Actin conformation dynamics precede force generation in cytotoxic T lymphocytes

All antibodies, drugs and dyes are found in the reagents table (see Table 1).

Table 1 Reagents used Full size table

Mice

Tg(TcraTcrb)1100Mjb Rag1tm1Bal/tm1Bal (MGI:3054907 and MGI:2448994 alleles, referred to as OTI) mice were bred on a C57BL/6 background. Mice were housed in University of Cambridge establishments in individually ventilated cages at room temperature between 21–24 °C, humidity of 55% ± 10% and a 12/12 light/dark cycle. Mice were provided with ad libitum normal mouse diet and water, sizzle nest bedding, and enrichment. Breeding and maintenance of transgenic mice were carried out under UK Home Office project licence PP5905963. This research has been regulated under the Animals (Scientific Procedures) Act 1986 Amendment Regulations 2012 following ethical review by the University of Cambridge Animal Welfare and Ethical Review Body. ARRIVE reporting guidelines have been followed.

Spleens were obtained from male and female mice aged between 12 and 30 weeks. OTI mice were used because they offer an easy source of highly uniform CD8+ T cells with a known TCR.

Plasmid design and cloning

For live fluorescence microscopy, the following plasmids were used:

Construct Labelled target Source LifeAct-eGFP F-actin Michael Davidson (Addgene #54610)33 utrnP2A mApple-utrnLAM and eGFP-utrnWT This study, adapted from ref. 36 mem-Electra2 Plasma membrane This study

To design utrnP2A, the cDNA sequences of utrnLAM and utrnWT36, were N-terminally fused with mApple and eGFP respectively. A GSG linker followed by a P2A sequence37 was inserted 3’ of eGFP-utrnWT (with the stop codon removed), followed by mApple-utrnLAM. This construct was synthesised by Twist Biosciences (San Francisco, USA) with BamHI and NotI flanking sites, which were used to subclone into the pEGFP.N1 vector for transient expression in primary mouse T cells.

To design mem-Electra2, the first 20 amino acids of Rattus rattus neuromodulin61 were fused to the N-terminus of blue fluorescent protein Electra2 via a short linker62. The sequence was codon-optimised for Mus musculus and flanked with BamHI and NotI sites. The construct was synthesised by Twist Biosciences (San Francisco, USA) and subcloned into the pHR vector for lentiviral transduction.

Cell culture

CTLs were from OTI mice, generated as previously described63. Briefly, OTI splenocytes were stimulated with 10 nM OVA 257-264 peptide (SIINFEKL) (Cambridge Bioscience) in mouse T-cell medium: RPMI-1640 medium (SigmaAldrich, #1640) supplemented with 10% heat-inactivated FBS (LabTech, #FBS-SA), 50 mM β-mercaptoethanol (ThermoFisher, #31350010), 10 U/ml recombinant murine IL-2 (Peprotech, #212-12), 2 mM L-Glutamine (Sigma-Aldrich, #G7513), 1 mM sodium pyruvate (ThermoFisher Scientific, #11360070), and 50 U/ml penicillin and streptomycin (Sigma-Aldrich, #P0781). After three days, OVA 257-264 was removed, and cells were thereafter maintained in mouse T-cell medium with daily replacement of medium via centrifugation and resuspension. Primary cells were incubated in a humidified atmosphere at 37 °C with 8% CO 2 .

Target cells for OTI CTL imaging experiments and killing assays were EL4 cells (ATCC: TIB-39; RRID: CVCL_0255) stably expressing mem-Electra2 (see below). Cell lines were maintained in DMEM (Sigma-Aldrich, #D5030) supplemented with 10% heat-inactivated FBS. Cell lines were incubated in a humidified atmosphere at 37 °C with 10% CO 2 .

Lentiviral transduction

To generate EL4 target cells stably expressing mem-Electra2, Lenti-X 293 T cells (Takara, # 632180) were transfected with mem-Electra2 in the pHR backbone, pMD2.G (AddGene #12259), and pCMV ΔR8.2 (AddGene #12263) in a 5:1:4 ratio using TransIT reagent (Mirus Bio, #MIR 2704). Virus was harvested from the supernatant at 48- and 72 hours post-transfection, filtered, and concentrated using Lenti-X Concentrator (Takara, #631232). EL4 cells (ATCC: TIB-39; RRID: CVCL_0255) in logarithmic growth phase were resuspended in cell line medium at 4 × 106/mL containing concentrated virus and polybrene at 1 μg/mL (Merck, #TR-1003-G). After 16 hours, the cells were diluted four-fold in cell line medium and cultured for three days before using FACS to isolate single-cell-derived clones based on Electra2 fluorescence.

Nucleofection

OTI CTLs were nucleofected on days 4–6 post-stimulation from splenocytes. 5 × 106 CTLs were washed in PBS via centrifugation (200 × g, 5 min) and resuspended, before final resuspension in nucleofection mix, comprised of 2.5 μg plasmid DNA and P3 Primary Cell Solution up to a total volume of 100 μL. Cells were electroporated in 100 µL Nucleocuvette™ vessels (Lonza, #V4XP-3024) with a 4D-Nucleofector® X unit (Lonza, #AAF-1003X) using pulse code DN-100. CTLs were immediately transferred, with micro-Pasteur pipette, to wells of a six-well plate containing 2 mL pre-warmed nucleofection recovery medium: calcium-free RPMI (US Biological, #R8999-02A), 5% heat-inactivated FBS, 2 mM L-glutamine, 32 μM 1-thioglycerol (Merck, #M6145), 1.7 mM sodium pyruvate, 20 μM bathocuproine disulfate (Merck, #B1125). CTLs were left to recover at 37 °C with 8% CO 2 for 6 hours, before topping up with 6 mL of pre-warmed mouse T-cell medium.

LDH killing assay

Killing of EL4 targets by OTI CTLs was determined by release of lactate dehydrogenase from lysed cells. Cells were seeded in round-bottomed 96-well plates at varied effector-to-target ratios, as indicated, in phenol red-free RPMI-1640 (ThermoFisher, #11835030) supplemented with 2% heat-inactivated FBS, and 50 U/mL penicillin and streptomycin. Plates were centrifuged for 5 minutes at 300 × g prior to incubation at 37 °C with 8% CO 2 for three hours. Cell lysis was measured via CytoTox 96® Non-Radioactive Cytotoxicity Assay (Promega, #G1780). Briefly, after incubation, the plate was centrifuged once more and the supernatant transferred to a mirror-image flat-bottomed 96-well plate. Solubilised substrate mix (Promega, #G179A and #G180A) was added and incubated in the dark for 30 min, allowing for the enzyme-coupled reaction to proceed, resulting in conversion of tetrazolium salt into a red formazan product proportional to the number of lysed cells. After 30 min, absorbance at 490 nm was measured with a spectrophotometric plate reader. Specific target lysis was calculated by subtracting absorbance from medium-only control wells and comparing absorbance in wells containing effectors and targets with vs without OVA 257-264 pulsing.

Degranulation assay

CTLs were resuspended in fresh pre-warmed mouse T-cell medium at 1 × 106 cells/mL. Fluorescently conjugated anti-mouse CD107a (LAMP1) (ThermoFisher, clone 1D4B, #12-1071) was added to the medium at 2 μg/mL, before seeding CTLs onto flat-bottomed plates that had been pre-coated with 1 μg/mL αCD3ε (‘stim.’) or PBS (‘unstim.’). Plates were incubated at 37 °C with 8% CO 2 for the indicated time. Plates were then kept on ice and stained with a live-dead marker before analysing with an Attune NxT flow cytometer (ThermoFisher). The gating strategy is shown in Fig. S1c.

Live imaging assays

CTLs were nucleofected with fluorescent protein constructs 24 hours prior to imaging. Prior to imaging, cells were resuspended at 1 × 106 cells/mL in imaging medium: phenol red-free RPMI supplemented with 10% heat-inactivated FBS, 2 mM L-glutamine, 25 mM HEPES (ThermoFisher, #15630080), 50 U/mL penicillin-streptomycin.

For low-magnification migration speed assays: Imaging dishes (Mattek, #P35G-1.5-14-C) were pre-coated with 0.5 μg/mL ICAM-1 (R&D Systems, #796-IC) in PBS. 250 μL CTL suspension was added per dish and allowed 30 minutes at 37 °C with 8% CO 2 for the cells to adhere and begin crawling. After this time, imaging medium was gently passed over the dish to remove non-adhered CTLs before imaging. Single confocal slices were captured with a ×20 objective lens (Leica HC PL APO ×20) in the mid-plane of the cells every 5 seconds for 5 minutes per field of view. Only the FITC channel was acquired (LifeAct-eGFP or eGFP-utrnWT). Three fields of view were taken per condition per experiment.

For high-magnification live imaging, CTLs were resuspended in imaging medium and dropped onto functionalized glass or TFM hydrogel. In the case of artificial synapses, cells were immediately imaged to capture them coming into contact with the activating surface. In the case of imaging migrating CTLs, cells were left to settle, adhere, and begin to crawl for 30 min prior to imaging through a ×100 objective lens (Leica HC PL APO ×100/1.40 OIL CS2). Confocal z-stacks with slices 0.8–1 μm apart were acquired over the whole volume of the conjugates every 10–30 seconds for up to 20 min.

For drug treatments, drugs were added to CTLs in imaging medium half an hour before imaging, and kept in throughout the assays. Working concentrations are given in the reagents table (Table 1). DMSO was used at a dilution of 1:1000.

In the case of imaging of conjugates, EL4 target cells expressing mem-Electra2 were pulsed with 1 μM OVA 257-264 peptide for half an hour and then washed thrice by centrifugation and resuspension in cell line medium. Target cells were ultimately resuspended in RPMI (serum-free) at 1 × 106 cells per mL, and 250 μL seeded onto imaging dishes that had been pre-coated with ICAM as before. After 10 minutes at 37 °C with 8% CO 2 , imaging medium was gently passed over the imaging dish to remove non-adhered target cells. Nucleofected CTLs were dropped onto target cells and allowed 10 minutes to settle before imaging.

All live microscopy used an Andor spinning-disk confocal system (Revolution; Andor) fitted with a CSU-X1 spinning-disk unit (Yokogawa) via a DMi8 microscope (Leica). Samples were excited with a combination of lasers at 405, 488, 561, and 637 nm wavelength. Images were captured using an iXon Ultra 888 camera and Fusion software (Andor). Imaging was performed with cells at 37 °C, 5% CO 2 in an airflow sample chamber on the stage (OkoLab).

Immunocytochemistry

To image p-MLC2, CTLs expressing utrnP2A were fixed by addition of paraformaldehyde to 4% and incubation at room temperature for 20 min. After washing with PBS, cells were permeabilised with 0.1% Triton X-100 for 10 min at room temperature. Cells were stained with an antibody against phospho-MLC2 (Ser19) (ThermoFisher, #MA5-15163) at a dilution of 1:200 in ICC buffer (PBS with 1% BSA) for one hour at room temperature. After washing, secondary staining was performed with an AF647-conjugated anti-mouse IgG antibody (ThermoFisher, #A32728TR) at a dilution of 1:400 in ICC buffer for 45 min at room temperature. Cells were washed and imaged in PBS without mounting.

TFM gel preparation

Gel preparation was adapted from published work38 as follows:

For glass preparation: Imaging dishes (Mattek, #P35G-1.5-14-C) were cleaned via three 5 min incubations with 100% ethanol. After final aspiration, the glass was left to air-dry completely. The glass was then silanized via 30 min incubation at room temperature with 0.5% (3-aminopropyl)trimethoxysilane (APTMS) (Sigma-Aldrich, #281778) in ultrapure sterile water. APTMS solution was then aspirated, and the glass was rinsed with water thoroughly. The surface was prepared for gel adhesion by incubation for 30 min at room temperature with 0.25% glutaraldehyde (Sigma-Aldrich, #354400). The glass was rinsed thoroughly with water again and left to air-dry. Top coverglasses were prepared by cleaning 12 mm circular glass coverslips (VWR, #630-2200) via water bath sonication in 100% ethanol for 10 min, then leaving them to air-dry on a clean Kimwipe.

For gel preparation: Gel premix was created by adding 500 μL of 40% acrylamide solution (SLS, #A4058) and 65 μL of N-hydroxyethyl acrylamide (Sigma-Aldrich, #697931) to an Eppendorf tube and mixing. 65 μL was then removed (leaving 500 μL), and 250 μL of 2% N,N’-methylenebisacrylamide solution (Sigma-Aldrich, #M1533) was then added and mixed by vortexing. To create TFM gels with an average stiffness of 2.17 kPa, 65 μL of gel premix was added to 425 μL of DPBS (Gibco, #14190144) and mixed. To this, 10 μL of dark red 0.2 μm FluoSpheres™ Carboxylate-Modified Microspheres (ThermoFisher, #F8807) was added. This mixture was vortexed and then sonicated in a water bath sonicator for 10 min, and subsequently degassed via a vacuum desiccator for 4 min.

For TFM gel synthesis: To polymerise the gel, 1.5 μL of N,N,N’,N’-tetramethylethylenediamine (Sigma-Aldrich, #T9281) and 5 μL of 10% (w/v) ammonium persulfate solution (Sigma-Aldrich, #A3678) were added to 500 μL of TFM gel mix and mixed by gentle pipetting. As quickly as possible, 3 μL was added to the centre of the prepared glass imaging dish. A prepared top coverglass was then inverted on top, causing the gel to spread to the edges of the 12 mm glass coverslip. The entire imaging dish was then inverted and left for 15 min at room temperature for the gel to polymerise. Flipping the dish back again, DPBS was added to cover the glass and left for 30 min. Then the top cover glass was removed gently with tweezers. Gels were treated with poly-D-lysine (Gibco, #A3890401) overnight at 4 °C. The gel could then be functionalized with a biologically relevant molecule.

For migration on TFM gels, poly-D-lysine was rinsed three times with PBS, then replaced with 50 μg/mL ICAM-1(R&D Systems, #796-IC) in PBS and incubated at 37 °C for 1 hour, before three PBS washes and cell mounting. For artificial immune synapses on TFM gels, the same process was performed, but substituting ICAM-1 for anti-CD3ε (ThermoFisher, #14-0033-82) at the same concentration of 50 μg/mL. High concentrations were used to compensate for poor functional molecule adhesion compared to regular glass coverslips, ensuring saturation of the surface.

Gel thickness was measured on the microscope by focusing on the fluorescent beads on the top and bottom of the gel, subtracting the stage positions accordingly (Supplementary Fig. 2d).

Image analysis

Migration assay data were analysed with ImageJ software using the TrackMate plugin. Cells were masked using a Laplacian of Gaussian detector, with an estimated diameter of 11 μm and threshold of 2.0. Migration was tracked between frames via a simple LAP tracker, with 5 μm maximum linking distance and zero tolerance for gaps. Tracks with fewer than 10 spots were excluded from analysis. Mean track speed was exported and plotted in Prism software (GraphPad).

TFM assays were analysed via MATLAB, customising a published script (GitHub: https://github.com/DanuserLab/u-inferforce40). For the TFM analysis of the bead movement, the slice of the confocal z-stack, where the surface of the gel was in focus, was selected. Frames were drift-corrected relative to each other by efficient subpixel registration, and bead displacement was calculated frame-to-frame. Using the empirically measured gel stiffness of 2.17 kPa, traction force was then calculated using Fourier transform traction cytometry, with the L-curve method employed for regularisation parameter selection, using L-optimal criteria. To calculate the amount of force in the cell area, a cell mask was created using the Otsu threshold method for both utrnWT and utrnLAM fluorescent channels. The utrophin masks were then summed up to generate a cell mask. The force in the cell area was calculated by overlaying the force field with the generated cell mask and summing the magnitudes of the force vectors inside the cell area.

To determine the probe ratio in the maximal force zone, a 30 px radius circle centred on the maximal force pixel was used to ascertain the average probe ratio in this area.

For TFM analysis based on utrn probe distribution, each frame was normalised to the maximum intensity pixel before histogram-matching in each probe channel. For ratiometric cell segmentation, the utrnWT/utrnLAM ratio was calculated and categorised as “utrnLAM dominant”, “utrnWT dominant”, or “neutral” based on a ratio value two standard deviations below 1, two standard deviations above 1, or between the two thresholds. For migrating cells, maximum intensity projections were used to create the ratiometric image. For synapses, only the bottom z-slice (in focus with the synapse) was analysed.

For analysis of utrn probe area over time at the synapse, we generated time-stacks of each probe in the plane of the gel and performed automatic Otsu thresholding (such that all timeframes were incorporated into the thresholding algorithm). Total area per probe for each frame was then calculated based on these thresholds.

For colocalization analysis between utrnWT and utrnLAM at the synapse, a time-dependent colocalization channel was generated in Imaris (version 11.0.1, Oxford Instruments) with dynamic automatic thresholding for each channel and time point. Pearson’s correlation coefficient for each time point was extracted. For analysis across replicates, these r values were z-scored for each movie. Spearman’s rank-order correlation was used to test for a trend in colocalization over time.

For analysis of probe bias in CTL-target conjugates, initially (Fig. 5b, c) manual ROIs were drawn over the synapse for each timepoint as a MIP and mean intensity for each probe was measured and then normalised across timepoints. Alternatively (Fig. 5d), a region of the CTL in close proximity to the target cell was generated in Imaris software by creating a generous segmentation surface based on target cell fluorescence and then segmenting CTL within that surface based on utrnLAM intensity. Sum intensities for each probe in 3D were then measured within that synapse-specific CTL volume (which moved across timeframes automatically). Spearman’s rank-order correlation was used to test for a trend in probe bias over time.

For analysis of probe bias in drug-treated and floating cells, boxes were manually drawn over the rear and the front half of migrating cells in ImageJ. The average intensities of each utrn probe were measured in each box and normalised to the whole-cell average per channel. The ratio of normalised probe intensities in the front and rear of the cells was then calculated and expressed as a “bias index” which represents utrnLAM enrichment at the front of the cell.

For p-MLC2 colocalization analysis, Pearson’s correlation coefficient was calculated in ImageJ between p-MLC2 and utrnWT or utrnLAM channels on a per-pixel basis.

Data from imaging of live conjugates were analysed and processed for export in Imaris software (BitPlane). Image channels were pseudo-coloured for optimal visual contrast.

Atomic force microscopy

AFM measurements were performed on a JPK CellHesion200 (Bruker), with a tipless cantilever (ArrowTL, NanoWorld) to which a 37 μm diameter polystyrene bead (PS-R-37.0–microParticles GmbH) had been glued. The spring constant of the cantilever, determined by thermal tuning, was 0.092 N/m, the setpoint 10 nN and the speed of the Z scanner was 10 μm/s. Force-distance curves were analysed in the JPK Data Processing software (Bruker), where the Hertz-Sneddon Model was applied to calculate Young’s Modulus, assuming Poisson’s ratio to be 0.5.

Data presentation

Data were processed, statistical tests applied, and data plotted in Excel (Microsoft) and/or Prism (GraphPad). Flow cytometry data were plotted in FlowJo (BD). Schematic diagrams were produced in BioRender. Figures were collated in Illustrator (Adobe). The manuscript was processed in Word (Microsoft).

Statistics and reproducibility

Number of cells (n) and biologically independent replicates (N), specific statistical tests, and p values are given in figure captions. Statistical tests were performed in Prism (GraphPad). All tests were two-tailed.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.