Quantifying NK-cell Cytotoxicity Potentiated by Chimeric Antigen Receptors and Therapeutic Antibodies Using Live-cell Imaging and Flow Cytometry

Summary

This protocol provides a standardized in vitro. workflow for consistent measurement of natural, chimeric antigen receptor–mediated, and antibody-dependent cytotoxicity of human natural killer cells using live-cell imaging and flow cytometry.

Abstract

Natural killer (NK) cells are innate lymphocytes that play a critical role in protective immunity against diverse intracellular pathogens and cancers. Their primary function is to kill target cells that are infected, malignantly transformed, or coated by antibodies via antibody-dependent cellular cytotoxicity (ADCC). NK cells can also be genetically engineered to express chimeric antigen receptors (CARs) that enable targeted recognition of specific antigens. Quantitative measurement of NK-cell cytotoxicity is essential for assessing baseline functionality and for preclinical evaluation of monoclonal antibodies and CAR-engineering strategies. However, in vitro functional assays remain highly variable across laboratories due to differences in cell preparation, target cells, effector-to-target ratios, co-incubation times, and readout methods, limiting reproducibility and cross-study comparisons. This article presents a standardized protocol for quantitative assessment of NK-cell cytotoxicity using flow cytometry and real-time, live-cell imaging. Primary human NK cells and CAR-expressing NK-92 cells were evaluated for their ability to kill cancer cells and antibody-coated target cells in a 96-well plate format to measure natural cytotoxicity, CAR-mediated killing, and ADCC. Target-cell survival was measured either continuously using live-cell imaging or at a defined time point by flow cytometry, which also enabled phenotypic characterization of NK cells and target cells. These protocols provide a robust framework using routine tissue culture, imaging, and flow cytometry methods to enable reproducible quantification of NK-cell effector functions for studies of innate immunity and NK cell–based immunotherapies.

Introduction

Natural killer (NK) cells are innate lymphocytes that play a central role in immune surveillance against virally infected and malignant cells, exhibiting rapid and potent cytotoxicity¹. Unlike B and T cells, NK cells do not express antigen-specific receptors and instead rely on an array of activating and inhibitory receptors to detect loss of major histocompatibility complex class I expression or cellular stress. This enables them to mediate direct cytotoxicity by releasing cytotoxic granules containing perforin and granzymes or, in some cases, by activating death receptors^1,2,3,4. In addition to natural cytotoxicity, NK cells express FcγRIII (CD16), a receptor that recognizes the Fc portion of immunoglobulin G (IgG) and mediates antibody-dependent cellular cytotoxicity (ADCC) against IgG-coated target cells^5,6.

Because of their intrinsic cytotoxic mechanisms and favorable safety profile, NK cells have emerged as promising candidates for cellular immunotherapy. With the potential for allogeneic use, NK cells can function as an “off-the-shelf” therapy with minimal risk of graft-versus-host disease, enabling broader clinical applicability and reduced costs⁷. Recent advances have further expanded the therapeutic potential of NK cells through genetic engineering strategies, including the introduction of chimeric antigen receptors (CARs) that redirect NK cells to recognize specific tumor antigens. CAR-NK cells have demonstrated antitumor efficacy in in vivo xenograft models and clinical trials, with reduced toxicity compared to CAR-T cells^8,9,10. Therefore, there is growing interest in developing engineered NK-cell platforms and antibody-based therapeutic strategies. In this context, accurate and quantitative measurement of NK-cell killing capacity is essential for understanding NK-cell biology, assessing the efficacy of CAR or other engineered NK-cell therapies, and evaluating monoclonal antibodies that rely on ADCC.

A variety of in vitro assays have been widely used across laboratories to evaluate NK-cell cytotoxicity. Common methods include radioactive chromium-51 release assays, lactate dehydrogenase release assays, and calcein–acetoxymethyl release assays^11,12. While these traditional methods have been widely used to measure cytotoxicity, they rely on indirect readouts of cell death, provide endpoint measurements rather than kinetic data, and offer limited sensitivity and dynamic range. They also do not provide information about the phenotype or activation state of effector cells. More recently, live-cell imaging platforms and flow cytometry have been increasingly adopted because they enable direct quantification of target-cell survival, monitoring of killing over time, and simultaneous phenotypic analysis of effector and target cells^13,14. However, even among these newer approaches, substantial variability exists across laboratories and published studies, including differences in target-cell preparation, effector-to-target (E:T) ratios, and co-incubation times, limiting reproducibility and cross-study comparisons, particularly in preclinical evaluations of NK cell–based immunotherapies.

This article presents a standardized workflow for quantitative assessment of NK-cell cytotoxicity using two independent, complementary platforms: real-time, live-cell imaging (IncuCyte SX5) and flow cytometry–based endpoint assays. Live-cell imaging enables continuous monitoring of target-cell survival during co-culture with effector cells, allowing kinetic evaluation of cytotoxicity across multiple E:T ratios and experimental conditions with minimal assay manipulation¹⁵. In contrast, flow cytometry provides a highly sensitive endpoint measurement of target-cell death, along with simultaneous phenotypic characterization of NK cells and target cells using a multimarker staining panel.

Additionally, this protocol incorporates standardized experimental parameters, including E:T ratios, incubation conditions, plate format, and analytical workflows, to improve reproducibility and facilitate comparison of cytotoxicity measurements across laboratories and studies. The workflow enables quantitative assessment of multiple NK effector functions, including natural cytotoxicity, CAR-mediated killing, and ADCC, and is compatible with primary human NK cells and NK cell lines. This protocol provides a practical framework for studies of NK-cell biology and preclinical evaluation of NK cell–based immunotherapy development.

Protocol

Use of the human donor samples for research was approved by the Institutional Review Board and conducted in accordance with institutional guidelines (Approval #2022P000699).

1. Real-time live-cell imaging killing assay

1. Coat plates
   1. Add 50 µL/well poly-D-lysine (0.1 mg/mL in phosphate-buffered saline [PBS]) to a 6 × 10 grid in a flat-bottom 96-well plate. Gently shake the plate to ensure the bottom of each well is fully covered.
   2. Incubate the plate for 2 h at room temperature (RT; 18°C–26°C) with the lid on.
   3. Wash wells 3× with 200 µL/well PBS and discard the supernatant.
      NOTE: Coated plates can be sealed with parafilm and stored at 4°C for up to 1 week with 200 µL of PBS added to each well. Ensure plates remain sterile, discard PBS, and allow them to equilibrate to RT before use.
2. Prepare target cells
   1. Prepare 10 mL of 2× SYTOX Orange (200 nM) by diluting SYTOX Orange stock solution (250 µM in DMSO) into NK-92 culture media (add 8 µL stock to 10 mL total volume) in a 15-mL conical-bottom centrifuge tube.
      NOTE: Culture media for NK-92 cells consists of Advanced RPMI-1640 medium, 10% fetal bovine serum (FBS), 2 mM L-alanyl-L-glutamine, 100 µg/mL primocin, and 500 U/mL human IL-2.
      CAUTION: SYTOX Orange is a fluorescent dye and may pose risks upon skin or eye exposure. Handle with appropriate personal protective equipment and in accordance with institutional safety guidelines.
   2. Count target cells using a dye exclusion–based cell counting method (e.g., trypan blue) or an equivalent validated method (e.g., automated cell counter).
      NOTE: In this assay, target cells were Ramos cells, a B-cell maturation antigen-expressing (BCMA⁺) B-cell lymphoma cell line, that we engineered to express the near-infrared (NIR) fluorescent protein emiRFP670.
   3. Transfer the required amount into a 15-mL conical-bottom centrifuge tube.
   4. Centrifuge (5 min, 300 × g, RT, acceleration = maximum; deceleration [brake] = maximum).
   5. Discard the supernatant.
   6. Resuspend target cells to a final concentration of 2.5 × 10⁵ cells/mL (2.5 × 10⁴ cells/100 µL) in 2× SYTOX Orange media.
      NOTE: Resuspended target cells can be maintained at 37°C with 5% CO₂ in the tube for up to 30 min while preparing effector cell dilutions.
      NOTE: For one full plate, 60 wells × 2.5 × 10⁴ cells/well = 1.5 × 10⁶ total target cells. Prepare 2.0 × 10⁶ cells to account for loss. Adjust cell numbers based on assay size and desired seeding density.
3. Prepare effector cells
   1. Assess the concentration and viability of NK-92 cells using a dye exclusion–based cell counting method (e.g., trypan blue) or an equivalent validated method. Proceed only if viability is ≥80%, as lower viability may impact assay performance.
      NOTE: NK-92 cells engineered to stably express the fluorescent protein ZsGreen, either alone (control) or with BCMA-targeted CAR constructs, were used as effector cells.
   2. Count and transfer the required number of cells into individual 15-mL conical-bottom centrifuge tubes.
      NOTE: Calculate the total number of effector cells required by summing the number of cells needed across all E:T ratios for one replicate, multiplying by the number of replicates, and adding 50% excess to account for cell loss. Round numbers as needed to facilitate preparation. For example, for E:T ratios ranging from 2:1 to 0.25:1, use 1.5 × 10⁵ NK cells per replicate. For triplicates, prepare 4.5 × 10⁵ cells per condition based on the plate layout described in Supplemental Figure 1A.
   3. Centrifuge cells (5 min, 300 × g, RT, acceleration = maximum; deceleration [brake] = maximum).
   4. Discard the supernatant.
   5. Resuspend cells to a final concentration of 5.0 × 10⁵ cells/mL (5.0 × 10⁴ cells/100 µL).
4. Prepare effector cell dilutions
   1. Prepare effector cell dilutions, including a target-only control, in a round-bottom 96-well plate (see Supplemental Figure 1A for a representative plate layout with four conditions in triplicates). Each condition is prepared in triplicate.
   2. Add 240 µL/well of effector cells (5.0 × 10⁵ cells/mL) to wells corresponding to an E:T ratio of 2:1.
   3. Add 120 µL/well of media to all remaining wells.
   4. Perform 2-fold serial dilutions from E:T = 2:1 wells through E:T = 0.25:1 wells. Mix thoroughly by pipetting ≥5 times and change tips between transfers.
      NOTE: If imaging runs have not been scheduled on the imaging system, schedule them before setting up the co-culture. Prepared effector cell dilutions can be maintained at 37°C with 5% CO₂ in a round-bottom 96-well plate for up to 30 min while setting up the live-cell imaging system.
5. Set up co-culture
   1. Obtain the poly-D-lysine-coated flat-bottom 96-well plate.
   2. Add 100 µL/well of target-cell suspension to all wells. Thoroughly resuspend target cells immediately before dispensing by pipetting up and down ≥5 times using an appropriate pipette (e.g., 200 µL) and avoid bubble formation during transfer.
   3. Add 100 µL/well of effector-cell suspension to the corresponding wells. Thoroughly resuspend effector cells immediately before dispensing by pipetting up and down ≥5 times using an appropriate pipette (e.g., 200 µL) and avoid bubble formation during transfer.
   4. Add 200 µL/well PBS to the surrounding unused wells.
   5. Centrifuge the plate (2 min, 100 × g, RT, acceleration = low [e.g., 2]; deceleration [brake] = low [e.g., 2]).
      NOTE: This soft spin is critical to rapidly and evenly settle cells across the bottom of the well for accurate imaging on the IncuCyte SX5. We use a swinging bucket rotor–compatible benchtop Eppendorf Centrifuge 5810R with the acceleration and deceleration settings set to 2 (out of 9).
   6. Place the plate in the live-cell imaging system and acquire images every 1 h for 24 h (extendable up to 72 h if required). Ensure imaging is performed under standard incubation conditions (37°C/5% CO₂).
      NOTE: To ensure that imaging at t = 0 h accurately reflects baseline target-cell counts, start image acquisition within 10–15 min following co-culture setup and centrifugation. For a full plate with three imaging channels (i.e., green, orange, and NIR), scan time is approximately 10 min per hour for 3 images/well at 20× magnification.
   7. Verify image quality after the first scan and adjust settings if necessary.
6. Instrument settings
   1. Set spectral unmixing to correct for fluorescence spillover of SYTOX Orange into the green channel. Adjust the unmixing value in the software by comparing the green and orange channels and confirming that Orange⁺ dead cells are not falsely detected as Green⁺. For the representative experiment (Figure 1), the orange channel contribution to the green channel was set to 8.5%. These values are dataset-specific and should be optimized for each experiment.
   2. Configure non-adherent Cell-by-Cell analysis to ensure accurate segmentation and masking of all cells within each field of view (see Supplemental Figure 1B for representative setup). The following parameters are recommended for the live-cell imaging system using suspension effector and target-cell line models:
      1. Set the expected object diameter to 15 µm (approximate cell diameter).
      2. Set threshold sensitivity and texture sensitivity to 5.
      3. Set edge sensitivity to 10 to facilitate separation of cell clusters.
      4. Set the accepted object area filter range to 50–2000 µm² to exclude debris and large objects.
      5. Set eccentricity maximum to 0.95 to exclude non-cellular objects.
   3. Configure classification thresholds (see Supplemental Figure 1C for representative gating). Use target-only control wells to define the baseline NIR⁺/Orange⁻ live target-cell population. Then use co-culture wells at later time points (e.g., t = 4 h) to confirm clear separation with minimal overlap between NIR⁺/⁻ and Orange⁺/⁻ populations. Example classification thresholds and analysis parameters are shown in Supplemental Figure 1C.
      1. Set Orange intensity threshold to approximately 15.
      2. Set NIR intensity threshold to approximately 0.1.
         NOTE: Parameter and threshold values may vary depending on the instrument, software, cell line model, and fluorescence signal intensity.
         NOTE: This protocol can be adapted for primary human NK-cell cytotoxicity or ADCC assays if target cells stably express NIR fluorescence (e.g., emiRFP670). It can also be adapted for flow cytometry–based killing assays by omitting poly-D-lysine coating and using flat-bottom or round-bottom 96-well plates.

Figure 1. Quantifying kinetics and potency of chimeric antigen receptor (CAR) NK-92 cell–mediated cytotoxicity using real-time live-cell imaging. (A) Schematic overview of the IncuCyte live-cell imaging assay for monitoring NK-cell–mediated cytotoxicity. BCMA⁺ Ramos target cells expressing emiRFP670 (near-infrared [NIR] channel) are co-cultured with NK-92 effector cells expressing ZsGreen (green channel), and cell death is detected using SYTOX Orange (orange channel). (B) Representative fluorescence image of a co-culture well at an effector-to-target (E:T) ratio of 1:1 at 12 h. Live target cells (emiRFP670⁺) are shown in blue, NK-92 effector cells (ZsGreen⁺) in green, and dead cells (SYTOX Orange⁺) in red. Images were acquired using the non-adherent Cell-by-Cell analysis module at 20× magnification. Scale bar = 100 µm. (C) Quantification of target-cell growth and survival over time at varying E:T ratios for control and CAR-expressing NK-92 cells. (D) Quantification of target-cell survival as a function of E:T ratio at multiple time points (4, 8, 16, and 24 h). Target-cell counts were normalized to target-only wells (E:T = 0:1) at each time point. (E) Schematic representation of quantitative metrics used to assess cytotoxicity: KR₅₀, killer-to-target ratio required to achieve 50% killing at 12 h; KT₅₀, time required to achieve 50% killing at an E:T ratio of 1:1; and IKI₅₀, integrated killer index defined as the area of the heatmap below the 50% survival isocline. (F) Heatmaps showing target-cell growth and survival as a function of E:T ratio and time for each effector condition. Quantitative values for KR₅₀, KT₅₀, and IKI₅₀ are indicated for each condition. Raw live target-cell counts were obtained from the gated NIR⁺/SYTOX Orange^– population using Cell-by-Cell classification analysis in IncuCyte 2021C software. All conditions were performed in triplicate (n = 3). Data are presented as mean ± standard deviation. Nonlinear regression analysis was used to model cytotoxicity trends in panels D and F. Please click here to view a larger version of this figure.

2. Quantification of antibody-dependent cellular cytotoxicity by flow cytometry

1. Prepare primary human NK cells
   NOTE: Primary human NK cells were isolated from leukocyte-enriched peripheral blood obtained from healthy adult donors using an NK-cell enrichment method followed by density gradient centrifugation, as previously described³. Purified NK cells were cryopreserved at 5 × 10⁶ cells/mL in freezing solution (90% FBS and 10% dimethyl sulfoxide [DMSO]) as 0.5-mL aliquots in cryovials and stored at less than −150°C until use³.
   1. Thaw one vial containing 5 × 10⁶ primary human NK cells by gently agitating the cryovial in a 37°C water bath for approximately 1 min, until only a small ice crystal remains.
      CAUTION: Handle all human-derived samples using appropriate biosafety procedures and personal protective equipment in accordance with institutional guidelines.
      CAUTION: DMSO facilitates skin absorption of chemicals. Wear appropriate protective equipment and avoid direct contact.
   2. Transfer the thawed cells into a 15-mL conical-bottom centrifuge tube containing 10 mL of prewarmed (37°C) NK-cell culture medium and mix by gentle swirling.
      NOTE: NK-cell culture media consists of Advanced RPMI-1640 supplemented with 10% FBS, 10% human AB serum, 2 mM L-alanyl-L-glutamine, 100 µg/mL primocin, and 50 U/mL human IL-2.
   3. Centrifuge cells (5 min, 300 × g, RT, acceleration = maximum; deceleration [brake] = maximum).
   4. Discard the supernatant.
   5. Resuspend the pellet in 5 mL of prewarmed (37°C) NK-cell culture medium.
   6. Centrifuge cells (5 min, 300 × g, RT, acceleration = maximum; deceleration [brake] = maximum).
   7. Discard the supernatant.
   8. Resuspend the pellet in 1 mL of prewarmed NK-cell culture medium.
   9. Transfer the suspension into one well of a 24-well tissue culture plate.
   10. Add 1 mL sterile PBS to surrounding wells to minimize evaporation.
   11. Incubate the 24-well plate at a 45° angle (using a plate rack or support to maintain the angle) at 37°C with 5% CO₂ for 16–24 h to allow NK-cell recovery.
       NOTE: Primary human NK cells require close cell-cell contact for maximal survival and recovery, which can be achieved by tilting the 24-well plate and allowing accumulation at the edge of wells. Maintain consistent incubation time across experiments.
       NOTE: Following the recovery period, proceed directly to target-cell preparation and co-culture setup.
2. Stain target cells with carboxyfluorescein succinimidyl ester (CFSE).
   1. Obtain 5 × 10⁵ target cells and transfer into a 15-mL conical-bottom centrifuge tube.
      NOTE: A CD38⁺ patient-derived lymphoma cell line was used as target cells¹⁶.
   2. Wash cells once with 10 mL PBS.
   3. Centrifuge cells (5 min, 300 × g, RT, acceleration = maximum; deceleration [brake] = maximum).
   4. Prepare a 0.5 µM working solution of CFSE in PBS using a 5 mM stock solution by diluting 1 µL of 5 mM CFSE stock into 10 mL of PBS in a 15-mL conical-bottom centrifuge tube.
      CAUTION: CFSE is a reactive amine-binding fluorescent dye and may pose risks upon skin or eye exposure. Handle with appropriate personal protective equipment and dispose of unused reagent according to local biosafety guidelines.
   5. Discard the supernatant.
   6. Resuspend the pellet at approximately 1 × 10⁶ cells/mL in the CFSE solution.
   7. Incubate for 15 min at 37°C with the tube wrapped in aluminum foil or placed inside an opaque container to protect from light.
   8. Quench the reaction by adding 5–10 volumes of culture medium.
   9. Incubate for 5 min at 37°C.
   10. Centrifuge cells (5 min, 300 × g, RT, acceleration = maximum; deceleration [brake] = maximum).
   11. Discard the supernatant.
   12. Resuspend cells at 2.5 × 10⁵ cells/mL so that 100 µL contains 2.5 × 10⁴ cells.
       NOTE: Proceed directly to plating to maintain consistent CFSE signal intensity and target-cell viability.
3. Plate target cells
   1. Obtain a round-bottom 96-well plate.
   2. Add 100 µL/well of target cells (2.5 × 10⁵ cells/mL) to designated wells using a multichannel pipette (see Supplemental Figure 2). Each condition is plated in triplicate.
      NOTE: Ensure the cell suspension is homogeneous before plating.
4. Prepare and add antibody
   1. Prepare 200 µL of 4× antibody solution (100 µg/mL) in NK-cell culture medium for each test antibody and matched isotype control.
      NOTE: This step is modular and can be adapted to any antibody that mediates ADCC against the chosen target cell line. Antibody stock concentrations and formulations may vary; adjust dilutions accordingly. Preservatives such as azide may affect target-cell survival and NK-cell functionality. While a range of antibody concentrations can be tested, a final concentration of 25 µg/mL is used here.
   2. Transfer antibody solutions to a deep-well plate.
   3. Add 50 µL/well of antibody solution to target cells and mix thoroughly by pipetting ≥5 times.
   4. Add 50 µL/well of NK-cell culture medium to no-antibody control wells and mix thoroughly by pipetting ≥5 times.
   5. Incubate for 25–30 min at 37°C with 5% CO₂.
5. Prepare and add NK cells
   1. Count and obtain 2 × 10⁵ NK cells using a dye exclusion–based cell counting method (e.g., trypan blue) or an equivalent validated method. The number is sufficient for three NK cell–containing conditions plated in triplicate, with excess to account for pipetting loss (see Supplemental Figure 2). Proceed if viability is ≥75%, as lower viability may impact assay performance.
   2. Resuspend cells to 2.5 × 10⁵ cells/mL in 800 µL of primary human NK-cell culture medium (final concentration = 1.25 × 10⁴ cells/50 µL).
   3. Add 50 µL/well of NK cells to the corresponding wells and mix thoroughly by pipetting ≥5 times.
      NOTE: Ensure the cell suspension is homogeneous before plating.
   4. Add 50 µL/well of culture medium to control wells and mix thoroughly by pipetting ≥5 times.
6. Set up final co-culture
   1. Carefully inspect the round-bottom 96-well plate to confirm that all wells contain 200 µL with uniform meniscus levels and no bubbles.
   2. Centrifuge the plate (2 min, 100 × g, acceleration = low [e.g., 2]; deceleration [brake] = low [e.g., 2]).
      NOTE: This soft spin is important to rapidly settle cells and increase cell–cell contact. We use a swinging bucket rotor–compatible benchtop Eppendorf Centrifuge 5810R with the acceleration and deceleration settings set to 2 (out of 9).
   3. Incubate for 16–18 h (overnight) at 37°C with 5% CO₂.
      NOTE: This incubation duration is optimized for cytotoxicity readout rather than degranulation, which is best assessed by measuring CD107a expression at earlier time points (2–6 h). As such, peak degranulation responses may be underestimated.
      NOTE: Do not extend incubation beyond 18 h, as prolonged co-culture may result in excessive target-cell death in control wells and reduced dynamic range between experimental conditions.
7. Perform flow cytometry
   1. Centrifuge (5 min, 300 × g, acceleration = maximum; deceleration [brake] = maximum).
   2. Discard the supernatant.
   3. Wash all wells with 200 µL/well of FACS buffer (2% FBS, 2 mM EDTA in PBS).
   4. Prepare a surface antibody master mix in FACS buffer by adding CD56-BV421 at 1:50 dilution, CD16-BV785 at 1:50 dilution, and CD107a-PE-Cy7 at 1:25 dilution.
      NOTE: Calculate total master mix volume based on the experimental plate design (number of wells × 50 µL/well + 20% excess). Antibody clones may vary in staining intensity and epitope accessibility. Clones used in the representative data are listed in the Table of Materials, and alternative clones may require titration. Additional antibodies may be included for NK-cell subset phenotyping (e.g., KIRs, NKG2A/C/D, and CD57) and activation markers (e.g., CD69 and HLA-DR).
   5. Resuspend cells in 50 µL/well of antibody master mix and mix thoroughly by pipetting ≥5 times.
   6. Incubate for 15 min at 4°C with the plate wrapped in aluminum foil or placed inside an opaque container to protect from light.
   7. Add 150 µL/well of FACS buffer.
   8. Centrifuge (5 min, 300 × g, acceleration = maximum; deceleration [brake] = maximum).
   9. Discard the supernatant.
   10. Resuspend cells in 100 µL/well of 5 µM 4′,6-diamidino-2-phenylindole (DAPI) in FACS buffer and proceed immediately to flow cytometry acquisition. No pre-acquisition incubation is required.
   11. Acquire data by flow cytometry and record fluorescence channels for CFSE, BV421, BV785, PE-Cy7, DAPI, and any additional fluorophores.
       NOTE: Acquisition settings (e.g., detector voltages, compensation, and gating) may require optimization. Use unstained and single-stained controls for compensation and gating setup, with compensation beads or representative cells as appropriate. Fluorescence-minus-one controls should be considered for markers with continuous or dim expression (e.g., CD107a). Maintain consistent settings and gating strategy, including live/dead gating and identification of target and effector populations, across experiments.
       NOTE: Both Step 1 and Step 2 can be adapted for adherent target cells. Seed target cells (with their respective culture media) in a flat-bottom 96-well plate one day prior to co-culture at a density that achieves approximately 70% confluency on the day of the assay. After confirming adherence, discard the target-cell media and replace with 100 µL of fresh NK-cell media, then add effector cells to initiate co-culture. Due to differences in cell size and morphology, IncuCyte Cell-by-Cell analysis parameters should be optimized for each adherent cell line.
       NOTE: For flow cytometry–based assays with adherent cells, following co-culture, centrifuge the plate (300 × g, 5 min) and discard the supernatant. Add 100 µL/well of TrypLE and incubate at 37°C with 5% CO₂ for approximately 5–15 min (depending on cell line) to detach cells. Add 100 µL/well of FACS buffer, mix by pipetting up and down in a circular motion, and transfer all cells to a round-bottom 96-well plate. Inspect wells microscopically to confirm complete detachment before proceeding with staining (Step 2, Sub-step 7). Optimize TrypLE incubation time as needed to achieve a single-cell suspension.

Results

To quantify NK-cell cytotoxicity in real time, NK-92 effector cells engineered to express anti-BCMA CARs were co-cultured with BCMA⁺ Ramos target cells in a black, clear flat-bottom 96-well plate and monitored using live-cell imaging (Figure 1A and Supplemental Figure 1A). Effector cells were added at E:T ratios of 0:1 (target only), 0.25:1, 0.5:1, 1:1, and 2:1, with each condition performed in triplicate. Three different CAR NK-92 constructs were evaluated: CAR1 with no co-stimulatory domain, CAR2 with a 4-1BB (CD137) co-stimulatory domain, and CAR3 with a DAP10 co-stimulatory domain, along with control NK-92 cells expressing a ZsGreen-only vector. All effector cells were generated via lentiviral transduction, as previously described⁴ using ZsGreen-P2A-CAR transgene cassettes. Ramos target cells were similarly transduced with an emiRFP670-encoding lentiviral vector. Target-cell growth and survival were tracked every hour over a 24-h period using the live-cell imaging instrument, measuring green fluorescence for ZsGreen⁺ NK-92 cells, orange fluorescence for SYTOX Orange⁺ dead cells, and NIR fluorescence for emiRFP670⁺ target cells (Figure 1A and Supplemental Figure 1B and 1C).

Microscopic images were acquired at defined time intervals to visualize effector cells, live target cells, and dead cells during co-culture (Figure 1B). Quantification of target-cell growth and survival demonstrated that CAR2 and CAR3 NK-92 cells exhibited faster killing kinetics than CAR1 NK-92 cells, particularly at earlier time points at an E:T ratio of 1:1. Target-cell killing increased with higher E:T ratios, including in control NK-92 cells, which demonstrated natural cytotoxicity (Figure 1C). These measurements highlight the ability of live-cell imaging to capture dynamic changes in cytotoxic activity. Target-cell survival was also analyzed as a function of E:T ratio at defined time points to compare CAR NK-92 potency. CAR2 and CAR3 NK-92 cells showed greater potency at lower E:T ratios at later time points (e.g., E:T = 0.25:1 at 24 h) and at intermediate E:T ratios at earlier time points (e.g., E:T = 1:1 at 4 h) (Figure 1D). No significant differences in killing were observed between CAR2 and CAR3 NK-92 effector cells across analyses.

A heatmap integrating E:T ratio, co-incubation time, and target-cell survival provides a comprehensive visualization of natural and CAR-mediated cytotoxicity across all effector populations and enables quantitative assessment. Three parameters were defined to quantify potency and kinetics: KR₅₀, the killer-to-target ratio that achieves 50% killing (at 12 h); KT₅₀, the co-incubation time required to achieve 50% killing (at an E:T ratio of 1:1); and IKI₅₀, the integrated killer index representing the area of the heatmap below 50% survival, defined by the 50% survival isocline (Figure 1E). These metrics enable comparison of CAR NK-92 cell potency (KR₅₀), killing kinetics (KT₅₀), and integrated cytotoxic performance (IKI₅₀) (Figure 1F). This analysis demonstrates how the assay can quantitatively compare engineered NK-cell platforms across multiple parameters within a single experiment.

Flow cytometry with cell counting was used to measure ADCC by co-culturing fluorescently labeled target cells with primary human NK cells in the presence or absence of therapeutic antibodies and quantifying the residual live target-cell population after 16–18 h of incubation (Figure 2A). Flow cytometers such as Cytek Aurora, BD Symphony, or BD Fortessa with a high-throughput sampler can be used. In the representative setup (Supplemental Figure 2), CD38⁺ patient-derived lymphoma target cells were plated in a round-bottom 96-well plate, and daratumumab (or isotype antibody) was added to appropriate wells. NK cells were then co-cultured at an E:T ratio of 0.5:1, a ratio optimized for these target cells to minimize natural killing while allowing detection of ADCC. After incubation, flow cytometry was performed to resolve cell populations by forward and side scatter, exclude non-viable events using DAPI, and separate CFSE⁺ target cells from CD56⁺CD16⁺ NK cells for parallel analysis (Figure 2B). This design enables simultaneous cytotoxicity measurement and phenotypic characterization.

Figure 2. Flow cytometry–based antibody-dependent cellular cytotoxicity (ADCC) assay enables simultaneous quantification of target-cell survival and NK-cell activation. (A) Schematic overview of the ADCC assay. CFSE-labeled CD38⁺ lymphoma target cells were plated at 2.5 × 10⁴ cells per well and cultured under five conditions: target cells alone; target cells with daratumumab (25 µg/mL); target cells with NK cells (E:T = 0.5:1); target cells with NK cells and daratumumab; and target cells with NK cells and isotype control antibody (25 µg/mL). (B) Representative flow cytometry gating strategy. Events were first gated by forward scatter and side scatter, followed by exclusion of non-viable cells using DAPI. Target cells were identified as CFSE⁺ events, and NK cells were identified as CD56⁺ and CD16⁺ events. NK-cell degranulation was assessed by CD107a expression. (C) Quantification of target-cell survival, expressed as a percentage relative to the target-only control condition (set to 100%). (D) Quantification of NK-cell degranulation, shown as the percentage of CD107a⁺ cells within the NK-cell population. Bars represent mean ± standard deviation, and overlaid points represent individual replicate wells (n = 3). Statistical analysis was performed using one-way analysis of variance with Tukey’s post hoc multiple-comparison test. Statistical significance is indicated as follows: *p < 0.05, **p < 0.01, ***p < 0.001. Please click here to view a larger version of this figure.

The primary endpoint is target-cell survival, calculated within the live CFSE⁺ gate and normalized to the target-only condition, which is set to 100% survival. This normalization controls for spontaneous target-cell loss and enables direct comparison across conditions. In representative data (Figure 2C), the addition of daratumumab alone did not reduce target-cell survival, confirming that the antibody has no direct effect on target-cell viability in the absence of effector cells. Co-culture with NK cells alone reduced target-cell survival to approximately 70%, indicating baseline natural cytotoxicity. The addition of daratumumab further reduced survival to approximately 37%, consistent with enhanced antibody-mediated killing, whereas the isotype control condition showed survival of approximately 62%, comparable to the NK-only condition. Successful experiments show clear separation between antibody-treated and control conditions with high reproducibility across replicates.

Secondary phenotyping of NK cells provides complementary confirmation of the link between target-cell killing and effector activation. NK-cell degranulation, measured as the percentage of CD107a⁺ NK cells, was lowest in NK cell–target cell co-culture without antibody and highest in the daratumumab-treated condition (approximately 31%), with the isotype control showing intermediate degranulation (approximately 19%), supporting antibody-mediated activation (Figure 2D). These results demonstrate that the assay can distinguish target-cell killing from effector activation. Although degranulation and survival trends are generally aligned, they do not fully overlap, indicating that they represent related but distinct biological processes.

Suboptimal experiments typically show reduced separation between conditions, variability in cell counts across wells, high target-cell death in control wells, weak CFSE staining, or elevated background degranulation. These outcomes may result from poor target-cell viability, inconsistent NK-cell recovery, inadequate mixing, or variability in staining and acquisition. Because the assay independently identifies target and effector compartments, it can be extended to include additional phenotyping of target cells (e.g., antigen density or ligand expression) and broader NK-cell characterization. Inclusion of an “NK only” condition can further quantify baseline degranulation and NK-cell viability. This multiparameter design is a key advantage, enabling functional cytotoxicity to be interpreted alongside mechanistic immune phenotypes.

Supplemental Figure 1. Representative plate layout and analysis workflow for live-cell imaging cytotoxicity assay. (A) Representative 96-well plate layout for the CAR NK-cell killing assay. Four effector cell conditions are arranged in quadrants, each tested across E:T ratios of 0:1 (target-only control), 0.25:1, 0.5:1, 1:1, and 2:1, with each condition performed in triplicate (total of 60 assay wells). (B) Representative configuration of non-adherent Cell-by-Cell analysis in IncuCyte 2021C software, illustrating spectral unmixing, segmentation, and masking parameters used to identify individual effector and target cells while excluding debris and aggregates. (C) Representative Cell-by-Cell classification gating strategy using NIR versus SYTOX Orange fluorescence to define the live target-cell population. Target-only wells and co-culture wells at an example time point (t = 4 h) are used to confirm separation between live and dead cell populations. Please click here to download this file.

Supplemental Figure 2. Representative plate layout for flow cytometry–based ADCC assay. Representative 96-well plate layout illustrating five experimental conditions plated in triplicate across rows B–D and columns 2–6 (total assay wells = 15): (1) target-only control, (2) target cells with daratumumab, (3) target cells with NK cells, (4) target cells with NK cells and daratumumab, and (5) target cells with NK cells and isotype control antibody. Target cells were plated at 2.5 × 10⁴ cells per well, and NK cells were added at 1.25 × 10⁴ cells per well to achieve an E:T ratio of 0.5:1. Antibodies were used at a final concentration of 25 µg/mL. Surrounding wells were filled with PBS to minimize evaporation. The target-only condition was used to normalize target-cell survival, and the isotype control accounted for antibody formulation effects independent of antigen specificity. Optimization of E:T ratio may be required depending on the susceptibility of the target cell line. Please click here to download this file.

Discussion

These protocols provide two complementary approaches for quantifying NK-cell cytotoxicity in a standardized 96-well plate format. Step 1 uses real-time live-cell imaging to continuously monitor target-cell loss and is well suited for assessing the kinetics of natural and CAR-mediated killing across multiple E:T ratios. Step 2 uses flow cytometry to quantify ADCC after a defined co-incubation period while simultaneously separating target and effector compartments for phenotypic analysis¹⁷. These methods, which are presented as complementary assay platforms that can be used to model distinct biological processes, are also readily interchangeable. Primary human NK cells can be used for live-cell-imaging-based ADCC assays and can also be engineered for CAR-mediated killing in flow cytometry assays. Furthermore, although wild-type NK-92 cells lack CD16 expression¹⁸, introduction of CD16 (e.g., via lentiviral transduction) readily enables NK-92 cells to mediate potent ADCC. As such, both assays can be tailored to individual scientific and translational needs.

There are several steps that are particularly important for reproducibility in both assays. Small changes in effector-cell number and phenotype can drastically alter target-cell survival. Accurate cell counting and thorough resuspension during plating are essential to minimize deviations in cell counts and incomplete mixing. Similarly, cell preparation must be kept consistent between experiments, including viability, recovery time, and prior cytokine exposure, to ensure uniform effector functionality. Plate geometry is another key determinant of assay performance. In the live-cell imaging workflow, flat-bottom coated plates support stable image acquisition and consistent field analysis, whereas in the flow cytometry workflow, round-bottom plates promote cell–cell contact and facilitate recovery of non-adherent co-cultures at the endpoint. In addition, the quality of fluorescent labeling must be confirmed before the assay, as poor target-cell discrimination by imaging or weak CFSE separation by flow cytometry will compromise quantification and interpretation.

These two workflows present distinct technical considerations. In the live-cell imaging assay, common causes of poor data quality include uneven cell plating, dense cell clustering or clumping, and suboptimal image-analysis thresholds, all of which are readily correctable. These issues may distort cell counts and result in inconsistent tracking over time. In the flow cytometry–based assay, suboptimal outcomes are more often caused by high background target-cell death, weak separation of target and effector populations, poor antibody staining, or inconsistent pipetting between replicate wells. Because these assays are designed for comparative analysis, inclusion of appropriate internal controls is essential for interpretation, including target-only controls for normalization and matched negative controls for antibody- or CAR-specific conditions.

Several shared and assay-specific limitations must be considered with these methods. The live-cell imaging assay provides temporal resolution but depends on accurate fluorescent labeling and image-analysis parameters, both of which may be influenced by cell clustering, spectral overlap, or changes in morphology during co-culture¹⁵. In some cases, particularly when assays involve primary effector cells, achieving adequate fluorescent labeling without altering killing dynamics is difficult or not feasible. The flow cytometry–based assay provides higher phenotypic resolution at a defined endpoint, but it does not independently distinguish early from late killing unless multiple matched time points are collected, which can be costly and time-consuming¹⁷. In addition, endpoint measurements may reflect not only target-cell death but also differences in target-cell proliferation, effector attrition, or recovery during sample handling. Mechanistic readouts such as CD107a are therefore best interpreted as complementary indicators of effector-cell activation rather than direct substitutes for target-cell elimination¹⁹. More broadly, neither platform fully reproduces the spatial architecture, stromal interactions, cytokine gradients, or trafficking constraints that shape NK-cell function in vivo.

Several biological factors must also be considered when interpreting NK-cell killing assays. Donor-to-donor variability is intrinsic to experiments with primary human NK cells, which differ in cytotoxic potential and subset composition due to donor age, sex, infectious disease history, and receptor polymorphisms²⁰. Use of NK-92 cells greatly reduces variability, but these cells can still exhibit source- and passage-related changes¹⁸. Our protocol mitigates technical or experimental differences through standardized parameters and same-plate comparisons to more accurately assess these biological factors. Where feasible, we recommend using defined NK-92 passage ranges with fixed viability thresholds. For primary NK cells, assays should use NK cells from multiple donors for each condition, with statistical analyses accounting for donor-to-donor variability (e.g., paired or mixed-effects models), given that relative differences between conditions are generally preserved across donors. Furthermore, additional NK-cell markers (e.g., KIRs, NKG2A/C/D, and CD57) can be incorporated into antibody panels used for flow cytometry to compare subset frequencies across donors and assess subset-specific NK-cell activation (i.e., CD107a and CD69 expression), further facilitating interpretation of results. Finally, for all NK-cell killing assays, certain additional scenarios must be considered. Expression of targeted antigens on NK cells, either by endogenous expression or by acquisition of antigens from target cells via trogocytosis, can result in NK-cell fratricide and confound results. For example, daratumumab can target CD38 expressed on NK cells and induce fratricide²¹, limiting the magnitude of ADCC against the intended targets. Fortunately, live-cell imaging and flow cytometry assays allow for quantification of NK cells to assess fratricide.

Despite these limitations, the combined use of live-cell imaging and flow cytometry provides a practical framework for reproducible NK-cell functional testing across a range of basic and preclinical applications. The imaging workflow is advantageous when temporal behavior is important, such as distinguishing rapid from delayed cytotoxic responses or comparing killing kinetics across CAR designs²². The flow cytometry–based workflow is advantageous when functional readouts must be interpreted alongside effector- and target-cell phenotypes within the same sample. Together, these approaches enable quantitative assessment of natural killing, CAR-mediated cytotoxicity, and ADCC using widely accessible experimental platforms. By establishing a standardized and scalable framework, this system facilitates rigorous benchmarking and mechanistic interrogation across NK cell–based modalities, accelerating the development and clinical translation of next-generation immunotherapies.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
Advanced RPMI-1640 medium	Thermo Fisher Scientific	12633020	Basal medium for NK-92 and primary human NK-cell culture media (see protocol for complete formulation)
Anti-human CD107a antibody, PE-Cy7 conjugated (clone H4A3)	BioLegend	328618	Degranulation marker for flow cytometry
Anti-human CD16 antibody, BV785 conjugated (clone 3G8)	BioLegend	302046	Flow cytometry
Anti-human CD56 antibody, BV421 conjugated (clone HCD56)	BioLegend	318328	Flow cytometry
Benchtop centrifuge with swinging bucket rotor compatible with 15-mL tubes and 96-well plates (Eppendorf Centrifuge 5810R)	Eppendorf	5811000015	Used for cell pelleting and plate centrifugation (100–300 × g)
Carboxyfluorescein succinimidyl ester (CFSE)	Thermo Fisher Scientific	C34554	Cell proliferation dye for target-cell labeling
Cytek Aurora spectral flow cytometer (or equivalent)	Cytek Biosciences	N7-00003	Used for flow cytometry acquisition and analysis
DAPI (4′,6-diamidino-2-phenylindole)	BioLegend	422801	Nuclear stain for live/dead discrimination
Deep-well 96-well plate	USA Scientific	1896-2110	Used for antibody preparation
Dimethyl sulfoxide (DMSO)	Fisher Scientific	BP231-100	Cryopreservation reagent
Fetal bovine serum (FBS)	Avantor	76419-584	Basal media supplementation; used in FACS buffer (2% FBS)
General Purpose Water Bath, 10 L (37 °C)	PolyScience	WBE10	For cell thawing
GlutaMAX (L-alanyl-L-glutamine supplement)	Thermo Fisher Scientific	35050061	Stable glutamine source
Human AB serum	Sigma-Aldrich	H4522-100ML	Supplement for primary human NK-cell culture media
IncuCyte SX5 live-cell imaging system	Sartorius	4816	Real-time live-cell imaging and Cell-by-Cell analysis
LUNA-FL dual fluorescence cell counter (or equivalent)	Logos Biosystems	L20001	Automated cell counting
Parafilm M (4 in × 250 ft roll)	Genesee	16-101	Plate sealing and storage
Phosphate-buffered saline (PBS), sterile (pH 7.2–7.4)	Corning	21-020-CV	Used for washes and FACS buffer preparation
Poly-D-lysine (0.1 mg/mL)	Thermo Fisher Scientific	A3890401	Plate coating reagent
Primocin	InvivoGen	ant-pm-2	Antibiotic
Recombinant human interleukin-2 (IL-2)	R&D Systems	202-IL-500	Cytokine
RosetteSep Human NK-cell enrichment kit	STEMCELL Technologies	15065	Human NK-cell enrichment cocktail
SYTOX Orange dead cell stain	Thermo Fisher Scientific	S34861	Viability dye for live-cell imaging
Trypan Blue 0.4%	Gibco	15250-061	Viability dye for cell counting
TrypLE Express Enzyme (1X), no phenol red	Thermo Fisher Scientific	12604039	For adherent cell detachment
UltraPure EDTA (0.5 M, pH 8.0)	Thermo Fisher Scientific	15575020	Used for FACS buffer preparation
15-mL conical-bottom centrifuge tubes	Corning	352097	Sterile
24-well tissue culture plate (tissue culture–treated)	Corning	3524	Used for NK-cell recovery
96-well clear round-bottom plate (tissue culture–treated)	Corning	3799	Used for co-culture and flow cytometry assays
96-well flat clear-bottom black polystyrene plate (tissue culture–treated)	Corning	3603	Compatible with live-cell imaging systems