Occupancy of Lymphocyte LFA-1 by Surface-Immobilized ICAM-1 Is Critical for TCR- but Not for Chemokine-Triggered LFA-1 Conversion to an Open Headpiece High-Affinity State

The LFA-1 integrin is the best studied adhesion molecule involved in lymphocyte arrest on endothelial cells as well as on various APC targets (1). Recent in vivo imaging suggests a key role for T cell LFA-1 interaction with dendritic cell ICAM-1 in long-lasting adhesions (2). LFA-1 is maintained in a low-affinity nonadhesive state in motile lymphocytes prior to their encounter with cognate Ag (3). In vitro, TCR agonists can rapidly trigger LFA-1–mediated T cell arrest and spreading on ICAM-1–bearing surfaces, processes associated with rapid segregation of the TCR ligands and ICAM-1 into central and peripheral supramolecular activation cluster (pSMAC)-like zones, respectively (4, 5).

Structural studies and epitope mapping suggest that LFA-1 exists in at least three distinct conformational states: bent, unfolded, and a high-affinity extended state. Inactive LFA-1 is compact and bent (6, 7). Constraints on LFA-1 activation can be relieved by both cytoplasmic events (inside-out activation) or by ligand binding (outside-in activation) (8). Headpiece activation involves the swinging out of the β subunit hybrid domain, which pulls on the C-terminal α helix of the β I domain (9). Use of mAbs that probe β subunit extension or β I domain opening has provided key insights into LFA-1 activation by chemokine signals (10–12). T cell spreading on ICAM-1 could be mediated by low-, intermediate-, and high-affinity LFA-1–ICAM-1 bonds (11). TCR-induced T cell spreading on ICAM-1 has traditionally been suggested to involve LFA-1 clustering within large patches rather than conformational switches of LFA-1 from low- to high-affinity states (13). Thus, it was assumed that both intermediate- and high-affinity LFA-1–ICAM-1 bonds are in situ triggered by TCR signaling. TCR signals were postulated to also drive LFA-1 release from cytoskeletal constraints, thereby enhancing macroclustering of the integrin with ICAM-1, a process commonly termed LFA-1 avidity modulation (14).

TCR activation of T cells can be induced by its ligation with anti-CD3 mAbs, widely used polyclonal TCR agonists that drive integrin-mediated T cell arrest and spreading on cognate ligands (15). Although nonphysiological ligation of the TCR with an extensively crosslinked anti-CD3 mAb can activate the LFA-1 headpiece (16), similar effects of CD3 occupancy by isolated mAb molecules, known to induce potent TCR signaling (17–19), have not been reported. Furthermore, the redistribution of differentially activated LFA-1 triggered by these stimuli and their mechanisms of activation have not been elucidated to date. Of note, many studies on LFA-1 inside-out activation by TCR signals were performed with lymphoblasts on which both LFA-1 expression and regulation are different from that of primary T cells. We therefore induced LFA-1 activation on human freshly isolated T lymphocytes using different configurations of the polyclonal TCR agonist, the anti-CD3 mAb OKT3, and addressed whether, when, and where conformational LFA-1 switches and LFA-1 clustering events took place. Using specific mAb probes for LFA-1 extension and headpiece opening (i.e., activation) as well as an α/β I-like allosteric antagonist of LFA-1, we found that potent TCR signals on their own, in contrast to chemokine signals on their own, failed to extend or activate the LFA-1 headpiece. Interestingly, occupancy of TCR-stimulated LFA-1 by surface-immobilized ICAM-1 rather than by soluble ICAM-1 was critical for its conversion to a headpiece- activated state. This ICAM-1–driven LFA-1 activation took place within numerous submicron focal dots that remained evenly scattered underneath spread T cells rather than in the classic peripheral assemblies, pSMACs. These focal dots, rather than large focal zones of LFA-1 (20), seem to function as the critical adhesive units of TCR-stimulated T cell spreading on ICAM-1–bearing surfaces.

Informed consent was obtained from each individual studied. This study was approved by the Institutional Review Board of the Rambam Medical Center, consistent with the provisions of the Declaration of Helsinki. All animal procedures were approved by the Institutional Animal Care and Use Committee at the Weizmann Institute of Science.

Human ICAM-1-Fc, human VCAM-1-Fc, murine ICAM-Fc, CXCL12, and CCL21 were purchased from R&D Systems (Minneapolis, MN). IL-4 and GM-CSF were obtained from Cytolab (Rehovot, Israel). BSA (fraction V), Ca²⁺/Mg²⁺-free HBSS, PMA, jasplakinolide, anti-talin (8d4), tetramethylrhodamine isothiocyanate-phalloidin, the superantigen staphylococcus enterotoxin A, polyinosinic-polycytidylic acid, and LPS were purchased from Sigma-Aldrich (St. Louis, MO). The cell-permeable Rho-inhibiting P23–40 peptide and its control P1 (penetratin peptide) were synthesized as described (21). Human serum albumin (HSA, fraction V), protein A, 4-amino-5-(4-chlorophenyl)-7-(t-butyl)pyrazolo[3,4-d]pyramidine (PP2), blebbistatin, cytochalasin D, and bisindolylmaleimide I were purchased from Calbiochem (San Diego, CA). Neutralite avidin (deglycosylated neutral avidin) was purchased from Pierce (Rockford, IL). Secramine A, a gift of Dr. T. Kirchhausen (Harvard University, Cambridge, MA), was synthesized as described (22). XVA143 was a gift from P. Gillespie (Roche, Nutley, NJ). The high-affinity β₂ reporter 327C and the β₂-blocking mAb TS1/18 were gifts from D. Staunton (ICOS, Bothell, WA). Biotinylated 327C mAb was prepared as described (23). The KIM127 mAb was a gift of M. K. Robinson (UCB Celltech, Slough, U.K.). Alexa 568 Fab anti-murine IgG1 was prepared using the Zenon Alexa Fluor 568 mouse IgG1 labeleing kit (Invitrogen, Carlsbad, CA). The nonblocker anti–LFA-1 (α_L, TS2/4) was purified from American Type Culture Collection hybridomas and conjugated to Alexa Fluor 568 according to the manufacturer’s instructions (Invitrogen). CD3-activating mAb (OKT3) and Alexa Fluor 488-conjugated OKT3 mAbs were from BioLegend (San Diego, CA). The function-blocking anti-human ICAM-1 (HA58) was from Serotec (Raleigh, NC). Allophycocyanin-conjugated anti-human CD45 as well as anti-human CD28 (37.51) were from eBioscience (San Diego, CA). R-PE- and HRP-conjugated goat anti-mouse Abs were from Jackson ImmunoResearch Laboratories (West Grove, PA). Goat anti-mouse Alexa Fluor 546 and streptavidin-Alexa Fluor 488 or 568 secondary Abs were from Invitrogen.

Human peripheral blood (PB) T lymphocytes were isolated from citrate-anticoagulated whole blood from healthy donors by dextran sedimentation, density separation over Ficoll-Hypaque, and nylon wool column separation as described (24) and consisted of >90% CD3⁺ T lymphocytes. The resulting PB T cells (>90% CD3⁺ lymphocytes) were cultured in RPMI 1640/10% FCS for 15–18 h before experiments. CD45RA⁺ T cells were isolated using a negative cell isolation kit (MACS; Miltenyi Biotec, Bergisch Gladbach, Germany).

Mice were from The Jackson Laboratory (Bar Harbor, ME). Splenocytes, obtained from spleens of 6- to 9-wk-old OT-1 transgenic mice (25) were cultured for 16 h, and CD8⁺ T cells were isolated using a negative cell isolation kit (MACS; Miltenyi Biotec). The K^b-OVA and control peptides were prepared as described (26).

Where indicated, lymphocytes were pretreated with the inhibitors PP2 (10 μM, 30 min), GF-109203x (5 μM, 15 min), blebbistatin (50 μM, 30 min), BAPTA-AM (25 μM, 30 min), cytochalasin D (10 μM, 20 min), jasplakinolide (0.5 μM, 60 min), or DMSO at 37°C. For inhibition of Cdc42, T cells were pretreated with secramine A (5 μM, 15 min). Inhibitors were present in the medium throughout the experiments.

For detection of agonist-induced high-affinity LFA-1 conformations, T cells were suspended in binding medium and incubated with the β₂ reporter mAbs 327C or KIM127 (10 μg/ml), either intact or biotinylated, for 5 min at 37°C. Unwashed T cells were then stimulated with the TCR-ligating mAb OKT3 (10 μg/ml), chemokines (10 nM), PMA (100 ng/ml), or Mg²⁺/EGTA (2 mM each) for an additional 5 min at 37°C, and stimulation was stopped by placing the cells on ice. Lymphocytes were washed, resuspended with Alexa Fluor 568- or PE-conjugated streptavidin or secondary Abs for an additional 30 min at 4°C and analyzed by FACScan (BD Biosciences, Erembodegem, Belgium). For TCR crosslinking experiments, T cells were incubated with the TCR-ligating mAb OKT3 (10 μg/ml) for 10 min on ice, washed, and then crosslinked with goat anti-mIgG2a Abs (Jackson ImmunoResearch Laboratories) at 37°C for 5–20 min. For detection of soluble ICAM-1 binding, T cells were incubated with ICAM-1-Fc for 30 min at room temperature in binding medium. Cells were then washed twice and incubated with PE-donkey anti-human IgG (Jackson ImmunoResearch Laboratories) for 20 min at room temperature followed by FACS analysis.

T lymphocytes were either left intact or stimulated for 5 min with 10 μg/ml OKT3 at 37°C before fixation with 4% paraformaldehyde/2% sucrose and subjected to staining with the anti-α_L mAb (TS2/4-Alexa Fluor 568). Serial Z-stacked (0.2 μm/section) confocal imaging was performed with DeltaVision (Applied Precision, Issaquah, WA) with an oil 60×/1.4 PlanApo (differential interference contrast [DIC]) objective. Cell images were acquired as serial Z-stacks (0.2 μm apart) and subjected to digital deconvolution and three-dimensional reconstructions with the softWoRX software (Applied Precision). An LFA-1 cluster was defined according to size (>0.5 μm²) and fluorescence intensity (at least 2-fold greater than the cell mean fluorescence intensity [MFI]).

For immunofluorescent staining, unless otherwise indicated, cells were first fixed in PBS containing 4% (w/v) paraformaldehyde and 2% sucrose, extensively washed with PBS, and blocked with TBS (25 mM Tris [pH 7.4], 150 mM NaCl) supplemented with 2% HSA or serum. Cells were incubated with either Alexa Fluor- or biotin-labeled mAbs (45 min, 37°C) or with unlabeled mAbs followed by Alexa Fluor-conjugated secondary reagents. For double staining of intracellular molecules (e.g., talin), cells and surface proteins were first permeabilized with saponin (0.1% w/v, 5 min) and blocked with goat serum. Subsequently, cells were incubated with primary Ab, washed, and incubated with secondary Alexa Fluor-labeled Abs in the presence of 0.05% saponin. Mouse serum was added to neutralize the saponin. In some experiments, T cells were stained with a trace of Alexa Fluor-labeled TS2/4 mAb for 5 min, washed, and subjected to OKT3 stimulation as described above. The mAb was verified to not interfere with time course of lymphocyte spreading on ICAM-1, and the numbers and distribution of focal LFA-1 dots under spread T cells visualized by this method were indistinguishable from those visualized by TS2/4 labeling following fixation.

Intracellular (cytosolic) calcium was monitored by the calcium-sensitive dye Fluo-4 AM (Molecular Probes/Invitrogen) using either flow cytometry or real-time fluorescent microscopy.

PB T cells were preloaded with 2 μM Fluo-4 plus 0.1% (w/v) Pluronic F-127 in HBSS (HBSS containing 2 mg/ml BSA, 10 mM HEPES [pH 7.4], 1 M CaCl₂, and 1 M MgCl₂) and incubated for 15 min in the dark at 37°C. Samples were washed and resuspended in HBSS (cation-free and without BSA) and incubated for an additional 10 min at room temperature. Cells were then resuspended in HBSS/CaMg with 0.2% BSA and analyzed using the FACScan flow cytometer (BD Biosciences) with an argon laser with a fixed output wavelength of 488 nm. Cell samples were divided into equal volumes, and the first aliquot was aspirated for 200 s to determine the baseline fluorescence of the Fluo-4–Ca²⁺ complex. For the stimulations, the aspiration of the baseline sample was paused after 20 s, either OKT3 (10 μg/ml) or CCL21 (10 nM) was added, and the acquisition was resumed with changes in intracellular Ca²⁺ concentration being recorded over a 200-s period. The samples were analyzed using FloJo software (Tree Star, Ashland, OR). Changes in the fluorescence intensity of the Fluo-4–Ca²⁺ complex were measured on the fluorescence 1 channel (voltage 600), and Fluo-4-Ca²⁺ fluorescence was plotted as a geometric mean moving average versus time. Fluorescence intensity (MFI) per 30 s was evaluated.

Briefly, T lymphocytes were preloaded with Fluo-4 and Pluronic F-127 as for flow cytometry. Washed lymphocytes were incubated with either BAPTA-AM or loading buffer for 15 min, washed, and further incubated for 10 min at room temperature. Cells were then centrifuged, resuspended in HBSS, and mounted into a microslide chamber (μ-Slide VI flat or μ-Slide VI; Integrated BioDiagnostics, Munich, Germany) coated with 2 mg/ml HSA. The Fluo-4 signal was imaged with the FITC filter. For TCR stimulation, cells were stimulated with soluble anti-CD3 mAb (10 μg/ml) and immediately introduced into a microslide chamber. Time-lapse images were collected at 3-s intervals for 5 min. The data were analyzed using the Volocity software (Improvision/PerkinElmer, Waltham, MA).

To examine the encounter rate and duration between T lymphocytes and ICAM-1–coated beads, magnetic protein A beads (Dynabeads; Dynal/Invitrogen) were coated with the indicated concentrations of recombinant ICAM-1/Fc cell adhesion molecule site densities, assessed using ¹²⁵I-labeled anti–ICAM-1 (HA58), as previously described (10), and then fully blocked with saturating (100 μg/ml) concentrations of human IgG. T cells either left intact or stimulated with agonist (PMA, 100 ng/ml; OKT3 mAb, 10 μg/ml) were injected together with the beads into microslides (ibidi, Martinsried, Germany) and assayed at 37°C. T cell–bead encounters were recorded for 15 min at six frames per minute using a 40×/0.95 NA DIC objective. Encounters were defined as any T bead contact lasting >10 s, and a productive contact was defined as any cell–bead encounter lasting >30 s. More than 95% of these productive contacts remained stable for at least 5 min.

Human T cells were either left untreated or incubated with OKT3 (10 μg/ml) and immediately perfused into a flow chamber mounted on a polystyrene or a glass slide coated with human ICAM-1-Fc or human VCAM-1-Fc overlaid on immobilized protein A (20 μg/ml) and subsequently blocked with human IgG. In other assays, protein A plates were coated with either ICAM-1-Fc, VCAM-1-Fc, or IgG control (1.5 μg/ml) followed by rabbit anti-mouse Ab (10 μg/ml) on which OKT3 (0.2–0.4 μg/ml) was captured. CD8⁺ purified OT-I splenocytes were allowed to spread on a polystyrene plate coated with murine ICAM-Fc overlaid on protein A coimmobilized with neutralite avidin. The MHC-bound OVA peptide fragment SIINFEKL (K^b-OVA) or irrelevant MHC-peptide were each biotinylated to enable their binding to the immobilized avidin. Human or murine cells were each settled on the substrate, and their spreading was monitored by videomicroscopy for 10–60 min. All spreading assays were performed at 37°C, and the number of cells spread was calculated as the percentage of initially settled cells on the purified substrate. Lymphocyte morphology (circumference, area, darkening) was determined using the Nikon NIS-Elements D 3.0 software (Nikon Instruments, Melville, NY). T cells that underwent both darkening for at least 3 min accompanied by an at least 40% increase in original cell circumference and a 2-fold increase in area were considered spread.

Fluorescence microscopy was carried out with the DeltaVision system (Applied Precision) using an oil 60×/1.4 PlanApo (DIC) objective. All fixed cell images were acquired as serial Z-stacks (0.2 μm apart) and subjected to digital deconvolution and three-dimensional reconstructions using the softWoRX software (Applied Precision). Spectrum analysis of relative fluorescence intensity was performed using modified Resolve3D and Prism software as described (27).

Liposomes containing biotinylated phosphatidylethanolamine with a caproyl spacer, 1,2-dioleoyl-sn-glycero-3-{[N(5-amino-1-carboxypentyl) iminodiacetic acid] succinyl}, and 1,2-dioleoyl-sn-glycero-3-phosphocholine lipids were deposited on clean glass coverslips in a parallel plate flow chamber (Bioptechs, Butler, PA), and a planar lipid bilayer was created on the substrate in HEPES/saline buffer as described (28). Then, 6-histidine–tagged Cy5-labeled ICAM-1 was conjugated to the bilayer in the presence of 100 μM NiCl₂. Streptavidin and biotinylated anti-CD3 mAbs were flowed sequentially on the biotin-PE–containing bilayer. The site density of the ICAM-1 ranged between 50 and 200 sites/mm². Freshly isolated human T lymphocytes were recorded at a rate of 1 frame per 8 s immediately upon perfusion into the chamber using an Olympus FluoView 1000 confocal microscope. T cells were considered spread if their original circumference was increased by 30% and their area was increased by 1.7-fold.

All data are reported as mean values ± SD and were analyzed by a two-tailed Student t test with equal sample variance. Analyses were performed using the statistics tool of Excel. Data sets were considered significantly different at p < 0.05.

To assess whether TCR signals induce inside-out conformational changes in the LFA-1 heterodimer, we exposed human resting PB T cells to saturating levels of the anti-CD3 mAb, OKT3, a prototypic TCR agonist (15, 29). T cells were coincubated with two fluorescently labeled reporter mAbs that detect conformational switches in the β₂ subunit of the LFA-1. Because T cells lack β₂ subunit-containing integrins other than LFA-1, these reporter mAbs are specific for LFA-1 extension and headpiece activation (10, 30). One of these mAbs, KIM127, detects an epitope on the β I-epidermal growth factor-2 domain, masked in the bent integrin (7) (Supplemental Fig. 1) and can therefore differentiate between bent and extended LFA-1 (30). The second mAb, 327C, has been used to probe the opening of the β I domain on LFA-1, a key headpiece rearrangement that stabilizes LFA-1 at a state favorable for high-affinity ICAM-1 binding (10, 12) (Supplemental Fig. 1). Strikingly, TCR ligation by a dimeric anti-CD3 mAb, OKT3, failed to trigger either LFA-1 extension or headpiece activation induced by the opening of the headpiece β I domain (Fig. 1A) despite high flux of Ca²⁺ triggered by this treatment (Fig. 1B). Notably, because the mAb reporters were present throughout the assay, any transient LFA-1 conformational switches would have been detected under these conditions. In contrast, brief T cell activation by phorbol esters or prototypic chemokine signals readily triggered both β subunit extension and β I domain activation in LFA-1 (Fig. 1A). In light of the lack of LFA-1 conformational switch by this TCR stimulus, we considered that CD3 ligation by a dimerizing mAb may instead patch LFA-1. Nevertheless, this CD3 ligation did not enhance any global LFA-1 macroclustering in resting human lymphocytes (Fig. 1C). Thus, in contrast to chemokine signals, TCR ligation failed on its own to trigger LFA-1 extension, headpiece opening, or macroclustering.

In accordance with the inability of TCR signals to switch LFA-1 into a headpiece-activated state, a soluble ICAM-1 dimer failed to stably bind TCR-stimulated T cells (Supplemental Fig. 2). To detect any short-lived conformational switches of LFA-1, we next incubated OKT3-stimulated T cells with the mAb reporter of open LFA-1 headpiece (i.e., 327C) and left the reporter mAb throughout the assay together with soluble ICAM-1 dimer, at a concentration known to saturate high-affinity LFA-1. Even under these conditions, which allow the integrin reporter mAb to irreversibly bind any TCR-stimulated LFA-1 transiently switched into its high-affinity conformation by ICAM-1, the potent TCR stimulus failed to shift the LFA-1 headpiece into the open conformational state (Fig. 2A). We next reasoned that the potent TCR signals triggered by OKT3 are insufficient to conformationally switch the LFA-1 headpiece into its open state stabilizing a high-affinity conformer in the presence of soluble ICAM-1. Nevertheless, we hypothesized that this very TCR stimulus may switch the LFA-1 headpiece into its open state in the presence of surface-immobilized ICAM-1. We therefore developed a highly sensitive videomicroscopy-based assay that monitors at high temporal resolution the earliest binding of T cells to ICAM-1–coated microbeads in the presence of minimal detachment forces (Fig. 2B, Supplemental Fig. 3, Supplemental Movies 1–3). TCR stimulation of resting T cells indeed promoted significant LFA-1 binding to ICAM-1 beads (Fig. 2C). To test whether this TCR-induced LFA-1 binding to ICAM-1 beads involves high- or intermediate-affinity LFA-1–ICAM-1 bonds (11), we used an α/β I allosteric inhibitor, XVA143, which discriminates between high- and low/intermediate-LFA-1–ICAM-1 interactions (30, 32). This inhibitor competes with the binding of the intrinsic ligand on the α I domain to the β I domain (Fig. 2D), disrupts communication between the β and α headpiece domains, and thereby restricts full opening of the α I domain and stabilization of high-affinity LFA-1 binding of ICAM-1 (30). Instead, the XVA143-occupied LFA-1 remains at extended conformation with its α I domain binding ICAM-1 via low/intermediate-affinity bonds (30, 32). In contrast to allosteric α I domain inhibitors, which act as direct blockers of all LFA-1–dependent interactions (10, 16), XVA143 is unique in that it acts as a partial gain-of-function modulator that unfolds bent LFA-1 and allows it to mediate weak-intermediate strength rolling adhesions (30). Surprisingly, in contrast to its proadhesive affects on LFA-1–mediated rolling, XVA143 totally eliminated the ability of PMA-stimulated T cell LFA-1 to bind ICAM-1 (Fig. 2C). Interestingly, all adhesive contacts between TCR-stimulated LFA-1 and surface-bound ICAM-1 were also completely blocked by XVA143 at both low and high ICAM-1 site densities (Fig. 2C). Thus, when TCR-stimulated LFA-1 was locked at its extended low/intermediate-affinity state (Fig. 2D), it completely failed to bind ICAM-1–coated beads. These results suggested that TCR-stimulated LFA-1 can acquire high-affinity headpiece conformation only in the presence of surface-bound ICAM-1 (Fig. 2D). Interestingly, CD28 coligation did not further stimulate TCR-stimulated LFA-1 adhesiveness to ICAM-1–coated beads (Supplemental Fig. 4A). Furthermore, TCR-stimulated LFA-1 binding to ICAM-1 beads was observed to similar extents with both naive and total CD3 T lymphocytes (Supplemental Fig. 4B). Notably, as ICAM-1 bead binding took place in the absence of any stirring, TCR stimulation of LFA-1 adhesiveness did not appear to require application of external forces, in sharp contrast to chemokine-mediated stimulation of LFA-1 (33). Interestingly, although CCL21 and CXCL12 induced potent extension and headpiece activation of LFA-1 (Fig. 1A), the LFA-1 on T cells stimulated by these chemokines failed to bind ICAM-1–coated beads (33). As expected, TCR-stimulated LFA-1 binding to ICAM-1–coated beads was entirely Src-dependent, as it was fully blocked by the Src inhibitor PP2 (Fig. 2E). Nevertheless, LFA-1 activation by TCR signals was insensitive to the 1,2-diacylglycerol (DAG)-dependent protein kinase C (PKC) inhibitor bisindolylmaleimide (Fig. 2E), ruling out a role for DAG-regulated PKCs in early LFA-1 activation by TCR signals.

We next assessed the role of this LFA-1 activation during various CD3/TCR-stimulated T cell spreading on ICAM-1, a more complex process downstream to the early LFA-1–ICAM-1 binding events probed by the ICAM-1 bead assay. As expected, the saturating TCR ligation protocol used in previous sections to probe LFA-1 conformation and ICAM-1 bead binding triggered robust and rapid T cell spreading on both medium- and high-density ICAM-1 (Fig. 3A, 3B, Supplemental Movies 4, 5). Interestingly, a much higher fraction of TCR-stimulated PB T cells could spread on ICAM-1 than bind ICAM-1 beads coated at similar site densities (Figs. 2C, 3B). Thus, active spreading allowed more TCR-stimulated LFA-1 to adhere to surface-coated ICAM-1. Nevertheless, the α/β I allosteric blocker, XVA143, still abolished all TCR-stimulated T cell spreading at both low and high ICAM-1 densities (Fig. 3B). Similar to the ICAM-1 bead assay, ligation of CD28 did not result in any acceleration of CD3/TCR-stimulated T cell spreading on ICAM-1 even at subsaturating levels of TCR ligation (Fig. 3C). TCR-stimulated LFA-1–mediated spreading was entirely Src-dependent, but it did not involve DAG-dependent PKCs or PI3K signaling, as was observed for TCR-stimulated LFA-1 binding to ICAM-1 beads (Figs. 2E, 3D and data not shown). Interestingly, LFA-1–mediated spreading also did not require cytosolic-free Ca²⁺, as both its magnitude and kinetics were insensitive to the cell-permeable Ca²⁺chelator BAPTA-AM (Fig. 3E, Supplemental Fig. 5A). This chelator fully abolished, however, the CD3/TCR-stimulated cytosolic-free Ca²⁺ signal (Supplemental Fig. 5B), as well as LFA-1 adhesiveness triggered by ionomycin (Supplemental Fig. 5C). Thus, both TCR-stimulated LFA-1 binding to ICAM-1 and T cell spreading on ICAM-1 do not require cytosolic-free Ca²⁺ or DAG-dependent PKCs and are not facilitated by coligation of T cell CD28.

Both intermediate- and high-affinity LFA-1–ICAM-1 bonds were suggested to promote T cell spreading on ICAM-1 (11). We next assessed whether physical proximity between the TCR stimulus and ICAM-1 could trigger numerous LFA-1–ICAM-1 interactions with low/intermediate- rather than high-affinity properties. Immobilized anti-CD3 mAb, although a weak stimulus for T cell spreading on its own, dramatically stimulated T cell spreading to both low- and high-density ICAM-1 (Fig. 4A). Nevertheless, the α/β I allosteric blocker, XVA143, which stabilized extended low- to intermediate-affinity LFA-1 conformations, eliminated all ICAM-1–dependent spreading stimulated by immobilized anti-CD3 mAb (Fig. 4A). Similarly, when fresh spleen-derived OT-I transgenic CD8⁺ T cells, specific for the OVA peptide 257–264 (15), were triggered by their cognate MHC–Ag complex to spread on coimmobilized ICAM-1 (Fig. 4B), this Ag-stimulated spreading was totally abrogated in the presence of XVA143 (Fig. 4B). Thus, extended intermediate-affinity LFA-1 conformers primed by TCR signals in murine T cells and artificially stabilized by the XVA143 allosteric inhibitor failed to support murine T cell spreading on ICAM-1. LFA-1 on TCR-stimulated T cells must therefore rearrange into its high-affinity ICAM-1–stabilized state to mediate effective spreading of both human and murine T cells on ICAM-1.

To gain further insights into the dynamics of TCR-triggered LFA-1 activation, we next analyzed LFA-1–mediated T cell spreading and occupancy of fluorescently tagged ICAM-1 embedded in a lipid bilayer together with a CD3-ligating mAb, both of which retain high mobility within the bilayer. Similar to T cell spreading on immobile ICAM-1 (Fig. 4A), freshly isolated human PB T cells encountering mobile ICAM-1 underwent very rapid spreading in response to CD3 ligation (Fig. 4C). Importantly, T cell spreading was associated with highly dynamic ICAM-1 microclusters that were entirely blocked by XVA143, suggesting that lymphocyte spreading in this system is mediated by high-affinity rather than by low/intermediate-affinity LFA-1–ICAM-1 bonds (Fig. 4C, 4D, Supplemental Movies 6–9). Interestingly, most LFA-1–driven ICAM-1 microclusters took place under the entire contact area of T cells with the lipid bilayer without bias toward a peripheral zone even after 10 min of spreading. Occasionally, during early phases of lymphocyte spreading, a fraction of ICAM-1 microclusters rearranged in short-lived pSMAC-like assemblies (Supplemental Movies 6, 7). These results suggest that microclusters of high-affinity LFA-1–ICAM-1 bonds continuously form and disassemble under the entire ventral side of lymphocytes spread on ICAM-1, regardless of whether ICAM-1 is mobile or immobile within the adhesive contact.

To directly label the TCR-stimulated headpiece-activated LFA-1, we next analyzed the distribution of the 327C β₂ open headpiece epitope during TCR-triggered T cell spreading on ICAM-1. Consistent with the evenly scattered pattern of ICAM-1 microclusters and their tight dependence on the XVA143 blocker (Fig. 4A), headpiece-activated LFA-1 conformers were detected within numerous submicron ventral dots underneath all spread T cells (>50 dots/cell; Fig. 4E), well segregated from TCR microclusters (Supplemental Fig. 6). Both the number and distribution of these dots were insensitive to chelation of cytosolic-free Ca²⁺, in support of the independence of LFA-1–mediated spreading from cytosolic Ca²⁺ (Fig. 4F). Identical insensitivity to chelation of cytosolic-free Ca²⁺ was also observed in Ag-stimulated LFA-1–mediated spreading of OT-I transgenic CD8⁺ T cells on ICAM-1 (Fig. 4B, right inset). Furthermore, this Ag-stimulated spreading was also insensitive to inhibition of DAG-dependent PKCs (not shown), as was observed in CD3/TCR-stimulated human PB T cells (Fig. 3D). Notably, no LFA-1 headpiece activation probed by the 327C reporter of the open β Ι domain could be detected in TCR-stimulated T cells spread on VCAM (Fig. 4G) or on high-density immobilized anti-CD3 mAb in the absence of ICAM-1 (Fig. 4G,ii). Since under these conditions the TCRs are assumed to undergo extensive crosslinking (34), these results suggested that the failure of LFA-1 to undergo conformational activation is not due to insufficient TCR crosslinking. Indeed, even when T cells were globally ligated with soluble OKT3 and then subjected to extensive crosslinking by secondary Abs, neither KIM127 or 327C epitopes were triggered within the time frames of our ICAM-1 bead binding and spreading assays (Figs. 2C, 3, 4H and data not shown). At prolonged incubation, however, significantly longer than these time frames, extensive crosslinking could trigger conformational LFA-1 activation in a subset of T cells (Supplemental Fig. 7), consistent with published data (16), whereas TCR ligation alone was still insufficient to trigger any detectable conformational activation (Supplemental Fig. 7). Interestingly, LFA-1 staining with TS2/4, a pan anti-α_L mAb that recognizes LFA-1 irrespective of its conformational states, was enriched within the headpiece-activated LFA-1 localized to the scattered ventral dots generated by TCR-stimulated T cells spread on ICAM-1 (Fig. 5A,i). In contrast, LFA-1 stained by TS2/4 remained peripheral on TCR-stimulated T cells spread on VCAM (Supplemental Fig. 8). Thus, in the absence of ICAM-1, any TCR-primed LFA-1 fails to acquire functionally adhesive conformation defined by three criteria: β I conformational activation (induction of the 327C epitope), susceptibility to XVA143, and microclustering within ventral focal dots. Notably, each microcluster of TCR-primed ICAM-1–rearranged LFA-1 localized within focal ventral dots was proximal to talin, visualized by intracellular staining (Fig. 5A,ii), consistent with the role of this adaptor in TCR-stimulated LFA-1 adhesiveness (29, 35). Nevertheless, only a small fraction of high-affinity LFA-1 dots colocalized with F-actin (Fig. 5A,iii). Instead, F-actin was mainly enriched in peripheral membrane ruffles, together with CD45 (Fig. 5A iv).

Actomyosin forces have been recently suggested to stabilize ligand-occupied integrins at high-affinity states (36). Interestingly, inhibition of myosin II, critical for TCR-transduced contractile forces (37), lymphocyte motility (33), as well as LFA-1 deadhesion (32), did not affect rate or magnitude of TCR-triggered T cell spreading on ICAM-1 (Fig. 5B). Similarly, the main effector region of RhoA, an upstream GTPase regulator of myosin II implicated in T cell motility on CCL21 (Supplemental Fig. 9A), cell protrusion (38), and LFA-1 activation by chemokines (10–12), was dispensable for TCR-triggered spreading on ICAM-1 (Fig. 5C). Inhibition of Cdc42, a key GTPase involved in cell protrusion, also did not affect the magnitude or kinetics of CD3/TCR-stimulated T cell spreading on ICAM-1 (Supplemental Fig. 9B). T cell spreading, however, required active actin turnover as it was completely eliminated by both cytochalasin D, an actin polymerization inhibitor, and by jasplakinolide, an actin-stabilizing drug (Supplemental Fig. 9C). Surprisingly, jasplakinolide-pretreated, but not cystochalasin D-pretreated, T cells were still able to mount TCR-mediated stimulation of LFA-1 adhesiveness (Supplemental Fig. 9Cii). Thus, actin polymerization rather than RhoA- or myosin II-driven contractile forces is the major driving element in TCR- stimulated, high-affinity LFA-1–mediated T cell spreading on ICAM-1, consistent with the suggested key role of retrograde actin flow in TCR-stimulated LFA-1–mediated T cell spreading on ICAM-1 (39) (Fig. 6). Taken together, these results suggest that focal dots of TCR-primed LFA-1 bidirectionally stabilized by ICAM-1 and F-actin are the quantal adhesive units of T cells undergoing spreading on APCs during early phases of immune synapse formation.

LFA-1 avidity modulation by TCR signaling is a key checkpoint in T cell arrest on APCs (40). Previous findings have suggested that TCR-mediated stoppage of T cells on ICAM-1 is initiated by transient microclusters of unstable LFA-1, which is rapidly mobilized and collapses into a peripheral zone termed the pSMAC (41). These earlier studies (5, 39) alluded to the possibility that firm TCR-triggered LFA-1–mediated adhesion requires LFA-1 segregation into this large assembly. It was also largely accepted that LFA-1 within both the early microclusters and the subsequent large pSMAC must undergo various conformational transitions between bent inactive, extended intermediate-, and high-affinity states (42). Nevertheless, the molecular nature of these conformational transitions has not been discerned or spatially monitored in any of these previous studies. Furthermore, it was hypothesized that the earlier TCR-driven LFA-1–ICAM-1 microclusters may consist mainly of short-lived intermediate-affinity bonds. Using specific probes together with an α/β I allosteric inhibitor, we demonstrate that various potent TCR signals trigger a switch in LFA-1 conformation that is readily stabilized by surface-bound ICAM-1. One of the most surprising results of this study is that unlike chemokine-stimulated LFA-1 (10, 12), TCR-stimulated LFA-1 does not get extended on its own and does not acquire open headpiece conformation unless it is properly rearranged by surface-bound ICAM-1 (Fig. 6). Furthermore, we do not find any evidence that the extended low/intermediate-affinity LFA-1 state, previously suggested to promote T cell protrusions on ICAM-1 (11), is sufficient for either the early or the later adhesive assemblies underlying TCR-triggered T cell spreading on ICAM-1. Indeed, when we locked LFA-1 in an extended low-affinity LFA-1 conformation using the α/β I allosteric inhibitor XVA143 (30), we could not detect early LFA-1–dependent T cell adhesion to ICAM-1–coated beads or T cell spreading on either low- or high-density ICAM-1. These findings sharply contrast with results reported in K562 cells, where XVA143 not only extended LFA-1 but stabilized extended low- or intermediate-affinity bonds with ICAM-1, which mediated LFA-1–driven rolling interactions of these cells on ICAM-1 (30). Thus, as opposed to XVA143-occupied LFA-1 on K562, XVA143-occupied LFA-1 on TCR-stimulated T cells can no longer adhere to ICAM-1 under any experimental conditions tested.

TCR-stimulated T cells generate, at their ventral side engaged with ICAM-1, numerous short-lived microclusters enriched with open headpiece LFA-1. These evenly scattered ICAM-1–LFA-1 dots were observed at both early (5–10 min) and late (15–60 min) periods of T cell spreading on either a lipid-embedded mobile form of ICAM-1 or an immobile form of ICAM-1. Based on their total elimination in the presence of the XVA143 inhibitor, we suggest that ICAM-1–stabilized high-affinity LFA-1 may favorably undergo microclustering, as previously observed in K562 cells (43). Indeed, ICAM-1–stabilized microclusters of headpiece-activated (327C-positive) LFA-1 are readily observed within evenly scattered submicron focal dots, during both early and late phases of T cell spreading on ICAM-1. These focal dots are reminiscent of focal adhesions and sites of force-regulated integrin unbending (36, 44). ICAM-1 occupancy of LFA-1 may not only activate the LFA-1 headpiece but may also restrict LFA-1 mobility within microclusters (8, 43). An intriguing possibility is that immobile ICAM-1 may first trap and stabilize the small pool of extended headpiece-activated LFA-1 already expressed by resting T cells. These postulated initiators of LFA-1 nascent adhesions may then nucleate additional pools of LFA-1 molecules to focal sites enriched with ICAM-1. Our data strongly suggest that these pools cannot shift into the high-affinity states in the presence of physiological TCR signals on their own but require an ICAM-1–triggered outside-in headpiece conformational switch. In contrast, LFA-1 in the presence of ICAM-1 alone without TCR signals that alter its cytoskeletal associations prior to its occupancy by ICAM-1 and immediately thereafter would fail to properly link to the actin cytoskeleton and thus be unable to undergo this critical outside-in switch and carry out its adhesive functions. Notably, although highly adhesive, these nascent adhesions, as well as the subsequent assemblies they nucleate, rapidly turn over during TCR-triggered T cell spreading and do not readily collapse into a peripheral ring or pSMAC. Taken together, these data imply that pSMAC ring-like structures may function as terminal adhesive assemblies of T cell-APC synapse rather than as the key adhesive elements of synapse formation (4, 39, 45, 46).

The hallmark of earliest TCR signaling is a rise in cytosolic Ca²⁺, and this secondary messenger was proposed to link TCR signaling to LFA-1 avidity modulation by releasing LFA-1 from cytoskeletal constraints (13). Our present results highlight critical roles for ICAM-1 driven LFA-1 conformational switching and anchorage within the cortical actin cytoskeleton rather than for LFA-1 release from this cytoskeleton in TCR-mediated LFA-1 adhesiveness. These results call into question the role of Ca²⁺-triggered LFA-1 release from cytoskeletal constraints (13), at least in early phases of immune synapse formation. The role of this key secondary messenger and its numerous potential targets in LFA-1 activation and redistribution in the immunological synapse has been indeed disputed (45, 47). TCR activation in lymphoblasts was shown to activate the Ca²⁺-dependent protease calpain, resulting in enhanced LFA-1 proteolytic cleavage of LFA-1–cytoskeletal linkages and accelerated T cell adhesion to ICAM-1 (13). This mechanism was never confirmed, however, in primary lymphocytes. Our work on these lymphocytes suggests that blocking all cytosolic Ca²⁺ and therefore inhibiting calpain activation by stimulated TCR does not interfere with TCR-triggered LFA-1–mediated T cell spreading on ICAM-1. Recently, TCR-triggered arrest of transgenic T blasts migrating through ICAM-1–coated filters was also shown to be partially inhibited by BAPTA-AM (47); however, in another study using T blasts, inhibition of cytosolic-free Ca²⁺ did not interfere with Ag-triggered arrest (45). Our work indicates that not only does blocking all cytosolic Ca²⁺ not attenuate TCR stimulated LFA-1–mediated T cell arrest and spreading on ICAM-1, but it does not affect the induction, stability, or distribution of high-affinity LFA-1 focal dots. Thus, although an artificial increase in cytosolic Ca²⁺ triggers LFA-1 adhesiveness to ICAM-1, the dramatic rise in cytosolic Ca²⁺ triggered by TCR ligation is not required for any of the LFA-1–ICAM-1 adhesive units we have defined as critical for T cell arrest and spreading.

In conclusion, our study suggests a previously unappreciated role for ICAM-1 in LFA-1 outside-in headpiece conformational activation event primed by TCR signaling. This conformational switch is critical for high-affinity LFA-1–ICAM-1 bond formation underlying TCR-triggered T cell arrest and spreading on ICAM-1. Notably, the focal contacts in which this outside-in conformational activation takes place form even in the absence of external forces or internal myosin II-driven contractile forces. This sharply contrasts the dependence of chemokine stimulation of LFA-1 on external forces (33). We propose that the binding of surface ICAM-1 to T cell LFA-1 has a triple role: it drives an outside-in switch of the LFA-1 into open high-affinity conformation; it drives the microclustering of ICAM-1–LFA-1 bonds (8); and it anchors LFA-1 to the cortical actin cytoskeleton (48), possibly via talin1, and thereby facilitates the outside-in switch via mechanical activation (36). LFA-1 clustering, in addition to stabilizing firm multifocal adhesions of TCR-stimulated T cells spread on ICAM-1–bearing surfaces (43), is likely to trigger outside-in Src signals (49) that further amplify the Src activation signals transduced by the TCR machinery. Our results also predict that T cells use actin-driven rather than myosin contractile forces to locally activate their ICAM-1–occupied LFA-1 with submicron microclusters. Although excluding a major role for the RhoA and Cdc42 GTPases in this actin-dependent process, our results elude to a major role for TCR-triggered Rac GTPase activities in T cell spreading on ICAM-1 (50). Our results are also consistent with recent studies that demonstrate that the arrest of T cells on ICAM-1 presented by APCs is proportional to both the magnitude of productive signals delivered by MHC–Ag complexes and the local surface density of the APC ICAM-1 (2). Future studies are necessary to delineate the active role of ICAM-1 in mechanochemical LFA-1 activation by TCR signals (51). It would also be interesting to address how ICAM-1 distribution and anchorage states on different APCs may affect the efficiency by which this key adhesive ligand translates signals from a given MHC–Ag complex to LFA-1 headpiece activation, microclustering, and macroclustering within different types of immune synapses.

We thank G. Shakhar for helpful discussions and S. Schwarzbaum for editorial assistance. We also thank T. Kirchhausen for the gift of secramine A. This manuscript is dedicated to the loving memory of Dr. Valentin Grabovsky, who passed away while the article was in press.

Disclosures The authors have no financial conflicts of interest.