Cellular determinants for preclinical activity of a novel CD33/CD3 bispecific T-cell engager (BiTE) antibody, AMG 330, against human AML

Abstract

CD33 is a valid target for acute myeloid leukemia (AML) but has proven challenging for antibody-drug conjugates. Herein, we investigated the cellular determinants for the activity of the novel CD33/CD3-directed bispecific T-cell engager antibody, AMG 330. In the presence of T cells, AMG 330 was highly active against human AML cell lines and primary AML cells in a dose- and effector to target cell ratio-dependent manner. Using cell lines engineered to express wild-type CD33 at increased levels, we found a quantitative relationship between AMG 330 cytotoxicity and CD33 expression; in contrast, AMG 330 cytotoxicity was neither affected by common CD33 single nucleotide polymorphisms nor expression of the adenosine triphosphate-binding cassette (ABC) transporter proteins, P-glycoprotein or breast cancer resistance protein. Unlike bivalent CD33 antibodies, AMG 330 did not reduce surface CD33 expression. The epigenetic modifier drugs, panobinostat and azacitidine, increased CD33 expression in some cell lines and augmented AMG 330-induced cytotoxicity. These findings demonstrate that AMG 330 has potent CD33-dependent cytolytic activity in vitro, which can be further enhanced with other clinically available therapeutics. As it neither modulates CD33 expression nor is affected by ABC transporter activity, AMG 330 is highly promising for clinical exploration as it may overcome some limitations of previous CD33-targeted agents.

Figure 1

Effect of CD33 expression on AMG 330-induced cytotoxicity. (A) Parental OCI-AML3 and (B) parental KG-1a cells were transduced with wild-type CD33 to generate sublines that expressed CD33 at different levels, as indicated by arbitrary median fluorescence units. Parental cells and corresponding sublines were incubated with increasing concentrations of AMG 330 and various E:T cell ratios using healthy donor T cells. After 48 hours, cell counts were determined and cytotoxicity was assessed with DAPI staining to quantify drug-specific cytotoxicity. Results (mean ± SEM) are shown as from 4 independent experiments (for quantification of CD33 expression) or from 3 independent experiments performed in duplicate wells using a single healthy donor as source for exogenous T cells (for determination of specific cytotoxicity; qualitatively similar results were obtained with T cells from 2 other healthy donors).

Figure 2

Drug-induced CD33 modulation. After exposure to a bivalent CD33 antibody (P67.6) or AMG 330 for 48 hours, CD33 cell-surface expression was quantified on ML-1, NB4, and TF-1 cells and compared with that of corresponding untreated control cells. Results are shown as a percentage of expression (mean ± SEM) relative to control cells from 3 independent experiments.

Figure 3

Effect of Pgp and BCRP expression on AMG 330-induced cytotoxicity. Parental (A) HL-60, (B) ML-1, and (C) NB4 cells and sublines that were transduced with either wild-type Pgp or BCRP at a MOI of 100 were incubated with increasing concentrations of GO or mitoxantrone without healthy donor T cells or AMG 330 and various E:T cell ratios using healthy donor T cells as indicated. After 48 hours, cell counts were determined and cytotoxicity was assessed with DAPI staining to quantify drug-specific cytotoxicity. Results are shown as mean ± SEM from 3 independent experiments (except 2 independent experiments in the case of AMG 330 cytotoxicity assays in ML-1 cells) performed in duplicate wells using a single healthy donor as source for exogenous T cells.

Figure 4

Effect of panobinostat pretreatment on CD33 expression and AMG 330-induced cytotoxicity. Parental OCI-AML3 (A) and KG-1a (B) cells were either left untreated or pretreated with panobinostat for 72 hours. Subsequently, CD33 expression was quantified, and cells treated with/without AMG 330 (0-250 pg/mL) and various E:T cell ratios using healthy donor T cells. Forty-eight hours later, cell counts were determined and cytotoxicity was assessed with DAPI staining to quantify drug-specific cytotoxicity. Results are shown as mean ± SEM from 3 independent experiments performed in duplicate wells using a single healthy donor as source for exogenous T cells.

Figure 5

Effect of azacitidine pretreatment on CD33 expression and AMG 330-induced cytotoxicity. Parental OCI-AML3 (A) and KG-1a (B) cells were either left untreated or pretreated with azacitidine for 72 hours. Subsequently, CD33 expression was quantified, and cells treated with/without AMG 330 (0-250 pg/mL) and various E:T cell ratios using healthy donor T cells. Forty-eight hours later, cell counts were determined and cytotoxicity was assessed with DAPI staining to quantify drug-specific cytotoxicity. Results are shown as mean ± SEM from 3 independent experiments performed in duplicate wells using a single healthy donor as source for exogenous T cells.

Figure 6

AMG 330-induced cytotoxicity in primary human AML cells. (A-C) Pretreatment specimens from 3 adult patients with normal karyotype AML were phenotypically assessed to enumerate endogenous T cells and quantify CD33 expression on myeloid blasts. Parallel aliquots of cells were incubated with increasing concentrations of AMG 330 and various E:T cell ratios using healthy donor T cells. After 48 hours, cell counts were determined and cytotoxicity was assessed with DAPI staining to quantify drug-specific cytotoxicity. Individual results for samples #1, #2, and #3 are depicted in panels A, B, and C, respectively. Mean values are shown from 1 experiment performed in duplicate wells using a single healthy donor as source for exogenous T cells.

Figure 7

Effect of autologous T cells on AMG 330-induced cytotoxicity in primary AML cells. (A-B) To test the effect of autologous T cells on AMG 330-induced cytotoxicity of primary human AML cells, frozen aliquots of pretreatment specimens from 2 AML patients were used; both patients presented with hyperleukocytosis, and research specimens were obtained via leukapheresis. One portion of the aliquot was labeled with CellVue Burgundy, and lymphocytes were isolated by fluorescence-activated cell sorting based on CD45 and side scatter properties. A second portion of the aliquot was left untreated and subsequently incubated with increasing concentrations of AMG 330 in the absence or presence of additional isolated autologous T cells in defined E:T cell ratios. After 48 hours, cell counts were determined and cytotoxicity was assessed with DAPI staining to quantify drug-specific cytotoxicity. Individual results for Samples #1 and #2 are shown as mean values from 1 experiment performed in duplicate wells, in panels A and B, respectively.