA, Tanner MJ. Isolation and morphology; Kipp Weiskopf for helpful comments on phagocytosis assay. In addition, characterization of CD47 glycoprotein: a multispanning membrane protein which we thank patients who consented to donate specimens. A part of this research was is the same as integrin-associated protein (IAP) and the ovarian tumour marker presented in the Lymphoma and Myeloma 2011 meeting. Dongkyoon Kim was OA3. Biochem J 1994; 304(Pt 2): 525–530. supported by the Irvington Institute Fellowship Program of the Cancer Research 6 Oldenborg PA, Zheleznyak A, Fang YF, Lagenaur CF, Gresham HD, Institute (initially by the Irvington Institute for Immunological Research and The Dana Lindberg FP. Role of CD47 as a marker of self on red blood cells. Science 2000; Foundation). This research was supported by co-sponsorship (SPO no.: 43710) of the 288: 2051–2054. Multiple Myeloma Research Foundation and the Leukemia Lymphoma Society, and 7 Oldenborg PA, Gresham HD, Lindberg FP. CD47-signal regulatory protein by the Ludwig Institute. Irving L Weissman is a Daniel K and Virginia Ludwig Professor alpha (SIRPalpha) regulates Fcgamma and complement receptor-mediated at Stanford. phagocytosis. J Exp Med 2001; 193: 855–862.

Leukemia (2012) 2538 – 2545 & 2012 Macmillan Publishers Limited Anti-myeloma effect of an anti-CD47 antibody D Kim et al 2545 8 Lagadec P, Dejoux O, Ticchioni M, Cottrez F, Johansen M, Brown EJ et al. 25 Chao MP, Alizadeh AA, Tang C, Jan M, Weissman-Tsukamoto R, Zhao F et al. Involvement of a CD47-dependent pathway in platelet adhesion on inflamed Therapeutic antibody targeting of CD47 eliminates human acute lymphoblastic vascular endothelium under flow. Blood 2003; 101: 4836–4843. leukemia. Cancer Res 2011; 71: 1374–1384. 9 Gresham HD, Goodwin JL, Allen PM, Anderson DC, Brown EJ. A novel member of 26 Kikuchi Y, Uno S, Kinoshita Y, Yoshimura Y, Iida S, Wakahara Y et al. Apoptosis the integrin receptor family mediates Arg-Gly-Asp-stimulated neutrophil phago- inducing bivalent single-chain antibody fragments against CD47 showed anti- cytosis. J Cell Biol 1989; 108: 1935–1943. tumor potency for multiple myeloma. Leuk Res 2005; 29: 445–450. 10 Brown E, Hooper L, Ho T, Gresham H. Integrin-associated protein: a 50-kD plasma 27 de Bruijn MF, Slieker WA, van der Loo JC, Voerman JS, van Ewijk W, Leenen PJ. membrane antigen physically and functionally associated with integrins. J Cell Biol Distinct mouse bone marrow macrophage precursors identified by differential 1990; 111(6 Pt 1): 2785–2794. expression of ER-MP12 and ER-MP20 antigens. Eur J Immunol 1994; 24: 2279–2284. 11 Jaiswal S, Jamieson CH, Pang WW, Park CY, Chao MP, Majeti R et al. CD47 is 28 Yamao T, Noguchi T, Takeuchi O, Nishiyama U, Morita H, Hagiwara T et al. upregulated on circulating hematopoietic stem cells and leukemia cells to avoid Negative regulation of platelet clearance and of the macrophage phagocytic phagocytosis. Cell 2009; 138: 271–285. response by the transmembrane glycoprotein SHPS-1. J Biol Chem 2002; 277: 12 Majeti R, Chao MP, Alizadeh AA, Pang WW, Jaiswal S, Gibbs Jr KD et al. CD47 is an 39833–39839. adverse prognostic factor and therapeutic antibody target on human acute 29 Okazawa H, Motegi S, Ohyama N, Ohnishi H, Tomizawa T, Kaneko Y et al. Negative myeloid leukemia stem cells. Cell 2009; 138: 286–299. regulation of phagocytosis in macrophages by the CD47-SHPS-1 system. 13 Chao MP, Alizadeh AA, Tang C, Myklebust JH, Varghese B, Gill S et al. Anti-CD47 J Immunol 2005; 174: 2004–2011. antibody synergizes with rituximab to promote phagocytosis and eradicate 30 Ide K, Wang H, Tahara H, Liu J, Wang X, Asahara T et al. Role for CD47-SIRPalpha non-Hodgkin lymphoma. Cell 2010; 142: 699–713. signaling in xenograft rejection by macrophages. Proc Natl Acad Sci USA 2007; 14 Chao MP, Tang C, Pachynski RK, Chin R, Majeti R, Weissman IL. Extranodal 104: 5062–5066. dissemination of non-Hodgkin lymphoma requires CD47 and is inhibited by 31 Kikuchi Y, Uno S, Yoshimura Y, Otabe K, Iida S, Oheda M et al. A bivalent single- anti-CD47 antibody therapy. Blood 2011; 118: 4890–4901. chain Fv fragment against CD47 induces apoptosis for leukemic cells. Biochem 15 Chan KS, Espinosa I, Chao M, Wong D, Ailles L, Diehn M et al. Identification, Biophys Res Commun 2004; 315: 912–918. molecular characterization, clinical prognosis, and therapeutic targeting of 32 Johansson U, Higginbottom K, Londei M. CD47 ligation induces a rapid caspase- human bladder tumor-initiating cells. Proc Natl Acad Sci USA 2009; 106: independent apoptosis-like cell death in human monocytes and dendritic cells. 14016–14021. Scand J Immunol 2004; 59: 40–49. 16 Willingham SB, Volkmer JP, Gentles AJ, Sahoo D, Dalerba P, Mitra SS et al. 33 Blazar BR, Lindberg FP, Ingulli E, Panoskaltsis-Mortari A, Oldenborg PA, Iizuka K et al. The CD47-signal regulatory protein alpha (SIRPa) interaction is a CD47 (integrin-associated protein) engagement of dendritic cell and macrophage therapeutic target for human solid tumors. Proc Natl Acad Sci USA 2012; 109: counterreceptors is required to prevent the clearance of donor lympho- 6662–6667. hematopoietic cells. JExpMed2001; 194: 541–549. 17 Campbell IG, Freemont PS, Foulkes W, Trowsdale J. An ovarian tumor marker with 34 Sanz-Rodriguez F, Hidalgo A, Teixido J. Chemokine stromal cell-derived factor- homology to vaccinia virus contains an IgV-like region and multiple transmem- 1alpha modulates VLA-4 integrin-mediated multiple myeloma cell adhesion to brane domains. Cancer Res 1992; 52: 5416–5420. CS-1/fibronectin and VCAM-1. Blood 2001; 97: 346–351. 18 Zhan F, Hardin J, Kordsmeier B, Bumm K, Zheng M, Tian E et al. Global gene 35 Gunn WG, Conley A, Deininger L, Olson SD, Prockop DJ, Gregory CA. A crosstalk expression profiling of multiple myeloma, monoclonal gammopathy of unde- between myeloma cells and marrow stromal cells stimulates production of DKK1 termined significance, and normal bone marrow plasma cells. Blood 2002; 99: and interleukin-6: a potential role in the development of lytic bone disease and 1745–1757. tumor progression in multiple myeloma. Stem Cells 2006; 24: 986–991. 19 McCune JM, Namikawa R, Kaneshima H, Shultz LD, Lieberman M, Weissman IL. 36 Epstein J, Yaccoby S. Consequences of interactions between the bone marrow The SCID-hu mouse: murine model for the analysis of human hematolymphoid stroma and myeloma. Hematol J 2003; 4: 310–314. differentiation and function. Science 1988; 241: 1632–1639. 37 Degrassi A, Hilbert DM, Rudikoff S, Anderson AO, Potter M, Coon HG. In vitro 20 Namikawa R, Ueda R, Kyoizumi S. Growth of human myeloid leukemias in the culture of primary plasmacytomas requires stromal cell feeder layers. Proc Natl human marrow environment of SCID-hu mice. Blood 1993; 82: 2526–2536. Acad Sci USA 1993; 90: 2060–2064. 21 Epstein J, Yaccoby S. The SCID-hu myeloma model. Methods Mol Med 2005; 113: 38 Damiano JS, Cress AE, Hazlehurst LA, Shtil AA, Dalton WS. Cell adhesion mediated 183–190. drug resistance (CAM-DR): role of integrins and resistance to apoptosis in human 22 Seegmiller AC, Xu Y, McKenna RW, Karandikar NJ. Immunophenotypic differ- myeloma cell lines. Blood 1999; 93: 1658–1667. entiation between neoplastic plasma cells in mature B-cell lymphoma vs plasma 39 Nefedova Y, Landowski TH, Dalton WS. Bone marrow stromal-derived soluble cell myeloma. Am J Clin Pathol 2007; 127: 176–181. factors and direct cell contact contribute to de novo drug resistance of myeloma 23 Adams J. The proteasome: a suitable antineoplastic target. Nat Rev Cancer 2004; 4: cells by distinct mechanisms. Leukemia 2003; 17: 1175–1182. 349–360. 40 Kukreja A, Radfar S, Sun BH, Insogna K, Dhodapkar MV. Dominant role of CD47- 24 Kim D, Park CY, Medeiros BC, Weissman IL. CD19-CD45low/-CD38high/CD138 þ thrombospondin-1 interactions in myeloma-induced fusion of human dendritic plasma cells enrich for human tumorigenic myeloma cells. Leukemia (in press). cells: implications for bone disease. Blood 2009; 114: 3413–3421.

Supplementary Information accompanies the paper on the Leukemia website (http://www.nature.com/leu)

& 2012 Macmillan Publishers Limited Leukemia (2012) 2538 – 2545