Leukocyte function-associated antigen 1-mediated adhesion stability is dynamically regulated through affinity and valency during bond formation with intercellular adhesion molecule-1.

Neutrophil rolling and transition to arrest on inflamed endothelium are dynamically regulated by the affinity of the beta(2) integrin CD11a/CD18 (leukocyte function associated antigen 1 (LFA-1)) for binding intercellular adhesion molecule (ICAM)-1. Conformational shifts are thought to regulate molecular affinity and adhesion stability. Also critical to adhesion efficiency is membrane redistribution of active LFA-1 into dense submicron clusters where multimeric interactions occur. We examined the influences of affinity and dimerization of LFA-1 on LFA-1/ICAM-1 binding by engineering a cell-free model in which two recombinant LFA-1 heterodimers are bound to respective Fab domains of an antibody attached to latex microspheres. Binding of monomeric and dimeric ICAM-1 to dimeric LFA-1 was measured in real time by fluorescence flow cytometry. ICAM-1 dissociation kinetics were measured while LFA-1 affinity was dynamically shifted by the addition of allosteric small molecules. High affinity LFA-1 dissociated 10-fold faster when bound to monomeric compared with dimeric ICAM-1, corresponding to bond lifetimes of 25 and 330 s, respectively. Downshifting LFA-1 into an intermediate affinity state with the small molecule I domain allosteric inhibitor IC487475 decreased the difference in dissociation rates between monomeric and dimeric ICAM-1 to 4-fold. When LFA-1 was shifted into the low affinity state by lovastatin, both monomeric and dimeric ICAM-1 dissociated in less than 1 s, and the dissociation rates were within 50% of each other. These data reveal the respective importance of LFA-1 affinity and proximity in tuning bond lifetime with ICAM-1 and demonstrate a nonlinear increase in the bond lifetime of the dimer versus the monomer at higher affinity.

Neutrophils circulate in the bloodstream to sites of inflammation where they adhere and transmigrate through the endothelium as the initial step in combating infection and to facilitate wound healing. Recruitment from the circulation involves a multistep process of cell rolling, activation, and arrest. The heterodimeric integrin receptor LFA-1 1 is composed of the ␣L (CD11a) and ␤ 2 (CD18) subunits and is constitutively expressed in a low affinity conformation on the plasma membrane of leukocytes (1)(2)(3). Neutrophils encountering chemokines on inflamed endothelium are activated to shift LFA-1 from the low to high affinity conformation, which supports tight binding to endothelial ICAM-1. Increases in integrin affinity correlate in time with adhesion function as recently demonstrated in aggregation of cells expressing ␣4␤1 and vascular cell adhesion molecule (4).
ICAM-1 recognizes LFA-1 through an inserted (I) domain in the ␣ subunit. There is strong evidence correlating shifts in I domain conformation to affinity changes in binding ICAM-1. Mutations in I domain residues stabilized distinct structural conformations correlating to LFA-1 affinity. ICAM-1 equilibrium binding constants increase over 4 orders of magnitude ranging between low (i.e. 1600 M), intermediate (i.e. 9 M), and high affinity (i.e. 0.15 M) (5). Further evidence linking allosteric shifts in I domain conformation to ICAM-1 binding is the activity of a class of allosteric small molecule antagonists engineered to inhibit LFA-1 function (6 -9). Statin-derived small molecules such as lovastatin and LFA703 target the I domain allosteric site (IDAS) and abrogate LFA-1 recognition of ICAM-1 (8,10). Another small molecule to the IDAS, BIRT377, was shown to inhibit rolling and adhesion of LFA-1 transfectants to ICAM-1 monolayers (11). BIRT377 and LFA703 appear to exert their actions through shifting LFA-1 into a bent conformation, rendering the I domain inaccessible and LFA-1 into a low affinity state (12). A second class of small molecules binds to the I-like domain in the ␤ subunit of LFA-1 and indirectly regulates I domain affinity and ligand binding (10,12). The small molecule XVA143 binds to the I-like domain and promotes cellular rolling by inducing an extended conformation that stabilizes an intermediate affinity associated with rolling of LFA-1 expressing transfectants in shear flow (11). We present here a new small molecule allosteric inhibitor that targets the IDAS and downshifts LFA-1 from a high to intermediate affinity. This small molecule is similar to the diaryl sulfide cinnamide antagonists (13). Allosteric small molecules provide a powerful tool for directing leukocyte adhesion; however, the interrelationships between bond kinetics, LFA-1 conformation, valence in binding ICAM-1, and adhesion stability remain ill-defined.
Concomitant with a shift in affinity is a rapid redistribution of LFA-1 into high density clusters on the plasma membrane.
We have reported recently that within seconds of activation, LFA-1 on neutrophils reorganizes from a uniform surface distribution to form both small punctate clusters (Ͻ1 m 2 ) and large caps (ϳ3 m 2 ) (1,14). Clustering of LFA-1 on leukocytes tethered to inflamed endothelium in shear flow is a key step in adhesion strengthening and the transition from cell rolling to arrest (1,14). To emulate affinity and molecular scale clustering of LFA-1 on activated leukocytes, we engineered a cell-free LFA-1 expression system by fusing the ␣-␤ subunits at the C terminus with an inserted leucine zipper motif. The C termini of two heterodimers were bound to each Fab arm of an antileucine zipper antibody covalently attached to the surface of a latex microsphere. We tested the hypothesis that two adjacent LFA-1 binding to an ICAM-1 homodimer could facilitate rebinding and exponentially prolong bond lifetime. Binding of fluorescent ICAM-1 to LFA-1 on beads was monitored in real time by flow cytometry while shifting LFA-1 conformation and affinity state with soluble agonists and antagonists.
Dissociation of monomeric ICAM-1 from high affinity LFA-1 was ϳ10-fold faster than dimeric ICAM-1. This difference was attributed to the ability of the dissociated leg of the ICAM-1 dimer to rebind to an adjacent LFA-1 as it is held in proximity by the remaining LFA-1 bond. Adhesion of neutrophils to beads presenting dimeric ICAM-1 in shear flow was sustained beyond 10 min, while monomeric ICAM-1 beads dissociated within 100 s. These data highlight the physiological significance of regulation of both LFA-1 affinity and LFA-1 spatial proximity in tuning bond lifetime and adhesion stability when binding to ICAM-1 homodimer.
Assembly of LFA-1 on the Microsphere-Amino microsphere latex beads (diameter ϭ 6 m, 2.7% solids, latex) (Biosciences, Piscataway, NJ) were mixed (Eppendorf Thermomixer R, Brinkmann Instruments) with a sulfosuccinimidyl maleimide-N-hydroxysuccinimide ester crosslinker (0.125 ϫ 10 Ϫ3 M) (Pierce) as described by the manufacturer's protocol. A sulfhydryl group was added to anti-leucine zipper 324C by incubating 324C with a sulfhydryl linker (Pierce). 324C-SH was mixed (450 rpm) with the amino cross-linker beads for 2 h at 8°C. Beads were washed and stored at a final concentration of 10 7 beads/ml at 4°C. Directly prior to experimentation, LFA-1 (20 g/ml) was mixed with 10 5 beads/100-l sample at 37°C at 450 rpm for 30 min. The bivalent binding structure of the antibody allows for clustering of the LFA-1 into a dimer-like configuration presented on the bead surface. The cell-free system offers the opportunity to prescribe LFA-1 distribution in a manner that shifts the binding spectrum toward bivalent driven adhesion of ICAM-1.
Binding kinetics of ICAM-1-AlexaFluor488 (10 g/ml) to beads were measured by adding ICAM-1 immediately prior to flow cytometry readings. ICAM-1 association and dissociation were measured for 3 ϫ 10 5 beads in 300 l of PBS aliquots for up to 40 min at room temperature. In the indicated samples, there was a 10-min incubation at room temperature with MgCl 2 (or MnCl 2, 3 mM) and/or 240Q (10 g/ml) before the addition of ICAM-1 with the activator remaining in the solution throughout the flow cytometry reading. Dissociation of bound ICAM-1 was continuously detected after the introduction of small molecule inhibitor IC487475 (1 M), small molecule lovastatin (100 M), addition of TS1/22 (100 g/ml), or dilution in the presence of unlabeled ICAM-1. The inhibitory reagents were introduced through polyethylene tubing inserted into the cytometer tube. For dissociation of bound ICAM-1 with dilution, 1 ml of PBS was added containing unlabeled dimeric ICAM-1/Ig 20 times the concentration of the labeled ICAM-1.
To detect conformation, LFA-1 beads were treated with by MgCl 2 (3 mM), 240Q (10 g/ml), or IC487475 (1 M) during a 15-min incubation. KIM127 was incubated with the samples for an additional 15 min. Samples were washed and incubated for 15 min with goat anti-mouse FITC, and mean fluorescent intensities (MFI) were read by flow cytometry (FACScan flow cytometer, Pharmingen).
Neutrophil Isolation and Activation-Whole blood was drawn from healthy subjects by venipuncture into sterile syringes with heparin (10 units/ml of blood, Elkins-Sinn, Inc., Cherry Hill, NJ) as described previously (1). Neutrophils were isolated from whole blood using a density gradient media (Robbins Scientific Corp., Sunnyvale, CA). Neutrophils were washed once with HEPES buffer (10 mM KCl, 110 mM NaCl, 10 mM glucose, 1 mM MgCl 2 , and 30 mM HEPES, pH 7.4) and were maintained at room temperature in a calcium-free HEPES buffer for up to 4 h. Neutrophils were activated by addition of antibody 240Q (10 g/ml) for 10 min at room temperature in HEPES buffer, human serum albumin (0.1%), and CaCl 2 (1.5 mM). Kinetic experiments were performed as described for the LFA-1 microspheres.
ICAM-1 Bead Assembly-Carboxylate microspheres (diameter ϭ 10 m) were purchased from Polysciences, Inc. (Warrington, PA). 500 l of beads were washed twice in 1.5 ml of MES buffer, pH 5.0 (Sigma), resuspended in 200 l of MES, and sonicated for 15 min. 1-Ethyl-3-(3dimethylaminopropyl)-carbodiimide, hydrochloride (Molecular Probes, Eugene, OR), was added at 1 mM, and beads were incubated for 5 min at room temperature and at 500 rpm. Monomeric ICAM-1 (70 g/ml) or dimeric ICAM-1/Ig (32 g/ml) was mixed with the beads for 1 h at room temperature and at 500 rpm. Glycine (Sigma) was added (10 mM), and beads were mixed for 30 min at room temperature at 500 rpm. Blockaid blocking solution (Molecular Probes) was added (200 l) and incubation continued for an additional 15 min. ICAM-1 beads were washed in PBS and resuspended in 1.5 ml of PBS. Site densities were obtained by Quantum Simply Cellular Beads (Bangs Laboratories, Fishers, IN) to be ϳ6000 sites/m 2 for monomeric and dimeric beads as identified by anti-CD54.
Adhesion of Neutrophils to Beads Presenting Monomeric and Dimeric ICAM-1-Neutrophils were mixed with 10-m fluorescent latex beads (Fluoresbrite Carboxyl YG 10 Micron Microspheres) with ICAM-1 derivatized to their surface. Samples contained 1 ϫ 10 6 neutrophils/ml, 2 ϫ 10 6 beads/ml, and a small magnetic stir bar. Mac-1 blocking antibody 2LPM19c (test volume 10 l) and/or 240Q (10 g/ml) was preincubated for 10 min with the cell suspension without beads. For samples stimulated with IL-8 (5 nM), beads and stimulus were added immediately prior to reading on the flow cytometer. Samples were maintained at 37°C within a mixing chamber with a magnetic motor as described previously (22). The magnetic motor coupled with a magnetic stir bar created a shear field (shear stress ϳ1.0 dyne/cm 2 ) within the test tube and initiated collisional interactions. Neutrophil capture of ICAM-1coated beads was continuously monitored by their characteristic forward and right angle light scatter properties and gated in order to exclude unbound beads. Neutrophil-bead adhesion was quantitated on green fluorescence on fluorescence histograms. Quantal increases in fluorescence appeared as peaks in the fluorescence histogram corresponding to populations of neutrophils binding increasing numbers of beads (22). To distinguish relative levels of bead capture within the stimulated neutrophil population, neutrophil-bead interactions were quantitated as the average number of beads per neutrophil according to Equation 1, where N represents the number of nonadherent neutrophils, and NB i represents the number of neutrophil-bead aggregates bound to between 1 and 5 beads. Aggregates larger than 5 beads were not notably seen in this assay.
Dissociation of ICAM-1 beads from neutrophils was induced after 2 min by removing the cytometer sample tube for not more than 10 s during reading and adding inhibitor (1 M IC487475, 100 M lovastatin, 100 g/ml TS1/22, or 4 l of 1:100 Me 2 SO:PBS). Cytometer reading resumed and dissociation was modeled as rate of bead/polymorphonuclear leukocyte disaggregation.
Data Analysis-Data were analyzed using Graphpad Prism version 4.0 for Windows (Graphpad Software Inc., San Diego). Constants k off (dissociation rate constant) and k obs (observed rate constant) were obtained by performing one-phase exponential decay (Y ϭ specific binding ϫ e ϪkT ϩ nonspecific binding) and one-phase exponential association (Y ϭ Y max (1 Ϫ e ϪkT )) curve fits of the real time data, respectively. The k on (association rate constant) was calculated as k on ϭ (k obs Ϫ k off )/(ICAM-1-Alexa). Data points were calculated by taking the average fluorescence over 5-25 s, depending on the rate of change. K D , the equilibrium affinity constant, was calculated as k off /k on for kinetic experiments with ICAM-1 binding to the LFA-1 beads. K D was calculated as the EC 50 value in a dose-response curve for ICAM-1 binding to the neutrophils. Statistical significance was determined (p Յ 0.05) by a one-way analysis of variance with a Newman-Keuls multiple comparison post-test or by two-way analysis of variance.

RESULTS
LFA-1 expressed on the leukocyte membrane can shift its conformation in response to inside-out signaling via chemokine stimulation or extracellularly by addition of divalent cations (23), antibodies (16), and small molecules (12,24). LFA-1 heterodimer captured at the C terminus by an antibody covalently attached to microspheres was predominantly expressed in a low affinity state but retained the flexibility to shift conformation and boost affinity for ICAM-1. Cell-free LFA-1 increased binding to dimeric ICAM-1/Ig by 150% over base line in response to the addition of divalent cations Mg 2ϩ or Mn 2ϩ , but by only 20% in response to Ca 2ϩ (Fig. 1). Addition of mAb 240Q, which is associated with allosterically stabilizing CD18 into a ligand binding conformation (1,6,15,16), did not itself induce activation of LFA-1 on beads but in conjunction with Mg 2ϩ augmented ICAM-1 binding by 50% above stimulation with Mg 2ϩ alone.
ICAM-1/Ig binding was inhibited by pretreating Mg 2ϩ -activated LFA-1 with mAb TS1/22, which when bound to its epitope on the I domain sterically blocks ICAM-1 recognition (17,20,25). Addition of the allosteric small molecule IC487475, which binds with high affinity to the I domain (i.e. ϳ10 nM), also abrogated ICAM-1 binding stimulated by Mg 2ϩ (Fig. 1b). Pretreatment with mAb TS1/18, which binds to an allosterically sensitive domain on the ␤ subunit (25), blocked 65% of ICAM-1 binding in the presence of Mg 2ϩ . A nonblocking control, anti-CD11a TS2/4 (18), increased ICAM-1 binding in the presence of Mg 2ϩ by 25% (data not shown). These data indicate that cell-free LFA-1 retains the capacity to increase affinity for the ICAM-1/Ig dimer through activation by divalent cation or by allosteric mAb 240Q. Moreover, ICAM-1 binding can be sterically blocked by TS1/22 or allosterically inhibited by binding of a small molecule to the I domain.
Antibodies that recognize specific epitopes on the ␤ subunit can report on integrin conformation and activation. KIM127 binding has been correlated recently with the extended conformation of LFA-1 and is associated with intermediate or high affinity ligand binding (11, 19 -21). This activation reporter was used to determine the mechanism by which IC487475 and 240Q allosterically alter the affinity for ICAM-1. In the absence of divalent cations, binding of KIM127 was 2-fold above an IgG isotype-matched control, indicating that a fraction of the LFA-1 adopted a conformation other than the bent low affinity state in the absence of stimulation (data not shown). Activation of LFA-1 with Mg 2ϩ increased KIM127 binding by 40%, which was augmented to 60% upon addition of 240Q (Fig. 1c). IC487475 did not significantly decrease LFA-1 recognition by KIM127 despite its ability to abrogate Mg 2ϩ -induced ICAM-1 binding. This suggests that IC487475 can downshift LFA-1 affinity at the IDAS in the absence of inducing a bent conformation of the heterodimer.
Kinetic Analysis of ICAM-1 Binding by Fluorescence Flow Cytometry-Leukocytes can shift LFA-1 conformation and increase avidity within seconds of contact at vascular sites of inflammation (26). We applied fluorescence flow cytometry to detect with sub-second resolution the binding kinetics of ICAM-1-Alexa488 to suspensions of LFA-1 beads following addition of ICAM-1 into the cytometry tube (Fig. 2a). Binding kinetics for monomeric and dimeric ICAM-1 were modeled (see "Materials and Methods" for details), and the k on values were found to be statistically equivalent (Fig. 2b). The rate of association of both dimeric and monomeric ICAM-1 was increased by ϳ 60% in the presence of the allosteric stabilizing mAb 240Q over association in the presence of Mg 2ϩ alone. Following equilibrium binding of dimeric ICAM-1-AlexaFluor488, dissociation kinetics were measured following injection of inhibitors or from dilution in 20-fold molar excess of unlabeled ICAM-1/Ig (Fig. 3a). Inhibitor was injected directly into the flow cytometry test tube, and dissociation curves were fit to a first order exponential decay equation to obtain the k off rate constant. Blocking reassociation of dimeric ICAM-1-AlexaFluor488 with antibody TS1/22 or unlabeled ICAM-1/Ig yielded an equivalent dissociation rate constant of ϳ0.003 s Ϫ1 , representative of the high affinity state (Table I). The dissociation rate was hastened ϳ10-fold to 0.03 s Ϫ1 by injection of IC487475, representative of dissociation from an intermediate affinity. Low affinity LFA-1 was detected upon injection of lovastatin, which resulted in a dissociation rate constant of ϳ4 s Ϫ1 , an ϳ1000-fold increase from high affinity dissociation (Fig. 3, b and c). Addition of the nonblocking TS2/4, which binds to the ␤-propeller of the ␣ subunit (18), did not induce dissociation of ICAM-1/Ig. Stabilizing CD18 by activation in the presence of Mg 2ϩ and 240Q decreased k off by 30% (Fig. 3c). These data demonstrate I domain-specific allosteric regulation of ICAM-1 bond lifetime from a high affinity of ϳ300 s down to ϳ0.4 s.
We next determined whether dimeric LFA-1 positioned on each Fab of the capture antibody bound more tightly to dimeric ICAM-1/Ig than to the monomeric ICAM-1. Dissociation kinetics in the presence of excess unlabeled ICAM-1/Ig or TS1/22 revealed that monomeric bonds are more transient than dimeric bonds (Fig. 4a). ICAM-1 dissociated at a rate of 0.04 s Ϫ1 , a value 11-fold faster than dissociation of dimeric ICAM-1/Ig in the presence of unlabeled ICAM-1 (Fig. 4b). Stabilization of activated LFA-1 by 240Q slowed dissociation of monomer ICAM-1 by ϳ30%, on par with that observed for stabilized dimer. Injection of IC487475 hastened dissociation by 4-fold, corresponding to a decrease in LFA-1 bond lifetime from ϳ25 to 7 s (Table I). Injection of lovastatin hastened dissociation by 100-fold from the high affinity state of monomeric ICAM-1.
We next tested the dissociation of monomeric and dimeric ICAM-1 from monovalent LFA-1 that was covalently attached directly to the bead surface presumably as a single heterodimer. Kinetic analysis revealed dissociation of monomeric ICAM-1 at 0.03 s Ϫ1 and dimeric at 0.05 s Ϫ1 from high affinity LFA-1. Because dissociation from monovalent LFA-1 occurred at statistically equivalent rates, we concluded that divalent LFA-1 presented on the anti-leucine zipper bead is the minimum configuration for forming dimeric bonds with ICAM-1/Ig. Together, monomeric and dimeric dissociation data suggested that monomeric ICAM-1 is 5-10-fold less stable than dimeric ICAM-1 at high and intermediate affinity but that dissociation becomes equivalent as LFA-1 adopts a low affinity conformation.
Regulation of Affinity of LFA-1 on Human Neutrophils-Neutrophils express Mac-1 (CD11b/CD18) and LFA-1 on their plasma membrane, both of which bind ICAM-1 (1). To discriminate the binding kinetics of the LFA-1/ICAM-1 interaction, neutrophils were preincubated with mAb 2LPM19c that blocks Mac-1 binding to ICAM-1. In contrast to cell-free LFA-1, ICAM-1 binding to neutrophils was not activated by addition of Mg 2ϩ alone. Therefore, both Mg 2ϩ and 240Q were added to cell suspensions, and binding was measured in real time by flow cytometry. AlexaFluor488-conjugated ICAM-1 binds both specifically and nonspecifically to neutrophils, the latter defined as that nonblockable with anti-CD18. Therefore, only dimeric ICAM-1/Ig, which yielded a significant increase in specific binding, was examined (Fig. 5a). Addition of ICAM-1/Ig to neutrophils elicited binding kinetics qualitatively similar to cell-free LFA-1 on beads. Dissociation of ICAM-1/Ig was induced by injection of IC487475 or TS1/22, which decreased signal down to the background level of fluorescence (Fig. 5b), whereas addition of nonblocking anti-LFA-1 TS2/4 did not induce dissociation from neutrophils (data not shown). Dissociation of ICAM-1/Ig from high affinity LFA-1 on neutrophils was ϳ0.02 s Ϫ1 , a value between those observed for monomer and dimer ICAM-1 on cell-free LFA-1 ( Table I). Addition of IC487475 hastened the rate of dissociation by 4-fold that of steric inhibition with TS1/22 (Fig. 5c). ICAM-1/Ig dissociation induced by binding of allosteric or steric inhibitors was ϳ2-fold faster on neutrophils as compared with beads. These data suggest that a fraction of the LFA-1 expressed on neutrophils either binds monovalently to ICAM-1/Ig or remains in a lower affinity state.
Equilibrium affinity constants were computed from binding kinetics of ICAM-1/Ig on beads yielding a K D of 19 nM for stabilized (240Q) dimer, a value ϳ10-fold lower than the K D of 221 nM obtained for stabilized monomeric ICAM-1 (Fig. 6). In  (3 mM). Curves were modeled to a one-phase exponential decay equation. b, dissociation rate constants, k off , for monomeric ICAM-1 from LFA-1 beads. Constants were calculated by Prism software from a one-phase exponential decay equation (mean Ϯ S.E., n Ն 3, * denotes a significant increase in k off at p Յ 0.05). the absence of 240Q, the K D for dimeric ICAM-1 was ϳ50 nM. By comparison the K D for neutrophil binding to dimeric ICAM-1 was ϳ150 nM.
Influence of LFA-1 Affinity and Valency on Neutrophil Adhesion-We have reported previously that activation of neutrophils through chemokine signaling, or allosterically by mAb 240Q, induces high affinity and high density clusters of LFA-1 on the plasma membrane. These two components were critical for eliciting optimum adhesion efficiency of neutrophils on ICAM-1 expressing substrates in shear flow (27)(28)(29). To examine how ICAM-1 presentation as a monomer or dimer influences the efficiency of LFA-1-mediated adhesion of neutrophils stimulated with IL-8 or 240Q, we examined the kinetics of capture of beads coated with equivalent numbers of monomeric and dimeric ICAM-1 in sheared cell suspension. Cytometric based detection of neutrophil capture of fluorescent beads coated with ICAM-1 provides a continuous readout of the kinetics of LFA-1 avidity. In this case, the rate of ICAM-1 bead capture was quantitated over the 1st min following injection of agonist (Fig. 7a). In the absence of stimulus, monomeric beads did not adhere, whereas dimeric ICAM-1/Ig beads exhibited a low level of bead capture (Fig. 7b). Dimeric ICAM-1 beads bound with a 1-fold higher capture rate than monomeric ICAM-1 beads in response to stimulation with IL-8 or 240Q. A remarkable difference was in the stability of bead adhesion activated with IL-8. Beyond the 1st min of bead capture, monomeric ICAM-1 beads steadily dissociated from neutrophils, whereas dimeric beads were captured and remained stably bound throughout 10 min of observation (Fig. 7a). Activation of neutrophils with 240Q effectively prevented monomeric bead dissociation and boosted the rate of capture up to 3-fold for either type of ICAM-1 beads. These data indicated that both the affinity of LFA-1 and its valency in binding ICAM-1 directly influenced the efficiency and stability of neutrophil adhesion.
We next examined the nature of bond stability in dissociation of the dimeric ICAM-1/Ig beads by injecting soluble inhibitors directly into the sheared neutrophil bead suspensions (Fig. 7, c  and d). We hypothesized that the stability of ICAM-1 bead adhesion is dependent on the continuous formation of LFA-1 bonds and that the rate of dissociation is indicative of these dynamics. Monomeric ICAM-1 beads dissociated at the same rate as those treated with TS1/22, suggesting that blocking rebinding of a single leg of ICAM-1/Ig facilitated bead dissociation at a rate expected for the decay of single LFA-1/ICAM-1 bonds. By comparison, IC487475 and lovastatin hastened the rate of dissociation by 40% and 2-fold, respectively, compared with TS1/22 dissociation. These data reveal how shifts into intermediate and low affinity influenced the stability of neutrophil adhesion. Furthermore, they revealed the importance of dimeric bond formation between LFA-1 and ICAM-1 in adhesion stability in shear flow. DISCUSSION Integrins may be thought of as gatekeepers in controlling the location and efficiency of leukocyte recruitment to vascular FIG. 5. ICAM-1/Ig binding to neutrophils. Experiments were performed in HHB buffer containing 1 mM Mg 2ϩ and 240Q, IC487475, TS1/22 as indicated, and 1:10 dilution with excess unlabeled ICAM-1/ Ig. a, representative curves of the full kinetics of ICAM-1/Ig binding to neutrophils. Dissociation was induced by injection of inhibitors through polyethylene tubing inserted into the cytometer tube. Dissociation was measured in the presence of 240Q, and inhibitor was added at respective arrows. b, dissociation of dimeric ICAM-1 from neutrophils. Curves were modeled by Prism software to a one-phase exponential decay equation. c, dissociation rate constants (mean Ϯ S.E., n Ն 3, * denotes a significant increase in k off at p Յ 0.05).
FIG. 6. Equilibrium binding affinity of cell-free and neutrophil LFA-1. K D of dimeric and monomeric ICAM-1 binding to LFA-1 beads in the presence of Mg 2ϩ (3 mM) with activating anti-CD18 240Q. K D was calculated as k off /k on for LFA-1 beads. Dissociation was induced by a 1:10 dilution with excess unlabeled ICAM-1/Ig (mean Ϯ S.E., n Ն 3). Equilibrium binding affinity of dimeric ICAM-1 binding to neutrophils in the presence of Mg 2ϩ (1 mM) and 240Q calculated from the EC 50 (mean Ϯ S.E., n ϭ 2, * denotes significant difference with p Յ 0.05).
sites of inflammation. To explore the relationships between integrin structure and function, we have engineered beads to express recombinant LFA-1 as a dimer at levels commensurate with that on leukocytes. This allowed a study of the influence of LFA-1 affinity and valence on bond lifetime and adhesion to ICAM-1 expressed as monomer and homodimer. Shifts in LFA-1 conformation from a low or intermediate to high affinity state were induced by divalent cations with the level of activation such that Mn 2ϩ and Mg 2ϩ exceeded Ca 2ϩ , in accordance to the hierarchy of activation observed previously (30,31). ICAM-1 dissociation from high affinity LFA-1 was ϳ10-fold faster for monomeric than dimeric ICAM-1. A downshift in LFA-1 affinity was rapidly triggered by the binding of IC487475 or lovastatin. Despite binding to a similar domain at the IDAS, these small molecules stabilized very different states in LFA-1.
Addition of IC487475 decreased bond lifetime for both the monomer and dimer ϳ5-fold from that of high affinity dimer and monomer, respectively. Lovastatin decreased bond lifetime 1000fold for the dimer and 100-fold for the monomer. These data reveal the capacity of neutrophils to regulate dynamically the affinity of LFA-1 over 4 orders of magnitude through shifts in conformation initiated at the I domain. This is greater than the 300-fold shift in dissociation rate observed for isolated I domain mutations (5). Our data indicate that a second mechanism for regulating adhesion kinetics is bond valency and that dimeric bond formation between LFA-1 and ICAM-1 homodimers can increase bond stability by an order of magnitude for high affinity binding. Supporting this is the ICAM-1 bead capture data. Although the rate of neutrophil adhesion to dimeric ICAM-1 beads was ϳ1-fold higher than for monomeric beads, adhesion stability exhibited greater sensitivity to bond valence. Conversion from transient (seconds) to stable adhesion (minutes) was only observed when neutrophils bound dimeric ICAM-1 beads.

Real Time Detection of LFA-1 Affinity Shifts on Beads and
Neutrophils-Flow cytometry operated in a continuous detection mode provides kinetic rates and affinity constants similar to those obtained by surface plasmon resonance (12,30,(32)(33)(34), but with at least an order of magnitude lower signal to noise ratio. This was evident for measurement on neutrophils, where affinity states of LFA-1 with K D Ͼ250 nM could not be detected. One advantage of flow cytometry is the ability to directly compare ICAM-1 binding between LFA-1 on beads in suspension and LFA-1 on neutrophils where conformation, affinity, and membrane redistribution are dynamically being regulated. Stabilized high affinity LFA-1 on beads corresponded to a k off of 0.002 s Ϫ1 for dissociation of dimeric ICAM-1 and 0.03 s Ϫ1 for dissociation of monomeric ICAM-1. These results are comparable with dissociations detected previously by surface plasmon resonance of 0.0016 s Ϫ1 for dimeric and 0.022 s Ϫ1 for monomeric ICAM-1 binding to immobilized high affinity LFA-1 I domain (33). The value for monomeric ICAM-1 binding affinity is also agreement with Jun et al. (33) where monomeric ICAM-1 was reported to have a K D of 168 nM when binding to immobilized high affinity LFA-1 I domain. Dimeric ICAM-1 was reported by Jun et al. (33) to bind with a K D of 102 nM (value multiplied by 2 to express the K D value in terms of binding sites). Cell-free LFA-1 as reported here had a lower K D value when binding to dimeric ICAM-1 (19 nM for stabilized high affinity LFA-1) than previously reported by Jun et al. (33). This is attributed to the presentation of LFA-1 as a dimer on each Fab of the anti-leucine zipper anchored to the bead, which effectively increases the likelihood of rebinding of each domain-1 of the ICAM-1/Ig dimer upon dissociation.
Affinity Is Dynamically Regulated through Conformational Shifts of the I Domain-How conformational shifts in ␤ 2 integrin translate into affinity changes that support distinct adhesive interactions such as cell rolling and arrest is an area of active investigation. Site-directed mutagenesis of the LFA-1 I domain has led to identification of stable conformations corresponding to low, intermediate, and high affinity states for binding ICAM-1. Dissociation rate constants reported for mutant LFA-1 expressed on transfected cells vary from 0.014 s Ϫ1 for high affinity to 4.6 s Ϫ1 for low affinity (5). ICAM-1 binding to cells is triggered from the inside-out as a result of cell signaling events that culminate in cytoplasmic domain shifts of the integrin (35). As shown here, binding can also be induced extracellularly by coordination of a Mg 2ϩ or Mn 2ϩ ion at the metal ion-dependent adhesion site (MIDAS) mapped to the top of the I domain (36). ICAM-1 binding is regulated by the IDAS, which is structurally linked to the MIDAS and found to bind statin-derived small inhibitory molecules such as lovastatin (8). Site-directed mutagenesis of amino acid residues within the IDAS are reported to increase LFA-1 affinity 6-fold (6,15). The I-like domain on the ␤ subunit also contains a MIDAS region that can regulate integrin affinity. The small molecule XVA143 binds to the I-like domain and induces the extended conformation of LFA-1 as indicated by recognition of the KIM127 epitope (37). XVA143 appears to stabilize an intermediate state that supports LFA-1-mediated cell rolling in shear flow but induces dissociation of soluble multimeric ICAM-1 (11). A member of a new class of IDAS inhibitors, IC487475, elicited an increase in ICAM-1 dissociation from ϳ0.003 to 0.03 s Ϫ1 , corresponding to a shift in LFA-1 from high to intermediate affinity. We found that IC487475 exerts its effect by altering the local conformation of the I domain while leaving the stalk region of the ␤ subunit in the open or extended conformation as confirmed by binding of mAb KIM127 (38). In separate studies, 2 we have observed that IC487475 can support neutrophil rolling on inflamed endothelium.
LFA-1 Forms Bivalent Bonds with Dimeric ICAM-1 and Prolongs Adhesive Lifetime-Published data (39) have demonstrated that ICAM-1 is expressed as a homodimer on inflamed endothelium. Spontaneous dimerization of ICAM-1 on the plasma membrane is believed to facilitate high avidity binding through LFA-1 (40,41). This is supported by ϳ10-fold faster dissociation rate constant (k off ) for LFA-1 binding to monomeric than dimeric ICAM-1 (33). In the present study, we show for the first time that the relative efficiency of neutrophil adhesion to ICAM-1 is ϳ60% greater in capture of dimer ICAM-1 than monomer at equivalent site density. This difference can be attributed to the capacity of the high affinity LFA-1 on neutrophils stimulated with IL-8, or allosterically induced by binding of mAb 240Q, to form tight molecular scale clusters capable of forming multivalent bonds to ICAM-1 homodimers (1,42). We found that a minimum, two LFA-1 engaging two ICAM-1 effectively prolonged adhesion lifetime. There was an exponential increase in the extension of the bond lifetime of the dimer over the monomer with increasing LFA-1 affinity. Stabilized in a low affinity state with lovastatin, dissociation of monomeric or dimeric ICAM-1 occurred at a closely equivalent and rapid rate. Stabilized in an intermediate affinity state with IC487475, dimeric ICAM-1 dissociated 4-fold slower than monomeric ICAM-1. At a high affinity state, dimeric dissociation was slowed 10-fold compared with monomeric dissociation. A simple computational model of dimeric ICAM-1/Ig high affinity dissociation implicates rebinding as a primary mechanism. In the absence of rebinding, the ratio of dimeric to monomeric ICAM-1 dissociation rate should be 1:2 instead of the observed 1:10. We applied kinetic rate equations for creation and destruction of single and double bonds between LFA-1 and ICAM-1/Ig to model the dissociation and included a rebinding frequency k rebind . Note that this rebinding applies to the dimeric LFA-1-ICAM-1-Ig bond complex, and dissociation k off for a single ICAM-1 was assumed to be comparable with dissociation from the monomeric complex (ϳ0.04 s Ϫ1 ). Our simple model (Equation 2) shows that the rate-limiting exponential decay of dimeric ICAM-1/Ig is governed by the frequency of rebinding.
ϳexp(Ϫk off 2 t/͑k rebind ϩ 2k off ͒) (Eq. 2) A rebinding frequency of 0.25 s Ϫ1 provided a best fit of the high affinity state dissociation kinetics for ICAM-1/Ig. Allosteric inhibition with IC487475 and lovastatin revealed the importance of rebinding in that the dimeric to monomeric ratio of k off remained at 5:1, even in the intermediate affinity state. However, at low affinity the ratio was ϳ1:1. We conclude that the increased frequency of rebinding of dimeric ICAM-1 afforded by molecular proximity of LFA-1 results in a macromolecular dissociation rate that is disproportionate to that of monomer ICAM-1 coming off of LFA-1 over a range of affinity. A more complete modeling of how molecular rate constants influence cellular adhesion kinetics is under investigation. The increase in bond stability for dimeric bonds was also evident for neutrophil capture of beads as monomeric ICAM-1 beads were captured 60% less efficiently than dimeric ICAM-1, despite a similar molecular rate of association. Thus, molecular scale clustering of LFA-1 and ICAM-1 is deemed critical to capture efficiency during collision between neutrophils and substrate, which are as brief as 5 ms at venular shear rates (29). Therefore, LFA-1 clustering on the neutrophil surface within seconds of chemotactic signaling provides a physiological mechanism for engaging homodimers of ICAM-1 and in turn transitioning leukocyte rolling to arrest in shear flow. We hypothesize that efficient rebinding of clustered LFA-1 to ICAM-1 expressed at a minimum as homodimers is critical to sustaining the stability of neutrophil adhesion. We observed that dissociation of dimeric ICAM-1 beads ensued at a rate equivalent to monomeric beads upon addition of TS1/22, whereas downshifting LFA-1 affinity by IC487475 or lovastatin triggered more rapid dissociation. The antibody and small molecules act very differently in that TS1/22 displaces exposed high affinity sites as they become available, whereas the small molecules allosterically induce dissociation. Most interestingly, despite a predicted 100-fold difference in the dimeric affinity stabilized by IC487475 versus lovastatin (i.e. Table I), the bead dissociation rates only differed by 1-fold. We conclude that the rate-limiting step in adhesion strengthening is governed by the rebinding of LFA-1 to ICAM-1, whereas dissociation may involve additional factors. Clustering on the surface of the neutrophil may also involve configurations other than the dimer configuration created on the surface of the bead. LFA-1 is mobile within the membrane, and densities are changing as a function of time (1).
In the current study, we define the relationships between LFA-1 conformation, bond valency with ICAM-1, and bond/ adhesion lifetime. A novel allosteric small molecule targeted to the IDAS revealed that LFA-1 dissociation can be dynamically regulated within this region into an intermediate affinity state. The data show for the first time that neutrophil regulation of LFA-1 affinity for ICAM-1 can be controlled via the I domain and that formation of molecular scale clusters of LFA-1 is critical for increasing avidity. In this manner, neutrophil adhesion kinetics can be tuned from rolling to arrest and transmigration. Regulation over this process can also be exerted at the endothelial cell level, which can dynamically control site density and mobility of ICAM-1. ICAM-1 is present on microvillus projections that extend from the endothelial surface to form