RESEARCH ARTICLE

Imaging of CD47 Expression in Xenograft and Allograft Tumor Models Alexander Zheleznyak, Oluwatayo F. Ikotun, Julie Dimitry, William A. Frazier, and Suzanne E. Lapi

Abstract CD47 functions as a marker of ‘‘self’’ by inhibiting phagocytosis of autologous cells. CD47 has been shown to be overexpressed by various tumor types as a means of escaping the antitumor immune response. The goal of this research was to investigate the utility of CD47 imaging using positron emission tomography (PET) in both human xenograft and murine allograft tumor models. Anti-CD47 antibodies were conjugated with p-isothiocyanatobenzyldesferrioxamine (Df-Bz-NCS) and labeled with 89Zr. We employed xenograft and allograft small-animal models of cancer in biodistribution and PET imaging studies to investigate the specificity and PET imaging robustness of CD47. Ab-Df-Bz-NCS conjugates were labeled with 89Zr with specific activity of 0.9 to 1.6 mCi/mg. Biodistribution studies in the xenograft and allograft model showed similar specific tumor uptake of the antihuman and antimouse CD47 antibodies. However, the tracer retention in the liver, spleen, and kidneys was significantly higher in the allograft-bearing animals, suggesting uptake mediated by the CD47 normally expressed throughout the reticular endothelial system. CD47, a marker of ‘‘self,’’ was evaluated as a diagnostic PET biomarker in xenograft and allograft cancer animal models. CD47 imaging is feasible, warranting further studies and immunoPET tracer development.

D47 (Figure 1A), also known as integrin-associated protein (IAP), is a widely expressed transmembrane protein characterized by its physical and functional association with multiple binding partners. Among them are thrombospondin-1 (TSP1); signal-regulatory protein alpha (SIRPa, CD172a); the integrin family members avb3, aIIbb3; and several beta1 and beta2 integrins.1 As a result, CD47 has been shown to participate in cellular processes such as apoptosis, proliferation, adhesion, and migration. In particular, CD47 functions as a marker of ‘‘self’’ by inhibiting phagocytosis of autologous cells through interaction with SIRPa expressed by professional phagocytes, such as macrophages (Figure 1B).2 SIRPa engagement by CD47 initiates a signaling cascade through an immunoreceptor tyrosine-based inhibitory motif (ITIM) present on the cytoplasmic portion of the SIRPa molecule that

C

From the Mallinckrodt Institute of Radiology and the Department of Biochemistry and Molecular Biophysics, Washington University School of Medicine, St. Louis, MO. Address reprint requests to: Suzanne E. Lapi, PhD, Mallinckrodt Institute of Radiology, Washington University School of Medicine, Campus Box 8225, 510 South Kingshighway Blvd, St. Louis, MO 63110; e-mail: lapis@ mir.wustl.edu.

DOI 10.2310/7290.2013.00069 #

2013 Decker Publishing.

results in the inhibition of phagocytosis. Inversely, loss of CD47 expression, for example, by aged or damaged cells, removes the inhibitory signal and leads to clearance of these cells by phagocytes. Consequently, blocking the interaction of CD47 with SIRPa, for example, with antiCD47 antibodies, removes the inhibitory signal and promotes phagocytosis. It has been shown that one of the mechanisms of cancer cell survival is the evasion of the immunosurveillance by phagocytic cells through overexpression of CD47.3–5 As early as 1992, Rosales and colleagues showed that myeloid cell lines HL-60 (human promyelocytic leukemia) and U937 (human histiocytic lymphoma) expressed up to fivefold more CD47 molecules than differentiated myeloid cells such as neutrophils and monocytes.6 More recently, CD47 overexpression was demonstrated on a number of hematologic malignancies,7,8 as well as solid tumors.3 Since a variety of tumors show increased expression of CD47 as a means of escaping phagocytosis, detecting such overexpression may be useful for cancer detection, diagnosis, and patient stratification for anti-CD47 therapies. Concurrently masking CD47 to promote phagocytosis of cancer cells and eliciting an adaptive immune response through direct antibodydependent cellular cytotoxicity (ADCC) may provide an effective therapeutic strategy.9–11 Furthermore, the magnitude of CD47 expression could be used as a prognostic

Molecular Imaging 2013: pp 1–10

1

2

Zheleznyak et al

Figure 1. A, Schematic representation of the structure of CD47. The extracellular region is a glycosylated IgV domain containing a conserved disulfide linker between C23 and C96.40 B, CD47 functions as a marker of ‘‘self’’ by activating the SIRPa phagocytosis-inhibiting signaling cascade.

marker for recurrence and metastasis in human breast cancer,12 as well as the aggressiveness of some leukemias.13 Together, the available data support the evaluation of CD47 as a target candidate for imaging and treatment of primary tumors and their metastatic dissemination. A number of monoclonal antibodies against both murine and human CD47 have been evaluated for their biochemical characteristics and ability to modulate the biological effects of CD47.1,14,15 B6H12 is a mouse monoclonal antibody that recognizes the extracellular domain of human CD47 and has a binding affinity (Kd) of 6 to 16 nM at 37uC.6 This antibody inhibits the migration and proliferation effects of CD47.16 Most importantly, B6H12 blocks SIRPa–CD47 binding and, therefore, inhibits the antiphagocytic signal transduced by this interaction.1 CD47 is not a biomarker unique to cancer but, rather, an overexpressed one. As a result, normally expressed CD47 will contribute to the signal obtained by positron emission tomography (PET)—a common problem in molecular imaging of overexpressed targets. Therefore, the background effect of normally expressed CD47 on the signal to noise characteristics of PET images of CD47-overexpressing tumors needs to be evaluated in an allograft tumor model. For this purpose, we selected a rat monoclonal antibody that binds to the extracellular domain of mouse CD47 (aM-CD47) with Kd of 2 nM (W.A. Frazier, unpublished data, 2011). The PET signal obtained with this antibody will be a combination of normally expressed in tissues and tumor-associated CD47. In addition, aM-CD47 cross-reacts with CD47 of human origin. This presents an intriguing opportunity for developing an experimental system that would allow for the evaluation of biomarkers in xenograft and allograft settings using the same tracer.

Zr (t1/2 5 78.4 hours; b+ 5 22.3%, c, 909 keV 5 99%) is a cyclotron-produced positron-emitting radionuclide with a 78.4-hour half-life. It is a member of a small group of radionuclides commonly used to label antibodies and their fragments: 124I (t1/2 5 100.2 hours; b+ 5 24%), 64 Cu (t1/2 5 12.7 hours; b+ 5 18%), and 18F (t1/2 5 1.83 hours; b+ 5 97%).17 By virtue of its relatively long halflife, 89Zr has emerged as a promising tool in antibody radiochemistry.18,19 In recent years, 89Zr has been successfully used to radiolabel trastuzumab (Herceptin, Genentech, South San Francisco, CA) for imaging of human epidermal growth factor receptor 2 (HER2),20,21 cetuximab (Erbitux, Ely Lilly, New York, NY) for imaging of epidermal growth factor receptor (EGFR),22 and bevacizumab (Avastin, Genentech) for imaging of vascular endothelial growth factor (VEGF).23 Together, available data provide strong evidence that 89Zr is the radionuclide of choice for pairing with an intact antibody. ImmunoPET technology, the use of radiolabeled antibodies as PET imaging agents, is a promising new tool in oncology.24–26 In recent years, efforts in the development of state-of–the-art PET/computed tomographic (CT) instrumentation, the production of chemically pure radionuclides with longer half-lives better matched to the pharmacodynamics of antibodies, and synthesis of improved chelates for radiometal chemistry have led to significant improvements in tracer sensitivity and hence image quality of these compounds. At the same time, more than 20 monoclonal antibodies have been approved for clinical use as anticancer agents, whereas over 200 new antibodies are being investigated for their therapeutic potential.27 Nanomolar affinities for antigens and welldeveloped methodology for generation and large-scale manufacturing of antibodies with various modifications, 89

3

Imaging of CD47

such as fragmentation, interspecies chimera construction, and partial or full humanization, make antibodies well suited to be used as imaging agents.28 Importantly, antibody-based radiopharmaceuticals allow single-patient determination of target abundance, thus providing the necessary information for therapy design and/or modifications in a clinical setting. Therefore, developing new cancer pathology–relevant target-tracer combinations is a worthwhile endeavor. Herein we describe an in vivo evaluation of CD47 as a ubiquitous cancer immunoPET imaging biomarker. We employed xenograft (human OV10 ovarian carcinoma) and allograft (murine B16F10 melanoma) models and complementary 89Zr-labeled anti-CD47 antibodies, B6H12 and aM-CD47, to investigate the specificity and imaging robustness of this tracer-target pair. These data complement the emerging new evidence, which illustrates the possibility of using CD47 as an immunotherapy target and supports future work to establish 89Zr-labeled anti-CD47 antibodies as radiopharmaceuticals for detection, staging, and therapy monitoring of various malignancies, as well as for optimization of patient selection to maximize the effectiveness of CD47 targeted therapies.

Materials and Methods

and is expressed as the avb3 integrin. The vector control that expresses avb3 without CD47 was also generated.29 The allogeneic tumor cell line, B16F10 murine melanoma, was used in combination with an antibody against murine CD47.30 4T1 murine mammary carcinoma and murine Lewis lung carcinoma were used in the immunophenotyping studies. Cell lines were maintained in complete media consisting of Iscove’s Modified Dulbecco’s Medium (IMDM) (Life Technologies, Grand Island, NY) supplemented with 10% fetal bovine serum and 50 mg/mL of gentamicin (Life Technologies). Doubly transfected OV10 cells were supplemented with 400 mg/mL of Geneticin and 100 mg/mL of Hygromycin, whereas single transfectants received either 400 mg/mL of Geneticin or 200 mg/mL of Hygromycin. All cells were routinely passaged by detachment with 5 mL of 0.05% trypsin/ethylenediamine tetraacetic acid (EDTA) (500 mg/mL trypsin, 200 mg/mL EDTA) (Life Technologies) and replating 1/100 of the suspension volume. B6H12 (IgG1) is a mouse monoclonal antibody against the extracellular domain of human CD47.6,31 aM-CD47 was raised against human placental CD47 in CD47-deficient mice.30 It competes with the antihuman monoclonal antibody B6H12 and thus shares a similar/overlapping epitope; however, this antibody also binds to rodent CD47 with an apparent Kd of approximately 2 nM (Frazier, unpublished data, 2011).30

Reagents All experiments involving the use of radioactive materials at Washington University are conducted under the authorization of the Radiation Safety Committee in accordance with the university’s Nuclear Regulatory Commission license. Activity was determined with a Capintec CRC-25 Dose Calibrator calibrated to a factor of 465 for 89Zr (Capintec, Ramsey, NJ). Cell lines were purchased from ATCC (Manassas, VA) unless specified otherwise. All chemicals were purchased from SigmaAldrich (St. Louis, MO) unless otherwise specified, and all solutions were prepared using ultrapure water with 18 MV-cm resistivity produced by a Millipore Integral 5 water purification system (Millipore, Billerica, MA). Cell Lines and Anti-CD47 Antibodies OV10 is a CD47-deficient clone of the human ovarian carcinoma parental cell line OVM1.16 This clone was stably transfected with the most commonly expressed CD47 splice variant 2 and a complementary deoxyribonucleic acid (cDNA) coding for the human CD61 (b3) monomer. The CD61 monomer forms a heterodimer with CD51 (av)

Animal Models Athymic nu/nu and C57Bl-6 5-week-old female mice were purchased from Charles River Laboratories (Wilmington, MA) and housed under pathogen-free conditions in an approved facility according to the guidelines of the Division of Comparative Medicine and the Animal Studies Committee of Washington University School of Medicine (WUMS). All animals were routinely acclimated to the WUMS animal facility environment prior to studies and fed ad libitum. Athymic mice were implanted subcutaneously (flank) with 5 3 106 CD47+ or 10 3 106 CD472 OV10 cells suspended in 100 mL of saline. C57Bl-6 mice (15–18 g) were implanted subcutaneously (flank) with 2 3 106 B16F10 (CD47+) murine melanoma cells suspended in 100 mL of saline. Animals developed tumors averaging 150 to 200 mm3 14 days from the time of implantation to the initiation of the imaging studies. 89 89

Zr Production and Purification

Zr was produced via the 89Y(p,n)89Zr reaction using the CS-15 cyclotron (Cyclotron Corporation, Berkeley, CA)

4

Zheleznyak et al

and separated via ion exchange chromatography as previously described,32,33 with a resulting specific activity of 22 to 134 mCi/mmol). Radiolabeling of Antibodies 89

Zr labeling was performed according to previously published methods.20 Briefly, anti-CD47 antibodies (. 5 mg/mL) were conjugated to isothiocyanatobenzyldesferrioxamine–NCS (Df-Bz-NCS) (Macrocyclics, Dallas, TX) maintained as 5 mg/mL stock in dimethyl sulfoxide. The antibody to chelator ratio was 1:20 in total reaction volume of 300 mL. The reaction was allowed to occur for 1 hour at 37uC in the presence of 0.1 M sodium carbonate buffer (pH 9). The unreacted chelate was removed by dialysis (10,000 MW cutoff mini dialysis units, Thermo Fisher Scientific, Rockford, IL) in three changes of 1 L phosphate-buffered saline (PBS). The Ab-Df-Bz-NCS conjugate was combined with 89Zr oxalate at a 1 mg/mCi ratio and a pH of 7.1. The mixture was incubated for 1 hour at 37uC. The labeling efficiency was confirmed by radio-TLC on C-18 plates with 50 mM diethylenetriaminepentaacetic acid (DTPA) eluent solvent and a Bioscan AR-2000 radioTLC scanner equipped with a 10% methane/argon gas supply. The 89Zr-labeled antibody remained at the baseline (Rf 5 0), whereas free 89Zr moved with the solvent front. The specific activity obtained was 0.9 to 1.6 mCi/mg with labeling efficiency of 98 to 100%; therefore, the compounds were used with no further purification.

time point). Briefly, 50 mCi 89Zr-Df-Bz-NCS-anti-CD47 antibodies were injected via the tail vein. Animals (n 5 3 for each condition) were sacrificed at selected time points postinjection, organs of interest were harvested and weighed, and the associated radioactivity was determined on a gamma-counter. After correcting for background and decay, the percent injected dose per gram (%ID/g) and percent injected dose per organ (%ID/organ) were calculated by comparison with a weighed, counted standard. Immunophenotyping OV10 and B16F10 cells were cultured in complete IMDM until < 75% confluent, harvested by mechanical dissociation with a rubber-tipped cell scraper (Sarstedt, Newton, NC), and washed once with complete IMDM. For each experimental condition, 1 3 106 cells were suspended in 50 mL of complete IMDM for each experimental condition, transferred to a V-bottomed well of a 96-well plate (Corning, New York, NY), and incubated with 1 mg/mL anti-CD47 antibodies for 1 hour at 4uC. The cells were washed three times with 200 mL of cold PBS and incubated with 50 mL of 1:100 dilution of secondary antihuman or antimouse IgG conjugated to fluorescein isothiocyanate fluorophore (FITC) (Santa Cruz Biotechnology, Santa Cruz, CA). After washing three times in 200 mL of cold PBS, the cell-associated fluorescence was detected with a Beckman Coulter Flow Cytometer (Becton Dickinson, Franklin Lakes, NJ).

In Vitro Cell Binding CD47+ and CD472 OV10 cells were grown to < 50% confluency on six-well tissue culture plates and incubated with 1 to 2 mCi (0.5 mg) of radiolabeled antibody in 500 mL of complete media for 35 minutes in humidified atmosphere supplemented with 5% CO2 at 37uC. Unbound antibody was removed by washing cells three times with PBS. Cells were then solubilized in PBS containing 0.1% sodium dodecyl sulfate, the lysate was transferred to microfuge tubes, and cell-associated activity was detected with a gamma-counter. Protein content was measured with the bicinchoninic acid assay (Thermo Fisher Scientific), and the data were expressed as counts per milligram of protein. In Vivo Biodistribution Biodistribution studies were conducted immediately after 3-, 5-, or 7-day PET/CT imaging studies (n 5 6 for each

Small-Animal PET/CT Imaging Prior to imaging, mice (n 5 3 for each condition) were injected intravenously (tail vein) with 40 to 70 mCi of either 89Zr-Df-Bz-NCS-B6H12 or 89Zr-Df-Bz-NCS-aMCD47. At 72, 120, or 168 hours, postinjection mice were anesthetized with 1 to 2% isoflurane and imaged with an Inveon small-animal PET/CT scanner (Siemens Medical Solutions, Knoxville, TN). Static images were collected for 20 minutes and reconstructed with the maximum a posteriori probability algorithm34 followed by coregistration with the Inveon Research Workstation (IRW) image display software (Siemens Medical Solutions). Regions of interest (ROI) were selected from PET images using CT anatomic guidelines, and the activity was determined with IRW software. Maximum standardized uptake values (SUVs) were determined as nCi/cc 3 animal weight/ injected dose.

Imaging of CD47

Statistical Analysis Data are expressed as mean 6 SD unless noted otherwise. Statistical significance was determined with unpaired, twotailed Student t-test and 95% confidence level, and p values less than .05 were considered significant. All data were analyzed and plotted using GraphPad Prism version 5.0 (GraphPad Software, La Jolla, CA).

Results Anti-CD47 Antibodies Are Efficiently Coupled to Df-Bz-NCS and Labeled with 89Zr Anti-CD47 antibodies were conjugated to Df-Bz-NCS as described in the Methods section. After purification, the Ab-Df-Bz conjugate was labeled with 89Zr, resulting in a specific activity of 0.9 to 1.6 mCi/mg and radiochemical purity of 98.2 6 1.3%. 89

Zr-Df-Bz-NCS-B6H12 Retained Specificity for CD47

To establish the conjugated antibodies that retained specificity for CD47, binding studies were conducted. CD47+ and CD472 OV10 human ovarian carcinoma clones were immunophenotyped to confirm either the presence or absence of CD47. As detailed in the Methods section, the cells were harvested and stained, in succession, with B6H12 and a FITC-conjugated goat antimouse

Figure 2. A, Immunophenotyping of CD47+ and CD472 OV10 cells with B6H12 antibody demonstrates cell surface expression of CD47. B, In vitro cell binding of 89Zr-B6H12 to CD47+ and CD472 OV10 cells demonstrates a 16-fold increase in activity associated with the CD47+ cells.

5

secondary antibody. The cell-associated fluorescence, indicative of the expression of CD47, was analyzed with a flow cytometer. As expected, CD47 expression was limited to the CD47+ clone (Figure 2A). When CD47+ and CD472 clones were incubated with 2 mCi (1.25–2 mg) of radiolabeled B6H12, approximately 16-fold excess of 89ZrDf-Bz-NCS-B6H12 (687,955 6 47,295 CPM/mg of protein vs 42,266 6 12,331 CPM/mg of protein) was associated with CD47+ cells (Figure 2B). Therefore, 89Zr-labeled antibody-chelator conjugate retained specificity for CD47. Biodistribution of 89Zr-Df-Bz-NCS-B6H12 Demonstrated Significant Tracer Accumulation in Xenograft Tumor Tissues The in vivo pharmacokinetic characteristics and tumortargeting efficiency of 89Zr-Df-Bz-NCS-B6H12 were investigated by performing biodistribution studies in mice bearing OV10 subcutaneous xenografts 168 hours after tracer administration. OV10 cells positive and negative for CD47 were implanted subcutaneously into the flanks of athymic mice and allowed to form small tumors. 89Zr-DfBz-NCS-B6H12 was prepared as described in the Methods section and administered systemically through the tail vein at 50 mCi per animal (25–40 mg of antibody). The results of these studies are summarized in Figure 3A. Analysis of tumor-associated activity revealed a significant 2.8 6 0.8-fold (mean 6 SEM) increase in accumulation of 89ZrDf-Bz-NCS-B6H12 in the CD47+ tumor tissue compared to the CD472 tissue, 4.92 6 0.48 %ID/g and 2.08 6 0.6 %ID/g (mean 6 SEM), respectively (p , .05). Nontarget and clearance organs, including the liver, lung, spleen, and kidney, retained similar amounts of the tracer in animals bearing both CD47+ and CD472 xenografts. In animals bearing CD472 xenografts, comparison of tumor-associated activity to that remaining in circulation revealed that the activity associated with tumors can be attributed to the tracer present in circulation, 2.08 6 0.85 %ID/g and 2.7 6 0.67 %ID/g, respectively. As expected, a relatively high amount of tracer was associated with the liver, 4.77 6 1.22 %ID/g (CD472) and 4.48 6 0.57 %ID/g (CD47+), the major clearance organ for antibody-based radiopharmaceuticals. In animals bearing CD47+ or CD472 tumors, activity detected in lung, spleen, and kidney, 1.7 6 0.3 %ID/g (CD47+)/1.7 6 0.4 %ID/g (CD472), 2.5 6 0.5 %ID/g (CD47+)/2.5 6 0.4 %ID/g (CD472), and 1.9 6 0.3 %ID/g (CD47+)/1.6 6 0.3 (CD472), respectively, was similar to that present in circulation, 2.5 6 0.4 %ID/g. In addition, a significant amount of the activity was associated with the bone tissue of animals bearing CD47+ tumors, 1.2 6 0.07 %ID/g. Our

6

Zheleznyak et al

Figure 3. A, Distribution of 89Zr-B6H12 radiotracer in athymic mice bearing CD47+ and CD472 xenograft tumors 168 hours after tracer administration shows that the activity accumulation in blood, liver, and kidneys is consistent with previously reported patterns of antibody-based tracer distribution and excretion and a 2.36-fold excess of 89Zr-B6H12 in tumor compared to circulation. B, Whole-body maximum-intensity projections of athymic mice implanted with CD47+ and CD472 tumors. Animals were injected with 50 mCi of 89Zr-B6H12 and imaged 72 and 168 hours later. Images show specific accumulation of the 89Zr-B6H12 in CD47+ tumor at days 3 and 7 after administration. C, Quantification of tumor-associated antibody shows two- to fourfold higher levels in CD47+ tumors.

data suggest that 89Zr-radiolabeled anti-CD47 antibody B6H12 specifically targets CD47 tumor xenografts and demonstrates predictable distribution to nontarget and clearance organs, thus supporting the feasibility of using CD47 as an imaging target. Quantification of Small-Animal PET/CT Images Shows Significant 89Zr-Df-Bz-NCS-B6H12 Accumulation in the Xenograft Tumor Tissue Small-animal PET/CT imaging was performed to evaluate the ability of 89Zr-Df-Bz-NCS-B6H12 to image tumor xenografts in a specific manner. For this purpose, a cohort of athymic mice was implanted with CD47+ or CD472 OV10 cells. When tumors reached an average volume of 200 mm3, animals were injected with 40 to 70 mCi (< 50 mg) of 89Zr-Df-Bz-NCS-B6H12 via the tail vein. Imaging studies were conducted 72 and 168 hours following tracer administration. Maximum-intensity projections (MIPs) of representative PET/CT images demonstrated high tumor to background contrast (Figure 3B). Interestingly, detectable amounts of activity were present in the periarticular bone tissue of both CD47+ and CD472 xenograft-bearing mice, likely due to the apparent bone tropism of free 89Zr following demetallation of the DFO complex. Activity associated with the liver was consistent with previously reported patterns of antibody-based tracer distribution and clearance.19,20,35 Quantification of the images presented in Figure 3C showed that CD47+ tumors yielded an SUV of 2.87 6 0.1 at 72 hours after tracer administration, which consequently increased to 8.02 6 0.53 (mean 6 SEM) after 168 hours. CD472 tumors had an SUV of 1.12 6 0.24 and 1.94 6 0.78 (mean 6 SEM) at 72 and 168 hours. Hence, CD47+ tumors retained significantly more radiolabeled antibody compared to CD472 tumors at both time points (p , .05

at 72 hours, p , .05 at 168 hours). Taken together, these data demonstrate effective imaging characteristics of 89Zrlabeled anti-CD47 immunoPET tracers for imaging of CD47-overexpressing malignancies. Biodistribution of 89Zr-Df-Bz-NCS-aM-CD47 Demonstrated Specific Tracer Accumulation in Allograft Tumor Tissues To evaluate a more realistic scenario in which substantial background expression of CD47 would contribute to the PET image, the in vivo target and nontarget/clearance organ distribution of 89Zr-Df-Bz-NCS-aM-CD47 was evaluated in B16F10 allograft–bearing mice. One of the challenges to investigating the biodistribution of monoclonal antibodies in mouse and other animal models is the lack of human protein in normal tissues. This model allowed examination of the tracer’s systemic distribution and uptake by the target receptor–overexpressing tumor in the context of endogenously expressed receptor. Immunophenotyping of B16F10 murine melanoma revealed that its expression of CD47 was about 1.5 times higher than that of the OV10 human ovarian carcinoma (Figure 4).

Figure 4. Immunophenotyping of selected murine tumor cell lines with aM-CD47 antibody.

Imaging of CD47

7

in the liver (p , .05), spleen (p , .05), and kidneys (p , .05) of tumor-bearing animals was significantly higher in the allograft-bearing animals, suggesting uptake mediated by the CD47 normally expressed throughout the reticular endothelial system (RES). Quantification of Preclinical PET Images of CD47 Allograft Model Demonstrated Altered Tumor Uptake Characteristics to Those of the Xenograft Model

Figure 5. A, Distribution of 89Zr-aM-CD47 radiotracer in C57Bl-6 mice bearing B16F10 allograft tumors 72 and 168 hours after tracer administration. B, Tumor to blood ratios calculated from the biodistribution data presented in (A) show progressive increase in tracer retention.

Allograft-bearing mice were injected with 50 mCi 89Zr-DfBz-NCS-aM-CD47 antibody via the tail vein and sacrificed at 72 or 168 hours. The biodistribution of 89Zr-Df-BzNCS-aM-CD47 is presented in Figure 5. The tracer accumulated in the clearance organs, with the liver retaining 19.8 6 3.2 %ID/g and 15.4 6 2.5 %ID/g at 72 hours and 168 hours after administration, respectively, whereas the kidneys retained 8.7 6 0.7 %ID/g and 15.9 6 2.2 %ID/g at 72 hours and 172 hours after administration, respectively. The spleen and bone retained significant amounts of the tracer at both time points, 12.5 6 5.1 %ID/g (72 hours) and 21.6 6 3.9 %ID/g (168 hours) for the spleen and 2.5 6 0.7 %ID/g (72 hours) and 3.0 6 0.9%ID/g (168 hours) for the bone, which may be indicative of an immune response to the Fc portion of the antibody molecule (see Figure 5A).36 The tracer activity in the tumor was higher than that in the circulation, with tumor to blood ratios of 1.67 6 0.19 at 72 hours and 2.91 6 1.01 at 168 hours, showing a 1.7-fold increase in retention with time (see Figure 5B). Of note, the tumorassociated activity decreased from 0.76 6 0.35 at 72 hours postinjection to 0.37 6 0.17 at 168 hours postinjection. Overall, these data suggest that the 89Zr-Df-Bz-NCS-aMCD47 distribution in tissues of the allograft model was altered to that of the xenograft model. The tracer retention

To determine whether the endogenous CD47 interferes with the in vivo tumor imaging efficiency of 89Zr-conjugated antibody, B16F10 murine melanoma–expressing murine CD47 (see Figure 4) was used to generate subcutaneous tumor allografts in C57Bl-6 mice. Animals were injected with 89 Zr-Df-Bz-NCS-aM-CD47 via the tail vein and whole body 20 minute static PET/CT images were obtained 72 hours later. MIP image reconstruction (Figure 6A) showed tumor to background contrast similar to that of the CD47+ OV10 xenograft images. SUV quantification demonstrated that 89 Zr-Df-Bz-NCS-aM-CD47 was retained similarly in the tumor allograft and xenograft models, with an SUV of 2.36 6 0.61 and 2.87 6 0.1 (mean 6 SEM) SUV in allograft and xenograft, respectively (p 5 .515) (Figure 6B). Comparison of mean tumor to muscle ratios demonstrated no significant difference between allograft and xenograft tumors, 4.507 6 1.18 and 4.782 6 0.6224 (mean 6 SEM), respectively (p 5 .8609) (Figure 6C). Overall, these data suggest that the normally expressed CD47 does not significantly affect the tumor uptake quantification obtained with 89Zr-labeled antiCD47 alloantibody. However, although the tumor to muscle ratio in the xenograft and allograft models was similar, substantial uptake in the liver, spleen, and kidneys might affect the unmodified tracer’s diagnostic value for tumors localized in or around these organs.

Discussion Herein we examine the feasibility of using CD47 as an immunoPET imaging target with potential applications in prescreening, diagnostics, staging, and therapy of hematologic and solid tumor malignancies. CD47 is a widely expressed transmembrane receptor belonging to the Ig superfamily. Its direct interaction with SIRPa expressed by myeloid cells results in inhibition of phagocytic clearance of autologous cells by professional phagocytes.37 Overexpression of CD47 by various cancer cell types has been demonstrated to be a survival strategy used by poorly differentiated cancer cells aimed at phagocytosis avoidance.9,37 To date, a number of

8

Zheleznyak et al

Figure 6. A, Whole-body maximum-intensity projections (MIPs) of C57Bl-6 mice implanted with B16F10 tumors. Animals were injected with 40 to 70 mCi 89Zr-aM-CD47 and imaged 72 hours later. Images show specific accumulation of the radiolabeled antibody. B, Mice bearing B16F10 allografts imaged with 89Zr-anti-mouse CD47 showed about 4.5-fold excess of the tracer associated with the tumor. C, Comparison of tumor to muscle standardized uptake value (SUV) ratios obtained from xenograft and allograft MIP demonstrates no significant difference.

malignancies have been shown to be responsive to treatment with an anti-CD47 antibody, B6H12.3,5,7–10 To our knowledge, this study is the first attempt to evaluate CD47 as a diagnostic biomarker. Binding of immobilized Ad22, B6H12, or 1F7 monoclonal antibodies to CD47 has been shown to induce caspase-independent apoptosis of several leukemia cell lines.38 This type of apoptosis does not require activation of the caspase cascade; instead, it involves the disruption of the mitochondrial membrane potential (DY) and the production of reactive oxygen species. Of the anti-CD47 antibodies, B6H12 has been shown to block CD47 on human cells, resulting in phagocytosis of the CD47expressing cells by macrophages.2 Moreover, a number of hematologic and solid tumors were shown to overexpress CD47 as a means of phagocytosis avoidance, suggesting that blocking tumor cells’ CD47 with B6H12 may cause tumor regression through enhanced phagocytosis by tissue macrophages. Thus, B6H12 would appear to be a good candidate to explore the suitability of CD47 as a cancer immunoPET biomarker and for characterizing potential CD47 targeted therapies.3,39 89 Zr-DFO-B6H12 binds specifically to the ovarian carcinoma cell line OV10. Biodistribution studies in xenograft animal models demonstrated specific uptake in the tumor tissue, whereas significant liver uptake could be attributed to the clearance pattern of intact antibody tracers described in a number of studies.18–20 The tracer uptake observed in nontarget clearance organs (spleen and kidney) is similar to that of the blood, suggesting that the detected activity may be attributed to the circulation of 89Zr-DFO-B6H12. Of note, 89Zr-DFO-B6H12 was taken up by the bone tissue with apparent specificity (see Figure 3A). In their excellent work, Holland and

colleagues addressed the issue of tissue tropism exhibited by 89Zr-chloride, 89Zr-oxalate, and 89Zr-DFO in a xenograft model.18 From their data, it was evident that 89 Zr-chloride preferentially localized to the liver and 89Zroxalate to the periarticular area of the extremities, and 89 Zr-DFO was cleared through the kidneys. However, little is known about the 89Zr-labeled tracer metabolism in vivo, and although it has been suggested that the 89Zr species might be absorbed by the bone matrix, it is possible that, additionally, in cases of intact ‘‘unmodified’’ antibodies, a significant amount of the tracer is taken up by lymphocytes of the RES, including liver, spleen, and bone marrow, expressing Fc receptor family members.36 The accumulation in the bone is likely the result of a combination of these factors. The specificity of 89Zr-DFO-B6H12 was also evident from the preclinical PET imaging studies performed where SUV for the CD47+ positive tumor tissues was two- to fourfold higher than that for the CD472 tumor tissues. Although the data obtained from the xenograft studies suggested that 89Zr-DFO-B6H12 would provide the required sensitivity for imaging of CD47, this is not optimal for the evaluation of an overexpressed, not a unique, target. Even with sequence modifications designed to minimize the tracer recognition by the recipient’s Fc receptors, the endogenous target would create some degree of systemic uptake. To evaluate the contribution of systemic uptake, we employed an allograft animal model combined with an antibody against murine CD47. As can be seen from the biodistribution data, the B16F10 allograft retained twofold more 89Zr-DFO-anti-mouse CD47 than was present in the circulation 78 and 168 hours after the dose administration. However, a significant amount of 89 Zr-DFO-antimouse CD47 was taken up by the liver,

Imaging of CD47

spleen, kidneys, and bone, indicating potential uptake by the cells of RES, which would lead to difficulties with image quantification in these areas. At the same time, although the endogenous CD47 expression was shown to be present in all tissues, little uptake was seen outside the RES organs and the tumor. This finding was reflected in the data obtained from preclinical PET imaging studies, where the MIP showed significant uptake in the RES and tumor tissues. These data suggest that tracer modification, such as removal of the Fc portion of the antibody, may be required to improve the imaging characteristics of B6H12. However, this modification may potentially diminish the ability of B6H12 to elicit an antitumor immune response. Overall, although the xenograft model demonstrated good selectivity of the tracer, the allograft model showed the necessity for tracer modification.

Conclusion We successfully labeled antihuman and antimouse antibodies against CD47 with 89Zr and employed them to evaluate CD47 as a PET imaging target in xenograft and allograft cancer animal models. PET image quantification demonstrated specific and similar tracer uptake in both xenograft and allograft tumors overexpressing CD47. Overall, CD47 imaging is feasible, warranting further studies and immunoPET tracer development.

Acknowledgments We would like to thank the Washington University isotope production team for routine production of 89Zr and the smallanimal imaging facility for generation of PET/CT data. Financial disclosure of authors: The production of 89Zr is supported by Department of Energy, Office of Sciences, Nuclear Physics, Isotope Program, under grant DESC0008657. William F. Frazier is a member of the board of Vasculox. Financial disclosure of reviewers: None reported.

References 1. Brown EJ, Frazier WA. Integrin-associated protein (CD47) and its ligands. Trends Cell Biol 2001;11:130–5, doi:10.1016/S0962-89 24(00)01906-1. 2. Oldenborg PA, Zheleznyak A, Fang YF, et al. Role of CD47 as a marker of self on red blood cells. Science 2000;288:2051–4, doi:10.1126/science.288.5473.2051. 3. Willingham SB, Volkmer JP, Gentles AJ, et al. The CD47-signal regulatory protein alpha (SIRPa) interaction is a therapeutic target for human solid tumors. Proc Natl Acad Sci U S A 2012;109:6662– 7, doi:10.1073/pnas.1121623109.

9

4. Edris B, Weiskopf K, Volkmer AK, et al. Antibody therapy targeting the CD47 protein is effective in a model of aggressive metastatic leiomyosarcoma. Proc Natl Acad Sci U S A 2012;109: 6656–61, doi:10.1073/pnas.1121629109. 5. Chao MP, Tang C, Pachynski RK, et al. Extranodal dissemination of non-Hodgkin lymphoma requires CD47 and is inhibited by anti-CD47 antibody therapy. Blood 2011;118:4890–901, doi:10. 1182/blood-2011-02-338020. 6. Rosales C, Gresham HD, Brown EJ. Expression of the 50-kDa integrin-associated protein on myeloid cells and erythrocytes. J Immunol 1992;149:2759–64. 7. Chao MP, Alizadeh AA, Tang C, et al. Therapeutic antibody targeting of CD47 eliminates human acute lymphoblastic leukemia. Cancer Res 2011;71:1374–84, doi:10.1158/0008-5472.CAN10-2238. 8. Jaiswal S, Jamieson CH, Pang WW, et al. CD47 is upregulated on circulating hematopoietic stem cells and leukemia cells to avoid phagocytosis. Cell 2009;138:271–85, doi:10.1016/j.cell.2009.05.046. 9. Chao MP, Weissman IL, Majeti R. The CD47-SIRPalpha pathway in cancer immune evasion and potential therapeutic implications. Curr Opin Immunol 2012;24:225–32, doi:10.1016/j.coi.2012.01.010. 10. Tseng D, Volkmer JP, Willingham SB, et al. Anti-CD47 antibodymediated phagocytosis of cancer by macrophages primes an effective antitumor T-cell response. Proc Natl Acad Sci U S A 2013; 110:11103–8, doi:10.1073/pnas.1305569110. 11. Zhao XW, Kuijpers TW, van den Berg TK. Is targeting of CD47SIRPalpha enough for treating hematopoietic malignancy? Blood 2012;119:4333–4; author reply 4334–5, doi:10.1182/blood-201111-391367. 12. Nagahara M, Mimori K, Kataoka A, et al. Correlated expression of CD47 and SIRPA in bone marrow and in peripheral blood predicts recurrence in breast cancer patients. Clin Cancer Res 2010;16: 4625–35, doi:10.1158/1078-0432.CCR-10-0349. 13. Majeti R, Chao MP, Alizadeh AA, et al. CD47 is an adverse prognostic factor and therapeutic antibody target on human acute myeloid leukemia stem cells. Cell 2009;138:286–99, doi:10.1016/ j.cell.2009.05.045. 14. Manna PP, Frazier WA. The mechanism of CD47-dependent killing of T cells: heterotrimeric Gi-dependent inhibition of protein kinase A. J Immunol 2003;170:3544–53. 15. Barazi HO, Li Z, Cashel JA, et al. Regulation of integrin function by CD47 ligands. Differential effects on alpha vbeta 3 and alpha 4beta1 integrin-mediated adhesion. J Biol Chem 2002;277:42859– 66, doi:10.1074/jbc.M206849200. 16. Lindberg FP, Gresham HD, Reinhold MI, et al. Integrin-associated protein immunoglobulin domain is necessary for efficient vitronectin bead binding. J Cell Biol 1996;134:1313–22, doi:10. 1083/jcb.134.5.1313. 17. Ikotun OF, Lapi SE. The rise of metal radionuclides in medical imaging: copper-64, zirconium-89 and yttrium-86. Future Med Chem 2011;3:599–621, doi:10.4155/fmc.11.14. 18. Holland JP, Divilov V, Bander NH, et al. 89Zr-DFO-J591 for immunoPET of prostate-specific membrane antigen expression in vivo. J Nucl Med 2010;51:1293–300, doi:10.2967/jnumed.110. 076174. 19. Borjesson PK, Jauw YW, Boellaard R, et al. Performance of immuno-positron emission tomography with zirconium-89labeled chimeric monoclonal antibody U36 in the detection of lymph node metastases in head and neck cancer patients. Clin

10

20.

21.

22.

23.

24.

25. 26.

27.

28.

29.

30.

Zheleznyak et al

Cancer Res 2006;12(7 Pt 1):2133–40, doi:10.1158/1078-0432.CCR05-2137. Holland JP, Caldas-Lopes E, Divilov V, et al. Measuring the pharmacodynamic effects of a novel Hsp90 inhibitor on HER2/neu expression in mice using Zr-DFO-trastuzumab. PLoS One 2010;5: e8859, doi:10.1371/journal.pone.0008859. Chang AJ, DeSilva R, Jain S, et al. 89Zr-radiolabled trastuzumab imaging in orthotropic and metastatic breast tumors. Pharmaceuticals 2012;5:79–83, doi:10.3390/ph5010079. Yeh HH, Ogawa K, Balatoni J, et al. Molecular imaging of active mutant L858R EGF receptor (EGFR) kinase-expressing nonsmall cell lung carcinomas using PET/CT. Proc Natl Acad Sci U S A 2011;108:1603–8, doi:10.1073/pnas.1010744108. Nagengast WB, de Korte MA, Oude Munnink TH, et al. 89Zrbevacizumab PET of early antiangiogenic tumor response to treatment with HSP90 inhibitor NVP-AUY922. J Nucl Med 2010; 51:761–7, doi:10.2967/jnumed.109.071043. van Dongen GA, Vosjan MJ. Immuno-positron emission tomography: shedding light on clinical antibody therapy. Cancer Biother Radiopharm 2010;25:375–85, doi:10.1089/cbr.2010.0812. Wu AM, Olafsen T. Antibodies for molecular imaging of cancer. Cancer J 2008;14:191–7, doi:10.1097/PPO.0b013e31817b07ae. Deri MA, Zeglis BM, Francesconi LC, et al. PET imaging with (8)(9)Zr: from radiochemistry to the clinic. Nucl Med Biol 2013; 40:3–14, doi:10.1016/j.nucmedbio.2012.08.004. Reichert JM. Monoclonal antibodies as innovative therapeutics. Curr Pharm Biotechnol 2008;9:423–30, doi:10.2174/1389201087 86786358. Reddy S, Robinson MK. Immuno-positron emission tomography in cancer models. Semin Nucl Med 2010;40:182–9, doi:10.1053/ j.semnuclmed.2009.12.004. Lindberg FP, Gresham HD, Schwarz E, et al. Molecular cloning of integrin-associated protein: an immunoglobulin family member with multiple membrane-spanning domains implicated in alpha v beta 3-dependent ligand binding. J Cell Biol 1993;123:485–96, doi:10.1083/jcb.123.2.485. Han X, Sterling H, Chen Y, et al. CD47, a ligand for the macrophage fusion receptor, participates in macrophage

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

multinucleation. J Biol Chem 2000;275:37984–92, doi:10.1074/ jbc.M002334200. Gresham HD, Goodwin JL, Allen PM, et al. A novel member of the integrin receptor family mediates Arg-Gly-Asp-stimulated neutrophil phagocytosis. J Cell Biol 1989;108:1935–43, doi:10.1083/ jcb.108.5.1935. Verel I, Visser GW, Boellaard R, et al. 89Zr immuno-PET: comprehensive procedures for the production of 89Zr-labeled monoclonal antibodies. J Nucl Med 2003;44:1271–81. Holland JP, Sheh Y, Lewis JS. Standardized methods for the production of high specific-activity zirconium-89. Nucl Med Biol 2009;36:729–39, doi:10.1016/j.nucmedbio.2009.05.007. Qi J, Leahy RM. Resolution and noise properties of MAP reconstruction for fully 3-D PET. IEEE Trans Med Imaging 2000;19:493–506, doi:10.1109/42.870259. Munnink THO, de Korte MA, Nagengast WB, et al. Zr-89trastuzumab PET visualises HER2 downregulation by the HSP90 inhibitor NVP-AUY922 in a human tumour xenograft. Eur J Cancer 2010;46:678–84, doi:10.1016/j.ejca.2009.12.009. Bruhns P. Properties of mouse and human IgG receptors and their contribution to disease models. Blood 2012;119:5640–9, doi:10. 1182/blood-2012-01-380121. Sick E, Jeanne A, Schneider C, et al. CD47 update: a multifaceted actor in the tumour microenvironment of potential therapeutic interest. Br J Pharmacol 2012;167:1415–30, doi:10.1111/j.14765381.2012.02099.x. Roue G, Bitton N, Yuste VJ, et al. Mitochondrial dysfunction in CD47-mediated caspase-independent cell death: ROS production in the absence of cytochrome c and AIF release. Biochimie 2003;85: 741–6, doi:10.1016/S0300-9084(03)00129-9. Hongrapipat J, Kopeckova P, Liu J, et al. Combination chemotherapy and photodynamic therapy with fab9 fragment targeted HPMA copolymer conjugates in human ovarian carcinoma cells. Mol Pharm 2008;5:696–709, doi:10.1021/mp800006e. Hatherley D, Graham SC, Turner J, et al. Paired receptor specificity explained by structures of signal regulatory proteins alone and complexed with CD47. Mol Cell 2008;31:266–77, doi:10.1016/ j.molcel.2008.05.026.