Conformational Changes in the Integrin (cid:1) A Domain Provide a Mechanism for Signal Transduction via Hybrid Domain Movement*

The ligand-binding head region of integrin (cid:1) subunits contains a von Willebrand factor type A domain ( (cid:1) A). Ligand binding activity is regulated through conformational changes in (cid:1) A, and ligand recognition also causes conformational changes that are transduced from this domain. The molecular basis of signal transduction to and from (cid:1) A is uncertain. The epitopes of mAbs 15/7 and HUTS-4 lie in the (cid:1) 1 subunit hybrid domain, which is connected to the lower face of (cid:1) A. Changes in the expression of these epitopes are induced by conformational changes in (cid:1) A caused by divalent cations, function perturbing mAbs, or ligand recognition. Recombinant truncated (cid:2) 5 (cid:1) 1 with a mutation L358A in the (cid:2) 7 helix of (cid:1) A has constitutively high expression of the 15/7 and HUTS-4 epitopes, mimics the conformation of the li-gand-occupied receptor, and has high constitutive ligand binding activity. The epitopes of 15/7 and HUTS-4 map to a region of the hybrid domain that lies close to an interface with the (cid:2) subunit. Taken together, these data suggest that the transduction of conformational changes through (cid:1) A involves shape shifting in the (cid:2)

Integrins mediate a wide variety of essential cell-matrix and cell-cell interactions and also participate in many common disease processes (1,2). Integrins are heterodimers containing non-covalently associated ␣ and ␤ subunits; each subunit has a large extracellular domain linked to a transmembrane segment and a short cytoplasmic tail. Integrins participate in bi-directional signaling; ligand recognition is dynamically regulated by "inside-out" signaling, and ligand occupancy leads to "outsidein" signals that affect cell migration, growth, differentiation, and survival (3)(4)(5). Modulation of integrin activity is essential in such processes as leukocyte migration to sites of tissue injury and the aggregation of platelets to form a hemostatic plug. Integrin activation can be mimicked in vitro by divalent cations such as Mn 2ϩ or Mg 2ϩ (6). Three major conformational states of integrins can be distinguished using monoclonal an-tibodies (mAbs) 1 : an inactive (resting or low affinity) state, an active (or high affinity) state, and a ligand-occupied state (7). The conformations of the inactive and active states are discriminated by low and high expression, respectively, of activation epitopes (such as those recognized by 12G10, 15/7, and 9EG7 for the ␤ 1 subunit, see Refs. 8 -10). The ligand-occupied conformer expresses high levels of ligand-induced binding site (LIBS) epitopes (which are generally also activation epitopes) and shows decreased expression of ligand-attenuated binding site (LABS) epitopes (such as mAb 13 for the ␤ 1 subunit, see Ref. 11). The conformational states are in equilibrium; therefore, antibodies that recognize activation epitopes or LIBS tend to cause activation and stabilize the ligand-occupied state. Conversely, antibodies that recognize LABS appear to block ligand binding by preventing conformational changes involved in ligand recognition (7,11,12).
The molecular basis of integrin function has been powerfully elucidated by the recent x-ray crystal structures of the extracellular domains of ␣ V ␤ 3 in both an unliganded state (13) and in complex with a small peptide ligand (14). Overall, the integrin structure resembles that of a "head" on two "legs." The ligand-binding head region of the integrin contains a sevenbladed ␤-propeller in the ␣ subunit, the top face of which is in close juxtaposition with a von Willebrand factor type A domain in the ␤ subunit (␤A). ␤A consists of seven ␣ helices encircling a central ␤-sheet and is connected at its N and C termini to an immunoglobulin-like "hybrid" domain and forms an extensive interface with it. The key regions involved in ligand recognition are loops on the upper surface of the ␤-propeller and the upper face of the ␤A, which contains a metal ion-dependent adhesion site (MIDAS) and an adjacent MIDAS cation-binding site (13)(14)(15). A small number of subtle conformational changes between the unliganded and liganded states were observed. The most important of these appeared to be a shift of the ␣1 helix in ␤A, and a slight closing up of the interface between the upper surface of the ␤-propeller and the upper face of the ␤A. A surprising feature of the crystal structures was that the two legs are severely bent at the "knees," such that the head is in close contact with lower legs. Because the peptide ligand was soaked into the crystals of unliganded ␣ V ␤ 3 , it is unclear whether the small conformational changes observed between the unliganded and liganded structures (13,14) are representative of those that take place upon ligand occupancy of the * This work was supported in part by grants from the Wellcome Trust (to M. J. H.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 1 The abbreviations used are: mAb, monoclonal antibody; ␤A, ␤ subunit von Willebrand factor type A domain; ␣A, ␣ subunit von Willebrand factor type A domain; MIDAS, metal-ion dependent adhesion site; BSA, bovine serum albumin; tr␣ 5 ␤ 1 -Fc, recombinant soluble integrin heterodimer containing C-terminally truncated ␣ 5 and ␤ 1 subunits (␣ 5 residues 1-613 and ␤ 1 residues 1-455) fused to the Fc region of human IgG␥1; LIBS, ligand-induced binding site; LABS, ligand-attenuated binding site. native integrin. Importantly, no pathway for the transduction of conformational changes from the head to the legs (or from legs to head) was evident. Hence, the molecular basis of both outside-in and inside-out signaling remains to be clarified.
Recently, evidence (16,17) has been presented that the bent form of the integrin is in the inactive state and that this may undergo a switchblade-like straightening to attain the active conformation. Nevertheless, precisely how this straightening is linked to activation of ligand binding in the head domain is uncertain. A key regulator of integrin activity is known to be the conformation of the ␤A domain (15,18), and we have shown that a movement of the ␣1 helix activates this domain (8). We hypothesized that the ␣1 helix could occupy two different positions: position 1 characterized by high binding of the mAb 12G10 (the active conformation), and position 2 characterized by low binding of 12G10 (the inactive conformation). The inward movement of ␣1 helix observed in the liganded ␣ V ␤ 3 structure (14) supports this proposal. Hence, position 1 appears to correspond to the "in" state and position 2 to the "out" state of the ␣1 helix. About half of the integrin ␣ subunits contain a similar domain (␣A or I), and in these domains an inward movement of the ␣1 helix is linked to rearrangement of cationcoordinating residues at the MIDAS and a dramatic downward shift of the C-terminal ␣7 helix and its preceding loop (19). However, no change in the position of the ␣7 helix of ␤A was observed between the two x-ray structures, and it was suggested that activation of ␤A does not involve ␣7 movement (13)(14)20).
Here we provide evidence that changes in the expression of activation epitopes in the hybrid domain are linked to shapeshifting in the ␣7 helix region of ␤A. This movement appears to participate in the conformational changes involved in both activation and ligand binding. Our data suggest that an outward swing of the hybrid domain is coupled to ␣7 helix motion, and hence lend support to a recent model of integrin activation (21). There are both strong similarities and some differences between ␤A and ␣A domain activation.
Expression Vector Construction and Mutagenesis-C-terminally truncated human ␣ 5 and ␤ 1 constructs encoding ␣ 5 residues 1-613 and ␤ 1 residues 1-455 fused to the hinge regions and C H 2 and C H 3 domains of human IgG␥1 (␣ 5 -(1-613)-Fc and ␤ 1 -(1-455)-Fc) were generated as described previously (24). To aid the formation of heterodimers, the C H 3 domain of the ␣ 5 construct contained a "hole" mutation, whereas the C H 3 domain of the ␤ 1 constructs carried a "knob" mutation as described (24,25). The L358A and S359A mutations in the ␤ 1 subunit were carried out using oligonucleotide-directed PCR mutagenesis, as described (24). Oligonucleotides were purchased from MWG Biotech (Southampton, UK). The presence of the mutations (and the lack of any other changes to the wild-type sequence) was verified by DNA sequencing.
For comparison of purified wild-type heterodimers with heterodimers containing the L358A or S359A mutations in ␤ 1 , 75-cm 2 flasks of sub-confluent CHOL761h cells were transfected with 5 g of wild-type or mutant ␤ 1 -(1-455)-Fc and 5 g of ␣ 5 -(1-613)-Fc DNA as described above. After 4 days, culture supernatants were harvested by centrifugation at 1000 ϫ g for 5 min. Wild-type or mutant heterodimers were purified using protein A-Sepharose essentially as described before (24).
Proteins-A recombinant fragment of fibronectin containing type III repeats 6 -10 (III 6 -10 ) was produced and purified as described previously (12). A mutant fragment in which the RGD integrin-binding sequence is replaced by the inactive sequence KGE (III 6 -10 KGE, see Ref. 26) was produced and purified in the same manner. III 6 -10 was biotinylated as before (8) using sulfo-LC-NHS biotin (Perbio, Chester, UK). Fab fragments of N29, TS2/16, and 12G10 were prepared by ficin cleavage of purified IgG, followed by removal of Fc-containing fragments using protein A-Sepharose, according to the manufacturer's instructions (Perbio). None of the Fab fragments showed any reactivity with goat anti-mouse IgG (Fc-specific) peroxidase conjugate (Sigma).
Effect of Divalent Cations on 15/7 and HUTS-4 Binding-96-Well plates (Costar 1 ⁄2-area EIA/RIA, Corning Science Products, High Wycombe, UK) were coated with goat anti-human ␥1 Fc (Jackson Immunochemicals, Stratech Scientific, Luton, UK) at a concentration of 2.6 g/ml in Dulbecco's phosphate-buffered saline (50 l/well) for 16 h. Wells were then blocked for 1Ϫ3 h with 200 l of 5% (w/v) BSA, 150 mM NaCl, 0.05% (w/v) NaN 3 , 25 mM Tris-Cl, pH 7.4 (blocking buffer). Blocking buffer was removed, and supernatant from cells transfected with wild-type ␣ 5 -(1-613) and ␤ 1 -(1-455)-Fc diluted 1:1 with 150 mM NaCl, 25 mM Tris-Cl, pH 7.4 (25 l/well), was added for 1-2 h at room temperature. Wells were then washed three times with 200 l of 150 mM NaCl, 25 mM Tris-Cl, pH 7.4, containing 1 mg/ml BSA (buffer A). Buffer A was treated with Chelex beads (Bio-Rad) to remove any small contaminating amounts of endogenous Ca 2ϩ and Mg 2ϩ ions. mAbs (1 g/ml) in buffer A with varying concentrations of Mn 2ϩ , Mg 2ϩ , or Ca 2ϩ were added to the plate (50 l/well). The plate was then incubated at 30°C for 2 h. Unbound antibody was aspirated, and the wells were washed three times with buffer A. Bound antibody was quantitated by addition of 1:1000 dilution of goat anti-mouse IgG (Fc-specific) peroxidase conjugate (Jackson Immunochemicals) in buffer A for 30 min at room temperature (50 l/well). Wells were then washed four times with buffer A, and color was developed using 2,2Ј-azinobis(3-ethylbenzothiazoline-6-sulfonic acid) substrate (50 l/well). Absorption at 405 nm was measured using a plate reader (Dynex Technologies). Background binding to mAbs to wells incubated with supernatant from mock-transfected cells was subtracted from all measurements. Measurements obtained were the mean Ϯ S.D. of four replicate wells.
For comparison of the effects of divalent cations on 15/7 and HUTS-4 binding to wild-type heterodimer and the L358A and S359A mutants, plates were coated with anti-human Fc, blocked as described above, and then incubated with supernatant from cells transfected with wild-type or mutant heterodimers. mAb binding was measured as described above in 2 mM EDTA, 2 mM Mn 2ϩ , 2 mM Mg 2ϩ , or 2 mM Ca 2ϩ . Measurements obtained were the mean Ϯ S.D. of four replicate wells.
Effect of mAbs and Ligand on 15/7 and HUTS-4 Binding-Plates were coated with anti-human Fc and blocked as described above. Wells were then incubated with supernatant from cells transfected with wildtype or mutant heterodimers for 1-2 h at room temperature as above. Wells were washed three times with 200 l of 150 mM NaCl, 1 mM MnCl 2 , 25 mM Tris-Cl, pH 7.4, containing 1 mg/ml BSA (buffer B). 15/7 or HUTS-4 (l g/ml in buffer B) was added to the plates (50 l/well) either alone or in the presence of Fab fragments of N29, TS2/16, or 12G10 (5 g/ml), mAb 13 IgG (10 g/ml), or III 6 -10 (20 g/ml). The plates were then incubated at 30°C for 2 h. Unbound antibody was aspirated, and the wells were washed three times with buffer B. Bound 15/7 or HUTS-4 was quantitated by addition of 1:2000 dilution of goat anti-mouse IgG (Fc-specific, precleared with rat serum proteins) peroxidase conjugate (Sigma) in buffer B for 30 min at room temperature (50 l/well). Wells were then washed four times with buffer A, and color was developed as above. Background binding of mAbs to wells incubated with supernatant from mock-transfected cells was subtracted from all measurements. Measurements obtained were the mean Ϯ S.D. of four replicate wells.
Comparison of Epitope Expression by Wild-type and Mutant Heterodimers-Plates were coated with anti-human Fc and blocked as described above. The blocking solution was removed, and cell culture supernatants were added (25 l/well) for 1-2 h. All supernatants were assayed in triplicate, and supernatant from mock-transfected cells was used as a negative control. The plate was washed 3 times with buffer B (200 l/well), and anti-␣ 5 or anti-␤ 1 mAb (5 g/ml) was added (50 l/well). The plate was incubated for 2 h and then washed 3 times in buffer B. Peroxidase-conjugated anti-rat or anti-mouse secondary antibodies (1:1000 dilution in buffer B; Jackson Immunochemicals) were added (50 l/well) for 30 min, and the plate was washed four times in buffer B, and color was developed as above. All steps were performed at room temperature. Results shown are the mean Ϯ S.D. of three separate experiments.
Effect of L358A and S359A Mutations on III 6 -10 Binding-Plates were coated with anti-human Fc and blocked as described above. Wells were then incubated with protein A-purified heterodimers diluted to ϳ1 g/ml with 150 mM NaCl, 25 mM Tris-Cl, pH 7.4 (25 l/well), for 1-2 h at room temperature. Wells were washed three times with 200 l of buffer B. Biotinylated III 6 -10 (0.1 g/ml) in buffer B was added to the plate (50 l/well) alone or in the presence of N29, TS2/16, or 12G10 (5 g/ml). The plate was then incubated at 30°C for 2 h. Unbound ligand was aspirated, and the wells were washed three times with buffer B. Bound ligand was quantitated by addition of 1:500 dilution of ExtrAvidin peroxidase conjugate (Sigma) in buffer B for 20 min at room temperature (50 l/well). Wells were then washed four times with buffer B, and color was developed using 2,2Ј-azinobis(3-ethylbenzothiazoline-6-sulfonic acid) substrate (50 l/well). Background binding to BSA-coated wells was subtracted from all measurements. Measurements obtained were the mean Ϯ S.D. of four replicate wells.
Mapping of the 15/7 and HUTS-4 Epitopes-Substitution of human residues with the corresponding residues in murine ␤ 1 within the hybrid domain sequence 361-425 was performed using a PCR-based mutagenesis kit (Gene Tailor, Invitrogen) according to the manufacturer's instructions. CHOL761h cells were transfected with wild-type or mutant constructs and supernatants harvested as described above. Binding of 15/7, HUTS-4, and TS2/16 to mutant heterodimers was performed as described above, relative to the wild-type control. Background binding to mAbs to wells incubated with supernatant from mock-transfected cells was subtracted from all measurements. Measurements obtained were the mean Ϯ S.D. of three replicate wells. Results shown are mean Ϯ S.D. of three separate experiments.
In each assay involving a comparison between different heterodimers, the binding of mAb 8E3 (5 g/ml) was used to normalize any differences between the amounts of the different heterodimers bound to the wells. For example, normalized A 405 for 15/7 binding ϭ (AM 15/7 Ϫ Am 15/7 ) ϫ ((AWT 8E3 Ϫ Am 8E3 )/(AM 8E3 Ϫ Am 8E3 )), where AM 15/7 ϭ mean absorbance of wells coated with mutant integrin; Am 15/7 ϭ mean absorbance of wells coated with mock supernatant; AWT 8E3 ϭ mean absorbance of 8E3 binding to wells coated with wild-type integrin; Am 8E3 ϭ mean absorbance of 8E3 binding to wells coated with mock supernatant, and AM 8E3 ϭ mean absorbance of 8E3 binding to wells coated with mutant integrin. 8E3 recognizes a non-functional epitope in the N-terminal region of the ␤ 1 subunit (24). Essentially identical results were obtained from normalization using mAb N29 against the PSI domain (Ref. 27, data not shown). In experiments using heterodimers captured from cell culture supernatants, similar results were obtained using protein A-purified heterodimers (data not shown). Each experiment shown is representative of at least three separate experiments.
Homology Modeling of the Head Region of ␣ 5 ␤ 1 -A model of the ␣ 5 -propeller and thigh domains and ␤ 1 A and hybrid domains was built based on an alignment against the ␣ V ␤ 3 crystal structure (13), using the same procedures as described previously (8). The PSI domain (residues 1-60 of ␤ 1 ) was not included in the model. Representation of the structure was produced using PyMol. 2

Expression of the 15/7 and HUTS-4 Epitopes Is Regulated by
Conformational Changes in the ␤A Domain-To investigate the mechanisms of integrin activation, we employed a recently described system for expression of recombinant soluble ␣ 5 ␤ 1 (24). For these particular studies, we have used a truncated version of ␣ 5 ␤ 1 , ␣ 5 -(1-613) ␤ 1 -(1-455), fused to the Fc region of human IgG␥1 (24) (hereafter referred to as tr␣ 5 ␤ 1 -Fc). This heterodimer contains the ␣ subunit ␤-propeller and thigh domain, and the ␤ subunit A, hybrid, and PSI domains (13), and has been shown to retain the properties of the full-length receptor (24). This system is particularly useful (a) because it permits the rapid analysis of the effects of mutations, and (b) because conformational changes in the head region can be studied in isolation, i.e. in the absence of any complicating effects due to the presence of the lower leg domains (e.g. unbending, see Refs. 16 and 17) or the cytoplasmic tails (5).
Activation of the integrin head is known to involve conformational changes in ␤A, and since ␤A is connected at its N and C termini to the hybrid domain, these changes must be transduced from and to the hybrid. HUTS-4 and 15/7 are two previously characterized mAbs whose epitopes lie within this region of the ␤ 1 subunit (9, 29 -31). Expression of the 15/7 and HUTS-4 epitopes by tr␣ 5 ␤ 1 -Fc was found to be cation-modulated (Fig. 1, A and B). Binding of each mAb was promoted by Mn 2ϩ and to a smaller extent by Mg 2ϩ , whereas Ca 2ϩ did not stimulate binding. Importantly, these effects parallel the effects of each divalent ion on the ligand-binding competence of the integrin (32), and they also mirror a conformational change in the ␤A domain reported by mAb 12G10 (8). These changes have been shown to be due to cation binding to the MIDAS (8), and in agreement with this, the MIDAS mutation D130A prevented the cation modulation of 15/7 and HUTS-4 binding (data not shown). Conformational changes in ␤A can also be induced by function-perturbing mAbs with epitopes in this domain. The epitopes of all function-altering anti-human ␤ 1 mAbs that map to the A domain include one or more residues in the sequence Asn 207 -Lys 218 (33), which is predicted to form the ␣2 helix region (13). The epitope of mAb 12G10 also includes two arginyl residues that lie near the base of the ␣1 helix (8). TS2/16 and 12G10 are examples of activating mAbs, whereas 13 is an example of a function-blocking mAb (11). The mAb N29, whose epitope lies in the PSI domain, was used as a control. 3 15/7 and HUTS-4 binding to tr␣ 5 ␤ 1 -Fc was increased by TS2/16 and 12G10 but markedly decreased by mAb 13 (Fig. 2). Hence, conformational changes in the ␣1/␣2 helix region of ␤A appear to modulate 15/7 and HUTS-4 binding. As ligand recognition is also known to cause shape-shifting in ␤A and to stabilize the active conformation of this domain (8), we tested the effect of ligand binding on the expression of the 15/7 and HUTS-4 epitopes. A recombinant fragment of fibronectin containing the ␣ 5 ␤ 1 recognition sites (26) stimulated 15/7 and HUTS-4 binding (Fig. 2) to a similar extent as TS2/16 and 12G10.
Taking these data together, the active conformation of ␤A (stabilized by Mn 2ϩ /Mg 2ϩ , activating mAbs, or ligand) leads to increased expression of the 15/7 and HUTS-4 epitopes. Conversely, the inactive conformation of ␤A (stabilized by Ca 2ϩ or function-blocking mAbs) leads to decreased expression of the 15/7 and HUTS-4 epitopes. These effects may be linked to a motion of the ␣1 helix (8). Our data on the effects of cations, function-perturbing mAbs, and ligand are in broad agreement with previous characterization of 15/7 and HUTS-4 as mAbs recognizing activation/LIBS epitopes (9,29,31).
A Mutation in the ␤A Domain ␣7 Helix (L358A) Results in Constitutively High Expression of the 15/7 and HUTS-4 Epitopes-The transduction of conformational changes from ␤A to the hybrid domain must take place at the interface between these two modules. At its C terminus ␤A is joined to the hybrid domain by the ␣7 helix, and mutations in this region of ␣ subunit A domains cause activation by favoring a downward shift of ␣7 (34,35). We therefore tested whether similar mutations in the ␣7 helix of ␤A could affect 15/7 and HUTS-4 binding. A mutation Leu 358 to Ala was found to cause increased expression of the 15/7 and HUTS-4 epitopes, whereas the control mutation S359A had little effect (Figs. 3 and 4). Expression of the 15/7 and HUTS-4 epitopes by the L358A mutant was constitutively high compared with the wild-type receptor, and Mn 2ϩ /Mg 2ϩ only caused a small enhancement of the level of binding seen in the absence of divalent ions (Fig. 3, A and B). The high level of 15/7 binding to the L358A mutant was only slightly increased by activating mAbs TS2/16 or 12G10 and was relatively resistant to inhibition by mAb 13 (Fig. 4); similar results were obtained for HUTS-4 (data not shown). In contrast to the wild-type receptor, ligand binding did not increase 15/7 epitope expression by the L358A mutant (Fig. 4). The mutation S359A had little effect on the ability of mAbs or ligand to modulate 15/7 binding (results similar those for wild-type tr␣ 5 ␤ 1 -Fc, see Fig. 2). These findings suggest that the L358A mutation constrains the ␤A domain in an active conformation and that the ␣7 helix region is involved in conveying conformational changes from ␤A to the hybrid domain.
The L358A Mutant Mimics the Ligand-occupied Conformation-We next tested whether the L358A mutant caused any other conformational changes associated with activation or ligand binding. For this purpose we used a panel of mAbs recognizing epitopes on both the ␣ 5 and ␤ 1 subunits (Fig. 5A). The results showed that the L358A mutation increased the expression of the 12G10 epitope and decreased the expression of the 13 and 4B4 epitopes in ␤A. The mutation also attenuated the expression of the epitopes of function-blocking mAbs SNAKA52, 16, and P1D6, which lie at or near the top of the ␣ 5 ␤-propeller domain (23), close to the ␤A/propeller interface (13). Although, as shown above, the L358A mutation increased the expression of the 15/7 and HUTS-4 epitopes, it did not alter the expression of the JB1A epitope, which maps to a different region of the hybrid domain (36). Furthermore, the expression of epitopes that are not affected by the activation state, such as TS2/16 and JB1A, was not altered by the L358A mutation. The control mutation S359A had no significant effect on the expression of any of the epitopes tested (Fig. 5B). Taken together, these results show that the L358A mutation specifically increases the expression of all activation/LIBS epitopes (12G10, 15/7, and HUTS-4, see Refs. 8, 9, and 31) and decreases the expression of all LABS epitopes (SNAKA52, 4 16, P1D6, 13, and 4B4, see Refs. 11, 12, and 37). Hence, the data suggest that L358A mutant adopts a conformation that is similar to the ligand-occupied state. Furthermore, a conformational change in the ␣7 helix region of ␤A caused by the mutation appears to be linked to movements in the ␣1/␣2 helix region (the location of the 12G10, 13, and 4B4 epitopes) and the proximity of the ␤A/propeller interface (the location of the SNAKA52, 16, and P1D6 epitopes).
The L358A Mutation Causes Activation of Ligand Binding-If the conformation of the L358A mutant is akin to the ligand-occupied state, it would be predicted that this mutant should be constitutively active for ligand binding (38). Tr␣ 5 ␤ 1 -Fc has low constitutive ligand binding activity when captured onto enzyme-linked immunosorbent assay plates using goat anti-human Fc, but the same protein has similar activity to recombinant integrin containing the complete extracellular domains of ␣ 5 and ␤ 1 when stimulated with mAbs such as 12G10 (24). We compared the ligand binding activity of wild-type tr␣ 5 ␤ 1 -Fc with the L358A and S359A mutants (Fig.  6). The results showed that compared with the wild-type receptor, the L358A mutant had high constitutive ligand binding activity, which was only slightly enhanced by activating mAbs TS2/16 or 12G10. In contrast, the S359A mutant had constitutively low activity, similar to wild-type levels.
The Epitopes of 15/7 and HUTS-4 Map to a Region of the Hybrid Domain Close to an Interface with the ␣ Subunit-The above results suggest that the transduction of conformational changes from ␤A to the hybrid domain involves a shift of the ␣7 helix. To understand these changes more fully, we fine-mapped the epitopes of 15/7 and HUTS-4. Both antibodies bind to human ␤ 1 but not to mouse ␤ 1 , and their epitopes have been shown to reside within amino acid residues 355-425 (30,31). These residues in human ␤ 1 show 10 differences with the equivalent sequence in murine ␤ 1 (Table I) using TS2/16 as a control. The mutations E371D and K417N abrogated both 15/7 and HUTS-4 binding, whereas the mutation S370P completely blocked 15/7 binding and strongly inhibited HUTS-4 binding. The other mutations had either no effect or showed a partial inhibition ( Table I). None of the mutations affected binding of TS2/16. Ser 370 and Glu 371 map to the C-D loop, and Lys 417 maps to the neighboring E-F loop of the hybrid domain (13). The spatial proximity of this triplet of residues is consistent with them forming an antibody epitope.
Comparison with the crystal structures of ␣ V ␤ 3 (13,14) shows that these epitopes map to a region of the hybrid domain that faces the ␣ subunit ␤-propeller and are very close to residues that form a small interface with it. To estimate the antibody-accessible surface, we rolled a 20-Å sphere over the structure (16). The results (not shown) demonstrated that Lys 417 would be accessible in this conformational state, but Ser 370 and Glu 371 would not. However, Ser 370 and Glu 371 would be available for antibody binding if the hybrid domain moves away from the ␤-propeller.

DISCUSSION
By using the conformation-sensitive mAbs 15/7 and HUTS-4 and site-directed mutagenesis, we have studied conformational changes in integrin ␤ 1 A domain, and we investigated how these relate to signal transduction in the integrin head region. Our results show the following: (i) ␤A domain activation involves a conformational change in the region of ␣7 helix; (ii) this shape-shifting results in increased exposure of the 15/7 and HUTS-4 epitopes in the hybrid domain and is also associated with other conformational changes in ␤A and the top of the ␣ subunit ␤-propeller; (iii) the 15/7 and HUTS-4 epitopes map to a portion of the hybrid domain that is likely to be partly masked (in the inactive receptor) due to its close proximity to the ␤-propeller. Taking these results together with previous data showing that a movement of the ␣1 helix is important for activation of ␤A (8), we propose the model of affinity regulation shown in Fig. 7.
Mechanism of ␤A Activation-Recent debate concerning the mechanism of activation of ␤A has centered on whether or not the ␣7 helix moves. Three distinct models have been proposed. (i) Based on the crystal structures of ␣ V ␤ 3 (13,14) in which there is no movement of ␣7, and the MIDAS site is unoccupied in the absence of ligand, it was suggested that ␤A is regulated by unmasking of the MIDAS site (20,39). (ii) Based on a separation of the head domains observed in RGD-occupied ␣ IIb ␤ 3 after rotary shadowing/electron microscopy (40), and on a similarity between the quaternary structures of integrins and heterotrimeric G-proteins (13), it was hypothesized that activation involves a rotation of ␤A away from its contact with the ␤-propeller (41). In this scenario, the position of the ␣7 helix is fixed but the rotation of ␤A resulted in the same net movement of ␣7 seen in ␣A domains. (iii) Based on studies of inactive (Ca 2ϩ -occupied) and active (Mn 2ϩ -occupied) ␣ V ␤ 3 by negative staining/electron microscopy (17), it was suggested that activa-tion of the head region is regulated by an outward swing of the hybrid domain, which is predicted to be coupled to downward shift of the ␣7 helix equivalent to that in ␣A domains (19,42). Model i appears unlikely because our results (this work and see Ref. 8) and the results of others (17,18) show that conformational changes take place through cation binding to the MIDAS in the absence of ligand occupancy. Model ii is improbable because ␤A and the ␤-propeller appear to move closer together, rather than farther apart, upon ligand recognition because ligand binding requires close apposition of residues on both subunits (14,39). Instead, our data supply strong support for model iii because they suggest that a movement in the ␣7 helix region is important for activation and that this movement is linked to a change in the position of the hybrid domain such that it moves away from the ␣/␤ subunit interface. The existence of the swing-out motion of the hybrid is further supported    reactivity with ␤ 1 hybrid domain substitution mutants CHO L761h cells were transfected with ␣ 5 -(1-613)-Fc and wild-type or mutant ␤ 1 -(1-455)-Fc. Cell culture supernatants were analyzed for reactivity with anti-␤ 1 mAbs by sandwich enzyme-linked immunosorbent assay. Results are expressed as a percentage of wild-type binding and are mean Ϯ S.D. from three separate experiments (except for the S422T mutant, from two separate experiments). A value of 0% indicates that mAb reactivity was identical to, or slightly lower than, reactivity with supernatant from mock-transfected cells. All the mutants bound well to the hybrid domain mAb JB1A, and none of the mutations affected recognition of the III 6 -10 fragment of fibronectin (data not shown). by the finding that a section of the ␤ 3 hybrid domain encompassing residues 393-423 (equivalent to residues 402-432 in ␤ 1 ) is exposed in the active but not the resting form of ␣ IIb ␤ 3 (43). This portion of the hybrid domain encompasses part of the 15/7 and HUTS-4 epitopes (Lys 417 ). Why was no movement of the ␣7 helix seen in the crystal structure of the liganded form of ␣ V ␤ 3 (13)? The likely explanation is that motion of ␣7 would be prevented because the hybrid domain is paralyzed by lattice contacts and by its contacts with the leg domains (16,17). Some conformational changes are observed in the liganded ␤A domain; these include rearrangement of the loops that coordinate the MIDAS cation, leading to an inward movement of the ␣1 helix. These changes are very similar, both in direction and form, to those seen in ␣A domains; however, as pointed out above, the subsequent downward motion of the ␣7 helix that takes place in ␣A domains is probably prohibited. Thus, the structural changes observed in the liganded ␤A also favor the hypothesis that the unliganded structure represents the inactive form of ␤A (44).
We found that a mutation in the ␣7 helix, L358A, caused activation of the ␤A domain. In ␣A domains mutation of a highly conserved isoleucine residue at the same position favors the active state. This is apparently because this residue fits into a hydrophobic pocket (known as "socket for isoleucine," SILEN), and this interaction favors the inactive form (34). However, unlike the ␣A domains, mutation of residues that form the hydrophobic pocket surrounding the ␣7 helix in the ␤ 1 A domain (Leu 125 , Leu 149 , Leu 253 , and Ile 314 ) did not affect the activation state of tr␣ 5 ␤ 1 -Fc. 5 Hence, the mechanism that regulates ␣7 movement may differ slightly from that of ␣A domains. Nevertheless, mutation of Leu 358 may favor the active state of ␤A because this residue is likely to be more exposed in the "down" position of the ␣7 helix than in the "up" position. Hence, mutation of the leucine residue to the less hydrophobic alanine would be predicted to lower the energy of the active state. We cannot rule out the possibility that the L358A mutation activates the receptor by altering the ␤A/hybrid interface. However, mutation of Ser 359 (also at the interface) did not cause activation, and furthermore, this paradigm would not explain how activating anti-␤A mAbs cause activation and hybrid domain movement in a similar manner to the L358A mutation. In contrast, a linked movement of the ␣1 and ␣7 helices as seen in ␤A domains can explain the mechanism of action of the anti-␤A domain mAbs (see below).
Similarities and Differences between Activation of ␤A and ␣A Domains-In the activation of ␣A domains a change in cation coordination at the MIDAS is linked to an inward movement of the ␣1 helix. This movement pinches the hydrophobic core, squeezing out residues in the loop that precedes the ␣7 helix, and results in an ϳ10-Å downward motion of ␣7 (19). A similar link between ␣1 and ␣7 helix movement in ␤A is suggested by our findings. For example, occupancy of the MIDAS by Mn 2ϩ or the binding of activating mAbs, which stabilize the in position of the ␣1 helix (8), also promotes the downward motion of the ␣7 helix (reported by increased exposure of the 15/7 and HUTS-4 epitopes).
Although the overall mechanisms of ␤A and ␣A activation now appear to be closely related, there are some subtle differences. For example, for ␣A domains Mn 2ϩ and Mg 2ϩ are equally effective for promoting ligand binding to the MIDAS (45,46). In contrast, Mn 2ϩ is much more effective than Mg 2ϩ for promoting ligand binding to the ␤A MIDAS (8). This property of Mn 2ϩ may be due to the fact that it binds with much higher affinity than Mg 2ϩ to the ␤A MIDAS, whereas the affinities of ␣A domains for these two ions are more comparable (44,45). Similar to ␣A domains, movement of the ␣7 helix appears to form an essential part of the activation mechanism of ␤A because ␣7 movement closely parallels the activation state. However, as noted above, the regulation of ␣7 motion may be slightly different to that in ␣A domains.
Allosteric Mechanism of Function-perturbing mAbs-We have shown previously (11,12,37) that most function-blocking anti-␣ 5 and anti-␤ 1 mAbs have an allosteric mode of action. They recognize epitopes attenuated by ligand recognition and appear to perturb ligand binding by preventing a conformational change involved in the formation of the integrin-ligand complex (7). The epitopes of function-blocking anti-␤ 1 A domain mAbs map to the ␣2 or ␣1 helix regions (13,33,47), which lie adjacent to each other. Similarly, the epitopes of functionblocking anti-␤ 2 A domain mAbs have been shown to include residues in the ␣1/␣2 or ␣7 helix regions (48,49). It is likely that these mAbs function allosterically by stabilizing ␣ 1 in the inactive (out) location and/or ␣7 in the inactive (up) position. The epitopes of function-blocking anti-␣ subunit mAbs map to loops on the top face of the ␤-propeller domain (indicated by arrowheads in Fig. 7), and the binding of these mAbs may prevent conformational rearrangements required for ligand binding (14). Hence both anti-␣ and anti-␤ subunit functionblocking mAbs appear to impede conformational changes involved in ligand recognition, as proposed previously (7).
The epitopes of activating anti-␤A domain mAbs map to the same regions as inhibitory mAbs (8,18,33) and are likely to function by stabilizing ␣1 in the active (in) location and/or ␣7 in the active (down) position. It has been suggested that 12G10 activates by affecting the ␤A/hybrid domain interface (44), rather than by an effect on the ␣1 helix (8). However, we consider this proposal to be incorrect because (i) the mechanism of 12G10 action is likely to be closely overlapping with that of other activating anti-␤A mAbs (whose epitopes are not close to the interface), and (ii) 12G10 has the same properties in a recombinant integrin that lacks a large part of the hybrid domain, suggesting that it directly affects the conformation of ␤A (24). The activating effect of the 15/7 and HUTS mAbs (9, 31) is likely to be due to their preferential binding to the active form of the integrin in which the hybrid domain is shifted away from the ␤-propeller.
Implications for Inside-out and Outside-in Signaling-A recent NMR study of a complex between the intracellular segments of ␣ IIb and ␤ 3 suggests that integrin activation is prevented by a "handshake" between ␣ and ␤ cytoplasmic tails, and that unclasping of this handshake by proteins such as talin represents the first stage of activation (5). Separation of the cytoplasmic domains may then be linked to unbending of the integrin legs, which in turn is coupled to activation of head domains (16,17). In this scenario, hybrid domain movement is severely restricted by its interface with the leg domains in the inactive, bent form of the integrin. Unbending causes activation because it is coupled to the release of the hybrid domain from these constraints, allowing swinging of the hybrid to take place, which, in turn, would cause pulling on the ␣7 helix of ␤A. Our findings generally support this model of activation. However, since removal of the lower leg domains (in the truncated integrin) did not favor the active state (24), hybrid domain movement (rather than unbending per se) appears to be the essential requirement for activation. Hybrid domain motion provides the conduit for the transduction of signals to and from the head region. The attenuation of the epitopes on the ␤-propeller in the L358A mutant also suggests that the outward swing of the hybrid domain is coupled to a conformational rearrangement of the ␤A/propeller interface that is involved in ligand recognition (14). Similarly, ligand binding reinforces and stabilizes the conformational changes associated with activation (e.g. the movements of the ␣1 and ␣7 helices and hybrid domain). This feature of ligand recognition may explain the well known ability of integrin ligands to cause activation that persists after dissociation of the complex (50,51). Outside-in signaling could result, in part, from stabilization of the active conformation, allowing the separated cytoplasmic tails to stably interact with cytoskeletal and signaling molecules. In addition, integrin clustering is also a major contributor to this signaling (52,53).
In summary, we have shown that a conformational shift in the ␣7 helix region of the ␤A domain is involved in the regulation of integrin activity. This movement is coupled to a swingout of the hybrid domain and thereby provides a pathway for signal transduction. Integrins are important therapeutic targets in many inflammatory and cardiovascular disorders (28), and our findings suggest a novel way in which highly specific regulators of integrin activity could be developed (e.g. by stabilizing the hybrid/propeller interface). A more complete understanding of the signaling mechanisms will require further characterization and crystallization of an integrin in defined conformational states.