Background

Natural killer (NK) cells are cytotoxic lymphocytes of the innate immune system . These cells express numerous germline-encoded activating and inhibitory receptors for assessing ligand levels on cells in the body to remove transformed or pathogen-infected cells.1 In addition, NK cells are directed to antigens on cellular targets through the recognition of antibodies to mediate antibody-dependent cell-mediated cytotoxicity (ADCC).2 Upon their activation, NK cells rapidly release cytolytic and apoptotic factors, as well as cytokines and chemokines that stimulate and recruit other leukocytes.3 Due to these assorted effector functions, there have been increasing clinical investigations into NK cells as an adoptive cell therapy (ACT) for cancer. Allogeneic NK cells have been a particular focus due to their off-the-shelf applications and differentiated safety profile compared with allogeneic T-cell therapies, including reduced graft versus host disease and cytokine release syndrome.1 4

ADCC by human NK cells is exclusively mediated by CD16A (FcγRIIIA).1 2 Inherent attributes of this IgG Fc receptor affect its binding affinity and avidity. For the latter, cell surface levels of CD16A undergo a rapid downregulation by a disintegrin and metalloproteinase-17 (ADAM17, CD156b) upon NK cell activation by CD16A signaling and various other stimuli as a negative feedback process.5 CD16A downregulation can also occur in the microenvironment of solid tumors, as has been reported in patients with ovarian cancer.6–8 Indeed, ADAM17 induction occurs under conditions of hypoxia,9 further linking its activity to the tumor microenvironment. CD16A allelic variants significantly affect receptor affinity for IgG. CD16A with a phenylalanine at position 158 binds to IgG1 with at least twofold lower affinity than CD16A with a valine at the same position.10 Of note is that approximately 80% of the population expresses a low affinity allele of CD16A.11–13

Clinical studies indicate that higher affinity and avidity interactions between CD16A and therapeutic monoclonal antibodies (mAbs) increases ADCC activity by NK cells.12 14–16 With the goal of augmenting ADCC potency by adoptive NK cell therapies, we generated the recombinant fusion FcγR CD64/16A for NK cell expression.17 Its extracellular region consists of human CD64 (FcγRI), the only high affinity IgG Fc receptor and mainly expressed by myeloid cell populations.18 CD64 binds to IgG1 with 30–100-fold higher affinity than CD16A depending on the CD16A allelic variant.10 CD64/16A contains transmembrane and cytoplasmic regions from CD16A. The latter associates with the signaling adaptors FcεRγ and CD3ζ and is a highly potent activating receptor in NK cells.19–21 CD64/16A also lacks the ADAM17 membrane proximal cleavage site of CD16A,17 which is intended to prevent its downregulation on NK cell thaw from cryopreservation, activation, proliferation, and by aberrant ADAM17 induction in the tumor microenvironment.22 The ability of CD64/16A to induce ADCC was initially investigated using NK-92 cells,17 a human NK cell line that lacks expression of endogenous FcγRs.23 Of interest is that due to its high affinity state, CD64/16 can stably bind to monomeric IgG1 and therefore provide a capturing element for antitumor therapeutic mAbs,17 22 which is referred to here as antibody arming.

Our approach in this study focused on induced pluripotent stem cell (iPSC) derived NK cells (iNK cells).24 Like human peripheral blood (PB) NK cells, iNK cells mediate their effector functions against tumor cells through granule release of perforins and granzymes, TRAIL and FasL production, and cytokine release, including interferon (IFN)-γ and tumor necrosis factor (TNF)-α.25 iNK cells have been shown to be equally or more effective than PB NK cells against various tumor cell targets including ovarian cancer cells in vitro and in xenograft mouse models.26 27 Other advantages of iNK cells are their homogeneity and clinically scalable production.28 Importantly, iPSCs are amendable to genetic modifications, allowing for the generation of multiplexed edited NK cells to create increasingly more functional, hypoimmunogenic, and persistent effector cells.25 29 We investigated antibody-armed iNK-CD64/16A cells targeting tumor antigens on ovarian cancer cell lines by in vitro and in vivo approaches. We demonstrate that these cells can be armed with different therapeutic mAbs immediately after expansion or on thaw from cryopreservation; antibody-armed iNK-CD64/16A cells thawed from cryopreservation are capable of continuous ADCC; and that antibody-armed iNK-CD64/16A cells can be repurposed to target new tumor antigens by addition of different therapeutic mAbs. These findings underscore the potential of iNK-CD64/16A cells as an off-the-shelf ACT strategy for multitumor antigen targeting to overcome tumor heterogeneity and antigen escape.

MethodsAntibodies

Antibodies used for the detection of CD64 (clone 10.1), CD3 (clone UCHT1), CD45 (clone 2D1), CD56 (clone HCD56), CD107a (clone H4A3), and CD16 (clone 3G8) were obtained from BioLegend (San Diego, California, USA). Therapeutic mAbs used were avelumab (Pfizer, New York, New York, USA), cetuximab (Lilly, Indianapolis, Indianapolis, USA), and trastuzumab (Genentech, San Francisco, California, USA). All immunophenotypic flow cytometry was performed using an FACSCelesta (BD Biosciences, San Jose, California, USA), and data was analyzed using FlowJo software (BD Biosciences). For controls, fluorescence minus one was used as well as appropriate isotype-matched antibodies since the cells of interest expressed FcRs.

Cell lines

The SKOV-3 (HTB-77) ovarian cancer cell line was purchased from American Type Culture Collection (Manassas, Virginia, USA) and were maintained in McCoy’s 5A media (Gibco, Waltham, Massachusetts, USA) supplemented with 10% fetal bovine serum (FBS) (Gibco) and 1× pen–strep (Gibco). SKOV-3 cells stably expressing firefly luciferase (SKOV-3-Luc), NucLightGreen (NLG), or NucLightRed (NLR) (Sartorius, Göttingen, Germany) were generated as previously described.30 The cancer cell lines OVCAR-4 (SCC258) and OVCAR-5 (SCC259) were purchased from MilliporeSigma (Burlington, Massachusetts, USA) and were maintained in Roswell Park Memorial Institute (RPMI)-1640 media (Gibco) supplemented with 10% FBS and 1× pen–strep. Cells were routinely tested for Mycoplasma with the MycoAlert Mycoplasma Test Kit (Lonza, Basel, Switzerland).

iNK and PB NK cell culture

iPSCs were generated by a transgene-free approach in feeder-free culture conditions that maintain pluripotency and genomic stability.24 iPSCs were engineered to express CD64/16A and an interleukin (IL)-15/IL-15Rα fusion (IL-15RF) protein by sequential lentiviral transduction, as previously described.17 31 For this preclinical study, transduced engineered pools instead of clonal lines were used and additional genomic stability analysis was not performed. The iPSCs were differentiated into iCD34 cells then iNK cells using established methodology.31 32 Differentiated iNK cells or PB NK cells were expanded for 2 weeks using irradiated K562-mbIL21-41BBL feeder cells.32 Briefly, this was done at a 1:2 NK cell:feeder cell ratio in B0 media:Dulbecco's Modified Eagle Medium (DMEM) and Ham’s F-12 (2:1 ratio) (Corning, Corning, New York, USA), 10% heat inactivated Human AB serum (Valley Biomedical, Winchester, Virginia, USA), 1% penicillin−streptomycin (Gibco), 20 µM β-mercaptoethanol (MilliporeSigma), 10 µg/mL ascorbic acid (MilliporeSigma), 1.5 ng/mL sodium selenite (MilliporeSigma), 50 µM ethanolamine (MilliporeSigma), and 10 mM HEPES (Gibco). Differentiated iNK cells were supplemented with 200 IU of rhIL-2/ml (R&D Systems, Minneapolis, Minnesota, USA), while PB NK cells were supplemented with 50 IU of rhIL-2/ml.

PB NK cells were obtained from healthy consenting adults at the University of Minnesota. Briefly, PB mononuclear cells were isolated using Lymphocyte Separation Medium (Corning, Tewksbury, Massachusetts, USA) per the manufacturer’s instructions. NK cells were enriched using a negative selection human NK Cell Isolation Kit (Miltenyi Biotec, Bergisch, Germany). Isolated NK cells were >95% pure, as determined by CD56+ CD3− staining for flow cytometry. Cryopreservation was performed using CryoStor CS10 Cell Freezing Medium (STEMCELL Technologies, British Columbia, Canada) at 1×107 to 2.5×107 cells per mL in cryogenic vials. Vials were placed in freezer storage containers with isopropyl alcohol for controlled rate freezing at −80°C for 24 hours. Vials were then moved to liquid nitrogen for storage. Recovery of cryopreserved cells was performed by floating a vial in a 37°C water bath for 1 min until partially thawed. Pre-warmed X-VIVO 15 media (Lonza, Basel, Switzerland) was slowly added to the vial and the cell suspension transferred to a 50 mL conical tube containing 10 mL of X-VIVO 15 media. Cells were centrifuged at 300×g for 5 min and resuspended in X-VIVO 15 for experiments unless otherwise noted.

NK cell antibody arming

iNK cells and PB NK cells were armed with therapeutic mAbs at a cell density of 5×106 cells/mL with 5 µg/mL of trastuzumab or cetuximab in X-VIVO 15 media (Lonza) supplemented with 200 or 50 IU/mL rhIL-2, respectively, for 2 hours at 37°C and 5% CO2. The antibody-armed cells were extensively washed with X-VIVO 15 media to remove excess unbound antibody.

Therapeutic antibodies were biotinylated using the EZ-Link Sulfo-NHS-Biotin Kit (Thermo Fisher) according to the manufacturer’s instructions. iNK cells or PB NK cells were armed with biotinylated antibodies as described above. To determine levels of antibody coupling, cells were incubated with a streptavidin conjugated fluorophore (BioLegend) for 15 min at room temperature. Cells were extensively washed and streptavidin staining was assessed by flow cytometry.

Cell cytotoxicity assays

Cell cytotoxicity was measured using the DELFIA EuTDA assay (PerkinElmer, Waltham, Massachusetts, USA) per the manufacturer’s instructions and as previously described.17 To measure cell cytotoxicity in real-time, NLR or NLG SKOV-3 cells were used as well as OVCAR-4 and OVCAR-5 labeled with CellTrace Far Red (Thermo Fisher) according to the manufacturer’s instructions. The tumor cells were plated at a density of 4×103 (SKOV-3) or 6×103 (OVCAR-4 and OVCAR-5) cells/well in a 96-well flat bottom, tissue culture-treated plate 24 hours prior to starting the assay. iNK cells or PB NK cells were added at the indicated effector:target (E:T) ratios in X-VIVO 15 media supplemented with 200 IU or 50 IU of rhIL-2, respectively, at 37°C and 5% CO2. Fluorescent images of live cells were obtained hourly for the duration of the assay using an IncuCyte SX3 live cell imaging and analysis system (Sartorius), as we have previously described.30 31 33 34 Data are presented as a single normalized frequency of target cells remaining. Area under the curve (AUC) was computed using GraphPad Prism (La Jolla, California, USA).

Analysis of CD107a expression and cytokine production

iNK-CD64/16A cells were co-cultured with SKOV3 cells at an E:T ratio of 1:1 for 5 hours at 37°C and 5% CO2. Fluorophore-conjugated anti-CD107a was added prior to NK cell stimulation, and GolgiStop (BD Biosciences) was added for the final 4 hours of co-culture. Cell surface CD107a expression was determined by flow cytometry. Additionally, iNK-CD64/16A cells were co-cultured with SKOV-3 cells at an E:T ratio of 2:1 for 4 and 18 hours at 37°C and 5% CO2. After incubation, the plate was centrifuged for 5 min at 2200×g and supernatant was removed and frozen at −80°C. The production of IFN-γ and TNF-α were quantified using LEGENDplex human IFN-γ and TNF-α capture beads (BioLegend) following manufacturer’s instructions for cell culture supernatant. Beads were acquired on an FACSCelesta and data analyzed using LEGENDplex software.

CRISPR/Cas9 knockout of HER2 and EGFR in SKOV-3 cells

Single guide RNAs (sgRNA) for human Human epidermal growth factor receptor 2 (HER2) or epidermal growth factor receptor (EGFR) were designed using Synthego’s Knockout Guide Design tool (https://design.synthego.com/#/); HER2 sgRNA sequence: UCUCUCCUGCCAGUGUGCAC, and EGFR sgRNA: CUUUUUCUUCCAGUUUGCCA. Alt-R CRISPR-Cas9 sgRNAs were ordered from Integrated DNA Technologies (IDT, Coralville, Iowa, USA). IDT sgRNAs contained both tracerRNA and crRNA sequences. Recombinant Cas9 protein was purchased from IDT. CRISPR/Cas9 ribonucleoprotein complexes (RNPs) were assembled by mixing 100 pmol of sgRNA with 30 pmol of Cas9. After assembly, RNP complexes rested at room temperature for 15 min. RNPs were combined with the Amaxa nucleofector solution from the SF Cell Line 4D-Nucleofector X Kit (Lonza). SKOV-3 NLR cells (2×105) were then resuspended in the RNP/nucleofector solution and transferred to a 20 µL Nucleocuvette strip and nucleofected with the Amaxa 4D Nucleofector program FE-132. Nucleofected SKOV-3 cells were gently resuspended with warm culture medium and transferred into a 24-well plate for initial expansion. SKOV-3 cells lacking HER2 and EGFR were enriched through negative sorting by staining the cells with biotinylated trastuzumab and cetuximab and using anti-biotin MicroBeads (Miltenyi Biotec) following the manufacturer’s instructions. SKOV-3 cells lacking HER2 and EGFR were confirmed by flow cytometry.

Xenograft mouse model

A short-term tumor xenograft model was used as previously described with some modifications.30 Briefly, 8–12 weeks old NOD-scid IL2Rgammanull (NSG) mice were intraperitoneal (i.p.) injected with 3×105 SKOV-3-Luc cells. Mice were administered iNK-CD64/16A or iNK-CD64/16A IL-15RF cells (1×107 cells) i.p. in 1× Hank's Balanced Salt Solution (HBSS). Where indicated, mice were given 50,000 IU of rhIL-2 i.p. three times a week for the duration of the experiment. rhIL-2 used for all in vivo assays was obtained from the Biological Resources Branch, National Cancer Institute (NCI), NIH. Bioluminescence imaging (BLI) was performed to quantify tumor burden using an in vivo imaging system (IVIS) Spectrum (PerkinElmer, Waltham, Massachusetts, USA). Images were analyzed using Living Image Software (PerkinElmer).

For the human IgG xenograft model, NSG mice received i.p. administration of GAMMAGARD (Takeda, Lexington, Massachusetts, USA), pooled human IgG, at the indicated doses. To assess serum levels of human IgG isotypes, blood was collected via check bleed, allowed to clot at room temperature for 30 min, and centrifuged at 2230×g for 10 min. Serum was collected and saved for analysis by LEGENDplex Human Immunoglobulin Isotyping kit (BioLegend) per the manufacturer’s instructions. Mice were i.p. injected with SKOV-3-Luc cells and iNK-CD64/16A IL-15RF cells as indicated above. rhIL-2 was not administered. Some mice received cetuximab injections of 100 µg/mouse i.p. as indicated. Mice were monitored by BLI and survival assessed based on mobility and morbidity behavior.

Statistical analysis

Data are presented as the mean±SD. In vitro data are from at least three independent experiments. Each short-term tumor xenograft experiment is representative of at least two independent experiments. The human IgG xenograft experiments were performed twice. All statistical analyses were performed using GraphPad Prism (La Jolla, California, USA). Comparison between two groups was computed using the Student’s t-test, while comparison among three or more groups was carried out using one-way analysis of variance followed by Tukey honest significance difference (HSD) post hoc test. Statistical significance for percent change in radiance for in vivo studies was determined using the Mann-Whitney test. The survival curve was analyzed using the log-rank (Mantel-Cox) test. Significance was set as p≤0.05 for all statistical tests performed.

ResultsProduction and antibody arming of iNK-CD64/16A cells

In this study, we used transgene-free iPSCs for iNK cell generation.31–33 We have previously shown that the derived iNK cells are phenotypically and transcriptionally similar to PB-derived NK cells.32 iPSCs were initially engineered to express CD64/16A, differentiated into iCD34+ hematopoietic progenitor cells, and then into early iNK cells. For the final step, iNK cells were co-cultured with K562-mbIL21-41BBL feeder cells for further maturation and expansion (figure 1A).32 The generated iNK cells (CD56+ CD3−) stained uniformly for CD64, whereas unmodified iNK control cells demonstrated no CD64 staining (figure 1A), as is also the case for PB NK cells (see figure 2A).17

Figure 1

Antibody-armed iNK-CD64/16A cells mediate ADCC in vitro and in vivo. (A) Schematic and phenotyping of iPSC differentiation to expanded iNK-CD64/16A cells. Early iNK-CD64/16A cells were stained on days 20 and 30 with an anti-CD64 mAb. After 2 weeks of expansion and differentiation with K562-mbIL21-41BBL feeder cells, iNK cells were stained with anti-CD45, CD56, CD3, and CD64 mAbs. Cell staining was analyzed by flow cytometry; histograms are a representative of several independent expansions. (B) The ovarian cancer cell lines SKOV-3, OVCAR4, and OVCAR5 were stained with trastuzumab (anti-HER2), cetuximab (anti-EGFR), or a second stage fluorophore and analyzed by flow cytometry. The x-axis=log 10 fluorescence. (C) iNK-CD64/16A cells fresh off of expansion, either unarmed (iNK-CD64/16A) or armed with trastuzumab (iNK-CD64/16A Tras) or cetuximab (iNK-CD64/16A Cetux), were co-cultured with SKOV-3, OVCAR4, or OVCAR5 cells at the indicated E:T ratios. Target cell cytotoxicity was determined at 2 hours by a DELFIA EuTDA assay. Data represented as mean±SD; n=4 replicates; ****, p≤0.0001 by Student’s t-test. (D) NSG mice were implanted i.p. with SKOV-3-Luc cells (3×105), administered iNK-CD64/16A cells (1×107) i.p., and tumor burden evaluated by BLI, as indicated in the schematic. Tumor implanted mice were randomized into three groups: tumor alone (n=3), unarmed iNK-CD64/16A cells (iNK-CD64/16A) (n=5), or trastuzumab-armed iNK-CD64/16A cells (iNK-CD64/16A Tras) (n=5). iNK-CD64/16A cells were antibody armed fresh off of expansion. All mice received i.p. injections o IL-2 every other day. The y-axis=percent change in radiance, calculated for each animal at each time point using the formula (radiance day x/radiance day 0)×100. Radiance values for each animal at day 0 are represented as 100%. Data represented as mean±SD; *, p≤0.05; **, p≤0.01 by Mann-Whitney test. ADCC, antibody-dependent cell-mediated cytotoxicity; BLI, bioluminescence imaging; EGFR, epidermal growth factor receptor; E:T, effector:target; HER2, human epidermal growth factor receptor 2; IL-2, interleukin-2; iNK, induced pluripotent stem cell-derived NK cell; i.p., intraperitoneal; iPSC, induced pluripotent stem cell; mAb, monoclonal antibody; NK, natural killer.

Figure 2

iNK-CD64/16A cells armed with antibodies after cryopreservation mediate ADCC. (A) Expanded then cryopreserved iNK-CD64/16A cells, iNK control cells, or PB NK cells were thawed and armed with biotinylated trastuzumab or cetuximab, washed, and stained with a streptavidin fluorophore. Unarmed cells were also stained with streptavidin fluorophore (no antibody). Cell staining levels were analyzed by flow cytometry. The x-axis=log 10 fluorescence. Flow cytometric analysis of CD16A and CD64 staining of PB NK cells was also performed (gray histograms). (B) iNK-CD64/16A, iNK control cells, or PB NK cells were thawed from cryopreservation, armed with trastuzumab or left unarmed, washed, and co-cultured with SKOV-3-NucLightGreen cells at the indicated E:T ratios. Cytotoxicity was assessed by live cell imaging. Percent of live target cells remaining are displayed relative to targets alone at time 0. The area under curve (AUC) of the remaining SKOV-3 was calculated for unarmed and antibody-armed NK cells. Data represented as mean±SD; n=4 replicates; ns, p>0.05; ****, p≤ 0.0001, by one-way ANOVA followed by Tukey HSD post hoc test. (C) Unarmed and trastuzumab-armed iNK-CD64/16A cells were incubated for 5 hours at 37°C in absence or presence of SKOV-3 cells at an E:T of 1:1, stained for CD107a, and examined by flow cytometry. Data represented as MFI±SD; n=3 replicates; ****, p≤ 0.0001, by one-way ANOVA followed by Tukey HSD post hoc test. ADCC, antibody-dependent cell-mediated cytotoxicity; ANOVA, analysis of variance; E:T, effector:target; HSD, honest significant difference; iNK, induced pluripotent stem cell-derived NK cell; MFI, mean fluorescence intensity; NK, natural killer; PB, peripheral blood.

Due to the high affinity state of CD64, it stably binds to free monomeric IgG1.35 We have shown that NK-92 cells expressing CD64/16A can be armed with antitumor therapeutic mAbs, which did not occur for NK-92 cells expressing equivalent levels of the higher affinity variant of CD16A.17 iNK-CD64/16A cells can also be stably armed with antitumor therapeutic mAbs, as shown below. To test the ability of antibody-armed iNK-CD64/16A cells to mediate ADCC, we used the ovarian cancer cell lines SKOV-3, OVCAR4, and OVCAR5 as target cells. Each cell line expressed HER2 and EGFR at different levels (figure 1B). Freshly expanded iNK-CD64/16A cells when armed with trastuzumab or cetuximab, which recognize HER2 and EGFR, respectively, lysed target cells at significantly (p≤0.0001) higher levels than unarmed iNK-CD64/16A cells. This was determined by a DELFIA EuTDA-based short-term cytotoxicity assay (figure 1C), as a quick screen for ADCC effector function.

To initially assess the antitumor function of the antibody-armed iNK-CD64/16A cells in vivo, we used a short-term intraperitoneal tumor xenograft model of ovarian cancer, as described earlier.30 SKOV-3 cells (3×105) expressing firefly luciferase were injected i.p. into NSG mice, the most common site for distant metastasis during ovarian cancer.36 Expanded iNK-CD64/16A cells (1×107) were either armed with trastuzumab or left unarmed, extensively washed, and then infused i.p. into the mice. Intraperitoneal administration of NK cells has been reported to be more effective than intravenous delivery in the xenograft mouse model,37 and is being evaluated in clinical trials to treat ovarian cancer.38 rhIL-2 was administered to all mice to support the proliferation and persistence of the iNK cells. Tumor burden was measured by BLI at 4 days post SKOV-3 cell implantation (Day 0) and at days 3 and 17 post-iNK cell infusion (as illustrated in figure 1D). By day 3, a significant reduction in tumor burden occurred in mice that received trastuzumab-armed iNK-CD64/16A cells when compared with mice that received tumor only or unarmed iNK-CD64/16A cells (p≤0.05 and p≤0.01, respectively; figure 1D). By day 17, unarmed CD64/16A cells effectively controlled tumor burden, consistent with their natural cytotoxicity effector function via germline receptors,32 and trastuzumab-armed iNK-CD64/16A cells demonstrated significantly greater tumor reduction (figure 1D). Taken together, the above findings show that freshly expanded, antibody-armed iNK-CD64/16A cells mediated ADCC in vitro and in vivo.

We next examined if the high affinity state of CD64/16A allowed for mAb arming immediately following their thaw from cryopreservation. For these studies, we examined iNK-CD64/16A cells, iNK control cells, and PB NK cells. All cells were expanded by K562-mbIL21-41BBL feeder cells and cryopreserved in a similar manner. All NK cells on thaw were incubated with biotinylated trastuzumab or cetuximab, extensively washed, stained with fluorophore-labeled streptavidin, and examined by flow cytometry. Though thawed PB NK cells retained high levels of CD16A, only the iNK-CD64/16A cells were capable of stable antibody coupling (figure 2A). Thawed iNK-CD64/16A cells, iNK control cells, and PB NK cells were also incubated with or without non-labeled trastuzumab, washed, and co-cultured with SKOV-3 cells at different E:T ratios. Target cell killing was assessed by an IncuCyte-based live cell imaging assay for 48 hours. rhIL-2 was added to the assay to support NK cell survival. Natural cytotoxicity of tumor cells occurred by the different unarmed NK cells, but only the iNK-CD64/16A cells demonstrated significantly (p≤0.0001) increased tumor cell lysis when armed with trastuzumab (figure 2B).

It has been reported that monomeric IgG attached to endogenous CD64 on leukocytes does not induce cell activation until it engages antigen.39 We evaluated this for antibody-armed iNK-CD64/16A cells by monitoring surface levels of CD107a, a marker of NK cell activation by CD16A signaling.40 As shown in figure 2C, unarmed and trastuzumab-armed iNK-CD64/16A cells treated for 5 hours at 37°C expressed equivalently low levels of CD107a. When co-cultured with SKOV-3 cells, only the trastuzumab-armed iNK-CD64/16A cells demonstrated a distinct upregulation of CD107a expression (figure 2C). This short-term assay detected robust iNK-CD64/16A cell activation and therefore we cannot rule out that antibody-arming of CD64/16A might cause a low level of tonic signaling in NK cells.

Cryopreservation and function of antibody-armed iNK-CD64/16A cells for off-the-shelf therapeutics

We next examined whether freshly expanded and antibody-armed iNK-CD64/16A cells could be cryopreserved and maintain their ADCC effector function on thaw. This was of interest since the use of cryopreserved antibody-armed iNK-CD64/16A cells would reduce the processing time between thawing and use in functional assays, as well as their administration into patients as an off-the-shelf cell therapy. To this end, expanded iNK-CD64/16A cells were armed with biotinylated-trastuzumab or cetuximab, excess antibody was washed away, and the cells were cryopreserved. We found that on thaw, the iNK-CD64/16A cells maintained high levels of antibody arming (figure 3A). In addition, their viability was similar to thawed, unarmed iNK-CD64/16A cells (data not shown). Thawed iNK-CD64/16A cells, either unarmed or armed with trastuzumab or cetuximab, were co-cultured with OVCAR-4 and OVCAR-5 cells labeled with CellTrace Far Red. Tumor cell lysis was determined by IncuCyte monitoring. A marked enhancement in tumor cell killing was observed by the thawed antibody-armed iNK-CD64/16A cells at different E:T ratios (figure 3B).

Figure 3

Cryopreserved antibody-armed iNK-CD64/16A cells mediate ADCC. (A) iNK-CD64/16A cells were armed with biotinylated trastuzumab or cetuximab, washed, cryopreserved, thawed, and stained with a streptavidin fluorophore. Unarmed cells were also stained with streptavidin fluorophore (no antibody). Cell staining levels were analyzed by flow cytometry. The x-axis=log 10 fluorescence. (B) Thawed trastuzumab-armed iNK-CD64/16A cells (iNK-CD64/16A Tras) or unarmed iNK-CD64/16A cells (iNK-CD64/16A) were co-cultured with OVCAR-4 or OVCAR-5 labeled with CellTrace Far Red at the indicated E:T ratios. Cytotoxicity was assessed by live cell imaging. Percent of live target cells remaining are displayed relative to targets alone at time 0. (C) Thawed trastuzumab-armed iNK-CD64/16A cells (iNK-CD64/16A Tras) or unarmed iNK-CD64/16A cells (iNK-CD64/16A) were co-cultured with SKOV-3 cells at an E:T ratio of 2:1. TNF-α and IFN-γ levels in the media supernatant were quantified at the indicated time points. Data represented as mean±SD; n=3 replicates; ****, p≤0.0001, by one-way ANOVA followed by Tukey HSD post hoc test. (D) Thawed trastuzumab-armed iNK-CD64/16A cells (iNK-CD64/16A Tras) or unarmed iNK-CD64/16A cells (iNK-CD64/16A) were co-cultured with SKOV-3-NLG cells at the indicated E:T ratios. Remaining effector cells were then harvested and co-cultured with additional SKOV-3-NLR cells for a second round. This was repeated for a third round using SKOV-3-NLG cells. SKOV-3 cells expressing NLG or NLR were alternated between rounds to exclude target cells carried over from the previous round. Cytotoxicity was assessed by live cell imaging. Percent of live target cells remaining are displayed relative to targets alone at time 0. AUC of the remaining SKOV-3 was calculated for unarmed and antibody-armed NK cells. Data represented as mean±SD; n=8 replicates; ****, p≤0.0001, by one-way ANOVA followed by Tukey HSD post hoc test. (E) CRISPR/Cas9 genome editing was used to generate HER2 and EGFR knockout SKOV-3 cells. HER2 and EGFR expression by wild type SKOV-3 cells and SKOV-3 HER2− EGFR− cells was determined by flow cytometry. In addition, SKOV-3 cells were stained with avelumab (anti-PD-L1) and/or a fluorophore-conjugated secondary antibody and analyzed by flow cytometry. The x-axis=log 10 fluorescence. (F) Thawed trastuzumab or cetuximab-armed iNK-CD64/16A cells (iNK-CD64/16A Tras or Cetux) or unarmed iNK-CD64/16A cells (iNK-CD64/16A) were co-cultured with SKOV-3-NLG cells or SKOV-3 HER2− EGFR−-NLR cells (E:T=8:1). In some wells, avelumab was added to the cell culture (+ Avelu). Cytotoxicity was assessed by live cell imaging. Percent of live target cells remaining are displayed relative to targets alone at time 0. AUC of the remaining SKOV-3 was calculated for unarmed and antibody-armed iNK cells. Data represented as mean±SD; n=4 replicates; ***, p≤0.001; ****, p≤0.0001, by one-way ANOVA followed by Tukey HSD post hoc test. ADCC, antibody-dependent cell-mediated cytotoxicity; ANOVA, analysis of variance; AUC, Area under the curve; EGFR, epidermal growth factor receptor; E:T, effector:target, HER2, human epidermal growth factor receptor 2; HSD, honest significant difference; IFN, interferon; iNK, iNK, induced pluripotent stem cell-derived NK cell; KO, knock-out; NK, natural killer; NLG, NucLightGreen; NLR, NucLightRed; PD-L1, programmed cell death ligand 1; TNF, tumor necrosis factor; WT, wild-type.

In additional assays, thawed iNK-CD64/16A cells armed with trastuzumab released significantly (p≤0.0001) higher levels of TNF-α and IFN-γ compared with unarmed iNK-CD64/16A when co-cultured with SKOV-3 cells (figure 3C). The latter iNK-CD64/16A cells did produce low levels of cytokines that was more apparent by 18 hours of activation, likely due to stimulation through their germline receptors (figure 3C). Thawed trastuzumab-armed iNK-CD64/16A cells also mediated higher levels of SKOV-3 cell killing (figure 3D). For this assay, we investigated ADCC durability by the thawed antibody-armed iNK CD64/16A cells. Unarmed or trastuzumab-armed iNK CD64/16A cells were collected at the conclusion of the assay and transferred to new wells containing fresh SKOV-3 cells for a second and then a third round of killing. To distinguish target cells in the new wells from any remaining live target cells carried over from previous wells, we alternated between SKOV-3 target cells expressing either NLG or NLR. For all rounds, the trastuzumab-armed iNK-CD64/16A cells maintained significantly (p≤0.0001) higher levels of killing than the unarmed iNK-CD64/16A cells at different E:T ratios (figure 3D). Hence, following a freeze/thaw cycle, antibody-armed iNK-CD64/16A cells retained their antigen specific effector function and could mediate sustained ADCC.

An important feature of iNK-CD64/16A cells is the usage of multiple therapeutic mAbs for directing these cells to different targets on tumor cells to address antigen escape. In addition to HER2 and EGFR, programmed cell death ligand 1 (PD-L1) is also uniformly expressed by SKOV-3 cells, and it can be targeted by the therapeutic mAb avelumab (figure 3E, left panel). Thawed unarmed iNK-CD64/16A cells mediated significantly (p≤0.0001) enhanced SKOV-3 cell killing when in the presence of added avelumab (figure 3F). Culturing SKOV3 cells with avelumab alone had no effect on cell viability (data not shown). To simulate the loss of tumor antigens, we used CRISPR/Cas9 genome editing to generate SKOV-3 HER2− EGFR− cells (figure 3E, right panel). In contrast to wildtype cells, SKOV-3 HER2− EGFR− cells did not undergo ADCC by thawed trastuzumab or cetuximab-armed iNK-CD64/16A cells (figure 3F). However, the addition of avelumab to a co-culture of unarmed iNK-CD64/16 A cells or antibody-armed iNK-CD64/16A cells with SKOV-3 HER2− EGFR− cells significantly (p≤0.0001) increased ADCC (figure 3F). These assays show that iNK-CD64/16A cells could be directed to tumor antigens either by the addition of therapeutic mAbs or when the antibodies were used to arm the iNK cells. Moreover, antibody-armed iNK-CD64/16A cells could be redirected by the addition of a new therapeutic mAb to target a different tumor antigen.

Generation of cytokine-autonomous iNK-CD64/16A cells to enhance therapeutic potential

Human NK cell adoptive transfer into NSG mice and patients requires repeated bolus infusions of cytokines, such as IL-2 or IL-15, for their expansion and persistence.41 To eliminate the variability, toxicity, and potential detrimental immunoregulatory responses of this approach, we have engineered iNK cells to express a membrane-bound IL-15/IL-15RF protein.33 IL-15RF complexes with IL-2/IL-15Rβ and the common γ chain to provide self-stimulating signals that induce iNK cell activation and proliferation in vitro and in vivo.31 33 34 The same modification of iPSCs expressing CD64/16A was performed here. Their cytokine independence was verified using a previously described IncuCyte assay.31 iNK-CD64/16A cells expressing or lacking IL-15RF demonstrated equivalent levels of antibody arming (figure 4A). Trastuzumab-armed iNK-CD64/16A lacking IL-15RF were thawed from cryopreservation and incubated with SKOV-3 cells at different E:T ratios. Inclusion of IL-2 in the culture markedly enhanced ADCC (figure 4B). Antibody-armed iNK-CD64/16A cells expressing IL-15RF on the other hand demonstrated equivalent robust ADCC of SKOV3 cells in either the presence or absence of IL-2 (figure 4B), demonstrating the intrinsic function of IL-15RF in these cells.

Figure 4

Cryopreserved antibody-armed iNK-CD64/16A cells expressing IL-15RF are cytokine-independent. (A) iNK-CD64/16A and iNK-CD64/16A IL-15RF cells were armed with biotinylated trastuzumab and cryopreserved. On thaw, the cells were stained with a streptavidin fluorophore, and analyzed for antibody arming by flow cytometry. The x-axis=log 10 fluorescence. (B) Unarmed or trastuzumab-armed iNK-CD64/16A or iNK-CD64/16A IL-15RF cells thawed from cryopreservation were co-cultured with SKOV-3-NucLightGreen cells at the indicated E:T ratios in the presence or absence of iIL-2. Cytotoxicity was assessed by live cell imaging. Percent of live target cells remaining are displayed relative to targets alone at time 0. (C) NSG mice were implanted i.p. with SKOV-3-Luc cells (3×105), administered iNK-CD64/16A IL-15RF cells (1×107) i.p., and tumor burden evaluated by BLI, as indicated in the schematic. Tumor implanted mice were randomized into three groups: tumor alone (n=6), unarmed iNK-CD64/16A IL-15RF cells (iNK-CD64/16A) (n=6), or trastuzumab-armed iNK-CD64/16A IL-15RF cells (iNK-CD64/16A Tras) (n=6). The y-axis=percent change in radiance, as described in figure 1D. Data represented as mean±SD; *, p≤0.05, **, p≤0.01 by Mann-Whitney test. BLI, bioluminescence imaging; E:T, effector:target; IL-2, interleukin-2; IL-15RF, interleukin-15Rα fusion; iNK, induced pluripotent stem cell-derived NK cell; i.p., intraperitoneal, NK, natural killer.

The cytolytic activity of antibody-armed iNK-CD64/16A IL-15RF cells was also examined using our tumor xenograft model. Following i.p. implantation of SKOV-3 cells, unarmed or trastuzumab-armed iNK-CD64/16A IL-15RF cells were thawed from cryopreservation and i.p. administered to the mice in the absence of cytokine support. Tumor BLI was measured 4 days post SKOV-3 cell implantation (Day 0) and at days 3 and 14 post-iNK cell infusion. Significant (p≤0.05) reductions in tumor burden were observed at both time points in mice tha