Ablation of SYK Kinase from Expanded Primary Human NK Cells via CRISPR/Cas9 Enhances Cytotoxicity and Cytokine Production

Natural killer cells are innate lymphocytes that can kill both cancerous and infected cells (1, 2). In peripheral blood, NK cells comprise 5–20% of lymphocytes, mainly surveying the bloodstream for potential target cells (3–5). NK cells kill target cells and signal neighboring cells through a variety of mechanisms, including natural cytotoxicity (e.g., missing self), FasL- or TRAIL-induced apoptosis, Ab-dependent cellular cytotoxicity (ADCC), and cytokine production. During ADCC and cytokine production, NK cells first recognize the Fc region of IgG bound to a target cell with CD16A (FcγRIIIA). CD16A utilizes signaling adapter molecules including the FcRγ-chain (gene: FCER1G, FcRγ) and/or the CD3ζ-chain (CD3ζ) for signal transduction (6). Despite their seemingly redundant functions, FcRγ and CD3ζ have distinct features. For example, FcRγ contains one ITAM, whereas CD3ζ contains three ITAMs (7). Crosslinking of CD16A by Abs bound to a target cell recruits Src-family kinases, such as Lck, to phosphorylate ITAMs on FcRγ and/or CD3ζ. This phosphorylation recruits the tyrosine kinases SYK and/or ZAP70 to ITAMs for subsequent signal transduction. SYK preferentially binds FcRγ, whereas ZAP70 preferentially binds CD3ζ (8). SYK and ZAP70 then phosphorylate their respective downstream targets (e.g., PI3K, VAV1, and phospholipase Cγ isoforms) to further propagate the CD16A signaling pathway, ultimately culminating in cytotoxicity and cytokine production (9, 10). Although the molecules participating in the CD16A-mediated signaling pathway are defined, the specific contributions of these signaling proteins to expanded primary human NK cell function were unclear.

Circulating NK cells are not a homogeneous population. Various factors, such as infection, can drive NK cell phenotypic and functional diversity (11, 12). CMV is the epitome of a pathogen that alters both NK cell phenotype and function (12–15). In mice, murine CMV (mCMV) infection results in expansion of a subset of NK cells expressing Ly49H, a receptor that binds mCMV peptide on MHC (12). This NK population persists, specifically expands upon mCMV reactivation, and has enhanced effector function, displaying adaptive-like qualities (12). In humans, CMV infection also results in expansion of NK cells expressing NKG2C and CD57 (13–15). These cells proliferate after recognizing CMV in vitro (16) and upon CMV reactivation in humans (17). After CMV reactivation, CD57⁺/NKG2C⁺/NKG2A⁻ NK cells also exhibit enhanced cytokine production (17). These results led to the classification of CD57⁺/NKG2C⁺/NKG2A⁻ NK cells as adaptive NK cells due to their adaptive qualities, such as longevity and Ag-specific expansion (albeit through a limited germline-encoded receptor). Adaptive NK cells are also hypothesized to play a protective role in cancer and transplant patients as a higher frequency of adaptive NK cells correlates with better clinical outcomes in individuals with reactivated CMV (18).

In addition to phenotypic differences, adaptive NK cells show increases in ADCC and cytokine production compared with conventional NK cells (19, 20). Zhang et al. (20) showed that CMV infection was associated with loss of the FcRγ-chain. FcRγ-negative NK cells showed increased degranulation (CD107a), IFN-γ, and TNF-α production in response to opsonized infected cells over conventional NK cells. In addition to FcRγ deficiency, adaptive NK cells have been shown to lack both SYK kinase and PLZF (21, 22). PLZF is a transcription factor that binds the promoter of genes encoding FcRγ and SYK and likely plays a role in regulation of these genes (22, 23). In addition to CMV, other infectious diseases, such as HIV and malaria, can induce adaptive NK cell formation (24–26). HIV- and malaria-induced adaptive NK cells also show enhanced ADCC and cytokine production (24–26). However, these cells have a different phenotype than do their CMV-induced counterparts, such as maintained SYK expression, highlighting adaptive NK cell heterogeneity (24–26).

The absence of FcRγ, PLZF, and/or SYK is hypothesized to be the mechanism responsible for enhancing the ADCC function of CMV-induced adaptive NK cells (17–22). However, it was unclear whether these molecules are responsible for enhanced function or only correlate with it. In this study, we investigated the role of molecules associated with CMV-induced adaptive NK cells, including FcRγ, PLZF, and SYK, in the ADCC signaling apparatus. We tested for redundancy in the ADCC signaling pathway by targeting molecules with similar functions as FcRγ and SYK, such as CD3ζ and ZAP70, respectively. We also sought to disentangle which molecules are associated with cytotoxic activity versus cytokine production after CD16A activation. Although using a murine model is a potential approach to answer these questions, the murine CD3ζ-chain does not associate well with CD16A, limiting the utility of murine signaling protein knockouts for ADCC studies (27). For humans, it has not been possible to ablate genes in expanded primary human NK cells, until recently. Therefore, to accomplish these goals, we developed a CRISPR/Cas9 gene-editing protocol for thawed and expanded human primary NK cells, ablated genes that encoded molecules in the CD16A pathway, and tested the effect of gene ablation on ADCC and cytokine production.

We show that SYK kinase ablation enhances ADCC, cytokine production, and target cell conjugation, whereas ZAP70 ablation results in the opposite phenotype. SYK ablation enhanced killing more so than degranulation, and the increased killing capacity was maintained for 4 wk postexpansion. These results point toward a mechanism involving SYK for the enhanced function of CMV-induced adaptive NK cells. Overall, this study provides a mechanistic understanding of the enhanced ADCC function observed in CMV-induced adaptive NK cells and provides a new strategy for improving NK cell therapies by leveraging the ADCC pathway.

Requests for information or resources should be directed to the lead contact (G.T.H.).

This study did not generate unique reagents. All reagents used are listed in Table I.

R 4.1 (28) was used to generate Supplemental Fig. 1G. The code was adapted from https://biocorecrg.github.io/CRG_RIntroduction/volcano-plots.html (29) and is available at https://github.com/jamesdahlvang/volcano. All raw data are available upon request.

The work done in this study was approved by the Institutional Review Board of the University of Minnesota. Primary RBCs and PBMCs were obtained and purified from deidentified adult blood donors (Memorial Blood Center). The CMV status of each donor was reported by the Memorial Blood Center. RBCs were purified from whole blood by filtration (Memorial Blood Center, 4C4300) and resuspended at 50% hematocrit in RPMI 1640 + 25 mM HEPES, l-glutamine, and 50 mg/L hypoxanthine (K-D Medical, 50-101-8907). PBMCs were isolated using Ficoll (MP Biomedicals, ICN50494) centrifugation and the remaining RBCs were lysed using ACK (ammonium-chloride-potassium) (Lonza, BP10-548E). Once isolated, PBMCs were counted and resuspended at 2 × 10⁸ cells/ml in 10 ml of PBS + 2% FBS (Peak Serum, PS-FB1) + 1 mM EDTA (SC buffer). NK cells were then isolated from these PBMCs using a magnetic negative selection kit (STEMCELL Technologies, 17955). Briefly, PBMCs were stained with 500 μl of the manufacturer’s Ab mixture, which is less Ab than the manufacturer recommends, and 30 μl of additional biotinylated anti-CD3 (STEMCELL Technologies, 18051) was added to the sample to improve purity. Samples were incubated at room temperature (RT) for 10 min. After incubation, 1 ml of magnetic beads was added to the sample, which was then vortexed two to three times and incubated for 10 min at RT. After incubation, 35 ml of SC buffer was added to the cells, which were then placed on a magnet and incubated at RT for 10 min. After incubation, the liquid, which contained purified NK cells, was removed from the tube without touching the magnet. NK cells were then added to a new 50-ml conical and placed back on the magnet for 10 min at RT. After 10 min, purified NK cells were removed and counted. For cryopreservation, NK cells were resuspended in RMPI 1640 (Fisher Scientific, SH3002701) + 50% FBS (Peak Serum, PS-FB1) at 10⁷ cells/ml. Five hundred microliters of the suspension was transferred to a Cryovial. When all tubes were ready to freeze, 500 μl of FBS (Peak Serum, PS-FB1) + 15% DMSO was added to cells in Cryovials. The cells were immediately transferred to a precooled Mr. Frosty (Thermo Fisher Scientific, 5100-0001) and then placed in a −80°C freezer. After 24–48 h, the cells were transferred to a vapor-phase liquid nitrogen freezer for later use.

Modified K562 artificial APCs (aAPCs) were provided by Dr. Branden Moriarity’s laboratory. These cells express 4-1BB ligand and IL-21 to aid NK cell proliferation (30). K562 aAPCs were maintained between 1 × 10⁵ and 1.5 × 10⁶ cells/ml in RPMI 1640 (Fisher Scientific, SH3002701) + 10% FBS (Peak Serum, PS-FB1) (RP10) + 1 μg/ml gentamicin (Sigma-Aldrich, G1272) and incubated at 37°C with 5% CO₂. Unless indicated otherwise, all 37°C incubations were conducted at 5% CO₂.

Both conventional SKOV-3 and fluorescent SKOV-3 NucLight Red (NLR) cells (SKOV-3 NLR) were provided by Dr. Jeffrey Miller’s laboratory. Once the cells became 80–95% confluent, the medium was aspirated from the flask and the cells were washed with PBS. After washing the cells, 5 ml of prewarmed 0.05% Trypsin-EDTA (Life Technologies, 25300062) was added to the flask. The cells were then incubated at 37°C for 5 min. Following incubation, the flask was tapped to dislodge cells from the plastic. The cells were collected and washed with 10 ml of prewarmed McCoy’s 5A medium (Life Technologies, 166600082) + 10% FBS (Peak Serum, PS-FB1) + 1% penicillin/streptomycin (Life Technologies, 1514022) + 1 μg/ml puromycin (Thermo Fisher Scientific, A1113830) to maintain NLR expression (SKOV-3 media). The resulting suspension was centrifuged, resuspended in prewarmed SKOV-3 media, and counted. Half of the cell suspension was used for counting. The SKOV-3 cells were then diluted to 6.5 × 10⁴ cells/ml with SKOV-3 media, transferred to a new flask, and incubated at 37°C.

Raji cells expressing NLR were provided by Dr. Jeffrey Miller’s laboratory. Every 2 d, Raji NLR cells were resuspended and centrifuged. After centrifugation, the cells were resuspended with the same volume in prewarmed RP10 and counted. The Raji NLR cells were then diluted to 2 × 10⁵ cells/ml with RP10 + 1 μg/ml of puromycin (Thermo Fisher Scientific, A1113830) to maintain NLR expression, transferred to a new flask, and incubated at 37°C.

NK-92 cells expressing GFP (parent line, NK-92-GFP), CD16A and GFP (NK-92-GFP-CD16A), and CD64/16A and GFP (NK-92-GFP-CD64/16A) were provided by Dr. Bruce Walcheck’s laboratory. NK-92-GFP-CD16A cells stably express GFP and CD16A (31), whereas NK-92-GFP-CD64/16A cells stably express GFP and the cytoplasmic tail of CD16A with the extracellular domain of CD64 (32). Every 2 d, NK-92 cells were resuspended and centrifuged. After centrifugation, the supernatant from the cell culture was saved. The cells were then resuspended with the same volume of prewarmed MEM-α (Life Technologies, 12561) + 10% heat-inactivated FBS (Life Technologies, 16140071) + 10% heat-inactivated horse serum (Life Technologies, 26050088) + 0.8% penicillin/streptomycin (Life Technologies, 1514022) + 200 IU/ml IL-2 (National Cancer Institute [NCI]) + 0.08 mM 2-ME (Life Technologies, 21985023) (NK-92 media) and counted. After counting, cells were resuspended at 2 × 10⁵ cells/ml in 50% NK-92 media and 50% of the previous culture’s supernatant, then incubated at 37°C. Cells were maintained between 2 × 10⁵ and 1.5 × 10⁶ cells/ml.

Cryopreserved vials of primary NK cells were thawed by placing them in a water bath at 37°C until the vial was 95% thawed. The cells were then transferred to a 15-ml conical. Prewarmed RP10 was added dropwise at a rate of 1 ml/min while the conical was swirled. After adding 3 ml of RP10, an additional 6 ml of prewarmed RP10 was added and the cells were centrifuged at 650 × g. The cells were washed once with RP10, centrifuged, and resuspended in 5 ml of X-VIVO 15 (Lonza, BE02-060F) + 10% FBS (Peak Serum, PS-FB1) (X-VIVO 10) + 2 ng/ml IL-15 (NCI). Cells were then transferred to a six-well plate and incubated overnight. After incubation, 3 × 10⁶ NK cells were mixed with 6 × 10⁶ irradiated K562 aAPCs in 30 ml X-VIVO 10 + 50 IU/ml IL-2 (NCI). K562 aAPCs were irradiated at 100 Gy (RS-2000 x-ray irradiator). The cells were added to a G-Rex six-well plate (Wilson Wolf, 80240M) and incubated at 37°C for 6 d.

Single-guide RNAs (sgRNAs) were designed using Synthego’s CRISPR design tool (https://design.synthego.com/#/). sgRNA sequences are listed in Supplemental Table I. Ribonucleoprotein (RNP) complexes were made by combining 100 pmol of sgRNA (IDT) with 30 pmol of Cas9 (IDT, 1081059). When the volume of the RNP complex was <5 μl/sample, nuclease-free H₂O was added after combining the RNP complex to bring the volume to 5 μl/sample. The RNP complexes were incubated at RT for 15 min. When multiple genes were ablated simultaneously, individual RNP complexes were made and combined after incubating for 15 min.

NK cells were centrifuged and resuspended in P3 primary cell Nucleofector solution (Lonza, V4XP-3032) at 3.33 × 10⁸ cells/ml. Cells (5 × 10⁶) were mixed with RNPs and transferred without forming bubbles to a 16-well Nucleocuvette strip (Lonza, V4XP-3032). Unless otherwise stated, the cells were nucleofected with an Amaxa 4D-Nucleofector (Lonza) using code CA137. After nucleofection, the cells were rested for 15 min at RT. Eighty microliters of prewarmed X-VIVO 10 was added to each nucleofection well. The cells were rested for a further 30 min at RT, and then the total volume was transferred to a G-Rex 24-well plate (Wilson Wolf, 80192M) with 1.9 ml of prewarmed X-VIVO 10. Each well was then washed with 100 μl of medium, which was transferred to the cell suspension. The cells were rested once more for 30 min at RT before 4 ml of X-VIVO 10 + 4.5 ng/ml IL-15 (NCI) was added to each well (3 ng/ml IL-15 final). The cells were then incubated at 37°C for 6 d.

RBCs were opsonized with polyclonal anti-RBC rabbit Ab (Rockland, 1094139) and combined with NK cells at a NK cell/RBC ratio of 1:5 in RP10 + 1 μg/ml brefeldin + 2 µM monensin + 1:200 anti-CD107a, which measures degranulation (BioLegend, 328625). As a negative control, NK cells were also incubated with nonopsonized RBCs. The cells were then incubated at 37°C for 5 h. After 5 h, the cells were centrifuged, washed, and stained for flow cytometry.

Cells were resuspended in a master mix made of PBS, viability dye (Tonbo, 13-0865-T500), and surface stain Abs, then incubated at RT in the dark for 20 min. The cells were washed and incubated in 2% formaldehyde (Thermo Fisher Scientific, 28908) at 37°C for 10 min. After incubation, the cells were washed and resuspended in 0.04% Triton X-100 (Fisher Scientific, BP151-500) at RT in the dark for 7 min. After 7 min, the cells were washed in PBS + 2% FBS (Peak Serum, PS-FB1) + 2 mM EDTA (Fisher Scientific, BP2482-1) + 2% BSA (MP Biomedicals, 810681) (FACS buffer) and stained with an internal master mix in FACS buffer. Depending on the Abs used, the cells were then incubated at RT in the dark for 30–120 min. After incubation, the cells were washed with FACS buffer, washed again with PBS, and resuspended in PBS for flow cytometry. Samples were run on a CytoFLEX flow cytometer (Beckman Coulter) and analyzed via FlowJo 10.8.0 (FlowJo) to determine the extent of gene ablation and the effect of gene ablation on cell function.

Synthego’s Inference of CRISPR Edits (ICE) (https://ice.synthego.com/) was used to analyze gene ablation. For ICE analysis, DNA from gene-ablated and control samples was isolated using Qiagen’s DNeasy kit (Qiagen, 69506) according to the manufacturer’s instructions. PCR primers were designed to amplify a region of 500 bp around the cut site. The forward primer was designed 100–200 bp upstream from the cut site to capture it with high-quality sequencing. Additionally, we designed sequencing primers that were nested 5–50 bp inside the amplicon to improve sequencing efficiency. After amplifying the sgRNA cut site, ∼100 ng of the PCR product was submitted to the University of Minnesota Genomics Center along with 6.4 pmol of a sequencing primer for Sanger sequencing. The primers used are listed in Supplemental Table I.

To assess the expression of ∼350 surface proteins at once, cells were analyzed using BioLegend’s LEGENDScreen human PE kit (BioLegend, 700007). Subjects were barcoded with combinations of CD45 Abs conjugated to three fluorophores, allowing us to test seven subjects in one well (A, B, C, AB, AC, BC, ABC). After barcoding and pooling, samples were stained for CD56, CD3, live/dead, and CD14, and then acquired on a BD LSRFortessa flow cytometer according to the manufacturer’s instructions.

Expanded primary NK cells were labeled with CellTrace Violet (CTV, Thermo Fisher Scientific, C34571) and SKOV-3 cells (HER2⁺) were labeled with CellTrace Far Red (CTFR; Thermo Fisher Scientific, C34564) per the manufacturer’s instructions. After labeling, the cells were counted and resuspended in RP10. SKOV-3 and Raji NLR cells were resuspended at 2 × 10⁶ cells/ml whereas NK cells were resuspended at 1 × 10⁶/ml. Prior to combining target cells with NK cells, Raji NLR cells were incubated with 1 μg/ml rituximab and SKOV-3 cells were incubated with 10 μg/ml anti-HER2 (trastuzumab) Ab. For a 1:2 E:T ratio, 100 μl of each cell type was mixed together, centrifuged for 1 min at 20 × g, and then incubated in a 37°C water bath for the indicated time points (0, 2, 10, 20, and 30 min). After each time point, the cells were vortexed for 3 s and immediately fixed with ice-cold 2% paraformaldehyde in PBS. Fixed cells were then used to acquire 5 × 10⁴ events on a flow cytometer. The percentage of conjugated NK cells was calculated by gating on the NLR and CTV double-positive events (33).

DELFIA killing assays were performed according to the manufacturer’s instructions (PerkinElmer, AD0116). Briefly, SKOV-3 cells were resuspended at 10⁶ cells/ml in SKOV-3 media. To label the cells, 1.5 μl/ml bis(acetoxymethyl)-2,2′:6′,2″-terpyridine-6′,6″-dicarboxylate (BATDA) was added to the suspension. The cell suspension was then incubated for 30 min at 37°C. After labeling the cells, they were washed and transferred to a round-bottom 96-well plate at a concentration of 8 × 10³ cells/well. Anti-HER2 was then added to the well at a concentration of 3 or 10 µg/ml along with either expanded primary NK or NK-92 (effector) cells at various E:T ratios. The plate was spun at 400 × g for 1 min and incubated at 37°C for 2 h with 5% CO₂. After 2 h, the plate was centrifuged at 500 × g for 5 min. Twenty microliters of supernatant was then transferred to a 96-well DELFIA yellow plate and combined with 200 μl of europium. Signal was measured by time-resolved fluorescence using a BioTek Synergy 2. To measure maximum release, BATDA-labeled target cells were incubated with 10 μl of lysis buffer per the manufacturer’s instructions. To measure spontaneous lysis, BATDA-labeled SKOV-3 cells were cultured in parallel without effector cells or Abs. When possible, samples were run in technical triplicates.

On the day prior to the killing assay, SKOV-3 NLR cells (HER2⁺) were resuspended at 4 × 10⁴ cells/ml in RP10. One hundred microliters of the cell suspension was added to each well of an Incucyte-compatible flat-bottom 96-well plate. The SKOV-3 NLR cells were then incubated overnight at 37°C to allow them to adhere to the bottom of the well.

The following day, NK cells were resuspended at 1.2 × 10⁶ cells/ml in RP10. Sixty microliters of NK cells was added to a new 96-well plate. Anti-HER2 was then diluted to 20 µg/ml in RP10 and 60 µl of the solution was added to the NK cells. To prevent cells from localizing to the edge of the well, both the 96-well plate containing the NK cells/anti-HER2 and the plate containing the SKOV-3 NLR cells were incubated at 37°C for 15 min. One hundred microliters of NK cells/anti-HER2 was then added to the plate containing SKOV-3 NLR cells for an E:T ratio of ∼15:1 and a final anti-HER2 concentration of 5 μg/ml. The cell mixture was then loaded into an Incucyte SX5 (Sartorius). Immediately after loading the plate, a t = 0 h image was taken to ensure that cells were not localized to the edge of the well. After confirming that cells were dispersed throughout the well, four consecutive images were taken per well once per hour for 24 h. As controls, SKOV-3 NLR cells and NK cells without anti-HER2 were imaged as well as SKOV-3 NLR cells without NK cells. Samples were run in technical triplicates.

Data points from samples with <25% protein ablation were excluded from all analyses to ensure that the gene-ablated population was substantial enough to analyze. For expansion experiments, the cutoff was only applied at day 6 (e.g., when a sample was >25% protein ablation at day 6 but <25% at day 13, the data from day 13 were included). Samples between 25 and 49% ablation were kept for ADCC assay analysis, as the gene-ablated cells can be gated on via flow cytometry. Unlike flow cytometry–based ADCC assays, killing assays used a target cell count readout, precluding specific analysis of gene-ablated cells. Thus, killing assay data from samples with <49% protein ablation were excluded.

Proteins that were expressed in <10% of cells from each subject at each time point were classified as not expressed in NK cells. These proteins were excluded from the final analysis.

Data were transformed using fold change and log₂ fold change to ease the visualization of trends. Fold change was calculated by dividing the gene-ablated value by the no RNP value for the measurement in question.

Flow cytometry was used to assess the extent of CRISPR/Cas9 gene ablation at the protein level. To calculate the protein ablation percentage for single-gene ablations, the following formula was applied: protein ablation % = (% positive RNP only − % positive ablated)/% positive RNP only.

To calculate the protein ablation percentage for double-gene ablations, the following formula was used: protein ablation % = 1 − [(1 − % double-negative ablated)/(1 − % double-negative RNP only)].

Incucyte base analysis software (Sartorius, v2019B) was used to analyze killing assays. Images with SKOV-3 NLR cells + NK cells, SKOV-3 NLR cells alone, and NK cells alone were chosen from the 0-, 12-, and 24-h images. Image parameters were then adjusted to ensure that the software detected SKOV-3 NLR cells but not NK cells. Graphs were created by normalizing each well’s SKOV-3 NLR cell count at each time point to its time 0 h count.

For each sample, specific lysis was used to report ADCC activity. The calculation for specific lysis is as follows: specific lysis % = [(experimental release − spontaneous release)/(maximum release − spontaneous release)] × 100.

Excel 16.54 (Microsoft) was used for data transformation and statistical analysis. Schematics were created with BioRender. R 4.1 (28) was used to generate Supplemental Fig. 1G. Prism 9.1.2 (GraphPad Software) was used to generate figures and run statistical analyses. Illustrator 25.4.1 (Adobe) was used to draft figures for publication. Sample sizes, independent experiments run, statistical tests, and significance thresholds are listed in each figure legend.

Previously published CRISPR/Cas9 protocols for expanded primary human NK cells have reported various platforms and settings (30, 34–36, and J. Rautela, E. Surgenor, and N. D. Huntington, manuscript posted on bioRxiv, DOI: 10.1101/406934). We aimed to optimize a CRISPR/Cas9 protocol for the Amaxa 4D-Nucleofector system, as it can be scaled up for a large numbers of cells, potentially allowing for future clinical applications. To optimize our CRISPR/Cas9 protocol, we first used 24 Amaxa 4D-Nucleofector codes (format: XX###) to ablate FcRγ (FCER1G) from expanded primary NK cells. At 6 d post-CRISPR, we measured gene ablation efficiency (Supplemental Fig. 1A) and cell recovery (Supplemental Fig. 1B) by flow cytometry. From these 24 electroporation codes, we picked six that balanced protein ablation and cell recovery. We also tested CA137, as it had been used for successful gene ablation in NK-92 cells (37). We observed >70% ablation of FcRγ (Supplemental Fig. 1C) for each code. However, viable cell recovery varied. Of the codes tested, CA137 showed the highest average recovery at 3.3 × 10⁶ cells (Supplemental Fig. 1D). We concluded that using CA137 results in the best balance of effective gene ablation and cell recovery, leading us to use CA137 for the rest of the experiments. By evaluating a compilation of experiments that used CA137, we found that cell counts at 6 d post-CRISPR often exceeded our starting cell count, indicating that the cells recovered well from nucleofection (Supplemental Fig. 1E). We also tested CRISPR/Cas9 deletion efficiency by molecular and protein-level methods to evaluate concordance. Indel analysis of sequencing data (ICE) and flow cytometry analysis of our target gene protein expression typically showed similar levels of gene ablation (Supplemental Fig. 1F). Representative gating schemes and protein ablation frequency for all data are shown in Supplemental Fig. 2.

We then assessed the phenotype and function of expanded primary human NK cells after CRISPR. To expand primary human NK cells, we incubated them for 7 d by coincubating them with IL-2, IL-15, and irradiated K562 aAPCs, which express 4-1BB ligand and IL-21 (see Table I for reagents, resources, and tools used in this study). However, expanding human primary NK cells in vitro can affect the NK cell phenotype (35). We were concerned that activation and expansion could trigger a disintegrin and metalloprotease 17 (ADAM17) activity, which cleaves CD16A from the NK cell surface (38, 39). Thus, to evaluate the effect of our expansion protocol on cell phenotype, we barcoded expanded primary human NK cells from seven subjects and examined the expression of 350 surface proteins pre- and postexpansion using BioLegend’s LEGENDScreen. As expected, we observed that many proteins were significantly upregulated and downregulated during expansion (Supplemental Fig. 1G). However, we did not see loss of CD16A expression, which indicated that our expanded cells would retain ADCC function. We confirmed this by measuring ADCC-induced degranulation and cytokine production with an anti-RBC ADCC assay, in which RBCs coated with anti-human RBC polyclonal Abs were used as target cells. On day 6 post-CRISPR, NK cells yielded high levels of degranulation (CD107a⁺) and IFN-γ production (Supplemental Fig. 1H, 1I). Overall, these results show that our expansion and CRISPR/Cas9 protocol generates gene-ablated expanded primary NK cells that can be used in functional assays (Fig. 1A).

We used our CRISPR/Cas9 gene-editing protocol to test whether the loss of proteins associated with adaptive NK cells affected ADCC and cytokine production. A representative gating scheme used for analyzing ADCC assays with gene-ablated NK cells is shown in Supplemental Fig. 2A. As a control, we first ablated both FcRγ and CD3ζ (CD247) from expanded primary NK cells and then tested the gene-ablated cells in anti-RBC ADCC assays. NK cells require expression of at least one of these adapter proteins for downstream CD16A signaling, so we predicted that ablation of both genes would preclude degranulation and cytokine production. As expected, we found that simultaneous ablation of FcRγ and CD3ζ decreased CD107a, IFN-γ, and TNF-α expression in every subject, with an average decrease of 80–90% for each marker (Fig. 1B, 1G). We were also interested in ablating genes from the NK-92 cell line to validate our results (32, 37). NK-92 cells lack the FcRγ-chain (37), meaning that ablation of CD3ζ should preclude any functional activity. Thus, we ablated CD3ζ from two NK-92 cell lines: one expressing CD16A (NK-92-GFP-CD16A) and the other expressing a recombinant high-affinity Fc receptor consisting of the CD16A cytoplasmic tail and CD64 extracellular domain (NK-92-GFP-CD64/16A). We then performed DELFIA killing assays with CD3ζ-ablated NK cells and SKOV-3 cells coated with trastuzumab (anti-HER2) as targets. Ablation of CD3ζ abrogated target cell killing in both NK-92-GFP-CD16A (Supplemental Fig. 3A, 3E–H) and NK-92-GFP-CD64/16A (Supplemental Fig. 3C, 3I–L) cell lines. These controls show that our CRISPR/Cas9 sgRNAs were on target and that differences in biological function could be revealed in both expanded primary NK and NK-92 cells.

We next used CRISPR/Cas9 to test whether adaptive NK cells have enhanced function because they lack the FcRγ-chain. FcRγ only has one ITAM whereas CD3ζ contains three ITAMs. Thus, we hypothesized that if FcRγ is absent from NK cells, they will signal exclusively through CD3ζ, resulting in a stronger ADCC signal due to the increased ITAM number on CD3ζ. We also hypothesized that if the lack of the FcRγ-chain enhances ADCC and cytokine production, lack of the CD3ζ chain may have the opposite effect. We ablated FcRγ and CD3ζ independently and then used the gene-ablated cells in anti-RBC ADCC assays. Surprisingly, we found that ablation of either the FcRγ- or CD3ζ-chain does not significantly affect degranulation or IFN-γ production (Fig. 1C, 1D, 1G). TNF-α was expressed in a significantly higher frequency of FcRγ-negative cells (Fig. 1C), but this was not true for CD3ζ-negative cells (Fig. 1D). Overall, although ablating the FcRγ-chain increases TNF-α production during ADCC, neither ablation of the FcRγ-chain nor the CD3ζ-chain consistently alters ADCC function.

Adaptive NK cells also lack PLZF (ZBTB16) expression (22). PLZF controls expression of genes associated with adaptive NK cell activity, such as FcRγ and SYK (SYK), leading to some speculation that PLZF may be the regulator of adaptive NK cell development (22). Without PLZF, FcRγ and other genes may not be transcribed, leading to stronger ADCC and cytokine production. We hypothesized that ablation of PLZF may trigger the adaptive NK cell transcriptional program, resulting in enhanced ADCC and cytokine production in PLZF-ablated expanded primary NK cells. To test this hypothesis, we ablated PLZF individually or PLZF together with FcRγ (FcRγ/PLZF). We found that there was no significant difference in cytotoxicity or cytokine production with PLZF- or FcRγ/PLZF-ablated cells at 6 d post-CRISPR/Cas9 in the anti-RBC ADCC assay (Fig. 1E–G). However, we were concerned that testing PLZF-ablated cells at 6 d post-CRISPR/Cas9 may be too soon to observe an effect. Because PLZF is a transcription factor, we speculated that changes in transcription could take longer than 6 d post-CRISPR/Cas9 to manifest. We therefore tested PLZF- and FcRγ/PLZF-ablated cells for ADCC function at 6, 13, and 20 d post-CRISPR/Cas9. Neither PLZF- nor FcRγ/PLZF-ablated NK cells showed enhanced degranulation (Fig. 1H, 1I), IFN-γ (Fig. 1J, 1K), or TNF-α production (Fig. 1L, 1M) at days 13 or 20 post-CRISPR. Overall, these results showed that ablation of PLZF did not significantly affect NK cell ADCC and cytokine production in this anti-RBC ADCC assay.

In addition to lacking FcRγ and PLZF, adaptive NK cells frequently lack the tyrosine kinase SYK (21, 22). Both SYK and the related ZAP70 transduce signals from FcRγ and CD3ζ for ADCC. However, loss of SYK has been correlated with enhanced ADCC function in primary NK cells (21). We next tested whether lack of SYK or ZAP70 would affect ADCC and cytokine production. We ablated SYK or ZAP70 (ZAP70) from expanded primary NK cells and then examined the function of the gene-ablated cells in anti-RBC ADCC assays. We found that SYK ablation significantly enhanced degranulation, with an average increase of 7% over no RNP controls (Fig. 2A, 2D). Of the 35 samples tested, 26 showed enhanced degranulation with SYK ablation (Fig. 2A, 2D). In contrast, ZAP70 ablation significantly diminished NK cell degranulation, with an average decrease of 11.9% versus no RNP controls and decreases in 15 of 22 subjects tested (Fig. 2A, 2D). For both SYK and ZAP70 ablation, cytokine production followed the same pattern as degranulation. SYK-ablated cells produced significantly more IFN-γ (Fig. 2B, 2D) and TNF-α (Fig. 2C, 2D) than did controls, whereas ZAP70-ablated cells expressed significantly less IFN-γ (Fig. 2B, 2D) and TNF-α (Fig. 2C, 2D).

Although CRISPR/Cas9 gene editing with expanded primary human NK cells is a valuable model for understanding NK cell function in vivo, high levels of intersubject variability can cloud biological differences. We speculated that CMV status may account for some of this variability between donors, so we analyzed ADCC data by stratifying the tested donors based on CMV status. No significant differences were observed in ADCC between CMV-positive and CMV-negative subjects after FcRγ or SYK ablation (Fig. 2E, 2F), indicating that CMV status did not affect the outcomes of gene ablation. Finally, to confirm the effects of SYK ablation on NK cell function, we ablated SYK in NK-92-GFP-CD16A and NK-92-GFP-CD64/16A cells and tested subsequent cytotoxicity using DELFIA killing assays with SKOV-3 as NK target cells. The results mirrored our expanded primary NK cell ADCC results, with SYK ablation enhancing killing activity in both cell lines (Supplemental Fig. 3B, 3D–L). Overall, these results indicate that SYK ablation enhances ADCC and cytokine production, whereas ZAP70 ablation has the opposite effect.

We followed these experiments by exploring the mechanism behind our results. First, we tested whether SYK or ZAP70 ablation was altering baseline CD16A surface expression. Because CD16A is the receptor that binds the IgG Fc region, upregulation or downregulation of CD16A could result in the enhanced or diminished function we observed in gene-ablated cells. We measured surface CD16A expression in FcRγ/CD3ζ-, SYK-, and ZAP70-ablated expanded primary NK cells prior to ADCC. FcRγ/CD3ζ ablation diminished CD16A expression (Supplemental Fig. 3P). This was expected, as NK cells require either FcRγ or CD3ζ to express CD16A on the cell surface (40–42). However, most SYK- and ZAP70-ablated cells expressed CD16A at similar baseline frequencies (Supplemental Fig. 3Q, 3R).

Although SYK-ablated samples showed no difference from control cells in CD16A surface expression before stimulation, we were curious whether differences would be detectable after ADCC. Stimulation of NK cells can downregulate surface CD16A expression via ADAM17 (38, 39, 43). To test this, we stained SYK-ablated samples for CD16A surface expression after ADCC assays were performed. Stimulated SYK-ablated samples showed a higher frequency of CD16A surface expression than no RNP controls (Supplemental Fig. 3S). This could indicate that SYK-ablated samples maintain CD16A surface expression better than canonical NK cells during stimulation, leading to enhanced ADCC function.

As another potential mechanism, we also tested whether SYK- or ZAP70-ablated expanded primary NK cells differed in their ability to form stable cell–cell conjugates. We postulated that SYK-ablated cells may conjugate more effectively to target cells, leading to enhanced cytotoxicity and target cell killing (32, 44). At 6 d post-CRISPR/Cas9, SYK- and ZAP70-ablated cells were labeled with CTV, then incubated for various periods with fluorescent Ab-tagged SKOV-3 (CTFR-labeled) or Raji (NLR-expressing) cells. After incubation, samples were fixed and analyzed by flow cytometry as shown in Supplemental Fig. 3T (SKOV-3) and S3U (Raji). NK cell–target cell conjugates were detected as single events that were positive for both CTV and NLR/CTFR. At 30 min postcombination, SYK-ablated expanded primary NK cells showed enhanced conjugation with both anti-HER2–labeled SKOV-3 (Fig. 3A) and rituximab-labeled Raji (Fig. 3B) cells as targets. In contrast, fewer ZAP70-ablated expanded primary NK cells conjugated to target cells relative to the no RNP controls (Fig. 3C, 3D). In NK cell–mediated ADCC, adhesion molecules, such as CD2 and LFA-1, mediate target cell conjugation to form the immunological synapse (45, 46). During ADCC, CD2 has also been shown synergize with CD16A, leading to enhanced cytotoxicity and cytokine production (45, 47). We hypothesized that because SYK ablation enhanced cytotoxicity, cytokine production, and target cell conjugation during ADCC, SYK-ablated cells would show increased CD2 or LFA-1 expression at baseline. To test this, unstimulated SYK- and ZAP70-ablated expanded primary NK cells were stained for CD2 or LFA-1 expression at 6 d post-CRISPR/Cas9. SYK-ablated cells expressed CD2 at a significantly higher frequency than control samples (Fig. 3E). Additionally, CD2 geometric median fluorescence intensity (gMFI) was trending higher in SYK-ablated samples, although this difference was insignificant (Fig. 3F). ZAP70-ablated samples showed the opposite phenotype of SYK-ablated cells with significantly fewer ZAP70-ablated samples expressing CD2 (Fig. 3G) and at a significantly lower gMFI (Fig. 3H) than no RNP controls. In contrast, SYK- and ZAP70-ablated samples showed no difference from control cells in expression of both the closed and open conformations of LFA-1 (data not shown). These results indicate that SYK ablation enhances target cell conjugation and that this could be due to increased CD2 expression.

We next explored other molecules in the CD16A signaling pathway, including SHP-1. SHP-1 is a phosphatase that can dephosphorylate ZAP70 (48, 49) and induce ITAM-mediated inhibition (ITAMi) (50–52). ITAMi is a process in which weakly ligated receptors recruit SHP-1 instead of SYK or ZAP70 to partially phosphorylated ITAMs, allowing SHP-1 to dephosphorylate the ITAM and terminate signaling (53, 54). Interestingly, ITAMi can also be facilitated by SYK dissociating from the partially phosphorylated ITAM of a weakly occupied receptor, allowing SHP-1 to dephosphorylate the ITAM (55, 56). We thus hypothesized that SHP-1 ablation would enhance NK cell–mediated ADCC and cytokine production. To test this, we ablated SHP-1 (PTPN6) from expanded primary NK cells and tested the SHP-1–ablated cells’ ADCC function and cytokine production. In line with our hypothesis, SHP-1 ablation robustly enhanced degranulation in the anti-RBC ADCC assay. On average, 43.55% of SHP-1–ablated samples were CD107a⁺, compared with 34.43% of subject-matched, no RNP samples in the assay (Fig. 4A, 4D). However, both IFN-γ (Fig. 4B, 4D) and TNF-α (Fig. 4C, 4D) were expressed at lower frequencies in SHP-1–ablated samples relative to no RNP controls. Because cytokine production reflected degranulation in both SYK- and ZAP70-ablated cells, we were surprised that SHP-1 ablation resulted in enhanced ADCC but reduced cytokine production. Overall, although our CD107a data suggest that SHP-1 ablation increases cytotoxicity, the effect of SHP-1 ablation on cytokine production is both confounding and interesting.

Next, we tested whether our ADCC degranulation results in the anti-RBC assay would correlate with ADCC toward a SKOV-3 target cell line. We hypothesized that gene-ablated cell killing would reflect CD107a expression in the anti-RBC ADCC assays. We used two complementary assays to measure target cell killing: a molecule release DELFIA assay and cell tracking with an Incucyte machine. In DELFIA assays, SKOV-3 target cells are loaded with BATDA, then incubated with effector cells (e.g., NK cells) for 2–6 h. After incubation, BATDA levels in the supernatant are measured, which reflects pores generated in the target cell or necrotic death releasing the BATDA into the supernatant. For DELFIA assays, we ablated FcRγ/CD3ζ, SYK, ZAP70, and SHP-1 from expanded primary NK cells. At 6 d post-CRISPR/Cas9, the gene-ablated cells were incubated with anti-HER2 and BATDA-loaded SKOV-3 cells. In parallel with the anti-RBC results, ablating FcRγ/CD3ζ significantly diminished ADCC function in DELFIA assays (Fig. 5A–D). We found that SHP-1 ablation resulted in significantly more specific lysis of target cells with 10 (Fig. 5A, 5B) and 3 μg/ml anti-HER2 (Fig. 5C, 5D). SYK-ablated cells killed significantly more target cells with 3 μg/ml anti-HER2 only (Fig. 5A–D). In the 10 μg/ml DELFIA assays, SYK ablation increased target cell killing in 8 of 10 subjects with a median specific lysis of 86.6%, an 8.1% increase over the no RNP group. However, these increases were insignificant (Fig. 5A, 5B). We also observed significantly less killing in the ZAP70-ablated samples (Fig. 5A–D). Although the 3 and 10 μg/ml DELFIA assays showed slightly different results, the trends are similar for each ablated gene.

To measure target cell killing over time, we used a microscopy-based live cell tracking Incucyte machine. Fluorescent SKOV-3 cells were opsonized with anti-HER2, incubated with gene-ablated and no RNP NK cells, and then imaged every hour for 24 h. We hypothesized that SKOV-3 death would correlate with degranulation and DELFIA readouts. Once again, we used FcRγ/CD3ζ-, SYK-, ZAP70-, and SHP-1–ablated cells in Incucyte assays at 6 d post-CRISPR/Cas9. FcRγ/CD3ζ-ablated samples killed significantly fewer target cells (Fig. 5E, 5F). In line with our other results, we saw that SYK ablation significantly enhanced killing function (Fig. 5E, 5F). Although SHP-1–ablated samples followed the same trends as in the DELFIA assays, these differences were not significant (Fig. 5E, 5F). ZAP70-ablated samples showed no difference from control cells in the Incucyte assay (Fig. 5E, 5F). Overall, degranulation, DELFIA-specific release, and Incucyte area under the curve results correlated with each other (Supplemental Fig. 3M–O). Of note, the magnitude of difference between SYK- and SHP-1–ablated cell killing of SKOV-3 cells versus no RNP cells was greater than the difference between the anti-RBC degranulation assay results (Figs. 2D, 4D, 5B, 5D, 5F). Taken together, these data support our hypothesis that SYK ablation enhances cytotoxic function in CMV-induced adaptive NK cells.

Ex vivo expansion of NK cells and subsequent transplantation is an exciting avenue for cancer treatment (57). However, the success of these therapies often depends on the ability of NK cells to survive and retain function postexpansion and posttransplant (58). Although our main goal was to understand the mechanism by which adaptive NK cells gain enhanced ADCC function, we also wanted to test whether SYK-ablated NK cells could survive and retain enhanced function postexpansion. Therefore, we ablated SYK from expanded primary human NK cells, then used them in anti-RBC degranulation and cytokine production assays at days 6, 13, 20, and 27 post-CRISPR. Across nearly all time points, SYK ablation resulted in enhanced CD107a (Fig. 6A), IFN-γ (Fig. 6B), and TNF-α (Fig. 6C) expression. SYK-ablated samples also retained enhanced killing in DELFIA (Fig. 6D–G) and Incucyte (Fig. 6H, 6I) assays with SKOV-3 cells as NK targets. Overall, these results show that SYK ablation’s effect on ADCC function is sustained after expanding the cells for 4 wk post-CRISPR, indicating that SYK-negative cells could be considered for NK cell–based ADCC therapies.

Adaptive NK cells show enhanced function relative to canonical NK cells (12, 19, 21, 22), and higher frequencies of adaptive NK cells correlate with protection against diseases such as CMV, HIV, and malaria (18, 24, 26). However, it was unclear which intracellular molecules are required to mediate enhanced adaptive NK cell function and whether the same molecules control both cytotoxicity and cytokine production. In this study, we show that ablating SYK kinase in expanded primary NK cells enhances ADCC, cytokine production, and target cell conjugation. SYK-ablated expanded primary NK cells retained their enhanced function up to 27 d post-CRISPR. Overall, these results indicate that a mechanism involving SYK kinase, including but not limited to increased conjugation through CD2 and perhaps involving decreased ITAM inhibition through SHP-1, could explain why CMV-induced adaptive NK cells show enhanced ADCC and cytokine production relative to conventional NK cells (17–22). These results also suggest that SYK-ablated NK cells may have increased ADCC function in vivo.

Previous work indicated that adaptive NK cells often lack expression of the transcription factor PLZF and FcRγ. In primary NK cells, loss of FcRγ has been associated with enhanced function, leading to hypotheses that the lack of FcRγ results in a stronger ADCC signaling (20, 22). Additionally, PLZF binds the FcRγ promoter region (22). Thus, it seemed possible that loss of PLZF may drive loss of FcRγ, which could lead to enhanced ADCC function. We found that FcRγ ablation led to a modest increase in TNF-α production but no significant increase in degranulation or IFN-γ expression. In PLZF-ablated samples, neither ADCC nor cytokine production was enhanced for up to 20 d post-CRISPR. Despite these results, PLZF may still be an important transcription factor for adaptive NK cell biology, as other transcription factors could have compensated for its loss in our study. In other work, Liu et al. (34) ablated FcRγ and PLZF individually in primary human NK cells and cross-linked CD16A to test ADCC function. Similar to Liu et al., we observed that ablating PLZF and FcRγ/PLZF did not enhance degranulation or cytokine production during ADCC. However, Liu et al. also showed that FcRγ ablation directly enhanced both IFN-γ and TNF-α production, but not degranulation, during ADCC. By cross-linking CD16A, Liu et al. directly stimulated the CD16A pathway. In contrast to Liu et al., we used two cellular targets to assess ADCC and cytokine production: RBCs and SKOV-3 cells. Interactions between surface proteins on target and NK cells can modulate NK cell ADCC activity (59, 60), which may explain the differences between our data and prior work. Given our findings, if gene-ablated NK cells are to be used in clinical settings, this emphasizes the need to test the target cell of interest in preclinical assays. Overall, our results show the potential for enhanced cytokine production in FcRγ-negative NK cells but they also indicated that lack of the FcRγ-chain does not universally enhance ADCC function.

In addition to lacking FcRγ, adaptive NK cells can lack SYK kinase (21, 22, 61). Loss of SYK was not observed in adaptive NK cells from malaria subjects in Mali (26), but it has been observed regularly during CMV infection. Similar to FcRγ-negative cells, SYK-negative primary NK cells exhibit enhanced ADCC function (21, 22). In ADCC, SYK preferentially binds the FcRγ-chain whereas ZAP70 preferentially binds the CD3ζ-chain (8). Thus, it seemed possible that lack of the FcRγ-chain in adaptive NK cells results in less SYK-mediated ADCC signaling, indicating a potential role for SYK in controlling ADCC signal strength. SYK ablation enhanced ADCC and cytokine production, whereas ZAP70 ablation decreased both, indicating that SYK and ZAP70 mediate ADCC signal strength.

By evaluating potential mechanisms behind this phenomenon, we observed that SYK ablation also increases stable NK cell conjugate formation. In addition, SYK-ablated NK cells expressed the adhesion molecule CD2 at a higher frequency than did controls. Previous reports have shown that CD2 expression predicts NK cell–mediated ADCC levels (45, 47). Thus, our findings could indicate that increased CD2 expression in SYK-ablated cells enhances target cell binding, contributing to increased cytotoxicity, cytokine production, and killing.

Beyond conjugation, we examined SHP-1 regulation of ADCC signal strength as a potential contributor to the functional behavior we observed when altering CD16A signaling partners. SHP-1 is a phosphatase that can modulate signaling in multiple pathways. Relevant to this study, SHP-1 can dephosphorylate ZAP70 in T cells (48, 49). In addition, SHP-1 has been shown to mediate ITAMi, in which SHP-1 is recruited to weakly ligated Fc receptor ITAMs and dephosphorylates them, dampening signaling (50, 52, 56). We show that SHP-1 ablation enhances degranulation and killing, in line with the phenotype observed for SYK-ablated NK cells. Because both of these components can inhibit ADCC signaling, it is probable that SHP-1 is partially responsible for the SYK ablation functional phenotype, which warrants further investigation. Interestingly, IFN-γ production from SHP-1–ablated NK cells was consistently lower than control NK cells, the opposite result of SYK ablation. One explanation could be different CD16A binding thresholds to initiate degranulation versus cytokine production. Fauriat et al. (62) proposed a model for NK cell function in which CD107a signaling requires a low threshold and triggers granule release immediately after stimulation whereas TNF-α secretion requires a moderate threshold and longer stimulation (∼3 h), and IFN-γ secretion requires the highest threshold and longest stimulation (∼5.5 h). Concurrently, strong activation through CD16A yields a self-regulation loop in which CD16A is enzymatically cleaved by ADAM17 (38, 39, 63, 64). Thus, it is possible that without SHP-1 negatively regulating CD16A signaling, SHP-1–ablated cells are rapidly activated and degranulate. This could lead to ADAM17 cleaving CD16A from the cell surface, resulting in a lack of signal for eventual IFN-γ production. Lack of SHP-1 has been associated with overactive immune cells (65). Thus, we hypothesize that SHP-1 ablation eliminates inhibition of both SYK (through ITAMi) and ZAP70 (by dephosphorylation) activity, resulting in robust CD16A signaling, rapid ADAM17-induced cleavage of CD16A, and loss of the hours-long CD16A stimulation required for IFN-γ production. Adhesion molecules could also contribute to this phenotype, as SHP-1 regulates adhesion in T cells (66) and SHP-1 phosphorylation is associated with more sequential contacts in NK cells (67). Additional studies beyond the scope of this study are needed to test these hypotheses. Lastly, one limitation of the CRIPSR/Cas9 method is that the cells need to be proliferating for the CRISPR/Cas9 to effectively ablate genes. No gene ablation was detected when the CRISPR was done on freshly isolated resting NK cells. This limits our study of testing the effect of gene ablation to only proliferating NK cells. In the future, to test the function of SYK and ZAP70 in resting NK cells, one could use drugs specific for SYK or ZAP70 or use a novel CRISPR method that is able to ablate genes in resting NK cells. We hypothesize the results would be similar to the proliferating cells. However, this would need to be directly tested.

Our results reveal that mechanisms involving a lack of SYK kinase, such as enhanced target cell conjugation or reduced SHP-1–mediated inhibition, could explain the enhanced ADCC function in CMV-induced adaptive NK cells. However, they do not explain adaptive NK cell behavior in all settings. Adaptive NK cells from malaria patients in Mali lacked FcRγ-chain and PLZF expression and showed enhanced ADCC and cytokine production in vitro (26). However, in contrast to CMV-induced adaptive NK cells, >90% of adaptive NK cells in the Malian cohort were SYK-positive and only 50% expressed NKG2C (26). The differences indicate that malaria-induced adaptive NK cells may contain novel origins and phenotypes and highlight the potential heterogeneity of adaptive NK cells induced by different infections (26). Although our study shows how SYK ablation can enhance degranulation and cytokine production with a CMV-induced adaptive NK cell phenotype, there may be multiple mechanisms driving this enhanced function. Understanding other mechanisms that increase cytotoxicity and cytokine production will be important toward understanding NK cell function and improving their use as a therapeutic intervention.

The CRISPR/Cas9 method we developed allowed us to ablate molecules in the CD16A pathway and define the consequences of the ablations on CD16A signaling in expanded primary human NK cells. These findings build on previous work in the NK and Fc receptor signaling fields as well as adaptive NK cell biology. With gene editing in expanded primary human NK cells, the field can address new questions on how adaptive NK cells function. Future studies will focus on understanding why SYK ablation enhances NK cell function, ZAP70 ablation decreases function, and SHP-1 ablation does both. In summary, these findings help define what drives enhanced cytotoxicity and cytokine production in CMV-induced adaptive NK cells and show that SYK ablation may be a viable therapy for chimeric Ag receptor NK cells that use the CD16A signaling.

The authors have no financial conflicts of interest.

We especially thank Patrick Willey and the rest of the staff at the University of Minnesota University Imaging Centers (SCR_020997), the University of Minnesota flow cytometry core, and the University of Minnesota Genomics Center. We also especially thank Peter Hinderlie (Dr. Jeffrey Miller’s laboratory) for technical support in setting up Incucyte assays and analyzing results. We thank Dr. Peter Morawski (Benaroya Research Institute) for helping with a confirmatory analysis (Infinity Flow R program) for our BioLegend LEGENDScreen results. We especially thank the National Cancer Institute for supplying IL-2 and IL-15 and Synthego for the guide RNA design tool for CRISPR/Cas9. We thank Dr. Jai Rutella and Dr. Jeffrey Miller’s NK Cell Immunology Research Forum group for expertise and advice. We also thank Dr. Joshua Baller for computational and statistical assistance with the data and manuscript. We thank Sloane Fussell for providing model figure graphics.