A real‐time digital bio‐imaging system to quantify cellular cytotoxicity as an alternative to the standard chromium‐51 release assay

Summary

Reliable measurement of cellular cytotoxicity is essential for the characterization of immune responses and for the monitoring of antibody treatment efficacy. Until now, the standard ⁵¹Cr‐release assay has remained the sole sensitive assay that measures cellular cytotoxicity. Alternative non‐radioactive assays have been developed but they do not provide accurate measurement of target cell cytotoxicity. The cost and hazard of handling radioactivity are strong incentives to find alternative solutions to ⁵¹Cr. We took advantage of the recent development of cell‐imaging multimode readers to develop a novel non‐radioactive and real‐time cytotoxic assay that demonstrates good reproducibility and sensitivity. The extent of target‐cell cytotoxicity is monitored over time by imaging and quantifying live fluorescent target cells in 96‐well plates. We have developed classical natural killer cell assays in the presence or absence of blocking antibodies and antibody‐dependent cell‐mediated cytotoxicity. We show that in these assays, cell killing occurs within the first 2 hr with half maximum killing reached after 30 min. This technology has numerous applications such as natural killer and T‐cell cytotoxicity assays and can be extended to cell survival and apoptosis measurement assays.

Abbreviations

ADCC

antibody‐dependent cell‐mediated cytotoxicity

CTL

cytotoxic T lymphocytes

LLT1

lectin‐like transcript 1

NK

natural killer

Introduction

Assessment of cell‐mediated cytotoxicity in research and in the clinic is commonly performed using the reference standard chromium 51 (⁵¹Cr) ‐release assay. This assay has so far remained the most sensitive cytotoxic assay that measures natural killer (NK) and T‐cell‐mediated cytotoxicity.1 It is also widely used to monitor the mechanism of action of therapeutic antibodies. However, the issue of the cost and hazard of handling radioactive isotopes is encouraging laboratories to use alternative assays. Over the last two decades, a number of non‐radioactive assays have been developed but none of them can substitute for the ⁵¹Cr‐release assay. A first group of assays is based on the measurement of the release of the enzymes lactate dehydrogenase,2 glyceraldehyde 3‐phosphate dehydrogenase,3 or specific serine dead cell proteases by dead targets. However, death of effector cells occurs during the assay and as they also release the enzymes, this leads to false measurements of target cell killing. Because non‐specific apoptosis of effector cells cannot be dissociated from the death of target cells, these assays also require the use of targets and effector cells displaying > 95% viability, adding further constraints. A second group of assays is based on the loading of target cells with non‐radioactive reagents such calcein‐AM4 or BAPTA, a fluorescence‐enhancing ligand of Europium,5 and measurement of the fluorescence released in the supernatant upon killing of targets. The major hurdle of the Europium assay is that most targets are refractory to loading with BAPTA and the assay lacks reproducibility. By contrast, calcein‐AM is highly lipophilic and therefore loaded in all target cells, but the measurement of fluorescence has proved to be much less sensitive compared with measurement of ⁵¹Cr, mainly due to the autofluorescence of the culture media.6, 7 Also, we could not obtain a correlation of the fluorescence intensity with the number of calcein‐loaded target cells using the bottom‐reading mode of a conventional plate‐reader as recently described.8 Finally, flow cytometry assays have also been developed that either detect degranulation of NK and T cells based on the detection of cell surface LAMP‐1 (CD107a),9 or that detect cytotoxicity of targets loaded with specific fluorescent cell trackers and markers of cell death.10 The degranulation assay is widely used as a readout of NK or T‐cell activity and has the advantage of discriminating degranulation at a single‐cell level. However, degranulation events do not systematically correlate with target cell lysis, precluding its use as a substitute of ⁵¹Cr‐release assay. Flow cytometric cytotoxicity assays can detect cell killing at a single cell level but they can only record cells undergoing apoptosis at a single time‐point and do not recapitulate target cell killing throughout the assay. The other main constraint is the length of time needed to register the individual samples, precluding the simultaneous measurement of all samples. Quantification to correct for volume and number of cells is also not straightforward, despite the possible use of calibration beads.

Based on the lack of reliable non‐radioactive cell‐mediated cytotoxic assays, and because of the recent technical developments available with cell‐imaging multimode readers, we sought to take advantage of this technology to develop a real‐time digital bio‐imaging cytotoxic assay.

Methods

Cells

Cells were maintained in complete RPMI‐1640 medium (Gibco, Waltham, MA) (C1R) and in Iscove's modified Dulbecco's medium (Gibco) (K562, SUDHL4, OciLy19), supplemented with 10% fetal bovine serum (GE Healthcare, Chalfont St Giles, UK), penicillin (100 U/ml) and streptomycin (100 μg/ml) (Gibco). C1R‐LLT1 cells were cultured in the presence of 1 mg/ml G418 (Euromedex, Strasbourg, France).11

Human peripheral blood mononuclear cells were separated by Ficoll‐Paque Plus density gradient centrifugation (GE Healthcare) from blood purchased from the Etablissement Français du Sang (Marseille, France). The NK cells were isolated with the NK Cell Isolation Kit (Miltenyi Biotec, Bergisch Gladbach, Germany). Purity was typically > 98%. The NK cells were either used immediately in the cytotoxic and antibody‐dependent cell‐mediated cytotoxicity (ADCC) assays or were subsequently co‐cultured with irradiated allogeneic peripheral blood mononuclear cells and B‐Epstein–Barr virus‐transformed feeder cells in X‐VIVO 15 medium (Cambrex, East Rutherford, NJ) supplemented with 10% fetal bovine serum and 500 units/ml interleukin‐2 (Novartis Pharma, Rueil‐Malmaison, France) to generate NK cell lines.

Monoclonal antibodies

In cytotoxic assays, isotype mIgG1 (Sigma‐Aldrich, St Louis, MO), blocking anti‐MHC class I (clone DX17) kindly provided by L. Lanier, USA and blocking anti‐LLT1 (clone 4F68) generated in house and described previously12, 13 were used at 10 μg/ml. In ADCC assays, isotype hIgG1 (Sigma‐Aldrich) and anti‐CD20 (Rituximab, Genentech, South San Francisco, CA) were used at the indicated concentrations.

Real‐time digital bio‐imaging cytotoxic assay

Target cells were labelled with 0·5 μm calcein‐AM (Molecular Probes, Eugene, OR) for 15 min at room temperature and washed twice before the assay. Then, 2 × 10⁴ calcein‐labelled target cells were incubated in each well of a 96‐well black plate with flat, clear bottom (Falcon), in the presence of an increasing number of effector cells (experimental well) or in the presence of an increasing number of unlabelled target cells (control well). Blocking monoclonal antibodies or antibodies triggering ADCC were incubated with calcein‐labelled targets for 15 min before the addition of effector cells. The plate was spun before being placed in the cell imaging multi‐mode plate reader Cytation™ 5 (Biotek, Winooski, VT). The assay was performed at 37° in an atmosphere supplemented with 5% CO₂. Images of each well were collected every 10–15 min over 4 hr, using a 4× phase‐contrast objective, a GFP filter cube (Excitation 469/35/Emission525/39, mirror 497) and 465 nm LED. Four images per well were stitched to cover most of the well. Cell images were processed using Gen5™ software (Biotek). The percentage of lysis from duplicates or triplicates was calculated as follow: % lysis = {1−[(experimental well at t/experimental well at t₀)/(control well at t/control well at t₀)]} × 100.

Statistical analysis

Statistical significance was determined using unpaired t‐test and Holm–Sidak method with α = 5%. Each row was analysed individually, without assuming a consistent SD. Analyses were carried out using Graphpad Prism (GraphPad, San Diego, CA) software version 6.0.

Results

A real‐time digital bio‐imaging cytotoxic assay provides kinetic analysis of cell‐mediated cytotoxicity.

A non‐radioactive cellular cytotoxicity assay was developed using standard NK cell targets efficiently killed by human NK cells. We labelled target cells with the cell‐permeable calcein‐AM dye, which is hydrolysed by intracellular esterases into calcein, a highly fluorescent compound retained in the cytoplasm. These fluorescent targets were incubated with primary human NK cells at different effector to target (E : T) ratios or without NK cells to correct for spontaneous unspecific target cell death. The NK‐cell‐mediated cytotoxicity was measured every 10–15 min over 4 hr using the cell imaging multi‐mode plate reader ‘Cytation™ 5’ developed by Biotek. This apparatus can provide fast and accurate live‐cell imaging, allowing the quantification of fluorescent cells in each well (Fig. 1 and see Supplementary material, Movie S1). To ensure accurate counting of fluorescent targets, four images per well are stitched and analysed. The assay corrects for non‐specific cell death using the control wells containing target alone and cell counts are normalized to the starting number of live targets at t₀. Importantly, ‘Cytation™ 5’ allows the assay to be performed in optimized cell culture conditions, at 37° and in the presence of 5% CO₂. We evaluated the requirements for a reproducible assay. We determined that a number of 2 × 10⁴ calcein‐labelled cells per well was optimum for reproducible cell counts (95% CI) and found that it was important to maintain a density of cells close to 90% to avoid cell aggregation at the centre of the wells (see Supplementary material, Movie S2). To reach this density, we used E : T ratios between 10 : 1 and 2·5 : 1. When a lower E : T ratio was needed, we increased calcein‐labelled cells up to 5 × 10⁴ per well. Importantly, we also included unlabelled targets in the control wells in which no NK cells were added and at a similar concentration to the effector cells to prevent cell aggregation. Using this assay, we measured over time NK‐cell‐mediated cytotoxicity towards MHC class I negative‐K562 (Fig. 1) and 721·221 cells (see Supplementary material, Fig. S1a) and towards MHC class I low‐C1R cells (see Supplementary material, Fig. S1b). The NK cell killing positively correlated with the number of effector cells used and statistical significance was reached for all E : T ratios at times 60–120 min and not at later time‐points (Fig. 1b). NK cell killing was found to be very rapid. Indeed, the half‐maximum cytotoxicity was reached after 30–40 min of incubation (Fig. 1a and see Supplementary material, Fig. S1). Similarly, the assay was used to assess cytotoxic T cell (CTL) lysis using a CTL line which specifically killed HLA‐A*0301⁺ B‐cell lines loaded with a 9‐mer peptide from influenza nucleoprotein (see Supplementary material, Fig. S1c).14 Significant lysis of the targets loaded with the influenza peptide was detected, demonstrating that the assay is also suitable for CTL. A direct comparison of this assay with the calcein‐release assay6, 7 revealed a higher sensitivity of the real‐time bio‐imaging cytotoxic assay, mostly due to the autofluorescence of the medium and a high spontaneous release ranging from 50% to 60% in calcein release assay (see Supplementary material, Fig. S1d–f).

Figure 1

Time–course natrual killer (NK) cell‐mediated lysis of K562 cells. (a) Kinetics of K562 cell killing, (b) statistical analysis of differential killing between E : T ratios, *P < 0·05, Multiple t‐test with Holm–Sidak correction, (c) visualization of calcein‐labelled K562 cells in the presence and absence of NK cells. The data are presented as mean ± SEM from one of five independent experiments.

Monitoring of the regulation of NK‐cell‐mediated cytotoxicity by inhibitory receptors

The modulation of NK‐cell‐mediated cytotoxicity by inhibitory receptors is a hallmark of NK cell regulation, with classical MHC class I molecules binding to inhibitory killer‐cell immunoglobulin‐like receptors15, 16 and HLA‐E binding to CD94/NKG2A receptors.17 Using the real‐time bio‐imaging cytotoxic assay, we therefore tested NK cell killing of the MHC class I‐expressing lymphoma cell line SUDHL4 (Fig. 2a) and OciLy19 (see Supplementary material, Fig. S2a), in the presence or absence of blocking anti‐MHC class I monoclonal antibody. SUDHL4 and OciLy19 cells were poorly lysed by NK cells but the addition of anti‐MHC class I monoclonal antibody restored significant target cell killing. This effect is rapid and occurs within 30 min. Similar results were obtained with Raji Burkitt cell line (data not shown).

Figure 2

Blocking inhibitory receptors restore natural killer (NK) cell lysis. Time–course NK cell‐mediated lysis of (a) SUDHL4 cells restored by addition of blocking anti‐MHC class I monoclonal antibody, (b) C1R‐LLT1 but not C1R cells restored by addition of blocking anti‐LLT1 monoclonal antibody. The data are presented as mean ± SEM from one of four independent experiments.

Natural killer cell function is also inhibited by MHC class I‐independent signals and among these signals, we previously identified the interaction of LLT1 with inhibitory CD161 receptors expressed by NK cells.11 Using the real‐time bio‐imaging cytotoxic assay, we show that expression of LLT1 in transfected C1R cells conferred significant protection from NK cell killing (see Supplementary material, Fig. S2b), consistent with previous finding using standard ⁵¹Cr‐release assays.11 Consequently, the addition of a blocking anti‐LLT1 antibody restored significant killing of C1R‐LLT1 cells and not of LLT1‐negative C1R cells (Fig. 2b).

Measurement of antibody‐dependent cell‐mediated cytotoxicity

The real‐time bio‐imaging cytotoxic assay was then used to measure ADCC of lymphoma target cells by primary human NK cells. We validated the assay using the therapeutic anti‐CD20 antibody Rituximab and the CD20⁺ MHCI⁺ OciLy19 and SUDHL4 lymphoma cell lines. Similar results were obtained with Raji cells (data not shown). Because of MHC class I expression, the OciLy19 cells were poorly killed by primary human NK cells (Fig. 3a). Addition of saturating concentrations of irrelevant control hIgG1 chimeric monoclonal antibody did not increase killing whereas addition of Rituximab induced significant lysis (P < 0·005), with half‐maximum cytotoxicity reached after 30 min incubation (Fig. 3a). ADCC of SUDHL4 by human NK cell lines increased with increasing concentrations of Rituximab and the calculated EC₅₀ value of 1 ng/ml is consistent with published results18 (Fig. 3b). An EC₅₀ of 5 ng/ml was also obtained with Ocily19 cells (Fig. 3c). Altogether, these results establish the robustness of the assay.

Figure 3

Antibody‐dependent cell‐mediated cytotoxicity (ADCC) of OciLy19 and SUDHL4 cells by primary human natrual killer (NK) cells. (a) Time–course ADCC in the presence of 1 μg/ml Rituximab (RTX), *P < 0·05, Multiple t‐test with Holm–Sidak correction for all time‐points, (b, c) ADCC in the presence of increasing concentrations of RTX. The calculated half maximum effective concentration (EC₅₀) value for RTX is 1 ng/ml for SUDHL4 cells (b) and 5 ng/ml for OciLy19 cells (c). The data are presented as mean ± SEM from one of three independent experiments.

Discussion

We have developed a novel non‐radioactive cell‐mediated cytotoxic assay that is robust, reproducible and sensitive and that can substitute for ⁵¹Cr‐release assay. This assay provides accurate measurement of cell cytotoxicity compared with non‐radioactive assays currently on the market. This is of major interest considering the hurdle of using radioactivity and the need to measure NK and T‐cell functions in basic and clinical immunology.

Importantly, this assay provides a real‐time measurement of cellular cytotoxicity based on accurate quantification of fluorescent target cells using the cell imaging multi‐mode plate reader ‘Cytation™ 5’. The analysis of the dynamics of target cell lysis in standard NK‐cell‐mediated cytotoxicity and ADCC assays revealed fast exponential kinetics of NK cell killing within the first hour and maximum lysis completed by 2 hr. Half of the maximum killing was reached after 30–40 min incubation. Interestingly, we show that the window between 1 and 2 hr provides greater and significant differences in lysis between E : T ratios. Measurement of the kinetics of NK cell killing is therefore a major improvement over current end‐point assays allowing the visualization of differential killing over time that may no longer be detectable after 4 hr.

Besides providing a kinetic analysis, the assay presents numerous advantages over current assays. Indeed, in contrast to most non‐radioactive assays and to the degranulation assay monitoring CD107a by flow cytometry, this real‐time bio‐imaging cytotoxic assay exclusively measures target cell lysis. The time of the assay can be shortened to a 2‐hr assay because we found that maximum lysis is generally reached within the first 2 hr. The procedure is simplified with a short labelling of targets, no requirement to harvest supernatants, and a very low cost of reagents. Although the assay is based on the imaging of fluorescent targets, both adherent and non‐adherent cells can be used. Of note, cells expressing a multidrug‐resistance system that can actively extrude calcein may be more difficult to use but inhibitors of multidrug resistance are currently being tested. Finally, the assay allows the visualization of cell killing at a single‐cell level and we detected the formation of clusters of target cells during cell killing. This is an active process which only occurs upon NK‐cell‐mediated lysis.

Future developments will most likely focus on new cell trackers that can improve sensitivity and reproducibility. This assay is the first that proposes an effective alternative to the standard ⁵¹Cr‐release assay and it has many applications beyond cytotoxicity.

Acknowledgements

VMB designed the study, JF, KT, ES, NHT and VMB performed the experiments, VMB wrote the article. This work was supported by the Centre National de la Recherche Scientifique, Cancéropole PACA, Fondation ARC pour la recherche sur le cancer and Agence Nationale de la Recherche (ANR).

Disclosures

VMB and JF benefited from a reimbursement from Biotek from attending a symposium. KT, ES, NT have no conflict of interest to disclose.