Metabolic Screening of Cytotoxic T-cell Effector Function Reveals the Role of CRAC Channels in Regulating Lethal Hit Delivery

Cytotoxic T lymphocytes (CTL) eliminate target cells, including virally infected and cancer cells, in an antigen-specific and contact-dependent manner. To induce tumor cell death, activated CTLs migrate into the tumor microenvironment (TME) and physically engage with target cells to sample peptide–MHC complexes via their T-cell receptor (TCR; ref. 1). In response to antigenic recognition, CTLs form a lytic immune synapse to release cytotoxic effector molecules toward the target cell (2). Cytotoxic contacts are terminated upon CTL detachment from the target cell followed by interstitial CTL migration and subsequent engagements with neighboring target cells (3, 4). The multistep cycle of CTL migration, adhesion, signaling, and degranulation defines the efficacy of (i) individual one-to-one contacts, (ii) sequential engagements with further target cells (serial killing), as well as (iii) CTL swarming and cooperation to deliver a sequence of sublethal hits that accumulate over time to induce target cell death (additive cytotoxicity; ref. 5). Consequently, interference at any step of the CTL migration and interaction cycle will compromise the overall CTL effector response in the tissue.

Metabolic perturbation of the TME negatively impacts CTL effector responses. After recruitment into tumors, CTLs are confronted with hypoxia and oxidative stress caused by tumor malperfusion and deregulated redox homeostasis (6, 7). Tumor-infiltrating CTLs further encounter elevated lactate levels caused by aerobic glycolysis in cancer cells (8). Metabolic perturbation can compromise several steps of the CTL effector response by compromising CTL entry into the tumor, labilizing CTL–target cell conjugates, impairing target cell recognition, and/or reducing CTL degranulation (9). Hypoxia upregulates the inhibitory immune checkpoint programmed death ligand 1 (PD-L1) on cancer cells (10), which weakens immune synapse stability and functions through binding to programmed death-1 (PD-1) on CTLs (11). Hypoxia further promotes extracellular matrix remodeling (12), which may misroute or exclude CTLs from tumor subregions (3). Similarly, deregulated redox homeostasis reduces CTL infiltration into tumor tissues (13) and impairs target cell recognition (14). Extracellular acidification, for example, by lactate, compromises the cytotoxic capacity of CTLs by reducing the availability of cytolytic effector molecules (15). Thus, through diverse and often complementary mechanisms, metabolic perturbation of the TME can impact the CTL effector cycle.

The effect of microenvironmental perturbation on the CTL effector response is typically studied in two-dimensional (2D) liquid cocultures of target cells with CTLs (16–19). This allows for convenient and low-cost analysis of CTL killing efficacy and immune-suppressive signatures with high throughput. However, due to the lack of a surrounding three-dimensional (3D) extracellular matrix, in which interstitial migration is balanced with stable conjugation to target cells, this approach fails to mimic the organotypic steps of the CTL effector response. Matrix-based spheroid models constitute a more physiologic system for studying the CTL effector response by mimicking the physical constraints and metabolic challenges present in cancer tissues (20). However, these 3D spheroids are optically convoluted and do not allow for live-cell monitoring of individual CTL–target cell interactions at the single-cell level.

We here developed a microscopy-based, 3D interface model suitable for medium-throughput screening for microenvironmental modulators of CTL effector responses against antigenic target cells. This coculture model allowed for quantitative single-cell monitoring of individual CTL contact phases, molecular effector signaling, and target cell death. Microenvironmental pathways were addressed by (i) directly mimicking metabolic features of the TME or (ii) pharmacologically modulating pathways triggered by microenvironmental factors. We identified metabolic and signaling conditions that compromise the CTL effector response and, using live-cell reporter detection of CTL–target cell contacts, delineated the underlying molecular events during individual contact phases.

Mouse melanoma B16F10 cells were kindly provided by Dr. Gosse Adema (year 2007). B16F10 cells expressing the ovalbumin-derived CTL epitope SIINFEKL were obtained by electroporation. Stable H2B-GFP- or H2B-mCherry–expressing B16F10/OVA cells were obtained by lentiviral transduction and blasticidin selection (10 μg/mL; Thermo Fisher Scientific, R2101). B16F10/OVA H2B-mCherry cells were lentivirally transduced to stably express the calcium sensor GCaMP6s followed by blasticidin selection (10 μg/mL). The cells were cultured in RPMI1640 medium (Gibco, 21875–034) supplemented with 10% FCS (Sigma, F7524), 1% sodium pyruvate (Gibco, 13360–039), and 1% penicillin and streptomycin (PAA, P11–010) at 37°C in a humidified 5% CO₂ atmosphere.

C57BL/6 mouse embryonic fibroblast-like cells (MEC1) expressing the costimulatory molecule B7.1 were kindly provided by Dr. Stephen Schoenberger (year 2002; ref. 21). Electroporation was used for stable expression of SIINFEKL, H2B-GFP, and GCaMP6s. The cells were cultured in T-cell medium (TCM), consisting of RPMI1640 (Gibco, 21875–034) supplemented with 10% FCS (Sigma, F7524), 10 mmol/L HEPES (Gibco, 15630–056), 500 mmol/L 2-mercaptopethanol, 1% penicillin and streptomycin (PAA, P11–010), 1% sodium pyruvate (Gibco, 11360–039), and 0.1 mmol/L nonessential amino acids (Gibco, 11140–035).

Unless stated otherwise, cell culture was performed at 37°C in a humidified 5% CO₂ atmosphere. All cell lines were cultured up to 30 passages. Identity of the B16F10 cells was verified by short tandem repeat (STR) DNA profiling (IDEXX BioResearch, last profiled on May 2018). No mammalian interspecies contamination was detected. MEC1 cells were not authenticated. Lack of contamination with Mycoplasma was routinely verified using the MycoAlert Mycoplasma Detection Kit (Lonza, LT07–318, last tested on August 2020).

Animal studies were approved by the Central Authority for Scientific Procedures on Animals (CCD) and performed in accordance with institutional guidelines. Splenocytes were derived from OT-I TCR transgenic mice (Jackson Laboratories, stock number: 003831) or from double-transgenic dsRed/OT-I mice obtained by crossing dsRed.T3 mice (Jackson Laboratories, stock number: 006051) with OT-I TCR transgenic mice (Jackson Laboratories, stock number: 003831). Splenocytes and OT-I T lymphocytes were cultured in TCM. Splenocytes were harvested by mashing the spleen tissue through a 100-μm nylon cell strainer (BD Biosciences, 352360). Erythrocytes were depleted by incubating the splenocytes in ammonium chloride (0.83% NH₄Cl, 0.1% KHCO₃, 0.37% Na₂EDTA; 5 minutes at room temperature). For the expansion of antigen-specific CTLs, splenocytes were cultured in 24-well plates (5 × 10⁵ cells/well) in the presence of SIINFEKL peptide (0.5 μg/mL; Sigma-Aldrich, S7951). After 3 days of culture, the cells were resuspended in a mixture of 50% CTL expansion culture supernatant with 50% fresh TCM and cultured in 24-well plates (4 × 10⁵ cells/well) in the presence of IL2 (IL2, 100 U/mL; ABD Serotec, PMP38). After 24 to 48 hours of culture, CTLs were isolated by Ficoll gradient centrifugation (Axis-Shield, 1114544). Purity typically exceeded 96% of Vα2⁺CD8⁺CD62L^lowCD44^hi cells, as determined by flow cytometry using PerCP-Cy5.5–conjugated rat anti-mouse CD8α (BD Biosciences, 551162), APC-conjugated rat anti-mouse Vα2 TCR (BD Biosciences, 560622), FITC-conjugated rat anti-mouse CD62L (BD Biosciences, 561917), and PE-conjugated rat anti-mouse CD44 (BD Biosciences, 553134).

Buffy coats from healthy volunteers were purchased from the Sanquin blood bank upon written informed consent in accordance with the Medical Ethics Committee of Sanquin Blood Supply and conform to the principles of the Declaration of Helsinki. Peripheral blood mononuclear cells (PBMC) were purified via density-gradient centrifugation using Lymphoprep (Axis-Shield, 1114544), followed by isolation of CD8⁺ T lymphocytes through negative selection against CD4, CD15, CD16, CD19, CD34, CD36, CD56, CD123, TCR γ/δ, and CD235a (Miltenyi Biotec, 130–096–495). Human CD8⁺ T lymphocytes were cultured in Iscove's modified Dulbecco's medium (Thermo Fisher Scientific, 21980–032) supplemented with 10% human serum (Sanquin), 1% penicillin and streptomycin (PAA, P11–010), 1% sodium pyruvate (Gibco, 11360–039), and 0.1 mmol/L nonessential amino acids (Gibco, 11140–035). Isolated CD8⁺ T cells were activated using anti-CD3/CD28 Dynabeads (Thermo Fisher Scientific, 11131D) in a 96-well round-bottom plate (8 × 10⁴ CTLs/well) at a bead-to-cell ratio of 1:1. After 3 days of culture, the Dynabeads were magnetically removed, and the cells were resuspended in a mixture of 50% CTL expansion culture supernatant with 50% fresh TCM, followed by culture in 96-well round-bottom plates (8 × 10⁴ CTLs/well) in the presence of recombinant human (rh) IL2 (50 U/mL; Chiron, NDC 53905–991–01) for 24 hours.

B16F10/OVA target cells were stimulated with IFNγ (200 U/mL; PeproTech, 315–05) for 24 hours prior to seeding to enhance SIINFEKL antigen presentation (22). Target cells were seeded in a 96-well flat-bottom plate (7,000 B16F10/OVA cells/well with 200 U/mL IFNγ; 10.000 MEC1/OVA cells/well without IFNγ) and allowed to adhere and spread overnight to form a subconfluent target cell layer. Preactivated OT-I CTLs were resuspended in bovine collagen solution (1.7 mg/mL; Advanced BioMatrix, 5005) and overlaid onto the target cell monolayer. After collagen polymerization (30 minutes at 37°C), TCM was added to each well, and the coculture was incubated for 30 hours (MEC1/OVA) or 48 hours (B16F10/OVA).

For proliferation analysis, preactivated OT-I CTLs were stained with 400 nmol/L CFSE (10 minutes at room temperature) prior to embedding in the collagen matrix. For screening, the 3D interface cocultures were treated with IL2 (ABD Serotec, PMP38), IL8 (R&D Systems, 208-IL), IL10 (eBioscience, 14–8101), IL12 (BD Biosciences, 555256), IFNγ (PeproTech, 315–05), TGFβ (R&D Systems, 246-LP/CF), TNFα (PeproTech, 315–01A), sucrose (Sigma-Aldrich, S7903), β-lactose (Sigma-Aldrich, L3750), H₂O₂ (Merck, 107209), cisplatin (Sigma-Aldrich, P4394), l-lactate (Sigma-Aldrich, L7022), hydrochloric acid (HCl; Boom, 76021889.2500), N-{4-[3,5-bis(Trifluoromethyl)-1H-pyrazol-1-yl]phenyl}-4-methyl-1,2,3-thiadiazole-5-carboxamide (BTP2; Sigma-Aldrich, 203890), 2-aminoethoxydiphenylborane (2-APB; Sigma-Aldrich, D9754), cyclosporin A (Cell Signaling Technology, 9973), kaempferol (TOCRIS, 3603), rotenone (Sigma-Aldrich, R-8875), and piericidin A (Enzo Life Science, ALX-380–235). The treatment concentrations are listed in the relevant figures or corresponding figure legends, and all compounds were added at the start of the 30-hour (for MEC1/OVA) or 48-hour (for B16F10/OVA) coculture assay. For 3D coculture in the presence of lactate or HCl, TCM was used that did not contain HEPES. Hypoxic cultures were performed at 2% oxygen (CB53 incubator, Binder GmbH).

Confocal images were taken after 30 hours (MEC1/OVA) or 48 hours (B16F10/OVA) of coculture by automated spinning disc confocal microscopy (BD Pathway 855 High-Content Bioimager; Olympus UApo/340 20x/0.75). For GFP signal detection, the 488/10 excitation filter and 520/35 emission filter were used. For propidium iodide (PI; Sigma-Aldrich, P4170) signal detection, the 548/20 excitation filter and 645/75 emission filter were used. To discriminate dead from live cells, PI (10 μL/well; 0.5 mg/mL in PBS; Sigma-Aldrich, P4170) was automatically added 20 minutes before imaging.

For image processing and quantification, we generated a macro in Fiji (version 2.0.0-rc-69/1.52p). The source code is available at https://github.com/Mverp/AutoPlot_/releases. In brief, fixed pattern noise was reduced by subjecting the maximum intensity projections of GFP and PI to a flat-field and background correction. Cell nuclei masks were generated by thresholding the summed GFP and PI images, and subsequently applying a 3 × 3 median filter, watershed separation algorithm, and particle analysis (all Fiji). The resulting regions of interest (ROI) corresponded to the individual target cell nuclei and were used to calculate the integrated fluorescent signal intensities per individual target cell nucleus for each flat-field corrected image.

A second FIJI plugin was created to gate cell populations from scatter diagrams and generate quantitative data output (https://github.com/Mverp/AutoPlot_/releases). This plugin plots dye intensity values of individual target cells in a 2D scatter graph (e.g., PI against GFP), as the basis for interactive gating of cell subsets. This plugin was used to count cells within a defined population, visualize the scatter plot profile of gated target cell subsets, and generate galleries of cropped cell images, aiding visual inspection of the morphology of the gated cells.

Plots and gated events were exported as images and histograms. CTL killing efficacy was calculated by normalizing the number of surviving cells in the CTL-containing culture to the number of viable cells in the CTL-free culture (0% value). Killing efficacies and statistical comparisons were displayed as heat maps using the GiTools software (www.gitools.org; version 2.2.2).

Cells were harvested from the 3D interface coculture by enzymatic digestion of the collagen gel using collagenase I (190 U/well, 30 minutes at 37°C; Sigma-Aldrich, C0130). The resulting cell suspension was collected, and the remaining adherent cells were detached using trypsin (0.075%) and EDTA (1 mmol/L). The cell suspension after trypsin/EDTA treatment was unified with the cell suspension after collagen digestion (without centrifugation), and the whole volume was analyzed by flow cytometry (BD FACSCalibur). For measuring the effect of microenvironmental and pharmacologic modulators on CTL viability, OT-I CTLs (2.4 × 10⁶ cells/mL) were cultured in 2D for 6 hours without target cells and analyzed by flow cytometry. To determine cell surface expression of CD107a or PD-1, cells were stained with Alexa Fluor 488–conjugated anti-mouse CD107a (BioLegend, 121608), anti–PD-1 (Thermo Fisher Scientific, 14–9985–82), or the respective isotype controls (BioLegend, 400525; Thermo Fisher Scientific, 14–4888–81). Primary antibodies were detected with Alexa Fluor 488–conjugated donkey anti-rat (Thermo Fisher Scientific, A-21208) or Alexa Fluor 488–conjugated goat anti-Armenian hamster (Jackson ImmunoResearch, 127–545–160). The cell suspension was analyzed by flow cytometry (BD FACSLyric or BD CyAn ADP). Cells were gated on intact morphology, viability by PI/SYTOX blue (Thermo Fisher Scientific, S11348) exclusion, and dsRed expression using FlowJo (Tree Star, version 10) or FCS Express (De Novo, version 4) software.

Time-lapse monitoring of CTL-mediated cytotoxicity was performed using a customized imaging chamber (23) or 96-well flat-bottom microplate (Greiner Bio-One, 665090). MEC1/OVA cells (90 cells/mm² in an imaging chamber) or B16F10/OVA cells (2,800 cells/well in a 96-well plate) were allowed to adhere and spread for 6 to 8 hours to form a subconfluent target cell layer. For direct comparison of 2D versus 3D coculture, B16F10/OVA cells were seeded at 2,800 cells/well in a 96-well plate to obtain an individualized low-density target cell distribution. Preactivated OT-I CTLs were resuspended in bovine collagen solution (1.7 mg/mL; Advanced BioMatrix, 5005) and overlaid onto the target cell monolayer. After collagen polymerization (30 minutes at 37°C), TCM was added, and the coculture was monitored by brightfield time-lapse microscopy (Leica DM IL inverted phase contrast microscope; Leica EF L 20x/0.30) or spinning disc confocal microscopy (BD Pathway 855 High-Content Bioimager; Olympus UApo/340 20x/0.75) for up to 24 hours (at 37°C) with a 30- to 99-second frame interval.

The duration, kinetics and killing outcome of individual CTL-target cell interactions was quantified by manual analysis. CTLs migrating in an out-of-focus plane above the target cell in the 3D collagen matrix were not included for analysis. Direct CTL–target cell contacts were identified based on the following morphologic and kinetic criteria: polarization/flattening of the CTL on the target cell surface, deviation of the CTL migration path along the target cell surface, and slowed CTL migration speed or migration arrest.

For analyzing the effect of microenvironmental and pharmacological modulators on CTL migration speed, preactivated OT-I CTLs were cultured under microenvironmental or metabolic challenge (as indicated in the “Cytotoxicity coculture” section) for 6 hours, followed by CTL migration speed analysis in target cell-free 3D collagen matrix culture. CTL migration was quantified for a 3-hour time period using computer-assisted cell tracking (AutoZell 1.0 Software; Center for Computing and Communication Technologies, University of Bremen, Bremen, Germany).

A 96-well flat-bottom microplate (Greiner Bio-One, 665090) was coated overnight with anti-mouse CD3ϵ (10 μg/mL; BioLegend, 100339) or anti-human CD3 (10 μg/mL; Thermo Fisher Scientific, 16–0037–85). Preactivated CTLs were incubated with DMSO or BTP2 (10 μmol/L, 1 hour at 37°C; Sigma-Aldrich, 203890) and seeded in anti-CD3-coated 96-well plates (150,000 CTLs/well; 37 °C) for 10 minutes for phospho-Src detection or 2 hours for NFAT1 detection. Cells were fixed with Bouin solution (15 minutes, room temperature; Klinipath, 64096) for phospho-Src detection or 2% paraformaldehyde (PFA; 15 minutes, room temperature; Merck, 104005) for NFAT1 detection. After permeabilization [1x PBS, 0.1% Triton X-100 (Sigma-Aldrich, T8787), 10% normal goat serum (Thermo Fisher Scientific, PCN5000), 1% BSA (Sigma-Aldrich, A9647); 15 minutes, room temperature] and blocking of nonspecific binding sites [1 hour, room temperature; 1x PBS, 0.05 Tween-20 (Sigma-Aldrich, P7949), 10% normal goat serum, 1% BSA], the cells were stained with anti–phospho-Src (Tyr-416) family kinase (2.5 μg/mL; Cell Signaling Technology, 2101), anti-NFATC2 (10 μg/mL; Thermo Fisher Scientific, MA1–025), anti-NFAT1 (2 μg/mL; Cell Signaling Technology, 5861), Alexa Fluor 647–conjugated anti-CD45 (2.5 μg/mL; BioLegend, 103124), or Alexa Fluor 488 Phalloidin (Thermo Fisher, A12379) overnight at 4°C in blocking solution. Background staining was detected using rabbit IgG (R&D Systems, AB-105-C) or mouse IgG1 isotype control (Thermo Fisher Scientific, 6–4714–82). Primary antibodies were detected by incubating the cells with Alexa Fluor 568–conjugated goat anti-rabbit (2.5 μg/mL; Thermo Fisher Scientific, A11036), Alexa Fluor 647–conjugated goat anti-rabbit (2.5 μg/mL; Thermo Fisher Scientific, A21245), or Alexa Fluor 647–conjugated goat anti-mouse (2.5 μg/mL; Thermo Fisher Scientific, A21236) overnight at 4°C in blocking solution.

Cells were imaged using a Zeiss LSM880 (63 × 1.4 NA oil immersion objective, Carl Zeiss). For phospho-Src analysis, confocal microscopy was performed with sequential 561 nm and 633 nm excitation. Emission light was collected using 568 to 642 nm and 646 to 756 nm filters for Alexa 568 and Alexa 647, respectively. For NFAT1 analysis, airyscan imaging was performed with sequential 405 nm, 488 nm, and 633 nm excitation. Emission light was collected using a BP420–480/BP495–620, 3P 495–550/LP 570 and 3P 570–620/LP645 for DAPI, Alexa 488, and Alexa 647, respectively. Raw images were reconstructed using the Zeiss Zen 2.1 Sp1 software. All images were processed using Fiji/ImageJ.

B16F10/OVA spheroids were generated in hanging drop culture (5,000 cells/drop in 5% methylcellulose) for 2 days and stimulated with IFNγ during the second day of aggregation (200 U/mL for 24 hours; PeproTech, 315–05). B16F10/OVA spheroids were resuspended in bovine collagen (1.7 mg/mL; Advanced BioMatrix, 5005) with 80,000 preactivated OT-I CTLs. After collagen polymerization (30 minutes, 37°C), TCM was added to each well and the coculture was maintained for 48 hours.

Cells were harvested from the 3D spheroid coculture by enzymatic digestion of the collagen gel using collagenase I (190 U/gel, 30 minutes at 37°C; Sigma-Aldrich, C0130) in FCS-free TCM. Cells were dissociated by adding trypsin (0.075%) and EDTA (1 mmol/L) to the cell suspension (20 minutes at 37°C), followed by trypsin/EDTA inactivation using FCS (10%). The cell suspension was analyzed by flow cytometry (BD FACSLyric). Cells were gated on intact morphology, viability by SYTOX blue exclusion, and dsRed expression using FlowJo (Tree Star, version 10) software.

Preactivated OT-I CTLs (3.6 × 10⁶ cells/mL) were incubated with Fura-2 AM (2 μmol/L, 30 minutes, 37°C, in the dark; Thermo Fisher Scientific, F1221), washed twice with medium, and embedded in the 3D interface coculture with unlabeled MEC1/OVA or B16F10/OVA target cells at an effector-to-target (E:T) ratio of 4:1. Ca²⁺ signals in the OT-I CTLs were monitored by spinning-disk confocal microscopy (BD Pathway 855 High-Content Bioimager; Olympus UApo/340 20x/0.75) using a frame interval of 18 to 106 seconds. To account for phototoxicity and bleaching, imaging periods were limited to 15 minutes. CTL calcium fluxes were analyzed by manual ROI selection using the Fura-2 340/380 ratiometric images (Fiji software, version 2.0.0-rc-69/1.52p).

MEC1/OVA and B16F10/OVA target cells were engineered to stably express the calcium sensor GCaMP6s (24). CTL-target cell conjugation and associated intracellular Ca²⁺ events were obtained by long-term confocal microscopy of GCaMP6s and OT-I CTLs at frame intervals of 8 to 15 seconds for up to 12 hours (Leica SP8 SMD; Leica HC PL APO CS 40x/0.85) using a fixed focus, 8.358-μm optical section, and an excitation power of 0.05 mW for each 488 and 561 nm laser line. Viability of CTLs was verified based on unperturbed CTL migration speed, polarity, and morphologic integrity. Viability of target cells was verified by unperturbed morphologic integrity, and proliferation rates compared to bright-field recordings of CTL-free target cell cultures. CTL-associated Ca²⁺ events were identified by manual frame-by-frame image intensity analysis (Fiji software, version 2.0.0-rc-69/1.52p) and displayed for cell populations of multiple independent experiments.

Parametric datasets were compared using a one-sample t test, unpaired t tests, or two-way ANOVA. Nonparametric datasets were compared using a Mann–Whitney U or Kruskall–Wallis test. Holm–Sidak, Bonferroni and Dunn multiple comparisons correction was performed for one-sample t tests, two-way ANOVA, and Kruskall–Wallis tests, respectively. All data were analyzed using GraphPad Prism 8.

A 3D culture approach was developed to combine high-quality microscopy with screening for modulators of CTL migration, CTL interaction with individual target cells, and CTL effector function within a collagen-based environment. Preactivated OT-I CTLs were embedded in 3D fibrillar collagen overlaid on a monolayer of ovalbumin (OVA)-expressing target cells, thus rendering CTL-target cell interactions dependent on active CTL migration (Fig. 1A). The interface topology of the matrix environment mimics 2D/3D tumor–matrix interfaces present in vivo (25). CTL engagement with antigenic target cells led to target cell death, including membrane blebbing, condensation of H2B-eGFP–labeled chromatin, and nuclear fragmentation (Fig. 1B; Supplementary Movie S1). Nuclear condensation and fragmentation were validated as marker for target cell death using PI (Fig. 1C). Cytotoxic effector function was quantified by automated 3D confocal microscopy of surviving target cells, based on nuclear H2B-eGFP for target cell detection and PI as viability marker (Fig. 1D; Supplementary Fig S1A). Images were automatically segmented to define the GFP and PI integrated intensity in single target cell nuclei, followed by quantification of viable target cells (Fig. 1D; Supplementary Fig. S1A). The image-based endpoint analysis was complemented by live-cell time-lapse microscopy to map changes in any key step of the CTL effector response, including interstitial migration toward and between target cells, contact acquisition, surface scanning of target cells, and cytotoxic hit delivery (Fig. 1A and B).

To determine the bulk CTL effector response in an effective antigen-dependent model, we confronted OT-I CTLs with transformed mouse embryonic fibroblasts expressing the OVA peptide (MEC1/OVA) and the costimulatory molecule B7.1 (21). With increasing CTL density, the fraction of surviving MEC1/OVA cells decreased, reaching >95% death rates at E:T ratios of 1:8 and higher (Fig. 1E). The viability of MEC1 control cells, not expressing the OVA peptide, remained near-100%, irrespective of the E:T ratio (Fig. 1E; Supplementary Fig. S1B). In contrast to matrix-free 2D coculture, which showed short-lived CTL interactions with target cells (median: 8.3 minutes; Supplementary Fig. S1C and S1D; Supplementary Movie S2) and inefficient killing of target cell populations (Supplementary Fig. S1E), the 3D interface model enabled more sustained CTL–target cell interactions in low-density coculture (median: 276 minutes; Supplementary Fig. S1C and S1D; Supplementary Movie S2) and enhanced CTL killing efficacy (by up to 2.6-fold; Supplementary Fig. S1E). Thus, the collagen-based 3D interface coculture combines the optical advantages of 2D culture with a 3D matrix environment, thereby allowing for optical assessment and automated microscopic quantification of the antigen-dependent CTL effector responses against target cells.

We performed a small-scale screen for microenvironmental perturbations and modulators of cell-intrinsic metabolic pathways that might modulate CTL effector responses against MEC1/OVA target cells. Based on the logarithmic phase of the MEC1/OVA killing curve (Fig. 1E), low E:T ratios ranging from 1:16 to 1:256 were sufficient for sensitive detection of cytotoxicity. The dynamic range of the 3D interface coculture was tested by using the immunostimulatory cytokine IL2 and the immunosuppressive drug cyclosporin A (CsA), both established modulators of the CTL killing efficacy (26, 27). As expected, the killing of MEC1/OVA cells was doubled in the presence of IL2 and reduced (by up to 70%) by CsA across a range of E:T ratios (Fig. 2A–C).

Next, we tested 44 conditions, representing metabolic perturbations (e.g., lactate, hypoxia, acidosis), immune modulatory cytokines (e.g., IFNγ, IL8, IL10, IL12, TGFβ, TNFα), and targeted perturbations of mitochondrial respiration (rotenone, piercidin A), ROS production (kaempferol), and calcium mobilization (BTP2, 2-APB). IFNγ increased MEC1/OVA killing up to 5-fold at a concentration of 100 U/mL (Fig. 2D, #1). Likewise, IL10 nearly doubled MEC1/OVA killing at low E:T ratio (Fig. 2D, #2). The β-galatoside lactose stimulated the CTL effector response against MEC1/OVA cells by up to 28% (Fig. 2D, #3). Lactose protects effector CTLs from inhibitory immune checkpoint signaling by antagonizing the ligand binding function of galectins (28). In contrast, the glycolytic metabolite lactate exerted an immunosuppressive effect on the CTL effector response. Extracellular lactate reaches concentrations of 30 mmol/L in solid tumors as a result of the Warburg effect (29). Using this lactate concentration in the 3D interface coculture, target cell killing was reduced by up to 47% (Fig. 2D, #4). Similarly, TGFβ compromised CTL-mediated cytotoxicity by 21% (Fig. 2D, #5). Elevated H₂O₂, a feature of tumors with deregulated redox homeostasis (30), reduced MEC1/OVA killing by 58% (Fig. 2D, #6). The mitochondrial complex I inhibitor piercidin A mimics the reduced oxidative metabolism observed in tumor-infiltrating CTLs (31) and compromised MEC1/OVA killing by 38% (Fig. 2D, #7). Similarly, pharmacologic inhibition of Ca²⁺ release-activated Ca²⁺ (CRAC) channels by BTP2 or 2-APB treatment reduced the CTL killing efficacy by up to 71% (for BTP2, Fig. 2D, #8) and 63% (for 2-APB, Fig. 2D, #9), respectively. CRAC signaling is required for initiating downstream CTL effector responses, and the CRAC channel machinery is compromised under conditions of oxidative stress and upon acidification of the extracellular TME (32, 33). Consistently, oxidative stress by H₂O₂ diminished the calcium flux into CTLs upon target cell conjugation (Supplementary Fig. S2; Supplementary Movie S3).

For validation in an antigenic tumor model, which may differ in metabolic reprogramming to the MEC1 mouse embryonic fibroblast model and show partial resistance to CTL-mediated killing, metabolic conditions were validated using mouse melanoma B16F10 cells expressing the OVA peptide (B16F10/OVA). Compared with MEC1/OVA cells, the killing of B16F10/OVA cells by OT-I CTLs required 8- to 64-fold higher E:T ratios (Supplementary Fig. S3). IL2 tripled the killing of B16F10/OVA cells, whereas CsA diminished B16F10/OVA cell killing by 41% (Fig. 2E, #1 and 2). Likewise, the inhibition of CTL-mediated killing in the presence of H₂O₂, piercidin A, BTP2, and 2-APB was confirmed in B16F10/OVA cells (Fig. 2E, #3–6). However, no effect was noted for IFNγ, TGFβ, lactose, and lactate (Fig. 2E, #7–10). The alkylating chemotherapeutic drug cisplatin reduced target cell killing (by 26%) only in the B16F10/OVA (Fig. 2E, #11), but not the MEC1/OVA model (Fig. 2D, #10). The majority of microenvironmental and metabolic conditions did not affect target cell density in control culture without CTLs, except for hypoxia, rotenone, 2-APB, and cisplatin (Supplementary Fig. S4A–S4C). CTL viability was unaffected by the screening hits validated in the B16F10/OVA model (Supplementary Fig. S4D), confirming that these metabolic stressors did not induce direct CTL death but impaired other steps of the CTL effector response. The general ability of CTLs to migrate was maintained under metabolically perturbed conditions, with only hypoxia and the antioxidant kaemferol compromising CTL migration speed (Supplementary Fig. S5). Thus, microscopy-based screening detected effector mechanisms enhancing or compromising CTL-mediated cytotoxicity, with or without compromising CTL migration. Because BTP2 and 2-APB compromised the CTL effector response against both MEC1/OVA and B16F10/OVA cells strongly and across a broad range of E:T ratios, we pursued this effect further to explore the involvement of CRAC signaling in the stepwise CTL effector response.

We investigated the effect of CRAC channel signaling on the individual substeps of the CTL effector cycle by time-lapse imaging of single CTL dynamics and target cell death (Supplementary Fig. S5; Supplementary Movie S4). The ability of CTLs to migrate within 3D collagen and to interact with target cells was not compromised by CRAC channel inhibition (Fig. 3A; Supplementary S5; Supplementary Movie S4). Similarly, neither the slow CTL dynamics during conjugation with individual target cells (Fig. 3B) nor the individual contact duration until CTL detachment was affected by BTP2 (Fig. 3C). BTP2 did not affect membrane proximal TCR signaling, as evidenced by the unaltered ability of OT-I CTLs to form phospho-Src–positive microclusters upon CD3 ligation (Fig. 3D and E). In control cultures, binding to target cells triggered a monophasic CTL Ca²⁺ influx within minutes of target cell binding, followed by a sustained elevated plateau level of intracellular calcium for the duration of the interaction (Fig. 3F and G). In contrast, Ca²⁺ influx into CTLs was reduced in amplitude, limited in duration and fluctuating in the presence of BTP2, despite persisting CTL–target cell interaction (Fig. 3F and G; Supplementary Movie S5). Likewise, nuclear localization of the calcium-regulated nuclear factor of activated T cells 1 (NFAT1), a downstream effector of TCR activation and calcium signaling, was reduced (by 72%) upon OT-I CTL restimulation during CRAC channel inhibition (Fig. 3H and I). Validation of decreased CTL restimulation by BTP2 in preactivated human CTLs showed a reduced amplitude and shorter duration of the calcium flux into the CTL (Supplementary Fig. S6A and S6B; Supplementary Movie S6), as well as decreased nuclear NFAT1 localization (by 66%, Supplementary Fig. S6C and S6D). Thus, both the strength and duration of the CTL calcium influx, as well as downstream NFAT signaling, were dampened by CRAC channel inhibition.

NFAT is a critical transcriptional regulator of the CTL effector response, mediating T-cell proliferation and inhibitory checkpoint expression upon persistent antigenic stimulation (34, 35). In OT-I CTLs, reduced NFAT1 localization to the nucleus during CRAC channel inhibition was associated with decreased proliferation (Fig. 3J; Supplementary Fig. S7A), decreased frequency of CTL subsets that exhibit high PD-1 expression (Fig. 3K and L; Supplementary Fig. S7B and S7C) and delayed (by 2.3-fold) induction of target cell death after initial CTL contact (from 39 to 106 minutes; Fig. 3M). Consistent with decreased CTL reactivation and compromised induction of target cell death, despite unperturbed duration of CTL–target cell interactions, CRAC channel inhibition reduced cell surface CD107a expression by 30%, both after short (6 hours) and long (24 hours) incubation in the 3D interface coculture (Fig. 3N and O).

We confirmed the effect of BTP2 in a 3D spheroid model, consisting of solid multicellular tumor-like aggregates. Here, CRAC channel inhibition completely abolished B16F10/OVA killing by OT-I CTLs (Fig. 3P and Q). Thus, although CRAC channel activity was dispensable for CTL migration and contact kinetics, it was important for sustaining TCR signaling, timely exocytosis of granules, and effective death induction.

To investigate why target cell killing was delayed despite stable CTL–target cell contacts, we monitored the functionality of the cytolytic immune synapse and cytotoxic hit delivery at the single-cell level. Our previously published data indicate that single CTL interactions with solid tumor cells, including B16F10/OVA cells, are typically nonlethal because (i) only half of the CTL conjugates are associated with perforin-based hit delivery and (ii) the intracellular damage induced by a single perforin hit is time limited and can be resolved by the target cell within minutes to hours (5). Consequently, tumor cells typically survive rare perforin hits, and require a timely series of multiple sublethal hits within a few hours to undergo apoptosis (5). We thus tested the effect of CRAC channel inhibition by BTP2 on the rate of sequential hits delivered by OT-I CTLs to target cells.

As a proxy for individual cytotoxic events, the calcium sensor GCamP6s (24) was introduced into MEC1/OVA cells. CTL degranulation induces transient perforin-mediated pores in the target cell membrane, which facilitates diffusion of extracellular calcium into the target cells and results in an increased brightness of the GCAMP6s sensor (1, 5). Transient Ca²⁺ events in the target cell (“perforin events”) during CTL engagement were monitored by long-term time-lapse confocal microscopy and identified by manual frame-by-frame image intensity analysis (Fig. 4A; Supplementary Movie S7). In control cultures, CTLs induced a perforin event during 80% of the CTL–MEC1/OVA cell interactions (Fig. 4B). The majority of these productive CTL contacts were efficient, as 82% of them induced two or more perforin events by the same CTL (Fig. 4C). BTP2 treatment reduced the probability to induce a perforin event by 61% (from 80% to 31%), with a 33% (from 81% to 54%) reduction in serial event induction by productive CTL contacts (Fig. 4C).

In coculture with B16F10/OVA cells, CTLs induced at least one perforin event during 50% of the interactions (Fig. 4D). This confirmed a decreased susceptibility of B16F10/OVA cells to lethal hit delivery, compared with MEC1/OVA cells (Supplementary Fig. S3). BTP2 reduced the probability to induce a perforin event in the B16F10/OVA cells by 70% (from 53% to 16%; Fig. 4D), with a 51% reduction in sequential perforin event induction by productive CTL contacts (Fig. 4E). Similarly, the CRAC channel inhibitor 2-APB reduced the probability to induce perforin events in B16F10/OVA cells (by 66%), with a concomitant decrease in serial perforin event induction by productive CTL contacts (by 61%; Supplementary Fig. S8). Thus, CRAC channel signaling is critical for timely CTL degranulation and serial perforin hit delivery to target cells.

Understanding the cellular and molecular mechanisms by which a metabolically perturbed TME compromises CTL effector function requires time-resolved analysis of the multistep CTL effector response at cellular and molecular level. The culture platform developed here allows for medium-throughput analysis of CTL effector responses in a 3D interface coculture model, which combines the optical strengths of 2D culture with a 3D matrix environment. Besides microenvironmental modulators, including hypoxia and H₂O₂, we identified the involvement of CRAC channel activity in controlling contact-dependent calcium influx into CTLs, timely CTL degranulation, and serial sublethal perforin hit delivery to target cells. Deciphering the response of individual CTL effector phases to environmental perturbation will improve targeted intervention to restore the full sequence of CTL effector function in 3D interface culture.

Microenvironmental modulators of the CTL effector response can be pursued by complementary strategies: (i) directly mimicking metabolic features of the tumor microenvironment or (ii) pharmacologic modulation of pathways triggered by microenvironmental factors. Metabolic conditions within the TME can impede CRAC channel signaling in T cells (32, 33), and the pharmacologic inhibitor BTP2 was used to inhibit CRAC channels in preactivated CTLs. However, besides CRAC channel inhibition, 10 μmol/L BTP2 can interfere with cellular calcium homeostasis by activating transient receptor potential melastatin 4 (TRPM4) channels and inhibiting transient receptor potential canonical 3 (TRPC3) and canonical 5 (TRPC5) channels (36, 37). This could compromise CTL effector responses by perturbing calcium signaling during CTL restimulation, independent of CRAC channel inhibition. Therefore, we validated our findings with the orthogonal CRAC channel inhibitor 2-APB. At 75 μmol/L, 2-APB exerts off-target activity on mitochondrial calcium export and transient receptor potential vanilloid (TRPV) channels (38, 39). These off-target effects are, however, different from those of BTP2, thereby minimizing the risk that off-target effectors were responsible for the observed impairment in CTL effector function.

In 3D interface culture, inhibition of CRAC channel activity compromised calcium signaling in the CTL upon conjugation and lowered the frequency of sequential sublethal hits delivered to the target cell, although contacts remained stable. This is consistent with the role of CRAC channel signaling in mediating the fusion of lytic granules with the plasma membrane at the immune synapse (40). CTL degranulation induces the formation of perforin-mediated pores in the target cell membrane, which facilitates diffusion of granzyme B and extracellular calcium into the target cell (41, 42). The calcium influx is transient and activates a cellular repair response to reseal the pores in the target cell membrane, thereby protecting the target cell from rapid death (43). Cytotoxicity is facilitated by CTL cooperation, in which multiple CTLs deliver a sequence of sublethal hits that accumulate over time to induce target cell death. When quantified, B16F10/OVA cells which receive two hits or less, as detected by transient calcium elevation, maintain a high survival probability of 80%, whereas three or more hits result in a more than 50% decrease in target cell survival probability (5). Sublethal hit delivery induces structural damage in the target cell, including nuclear envelope defects and DNA damage, which is reversible over minutes to hours (5), indicating that target cells may resist CTL attack by repairing CTL-induced sublethal hits. The delay of CTL degranulation and consecutively decreased frequency of delivered perforin hits upon CRAC channel inhibition suggests that target cells experience a prolonged time span between hits, and thus are more likely to recover from individual sublethal events and survive. Thus, the ability of CTLs to deliver sublethal hits at high frequency depends on CRAC channel function. Perturbed accumulation of lethal events may underlie the inability of CRAC channel–deficient CTLs to control the growth of melanoma and colon carcinoma tumors (44). Therapeutic stimulation of calcium influx in CTLs can restore impaired lytic granule release caused by an immunosuppressive TME (45). Thus, the ability of CTLs to deliver sublethal hits at high frequency depends on CRAC channel function.

The image-based 3D interface coculture is a versatile model to study modulators of the multistep CTL effector cycle. Both interference with mitochondrial respiration and induction of oxidative stress compromised CTL effector function. Piercidin A inhibits mitochondrial complex I and thereby mimics the reduced oxidative metabolism observed in tumor-infiltrating CTLs (31). The compromised CTL effector response against MEC1/OVA and B16F10/OVA cells in the presence of piercidin A is consistent with a generally reduced lytic granule release (46). Oxidative stress interferes with proximal TCR signaling by lowering CD3ζ chain expression (47), and coexpressing the H₂O₂-degrading enzyme catalase in chimeric antigen receptor (CAR) T cells maintains their cytotoxic effector response against antigenic target cells despite H₂O₂-rich conditions (48). Consistently, H₂O₂ diminished CTL reactivation upon target cell conjugation and compromised MEC1/OVA and B16F10/OVA killing by OT-I CTLs. In-depth analysis will require deconstruction of the CTL effector cycle to define the specific effector mechanisms perturbed by piercidin A and H₂O₂.

In contrast to matrix-based spheroid models, which are optically convoluted and do not allow to resolve the kinetics of individual CTL–target cell interactions (20), the interface geometry used in the 3D interface coculture provides high-quality microscopic monitoring of the timing and spacing of sublethal hit delivery in the same focal plane. The matrix environment of the 3D interface model supports CTL motility, CTL conjugation, and effective killing of target cell populations, as opposed to the labile and inefficient interactions in 2D matrix-free culture. Additional information on the impact of individual CTL hits and the recovery of target cells between sublethal hits can be provided by using probes for nuclear lamina integrity and DNA damage (5). The 3D interface coculture can further be combined with single-cell biosensors for real-time monitoring of signaling pathways and microenvironmental challenges at cellular and subcellular resolution. This will allow to identify target cell subset behavior, including resistance development. Metabolic stressors may induce a mesenchymal phenotype in cancer cells, and mesenchymal programs may compromise immune synapse signaling and mediate resistance toward lymphocyte-mediated lysis (49). In addition, aggregation of cancer cells may increase their susceptibility to immune attack by spreading death signal across adjacent target cells (50). The 3D interface coculture can further be developed to better mimic the complexity of the TME by adding additional cell types that modulate CTL effector responses, including helper and regulatory T cells, myeloid-derived suppressor cells, and/or fibroblasts. Cells within the 3D interface coculture can also be harvested to perform in-depth analysis of target cell resistance signatures or CTL dysfunctional phenotypes by single-cell RNA sequencing and metabolomics analysis. Although the 3D interface coculture allows for time-resolved analysis of the multistep CTL effector response, including interstitial CTL shuttling between target cells, this model lacks the cell density and physical constraints, as well as vascularization and a reactive tumor stroma present in cancer tissues.

The 3D interface coculture model will be useful to address therapeutic approaches that reinvigorate defined phases of the CTL effector cycle within metabolically perturbed tumors and guide the rational design of combinatorial therapeutic strategies. Thus, microscopic analysis of CTL function in 3D interface coculture will facilitate the identification of immune-suppressive mechanisms and targeted intervention strategies to restore the CTL effector response.

J. Slaats reports support by a RadboudUMC PhD fellowship (RvB15.52139). C.E. Dieteren reports grants from NWO-Veni (863.13.019) during the conduct of the study. J.A. van der Laak reports personal fees from Philips, the Netherlands and ContextVision, Sweden, and grants from Philips, the Netherlands, ContextVision, Sweden, and Sectra, Sweden outside the submitted work. C. Figdor reports support by the Dutch Cancer Foundation (KWF 2008–4031). P. Friedl reports grants from the Dutch Cancer Foundation (KWF 2008–4031), Cancer Genomics Cancer (CGC), The Netherlands, and NWO (investment grant, 834.13.003) during the conduct of the study. No disclosures were reported by the other authors.

J. Slaats: Formal analysis, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. C.E. Dieteren: Conceptualization, data curation, software, formal analysis, supervision, validation, investigation, visualization, methodology, project administration, writing–review and editing. E. Wagena: Data curation, formal analysis, validation, investigation, visualization, methodology, writing–review and editing. L. Wolf: Software, validation, investigation, methodology, writing–review and editing. T.K. Raaijmakers: Formal analysis, validation, investigation, visualization, methodology, writing–review and editing. J.A. van der Laak: Software, writing–review and editing. C.G. Figdor: Conceptualization, funding acquisition, writing–review and editing. B. Weigelin: Supervision, methodology. P. Friedl: Conceptualization, resources, formal analysis, supervision, funding acquisition, methodology, writing–original draft, project administration, writing–review and editing.

The authors thank Dr. Stephen P. Schoenberger for providing the MEC1 cell line and Dr. Gosse Adema for providing the B16F10 cell line. They thank Dr. Yingxin Yu for providing the human CTLs.

This work was supported by the Dutch Cancer Foundation (KWF 2008-4031, to C.G. Figdor and P. Friedl), NWO-Veni (863.13.019, to C.E. Dieteren), a personal RadboudUMC PhD fellowship (to J. Slaats), and Cancer Genomics Cancer, the Netherlands (to P. Friedl). Time-lapse confocal microscopy was enabled by an NWO investment grant (834.13.003).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.