The C-Terminal αI Domain Linker as a Critical Structural Element in the Conformational Activation of αI Integrins*

The activation of α/β heterodimeric integrins is the result of highly coordinated rearrangements within both subunits. The molecular interactions between the two subunits, however, remain to be characterized. In this study, we use the integrin αLβ2 to investigate the functional role of the C-linker polypeptide that connects the C-terminal end of the inserted (I) domain with the β-propeller domain on the α subunit and is located at the interface with the βI domain of the β chain. We demonstrate that shortening of the C-linker by eight or more amino acids results in constitutively active αLβ2 in which the αI domain is no longer responsive to the regulation by the βI domain. Despite this intersubunit uncoupling, both I domains remain individually sensitive to intrasubunit conformational changes induced by allosteric modulators. Interestingly, the length and not the sequence of the C-linker appears to be critical for its functionality in α/β intersubunit communication. Using two monoclonal antibodies (R7.1 and CBR LFA-1/1) we further demonstrate that shortening of the C-linker results in the gradual loss of combinational epitopes that require both the αI and β-propeller domains for full reactivity. Taken together, our findings highlight the role of the C-linker as a spring-like element that allows relaxation of the αI domain in the resting state and controlled tension of the αI domain during activation, exerted by the β chain.

Integrins are a large family of ␣/␤ heterodimeric cell surface receptors that mediate interactions with other cells or the extracellular matrix. They are important therapeutic targets in a wide range of diseases, including cardiovascular and immune disorders (1). Integrin activation is dynamically regulated by signals from within the cell in a process termed inside-out signaling. In addition, outside-in signaling induced by ligand binding directs signals from the extracellular domains to the cytoplasm. This bidirectional signaling is associated with highly coordinated domain rearrangements in both the ␣ and the ␤ subunits (2).
Several studies indicate that inside-out signaling converts integrins from a bent conformation with a closed headpiece into an extended conformation with an open headpiece and thereby activates ligand binding (2) (Fig. 1, A-C). Central to ligand recognition are von Willebrand factor type A domains that are present in all integrin ␤ subunits (termed ␤ inserted (␤I) domains) and in some ␣ subunits (termed ␣ inserted (␣I) domains). In integrins that lack ␣I domains, the activated ␤I domain directly interacts with the ligand through a metal ion-dependent adhesion site (MIDAS). 3 In ␣I integrins, the ␣I domain serves as the ligand-binding domain instead of the nearby ␤I domain. Interestingly, in these integrins, the ␤I domain appears to regulate the activation of the ␣I domain and, thus, ligand binding. In the current model, an invariant Glu residue (Glu-310 in ␣ L ) located at the C-terminal end of the ␣I domain is thought to bind to the MIDAS of the active ␤I domain (Fig. 1C). This interaction is hypothesized to lead to an axial displacement of the ␣7 helix of the ␣I domain in the C-terminal direction, as seen in crystal structures of isolated ␣I domains (3). As a consequence, the ␣I MIDAS is turned into a high-affinity, ligand-binding state (2).
Additional support for a structural link between the ␣I and ␤I domains comes from studies that characterize small molecule inhibitors of the integrin ␣ L ␤ 2 at a molecular level (4). One class of ␣ L ␤ 2 inhibitors, termed ␣I allosteric inhibitors, has been shown to bind underneath the C-terminal ␣7 helix of the isolated ␣I domain and stabilize it in the closed, low-affinity state (4). On the basis of this finding it is hypothesized that these inhibitors lock the integrin in its inactive form by preventing the downward axial shift of the ␣7-helix required for ␣I domain/␤I domain interactions (4) (Fig. 1A). Another class of ␣ L ␤ 2 allosteric inhibitors, termed ␣/␤ I allosteric inhibitors, appear to bind to the MIDAS of the ␤I domain (4,5). Thereby they are hypothesized to competitively antagonize the binding of ␣ L Glu-310 to the ␤I domain. As a result, the ␣I domain remains in an inactive state, whereas the ␤I domain together with the "leg" region of the integrin is stabilized in a pseudoliganded, active state, as shown by the induction of activationdependent epitopes and induction of the extended conformation with the open headpiece (as seen with electron microscopy) (6) (Fig. 1D).
Crystal structures of the ␣I integrin ␣ X ␤ 2 revealed unanticipated flexibility of the ␣I domain (7). One possible function of this flexibility would be to enable two ␤I domain conformational states to couple with three ␣I domain states (7). A key * This work was supported, in whole or in part, by National Institutes of Health question remaining to be answered is which structural features enable the conformational flexibility of the ␣I domain needed for the activation of I domain integrins. Linkers connect the N termini (N-linker) and C termini (C-linker) of the ␣I domain to the ␤-propeller domain in which the ␣I domain is inserted (7). The role of the N-linker appears to be limited by its short length of three residues. In contrast, the C-linker, which follows the ␣7-helix of the ␣I domain and contains the invariant Glu-310, is ten residues long and is flexible, as shown by weak electron density in the crystal structure (7).
Here we use the leukocyte integrin ␣ L ␤ 2 to test the hypothesis that the C-linker acts as a spring-like element that, when mutationally shortened, activates the ␣I domain. ␣ L ␤ 2 is selectively expressed on all leukocytes and is among the best characterized of ␣I integrins (8). The ligands of ␣ L ␤ 2 are members of the Ig superfamily, including intercellular adhesion molecule-1 (ICAM-1). The ␣ L ␤ 2 /ICAM-1 interaction plays a major role in inflammatory and immune responses by regulating cell adhesion, leukocyte trafficking, and T cell costimulation (8). The present study provides important insights into how the C-linker regulates ␣I integrin adhesiveness.
Cell Culture-Human embryonic kidney 293T cells were cultured in Dulbecco's modified Eagle's medium supplemented with 2 mM L-glutamine, 10% fetal bovine serum, nonessential amino acids, and penicillin-streptomycin at 37°C with 5% CO 2 (all reagents from Invitrogen). The day before transfection lowpassage 293T cells were transferred into 24-well plates.
cDNA Constructs and Transient Transfections-cDNA of WT ␣L was inserted into pcDNA3.1/Hygro(-) (Invitrogen) and used as template for mutagenesis. ␣L C-linker deletion and swap mutants were generated by overlap extension PCR (15). For the ␣L C-linker deletion mutants ⌬ 8 and ⌬ 10, the mutant ⌬ 6 was used as a template. Human-mouse ␣L chimeras in expression vector AprM8 and I-less ␣ L ␤ 2 (lacking residues 129 -308) in the same expression vector were described previously (9,16). The chimeras were named according to the species origin of their segments. For example, h217m248h indicates that residues 1-217 are from human (h) ␣L, residues from 218 to 248 are from mouse (m) ␣L, and residues from 249 to the C terminus are from human ␣L. ␣L I domain expressed on the cell surface with N-terminal or C-terminal transmembrane domains have been described previously (17,18). Resequencing of these vectors demonstrated that they contain ␣L residues Val-130 to Val-339 and Gly-128 to Tyr-307, respectively. All constructs were confirmed by sequencing. 293T cells (80% confluent) were transfected with empty vector (mock) or cotransfected with mutant ␣L and wild-type (WT) ␤2 plasmids (either inserted into pcDNA3.1/Hygro(-), AprM8, or pcDNA3.1(ϩ)) using Lipofectamine 2000 according to the manufacturer's instructions. Two days after transfection the cells were harvested for flow cytometric analysis, adhesion, and binding assays.
Immunofluorescence Flow Cytometry-Immunofluorescence flow cytometry was performed as described previously (19). Briefly, transfected 293T cells were detached and washed once in 20 mM HEPES (pH 7.3) containing 150 mM NaCl, 1 mM CaCl 2 , 1 mM MgCl 2 , and 1.5% BSA (assay buffer A). Cells were then resuspended in assay buffer A containing 10 g/ml primary antibody and incubated on ice for 30 min. mAb KIM127 was used at a concentration of 7 g/ml and incubated at 37°C for 20 min. After a washing step, the cells were exposed to FITC-conjugated goat anti-mouse IgG (Invitrogen) diluted 1:500 in assay buffer A for 20 -30 min on ice. After two washing steps, cells were resuspended in cold assay buffer A and analyzed on a FACScan (BD Biosciences). Mean fluorescence intensities were calculated using the CellQuest software.
Cell Adhesion to ICAM-1-The cell adhesion assay was performed in V-bottom 96-well plates (Corning) as described previously (20). Briefly, the plates were coated with 10 g/ml recombinant human ICAM-1 or 10 g/ml BSA as a control in 20 mM Tris (pH 8), 150 mM NaCl, and 2 mM MgCl 2 at 4°C overnight (or 37°C for 2 h) and then blocked with 20 mM Tris (pH 7.5) containing 150 mM NaCl, 1.5% BSA, and 5 mM glucose (assay buffer B) at 37°C for 2 h. The transfected 293T cells were detached, resuspended in assay buffer B, and labeled with 1-2 g/ml 2Ј,7Ј-bis-(carboxyethyl)-5-(and-6)-carboxyfluorescein acetoxymethyl ester (BCECF AM) (Invitrogen) at 37°C for 30 min in the dark. After this labeling step, the cells were washed once and resuspended in assay buffer B containing 1 mM CaCl 2 and 1 mM MgCl 2 (resting condition) or 1 mM CaCl 2 , 1 mM MgCl 2 , and activating mAbs (10 g/ml KIM127 and 10 g/ml CBR LFA-1/2) or 1 mM MnCl 2 alone (activating conditions). After incubation at 37°C for 25 min in the dark, cells were vigorously pipetted up and down and added to the ICAM-1and BSA-coated plates (the cell number varied from 3000 to 10,000 cells/well from experiment to experiment). The plates were immediately centrifuged at 100 ϫ g for 10 min (Beckman CS centrifuge, brake off). After centrifugation, nonadherent cells that accumulated in the center of the V bottom were quantified using the Fluoroskan Ascent microplate fluorometer (Thermo Scientific) with the "small beam" setting and filter sets allowing excitation at 485 nm and quantification of emission at 535 nm. The percentage of cell adhesion was calculated according to the following formula: where Fl ICAM-1 is the fluorescent signal (arbitrary units) when cells bind to ICAM-1 (low signal) and Fl BSA is the fluorescent signal in absence of ICAM-1 and presence of BSA (high signal).
Binding of Multimeric Soluble ICAM-1-The binding of soluble ICAM-1 was assessed as described previously (21). Transfected 293T cells were detached using 20 mM HEPES (pH 7.3) supplemented with 150 mM NaCl and 5 mM glucose (assay buffer C) and transferred into V-bottom 96-well plates (Corning). The cells were washed in assay buffer C and resuspended in assay buffer C containing 2 mM CaCl 2 and 2 mM MgCl 2 (50 l/well). Multimeric ICAM-1 complexes were prepared by mixing human ICAM-1/Fc (R&D Systems) with affinity-purified goat anti-human IgG (HϩL)-FITC antibodies (Invitrogen) (1:10 w/w) and incubated at room temperature for 30 min. The ICAM-1 complexes were diluted 1:6 in assay buffer C and added to the plates (50 l/well), yielding a final concentration of 1 mM for each cation. The cells were incubated at room temperature for 30 min, washed in assay buffer C containing 1 mM CaCl 2 /1 mM MgCl 2 and subjected to immunofluorescence flow cytometry. As a control, soluble multimeric human myeloma IgG1 complexes with anti-human IgG were prepared as described above and exposed to the transfected cells.

Design and Cell Surface Expression of ␣ L ␤ 2 C-linker Mutants-
We designed five ␣ L ␤ 2 mutants in which the C-linker of the ␣I domain (residues 309 -318) (Fig. 1F) was shortened by 2, 4, 6, 8, or 10 amino acids (Fig. 1G). In ⌬ 10, the C-linker is completely removed, and thus, this mutant is the only mutant that lacks the invariant Glu-310 residue thought to be important for ␣I/␤I domain communication (Fig. 1F). The C-linker of ␣ L ␤ 2 was also FIGURE 1. Model of conformational activation of ␣ L ␤ 2 with the C-linker acting as a spring-like element (modified from (2)). A, bent conformation of WT ␣ L ␤ 2 with closed headpiece, stabilized by the ␣I allosteric antagonist LFA878 (low affinity). B, extended conformation of WT ␣ L ␤ 2 with closed headpiece and epitopes of mAbs KIM127 and CBR LFA-1/2 exposed (low affinity (23)). TM, transmembrane; PSI, plexin-semaphorin-integrin. C, extended conformation of WT ␣ L ␤ 2 with open headpiece and ligand bound (high affinity). D, extended conformation of WT ␣ L ␤ 2 with open headpiece but closed ␣ I domain induced by ␣/␤ ⌱ allosteric inhibitor XVA143. E, ␣ L ␤ 2 C-linker mutant ⌬ 10 shown in its extended conformation with a constitutively active ␣ I domain that is no longer responsive to regulation by the ␤ I domain. F, close-up of the ␣ L ␤ 2 C-linker region. The C-linker is shown in black. Upon activation, residue Glu-310 (shown in stick) is hypothesized to bind to the MIDAS Mg 2ϩ ion shown as a green sphere. The homology model is based on the ␣ X ␤ 2 crystal structure (7). G, C-linker deletions ⌬ 2 to ⌬ 10 and replacement of the ␣ L ␤ 2 C-linker by ␣ X ␤ 2 C-linker residues.
replaced with the C-linker of ␣ X ␤ 2 (CX mutant) to differentiate the importance of its sequence as compared with its length (Fig.  1G). Wild-type (WT) and mutated ␣ subunits were coexpressed with WT ␤2 in 293T cells. Immunofluorescence flow cytometry using mAb TS2/4, which binds the ␣L ␤-propeller domain and requires ␤2 association for reactivity, demonstrated that all ␣ L ␤ 2 C-linker deletion mutants were correctly assembled and expressed at WT levels ( Fig. 2A). In contrast, swapping the ␣ L ␤ 2 C-linker for the ␣ X ␤ 2 C-linker reduced cell-surface expression by 2-fold ( Fig. 2A). However, expression at WT level was achieved when ␣X-Glu-313 of the CX mutant was replaced by the corresponding ␣L C-linker Ser residue (mutant CX-E) (Figs. 1G and 2A).
Impact of C-linker Shortening on Global Conformation of the Mutants-To assess the global conformation of the deletion mutants, we tested the binding of mAb KIM127. The epitope of this activation-dependent antibody involves residues on the ␤2 subunit that are masked in the bent (inactive) and exposed in the extended (active) conformation (22). Further, the KIM127 epitope is known to be induced by ␣/␤ I allosteric inhibitors such as XVA143 (5). Under resting conditions in the presence of Ca 2ϩ and Mg 2ϩ , only basal binding of mAb KIM127 to the mutants was noted (Fig. 2B). This result suggests that the mutants are largely in the bent conformation in the absence of activating agents. Interestingly, KIM127 epitope exposure in the mutants was induced by XVA143 to a degree comparable with the WT receptor (Fig. 2B). These results show that despite C-linker shortening, all mutants are basally bent and are able to bind XVA143 and undergo conversion to an extended conformation.
Impact of C-linker Shortening or Swapping on ␣ L ␤ 2 Function-The function of the mutants was studied by investigating the adhesion of 293T cell transfectants to immobilized ICAM-1 using the V well assay format. Under activating conditions in the presence of Mn 2ϩ , the mutants ⌬ 2 and ⌬ 10 adhered comparably to the WT, whereas adhesion of 293T cells expressing  In contrast, all mutants bound at WT levels when mAb CBR LFA-1/2 and mAb KIM127 were used to stimulate adhesion (Fig. 3A). CBR LFA-1/2 is an activating antibody that binds to the leg region of the ␤2 subunit and maximally activates ␣ L ␤ 2 in combination with higher concentrations of mAb KIM127 (22,23). The CX-E transfectants were activated by Mn 2ϩ to levels observed for WT transfectants (Fig. 3A). The reduced adhesion of the CX mutant in presence of Mn 2ϩ is explained by lower cell surface expression (Figs. 3A and 2A).
Under resting conditions, in presence of physiological concentrations of Mg 2ϩ and Ca 2ϩ , the C-linker mutants ⌬ 8 and ⌬ 10 constitutively adhered to immobilized ICAM-1, whereas the rest of the mutants exhibited weak or no adhesion to ICAM-1 (Fig. 3B). The level of constitutive adhesion was comparable with the level observed for activated WT ␣ L ␤ 2 (Fig. 3, A and B). Selective binding of soluble ICAM-1 complexes to the ⌬ 8 and ⌬ 10 mutants confirmed their constitutively active phenotype (Fig. 3, C and D).
Susceptibility of the Constitutively Active Mutants to Small Molecule Antagonists and Inhibitory Antibodies-Interestingly, the adhesion of the constitutively active mutants ⌬ 8 and ⌬ 10 to immobilized ICAM-1 was not affected by the ␣/␤ I allosteric inhibitor XVA143 (Fig. 4), although the compound clearly binds to the mutants, as shown by induction of KIM127 epitope exposure (Fig. 2B). This result suggests that the active state of the ␣I domain is no longer regulated by the ␤I domain in these mutants. In contrast, inhibitors that bind underneath the ␣7-helix of the ␣I domain, such as LFA878, abolished the proadhesive state of the ⌬ 8 and ⌬ 10 mutants, demonstrating that shortening of the C-linker does not irreversibly activate the ␣I domain (Fig. 4). As expected, the control mAb TS1/22 completely blocked the binding of the two mutants to ICAM-1 (Fig.  4). TS1/22 was mapped to the ligand binding region of the ␣I domain (residues Gln-266 and Ser-270) and competitively inhibits ␣ L ␤ 2 (24,25).

C-Terminal ␣I Domain Linker and Activation of ␣I Integrins
The reactivity of the mAbs was further checked with ␣I-less ␣ L ␤ 2 , which lacks ␣ L residues 129 -308, i.e. the ␣I domain (16). Reactivity of both CBR LFA-1/1 and R7.1 was greatly decreased with ␣I-less ␣ L ␤ 2 compared with wild-type ␣ L ␤ 2 (Fig. 5B). However, the binding of mAb R7.1 was significantly above the reactivity of the control mAb TS1/22 (p ϭ 0.012) whose epitope exclusively maps to the ␣I domain (Fig. 5B). Reactivity of CBR LFA-1/1 with the ␣I-less construct was also greater than TS1/ 22, although this difference did not reach significance (Fig. 5B).
Furthermore, the binding of CBR LFA-1/1 and R7.1 to type I or type II-anchored ␣I domains was assessed. The type I construct (C-terminal anchorage of the I domain) includes ␣L residues 130 -339, i.e. almost all the ␣I domain (129 -308), the C-linker (309 -318), and an adjacent ␤-propeller segment (319 -339), whereas the type II construct (N-terminal anchorage) contains residues 128 -307, i.e. the ␣I domain only (17,18). CBR LFA-1/1 and R7.1 recognized type I-but not type II-anchored ␣L I domains (Fig. 5C). The binding of the antibodies to the type I TM-anchored ␣I domain was slightly but significantly reduced relative to TS1/22 (Fig. 5C). These results suggest that the C-linker and/or the ␤-propeller domain, in addition to the ␣I domain, contribute to the CBR LFA-1/1 and R7.1 epitopes.
Next we explored the effect of LFA878, an ␣I allosteric antagonist, and XVA143, an ␣/␤ I allosteric antagonist, on R7.1 and CBR LFA-1/1 reactivity. It has been described previously that ␣I and ␣/␤ allosteric inhibitors reduce or do not affect the binding of R7.1 to ␣ L ␤ 2 , respectively (26,27). The effects of the compounds on R7.1 binding were as described previously (Fig.  5D). Similarly to R7.1, CBR LFA-1/1 reactivity was reduced in the presence of LFA878. However, in contrast to R7.1, CBR LFA-1/1 reactivity was marginally (but significantly (p ϭ 0.021)) influenced by the ␣/␤ I allosteric antagonist XVA143 (Fig. 5D). Taken together, these unique properties indicated that the mAbs were well suited for the characterization of the C-linker mutants.
Impact of C-linker Shortening on mAb R7.1 and CBR LFA-1/1 Reactivity-Shortening of the ␣ L ␤ 2 C-linker progressively decreased the binding of the R7.1 and CBR LFA-1/1 antibodies to different degrees (Fig. 6). The CBR LFA-1/1 epitope was already affected in the ⌬ 2 mutant and almost lost in the ⌬ 4 mutant, whereas the epitope of mAb R7.1 was still preserved in ⌬ 2, was affected in ⌬ 4, and almost lost in ⌬ 6. Intriguingly, the binding of both antibodies could be reconstituted by the insertion of the ␣ X ␤ 2 C-linker (CX mutant) and its modified version (CX-E mutant), which differ from the ␣ L ␤ 2 C-linker in six and five amino acids, respectively (Figs. 6 and 1B). Thus, the data suggest that these C-linker residues, 313-318, do not directly contribute to the R7.1 or CBR LFA-1/1 epitopes. In summary, our data show that shortening of the C-linker results in the gradual (yet differential) loss of two combinational epitopes that require both the ␣I domain and the C-linker/␤-propeller domain for full reactivity.

DISCUSSION
The C-linker is defined as the C-terminal connection between the ␣I domain and the ␤-propeller domain, which belongs structurally to neither the ␣I nor the ␤-propeller domains. The C-linker is located at the interface with the ␤I domain, and mutations in the C-linker including Glu-310, Lys-314, and Leu-317 have been shown to regulate I domain affinity (28,29). In this study, we systematically addressed the functional role of the C-linker of the ␣I domain integrin ␣ L ␤ 2 by mutational analysis. Stepwise shortening of the C-linker neither impaired expression of ␣ L ␤ 2 nor its adhesive function induced by activating antibodies. However, differences in the level of adhesion were observed when the divalent cation Mn 2ϩ was used as a stimulus. The mechanism of ␤2 integrin activation by Mn 2ϩ is not well understood, and these differences may suggest two distinct mechanisms that are differentially affected by C-linker shortening.
Most interestingly, our studies demonstrate that substantial shortening of the C-linker by 8 and 10 amino acids constitutively activates ␣ L ␤ 2 . As evidenced by mutant ⌬ 10, which lacks residue Glu-310, this constitutive activation does not require the putative interaction of ␣L Glu-310 with the ␤I MIDAS (Fig.  1E). In agreement with this assumption, ⌬ 8 and ⌬ 10 were resistant to inhibition by the ␣/␤ I allosteric inhibitor XVA143. This class of inhibitor is thought to block the binding of ␣L residue Glu-310 to the ␤I MIDAS (5). However, the ⌬ 8 and ⌬ 10 mutants only marginally expressed the KIM127 activation epitope on the ␤ chain, which is normally exposed upon ␣I/␤I domain interaction (25). These results suggest that a "pull" triggered by shortening of the C-linker is sufficient to keep the ␣I domain in a high-affinity conformation independent of a regulative function of the ␤2 chain. Similarly, the previously described ␣L-E310C/␤2-A210C double mutant is constitutively active in the absence of Glu-310, does not express activation epitopes on the ␤ chain, and is resistant to XVA143 inhibition (although binding of the inhibitor to the mutant was demonstrated) (30). However, in the latter case, the pull keeping the I domain of ␣L-E310C/␤2-A210C in a high-affinity state is exerted by a disulfide connecting C-linker residue 310 to a residue in a ␤I MIDAS-coordinating loop (30). Intriguingly, ␣I allosteric inhibitors that bind underneath the ␣7-helix of the ␣I domain were still able to convert the constitutively active state of the ␣L-E310C/␤2-A210C mutant into an inactive state (30). The same property was observed for the C-linker ⌬ 8 and ⌬ 10 mutants in this study.
Inhibition by ␣⌱ allosteric antagonists demonstrates that there is still enough "play" in the ⌬ 8 and ⌬ 10 C-linker mutants to enable the reversion of the open ␣I domain conformation to the closed conformation. Because in ⌬ 8 and ⌬ 10 the C-linker is largely or completely removed, respectively, this play must come from somewhere else. This remaining play may come both from changes in ␣I domain orientation with respect to the ␤-propeller and ␤I domains and from unwinding of the C-terminal portion of the ␣7-helix. Unwinding of the C-terminal portion is feasible because crystal structures of isolated ␣ L I domains complexed with this class of inhibitors show interaction with residues located only in the N-terminal portion of the ␣7-helix (31). For example, the main contacts of LFA878 are formed with ␣7-helix residues Glu-301, Leu-302, and Lys-305 (31). Interestingly, the constitutively active ␣ L ␤ 2 mutant ␣ L -K287C/␣ L -K294C is resistant to inhibition by ␣I allosteric inhibitors (5,25). This latter mutant contains an ␣I domain which, in contrast to the above described mutants, is "locked" in the high-affinity form by a disulfide bond introduced prior to the ␣7-helix in the ␤6-␣7 loop (25). However, similarly to the C-linker mutants and ␣L-E310C/␤2-A210C, there was only little expression of the KIM127 epitope in this mutant (25).
Thus, constitutive activation of the ␣I domain via C-linker shortening or introduction of disulfide bonds is not associated with a global conformational change from the bent to the extended form. Does this mean that bent ␣ L ␤ 2 can be adhesive? Most likely this is not the case. All of the mutants described above express the KIM127 epitope to a certain degree, suggesting that some receptors are transiently extended and sampling extended conformational space. Even wild-type LFA-1 has basal adhesive activity. Taken together, our findings demonstrate that the ␣I domains of the C-linker mutants ⌬ 8 and ⌬ 10 are able to respond to ␣I allosteric inhibitors similarly to WT ␣ L ␤ 2 . Furthermore, ␣/␤ I allosteric inhibitors that bind to the MIDAS of the ␤I domain were still able to convert the ␤2 chain of the mutants from an inactive into an active conformation, as shown by the exposure of the KIM127 activation epitope. These observations demonstrate that intrachain conformational change within the ␣I domain and within the ␤I and hybrid domains is still preserved in the C-linker mutants, whereas communication between the ␣I and ␤I domains is disrupted.
Conformational changes because of C-linker shortening were also detectable by the anti-␣L mAbs R7.1 and CBR LFA-1/1. Our study indicates that the epitopes of these antibodies involve both the ␣L I domain and the ␤ propeller domain. Two results indicate that CBR LFA-1/1 requires the ␣I domain for binding: deletion of the ␣I domain (129 -308) led to greatly reduced reactivity, and mapping studies demonstrate that ␣I domain residues 250 -303 contribute to the epitope recognized by CBR LFA-1/1. The lack of reactivity with cell surface expressed ␣I domain alone shows that another component is required to constitute the CBR LFA-1/1 epitope. The reactivity of the antibodies with a cell surface-expressed ␣L fragment containing the ␣I domain, the C-linker, and part of the ␤-propeller domain (residues 130 -339) indicates that residues of the C-linker and ␤-propeller region may constitute this second component. This notion is supported by mapping studies that show that residues 301-442 (in this study) and residues 301-359 (as reported previously (9)) are involved in the epitope. The observation that CBR LFA-1/1 only marginally binds to the latter region in the absence of the ␣I domain in I-less ␣ L ␤ 2 are in line with this interpretation. Moreover, the dual specificity is consistent with the previous observation that CBR LFA-1/1 had properties intermediate between antibodies to the ␣I and ␤-propeller domains in its requirement for folding of these domains for immunoprecipitation (32). R7.1 similarly required both the ␣I and ␤-propeller domains for full reactivity but recognizes an overlapping but distinct epitope on the basis of mapping the human residues for which it is specific with chimeras.
C-linker shortening gradually reduced expression of the R7.1 and CBR LFA-1/1 epitopes, with the CBR-LFA-1/1 epitope more sensitive to shortening than the R7.1 epitope. Both epitopes were abolished after complete C-linker deletion. Intriguingly, the epitopes of R7.1 and CBR LFA-1/1 could be reconstituted by inserting the C-linker of an integrin that is not recognized by the antibodies, i.e. ␣ X ␤ 2 . The sequence exchanged is identical to the smallest segment of residues that was deleted to almost completely abrogate antibody binding (i.e. ⌬ 6). In other words, deletion of residues 313-318, SKQDLT, in ⌬ 6 abolished reactivity, and replacement with residues ETTSSS in CX-E restored activity. Thus, R7.1 and CBR LFA-1/1 do not appear to recognize this portion of the C-linker directly. Instead, they appear to recognize a combinatorial epitope comprised of portions of the ␣I and ␤-propeller domains and possibly conserved linker residues 309 -312. Thus, the length but not the sequence of C-linker residues 313-318 is important for enabling a specific orientation between the segments recognized by the antibodies.
Consistent with the assumption that R7.1 and LFA CBR-1/1 are sensitive to conformational changes in the C-linker region, binding of both antibodies was perturbed by allosteric inhibitors that target this area. Notably, CBR LFA-1/1 blocks ligand binding by wild-type ␣ L ␤ 2 but not by the disulfide-locked highaffinity mutant, suggesting inhibition by an allosteric mode of action (25). Our studies establish R7.1 and CBR LFA-1/1 as a new class of ␣ L ␤ 2 antibodies sensitive to conformational change across the important interface between the ␣I and ␤-propeller domains and as important tools to study conformational alterations within the C-linker region of ␣ L ␤ 2 . In contrast, the allosteric inhibitory ␣ L ␤ 2 antibodies TS2/14 and 25-3-1 have been mapped to species-specific residues contained wholly within the ␣ L I domain in the ␤5-␣6 loop and ␣6 helix (24).
Functionally, the replacement of the ␣ L ␤ 2 C-linker by the ␣ X ␤ 2 C-linker, altering residues 313-318, resulted in mutants with properties comparable with the WT. This result indicates that the length of the C-linker is more critical for its function than the sequence. In support of this notion we also found that single point mutations in the C-linker region (K314A and L317A) did not affect expression and function of ␣ L ␤ 2 in 293T cells (data not shown). It is not known why this normal pheno-type in 293T cells differs from the constitutively active phenotype of the same mutants reported in COS-7 cells (28). It is noteworthy, however, that a similar discrepancy between COS-7 cells and another cell line has been reported with an ␣ L ␤ 2 receptor mutant before (33). Our findings suggest that the C-linker functions as a spring-like element that allows relaxation of the ␣I domain in the resting state and controlled tension of the ␣I domain during activation. In agreement with this notion, the crystal structure of the ␣I domain integrin ␣ X ␤ 2 in the bent, inactive conformation reveals that the C-linker is not in contact with the surrounding regions (7). The C-linker polypeptide on the ␣ subunit is not the only spring-like element in integrins that fine-tunes an equilibrium between the resting and active states. Recently, it has been demonstrated in ␤3 integrins that the length of a loop expressed on the ␤ subunit at the integrin "knee" modulates the equilibrium between lowand high-affinity states of these integrins (34). In fact, disordered regions such as the C-linker are quite common in eukaryotic proteins. They can be found in transcription factors and cell signaling molecules (35). A recent mechanistic model suggests that intramolecular site-to-site allosteric coupling is optimized when intrinsic disorder is present in the domains containing one or both of the coupled sites (36). Thus, it appears to be an emerging theme that polypeptide stretches such as the C-linker in ␣ L ␤ 2 are critical elements for conformational flexibility and functionality. It is intriguing to speculate that such disordered regions and their vicinity could become new targets for drug development and therapeutic intervention.