An engineered NKp46 antibody for construction of multi-specific NK cell engagers

Abstract

Recent developments in cancer immunotherapy have highlighted the potential of harnessing natural killer (NK) cells in the treatment of neoplastic malignancies. Of these, bispecific antibodies, and NK cell engager (NKCE) protein therapeutics in particular, have been of interest. Here, we used phage display and yeast surface display to engineer RLN131, a unique cross-reactive antibody that binds to human, mouse, and cynomolgus NKp46, an activating receptor found on NK cells. RLN131 induced proliferation and activation of primary NK cells, and was used to create bispecific NKCE constructs of varying configurations and valency. All NKCEs were able to promote greater NK cell cytotoxicity against tumor cells than an unmodified anti-CD20 monoclonal antibody, and activity was observed irrespective of whether the constructs contained a functional Fc domain. Competition binding and fine epitope mapping studies were used to demonstrate that RLN131 binds to a conserved epitope on NKp46, underlying its species cross-reactivity.

Keywords: NCR1; NK cell engager; NKp46; antibody engineering; bispecific antibody; species cross reactive.

Figure 1

Engineering a species cross-reactive anti-NKp46 antibody. (a) Antibody discovery scheme. A phage library was alternately panned against hNKp46 and mNKp46. The enriched library was transformed into Saccharomyces cerevisiae and the resulting library was sorted using FACS. More details can be found in Fig. S1a. (b) Binding of the top yeast-displayed scFv clone from the phage library to soluble hNKp46 and mNKp46, measured by flow cytometry. Binding to his-tagged ligands is shown as a fraction of maximum binding. Mean ± SD is shown, n = 3. (c) FACS sorting of CDRH3-mutated variant library. A second library was generated by mutating four residues in the CDRH3 with NNT degenerate codons. FACS plots of the library sorting against decreasing concentrations of mNKp46 are shown, with gates denoting collected cells. Expression was detected using an anti-Myc antibody, followed by secondary antibody detection with an AF488-conjugated antibody. Binding was detected with an anti-his-AF647 antibody. (d) Binding of yeast-displayed RLN131 to soluble hNKp46 and mNKp46, measured by flow cytometry. Binding to his-tagged ligands is shown as a fraction of maximum binding. Mean ± SD is shown, n = 3. (e) BLI affinity measurement of RLN131 (mIgG1 isotype) to hNKp46, mNKp46, and cNKp46. His-tagged hNKp46, mNKp46, and cNKp46 were loaded on Octet NTA biosensors for 150 s. Following rinse of the biosensors, 2-fold dilutions of RLN131 were associated for 350 s, followed by dissociation for 1800 s, starting at 6.25 nM antibody for hNKp46 or mNKp46 (600 s, 1600 s, and 5 nM, respectively, for cNKp46). Affinity was calculated with Octet data analysis software.

Figure 2

NK cell binding and activation with RLN131. (a) Binding of RLN131 (mIgG1 isotype) to primary human NK cells and mouse NK cells. Binding is shown as a fraction of maximum binding. Mean ± SD is shown. Data for human NK cell binding was collected from NK cells from 3 separate donors (n = 1 per donor). Binding of mouse NK cells was collected from NK cells purified from the splenocytes of one C57BL/6 mouse, n = 3. (b) NK cell activation with RLN131. Primary human NK cells were incubated in microtiter plates coated with RLN131 antibody (mIgG1 isotype) or isotype control antibody for 4 h. Cells were then fixed, stained, and analyzed by flow cytometry. Data is representative of two donors. (c) Redirected lysis assay of tumor cells using primary NK cells. Cytotoxicity of primary human NK cells toward THP-1 cells was measured using a calcein release assay. NK cells and THP-1 cells were cocultured for 4 h at various E:T ratios in the presence of increasing concentrations of RLN131 (mIgG1). Mean ± SD from one donor is shown, n = 3. Data is representative of 3 separate donors. (d) Proliferation of primary human NK cells treated with RLN131 (mIgG1). Primary human NK cells were labeled with CFSE and incubated in wells coated with RLN131 or isotype control antibody for 11 d in the presence of IL-2, IL-15, or both, and analyzed by flow cytometry. The % of the NK cell sample that proliferated (left) and change in mean CFSE fluorescence from nonproliferating controls (right) is shown. Matched data from three donors is shown.

Figure 3

Cytotoxicity of NKp46-engaging NKCEs incorporating RLN131. NKCE mediated cytotoxicity was measured via calcein release from CD20-expressing Raji cells cocultured with primary human NK cells and NKCEs. Primary human NK cells were incubated with Raji cells and antibodies at an E:T ratio of 2:1 for 4 h. In all figures, a cartoon representation of the construct tested is shown (RLN131, green; anti-CD20, purple). Constructs denoted Fc- (orange lines) contain LALA-PG mutations to eliminate Fc effector function. Comparison to an unmodified anti-CD20 antibody (black lines, identical across all panels) is shown. Mean ± SD from one donor is shown, n = 3. Data is representative of 3 donors. Cytotoxicity of: (a) NKCE1 and NKCE1^Fc-, N-terminal heavy chain fusions of RLN131; (b) NKCE2 and NKCE2^Fc-, N-terminal light chain fusions of RLN131; (c) NKCE3 and NKCE3^Fc-, C-terminal heavy chain fusions of RLN131; (d) NKCE4 and NKCE4^Fc-, C-terminal light chain fusions of RLN131; (e) NKCE5 and NKCE5^Fc-, bispecific antibody constructed using knobs-into-holes mutations and Crossmab engineering; (f) NKCE6, a bispecific killer cell engager (BiKE) made by fusing anti-CD20 and RLN131 scFvs.

Figure 4

Further characterization of NKCE1. (a) NKCE1 and NKCE1^Fc- mediated cytotoxicity of primary human PBMCs toward CD20-expressing Raji cells. PBMCs were incubated with Raji cells and NKCEs at an E:T ratio of 10:1 for 4 h. Comparison to an unmodified anti-CD20 antibody (black) and an anti-CD20 antibody with the LALA-PG mutations (black, dotted) is shown. Mean ± SD from one donor is shown, n = 3. Data is representative of 3 donors. (b) NKCE1 and NKCE1^Fc- mediated activation of primary human NK cells during coculture with CD20-expressing Raji cells. Primary human NK cells were incubated with 0.01 nM NKCE and Raji cells at an E:T ratio of 1:1 for 4 h. Cells were then fixed, stained, and analyzed with flow cytometry. Comparison to an unmodified anti-CD20 antibody is shown. (c) NKCE1-mediated cytotoxicity of primary human NK cells toward CD20-expressing Raji cells as a function of linker length. NKCE1 was expressed with 0 aa, 5aa, 10aa, and 17aa (G₄S)_n linkers between RLN131 and the anti-CD20 antibody. Comparison to an unmodified anti-CD20 antibody (black) is shown. Mean ± SD from one donor is shown, n = 3. Data is representative of 3 donors.

Figure 5

Epitope mapping of RLN131. (a) Epitope binning of RLN131. Competition binding of RLN131 with various hNKp46 antibodies was measured with BLI. hNKp46 was immobilized on the sensor tip, followed by association of 100 nM RLN131 for 350 s. After washing, 100 nM D1, D2, or D1-2 anti-hNKp46 antibodies were associated for another 350 s. (b) Binding of D2, D1-2, and RLN131 IgGs to hNKp46 K41S/E42A and Y121A D121A expressed on the surface of Expi293 cells. Data represents the binding of each antibody to hNKp46 mutants as a fraction of binding to wild-type hNKp46, also expressed on the surface of Expi293 cells. Mean ± SD is shown, n = 3. ^*P < 0.0005. (c) Alanine scan of hNKp46 and mNKp46. Binding of 1 nM RLN131 to hNKp46 (top) and mNKp46 (bottom) with single alanine mutations is shown. Residues that constitute the binding epitope are highlighted (green). Mean ± SD is shown, n = 3. MFI, mean fluorescence intensity. (d) Binding epitope of RLN131 mapped to the three-dimensional structure of hNKp46 (PDB 6IAP) and mNKp46 (structure predicted with Alphafold 2.0) as seen from a side-view perspective. Residues that constitute the binding epitope and residues that were tested but are not part of the binding epitope as determined from (c) are labeled (green and yellow, respectively). The N- and C-termini of the respective proteins are shown.