Identification of novel anti-CD16a antibody clones for the development of effective natural killer cell engagers

ABSTRACT

Natural killer (NK) cells are key players in human innate immunity. Cell engager antibody formats that recruit and activate NK cells more effectively have emerged as a promising immunotherapy approach to target cancer cells through more effective antibody-dependent cell-mediated cytotoxicity (ADCC). Monoclonal antibody drugs with ADCC activity have shown clinical benefit and improved outcomes for patients with certain types of cancer. CD16a, a Fc gamma III receptor, is the major component that is responsible for the ADCC activity of NK cells. Screening AvantGen’s yeast displayed human antibody libraries led to the isolation of 2 antibody clones, #1A2 and #2-2A2, that selectively recognize both isoforms (F and V) of CD16a on primary NK cells with high affinity, yet minimally (#1A2) or do not (#2-2A2) cross-react with both allelotypes of CD16b (NA1 and NA2) expressed by neutrophils. Epitope mapping studies revealed that they bind to an epitope dependent on residue Y158 of CD16a, since mutation of Y158 to the corresponding CD16b residue H158 completely abolishes binding to CD16a. When formatted as bispecific antibodies targeting CD16a and a tumor-associated antigen (TAA, e.g. CD19), they exhibit specific binding to NK cells and induce potent NK cell activation upon encountering tumor cells, resulting in effective tumor cell killing. Notably, these bispecific antibody engagers stimulate NK cell cytokine release during co-culture with target cells, resulting in target cell cytotoxicity. These anti-CD16a antibody clones are promising candidates for combination with any TAA of interest, offering the potential for novel NK cell engager-based cancer therapeutics that are minimally affected by the high concentrations of human IgG in the circulation.

Introduction

Natural killer (NK) cells are part of the innate immune system and play a key role in the first line of self-defense against viral infections and transformed, malignant cells both through direct cytotoxic effects and through production of pro-inflammatory cytokines. NK cells express a variety of activating and inhibitory receptors and the overall balance of the two results in either a response to or tolerance of the target cells.^1,2

Two classes of NK cells, CD56^bright and CD56^dim, have been identified based on the level of expression of CD56. CD56^bright NK cells are less frequently found in the circulation and regulate adaptive immunity through cytokine secretion. CD56^dim NK cells are found at a much higher frequency in the circulation than CD56^bright cells, and both secrete pro-inflammatory cytokines and are cytotoxic. CD56^dim NK cells also express much higher levels of the low-affinity IgG Fc receptor IIIa (FcγRIIIa; CD16a).³ CD16 is the most potent activating receptor expressed by NK cells¹ and unlike other activating and inactivating NK receptors, it does not need to act in concert with other receptors but can be activated alone to mediate antibody-dependent, cell-mediated cytotoxicity (ADCC), and cytokine secretion. The affinity of the Fc region of monomeric IgG for CD16 is relatively low, in the µM range. Therefore, it has low activation in the presence of circulating monomeric IgGs, rather it is activated upon cross-linking via clustered Fc regions decorating opsonized cells and circulating immune complexes. Two forms of CD16, CD16a, and CD16b, exist. CD16a (UniProt #P08637), a 50–65 kDa heterooligomeric polypeptide-anchored transmembrane protein, is expressed largely on NK cells, subsets of monocytic cells, macrophages, and dendritic cells. CD16b (UniProt # O75015) is a GPI-anchored membrane protein expressed on granulocytes.^4–6 Substantial effort has been dedicated to improving the therapeutic outcomes of antibody drugs by increasing the affinity of the Fc region to CD16a.^7,8

While the presence of a high number of NK cells within tumors often correlates with good prognosis,^9–12 many tumors have low NK cell numbers, and anti-tumor activity of NK cells can be adversely influenced by an immunosuppressive tumor microenvironment.¹³ This has led to the development of strategies to override the negative signaling within the tumor microenvironment and activate NK cell anti-tumor targeting by generating bi- or tri-specific antibodies that combine activating NK receptor targeting with tumor targeting.^14–17 However, most activating receptors appear to require coordinated activity of another anti-activating receptor domain within a tri-specific format to effectively couple tumor targeting with enhanced NK cell activity.^15,18 As noted above, CD16a is unique among the activating NK receptors in that it does not need coordinated activation with other activating receptors to trigger cytotoxicity. This enables an effective NK cell engager to be produced in a bispecific format composed of an anti-CD16a arm with high affinity for the non-Fc binding portion of the extracellular domain (ECD) of CD16 coupled with an arm directed against the tumor antigen to enhance the interaction between the NK and target tumor cell.

One hurdle to generating effective CD16a-based NK cell engager bispecific antibodies (BsAbs) is that the ECDs of CD16a and CD16b are highly homologous and share approximately 97% identity and 98% similarity (Figure S1). Therefore, a highly specific antibody against CD16a that can distinguish between the ECDs of these two forms is preferred to avoid untoward activation of neutrophils and other granulocytes.^19–22 Another hurdle is that two different alleles for CD16a occur in humans, a F176 and a V176 (numbering based on the full-length UniProt entry) variant that differ in affinity for IgG,^23,24 and three different alleles exist for CD16b, NA1, NA2, and the rare SH,²⁵ where NA1 and NA2 are detectable with antibodies against the biallelic neutrophil-specific antigen system NA²⁶ (Figure S1). Therefore, an ideal antibody should bind both the F176 and the V176 form of CD16a but not bind to the different allelic forms of CD16b.

Here, we describe the identification of two novel anti-CD16a human antibody clones that specifically recognize both CD16a-F176 and CD16a-V176 with negligible binding to CD16b. Importantly, their binding to human NK cells is not competed by high concentrations of nonspecific human polyclonal IgG (hIgG) at levels equivalent to those found in the circulation. In the bispecific engager formats, with one arm targeting CD16a and the other arm targeting a tumor-associated antigen (TAA), these CD16a-specific clones exhibit potent tumor cell-dependent activation of NK cells associated with the release of cytokines and effective killing of tumor cells in both monovalent and bivalent formats. Therefore, these anti-CD16a antibody clones are ideal candidates for developing effective NK cell engagers for anti-tumor therapy.

Results

Identification of a specific anti-CD16a antibody clone from the yeast display synthetic human antibody libraries

To isolate an antibody that specifically targets CD16a, AvantGen’s existing yeast display synthetic human antibody libraries were used. These libraries were constructed based on the analysis of the complementarity-determining region sequences by germline variable region subfamily of a deep sequencing data set of antibody clones from the B cells of 500 individuals performed in collaboration with IMGT. The existing libraries consist of five different VH families, each combined with four V$\mathit{\lambda }$ libraries and one Vλ library.

Recombinant human CD16a and CD16b ECDs were used to screen for positive binders to both F and V variants of CD16a that were negative for binding activity to both CD16b-NA1 and CD16b-NA2 ECDs. From a total of 10¹¹ antibody clones, a panel of 9 antibody clones were identified that met our initial criteria, specifically that they bind to both CD16a-F and CD16a-V, but not to CD16b-NA2 recombinant proteins in enzyme-linked immunosorbent assays (ELISAs) (Figure S2A), and to CD16a-F-expressing 293F cells, but not CD16b-SH-expressing or untransfected 293F cells (Figure S2B). These clones were then subcloned into mammalian expression vectors and purified as IgG1 antibodies (Figure 1(a)) from transfected 293 cell culture media, and their binding affinity was determined by ELISA. Among these clones, clone A2 showed the highest affinity to both variants of CD16a, 0.25 nM and 0.19 nM to CD16a-F and CD16a-V, respectively, and the lowest binding (only marginally above background at the highest concentration tested, 100 nM) to both CD16b ECDs (Figure 2(a)). Upon further characterization, this clone exhibited positive binding to NK cells (Figure S3A), and negative binding to neutrophils (Figure S3B-D) and could activate NK cells in a bispecific engager format (Preliminary data not shown). Therefore, it was selected for affinity maturation.

Figure 1.

Formats of the anti-CD16a antibody clones used in this study. (a) and (b) are anti-CD16a antibody clones only, (c-f) are monovalent or bivalent NK cell engagers. (a) and (e) have a silenced Fcγ.

Figure 2.

Binding of anti-CD16 antibody clones to recombinant CD16. (a-c) ELISA to determine the binding of antibody clones A2-IgG1 (a), #1A2-IgG1 (b) and #2-2A2-IgG1 (c) to human CD16a-F and V and CD16b-NA1 and NA2. (d) and (e) ELISA to compare the binding of antibody clones #1A2-IgG1 (d) and #2-2A2-IgG1 (c) to the extracellular domains of CD16a-V compared with CD32a, CD32b and CD64. (f-i) binding of anti-CD16-IgG1 antibody clones to Expi293-F cell-expressed CD16 by flow cytometry. (f) and (g) binding to CD16a-F and CD16-V, respectively; (h) and (i) binding to CD16b-NA1 and CD16b-NA2, respectively; (j) binding to cynomolgus CD16. The pan anti-CD16 antibody clone, 3G8, and a non-relevant antibody clone were used as reference binding controls.

Determination of the binding affinity of the anti-CD16a antibody clones

Two antibody clones, #1A2 and #2-2A2 were isolated from the affinity maturation campaign. They were produced as IgG1s harboring the Fc-silenced mutations in the CH2 domain (Leu234Phe, Leu235Glu, and Asp265Ala), which effectively abolishes the binding of Fc to CD16.^27–29 The binding activity of these two antibody clones to recombinant CD16 ECDs was assessed by ELISA. As shown in Figure 2(b,c), both clones showed improved binding to both F and V variants of CD16a compared to the parental A2 clone (Figure 2(a)), with minimal binding to CD16b-NA1 and NA2. No binding to other types of Fcγ receptors, such as CD32a, CD32b, and CD64, was detected (Figure 2(d,e)). Next, their binding to CD16-expressing 293 cells, compared to that of A2 was assessed using a pan CD16-specific antibody clone, 3G8.³⁰ As shown in Figure 2(f–i), clones #1A2 and #2-2A2 recognized the native CD16a-F and V variants expressed on 293 cells with equivalent affinity, similar to that of 3G8 antibody, whereas their binding to 293 cells expressing CD16b-NA1 and NA2 was significantly lower than that of 3G8. Interestingly, the parental A2 clone exhibited similar affinity to the two A2 derivatives in binding activity to full-length CD16a-F expressed by 293 cells but had a 2-fold reduction in apparent affinity to cells expressing CD16a-V. In addition, both #1A2 and #2-2A2 show strong binding to cynomolgus CD16 expressed on 293 cells (Figure 2(j)). Neither #1A2 and #2-2A2 exhibited any binding above background to untransfected 293 cell controls, confirming the specificity of binding (data not shown)

These two clones were produced in various formats for characterization in addition to IgG1 with a silenced Fc (Figure 1(a)), including as Fab (Figure 1(b)), monovalent single-chain variable fragment (scFv) bispecific killer engager (BiKE) (Figure 1(c)), bivalent BiKE-IgG4 (Figure 1(d)), and bivalent IgG-like BsAb with a silenced Fcγ1 (Figure 1(e)). For the BiKE, the orientation was anti-CD19-(LC-HC)-linker-anti-CD16a-(LC-HC) followed by a hexa-His-tag to facilitate purification. For the bivalent BiKE-IgG4, the C-terminal of BiKE was linked to the N-terminal of the Fcγ4 domain. For the IgG-like BsAb, the N-terminal of the anti-CD16a scFv was linked by a linker of 5 G₄S repeat to the C-terminal of anti-CD19-IgG1 with the silenced Fc.

The binding kinetics of both #1A2 and #2-2A2 antibody clones in various formats to recombinant CD16 were determined by biolayer interferometry (BLI). The monoclonal antibodies and bivalent engagers that carry either IgG1-Fc (Figure 3(a,e)) or IgG4-Fc (Figure 3(d)) were captured by anti-human Fc biosensors, and their binding to the relevant CD16 isoform as a monovalent His-tagged protein was evaluated. For monovalent anti-CD16 Fab (Figure 3(b)) and BiKE (Figure 3(c)), biotinylated CD16a were captured by streptavidin-coated biosensors, and their binding to anti-CD16a-Fab or BiKE was evaluated. As shown in Figure 3, the affinity (K_D) ranged from 2.15 nM to 6.38 nM for the #1A2 clone and from 5.68 nM to 9.89 nM for the #2-2A2 clone for the IgG1, Fab, BiKE, and IgG-like BsAb with each format exhibiting equivalent affinity to both CD16a-F and CD16a-V. The bivalent BiKE-IgG4 format where the anti-CD16a scFv is between the anti-target antibody and the IgG4-Fc, which exhibited an affinity to CD16a that was approximately 4 to 6-fold-lower (Figure 3(d)). No measurable binding to CD16b-NA2, and only minimal binding to CD16b-NA1 was detected for either of the two affinity-matured clones. Finally, both clones bound to cynomolgus CD16a with similar affinity as to human CD16a (Figure 3(a)).

Figure 3.

Measurement of binding constant (K_D), association constant (K_on) and dissociation constant (K_off) of anti-CD16 antibody clones by BLI. (a) anti-CD16a clones in IgG1 format; (b) anti-CD16a in fab format, in the absence or presence of 10 mg/mL of hIgG; (c) anti-CD16a in monovalent BiKE format; (d) anti-CD16a in the bivalent IgG-like BsAb format; (e) anti-CD16a in the bivalent BiKE-IgG4 format. An anti-HIgG Fc biosensor was used to capture Fc-containing antibodies (a, d and e) or streptavidin biosensor was used to capture biotinylated CD16a (b and c). Non-biotinylated CD16a (a, d and e), CD16b (a), fab (b) or BiKE (c) at serial dilutions was added to the biosensors, in the presence (b) or absence (a, b, c, d and e) of 10 mg/mL of hIgG.

In normal human serum, the concentration of human IgG is in the range of 10 gm/L. Therefore, an important consideration is that the binding of the anti-CD16a clones to their CD16a epitope on NK cells is not adversely affected by binding of nonspecific human IgG to the normal Fc binding site on CD16. To assess the impact, the assays were performed in the absence or presence of 10 mg/mL of hIgG. The data showed that the binding of these two variant A2 clones to CD16a was only affected by 1.2 to 4.8-fold in either Fab or IgG1 formats (Figure 3(b), Table S1). Similar shifts were observed for the BsAb engager formats (Table S2). Similarly, in the ELISA and BLI assays performed to determine the binding of these antibody clones to recombinant CD16a, Clone #1A2 shows strong resistance to hIgG competition and retains full binding activity (no shift in EC₅₀ values). Since a CD16a-specific scFv clone was previously reported by Affimed,¹⁴ it was of interest to also include this clone in our study as a comparator. This clone was generated in-house using the published sequence of the variable regions produced with the same constant regions as the #1A2 and #2-2A2 clones. Affimed’s anti-CD16a clone exhibited a greater shift (2.0 to 11.2-fold) in the presence of 10 mg/mL hIgG than clone #1A2 in all formats tested (Table S1 and S2 and Figure S4A to E for IgG1 format). Collectively, these results indicate that these anti-CD16a antibodies recognize an epitope on CD16a that is distinct from the Fc binding site.

Defining the #1A2 and #2-2A2 epitope binding site on CD16a

First, an epitope binning study was performed to assess whether these two antibody clones would compete with the pan CD16-specific antibody clone, 3G8, which has been shown to block Fc binding.³¹ Biotinylated CD16a-F was first captured by streptavidin-coated biosensors, and its association with the first monoclonal antibody was monitored until it reached a plateau. The second antibody clone was then added to determine whether it could show additional binding to CD16a-F as evidenced by the increase in the refractive signal, i.e., its epitope was not masked by the first antibody clone. As shown in Figure 4A, #1A2 or #2-2A2 were able to bind to CD16a, independently of 3G8 binding activity, regardless of the order in which the clones were added. Since neither clone competed with 3G8 antibody binding, this indicated that their binding site does not overlap with that of 3G8 to the Fc binding site of CD16a.

Figure 4.

Epitope binning and mapping for the anti-CD16a antibody clones. (a) Epitope binning between anti-CD16 antibody clones with the pan anti-CD16 antibody clone, 3G8. Two streptavidin biosensors were first loaded with biotinylated CD16a-F, which were then associated with the first IgG1 antibody clone (these harbor a silenced Fcγ1) or the murine antibody 3G8 at 100 nM to reach the saturation binding. They were further associated with either the first antibody clone, a second antibody clone or 3G8 at 100 nM to detect whether any additional CD16a-F:antibody association was observed. (b) ELISA to determine the binding of anti-CD16 antibody clones in IgG1 format to CD16a mutants, G147D or Y158H4.

Additionally, when a control IgG clone with normal Fc domain was added after reaching maximal binding with either the #1A2 or #2-2A2 clone or vice versa, further binding was observed as evidenced by the increase in the refractive index signal (Figure S5A). In contrast, no additional signal was observed when control IgG was added to bound 3G8, providing further evidence that the two affinity-matured clones did not bind the Fc binding site on CD16a competed by Clone 3G8.²⁸ When Affimed’s clone was added to bound #1A2 or #2-2A2 or vice versa, no increase in bound protein signal was observed, suggesting that the two clones we isolated share an overlapping binding site to the Affimed’s antibody clone (Figure S5B).

The ECD sequences of the CD16a and CD16b variants differ in only three amino acids, namely the amino acid residues 147, 158, and 203 (Figure S1). Phe203 in CD16a determines the start of the transmembrane domains, whereas Ser203 in CD16b determines the GPI-anchoring site. Therefore, only the two other amino acids, 147 and 158, could contribute to the differential binding to CD16a and CD16b by these two antibody clones. The CD16a-F-G147D or Y158H variants were generated as CD16a-F ECD-Fc fusion proteins and purified from transfected 293 cells culture supernatants. Analysis of binding specificity by ELISA of both clones to these mutant antigens indicated that minimal (#1A2) or no (#2-2A2) binding was observed to the CD16a-F-Y158H variant, whereas binding to the CD16a-F-G147D variant was similar to that of CD16a-F-WT (Figure 4(b)), while the control antibody (3G8) retains its binding activity to CD16a-F-Y158H (data not shown), suggesting the CD16a-F-Y158H is folded correctly. Furthermore, neither antibody clone bound to native CD16a-F-Y158H expressed on transfected 293 cells (data not shown) in contrast to 293 cells expressing wildtype CD16a (Figure 2(f,g)). This result indicates that amino acid Y158 is important for the binding of these two antibody clones to CD16a, and the single amino acid variation between CD16a and CD16b at position 158 contributes to the high degree of specificity of these two antibody clones to CD16a compared to CD16b.

Binding activity to primary NK cells and neutrophils

These two antibody clones were then tested for their binding activity to primary NK cells and neutrophils. The #1A2 and #2-2A2 IgGs with the silenced Fc were biotinylated, with a final biotin:antibody ratio between 2 and 3:1. They were then incubated with NK cells from two different donors, in the presence or absence of 10 mg/mL of hIgG. As shown in Figure 5(a), both antibody clones recognize CD16a expressed by NK cells, with apparent K_D values of 8–10 nM, similar to the binding profile to CD16a-expressing 293 cells (Figure 2(f,g)). In the presence of 10 mg/mL of hIgG, they retained binding to the NK cells, albeit with a <3.7-fold decrease in apparent affinity (apparent K_D values of 20–30 nM), which is considerably higher affinity than the typical µM affinity of IgG to the Fc binding site on CD16a (Figure 5(b)), indicating that these clones are resistant to hIgG competition. In contrast, the 3G8 antibody binding to CD16a on NK cells was significantly affected by the presence of hIgG, which is consistent with data previously published by others.³¹

Figure 5.

Binding of anti-CD16a antibody clones to primary NK cells and neutrophils. (a) and (b) binding of anti-CD16a antibody clones in IgG1 format to primary NK cells isolated from 2 donors with a CD16a-F/V genotype in the absence (a) or presence (b) of 10 mg/mL of hIgG; (c) binding to anti-CD16a antibody clones in IgG1 format to neutrophils isolated from 2 donors, both with a CD16b-NA1/NA2 genotype. The 3G8 antibody and a control antibody serve as a positive or negative binding control, respectively.

For the neutrophil binding study, neutrophils were isolated from normal healthy donors and the CD16b alleles genotyped as described above for the parental A2 clone. As shown in Figure 5(c), clone #1A2 showed minimal binding to neutrophils only at concentrations >100 nM, whereas clone#2-2A2 showed no binding above background to the neutrophils from two separate subject samples up to 800 nM.

In vitro cytotoxicity assay

These antibody clones were then produced in monovalent and bivalent BsAb formats with CD19 selected as the TAA to target for proof-of-concept studies (Figure 1(c–e). The anti-CD19 antibody was constructed based on the sequence of blinatumomab (Drug Bank accession number DB09052). To ensure that the various NK cell engager constructs were stable and monodisperse, they were analyzed by both sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and high-performance liquid chromatography-size-exclusion chromatography (HPLC-SEC). As shown in Figure S6, the three constructs all yielded a monodisperse peak, with no evidence of aggregation.

These BsAbs were tested for their binding activity to primary NK cells. As shown in Figure 6, all three formats of the BsAb constructs bind to NK cells with high affinity in the sub nM to single-digit nM range, similar to the robust binding to NK cells observed with the parental A2 clone in a BiKE format (Figure S7). However, their binding to neutrophils is either minimal (#1A2-BiKE-IgG4) or close to background levels (all other BsAb variants) (Figure S8), similar to what was previously observed for the anti-CD16a clones as normal IgG1 antibodies (Figure 5(c)). The addition of 10 mg/mL of hIgG to BiKE and IgG-like BsAb resulted in a 2-3-fold shift in binding affinity for NK cells (Figure 6).

Figure 6.

Binding of bispecific NK cell engagers to primary NK cells. The NK cell engager format is shown above each set of binding curves. (a) Binding of clones #1A2 and #2-2A2 BiKEs to primary NK cells in the presence or absence of 10 mg/mL of hIgG; (b) binding of clones #1A2 and #2-2A2 bivalent BiKE-IgG4s to NK cells in the absence of hIgG; (c) binding of clones #1A2 and #2-2A2 IgG-like BsAbs to primary NK cells in the presence or absence of 10 mg/mL of hIgG.

Two CD19-positive cell lines, Raji,³² a Burkitt lymphocyte cell line, and NALM6, a B cell precursor leukemia cell line, were used as target cells. Primary NK cells from several individuals were isolated and co-cultured with target cells in the presence of a serial dilution of BiKE. As shown in Figure 7(a,d), this monovalent NK cell engager effectively resulted in both Raji and NALM6 cell cytolysis, with EC₅₀ values in the 10⁻¹² M range (3.3 pM and 2.4 pM for #1A2, respectively, and 8.1 pM and 4.1 pM for #2-2A2, respectively).

Figure 7.

ADCC assays with different formats of NK cell engagers. in vitro cytotoxicity assay with target cancer cells, Raji (a-c and g) or NALM6 cells (d-f and h) added to primary NK cells in the presence of a serial dilution of the different formats of NK cell engagers. Target tumor cells were pre-loaded with CFSE. Serial dilutions of the NK cell engager were added to a 5:1 effector NK to target cell ratio. After 20 hours incubation the number of viable target cells was compared to control wells to give percent killing. The percentage of cell death was determined by the loss of viable cells (CFSE labeled) compared to numbers of viable cells in the presence of NK cells without the engagers.

The bivalent-BiKE-IgG4 and IgG-like BsAbs exhibited potent Raji and NALM6 cytotoxicity, with EC₅₀ values in the 10⁻¹³ M range (Figure 7(b–f)). The bivalent A2-BikE-Ig4 was also tested using NK cells from patients with different genotypes and exhibited a slightly higher potency with F/F homozygotes (EC₅₀ value of 8.3 pM; Figure S9A) compared to the F/V heterozygotes (EC₅₀ value of 13 pM; Figure S9B) or V/V homozygotes (EC₅₀ values of 29 pM; Figure S9C and D).

Interestingly, the analogous constructs using the Affimed’s anti-CD16a clone consistently exhibited less potent activity than the #1A1 and #2-2A2 clone constructs; the monovalent BiKE format with Raji cells showed a 13- and 7-fold decrease in EC₅₀ values compared to Clone #1A2 and #2-2A2 BiKEs, respectively (Figure 7(g)). In the case of the BiKE-IgG4 format, Affimed’s BiKE-IgG4 exhibited a two-fold decrease in cytotoxicity with NALM6 cells (Figure 7(h)).

The same difference in efficacy was also observed in reporter assays (Figure S10), which also showed that the activity of the Affimed’s clone-based constructs was much more negatively impacted by the presence of 10 mg/mL hIgG. This data suggested that the #1A2 and #2-2A2 clones support superior activity in an NK cell engager format.

To test if engagers based on the #1A2 and #2-2A2 antibody clones exhibit cytotoxicity against other types of tumor cells when coupled to antibody clones directed against other TAAs, IgG-like BsAbs targeting B-cell maturation antigen (BCMA) or human epidermal growth factor receptor 2 (Her2) were produced with an anti-BCMA antibody clone developed in-house using the same synthetic human antibody libraries (data not shown) and with anti-Her2 trastuzumab.³³ Two cancer cell lines, H929, a multiple myeloma cell line³⁴ and SKBR3, a human breast cancer cell line, that are positive for BCMA or Her2, respectively, were used as target cells. As shown in Figure 8(a,b), these NK cell engagers also effectively induced H929 and SKBR3 cell cytolysis. Significant cytotoxicity was only observed when CD19-IgG-like BsAb or Her2-IgG-like BsAb NK cell engagers were incubated with the relevant TAA-expressing cells. No cytotoxic effect of NK cells was observed in the absence of an IgG-like BsAb (Figure 8(c)). These results indicate both that the cytotoxic effect is dependent on the presence of the relevant format targeting the relevant TAA and that this approach could be used to target a wider range of TAAs for enhanced NK cell-mediated ADCC.

Figure 8.

ADCC assays with different TAA-NK cell engagers. in vitro cytotoxicity assay with (a) H929 cells using anti-BCMA-IgG-like BsAb or (b) SKBR3 cells using trastuzumab (Her2)-IgG-like BsAb. Target tumor cells were pre-loaded with CFSE. Serial dilutions of the indicated IgG-like BsAb NK cell engager were added to a 5:1 effector NK to target cell ratio. After 20 hours incubation the number of viable target cells was compared to control wells to give percent killing. (c) Comparison of NK-mediated cytotoxicity of either NALM6 or SKBR3 target cells in the presence of 10 pM of CD19-IgG-like BsAb, Her2-IgG-like BsAb or no IgG-like BsAb. NK effector (E) cells were mixed in a 2:1 or 4:1 ratio with NALM6 or SKBR3 target (T) cells in the presence or absence of the indicated BsAb antibody constructs and incubated in a humidified 5% CO2 atmosphere for 20 hours. Then the viable cells were counted and compared to the number in the samples with no BsAb NK cell engagers, which was set at 100%. The data are presented as mean ± SD (n = 3). *p < 0.05 in the paired T-test.

Finally, to assess whether the ADCC activities induced by the NK cell engagers were the result of paracrine action of cytokines released by NK cells, bivalent CD19-BiKE-IgG4 was added to NK cells in the presence or absence of NALM6 cells for 24 h. The levels of human interferon-gamma (IFN-$\mathit{\gamma }$) and tumor necrosis factor (TNF), which are important in the NK cells immune responses,³⁵ were then measured in the culture media pre and post addition of the target cells and the NK cell engager constructs. As shown in Figure 9, both #1A2 and #2-2A2 BiKE-IgG4 engagers were associated with the release of IFN-$\mathit{\gamma }$ (Figure 9(a)) and TNF (Figure 9(b)) only in the presence of NALM6 cells. No detectable IFN-$\mathit{\gamma }$ or TNF was produced by NK cells alone under these conditions.

Figure 9.

NK cell engagers trigger the release of cytokines by NK cells in the presence of target cells. Bivalent BiKE-IgG4 was added to the NK cells in the presence (■) or absence (□) of NALM6 cells at the E:T ratio of 5:1. NALM6 cells mixed with NK cells without an engager serves as background cytokine level. After 24 hours incubation, IFN-$\mathit{\gamma }$ (a) and TNF (b) in the culture supernatants were quantified. The data are presented as mean ± SD (n = 3). *p < 0.05 in the paired T-test.

Discussion

The engagement of CD16a expressed on NK cells in immunotherapy is of critical importance. Monoclonal antibodies utilize CD16a receptors on NK cells and macrophages to initiate ADCC and ADCP (antibody-dependent cell phagocytosis) mechanisms.^36,37 These processes are vital in the anti-tumor effects of these antibodies, leading to enhanced outcomes in cancer immunotherapy. As demonstrated by studies with rituximab, a chimeric anti-CD20 monoclonal antibody, and trastuzumab, an anti-Her2 monoclonal antibody, the modulation of CD16a receptors triggers NK cells and macrophages to induce ACDD and ADCP activities against malignant cells.^37–40 However, since the binding affinity of normal monodisperse IgG is in the uM range, effective NK engagement typically relies on aggregated immune complexes for effective ADCC. Consequently, approaches using mutated Fc with enhanced CD16a binding activity have been developed to enable greater cytotoxicity activity.⁴⁰ This emphasizes the need for strategies that bolster the effectiveness of these processes and ultimately improve the success of cancer immunotherapy. It is important to note that the F176 vs V176 allelic difference of CD16a, which display different affinity to Fc, may result in differences in CD16a engagement between patients, which in turn affects the outcome of monoclonal antibody therapy.²⁴

In this study, we aimed to isolate anti-CD16a-specific antibody clones that recognize both allelic variants of CD16a and evaluate their roles in ADCC activity in the corresponding monovalent and bivalent engager formats. From our synthetic human antibody library using yeast display technology,⁴¹ we first isolated a panel of antibody clones that meets the screening criteria: specific binding to human CD16a, both F and V variants with equivalent affinity, and low or no binding to both dominant allelotypes of human CD16b, NA1, and NA2 (Figure S2A and B). Upon further analysis, clone A2 was selected since it exhibited robust binding to NK cells (Figure S3A), no binding to neutrophils (Figure S3D) and in the monovalent and bivalent engager formats, it could induce target cell-specific cell cytotoxicity (Figure S9).

The A2 clone then underwent an affinity maturation campaign to identify variants that exhibited higher affinity to CD16a and were also better able to resist the negative competition effect of high concentrations of circulating human IgG binding to the Fcγ binding site on CD16a. Two antibody clones have been isolated from this campaign, #1A2 and #2-2A2 which showed improved affinity to CD16a in the Fab format, while exhibiting minimal binding to CD16b, as measured by BLI (Figure 3). Importantly, both clones show improved binding to NK cells (Figures 5(a) and 6), while maintaining minimal (#1A2) or no (#2-2A2) binding to neutrophils (Figure 5(c)). Both clones specifically recognize the amino acid residue Y158 on the CD16a, but not H158 present in the CD16b (Figure 4(b) and S1), explaining their minimal to no binding to CD16b (Figure 3(a)). Because this residue is further away from amino acid G147, which is considered in close proximity to the Fc binding site,¹⁴ both clones show resistance to hIgG competition to a similar degree, in both monovalent and bivalent formats (Figures 3(b), 5(a,b), Table S1 and S2).

These two clones were then used to construct three potential NK cell engager formats, using an anti-CD19 arm derived from the published sequence of blinatumomab: 1) a bispecific scFv BiKE, 2) a bivalent bispecific IgG4 (BiKE-IgG4) and 3) an Fc-silenced, IgG1-like BsAb. BLI analysis indicated that the BiKE and IgG-like BsAb formats exhibited similar affinities to CD16a to their parental clones in a regular IgG1 format in the 2.84 and 9.31 nM range, but the BiKE-IgG4 bivalent format exhibited a decreased affinity of about 15 nM for the CD16a variants for the #1A1 construct and 23–27 nM for the #2-2A2 construct (Figure 4: panel d compared to panels a, b, c, and e); this lower affinity for CD16a is possibly due to the position of the anti-CD16a moiety being in the middle of sandwich position (Figure 1).

Strong binding to NK cells in the presence of hIgG and minimal binding to neutrophils are desired features in clinical settings.^19,20 In all three formats, these NK cell engagers retained strong binding activity to primary NK cells, while this binding for the BiKE and IgG-like BsAb formats is largely unaffected by the presence of high concentrations of hIgG, showing only 2-3-fold shift in binding affinity (Figure 6(a,c)). Furthermore, they exhibit negligible binding to neutrophils, except in the case of #1A2 in the BiKE-IgG4 format (Figure S8B), where some binding was observed at high concentrations (>100 nM). However, these concentrations are much higher than the anticipated clinical doses of these NK cell engagers.

In all three formats, these CD19-directed NK cell engagers could activate NK cells and tumor-cell cytotoxicity against both Raji and NALM6 in a dose-dependent manner (Figure 7(a–g)) with no effect against SKBR3 cells, which do not express CD19 (Figure 8(c)). Similar results were observed using BCMA and Her2 as the target antigens expressed on H929 multiple myeloma and SKBR3 breast cancer cells, respectively (Figure 8(a,b)). Further analysis revealed that this cytotoxicity correlates with cytokine release by the NK cells in a target cell-dependent manner (Figure 9). This is important because NK cell activation without the presence of target cells may result in NK cell fratricide, resulting in NK cell depletion.^42,43

Among these three formats, monovalent BiKE in general shows the lowest cell killing activity, which is consistent with published data for the CD16a-specific antibody from Affimed.¹⁴ For the bivalent formats, the IgG4 format seems to offer higher cell killing efficacy than the IgG-like BsAb format when Raji cells were the target cells. When NALM6 cells were the target cells; however, no significant differential cell-killing effect was observed between the bivalent BiKE-IgG4 format and the IgG-like BsAb format (Figure 7(e,f)). One possibility for this difference is that the distance between the anti-target antibody moiety and the anti-CD16a antibody moiety may be crucial to achieve the optimal activation of NK cells.^44,45 Other factors, such as the density of target protein on the cell surface, may also contribute to the cell-killing activity of these NK cell engagers.⁴⁶

Several studies have reported the identification of anti-CD16a antibody clones.^14–47–49 Notably, the NK cell engager developed by Affimed based on their anti-CD16a antibody is in clinical development.^50,51 The anti-CD16a VHH nanobody developed by Nikkhoi et al., however, recognizes CD16b-NA2,⁴⁸ which may limit its scope of application.⁵²

The #1A2 and #2-2A2 clones share many similar characteristics as those of Affimed’s anti-CD16a clone.¹⁶ The gene encoding this clone was generated based on the sequence in a European patent (EP2450380A1)⁵³ and produced from transfected 293 cells to serve as a reference clone for comparison with the clones we isolated. For example, they compete for the same binding epitope on CD16a with Affimed’s clone as they bind to the same epitope on CD16a (Figure S5B, S5C, and Figure 4(b)).¹⁴ As a result, they all resist hIgG competition (Table S1, Figure S10B) and exhibit low binding to the dominant NA1 and NA2 CD16b allotypes.

However, in a head-to-head comparison with Affimed’s anti-CD16a clone, the #1A2 and #2-2A2 clones exhibited higher binding affinity to recombinant CD16a-F/V and notably, clone #1A2 exhibited only a minimal shift in affinity in the presence of 10 mg/mL of hIgG compared to #2-2A2 and Affimed’s clone either in Fab or IgG format (Table S1). In the BsAb formats, as shown in Table S2, both #1A2 and #2-2A2 show higher binding affinity than that of the Affimed’s clone, suggesting that both clones may support more potent NK cell engagers. Surprisingly, in cytotoxicity assays, these two clones have more than a 10-fold improvement of EC₅₀ values in both the BiKE and IgG-like BsAb formats compared to those of Affimed’s clone in the same format (Figures 7(g), 8(a,b)). In the ADCC reporter assay, these two clones also outperformed Affimed’s clone at both activation of the reporter cells and resistance to hIgG competition (Figure S10A-D). It is worth noting, however, that the format of the Affimed’s NK cell engagers that have been used in clinical settings differs from the formats we tested.^50,51 It remains to be seen whether the optimal properties for the #1A2 and #2-2A2 clones described here will translate to a clinical advantage.

In summary, here we report the identification of two novel anti-CD16a antibody clones that have unique characteristics, such as high binding affinity and specificity to CD16a and high resistance to human IgG competition compared to other reported anti-CD16a clones, to be used as effective NK cell engagers. Future studies will determine whether the advantages of the #1A2 and #2-2A2-based NK cell engagers reported here can better leverage a patient’s innate immunity to fight cancer.

Materials and methods

Production of recombinant CD16 and antibodies

The following biotinylated or non-biotinylated recombinant ECDs were purchased from Acrobiosystems (Newark, DE). CD16a-F176 (Cat.: CDA-H82E8 or CDA-H5220, respectively) and CD16a-V176 (Cat.: CDA-H82E9 or CDA-H52S1, respectively) ECDs encompassing residues Gly 17 to Gln 208 followed by an Avi and/or a His tag at the C-terminus. CD16b-NA1 (Cat.: CDB-H82E4 or CDB-H5227, respectively) and NA2 (Cat.: CDB-H82Ea or CDB-H5222, respectively) ECDs encompassing residues Gly 17 to Ser 200 followed by an Avi and/or a His tag at the C-terminus. Non-biotinylated cynomolgus CD16 (Cat.: FC6-C52H9) ECD encompassing residues Gly 17 to Gln 208 followed by a His tag at the C-terminus. Biotinylated human CD32a (Cat.: CDA-H82E6), CD32b (Cat.: CDB-H82E0) and CD64 (Cat.: FCA-H82E8) ECDs were also from Acrobiosystems.

Gene fragments of different recombinant proteins were synthesized by Integrated DNA Technologies (San Diego, CA) or derived by polymerase chain reaction (PCR) amplification. The anti-CD19 antibody was constructed based on the sequence of blinatumomab (Drug Bank accession number DB09052). The trastuzumab (Herceptin®) sequence was also obtained from DrugBank (Accession number SB00072). The anti-BCMA antibody clone was isolated in-house from the same synthetic human antibody libraries used to isolate clone A2.

Expression vectors encoding these recombinant proteins were generated by cloning the respective sequence elements into a modified version of the mammalian expression vector, pcDNA3.4 (Life Technologies), which includes the N-terminal signal peptides to facilitate the secretion of recombinant proteins into the culture media. Soluble recombinant CD16 variants were constructed as fusion proteins of the ECD sequence coupled to human IgG1-Fc. For cell surface-anchored CD16 expression, the CD16a-F (Cat.: HG10389-UT) and CD16b-SH (Cat.: HG11046-UT) expression plasmids were purchased from Sino Biological (Wayne, PA) and variants, CD16a-V, CD16b-NA1, and CD16b-NA2 were created by site-directed mutagenesis. The subcloning was carried out using Gibson Assembly Master Mix (Cat.: E2611, New England Biolabs).

For IgG1 and the IgG1-BsAb, a silenced Fcγ region harboring the Leu234Phe, Leu235Glu, and Asp265Ala mutations was used.

Soluble antigens or antibodies, or cell surface-anchored CD16b-NA1 or CD16b-NA2 were expressed in Expi-293F cells or ExpiCHO-S cells. Target soluble proteins were purified from cell culture supernatant using standard affinity chromatographic methods, such as Protein A resin (Cat.: 17519902, Cytiva) for Fc-fusion proteins or antibodies, or Nickel resin (Cat.: 88221, ThermoFisher Scientific) for His-tagged Fab and BiKE. They were then further polished by FPLC-SEC using a HiLoad 16/600 Superdex 200 pg column (Cat.: 28-9893-35, Sigma-Aldrich) and analyzed using SDS-PAGE and an HPLC system (Agilent Technologies) using a SEC mAb column from BioResolve (P/N: 186009439).

Cell lines and cell culture

Raji (Cat.: CCL-86), NALM6 (Cat.: CRL-3273), H929 (Cat.: CRL-3580), and SKBR3 (Cat.: HTB-30) cells were purchased from ATCC and cultured in their respective media: RPMI-1640 medium (Cat.: 11875–085, ThermoFisher Scientific), or McCoy’s 5A medium (Cat.: 30–2007, ATCC) containing 10% fetal bovine serum (FBS, Cat.: 10437028, ThermoFisher Scientific). The FBS with ultra-low IgG was purchased from ThermoFisher Scientific (Cat.: 16250078). Primary NK cells were purchased from Astarte Biologics (Cat.: PB56NV–2, now Charles River) and ALLCELLS (Item#260370.9). Expi293F (Cat.: A14635) and ExpiCHO-S (Cat.: A29133) cells were purchased from ThermoFisher Scientific (Carlsbad, CA) and cultured in their respective medium (Cat.: A14351–01 and Cat.: A29100–01, respectively) also from ThermoFisher Scientific.

ELISA to determine the binding of antibody clones to CD16

ELISAs were carried out by coating the 96-well high-binding microplates (Cat.: 655061, Greiner) with antigens or antibodies at 1 or 5 µg/mL in 100 mM carbonate-bicarbonate buffer overnight at 4°C. Wells were then blocked with 1× phosphate-buffered saline (PBS) containing 3% bovine serum albumin (BSA) for 1 h at room temperature. For screening antibody clones from the yeast-displayed human antibody libraires, 20 µl yeast clone culture media containing secreted Fab were added to 30 µl 2× binding buffer (0.5 M HEPES, 0.5 M NaCl, 5% BSA, pH 7.5), and the plates incubated for 1 h at room temperature. Bound Fab was detected with horseradish peroxidase (HRP)-labeled goat anti-human kappa or lambda light chain antibody (Cat.: A80-115A or 80-116A, respectively, Bethyl Laboratories). For determining binding of purified antibody to recombinant CD16, a serial dilution of antibody (when coated with CD16) or biotinylated CD16 (when coated with antibody) was added to the relevant wells, which were then detected with HRP-labeled goat-anti-human Fcγ fragment antibody (Cat.: 109-035-008, Jackson ImmunoResearch) or with HRP-labeled streptavidin (Cat.: 016-030-084, Jackson ImmunoResearch), respectively. The colorimetric reaction was detected as the reading at OD450 nm using a FilterMax F5 plate reader (Molecular Devices). Specific binding was plotted and curve fitted using GraphPad Prism software (GraphPad, La Jolla).

Determination of binding kinetics and competition using BLI

The binding affinity (K_D) of anti-CD16 antibody clones was determined with a Gator Prime System (Gator Bio, Palo Alto, CA), which performs the label-free kinetic characterization of biomolecular binding interactions. A Gator Anti-HIgG Fc biosensor (Cat.: PL168–160003) was used to capture IgG1 antibody or IgG4 bivalent engager to detect its binding to CD16, or a streptavidin biosensor (Cat.: 160002) to capture the biotinylated CD16 antigen for detection of binding to monovalent antibodies, e.g., Fab, monovalent BiKE. The biosensor was loaded with the first protein (ligand) until about 1 nm shift in the binding buffer (1× PBS containing 0.002% Tween-20, 0.2% BSA, pH7.4), which was then associated with serial dilution of the second protein (analyte) for 60–120 s, followed by dissociation for 180–300 s. Human polyclonal IgG (hIgG, Cat.: 16-16 -090,707, Athens Research & Technology) at 10 mg/mL was added to the flow solution when testing the effect of hIgG on antibody binding to CD16a. Additionally, two biosensors loaded either with or without the first molecule were associated with no analyte or the analyte at the lowest concentration, respectively, to serve as negative binding controls. After disassociation, the Gator system calculated K_on, K_off, and K_D values with its built-in software.

For the epitope binning assay, two streptavidin biosensors were first loaded with biotinylated CD16a-F until about a 1 nm shift in refractive index was reached. They were then associated with the first IgG1 clone with a silenced Fcγ or the murine mAb, 3G8, at 100 nM for 300 s to reach the saturation of binding. The loaded sensors were then immediately dipped into the second antibody solution, either more of the first antibody clone, a second antibody clone, or 3G8 at 100 nM, for 180 s to detect if additional antigen:antibody association was observed. Lack of further association of the first antibody in this extended period indicates that saturation of antigen:1^st-antibody binding has been reached, whereas any further association upon addition of the second antibody to the biosensor suggests that the first antibody and the second antibody do not share the same binding epitope. The pan anti-CD16 antibody, 3G8 (Cat.: 302002, BioLegend) was used as a reference clone; it also serves as a known binder of the Fcγ binding site on CD16.

Cell binding assays and flow cytometric analysis

Primary NK cells, neutrophils, or Expi293-F cells transfected with pCMV3-C expression vector encoding respective CD16 were seeded in 96-well U-shape microplate (Cat.: 650101, Greiner) in the binding buffer (1×PBS containing 0.5% BSA, pH7.4) at a density of 1.0–5.0×10⁵/well. It was then added with a serial dilution of antibody in the absence or presence of 10 mg/mL of hIgG and incubated for 30 min at 4°C. After repeated washing with binding buffer, the relevant detection antibodies were added, namely fluorescein isothiocyanate (FITC)-labeled goat anti-human Fc F(ab)2 fragment (Cat.: A24477, Invitrogen) for the detection of IgG1 and IgG4 antibody formats, Alexa Fluor 647-conjugated rabbit anti-His-Tag mAb (D3I1O, Cat.: 14931, Cell Signaling Technology) for the Fab, BiKE, or IgG-like BsAb, which each contain a C-terminal His-tag, or allophycocyanin (APC)-labeled streptavidin (Cat.: 016-130-084, Jackson ImmunoResearch) for detection of biotinylated antibodies. The FITC-labeled 3G8 antibody (Cat.: MHCD1601, ThermoFisher Scientific) was used as a reference binding control. The bound detection reagent was then detected using a 96-well IntelliCyt® iQue Screener PLUS cytometer and analyzed with the ForeCyt Software (Sartorius). Curve fitting of the MFI values vs concentration was performed using Prism software.

Isolation of NK cells

Peripheral blood mononuclear cells (PBMCs) were isolated from the whole blood of healthy donors (obtained from the San Diego Blood Bank) by density gradient centrifugation using Lymphoprep^TM (Cat.: 07851/07861, Stemcell Technologies) and SepMate^TM-50 (Cat.: 85450, Stemcell Technologies). NK cells were negatively purified from the PBMCs with an NK Cell Isolation Kit (Cat.: 130-092-657, Miltenyi Biotec) according to the manufacturer’s protocol. The purity of the NK cell population was confirmed by staining for the presence of CD56 (Cat.: MHCD5604, ThermoFisher Scientific) and CD16 (Cat.: MHCD1601, ThermoFisher Scientific) and the absence of CD3 (Cat.: 555329, BD Biosciences), which showed the cells to be more than 80% CD56 and CD16 positive (Figure S7A), and CD3 negative (data not shown).

The genotyping of CD16a-F and V variants was carried out by PCR using the following primers: CD16a-Forward: 5’-ACATATTTACAGAATGGCAAAGG-3’ and CD16a-F-Reverse: 5’-TGAAGACACATTTTTACTCCCAAA-3 for amplification of CD16a-F variant. CD16a-Forward and CD16a-V-Reverse: 5’- TGAAGACACATTTTTACTCCCAAC-3’ for amplification of CD16a-V variant. The amplification of the respective PCR fragment indicates the presence of the respective CD16a variant. The amplified PCR product of primers: CD16a-Forward and CD16-Reverse: 5’- GGTGATGTTCACAGTCTCTG-3’ were sequenced to further confirm the genotypes of CD16a of the donor. The genomic DNA of individual subjects was extracted from the blood sample using Puregene Blood Core Kit A (Cat.: 1042604, Qiagen). The PCR was performed in a reaction mixture (25 μl) containing MgCl₂ (1.5 mM), deoxyribonucleotide triphosphates (dNTP; 0.2 mM), and 1 units of JumpStart Taq DNA Polymerase (Cat.: D9307, Sigma) under the following conditions (30 cycles): denaturation at 94°C for 30 sec; and annealing for 30 sec at 60°C and extension at 72°C for 30 sec.

Isolation of neutrophils

Human neutrophils were isolated from human blood samples obtained from the San Diego Blood Bank using an EasySep Neutrophil Isolation kit (Cat.: 19666, Stemcell Technologies). Staining an aliquot with APC-labeled clone HI98, an anti-CD15 clone (the sialyl LewisX glycan marker; Cat.: 301908, BioLegend) and a phycoerythrin (PE)-labeled CD16 clone, 3G8, raised against neutrophils that recognizes both CD16a and CD16b as a positive control (Cat.: 302056, BioLegend) confirmed that the neutrophil population expressed CD16b.

The genotyping of CD16b-NA1 and NA2 was carried out by multiplex PCR with the following primers: CD16b-NA1-F: 5’-CAGTGGTTTCACAATGAGAA-3’; CD16b-NA2-F: 5’-CAATGGTACAGCGTGCTT-3’; CD16b-R: 5’-ATGGACTTCTAGCTGCAC-3’,⁵⁴ to amplify an NA1 fragment of 141 bp, and NA2 fragment of 219 bp, respectively. The primers, L19-F: 5’-ATTCTTGTGCCCTGATTTGT-3’ and L19-R: 5’-CCAGACTATACAAGACACCA-3’ were used to amplify an RPL19 (ribosome protein L19) fragment of 340 bp. The PCR was carried out as described above.

Cytotoxicity and ADCC reporter assays

In vitro cytotoxicity assays were carried out by measuring the viable target cells labeled with carboxyfluorescein succinimidyl ester (CFSE, Cat.: C34554, Invitrogen) and 7-amino-actinomycin D (7-ADD; Cat.: 00-6993-50, ThermoFisher Scientific). Briefly, tumor cells were first pre-loaded with CFSE according to the manufacturer’s recommended protocol and seeded in 96-well tissue culture treated assay plates (Cat.: 3610, Corning Incorporated) at 20,000 cells/well in RPMI-1640 medium containing 10% ultra-low IgG FBS. NK cell engager at serial dilutions was added to the wells followed by primary NK cells purified from healthy donors were added at 100,000 cells/well to give a final effector:target (E:T) ratio of 5:1. The plates were incubated for 20 h at 37°C in a humidified 5% CO₂ incubator. CountBright Absolute Conting Beads (Cat.: C36950, Invitrogen) were added at 10,000 beads/well as a sample loading control. The dead cells were then labeled with 7-ADD to exclude the nonviable cells, and viable cells were detected with a 96-well IntelliCyt® iQue Screener PLUS and analyzed with the ForeCyt Software (Sartorius). The percentage of cell death was determined by the loss of viable cells compared to the number of viable cells in the presence of NK cells without the NK cell engagers. The cytotoxicity of NK cells was calculated based on the number of live target cells (negative staining of 7-ADD) within the population of CFSE⁺ cells and in vitro potency (eEC₅₀) was determined by fitting the nonlinear regression model to sigmoidal dose-dependent curves (variable slope) using GraphPad Prism software (GraphPad, La Jolla, CA).

The ADCC reporter assay was carried out using the ADCC Reporter Bioassay and Core Kit from Promega (Cat.: G7010). Briefly, tumor cells were first seeded in 96-well tissue culture treated assay plates (Cat.3610, Corning Incorporated) at 25,000 cells/well in RPMI-1640 medium containing 10% ultra-low IgG FBS. Serial dilutions of NK cell engager were added in the presence or absence of hIgG at a final concentration of 10 mg/mL. Jurkat-CD16a(V)-NFAT reporter cells were added at 75,000 cells/well with an E:T cell ratio of 3:1. The plates were incubated for 20 h at 37°C in a humidified 5% CO₂ incubator. Bio-Glo Luciferase assay substrate was added to wells, incubated for 5–30 min, and then the luminescence intensity detected with a FilterMax F5 plate reader (Molecular Devices). The luminescence intensity readings (Relative Light Unit) were plotted, and curve fitted using GraphPad Prism software (GraphPad, La Jolla, CA).

Cytokine release assay

NALM6 cells were seeded in tissue culture-treated 96-well plates (Cat.: 3879, Corning Incorporated) at a density of 10,000 cells/well. NK cell engagers were added at a final concentration of 10 nM and incubated at 37°C for 30 min. NK cells were then added at E:T ratio of 5:1, and the plates were incubated for 24 h at 37°C and in a 5% CO₂ atmosphere in a humidified incubator. Plates were centrifuged at 1,000×g for 5 min, and cell culture supernatants were harvested for quantification of human IFN-$\mathit{\gamma }$ (Cat.: DIF50C, R&D Systems) and human TNF (Cat.: DTA00D, R&D Systems) according to the manufacturer’s protocols.

Funding Statement

The author(s) reported that there is no funding associated with the work featured in this article.

Abbreviations

ADCC

antibody-dependent cell-mediated cytotoxicity

ADCP

antibody-dependent cellular phagocytosis

BCMA

B-cell maturation antigen

BiKE

bispecific killer engager

BLI

biolayer interferometry

BsAb

Bispecific antibody

CD

Cluster of differentiation

CD16a

Cluster of differentiation 16a

CD16b

Cluster of differentiation 16b

CHO

Chinese hamster ovary

ECD

extracellular domain

ELISA

enzyme-linked immunosorbent assay

Fab

antibody binding fragment

FBS

fetal bovine serum

Fc

fragment, crystallizable

FcR

Fc receptors

Fv

variable fragment

GPI

glycosylphosphatidylinositol

Her2

epidermal growth factor receptor 2

hIgG

human polyclonal IgG

IFN-$\mathit{\gamma }$

interferon-gamma

K_D

equilibrium dissociation constant

mAb

monoclonal antibody

NA

neutrophil-specific antigen system NA

NK

natural killer

PBMC

peripheral blood mononuclear cell

PBS

phosphate-buffered saline

PCR

polymerase chain reaction

scFv

single-chain variable fragments

SDS-PAGE

sodium dodecyl sulfate-polyacrylamide gel electrophoresis

SEC

size-exclusion chromatography

TAA

tumor-associated antigen

TNF

tumor necrosis factor

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/19420862.2024.2381261