Accumulation of Dynamic Catch Bonds between TCR and Agonist Peptide-MHC Triggers T Cell Signaling  Baoyu Liu, Wei Chen, Brian D. Evavold, Cheng Zhu  Cell  Volume 157, Issue 2, Pages 357-368 (April 2014) DOI: 10.1016/j.cell.2014.02.053 Copyright © 2014 Elsevier Inc. Terms and Conditions

Cell 2014 157, 357-368DOI: (10.1016/j.cell.2014.02.053) Copyright © 2014 Elsevier Inc. Terms and Conditions

Figure 1 In Situ Analysis of Force-Dependent TCR-pMHC Bond Kinetics by BFP (A) BFP schematic. A micropipette-aspirated RBC with a probe bead attached to the apex (left) was aligned against a T cell held by an apposing pipette (right). (B) BFP functionalization. The probe bead was covalently linked to SA to capture pMHC (left) to interact with TCR (right). (C–E) Representative force traces of measurement cycles showing adhesion that survived ramping and sustained a preset level of force until dissociation (marked by a red star), enabling bond lifetime measurement (C), adhesion ruptured by a ramp force (marked by a magenta star) before reaching the set force or in force-ramp assay (D), or no adhesion (E). (F) Binding specificity. Mean ± SEM of adhesion frequencies of >10 T cell-bead pairs with 50 contacts for each. Densities of pMHCs (ml) are indicated inside of each bar. N.D., not detected. Error bars represent SEM. See also Figure S1 and Movies S1, S2, and S3. Cell 2014 157, 357-368DOI: (10.1016/j.cell.2014.02.053) Copyright © 2014 Elsevier Inc. Terms and Conditions

Figure 2 TCR Forms an Agonist-Specific Catch-Slip Bond (A) Lifetimes of bonds of OT1 T cells with probe beads coated with indicated pMHCs at 0 (white) and 10 (black) pN. (B and C) Lifetime versus force curves showing that OT1 TCR formed catch-slip bonds with progressively weaker agonists OVA (red open square), A2 (light blue open triangle), and G4 (inverted magenta open triangle) (B) but slip-only bonds with antagonists R4 (blue open circle) and E1 (green open diamond) (C). Error bars in (A)–(C) represent SEM. (D) Force regulation of antigen discrimination, measured by the average lifetime ratio of TCR bonds with OVA to another peptide. See also Figure S2. Cell 2014 157, 357-368DOI: (10.1016/j.cell.2014.02.053) Copyright © 2014 Elsevier Inc. Terms and Conditions

Figure 3 Single-Cell Concurrent Measurement of Ca2+ Flux and In Situ TCR-pMHC Bond Kinetics (A and B) Representative wide-field pseudocolored images of two types of intracellular Ca2+ signals. (C and D) Representative time courses of relative fura-2 ratio of type α (C, magenta curve, left ordinate) or type β (D, green curve, left ordinate) intracellular Ca2+ signal synchronized with concurrent measurement of rupture (red open circle) and lifetime (blue solid triangle) events and the cumulative lifetime (blue dashed curve, right ordinate). Note that there are no rupture events for (C). A dashed-dot horizontal line indicates a 10 s threshold of cumulative lifetime. See also Figure S3 and Movies S4 and S5. Cell 2014 157, 357-368DOI: (10.1016/j.cell.2014.02.053) Copyright © 2014 Elsevier Inc. Terms and Conditions

Figure 4 TCR- and pMHC-Specific Ca2+ Flux Requires both Force and Lifetime (A) Ca2+ flux requires durable force on TCR bond applied by antigen pMHC. Percent increase of fura-2 ratio in OT1 T cells tested without force or lifetime via VSV (green open diamond, cf. Figure 1E), with force but no lifetime via OVA (red solid square, cf. Figure 1D), or with both force and lifetime (cf. Figure 1C) via OVA (red open square) or G4 (inverted light blue open triangle) or via anti-LFA-1 beads (orange open circle). (B and C) Ca2+ was triggered by an optimal force. Percent increase of fura-2 ratio (points, left ordinate) in OT1 T cells triggered by lifetimes (gray bars, mean ± SEM of >50 measurements, right ordinate) of TCR bonds with OVA (B) or anti-OT1 (C) measured at 0 (inverted magenta open triangle), 5 (green open circle), 10 (red open square), or 20 (blue open triangle) pN force. The asterisk (∗) denotes p < 0.01 on Ca2+ signals. Error bars represent SEM. Cell 2014 157, 357-368DOI: (10.1016/j.cell.2014.02.053) Copyright © 2014 Elsevier Inc. Terms and Conditions

Figure 5 Correlating Ca2+ Signals with Kinetic-Associated Statistics from Each Cell’s 600 s Event Sequence (A–F) Percent increase of fura-2 ratios versus number of adhesions Na (A), number of lifetimes Nlt (B), adhesion frequency Pa (C), average lifetime < t > (D), longest lifetime tmax (E), and cumulative lifetime Σti (F) of OT1 TCR bonds with OVA at 0 (brown solid square), 5 (orange open square), 10 (magenta open circle or green open triangle for type α or β Ca2+, respectively) or 20 (inverted open red triangle) pN, or with G4 at 0 (yellow solid diamond) or 10 (light blue open diamond) pN for each T cell calculated from the binding events in the entire 10 min experimental period. Dashed lines are linear fits to data. The Pearson coefficient of the correlation (R) is indicated in each subpanel. (G) Summary of Pearson coefficients. Cell 2014 157, 357-368DOI: (10.1016/j.cell.2014.02.053) Copyright © 2014 Elsevier Inc. Terms and Conditions

Figure 6 Ca2+ Best Correlates with TCR-pMHC Bond Lifetimes Accumulated in the First Minute of Successive Force Applications (A) Schematic of a sliding window (highlighted), starting at T0 with length TL and containing different binding events (red, no adhesion; green, rupture force; blue, lifetime). (B and C) Percent increase of fura-2 ratio versus cumulative lifetime of OT1 TCR bonds with OVA at 0 (brown solid square), 5 (orange open square), 10 (magenta open circle or green open triangle, for type α or β, respectively), or 20 (inverted open red triangle) pN, or with G4 at 0 (yellow solid diamond) or 10 (light blue open diamond) pN for each T cell accumulated in windows of indicated lengths and starting times. Dashed lines are linear fits to data, and the Pearson coefficients (R) are indicated. The horizontal and vertical dotted lines in the subpanel with the best correlation (highlighted) denote the demarcation of types α (magenta open circle) and β (brown solid square, orange open square, green open triangle, inverted open red triangle, yellow solid diamond, light blue open diamond) Ca2+ and the 10 s threshold of cumulative lifetime for triggering type α Ca2+. (D–F) Pearson coefficient analysis to search for the window for the kinetic parameters to achieve the best correlation with Ca2+. (D) Maximal Pearson coefficients Rmax versus TL of the initial window (T0 = 0) within which the kinetic parameters best correlate with Ca2+. (E) Pearson coefficient for cumulative lifetime versus sliding window size TL for the indicated starting times T0. (F) Pearson coefficients for indicated kinetic parameters calculated in a 60 s sliding window versus its starting time T0. Different symbols in (D) and (F) denote number of adhesions Na (purple open circle), number of lifetimes Nlt (magenta open square), adhesion frequency Pa (blue open triangle), average lifetime < t > (green solid square), longest lifetime tmax (orange solid circle), and cumulative lifetime Σti (inverted red solid triangle). See also Figures S4, S5, and S6. Cell 2014 157, 357-368DOI: (10.1016/j.cell.2014.02.053) Copyright © 2014 Elsevier Inc. Terms and Conditions

Figure 7 Long TCR-pMHC Bond Lifetime Accumulated after Many Short Ones Fails to Induce Ca2+ (A) Maximal cumulative lifetime max{Σti} of TCR bonds with OVA at 0 (brown solid square), 5 (orange open square), 10 (magenta open circle or green open triangle, for type α or β Ca2+, respectively), or 20 (inverted red open triangle) pN or with G4 at 0 (yellow solid diamond) or 10 pN (light blue open diamond) for each T cell accumulated in a 60 s window versus its starting time T0. (B) Percent increase of fura-2 ratio versus maximal cumulative lifetime max{Σti} calculated for each cell in its own 60 s window with starting time adjusted to allow Σti to achieve maximum. Demarcation of 50% increase of fura-2 ratio and threshold of 10 s cumulative lifetime (horizontal and vertical red dotted lines) are used to segregate cells into three groups. Group A cells accumulated >10 s lifetime in the initial 60 s window and generated type α Ca2+ (magenta open circle). Group B cells accumulated >10 s lifetime in later 60 s windows but generated type β Ca2+ (light blue open square). Group C cells accumulated <10 s lifetime and generated type β Ca2+ (green open triangle). Dashed lines in (A) and (B) are linear fits to data with R values indicated. (C–F) Normalized lifetime histogram (C and E) and number (left ordinate) and fraction (right ordinate) of short lifetimes (D and F) before Ca2+ peak pooled from cells in groups A (magenta) and B (cyan). (E and F) are similar to (C) and (D) except that lifetimes were collected from an initial 60 s window for group A; for group B, the starting time was adjusted for each cell to allow Σti to achieve maximum. Error bars in (D) and (F) represent SEM. See also Figure S7. Cell 2014 157, 357-368DOI: (10.1016/j.cell.2014.02.053) Copyright © 2014 Elsevier Inc. Terms and Conditions

Figure S1 Representative Raw Data of BFP Experiments and Controls, Related to Figure 1 (A) Mean ± SEM of adhesion frequencies between ∼10 pairs of beads and OT1 T cells contacting 50 times each at 25 or 37 °C. The beads were coated without (SA-only) or with OVA:H2-Kbα3A2 at the site densities ml indicated inside each bar adjusted to achieve low adhesion frequencies. The cells were loaded with or without fura-2. (B) Mean ± SEM of > 100 lifetimes of single-bonds between OT1 T cells and probe beads bearing OVA:H2-Kbα3A2 captured by monomeric or tetrameric SA measured at 10 pN by the lifetime assay (filled bars). For comparison, bond lifetimes of biotin-SA interaction are also shown (mean ± SEM of > 50 measurements [open bars]). ∗ = p < 0.01 by two-tailed Student t test. NS = Not Significant (p > 0.1). Error bars in (A) and (B) represent SEM. (C and D) Representative data of displacement versus time (C) and 100-point sliding standard deviation of displacement versus time (D) curves of the thermal fluctuation assay. This assay differs from the force-clamp assay for bond lifetime measurement in which the T cell retraction stopped at the zero force position (C) so bond formation is not detected by a tensile force (Figure 1C). Instead, it is detected by the reduction of the sliding standard deviation below a threshold (D). Further, bond dissociation is not detected by a sudden drop of the tensile force to zero (Figure 1C), but by the resumption of the sliding standard deviation above the threshold (D). Cell 2014 157, 357-368DOI: (10.1016/j.cell.2014.02.053) Copyright © 2014 Elsevier Inc. Terms and Conditions

Figure S2 Force-Dependent Lifetimes of TCR and LFA-1 with Their Respective Ligands, Related to Figure 2 (A–E) Distributions of lifetimes of OT1 TCR bonds with five pMHC ligands at representative forces. Plots of normalized ln(# of events with a lifetime > t) versus lifetime of OT1 TCR bonds with the indicated peptides – OVA (A), A2 (B), G4 (C), E1 (D), and R4 (E) – presented by H2-Kbα3A2 at the indicated forces. The catch (A and B), catch-slip (C), and slip (D and E) bonds can be identified from the changes in decay rate with increasing force in these plots because the steeper the decay, the faster the dissociation. (F) 2C TCR forms catch-slip bonds with an agonist (SIYR:H2-Kbα3A2). Data are shown as mean ± SEM of > 50 measurements. (G) LFA-1 forms a slip bond with its mAb. Instead of pMHC, a biotinylated mAb M17/4 captured on streptavidinylated beads was allowed to form bonds with LFA-1 (instead of TCR) on T cells whose lifetimes were measured by BFP with the force-clamp assay. Although the lifetimes of LFA-1–M17/4 bonds are much longer than those of the TCR–pMHC bonds, they decreased monotonically with increasing force. Data are shown as mean ± SEM of > 20 measurements. Error bars in (F) and (G) represent SEM. Cell 2014 157, 357-368DOI: (10.1016/j.cell.2014.02.053) Copyright © 2014 Elsevier Inc. Terms and Conditions

Figure S3 Schematic of Fluorescence Biomembrane Force Probe, Related to Figure 3 (A) Schematic of home-built fluorescence biomembrane force probe (fBFP). (B) Area under the calcium curve (AUC) correlates with the maximal percent increase of fura-2 ratio. Scatter plot of two measures of single-cell calcium signals showing correlation between AUC and maximal percent increase of fura-2 ratio from the initial value, which is fitted by a dashed line with the Pearson coefficient R indicated. Cell 2014 157, 357-368DOI: (10.1016/j.cell.2014.02.053) Copyright © 2014 Elsevier Inc. Terms and Conditions

Figure S4 Correlations of Ca2+ with Kinetic-Associated Statistics Calculated in Initial Windows of Different Lengths, Related to Figure 6 (A–E) Percent increase of fura-2 ratios versus the number of adhesions Na (A), the number of lifetimes Nlt (B), the adhesion frequency Pa (C), the average lifetime < t > (D), and the longest lifetime tmax (E) of OT1 TCR bonds with OVA:H2-Kbα3A2 at 0 (brown solid square), 5 (orange open square), 10 (magenta open circle or green open triangle, for type α or β Ca2+, respectively), or 20 (inverted red open triangle) pN, or with G4:H2-Kbα3A2 at 0 (yellow solid diamond) or 10 (light blue open diamond) pN within initial time (T0 = 0 s) windows of indicated length (TL = 120, 60, or 30 s). Dashed lines are linear fits to data. The goodness-of-fit is indicated by the Pearson coefficient (R) in each panel. Cell 2014 157, 357-368DOI: (10.1016/j.cell.2014.02.053) Copyright © 2014 Elsevier Inc. Terms and Conditions

Figure S5 Changes of Pearson Coefficients for Correlations between Ca2+ and Kinetic Parameters with Different Window Lengths and Starting Times, Related to Figure 6 (A–E) Each panel corresponds to one parameter – number of adhesions Na (A), number of lifetimes Nlt (B), adhesion frequency Pa (C), average lifetime < t > (D), and longest lifetime tmax (E). For all parameters, Pearson coefficient progressively decreases as windows slide from the beginning to the end of the 10 min experiment. Cell 2014 157, 357-368DOI: (10.1016/j.cell.2014.02.053) Copyright © 2014 Elsevier Inc. Terms and Conditions

Figure S6 Correlations of Ca2+ with Kinetic-Associated Statistics Calculated in 60 s Windows of Different Starting Times, Related to Figure 6 (A–E) Percent increase of fura-2 ratios versus the number of adhesions Na (A), the number of lifetimes Nlt (B), the adhesion frequency Pa (C), the average lifetime < t > (D), and the longest lifetime tmax (E) of OT1 TCR bonds with OVA:H2-Kbα3A2 at 0 (brown solid square), 5 (orange open square), 10 (magenta open circle or green open triangle, for type α or β Ca2+, respectively), or 20 (inverted red open triangle) pN, or with G4:H2-Kbα3A2 at 0 (yellow solid diamond) or 10 (light blue open diamond) pN within windows of the same size (TL = 60 s) but different starting time (T0 = 30, 60, 120 s). Dashed lines are linear fits to data. The goodness-of-fit is indicated by the Pearson coefficient (R) in each subpanel. Cell 2014 157, 357-368DOI: (10.1016/j.cell.2014.02.053) Copyright © 2014 Elsevier Inc. Terms and Conditions

Figure S7 Sliding Window Analysis Reveals No Correlation between Maximal Kinetic-Associated Statistics and Window Starting Times, Related to Figure 7 (A) Plot of cumulative lifetime within a window of fixed length (TL = 60 s) versus its starting time T0 for a representative T cell. Such plot identifies the maximal cumulative lifetime for the cell, which is max{Σti} = 16.9 s that occurred when the window starting point is T0 = 135 s. (B–F) No correlation between maximal kinetic-associated statistics from 60 s windows and their starting times. The maximal number of adhesions max{Na} (B), the maximal number of lifetimes max{Nlt} (C), the maximal adhesion frequency max{Pa} (D), the maximal average lifetime max{ < t > } (E), and the maximal longest lifetime max{tmax} (F) of OT1 TCR bonds with OVA:H2-Kb3A2 at 0 (brown solid square), 5 (orange open square), 10 (magenta open circle or green open triangle, for type or Ca2+, respectively), or 20 (inverted red open triangle) pN, or with G4:H2-Kb3A2 at 0 (yellow solid diamond) or 10 (light blue open diamond) pN calculated in 60 s windows versus the starting time (T0) when these parameters achieve maxima for each cell. Dashed lines are linear fits to data. The goodness-of-fit is indicated by the Pearson coefficient (R) in each panel. Cell 2014 157, 357-368DOI: (10.1016/j.cell.2014.02.053) Copyright © 2014 Elsevier Inc. Terms and Conditions