Pre-Clinical Development of a Humanized Anti-CD47 Antibody with Anti-Cancer Therapeutic Potential

Abstract

CD47 is a widely expressed cell surface protein that functions as a regulator of phagocytosis mediated by cells of the innate immune system, such as macrophages and dendritic cells. CD47 serves as the ligand for a receptor on these innate immune cells, SIRP-alpha, which in turn delivers an inhibitory signal for phagocytosis. We previously found increased expression of CD47 on primary human acute myeloid leukemia (AML) stem cells, and demonstrated that blocking monoclonal antibodies directed against CD47 enabled the phagocytosis and elimination of AML, non-Hodgkin's lymphoma (NHL), and many solid tumors in xenograft models. Here, we report the development of a humanized anti-CD47 antibody with potent efficacy and favorable toxicokinetic properties as a candidate therapeutic. A novel monoclonal anti-human CD47 antibody, 5F9, was generated, and antibody humanization was carried out by grafting its complementarity determining regions (CDRs) onto a human IgG4 format. The resulting humanized 5F9 antibody (Hu5F9-G4) bound monomeric human CD47 with an 8 nM affinity. Hu5F9-G4 induced potent macrophage-mediated phagocytosis of primary human AML cells in vitro and completely eradicated human AML in vivo, leading to long-term disease-free survival of patient-derived xenografts. Moreover, Hu5F9-G4 synergized with rituximab to eliminate NHL engraftment and cure xenografted mice. Finally, toxicokinetic studies in non-human primates showed that Hu5F9-G4 could be safely administered intravenously at doses able to achieve potentially therapeutic serum levels. Thus, Hu5F9-G4 is actively being developed for and has been entered into clinical trials in patients with AML and solid tumors (ClinicalTrials.gov identifier: NCT02216409).

Conflict of interest statement

Competing Interests: JL, RM, and ILW have filed International Patent Application WO 2011/143624 A2 entitled ‘Humanized and chimeric monoclonal antibodies to CD47’. JL, RM, ILW, SW, JV, MH, and SP have filed U.S. Patent Application number PCT/US2014/018743 entitled ‘Methods for achieving therapeutically effective doses of anti-CD47 agents’. This does not alter the authors' adherence to PLOS ONE policies on sharing data and materials.

Figures

Fig 1. Cloning, humanization, and characterization of anti-human CD47 monoclonal antibody 5F9.

(A-B) Comparison of mouse and humanized 5F9 variable heavy (A) and light (B) regions. CDRs are marked as indicated. (C) Rat YB2/0 cells stably transfected with human CD47 and parental YB2/0 cells were stained with human IgG4 isotype control or Hu5F9-G4 and analyzed for surface binding by flow cytometry. (D) Biotinylated mouse 5F9 binding to human CD47/mFc was detected by ELISA in the absence or presence of increasing concentrations of unlabeled Hu5F9-G4 or isotype control. Each sample was assayed in duplicate and SEM values are presented. Indicated p-values were determined by Two-Way ANOVA in comparison. (E) Binding of Hu5F9-G4 to monomeric or dimeric CD47 was analyzed by surface plasmon resonance yielding the indicated traces and binding constants. Data were processed by subtracting responses from the reference surface as well as an average of buffer injections. The responses were globally fit to a 1:1 interaction model including a term for mass transport. The number in parentheses represents the standard error in the last reported digit.

Fig 2. Hu5F9-G4 induces potent macrophage-mediated phagocytosis of AML.

(A) Recombinant SIRPα/human Fc fusion was coated in 96-well plates and CD47/mouse Fc fusion protein was added either in the absence or presence of an equal amount, 5, 10, 25, 50 or 100-times more Hu5F9-G4. The binding activity of CD47 to SIRPα was measured by an HRP-conjugated anti-mouse Fc specific secondary antibody. Each sample was performed in triplicate and SEM values are presented. (B) HL-60 and seven primary human AML cells were labeled with CFSE and incubated with human peripheral blood-derived macrophages in the presence of 10 ug/ml Hu5F9-G4 or isotype control. Two hours later, macrophages were imaged by fluorescence microscopy to determine the phagocytic index (number of target cells ingested per 100 macrophages) in duplicate. Lines indicate mean values and the indicated p-value was determined using a two-sided t-test. (C) The relationship between CD47 receptor occupancy and phagocytosis was determined with primary human AML cells. AML cells were first incubated with increasing concentrations of unlabeled Hu5F9-G4 (0.01–10 ug/ml), and after 2 hours, free CD47 receptor was measured using a saturating concentration (40 ug/ml) of Alexa 488-labeled Hu5F9-G4 followed by flow cytometry. In parallel, the AML cells were evaluated for susceptibility to phagocytosis via addition of human peripheral blood-derived macrophages and determination of the phagocytic index for each condition. (D) HL-60 cells were incubated with increasing concentrations of the indicated isotype control antibodies, chimeric B6H12-IgG1 as a positive control, or Hu5F9-G4, and ADCC activity was determined with human PBMC effector cells. The experiment was repeated 3 times. *** p

Fig 3. Hu5F9-G4 eradicates primary human AML and synergizes with rituximab to eliminate lymphoma.

(A) Primary human AML cells (5e6) from samples SU028 and SU048 were engrafted into 10 NSG mice per the indicated scheme. Engrafted mice were assigned to treatment with daily injections of either control IgG or Hu5F9-G4 for 2 weeks (n = 5 mice per group). The percent of hCD45+CD33+ leukemic blasts in the bone marrow was determined pre- and post-treatment in all mice by flow cytometry. The indicated p-values were determined by paired t-test. (B) At the end of treatment, mice were monitored for an additional 22 weeks with periodic assessment of human leukemic engraftment in the bone marrow. For the two SU028 mice with recurrence of disease, additional treatment and response is indicated. (C) Kaplan-Meier plot of overall survival of SU048 and SU028 cohorts treated with control IgG or Hu5F9-G4 is indicated. The indicated p-values were derived by log rank test. All mice treated with Hu5F9-G4 were sacrificed on Day 159 for histology and bone marrow analysis. (D) H&E sections of representative mouse bone marrow from mice engrafted with SU028 or SU048 post-treatment with either control IgG or Hu5F9-G4 antibody. (E) Luciferase-labeled Raji cells were transplanted intravenously into NSG mice as a model of disseminated lymphoma. 7 days after transplantation, mice were assigned to daily treatment with control IgG (200 ug), Hu5F9-G4 (100 ug), rituximab (200 ug), or combination of Hu5F9-G4 (100 ug) and rituximab (100 ug) for 21 days. Mice were imaged repeatedly up to more than 200 days to determine the bioluminescent radiance. Relative to the IgG control, p-value of tumor growth inhibition is 0.0013, 0.0065, and 0.0013 on day 21 in Hu5F9-G4-, rituximab-, and combination-treated groups, respectively. Relative to Hu5F9-G4- and rituximab-treated groups, p-value of tumor growth inhibition is 0.0112 and 0.0003 on day 21 in combination-treated groups, respectively. On day 28, p-value of tumor growth inhibition is 0.0059 in Hu5F9-G4 and combination-treated groups. P-values were determined by t-test. (F) Kaplan-Meier plot of overall survival of Raji-engrafted cohorts from panel E. The indicated p-values were derived by log rank test.

Fig 4. Non-human primate pharmacokinetic and toxicology studies show no serious adverse events associated with Hu5F9-G4.

(A) Individual cynomolgus monkeys were administered single intravenous infusions of Hu5F9-G4 at the indicated doses. The hemoglobin level was monitored over 3 weeks. The shaded bar indicates the range of hemoglobin that might trigger transfusion in humans. (B) Serum Hu5F9-G4 levels were determined from the monkeys dosed in panel A. The shaded bar indicates the range of serum Hu5F9-G4 associated with efficacy against human cancer in xenograft studies. (C) Individual cynomolgus monkeys that received either no pre-treatment (red) or pre-treatment with a single dose of EPO (black) were administered Hu5F9-G4 in a dose escalation study with the doses and time points indicated. Hemoglobin was serially measured to monitor anemia. The shaded bar indicates the range of hemoglobin in humans that might trigger transfusion. (D) Serum Hu5F9-G4 levels were determined from the monkeys dosed in panel C. The shaded bar indicates the range of serum Hu5F9-G4 associated with potent efficacy against primary human AML in xenograft studies. (E) Cynomolgus monkeys were administered a priming dose (PD) on Day 1 of either 1 or 3 mg/kg, followed by once weekly maintenance doses (MD) of 30 mg/kg at the indicated time points. Hemoglobin was serially measured to monitor anemia, and the shaded bar indicates the range of hemoglobin in humans that might require transfusion. (F) Serum Hu5F9-G4 levels were determined from the monkeys dosed in panel E; the shaded bar indicates the range of serum Hu5F9-G4 associated with efficacy against human cancer in xenograft studies. (G) Pharmacokinetic parameters in cynomolgus monkeys dosed in panel E. T1/2: half-life, Tmax: time of Cmax, Cmax: maximum observed concentration, AUC: area under the curve, CL: clearance.