The magnitude of LFA-1/ICAM-1 forces fine-tune TCR-triggered T cell activation

T cells defend against cancer and viral infections by rapidly scanning the surface of target cells seeking specific peptide antigens. This key process in adaptive immunity is sparked upon T cell receptor (TCR) binding of antigens within cell-cell junctions stabilized by integrin (LFA-1)/intercellular adhesion molecule–1 (ICAM-1) complexes. A long-standing question in this area is whether the forces transmitted through the LFA-1/ICAM-1 complex tune T cell signaling. Here, we use spectrally encoded DNA tension probes to reveal the first maps of LFA-1 and TCR forces generated by the T cell cytoskeleton upon antigen recognition. DNA probes that control the magnitude of LFA-1 force show that F>12 pN potentiates antigen-dependent T cell activation by enhancing T cell–substrate engagement. LFA-1/ICAM-1 mechanical events with F>12 pN also enhance the discriminatory power of the TCR when presented with near cognate antigens. Overall, our results show that T cells integrate multiple channels of mechanical information through different ligand-receptor pairs to tune function.

. Qualitative fluorescence microscopy for density calibration of tension probe/ICAM-1. A) Brightness of samples and TR-DPHE liposome were measured using fluorescence microscopy. B) A scaling factor F was generated to relate the brightness of sample to the TR-DPHE by from the ratio of slopes from solution calibrations. C) A TR-DHPE bilayer calibration curve with known molecular densities. D) F factor was used to scale the slope of the TR bilayer calibration to estimate the relation between the sample intensity on a surface and its surface density. Density of ICAM-1 and Cy3B-tension probe used in respective experiments are summarized in Table S3. showing the lateral mobility of A) Alexa647 tagged streptavidin and B) dimeric ICAM-1 can be manipulated using bilayers with different fluid-to-gel transition temperatures (99 mol% DOPC or DPPC + 0.1mol% biotinyl-Cap PE). Right are FRAP plots showing recovery of GFP or Alexa647 fluorescence over the course of 600 s. The Lateral diffusion coefficients (D) were calculated by: D = w 2 /4t1/2, where w is the radius of the Gaussian bleaching area and t1/2 is the time for 50% recovery obtained from the fit. Experiments were performed at 25 o C. Scale bar = 10 µm.

Fig. S5
. Quantifying probe quenching efficiency. A) Artificial opening of DNA force probes using a full complement (teal) that binds to the hairpin stem-loop, and this hybridization subsequently leads to de-quenching of the dye. B) Quantification of the fluorescence intensity of the "closed" and "opened" DNA probes immobilized on the surface via copper-free click reaction.

Fig. S6
. Image analysis pipeline to process the force signal. Briefly, time-lapse tension videos were subjected to photobleaching correction and drift correction. Then, a region of ROI containing at least a cell was isolated in both the tension and RICM images. From the cropped ROI, a manual local background subtraction was performed to remove signals contributed from DNA probes that were not subjected to cell pulling. The corrected image displaying tension signals were isolated using cell masks.      LFA-1 transmitted forces on tension probe substrate (F1/2 = 4.7 pN) was imaged in the Cy3B channel and adhesion was measured using RICM. Scale bar = 5 µm.

Fig. S13. TCR-pMHC tension magnitude enhances TCR activation. A)
Representative RICM and pY-ZAP70 immunofluorescence image (using a 100x oil objective) of naïve OT-1 cells seeded on 12 pN TGT or 56 pN TGT substrates for 1h. The surfaces presented OVA-N4 pMHC. B) Quantification both average and integrated pY-ZAP70 intensity of cells seeded on 12 pN or 56 pN TGT substrates using cell masks with same region of interest. N > 30 cells from three independent experiments. ****P < 0.0001. C) Representative brightfield and immunofluorescence images (imaged using a 40x objective) of naïve OT-1 cells seeded on 12 pN TGT or 56 pN TGT substrates for 1h. D) Quantification both average and integrated pY-ZAP70 intensity of cells seeded on 12 pN or 56 pN TGT substrates using cell masks with same region of interest. N > 50 cells from three independent experiments. **** P < 0.0001. Scale bars = 10 µm. For all experiments, Alexa-647 phospho-ZAP70 antibody was used to stain the cells and BF-defined ROI to compare the average intensity versus integrated intensity of pY-ZAP70 of cells seeded on 12 pN or 56 pN N4 TGT surfaces. This confirms integrated intensity of pY-ZAP70 is a valid readout for the degree of early T cell activation.

Fig. S14. Mapping LFA-1/ICAM-1 peak tension using turn-on TGT probes. A) Schematic
showing the design of turn-on TGT. B) Representative RICM time-lapse images and the corresponding turn-on TGT signal of αCD3 primed OT-1 cells. These cells were seeded on 12 pN ICAM-1-TGT or 56 pN ICAM-1 TGT substrates, and we started recording the time-lapse movies at ~5 min. C) αCD3 primed cells on ICAM-1 TGTs at t = ~30 min after seeding. Tension Images were normalized to the probe background for better visualization. Calibration bar represents fold change in fluorescence over background. Scale bar = 5 µm.     Table S1. Annotated amino acid sequences of the soluble ICAM-1 constructs.  Figure S3D. Since these molecules attached to only the upper leaflet of bilayer or the DNA probe is attached on the glass coverslip, unlike the Texas Red lipid probe which is presented on both bilayer leaflets. A factor of 2 is used to "correct" the protein density.

T A A G I T H G M D E L Y K G L N D I F E A Q K I E W H E G G G G S H H H H H H H H H H Dimeric ICAM-1 M A S T R A K P T L P L L L A L V T V V I P G P G D A Q V S I H P R E A F L P Q G G S V Q V N C S S S C K E D L S L G L E T Q W L K D E L E S G P N W K L F E L S E I G E D S S P L C F E N C G T V Q S S A S A T I T V Y S F P E S V E L R P L P A W Q Q V G K D L T L R C H V D G G A P R T Q L S A V L L R G E E I L S R Q P V G G H P K D P K E I T F T V L A S R G D H G A N F S C R T E L D L R P Q G L A L F S N V S E A R S L R T F D L P A T I P K L D T P D L L E V G T Q Q K L F C S L E G L F P A S E A R I Y L E L G G Q M P T Q E S T N S S D S V S A T A L V E V T E E F D R T L P L R C V L E L A D Q I L E T Q R T L T V Y N F S A P V L T L S Q L E V S E G S Q V T V K C E A H S G S K V V L L S G V E P R P P T P Q V Q F T L N A S S E D H K R S F F C S A A L E V A G K F L F K N Q T L E L H V L Y G P R L D E T D C L G N W T W Q E G S Q Q T L K C Q A W G N P S P K M T C R R K A D G A L L P I G V V K S V K Q E M N G T Y V C H A F S S H G N V T R N V Y L T V L Y H S Q N N S K G E E L F T G V V P I L V E L D G D V N G H K F S V R G E G E G D A T N G K L T L K F I C T T G K L P V P W P T L V T T L T Y G V Q C F S R Y P D H M K R H D F F K S A M P E G Y V Q E R T I S F K D D G T Y K T R A E V K F E G D T L V N R I E L K G I D F K E D G N I L G H K L E Y N F N S H N V Y I T A D K Q K N G I K A N F K I R H N V E D G S V Q L A D H Y Q Q N T P I G D G P V L L P D N H Y L S T Q S V L S K D P N E K R D H M V L L E F V T A A G I T H G M D E L Y K P K S C D K T H T C P P C P A P E L L G G P S V F L F P P K P K D T L M I S R T P E V T C V V V D V S H E D P E V K F N W Y V D G V E V H N A K T K P R E E Q Y N S T Y R V V S V L T V L H Q D W L N G K E Y K C K V S N K A L P A P I E K T I S K
where intensity is the surface fluorescent signal obtained with the same CCD camera setting as the lipid calibration experiment, and Isample(bilayer) is defined in Fig. S2D. Table S3. Oligonucleotide sequences used in this study Name Sequence (

Supplementary Note 1: Orientation of external force does not affect the rupture force of a surface anchored DNA duplex
To test whether the rupture force of a DNA duplex is sensitive to force orientation, we used the course grain modeling software package (75). oxDNA2 implements a DNA model with major and minor grooves, coarse-grained at the level of single nucleotides with interactions to mimic hydrogen-bonding, stacking, chain connectivity and excluded volume. This allows the model to reproduce the thermodynamic, structural, and mechanical properties of DNA that match experimental data.
The DNA duplex can be designed with geometries that tune the rupture force: the unzipping geometry and the shearing geometry. In unzipping geometry (estimated Ttol = ~12 pN, Fig. S18A), force is applied to the nucleotide at the same end of the nucleotide anchoring to the coverslip. While in shearing geometry (estimated Ttol = ~56 pN, Fig. S18B), force is applied to the nucleotide in an orientation that is parallel to the long axis of the double helix. We ran MD simulations on oxDNA to predict the rupture force of DNA duplexes under force loading from different orientations. For each configuration, we applied force along four different orientations [(x,y,z) = (1,0,0) , (0,0,1) , (1,1,1) , (0,-1,1)] at the center of mass of the nucleotide highlighted in pink while keeping the nucleobase on the opposite strand fixed at its initial position. These 4 force orientations represent the application of pure lateral force, pure vertical force and two randomly chosen force vectors. The system is evolved as per Newtonian mechanics for a given number of steps (newtonian_steps=103) after which particles below a certain threshold were randomly assigned velocities and momenta from a Maxwell distribution (dictated by the value diff_coeff=2.5). To emulate this Brownian dynamics, john thermostat in oxDNA was employed. Temperature and [Na + ] were set to 37 o C and 0.157M respectively to mimic in vitro experimental conditions. The configuration for force-extension was adopted from the protocol described by Engel and colleagues used for simulating DNA origamis under force (77).
For all the simulations, harmonic traps of stiffness of 11.40 pN/nm were placed at the nucleotides of interest (i.e. the pink nucleotide and the fixed nucleotide highlighted in Fig. S19). The effective trap stiffness can be calculated using the following equation: An extension rate of 7.03x10 3 nm/s was used to move the trap on the pink nucleotide along the indicated direction. A total of 2x10 9 MD steps were run for each simulation to generate 2x10 4 data points. To in silico generate a force-extension curve in each pulling direction, we extract the trap extensions from their corresponding nucleotides and then project it along the force axis. The force at a given point is caluclated by multiplying the total projected extensions with keff (5.71 pN/nm). Force is then plotted against the projected displacement of the two traps from their initial location along with a 100 point expontential moving average (EMA) of the data points using python (78). The rupture force was estimated by picking the peak at the point of rupture using SciPy find_peaks module (79). From the simulations, the Ttol of DNA duplex subjected to unzipping or shearing remains relatively constant in the four tested pulling orientations [(x,y,z) = (1,0,0) , (0,0,1) , (1,1,1) , (0,-1,1)] indicating that the direction of applied force has a minimal impact on the Ttol (Fig. S20).
The experimentally determined Ttol of the DNA duplex was derived from an early magnetic tweezer-based force clamp measurement by Prentiss and co-workers where the rupture force was determined by applying a constant force to shear the DNA duplex for a defined interval (e.g. 2 sec) and measuring the fraction of DNA duplex separated at that force (80). The constant force is incrementally increased and the Ttol is defined as the force that leads to 50% of duplexes dissociated within its application window, while the simulation experiments were performed by applying a defined loading velocity to the nucleotide. In a typical "force ramp" setup, the rupture force of a DNA duplex is highly dependent on the loading rate (i.e. the higher the loading rate, the higher the Ttol). It is also worth to note that oxDNA runs have an inherent level of stochasticity that results in small variations in force estimation between runs.
We also extracted the number of hydrogen bonds for the system throughout the course of the simulation. oxDNA considers hydrogen bonds as broken when the hydrogen bonding energy between a base pair is less than 10% of that of a fully formed hydrogen bond (Fig. S21).

Supplementary Note 2: Cofirming pY-ZAP70 intensity is a faithful readout of early T cell activation
As a confirmatory study, we performed immunostaining experiment of pY-ZAP70 using a 40x air objective and a 100x oil objective. The axial resolution of an objective can be estimated by the following formula: where λ is the wavelength of light and NA is the numerical aperture of the objective.
Considering our 100x objective has a NA of 1.49 and 40x objective has a NA of 0.65. The axial resolution for resolving a Alexa 647 molecule would be 595 nm for the 100x objective and 3,124 nm for the 40x objective. In other words, the 40x objective samples the entire thickness of the T cell and hence allows quantifying total pY-ZAP70 levels (81). In addition, we used the BF images to define the area occupied by the T cell. We used this BF-defined ROI to compare the average intensity versus integrated intensity of pY-ZAP70 of cells seeded on 12 pN or 56 pN N4 TGT surfaces (Fig. S13). We found the use of a lower magnification objective (i.e. 40x) and BF to define the cell area led to the same conclusions as those reported in the main text in Fig. 4 (i.e. 56 pN TGT substrates support higher ZAP70 phosphorylation of T cells when compared to 12 pN substrate).