NK cells switch from granzyme B to death receptor-mediated cytotoxicity during serial killing

Abstract

NK cells eliminate virus-infected and tumor cells by releasing cytotoxic granules containing granzyme B (GrzB) or by engaging death receptors that initiate caspase cascades. The orchestrated interplay between both cell death pathways remains poorly defined. Here we simultaneously measure the activities of GrzB and caspase-8 in tumor cells upon contact with human NK cells. We observed that NK cells switch from inducing a fast GrzB-mediated cell death in their first killing events to a slow death receptor-mediated killing during subsequent tumor cell encounters. Target cell contact reduced intracellular GrzB and perforin and increased surface-CD95L in NK cells over time, showing how the switch in cytotoxicity pathways is controlled. Without perforin, NK cells were unable to perform GrzB-mediated serial killing and only killed once via death receptors. In contrast, the absence of CD95 on tumor targets did not impair GrzB-mediated serial killing. This demonstrates that GrzB and death receptor-mediated cytotoxicity are differentially regulated during NK cell serial killing.

Graphical abstract

Figure 1.

Different killing kinetics of target cells showing GrzB versus Casp8 activity. HeLa cells were cotransfected with NES-RIEADS-mCherry (for GrzB), NES-ELQTD-GFP (for Casp8), and CD48. Confocal time-lapse microscopy started immediately after NK92-C1 cell exposure (E/T = 3). From two independent experiments, 52 cells were classified according to strength of RIEADS- and ELQTD-cleavage. Three dying cells showed no (<10%) reporter cleavage and were excluded. (A) Responses of example single cells (upper row, real time) or of cell subpopulations (mean ± SD, time point 0 set to time of cell death) with more Casp8 than GrzB-reporter cleavage (left), equal reporter cleavage (center), or more GrzB- compared with Casp8-reporter cleavage (right), from one experiment (time resolution, 4 min). (B) Time of target cell death (Cell death), time point of NK cell contact with target cell (Cell contact), and the time span (from NK cell contact to target cell death) for the three cell subpopulations of both experiments is shown. ****, P ≤ 0.0001; *, P ≤ 0.05 (unpaired, two-tailed t test with Welch’s correction).

Figure 2.

Cell death morphology, GrzB, and Casp8 activity in target cells killed upon co-culture with primary NK cells. Human primary NK cells were isolated from blood of three healthy donors. For each donor NK cell population, two experiments at different NK cell doses (E/T = 1 and E/T = 3) were performed. HeLa-CD48 target cells expressing the NES-RIEADS-mCherry (GrzB) and NES-ELQTD-GFP (Casp8) reporters were imaged over time by confocal microscopy. (A) Images show alive cells (top) and three different cell death morphologies that were visually distinguished (with reporter fluorescence, center; or transmission light, bottom): an apoptosis-like phenotype with cell blebbing, a non-apoptosis-like phenotype with lack of cell blebbing, and an intermediate phenotype with cell shrinkage but no or few cell blebbing. Scale bars, 10 µm. (B) Pie charts show the frequency of different cell death morphology for cells induced with the three donor NK cells (E/T = 1, in total 94 cells; E/T = 3, in total 119 cells). Bar graphs show this distribution as relative frequencies for cells grouped for the balance of GrzB- and Casp8-reporter cleavage and for E/T ratio (mean ± SEM, n = 3). (C–E) Plots of single cell data of the time of target death (C), the time point of NK cell contact with target cell (D), and the time between NK cell contact and target cell death (E). The median is indicated for pooled data from three independent experiments using NK cells from three donors. (F) Pie charts show the proportion of reporter cleavage at the two different NK cell doses for the NK cells of each donor. ****, P ≤ 0.0001; **, P ≤ 0.01; *, P ≤ 0.05 (unpaired, two-tailed t test with Welch’s correction).

Figure 3.

NK cells differentially induce GrzB- or Casp8-mediated target cell death during serial killing. Serial killing activity of primary human NK cells was evaluated against HeLa-CD48 cells stably expressing NES-ELQTD-GFP-T2A-NES-VGPD-mCherry. (A–G) Three independent experiments using IL-2–activated NK cells from three donors (A–E) and three independent experiments using resting, freshly isolated NK cells from three different donors (F and G) were conducted. A total of 347 activated and 63 resting NK cells was tracked over 15–17 h, and each of their killing events was characterized in terms of GrzB and Casp8 activity as well as morphology of cell death. (A) Example of five consecutive killing events by a serial killer NK cell (NK cell visualized in brightfield, arrowheads, lower row) with corresponding enzymatic activity (red and green nuclear signal indicates GrzB and Casp8 activity, respectively). Images have been background-subtracted and adjusted for brightness and contrast using ImageJ. See Video 3 for details. Scale bar, 15 µm. (B and C) Diagrams display the activity of GrzB and Casp8 (B) and the morphology of target cell death (C) in single killing events. Each row displays the killing sequence of one individual NK cell. Cells for which fluorescent signal could not be reliably measured are denoted by “unstained.” Some target cells showed strong Casp8 reporter activity without displaying significant morphological signs of cell death during the observation period (No death). (D–G) Target cell deaths were categorized according to their respective number in the corresponding NK cell killing sequence. Histograms show the proportion of target cell deaths displaying activity of GrzB, Casp8, or absence of reporter cleavage (D and F) and the proportion of target cell deaths with an apoptosis-like or non-apoptosis-like phenotype (E and G). Cells showing clear enzymatic activity following an NK cell interaction, yet surviving the entire time span of the assay, are denoted by “No death.” The non-apoptosis-like phenotype is classified as described in Fig. 2, and the apoptosis-like fraction also contains cells with the “intermediate” phenotype.

Figure 4.

Timing of killing events. (A) Each killing event shown in Fig. 3 B was analyzed according to time of first contact, NK cell commitment, reporter cleavage, and target cell death. (B–E) The time span between NK cell contact and NK cell commitment (B), reporter cleavage (C), and target cell death (D and E) was plotted for individual killing events. Data were grouped according to the respective number in the corresponding NK cell killing sequence (E) or according to the reporter activity (B–D). Black lines indicate median values.

Figure 5.

Changes in granule content and surface CD95L on NK cells during serial killing. Primary human NK cells were incubated at an E/T ratio of 2 with HeLa-CD48 cells in the presence of an anti-CD107a antibody. After the indicated times, NK cells were harvested and stained for CD45, CD95L, and TRAIL followed by intracellular staining for perforin and GrzB. Unstained NK cells (control) and stained NK cells incubated for 4 h without target cells (NK cells) were used as controls. (A and C) Histograms for the staining of CD95L, TRAIL, perforin, and GrzB on CD107a⁺ (A) and CD107a⁻ (C) NK cells (gated on CD45⁺). (B and D) Summary of three independent experiments showing the mean fluorescence staining intensity (MFI) for CD95L or the relative fluorescence staining intensity (RFI, normalized to the NK cell only control) for TRAIL, perforin, and GrzB at the different time points for CD107a⁺ (B) and CD107a⁻ (D) NK cells. Values shown are the mean ± SD; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001 (one-way ANOVA).

Figure 6.

Perforin is essential for NK cell serial killing. (A) Intracellular FACS staining of perforin in human NK cells before (control) and after CRISPR/Cas9-mediated KO. (B) Example for intracellular perforin staining in KO and WT clones. (C) Perforin WT (C) and KO (D) NK cell clones (NK cells from two independent clones each) were evaluated against HeLa-CD48 cells stably expressing NES-ELQTD-GFP-T2A-NES-VGPD-mCherry. NK cells were tracked over 15–17 h, and each of their killing events was characterized in terms of GrzB and Casp8 activity as well as the kinetics of reporter cleavage and cell death.

Figure 7.

Serial killing of CD95KO targets. Serial killing activity of primary human NK cells was evaluated against CD95KO HeLa-CD48-NES-ELQTD-GFP-T2A-NES-VGPD-mCherry targets. Three independent experiments using IL-2–activated NK cells from three donors were conducted. A total of 60 NK cells were tracked over 15–17 h, and each of their killing events was characterized in terms of GrzB and Casp8 activity. (A) Diagram displaying the activity of GrzB and Casp8 in single killing events. Each row shows the killing sequence of one individual NK cell. (B) Relative distribution of reporter activity of all killing events for control and CD95KO HeLa targets. (C) Relative distribution of NK cells killing one to six different control or CD95KO HeLa targets.