Control of acute myeloid leukemia by a trifunctional NKp46-CD16a-NK cell engager targeting CD123

Abstract

CD123, the alpha chain of the IL-3 receptor, is an attractive target for acute myeloid leukemia (AML) treatment. However, cytotoxic antibodies or T cell engagers targeting CD123 had insufficient efficacy or safety in clinical trials. We show that expression of CD64, the high-affinity receptor for human IgG, on AML blasts confers resistance to anti-CD123 antibody-dependent cell cytotoxicity (ADCC) in vitro. We engineer a trifunctional natural killer cell engager (NKCE) that targets CD123 on AML blasts and NKp46 and CD16a on NK cells (CD123-NKCE). CD123-NKCE has potent antitumor activity against primary AML blasts regardless of CD64 expression and induces NK cell activation and cytokine secretion only in the presence of AML cells. Its antitumor activity in a mouse CD123⁺ tumor model exceeds that of the benchmark ADCC-enhanced antibody. In nonhuman primates, it had prolonged pharmacodynamic effects, depleting CD123⁺ cells for more than 10 days with no signs of toxicity and very low inflammatory cytokine induction over a large dose range. These results support clinical development of CD123-NKCE.

Fig. 1. The expression of high-affinity FcγR CD64 on AML cells inhibits the ADCC activity of the anti-CD123 antibody in vitro.

a, Cytotoxicity of the anti-CD123 antibody (CD123-IgG1⁺) against AML blasts from patients. Malignant cells from seven patients with AML were used as targets and purified NK cells from ten healthy donors were used as effectors. Results are shown for all healthy donor NK cells tested. b, Phenotype of the malignant AML cells from patients used in a showing the expression of CD33, CD123, CD32a/b and CD64. FI, fluorescence intensity. c, Upper panels show the cytotoxicity of anti-CD123 antibody (CD123⁻IgG1⁺) against AML cell lines with and without expression of CD32 and CD64. MOLM-13 (CD32^lowCD64⁻) and THP-1 (CD32⁺CD64⁺) cells and THP-1 subclones with inactivated CD32 (CD32-KO CD64⁺) or CD64 (CD32⁺ CD64-KO) expression were used as the target cells, with purified resting NK cells from healthy donors as effectors (n = 3). Data of a and c are presented as mean values ± s.d. Lower panels show the phenotype of the AML cell lines expressing CD123, CD32 and CD64. Ab, antibody.

Fig. 2. CD123-NKCE displays strong cytotoxic activity against AML cells, strong activation of NK cells and no off-target effects.

a, Diagrams showing the molecular organization of the NKCE molecules. The top shows the CD123-NKCE trifunctional molecule built with an unmodified human IgG1-Fc (red), targeting CD123 (orange) and coengaging NKp46 (green) and CD16a on NK cells. The bottom shows the bifunctional NKCE containing a human IgG1-Fc silenced for binding to all FcγRs (Fc null; purple). b, Comparison of the cytotoxicities of NKCEs targeting CD123 and engaging CD16a only (IC-Fc-CD123), NKp46 only (NKp46-Fc null-CD123) or coengaging NKp46 and CD16a (NKp46-Fc-CD123; CD123-NKCE). MOLM-13 cells were used as the targets and purified resting NK cells as effectors. Results for two healthy donors are shown (n = 3). Data are presented as mean values ± s.d. c, Cytotoxicity of CD123-NKCE against the AML cell line MOLM-13. Results for five healthy donors are shown. Data are presented as mean values ± s.d. d, Evaluation, by flow cytometry, of CD107, CD69, TNF-α, IFN-γ and MIP-1β expression by NK cells treated with CD123-NKCE. NK cells alone are compared with NK cells cocultured with MOLM-13 cells. Results for one representative donor are shown (n = 3).

Fig. 3. CD123-NKCE displays strong cytotoxic activity against AML cells that is not affected by expression of CD64.

a, Comparison of the cytotoxicities of NKCE molecules engaging only NKp46 (NKp46-Fc null-CD123) or coengaging NKp46 and CD16 (CD123-NKCE) against AML cell lines with and without CD32 and CD64 expression. THP-1 (CD32⁺ CD64⁺) cells and THP-1 subclones with inactivated CD32 (CD32-KO CD64⁺) or CD64 (CD32⁺ CD64-KO) expression were used as the targets, and purified resting NK cells from healthy donors were used as effectors. The data of one representative experiment among three are shown. b, Comparison of the cytotoxicities of CD123-IgG1⁺(black lozenge), NKp46-Fc null-CD123-NKCE (green circle) and CD123-NKCE (red square) against primary AML blasts expressing or not CD64. Primary AML blasts CD64-negative (AML no. 2) and CD64-positive (AML no. 4) were used as the targets and purified resting NK cells as effectors. The data for three healthy donor NK cells are shown. c, Cytotoxicity of CD123-NKCE against blasts from patients with AML. five blasts from patients with AML (AML nos. 1, 3, 5, 6 and 7) were used as targets, and purified NK cells from healthy donors (n = 9) were used as effectors. Results are shown for all the healthy NK cell donors tested. Data of a and c are presented as mean values ± s.d. d, Maximum cytotoxic activities of CD123-IgG1⁺ and CD123-NKCE molecules against blasts from patients with AML. Blast cells from seven patients with AML were used as targets and purified NK cells from ten healthy donors were used as effectors. Delta (Δ) maximum lysis, defined as percentage maximum lysis of the compound minus percentage background lysis of the isotype control molecule (IC-IgG1⁺ or IC-NKCE), were monitored from the dose response of each compound, and plotted separately for all couples of primary CD64-positive and CD64-negative AML sample/NK donor. (*P ≤ 0.05; **P ≤ 0.005; two-sided Wilcoxon matched-pairs signed rank test).

Fig. 4. CD123-NKCE promotes tumor growth control in vivo.

a, Schematic diagram of the experimental setting used in b. i.p., intraperitoneal. b, Mice engrafted with MOLM-13 cells i.v. were treated, on the day after cell injection, with 5 mg kg⁻¹ (left panel), 0.5 mg kg⁻¹ (middle panel) or 0.25 mg kg⁻¹ (right panel) surrogate CD123-NKCE (red), anti-CD123 antibody (CD123-IgG1⁺; black), or vehicle (gray). Kaplan–Meier curves were plotted for the analysis of mouse survival. Endpoint significance was calculated in a log-rank test. n = 10 to 20 per group. *P < 0.05; **P < 0.01; ***P < 0.001;****P < 0.0001. NS, not significant. c, Schematic diagram of the experimental setting used in d. d, Mice were split into two groups, one treated with anti-asilo GM1 1 day before engraftment (day −1) and on day 5 to deplete NK cells, and the other left untreated. MOLM-13 cells were transplanted i.v. into the mice of the two groups on day 0, and the mice were then treated, the day after cell injection, with 0.5 mg kg⁻¹ surrogate CD123-NKCE (red) or vehicle (gray). Kaplan–Meier curves were plotted to analyze mouse survival. Dashed lines correspond to the groups treated with anti-asialo GM1 antibody. Endpoint significance was calculated in a log-rank test. n = 10 per group. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. NS, not significant.

Fig. 5. CD123-NKCE mediates pharmacodynamics effects in human PBMCs with negligible cytokine release as compared to CD123-TCE.

a, IL-1β, TNF-α, IFN-γ and IL-6 cytokine release in vitro by PBMCs from healthy donors (n = 10) following the administration of CD123-NKCE (dose range 0.1 to 10 µg ml⁻¹), control NKCE, or a bispecific T cell engager tool targeting the same antigen (CD123-TCE, 0.1 µg ml⁻¹). Individual (dot) and median (bar) values are shown. b, CD123-positive basophil depletion activity in healthy donor PBMCs (n = 10) following the administration of CD123-NKCE (dose range 0.001 to 10 µg ml⁻¹), control NKCE (IC-NKCE) or CD123-TCE (dose range 0.001 to 0.1 µg ml⁻¹). c, Left panel shows a boxplot, with whiskers showing minimal and maximal value and upper and lower quartile box limits, of CD123-NKCE maximum depletion activity of the ten donors at the highest dose tested (10 µg ml⁻¹, 68 nM). Right panel shows a boxplot, with whiskers showing minimal and maximal value and upper and lower quartile box limits, of EC₅₀s for CD123-positive basophil depletion calculated from CD123-NKCE dose responses for six healthy donors among ten.

Fig. 6. CD123-NKCE is safe and induces pharmacodynamic effects through the sustained depletion of CD123-positive cells in NHPs.

a, Depletion of CD123-positive basophils (gated population) from the blood of monkeys M3 and M4 treated at the low dose of 3 µg kg⁻¹ was analyzed by flow cytometry before dosing (predose, pd) and 24 h after the start of the infusion. b, Numbers of circulating CD123-positive basophils (left panel) and total CD123-positive leukocytes (right panel) at time of study in monkeys M1 (orange) and M2 (purple) treated with 3 mg kg⁻¹, and monkeys M3 (black) and M4 (blue) treated with 3 µg kg⁻¹. c, IL-6 concentration in plasma of monkeys M1, M2, M3 and M4 are shown before dosing (0), and 1.5, 5 and 24 h after the start of the treatment. d, Toxicokinetics of the CD123-NKCE molecule in male monkey M5 weekly treated at a dose of 3 mg kg⁻¹ per administration for 4 weeks (on days 1, 8, 15 and 22). Plasma CD123-NKCE concentrations were determined before dosing (predose) and 1, 1.5, 5, 24 and 72 h after the start of the 1-h infusion on days 1, 8, 15 and before dosing and, 1, 1.5, 5, 24 and 168 h after the start of the last infusion on day 22. Values below the lower limit of quantification (LLOQ, 0.25 ng ml⁻¹) are not reported on the graphs. Infusion days are indicated by vertical dotted lines. e, Plasma IL-6 concentrations of monkey M5 were monitored before dosing and 1, 1.5, 5 and 24 h after the start of the 1 h infusion on days 1, 8, 15 and before dosing and 1, 1.5, 5, 24 and 168 h after the start of the last fourth infusion on day 22. f, Upper panels show the number of circulating CD123-positive basophils (open symbols) and total CD123-positive leukocytes (closed symbols) in blood (left panel) or bone marrow (right panel), by timepoint in the study, for monkey M5, treated at a dose of 3 mg kg⁻¹ per week. Lower panels show the number of circulating CD3-positive T cells (green square) in blood (left panel) or bone marrow (right panel).

Extended Data Fig. 1. CD123⁻NKCE promotes strong killing activity against AML cells that is not affected by the expression of CD64.

a, Phenotype of the 14 AML cell lines used in the study showing the expression of CD64 and CD123 at the cell surface by flow cytometry. Antibody binding capacity (ABC) values for CD123 are indicated. b, Maximum cytotoxic activities of the anti-CD123 antibody (CD123-IgG1+; black), NKp46-Fc null-CD123 NKCE (red), and IgG1 isotype control (IC-IgG1+; white) against AML cell lines. AML cell lines were used as targets and purified NK cells from 4 healthy donors were used as effectors. Data are presented as mean values ± s.d. c, Delta (Δ) maximum lysis, defined as percent maximum lysis of the compound minus percent background lysis of the isotype control molecule (IC-IgG1⁺) at the corresponding concentration, were monitored from the dose response of each compound, and were plotted separately for all couples of CD64⁻positive and CD64-negative AML cell lines/NK donor. (ns P > 0.05; **** P ≤ 0.0001; two-sided Wilcoxon matched-pairs signed rank test).

Extended Data Fig. 2. CD123⁻NKCE-mediated activation of autologous NK-cells against AML blasts is not affected by the expression of CD64.

a, Phenotypes of NK and malignant cells from AML patients. Upper panels, Expression of CD123 on AML blasts (gated on the CD33⁻positive population); middle panels, expression of CD64 (CD64 staining in black and isotype control in gray) on CD123-positive AML blasts; lower panels, expression of NKp46 and CD16a on AML sample NK cells. b, Measurement, by flow cytometry, of CD107a/b expression by NK cells after the overnight treatment of PBMCs from AML patients with 120 ng/mL CD123-NKCE (red), anti-CD123 antibody (CD123-IgG1⁺; black), and IC-IgG1⁺ (white) and IC-NKCE (gray) negative control molecules.

Extended Data Fig. 3. CD123-NKCE is safe and induces pharmacodynamic effects through the sustained depletion of CD123-positive cells in NHPs.

a, Cytokine production in cynomolgus monkeys treated with the high and low doses of 3 mg/kg and 3 µg/kg as single 1-hour intravenous infusion, respectively. Plasma IL-10 concentrations are shown before dosing (0), and 1.5, 5 and 24 hours after the start of the treatment. b, Numbers of circulating CD123-positive basophils (close symbols) and total CD123-positive leukocytes (open symbols) at time of study in monkeys M6 (pink) and M3 (black) treated with 0.5 and 3 µg/kg as single 1-hour intravenous infusion, respectively. c, Pharmacokinetics of the CD123-NKCE molecule in monkeys M1 (orange) and M2 (purple) treated with 3 mg/kg (left panel), and monkeys M3 (black) and M4 (blue) treated with 3 µg/kg (right panel). Plasma CD123-NKCE concentrations were monitored 1.5, 5, 24, 48, 72, 168, 240, 336, 504 and 672 hours (that is 0.04, 0.06, 0.21, 1, 2, 3, 7, 10, 14, 21 and 28 days) after the start of the one-hour infusion. The lower limit of quantification (LLOQ; 0.25 ng/mL) is indicated by the horizontal dotted line. d, Individual Anti-Drug antibody (ADA) ratio of monkeys M1 (orange) and M2 (purple) treated with 3 mg/kg, and monkeys M3 (black) and M4 (blue) treated with 3 µg/kg. Presence of ADA in plasma was monitored at predose (baseline) and at day 1, 7, 10, and 28 after the start of the one-hour infusion.

Extended Data Fig. 4. Expression of CD64 on target cells inhibits anti-B7-H3 antibody-mediated ADCC in vitro.

a, Comparison of the cytotoxicities of an anti-B7-H3 antibody (B7-H3-IgG1) and of an anti-B7-H3 NKCE molecule co-engaging NKp46 and CD16a (B7-H3-NKCE). CD64⁻negative A375 and LN229 cells, and CD64-positive THP-1 cells were used as the targets and purified resting NK cells as effectors. The data shown are representative of three independent experiments. b, Phenotype of the target cells used in a showing the expression of B7-H3, CD32a/b, CD64 and CD123 by flow cytometry.