Background

Natural killer (NK) cells are innate lymphocytes that use germline-encoded receptors to interrogate cells of the body for ligands that are either aberrantly upregulated or downregulated to mediate natural cytotoxicity. Human NK cells also target cells by recognizing cell-bound IgG antibodies via their Fc receptor CD16A (FcγRIIIA), inducing antibody-dependent cell-mediated cytotoxicity (ADCC).1 Due to their broad antitumor effector functions, there has been a growing emphasis on invoking endogenous NK cells in patients with cancer and the adoptive transfer of autologous or allogeneic NK cells.2–4 Key challenges in using NK cells for cancer therapies are that they compose a minor population of circulating lymphocytes and their persistence is limited following adoptive transfer.5 6 Cytokines such as IL-15 are commonly used to promote human NK cell priming, cytotoxicity, and expansion ex vivo and in patients.7

NK cell activation induces the activity of the membrane-associated protease ADAM17 (a disintegrin and metalloproteinase-17),8 9 which has a variety of substrates.10 11 Their cleavage typically occurs in a cis manner at an extracellular location proximal to the cell membrane, referred to as ectodomain shedding.10 CD16A on NK cells is a well-described ADAM17 substrate that is rapidly cleaved following NK cell activation as a negative feedback mechanism10, and upon its engagement of cell-bound antibodies, as a detachment process.12

We have reported that blocking ADAM17 function with a monoclonal antibody (mAb) markedly enhanced IL-15-driven NK cell expansion ex vivo and in vivo.9 In this study, we sought to elucidate the underlying mechanism of this process. We show that Medi-1, a fully human IgG1 mAb that selectively blocks ADAM17 function, is engaged by CD16A while blocking its shedding. This signaling synergizes with IL-15 to enhance CD137 upregulation and NK cell proliferation when in the presence of peripheral blood mononuclear cell (PBMC) accessory cells or an anti-CD137 agonist mAb.

MethodsAntibodies

The anti-human ADAM17 mAb MEDI3622 has been previously described.13 The following Fc variants were generated from its nucleotide sequence. Human isotypes IgG1 and IgG4 were generated by Syd Labs (Natick, Massachusetts, USA) and are referred to as Medi-1 and Medi-4, respectively. The mutations Pro329Gly, Leu234Ala, and Leu235Ala (PGLALA) were engineered into the Fc region of Medi-1 (Syd Labs) to impair FcγR binding and is referred to as Medi-PGLALA. The Fc region of Medi-1 was proteolytically removed (SouthernBiotech, Birmingham, Alabama, USA) and is referred to Medi-F(ab′)2. The Trispecific Killer Engager (TriKE) containing a human IL-15 moiety and humanized camelid single-domain antibodies against CD16A and B7H3 (cam16-IL15-camB7H3), as well as a Targeted ADAM17 Blocker-16 (TAB16), containing a single-chain fragment variable (scFv) from Medi-1 and a humanized camelid single-domain antibody against CD16, were generated similar to as previously described.14 All commercially available antibodies are listed in table 1. Any antibodies used in functional assays that contained a preservative, such as sodium azide (NaN3), were subjected to a Zeba Spin Desalting Column 40K (Thermo Fisher Scientific, Waltham, Massachusetts, USA), according to the manufacturer’s instructions, and resuspended in PBS buffer without Ca+2 and Mg+2 (Gibco, Waltham, Massachusetts, USA).

Table 1

Commercial antibodies

Flow cytometric analyses

NK cell phenotypic analyses were performed as previously described.9 Briefly, fluorescence minus one (FMO) and appropriate isotype-matched antibodies were used for controls. An FSC-A/SSC-A plot was used to set an electronic gate on leukocyte populations, and an FSC-A/FSC-H plot was used to set an electronic gate on single cells. The cell viability dye Ghost Dye Red 710 (Tonbo Bioscience, San Diego, California, USA) was used to distinguish live vs dead cells per the manufacturer’s instructions. Cell samples were acquired on a BD FACSCelesta Cell Analyzer (BD Biosciences, San Jose, California, USA) and analyzed using FlowJo V.10.9.0 Software (BD Bioscience).

Cell isolation and culture

PBMCs were obtained from whole blood, collected from healthy consenting adults at the University of Minnesota (IRB protocol # 9708M00134) in sodium heparin tubes (BD Bioscience), and from plateletpheresis products from Innovative Blood Resources (St. Paul, Minnesota, USA). PBMCs were isolated using Lymphocyte Separation Medium (Promocell, Heidelberg, Germany). NK cells were enriched using a negative selection human NK Cell Isolation Kit (Miltenyi Biotec, Bergisch, Germany) or (StemCell Technologies, Vancouver, BC, Canada). Isolated NK cells were ≥90% pure, as determined by CD56+ CD3− staining and flow cytometry. PBMCs or enriched NK cells, either unlabeled or labeled with CellTrace Violet (CTV) Cell Proliferation Dye (ThermoFisher Scientific) as per the manufacturer’s instructions, were cultured for up to 7 days at 37°C in 5% CO2 in RPMI 1640 media (Gibco) supplemented with 10% Heat Inactivated FBS (Gibco) and 1x Anti-Anti (Gibco), in the presence or absence of rhIL-15 or rhIL-2 (R&D Systems, Minneapolis, Minnesota, USA) and the indicated antibodies. NK cell proliferation was analyzed via flow cytometry and a division index was calculated using FlowJo.

SKOV-3 cells (HTB-77), an ovarian adenocarcinoma cell line, K562 cells (CCL-243), a chronic myelogenous leukemia cell line, and NK-92 cells (CRL-2407), a malignant non-Hodgkin’s lymphoma NK cell line, were purchased from ATCC (Manassas, Virginia, USA). SKOV-3 cells stably expressing NucLightGreen (SKOV-3/NLG) were generated as previously described.15 SKOV-3 cells were maintained in McCoy’s 5A media (Gibco) supplemented with 10% HI FBS and 1X Anti-Anti. K562 cells were maintained in 10% RPMI 1640 media, as described above. NK-92 cells were maintained in MEM Alpha media (Gibco) with 12.5% HI FBS (Gibco), 12.5% Horse Serum (Gibco), 1x Anti-Anti, and 0.1 mM 2-Mercaptoethanol (Gibco). All cell lines were routinely tested for mycoplasma with the MycoAlert Mycoplasma test kit (Lonza, Basel, Switzerland).

Cellular assays

PBMCs were cultured with SKOV-3 or K562 cells at an effector:target (E:T) ratio of 1:1 at 37°C and 5% CO2 for 5 hours. CD107a and IFN-γ staining were determined as previously described.8 To measure cell cytotoxicity in real-time, SKOV-3/NLG cells were plated at a density of 4×103 cells/well in a 96-well flat bottom, tissue culture-treated plate (Falcon, cat# 353072, Corning, NY) 24 hours prior to starting the assay. Enriched NK cells were added at the indicated E:T ratios in RPMI 1640 media at 37°C in 5% CO2. Fluorescent images of live cells were obtained hourly for 7 days using an IncuCyte SX3 live cell imaging and analysis system (Sartorius Göttingen, Germany), as previously described.15

NK-92 cells expressing human CD16A were stimulated with phorbol 12-myristate 13-acetate (PMA), as described previously.16 Briefly, NK-92/CD16A cells (3×105/mL) were treated with or without 10 ng/mL PMA (Sigma Aldrich, St. Louis, Missouri, USA) and 1 µg/mL ionomycin (AdipoGen Life Sciences, San Diego, California, USA) for 2 hours at 37°C in the presence or absence of Medi-1, Medi-4, Medi-PGLALA, Medi-F(ab′)2, D1(A12), or an isotype-matched negative control antibody at the indicated concentrations. The cells were then washed twice with PBS containing 0.05% FBS, treated with cell viability dye, and stained for CD16A. All samples were fixed with 2% paraformaldehyde in PBS without Ca+2 and Mg+2 and subjected to flow cytometric analyses.

Immunoblotting

Enriched NK cells were left untreated or treated with Medi-1 or an isotype-matched negative control antibody (5 µg/mL) at 37°C in serum-free RPMI 1640 (Gibco). Cells were then washed, lysed with detergent, and prepared for immunoblotting as previously described.17 Briefly, ~2.5×105 cells were run in each lane of a 7% NuPAGE tris-acetate gel (Invitrogen, Carlsbad, California, USA) and transferred to Immobilon-FL polyvinylidene difluoride membrane (EMD Millipore, Burlington, Massachusetts, USA). REVERT Total Protein Stain (LI-COR Biosciences, Lincoln, Nebraska, USA) was used for normalization to the total protein content of each sample. Membranes were treated with Intercept (TBS) Blocking Buffer (LI-COR Biosciences), incubated with the indicated primary antibodies, and then incubated with the appropriate near-infrared secondary antibodies. Blots were visualized using an Odyssey CLx near-infrared imager (LI-COR Biosciences), and signals were background-subtracted using ImageStudio Software (LI-COR Biosciences). After correction for whole-lane protein content, samples were normalized to the t=0 signal.

Statistical analysis

Data are presented as the mean±SD or, for immunoblotting, SEM. Compiled in vitro data are from at least three independent experiments using separate donors. Statistical analysis was conducted, and significance was determined using GraphPad Prism Software V.10.0.3 for IOS or V.10.2.3 (403) for Windows (La Jolla, California, USA). Comparison between two treatments was computed using the paired two-tailed Student’s t-test. Comparison among three or more treatments was made using one-way or two-way analysis of variance, as appropriate, followed by the indicated post hoc test. All data presented as geometric mean fluorescent intensity (MFI) were log-transformed before statistical analysis. Outlier analyses were performed on immunoblotting data using unbiased ROUT with Q=1% followed by a mixed-effects analysis approach with correction for multiple comparisons/Fisher’s least significant difference test. The area under the curve was calculated to compare the killing kinetics of the live imaging in vitro killing assays. The data generated in this study are available on request from the corresponding author.

ResultsMedi-1 treatment selectively amplifies NK cell proliferation in the presence of IL-15

It is well established that IL-15 promotes the proliferation of NK cells and T cells, especially in the presence of monocytes which transpresent IL-15.18 To determine the effects of the ADAM17 function-blocking mAb Medi-1 on IL-15-driven proliferation of these populations, PBMCs labeled with CTV were cultured with IL-15 at 10 ng/mL, a standard concentration used for ex vivo lymphocyte expansion,9 19 ±Medi-1 (5 µg/mL) for up to 7 days. As shown in figure 1A, IL-15 stimulation along with Medi-1 markedly enhanced NK cell proliferation, which was dose-dependent (online supplemental figure 1A), whereas Medi-1 treatment alone had a negligible effect on NK cell proliferation (figure 1A). An isotype-matched control for Medi-1 did not augment IL-15-driven NK cell proliferation (online supplemental figure 1B). Medi-1 treatment also increased NK cell proliferation at a low concentration of IL-15 (1 ng/mL) (online supplemental figure 1C). Enhanced IL-15-driven NK cell proliferation by Medi-1 caused a corresponding increase in the proportion of these cells in the PBMC population after 7 days of culture (figure 1B), resulting in a 4.5-fold expansion, with no effect on cell viability, compared with a 1.5-fold expansion exhibited by cells stimulated with IL-15 alone (figure 1C). In contrast, T cells within the IL-15-stimulated PBMC population were only marginally affected by Medi-1 treatment, with NKT cells showing minor augmentation of proliferation while CD4 and CD8 T cells showed no benefit (figure 1D). Hence, the Medi-1 effect was primarily NK cell-specific and it increased their sensitivity to IL-15 stimulation.

Figure 1

Medi-1 treatment selectively increases NK cell proliferation by IL-15. Freshly isolated PBMCs were labeled with CellTrace Violet (CTV) to track cell division and cultured for 7 days±IL-15 (10 ng/mL)±Medi-1 (5 µg/mL). (A) CD56+ CD3− NK cells were analyzed for CTV dilution by flow cytometry. Representative data are shown (left panels). Division index (right panel), mean±SD, n=6 donors. (B) Percentage of NK cells for the indicated conditions and time points. Data are representative of 6 independent experiments. (C) Fold increase in NK cells and viability after 7 days of expansion for the indicated conditions. Mean±SD, n=6 donors. (D) The indicated lymphocyte populations were analyzed for CTV dilution by flow cytometry. NK cells (CD56+ CD3−), NK T cells (CD56+ CD3+), CD8 T cells (CD56− CD3+ CD8+) and CD4 T cells (CD56− CD3+ CD4+). Data comparing NK cells to T cells are representative of at least three independent experiments using leukocytes from separate donors. **p<0.01; ns, not significant. Statistical significance was determined by paired two-tailed Student’s t-tests. NK, natural killer; PBMC, peripheral blood mononuclear cell.

D1(A12) is another human IgG1 mAb that blocks ADAM17 function20 (online supplemental figure 2). Interestingly, D1(A12) treatment of PBMCs marginally, but not significantly, enhanced NK cell proliferation by IL-15 stimulation (online supplemental figure 3). This could be due to distinct aspects of the Fab and/or Fc regions of D1(A12) or its targeted epitopes. The variable heavy and light chains of this engineered mAb were designed to recognize distinct epitopes in different regions of ADAM17.20 Moreover, the strength of CD16A binding to IgG1 is affected by glycosylation and amino acid polymorphisms in its Fc region,21 which could vary between D1(A12) and Medi-1.

IL-15 is used therapeutically in various forms, including its incorporation into multiengager complexes, such as a TriKE.22 To assess whether Medi-1 treatment enhanced NK cell proliferation by IL-15 in other contexts, we used it in conjunction with a TriKE consisting of two humanized camelid nanobodies against CD16 and the tumor-associated antigen B7-H3 attached to IL-15, referred to as cam16-IL15-camB7H3. We observed that Medi-1 treatment enhanced NK cell proliferation mediated by the TriKE as well (online supplemental figure 1D). Moreover, the effects of Medi-1 were not restricted to IL-15, as the presence of Medi-1 also enhanced IL-2-driven NK cell proliferation (online supplemental figure 1E).

NK cells undergoing IL-15-driven proliferation in the presence or absence of Medi-1 express similar functional markers

Overstimulation of NK cells with IL-15 or other stimuli can cause changes in their phenotypic profile, including the downregulation of activating receptors and the upregulation of inhibitory receptors.23 Stimulation of NK cells with a single 10 ng/mL dose of IL-15 for 7 days does not induce their exhaustion.19 To assess whether this is altered by Medi-1 treatment, PBMCs were stimulated in this manner for 7 days±Medi-1 and then NK cells were compared for their staining levels of well-described activating and inhibitory receptors by flow cytometry. As shown in figure 2A, we observed no statistical differences in the percentage of NK cells expressing NKp44, NKp46, NKG2D, KIR2DL1, KIR2DL2/L3, KIR3DL1, or NKG2A when stimulated with IL-15±Medi-1. The percentage of NK cells expressing CD16A, an ADAM17 substrate,10 was slightly but significantly increased in the presence of Medi-1 (figure 2A). The geometric MFI of positively staining NK cells for NKG2D, KIR2DL1, KIR2DL2/L3, and NKG2A was also marginally increased for IL-15-stimulated NK cells in the presence of Medi-1 (figure 2A). The inhibitory receptors LAG-3, TIGIT, and TIM-3 have been reported to be upregulated in expression and activity in exhausted T cells.23 24 These markers can be upregulated on activated NK cells, but whether their expression signifies exhaustion is less clear.23 LAG-3, TIGIT, and TIM-3 staining levels on resting NK cells varied; therefore, we included FMO controls in the flow cytometric plots to better distinguish changes in their staining during NK cell activation. When NK cells were cultured in the presence of IL-15 and Medi-1, the percentage of LAG-3+ cells was significantly increased, as was the geometric MFI of cells positively staining for LAG-3 and TIGIT (figure 2B).

Figure 2

Proliferating NK cells treated with IL-15±Medi-1 express similar functional markers. Freshly isolated PBMCs were cultured in the presence or absence of IL-15 (10 ng/mL) and Medi-1 (5 µg/mL), as indicated. At day 0 or day 7, CD56+ CD3− NK cells were analyzed for their expression levels of the indicated markers by flow cytometry. (A) Activating receptors (top panels) and inhibitory receptors (bottom panels). Representative histogram plots (left panels). y-axis=cell number. Cumulative data showing the percentage of NK cells positively staining for each marker (mean is indicated by the horizontal line) and their geometric MFI (right panels). For the latter, data were log-transformed prior to analysis. n=6 donors. (B) Exhaustion markers. Representative histograms are shown in the left panel. Unstimulated=unstimulated cells stained at day 7. Fluorescence minus one (FMO) of cells at day 0. Cumulative data are shown as the percentage of NK cells positively staining for each marker and the geometric MFI of staining. For the latter, data were log-transformed prior to analysis. n=4–5 donors. (C) PBMCs were cultured in IL-15±Medi-1 for 7 days, washed, and then incubated with K562 or SKOV-3 cells for 5 hours (effector:target ratio=1:1). PBMCs were stained for CD56, CD3, CD107a, and IFN-γ and examined by flow cytometry (CD56+ CD3− cells are shown). Representative histogram plots (left panels) and cumulative data (right panels) are shown, n=3–4 donors. *p<0.05; **p<0.01; ****p<0.0001; ns, not significant. Statistical significance was determined by a one-way ANOVA with a Dunnett post hoc test. ANOVA, analysis of variance; MFI, mean fluorescent intensity; NK, natural killer; PBMCs, peripheral blood mononuclear cells.

To assess the significance of the modest increase in expression of certain activating and inhibitory receptors by NK cells treated with IL-15 and Medi-1, we evaluated the functional readouts of degranulation and cytokine production via CD107a and IFN-γ detection, respectively, which are diminished upon NK cell exhaustion.23 PBMCs were stimulated with IL-15±Medi-1 for 7 days and then co-cultured with the ovarian cancer cell line SKOV-3 or the myelogenous leukemia cell line K562 for 5 hours. With both tumor cell targets, NK cells expanded by IL-15±Medi-1 upregulated CD107a and IFN-γ at equivalent levels (figure 2C), though their response to K562 cells was much more robust. Hence, though Medi-1 treatment augmented IL-15-driven NK cell proliferation and increased the expression of certain inhibitory markers, this did not lead to broad changes in well-recognized functional phenotypic markers.

Early activation markers distinguish IL-15-stimulated NK cells treated with Medi-1

We next examined the expression of the early activation markers CD69, CD25 (IL-2Rα), and CD137 (4-1BB) on NK cells following the stimulation of PBMCs with IL-15±Medi-1. Upon IL-15 stimulation, NK cells rapidly and uniformly upregulated CD69, which remained elevated for the 7-day culture (figure 3). IL-15 stimulation along with Medi-1 increased CD69 upregulation on day 1, though its expression modestly decreased at the remaining time points (figure 3). CD25 levels on NK cells following IL-15 stimulation were increased at all time points compared with unstimulated NK cells, and this was markedly enhanced in the presence of Medi-1 on days 1–4 (figure 3). Similarly, Medi-1 treatment significantly enhanced CD137 upregulation on NK cells by IL-15 (figure 3). Hence, CD25 and CD137, but not CD69, demonstrated distinct expression patterns on NK cells treated with IL-15 and Medi-1 when compared with IL-15 treatment alone, which was apparent by day 1 of stimulation (figure 3).

Figure 3

Medi-1 treatment augments the upregulation of early activation markers on NK cells. Freshly isolated PBMCs were treated as described in figure 2. CD56+ CD3− NK cells were analyzed for their CD69, CD25, and CD137 expression levels by flow cytometry on days 0, 1–4, and 7 of culture. Representative histogram plots (left panels). y-axis=cell number. Cumulative data (right panels) are shown as an averaged log-transformed geometric MFI fold increase from day 0. Mean±SD, n=4 donors. *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001; ns, not significant. Statistical significance was determined by a two-way ANOVA with Tukey post-hoc test. ANOVA, analysis of variance; MFI, mean fluorescent intensity; NK, natural killer; PBMCs, peripheral blood mononuclear cells.

The kinetics of CD69, CD25, and CD137 upregulation were also examined using enriched NK cells treated with IL-15±Medi-1 for up to 7 days. Again, CD25 and CD137 demonstrated a distinct pattern of upregulation by IL-15-stimulated NK cells treated with Medi-1 (figure 4A). Since CD137 achieved peak expression by day 1, we examined earlier time points of stimulation at 4, 8, and 18 hours. CD25 and CD69 demonstrated significantly higher levels of upregulation as early as 4 hours after NK cell treatment with IL-15 and Medi-1 compared with NK cells stimulated with IL-15 alone (figure 4B). Medi-1 treatment significantly increased CD137 as early as 8 hours after stimulation relative to NK cells treated with IL-15 alone (figure 4B). These data demonstrate that Medi-1 treatment played a direct role in enhancing NK cell activation by IL-15 (ie, independent of other PBMC populations).

Figure 4

Direct effects of Medi-1 on enriched NK cells. Freshly isolated NK cells (>90% enriched) were treated and analyzed as described in figure 2. (A) CD56+ CD3− NK cells were analyzed for their expression levels of CD69, CD25, and CD137 by flow cytometry on days 0, 1–4, and 7 of culture. (B) CD56+ CD3− NK cells were similarly examined at hours 0, 4, 8, and 18. Data are shown as the averaged log-transformed geometric MFI fold increase from day 0 (A) hour 0 (B). Mean±SD, n=4 donors. *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001. Statistical significance was determined by a two-way ANOVA with a Šídák (A) or Tukey post hoc test (B). ANOVA, analysis of variance; MFI, mean fluorescent intensity; NK, natural killer.

The Fc region of Medi-1 enhances NK cell activation by IL-15

The distinct expression patterns of CD25 and CD137 by IL-15-stimulated NK cells when treated with Medi-1 were indicative of its synergetic effect. CD16A is the main IgG Fc activating receptor expressed by human NK cells and it exclusively mediates ADCC.10 As Medi-1 is a human IgG1 mAb, it can potentially be engaged by CD16A on binding NK cells. To test whether the Fc region of Medi-1 contributed to its synergistic effect, we generated several other Fc variants of the mAb, including Medi-F(ab′)2 in which its Fc region was removed; Medi-PGLALA containing the mutations Pro329Gly, Leu234Ala, and Leu235Ala that also abolish FcγR binding, but with less potential effects on antibody structure and function25; and Medi-4 containing the Fc region of IgG4, which binds to CD16A with~10 fold lower affinity than IgG1.26 The predicted rank order of CD16A binding affinity for the generated Medi Fc variants was Medi-1>Medi-4> Medi-PGLALA>/=Medi F(ab′)2. All versions of Medi, but not a human IgG control mAb, equivalently blocked CD16A downregulation on cell activation (online supplemental figure 2), demonstrating that their ability to block ADAM17 function was not altered. PBMCs were stimulated with IL-15±a Medi Fc variant at the same molar concentration for 24 hours. The expression of CD25 and CD137 was then determined by flow cytometry. As shown in figure 5A, IL-15-stimulated NK cells treated with Medi-1 versus the other Fc variants demonstrated significantly higher levels of these activation markers. The same effects were observed upon IL-15-driven NK cell proliferation (figure 5B).

Figure 5

Medi-1 synergizes with IL-15 stimulation through its Fc region. (A) Freshly isolated PBMCs were cultured with IL-15 (10 ng/mL)±molar equivalents of Medi-1, Medi-4, Medi-PGLALA, or Medi-F(ab′)2, as indicated. CD56+ CD3− NK cells were analyzed for their expression levels of CD25 and CD137 by flow cytometry at 18 hours of culture. Data are shown as the averaged log-transformed geometric MFI fold increase from time 0. Mean±SD, n=9 donors. (B) PBMCs were labeled with CTV and cultured for 7 days with IL-15±Medi-1, Medi-4, Medi-PGLALA, or Medi-F(ab′)2, as indicated. CD56+ CD3− NK cells were analyzed for CTV dilution by flow cytometry. Representative histograms (left panel) and division index (middle panel) are shown. Percentage of NK cells for the indicated conditions (right panel). Mean±SD, n=5 donors. (C) Enriched NK cells were cultured with no further treatment (NT), with Medi-1, or with isotype-matched negative control antibody for the indicated times. Immunoblotting shows pZAP-70 (Y319)/pSYK (Y352), direct targets of Src-family kinases on ligation of ITAM-coupled receptors50; total SYK protein is shown as a loading control. Representative (left panel) and cumulative data (right panel) are shown. Mean±SEM, three donors (two for isotype) used for a total of nine independent treatments (six for isotype). (D) PBMCs were treated as described in panel B with IL-15±Medi-1 or TAB16 at equivalent molarity (33 nM). Representative histograms (left panel) and division index (right panel) are shown. Mean±SD, n=3 donors. *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001. Statistical significance was determined by a one-way ANOVA with a Dunnett (A, D) or Tukey (B) post hoc test, or a two-way mixed-effects analysis with a Šídák post-hoc test (C). ANOVA, analysis of variance; CTV, CellTrace Violet; NK, natural killer; PBMCs, peripheral blood mononuclear cells.

The above findings indicate that CD16A engages the Fc region of Medi-1 when bound to ADAM17 on NK cells. CD16A non-covalently associates with the signaling adaptors FcRγ and CD3ζ.27 28 These immunoreceptor tyrosine-based activation motif (ITAM)-containing signaling adaptors become phosphorylated and associate with the ZAP-70 and SYK protein tyrosine kinases, which are also phosphorylated, leading to downstream signaling cascades for cell activation.29 We observed that treating NK cells with Medi-1 alone led to significantly higher levels of ZAP-70/SYK phosphorylation than NK cells treated with an isotype-matched negative control antibody or no antibody (figure 5C).

As an approach to directly engage CD16A and simultaneously block ADAM17 function, we generated a bispecific antibody engager that consists of a humanized camelid nanobody against CD16 and a scFv from Medi-1, referred to as a Targeted ADAM17 Blocker-16 (TAB16). PBMCs were labeled with CTV and cultured with IL-15 alone or with molar equivalents of Medi-1 or TAB16 for 7 days. TAB16+IL-15 induced significantly higher levels of NK cell proliferation than IL-15 alone and, interestingly, higher levels of proliferation than IL-15+Medi-1 (figure 5D).

A potential adverse effect of CD16A engagement of the Fc region Medi-1 would be the induction of NK cell degranulation and fratricide. For example, daratumumab, a human IgG1 mAb that recognizes CD38 on NK cells, has been shown to induce fratricide.30 Indeed, we observed impaired NK cell proliferation and a greatly decreased representation within the PBMC population when CTV-labeled PBMCs were cultured for 7 days with IL-15 and daratumumab, which was not seen with Medi-1 treatment (figure 6A), indicating it did not induce a critical level of fratricide or clearance by macrophage present in the culture. In other assays, enriched NK cells cultured with daratumumab for 24 hours demonstrated a significant upregulation in cell surface levels of CD107a, a sensitive marker of NK cell degranulation upon CD16A signaling,8 compared with control cells. This was not observed for NK cells cultured with Medi-1 (figure 6B). To assess the effects of Medi-1 treatment on the functional state of NK cells, we examined their natural cytotoxicity against tumor cells. For this assay, enriched NK cells were used to exclude antitumor activity by other leukocytes. NK cells were cultured with SKOV-3/NLG cells at various E:T ratios in the presence of IL-15±Medi-1, and tumor cell lysis was assessed by IncuCyte live cell imaging for 7 days. Interestingly, in the presence of Medi-1, NK cells mediated significantly higher levels of SKOV-3 killing (figure 6C). Taken together, our results show that the Fc region of Medi-1 contributed to the activation and proliferation of IL-15-stimulated NK cells while not inducing apparent adverse effects.

Figure 6

Medi-1 treatment of IL-15-stimulated NK cells does not induce fratricide or impair natural cytotoxicity. (A) PBMCs were labeled with CTV and cultured for 7 days with IL-15 (10 ng/mL)±Medi-1 or daratumumab at 5 µg/mL, as indicated. CD56+ CD3− NK cells were analyzed for CTV dilution by flow cytometry. Representative data are shown (left panel) as well as the percentage of NK cells for the indicated conditions (right panel). Mean±SD, n=4 donors. (B) PBMCs cells were cultured for 24 hours in IL-15±an isotype-matched negative control mAb (human IgG1), Medi-1 (5 µg/mL), or daratumumab at (5 µg/mL), as indicated. CD56+ CD3− NK cells were analyzed for their expression levels of CD107a by flow cytometry. Cumulative data are shown as geometric MFI±SD, n=7 donors. (C). Enriched NK cells were co-cultured with SKOV-3-NLG cells at the indicated E:T ratios for 7 days in the presence of IL-15 (10 ng/mL)±Medi-1 (5 µg/mL). Cytotoxicity was assessed by IncuCyte-based live cell imaging. The percentage of viable SKOV-3 cells was double normalized to SKOV-3 cells alone (left panel). The area under the curve (AUC) of the remaining SKOV-3 cells was calculated for each condition. Mean±SD, n=4. *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001. Statistical significance was determined by a one-way ANOVA with Tukey post hoc test (A, B) or two-way ANOVA with Šídák post-hoc test (C). ANOVA, analysis of variance; CTV, CellTrace Violet; E:T, effector:target; MFI, mean fluorescent intensity; NK, natural killer; PBMCs, peripheral blood mononuclear cells.

Enhanced IL-15-driven NK cell proliferation by Medi-1 involves CD137 and PBMC accessory cells

Circulating NK cells are composed of CD56 bright and dim populations, and CD16A is primarily expressed by the latter cells.31 It was mainly these cells that upregulated CD137 after stimulation with Medi-1 and IL-15 (online supplemental figure 4). It is well established that CD137 stimulation in T cells promotes their activation and proliferation.32 However, much less is known about the biological importance of CD137 in NK cells. Engagement of this coactivation receptor by agonistic antibodies or by CD137L-expressing artificial feeder cells induces NK cell proliferation.33 34 To investigate the role of CD137 in the augmented proliferation of Medi-1+IL-15-stimulated NK cells, we blocked its function. Others have reported that the addition of soluble CD137L to PBMC cultures can inhibit T cell proliferation by acting as a competitive inhibitor.35 36 We attempted this for NK cells using a recombinant human CD137L extracellular domain-Fc fusion protein (sCD137L-Fc). We found that sCD137L-Fc diminished NK cell proliferation by IL-15 as well as IL-15+Medi-1 (figure 7A), which occurred in a dose-dependent manner (online supplemental figure 5). A caveat of this approach is that sCD137L-Fc, when attached to NK cells, might be engaged by CD16A, reducing its interaction with Medi-1. However, the presence of sCD137L-Fc did not affect the enhanced upregulation of CD25 by NK cells treated with IL-15 and Medi-1 (figure 7B), indicating that Medi-1 engagement by CD16A was not affected. We could not assess CD137 staining since sCD137L-Fc competed with our anti-CD137 mAb used for flow cytometry (data are not shown), demonstrating its attachment to CD137.