Novel NKG2D-directed bispecific antibodies enhance antibody-mediated killing of malignant B cells by NK cells and T cells

Abstract

The activating receptor natural killer group 2, member D (NKG2D) represents an attractive target for immunotherapy as it exerts a crucial role in cancer immunosurveillance by regulating the activity of cytotoxic lymphocytes. In this study, a panel of novel NKG2D-specific single-chain fragments variable (scFv) were isolated from naïve human antibody gene libraries and fused to the fragment antigen binding (Fab) of rituximab to obtain [CD20×NKG2D] bibodies with the aim to recruit cytotoxic lymphocytes to lymphoma cells. All bispecific antibodies bound both antigens simultaneously. Two bibody constructs, [CD20×NKG2D#3] and [CD20×NKG2D#32], efficiently activated natural killer (NK) cells in co-cultures with CD20+ lymphoma cells. Both bibodies triggered NK cell-mediated lysis of lymphoma cells and especially enhanced antibody-dependent cell-mediated cytotoxicity (ADCC) by CD38 or CD19 specific monoclonal antibodies suggesting a synergistic effect between NKG2D and FcγRIIIA signaling pathways in NK cell activation. The [CD20×NKG2D] bibodies were not effective in redirecting CD8+ T cells as single agents, but enhanced cytotoxicity when combined with a bispecific [CD19×CD3] T cell engager, indicating that NKG2D signaling also supports CD3-mediated T cell activation. In conclusion, engagement of NKG2D with bispecific antibodies is attractive to directly activate cytotoxic lymphocytes or to support their activation by monoclonal antibodies or bispecific T cell engagers. As a perspective, co-targeting of two tumor antigens may allow fine-tuning of antibody cancer therapies. Our proposed combinatorial approach is potentially applicable for many existing immunotherapies but further testing in different preclinical models is necessary to explore the full potential.

Keywords: CD20; FcγRIIIA; NKG2D; antibody therapy; bispecific antibody; lymphoma; phage display.

Figure 1

Isolation of NKG2D-specific human scFv antibodies and sequence analysis of the V regions. (A) ScFv phage, which had been isolated from a naïve antibody library by panning against the human NKG2D antigen, were analyzed for specific antigen binding by phage ELISA using an NKG2D-Fc fusion protein and the analogously constructed control protein NKp30-Fc. ScFv phage binding an irrelevant antigen were used as controls. (B) The isolated NKG2D-specific scFvs were grouped via sequence analysis into 3 different groups according to different germline gene segment families as well as their VH/VL combinations (detailed sequence information is available (47):). The further characterized clones #3 (blue) and #32 (red) and the later used control scFv #24 (green) are highlighted.

Figure 2

Generation and characterization of bispecific [CD20×NKG2D] bibodies. (A) Schematic illustrations of the expression cassettes of bispecific [CD20×NKG2D] antibodies in the bibody format (Fab-scFv). CMV, cytomegalovirus promotor; Ig_κ, human Ig kappa secretion leader; VH_A, VL_A, sequences coding for the variable regions of the immunoglobulin heavy and light chains of the CD20 antibody rituximab (RTX), respectively; CH1, CL, sequences coding for the human immunoglobulin heavy chain constant region 1 and the human immunoglobulin kappa-light chain constant region, respectively; VH_B, VL_B, cDNA sequence coding for the variable heavy and light chain regions of the NKG2D-specific scFv; L₁, L₂, sequence coding for a linker peptides; c-myc, 6×His, sequence coding for the c-myc epitope and a hexahistidine tag, respectively. (B) Block structure of the produced bispecific antibodies in the bibody format. The NKG2D-specific scFvs were fused to a CD20 directed Fab. S-S, disulfide bridge. Purity and integrity of purified bispecific antibodies, consisting of a light chain (LC, approx. 25 kDa) and a heavy chain derivate (Fd-scFv, approx. 60 kDa), were analyzed by Coomassie stained SDS-PAGE under reducing (10% PAA) (C) and non-reducing conditions (4 – 15% PAA) (D). Of note: Only 36 of the initially sequenced 38 NKG2D scFvs could be successfully expressed as recombinant protein. Bibodies 35 and 36 expressed at very low levels and were not analyzed in all assay conditions. The numbering of the lanes represents the clone numbers #1 - #36 introduced in Figure 1. One representative experiment out of three is shown.

Figure 3

Simultaneous antigen binding of [CD20×NKG2D] bibodies and induction of NK cell activation. (A) CD20⁺ Raji lymphoma cells were first incubated with the different [CD20×NKG2D] bibodies and then reacted with NKG2D-Fc or with the control protein NKp30-Fc. Dual antigen binding of the bibodies was visualized by a FITC-coupled antibody against human Fc via flow cytometry. As a control, cells were incubated with either NKG2D-Fc or NKp30-Fc (control) in absence of the [CD20×NKG2D] bibodies or with the FITC-coupled detection antibody alone (buffer). The exemplary results are shown for the bibody [CD20×NKG2D#3], which specifically interacts with CD20 and NKG2D-Fc but not with NKp30-Fc. Note: in the left panel buffer control and NKG2D-Fc stainings are superimposed. (B) Abilities of various individual [CD20×NKG2D] bibody constructs containing different NKG2D scFv clones to simultaneously bind CD20 and NKG2D. Each data point represents an individual construct and indicates the mean fluorescence intensity value from three independent experiments. Horizontal lines show medians with interquartile range. The further characterized clones #3 (blue) and #32 (red) and the later used control scFv #24 (green) are highlighted. (C) NK cells were incubated with the [CD20×NKG2D] bibodies (10 µg/ml) in the presence of GRANTA-519 mantle cell lymphoma cells. As a control, NK cells and lymphoma cells were incubated in absence of a bibody. After 4 h, the induced expression of the activation marker CD69 was analyzed on CD56⁺/CD3^- NK cells via flow cytometry and mean fluorescent intensities were determined. Data points were normalized to CD69 expression induced by the bibody [CD20×NKG2D#3] and indicate mean values from 3 independent experiments. Horizontal lines indicate medians with interquartile range. The further characterized bibody constructs based on clones #3 (blue) and #32 (red) and the later used control construct #24 (green) are highlighted.

Figure 4

Cytotoxicity of the bibodies [CD20×NKG2D#3] and [CD20×NKG2D#32] and synergy with the CD38 specific mAb daratumumab. (A) CD20⁺/CD38⁺ GRANTA-519 MCL cells were incubated either with daratumumab, with the bispecific [CD20×NKG2D] antibodies or with their combinations, respectively, in presence of mononuclear cells (MNC; E:T ratio: 40:1) or NK cells (E:T ratio = 10:1) as effector population. After 4 h lysis of target cells was analyzed. The data points represent mean values of three independent experiments ± SEM. (*, statistically significant differences to treatment with daratumumab only; p ≤ 0.05). (B) [CD20×NKG2D#3] enhances ADCC triggered through daratumumab against tumor cells derived from two different MCL patients (p). NK cells were used as effector population. Data points represent the mean value from two independent experiments ± SEM (*, statistically significant differences to treatment with daratumumab only; p ≤ 0.05). **, P values between 0.001 and 0.01; ***, P values between 0.0001 and 0.001; ****, P values less than 0.0001.

Figure 5

Cytotoxicity of combinations of bibody [CD20×NKG2D#3] and [CD20×NKG2D#32] with an Fc-engineered CD19 mAb (CD19-DE). The cytotoxic function of the bispecific antibodies [CD20×NKG2D#3] (A) and [CD20×NKG2D#32] (B) alone, or in combination with the Fc-engineered CD19-DE mAb, was analyzed in 4 h ⁵¹Cr release assays. GRANTA-519 MCL cells (CD19⁺, CD20⁺) were used as target cells and MNC isolated from healthy donors were applied as effector population (E:T ratio = 40:1). A non-binding monoclonal IgG1 Ab was used as a control. The data points represent the mean value of four independent experiments ± SEM. (*, statistically significant differences against the treatment with CD19-DE only; p≤ 0.05). **, P values between 0.001 and 0.01; ***, P values between 0.0001 and 0.001; ****, P values less than 0.0001.

Figure 6

Cytotoxic activity of bispecific [CD20×NKG2D] antibodies with CD8-positive αβ T cells as effector population. CD8⁺ αβ T cells were isolated via MACS. The purity was determined by flow cytometry using CD3, CD8, CD16 and CD56 antibodies labelled with appropriate fluorescent dyes. Purified T cells were stimulated with IL-2 (300 U/ml) for three days and were tested as effector cells (E:T ratio: 20:1) for the bispecific antibodies [CD20×NKG2D#3] (left graph) and [CD20×NKG2D#32] (right graph) as well as their combinations with a [CD19×CD3] BiTE in a 4 h ⁵¹Cr release assay. GRANTA-519 MCL cells were used as target cells. The data points represent the mean value of three independent experiments ± SEM. (*, statistically significant differences against the treatment with [CD19×CD3] only; p ≤ 0.05). A [HER2×CD3] BiTE construct that does not bind to the target cells was used as a negative control. ***, P values between 0.0001 and 0.001; ****, P values less than 0.0001.