Integrin Activation Involves a Conformational Change in the α1 Helix of the β Subunit A-domain

The ligand-binding region of integrin β subunits contains a von Willebrand factor type A-domain: an α/β “Rossmann” fold containing a metal ion-dependent adhesion site (MIDAS) on its top face. Although there is evidence to suggest that the βA-domain undergoes changes in tertiary structure during receptor activation, the identity of the secondary structure elements that change position is unknown. The mAb 12G10 recognizes a unique cation-regulated epitope on the β1 A-domain, induction of which parallels the activation state of the integrin (i.e. competency for ligand recognition). The ability of Mn2+ and Mg2+ to stimulate 12G10 binding is abrogated by mutation of the MIDAS motif, demonstrating that the MIDAS is a Mn2+/Mg2+ binding site and that occupancy of this site induces conformational changes in the A-domain. The cation-regulated region of the 12G10 epitope maps to Arg154/Arg155 in the α1 helix. Our results demonstrate that the α1 helix undergoes conformational alterations during integrin activation and suggest that Mn2+ acts as a potent activator of β1integrins because it can promote a shift in the position of this helix. The mechanism of β subunit A-domain activation appears to be distinct from that of the A-domains found in some integrin α subunits.

Integrins are ␣/␤ heterodimeric transmembrane receptors that have widespread essential functions in development, tissue organization, and the immune system (1). Integrins recognize a variety of extracellular matrix and cell-surface ligands; however, ligand recognition is frequently not constitutive but is instead under strict cellular control by "inside-out" signaling. Acquisition of the active state has also been shown to require divalent cations. For ␤ 1 integrins, ligand binding is promoted by Mg 2ϩ or Mn 2ϩ but only weakly by Ca 2ϩ (2). A well known but unexplained property of Mn 2ϩ is its ability to mimic the process of inside-out signaling to strongly up-regulate integrin function (3,4).
The molecular basis of integrin-ligand interactions has been greatly elucidated by the recent x-ray crystal structure of ␣ V ␤ 3 (5). The ligand binding "head" of the integrin is seen to contain a seven-bladed ␤-propeller fold in the ␣ subunit and a von Willebrand factor type A-domain in the ␤ subunit (␤A-domain). 1 Cation-binding sites are present on the lower face of the ␤-propeller domain and the upper face of the ␤A-domain (5). The key regions involved in ligand recognition are loops on the upper surface of the ␤-propeller and the upper face of the ␤A-domain, which contains a metal ion-dependent adhesion site (MIDAS) (5)(6)(7). Nevertheless, as the crystal structure only provides a "snapshot" of one integrin conformation, attention is now focused on understanding the conformational changes that occur during the transition from the inactive to active state (8). These changes are thought to include shape shifting in the ␤A-domain (4,7).
The A-domain contains a central hydrophobic ␤ sheet encircled by seven ␣ helices (␣1-␣7) (5). Some ␣ subunits also contain an A-domain and a key feature of the activation of these domains has been shown to be a large movement of the ␣7 helix (9). Here we investigate conformational changes in the ␤A-domain using the anti-␤ 1 mAb 12G10, which recognizes a cation-and ligand-induced epitope (10,11). We show that movements in the ␣1 helix of the ␤A-domain parallel changes in the activation state of the integrin. Our results provide insights into the mechanisms of Mn 2ϩ and Ca 2ϩ -induced shape changes in the ␤ 1 subunit, and therefore into the opposing roles played by these divalent ions in regulating integrin function. Our findings also imply that the mechanism of ␤A-domain activation is different to that of ␣A-domains.
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 were generated as previously described (13). To aid heterodimerization, the C H 3 domain of the ␣ 5 construct contained a "hole" mutation, whereas the C H 3 domain of the ␤ 1 construct carried a "knob" mutation as described (13,14). Mutations in the A-domain of the ␤ 1 subunit were carried out using oligonucleotide-directed PCR mutagenesis, as described (13). Oligonucleotides were purchased from MWG Biotech (Milton Keynes, UK). The presence of the mutations was verified by DNA sequencing.
Effect of Divalent Cations on 12G10 Binding-Purified integrin was diluted to approximately 1 g/ml in Dulbecco's PBS and added to the wells of a half-area enzyme immunoassay/radio immunoassay plate (Costar, Corning Science Products, High Wycombe, UK; 25 l/well) for 16 h at room temperature. Wells were blocked for 1-3 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). 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, Hemel Hempstead, UK) to remove any small contaminating amounts of endogenous Ca 2ϩ and Mg 2ϩ ions. 12G10 (0.1 g/ml) in buffer A with varying concentrations of Mn 2ϩ , Mg 2ϩ , or Ca 2ϩ was 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 washed three times with buffer A. Bound antibody was quantitated by addition of 1:500 dilution of ExtrAvidin ® peroxidase conjugate (Sigma, Poole, UK) in buffer A for 20 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-ethylbenzthiazoline-6-sulfonic acid) substrate (50 l/well). Background binding to BSA 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 12G10 binding to wild-type tr␣ 5 ␤ 1 -Fc and the R154R/AS mutant, the assay was performed using 12G10 concentrations that gave approximately half-maximal antibody binding in 1 mM Mn 2ϩ (0.1 g/ml for wild-type tr␣ 5 ␤ 1 -Fc, 10 g/ml for R154R/AS mutant). Binding was measured in 2 mM EDTA, Mn 2ϩ , Mg 2ϩ , or Ca 2ϩ . Measurements obtained were the mean Ϯ S.D. of four replicate wells.
Effect of Divalent Cations on III 6 -10 Binding-Measurement of the binding of III 6 -10 to purified wild-type or mutant tr␣ 5 ␤ 1 -Fc was performed exactly as described for biotinylated 12G10 (see above), except that biotinylated III 6 -10 was incubated with integrin for 3 h at 30°C. All assays were performed using a concentration of biotinylated III 6 -10 that gave approximately half-maximal ligand binding in 1 mM Mn 2ϩ (0.1 g/ml).
Sandwich ELISA for Epitope Expression-A 96-well plate (Costar half-area enzyme immunoassay/radio immunoassay) was coated with goat anti-human ␥ 1 Fc (Jackson Immunochemicals, Stratech Scientific, Luton, UK) at a concentration of 2.6 g/ml in Dulbecco's PBS (50 l/well) for 16 h. The coating solution was replaced with blocking buffer for 1 h. The blocking solution was removed, and cell culture supernatants were added (25 l/well) for 1 h. All supernatants were assayed in triplicate, and supernatant from mock-transfected cells was used as a negative control. The plate was washed three times in buffer A containing 1 mM MnCl 2 (buffer B; 200 l/well), and anti-␣ 5 or anti-␤ 1 mAbs (10 g/ml, or 1 g/ml for SNAKA52) were added (50 l/well). The plate was incubated for 2 h and then washed three times in buffer B. Peroxidaseconjugated anti-rat or anti-mouse secondary antibodies (1:1000 dilution in buffer B; Jackson Immunochemicals) were added (50 l/well) for 30 min, the plate washed four times in buffer B, and color was developed using 2,2Ј-azinobis(3-ethylbenzthiazoline-6-sulfonic acid) substrate (50 l/well). All steps were performed at room temperature.
Each experiment shown is representative of at least three separate experiments.

FIG. 1. Effect of divalent cations on the binding of mAb 12G10
(A) and III 6 -10 fragment of fibronectin (B) to tr␣ 5 ␤ 1 -Fc. Binding of 12G10 or III 6 -10 was measured in the presence of varying concentrations of Mn 2ϩ (q), Mg 2ϩ (f), or Ca 2ϩ (OE). For tests of specificity, the binding of biotinylated 12G10 to tr␣ 5 ␤ 1 -Fc could be inhibited Ͼ95% by a hundredfold excess of unlabelled 12G10; III 6 -10 binding to tr␣ 5 ␤ 1 -Fc could be inhibited Ͼ90% by the anti-␣ 5 mAb 16 (data not shown).

RESULTS
Induction of the 12G10 Epitope on the ␤ 1 A-domain Correlates with Competency for Ligand Binding-To investigate the mechanisms of integrin activation, we employed a recently described system for expression of recombinant soluble ␣ 5 ␤ 1 (13). 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 (hereafter referred to as tr␣ 5 ␤ 1 -Fc). This heterodimer contains the ligand-binding head and thigh domains of the integrin (5) and has been shown to retain the properties of the full-length receptor (13). In contrast to previous mutagenesis-based analyses of integrin function, which have largely employed cell-expressed integrins, this system is ideal (a) because it permits the rapid analysis of the effects of mutations and (b) because the effects of mutations that normally preclude expression at the plasma membrane can be studied. 12G10 is a previously characterized activating mAb directed against the ␤ 1 A-domain, whose binding to ␣ 5 ␤ 1 is modulated by divalent cations and ligand (10,11). The binding of 12G10 to tr␣ 5 ␤ 1 -Fc was promoted by Mn 2ϩ and to a lesser degree by Mg 2ϩ , whereas Ca 2ϩ was inhibitory (Fig. 1A). The effects of these cations on 12G10 binding closely paralleled their effects on ligand binding (Fig. 1B). Importantly, as for the native integrin, ligand binding is strongly activated by Mn 2ϩ and more weakly by Mg 2ϩ , whereas Ca 2ϩ is a very poor activator (16). These results show that the ␤A-domain undergoes conformational changes in response to cation binding (reported by modulation of the 12G10 epitope) that correspond with changes in the activation state of the integrin.
Mutation of the MIDAS Site in the ␤A-domain Leads to Loss of Induction of 12G10 Binding by Mn 2ϩ /Mg 2ϩ -The identity of the cation-binding site(s) involved in activation of ␤ 1 integrins by Mn 2ϩ and Mg 2ϩ is unknown. The MIDAS is a strong candidate for this site, but this has been difficult to test experimentally because mutation of the MIDAS residues completely abrogates ligand recognition (although expression is unaffected; Refs. [17][18][19]. In agreement with these previous studies, tr␣ 5 ␤ 1 -Fc with MIDAS mutations did not bind ligand under any cation conditions, even though such mutants (e.g. D130A) retained all the epitopes of conformation-sensitive ␣ 5 and ␤ 1 mAbs (Table I). Because mAb binding was retained, we tested the effect of the D130A mutation on the ability of divalent cations to regulate 12G10 binding to tr␣ 5 ␤ 1 -Fc (Fig. 2). The binding of 12G10 to the D130A mutant in the absence of divalent cations was similar to the wild-type integrin (comparing Fig. 1A with Fig. 2); however, the ability of Mn 2ϩ and Mg 2ϩ to stimulate 12G10 binding was totally lost in the MIDAS mutant. Interestingly, the inhibition of 12G10 binding by Ca 2ϩ seen for the wild-type integrin was enhanced in the MIDAS mutant. Similar results were obtained with a "double" MIDAS mutation D130A/S132A (data not shown). Conversely, muta-tion of cation-binding sites in the ␣ 5 subunit ␤-propeller did not affect the capacity of Mn 2ϩ or Mg 2ϩ to modulate 12G10 binding (20). Hence, these data demonstrate that the MIDAS is a Mn 2ϩ / Mg 2ϩ -binding site and that occupancy of this site induces conformational movements that are detected by changes in 12G10 binding.
The Cation-responsive Portion of the 12G10 Epitope Maps to Arg 154 /Arg 155 in the ␣1 Helix of the ␤A-domain-The epitopes of all function-altering mAbs that map to the ␤ 1 A-domain include one or more residues in the sequence Asn 207 -Lys 218 (21), which based on homology to ␤ 3 is predicted to form the ␣2 helix (5,22). The epitope of 12G10 also maps to this region and includes Lys 218 as part of its epitope (11); however, among these regulatory mAbs, 12G10 has the unique property of showing strong cation modulation of binding (11). Therefore, part of the 12G10 epitope may be distinct from that of the other ␤A-domain mAbs. While investigating the mechanism of integrin activation using alanine-scanning mutagenesis, 2 we found two mutations that selectively perturbed 12G10 binding ( Table  I). Mutation of Arg 154 or Arg 155 to Ala reduced 12G10 binding by ϳ65 and ϳ35%, respectively, whereas mutation of surround-TABLE I Summary of mAb reactivity with ␤ 1 A-domain 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-␣ 5 and anti-␤ 1 mAbs by sandwich ELISA. The anti-␣ 5 mAbs recognize the ␤-propeller domain (12), and the anti-␤ 1 mAbs are all directed against the ␤A-domain (11,21). ϩϩϩ, reactivity 70 -100% of wild-type integrin; ϩϩ, reactivity 50 -70% of wild-type integrin; ϩ, reactivity 20 -50% of wild-type integrin; ϩ/Ϫ, reactivity Ͻ 20% of wild-type integrin. None of the mutations (except D130A) affected recognition of the III 6 -10 fragment of fibronectin (data not shown). ing residues (Asn 151 , Met 153 , Ile 156 ) had no effect. A double mutation R154R/AS reduced 12G10 binding by Ͼ80% but did not perturb the binding of other function-modulating mAbs against the ␤ 1 A-domain (Table I), and also had no effect on the apparent affinity of ligand binding (data not shown). Therefore, Arg 154 and Arg 155 appear to form part of the 12G10 epitope. 3 By homology to the structure of the ␤ 3 A-domain (5), Arg 154 / Arg 155 lie at the base of the ␣1 helix and these residues would be predicted to be in sufficiently close proximity to Lys 218 in the ␣2 helix for all three residues to contribute to the 12G10 epitope (23).
The above data suggest that Arg 154 and Arg 155 form part of the 12G10 epitope, but these residues do not contribute to other A-domain epitopes. Hence, do Arg 154 /Arg 155 form the cationregulated region of the 12G10 epitope? To test this proposal, we compared the effects of divalent cations on 12G10 binding to the R154R/AS mutant and wild-type tr␣ 5 ␤ 1 -Fc. The results (Fig. 3A) showed that the abilities of Mn 2ϩ , Mg 2ϩ , and Ca 2ϩ to modulate 12G10 binding were strongly attenuated by the R154R/AS mutation. The mutation did not affect the cation regulation of ligand binding (Fig. 3B) or of ␣ 5 epitopes (Ref. 11; data not shown), suggesting that the mutation does not itself affect cation-induced conformational changes but rather that the ability of 12G10 to detect these changes is specifically compromised by the mutation. Therefore, the portion of the 12G10 epitope that is responsive to cation binding lies in the ␣1 helix, indicating that the position of this helix is different in the active and inactive states. DISCUSSION Using the anti-␤ 1 mAb 12G10 as a probe of ␤A-domain conformation, we have shown that: (i) the ␤A-domain undergoes shape changes that correlate with changes in the activation state of the integrin, (ii) occupancy of the MIDAS site in the ␤A-domain by Mn 2ϩ or Mg 2ϩ induces these changes, and (iii) ␤A-domain activation involves movement of the ␣1 helix. Taking these results together, we propose that the ␤A-domain can exist in at least two conformational states: an "active" conformation with the ␣1 helix in a position characterized by high 12G10 binding and an "inactive" conformation with the ␣1 helix in a different position, characterized by low 12G10 binding.
Movement of the ␣1 helix appears to form an essential part of the activation mechanism of the ␤A-domain because ␣1 movement closely parallels the activation state and a lack of ␣1 Binding of 12G10 or III 6 -10 to wild-type tr␣ 5 ␤ 1 -Fc or tr␣ 5 ␤ 1 -Fc with the mutation R154R/AS in ␤ 1 was measured in the presence of 2 mM EDTA (white bars), 2 mM Mn 2ϩ (black bars), 2 mM Mg 2ϩ (gray bars), or 2 mM Ca 2ϩ (hatched bars). In A, 12G10 was used at concentration of 0.1 g/ml for wild-type tr␣ 5 ␤ 1 -Fc or 10 g/ml for the R154R/AS mutant (conditions that gave approximately half-maximal 12G10 binding in the presence of 2 mM Mn 2ϩ ). The control mutation M153A had no effect on the cation modulation of 12G10 binding (data not shown). movement (in the Ca 2ϩ -occupied integrin) corresponds to low activity. Furthermore, the epitopes of function-blocking antichicken ␤ 1 mAbs have been shown to include residues in the ␣1 helix (24), and the epitopes of function-altering anti-human ␤ 1 mAbs include residues in the ␣2 helix, which lies adjacent to ␣1 (5,21). Based on previous analyses of the mode of action of regulatory anti-integrin mAbs (2,25), it appears that they are likely to function allosterically by stabilizing the position of ␣1 in either the active or inactive conformation. Additionally, it has been shown that mutation of residues in the ␣1 helix can activate ligand binding (26).
Our data provide evidence that the MIDAS is primarily a Mn 2ϩ /Mg 2ϩ binding site and suggest an explanation for the opposing effects of Mn 2ϩ and Ca 2ϩ on ␤ 1 integrin function. Mn 2ϩ can induce a large shift in the equilibrium between active and inactive states because of its ability to promote ␣1 helix movement upon binding to the MIDAS site. On the other hand, Ca 2ϩ is unable to cause the same conformational change. It is likely that Ca 2ϩ can occupy the MIDAS site because Ca 2ϩ can support low affinity ligand binding to ␣ 5 ␤ 1 and high affinity binding of activation-independent ligands to ␣ 4 ␤ 1 (34). However, Ca 2ϩ binding to sites other than the MIDAS appears to shift the equilibrium toward the inactive conformation (low 12G10 binding), as shown by the strong inhibition of 12G10 binding by Ca 2ϩ in the D130A mutant.
Some integrins contain an A-domain in their ␣ subunits (e.g. the ␤ 2 family). These domains can exist in inactive ("closed") or active ("open") states dependent upon movement of the Cterminal helix (␣7). The open form can be induced in the presence of a ligand or pseudo-ligand, or by locking the position of ␣7 (9,(27)(28)(29). There appear to be some differences between the activation mechanism of ␣A-domains and the ␤A-domain. First, the nature of the metal ion at the MIDAS does not directly influence the equilibrium between inactive and active states in ␣A-domains (4, 28), whereas, based on data reported here for the ␤A-domain, the nature of the divalent ion can markedly affect this equilibrium. Second, in contrast to the ␤A-domain, there is no evidence for allosteric regulation of activity by mAbs to ␣A-domains whose epitopes include residues in the ␣1 helix (30). Third, although in ␣A-domains the open form can be induced by mutation of residues that form a hydrophobic pocket surrounding ␣7 (31), mutation of the equivalent residues in ␤ 1 does not alter integrin activity. 2 Fourth, the crystal structure of the ␤ 3 A-domain indicates that the ␣7 helix is unlikely to undergo large conformational movements (5). All these findings suggest that the ␤A-domain is regulated differently to the ␣A-domains in that movement of the ␣1 helix (rather than ␣7) is a key feature of ␤A-domain activation. Nevertheless, comparison of the open and closed forms of ␣Adomains shows that there is an inward shift of the ␣1 helix in the open form (9), and a similar movement could take place in the ␤A-domain (Fig. 4).
In integrins that contain an A-domain in the ␣ subunit, the ␤A-domain does not participate directly in ligand binding (4). Nevertheless, Mn 2ϩ and mAbs to the ␤A-domain can strongly modulate the activity these integrins (3,4). The epitopes of activating and inhibitory anti-␤ 2 mAbs have also been shown to contain residues in the ␣1 helix of the ␤ 2 A-domain (4,32,33). Therefore, movement of the ␣1 helix may also regulate the activation state of this class of integrin.
Is ␣1 helix movement involved in the activation of integrins by inside-out signaling? It has been shown that the expression of the 12G10 epitope correlates with the activity of cell-surface ␤ 1 integrins, whereas expression of other ␤ 1 A-domain epitopes is constitutive (35,36). Because 12G10 differs from the other A-domain mAbs in having part of its epitope in the ␣1 helix, these data imply that inside-out signaling alters the position of ␣1. Inside-out signaling may also cause a shift in the conformation of the ␣1 helix in ␤ 2 integrins. For ␣ L ␤ 2 and ␣ M ␤ 2 , activating cytoplasmic domain mutations led to the induction of the mAb 24 epitope, which includes Arg 122 in the ␣1 helix of the ␤ 2 A-domain (4,37). It has often been questioned whether Mn 2ϩ -induced integrin activation accurately mimics physiologic activation. However, a common feature of both types of activation appears to be movement of the ␣1 helix; hence, their molecular mechanisms may be very similar.
Finally, how might ␣1 helix movement be important for activation? The top (MIDAS) face of the ␤A-domain interacts closely with the upper surface of the ␣ subunit ␤-propeller domain (5). In particular, loops on the top face of the A-domain close to the ␣1 helix (notably the ␣2-␣3 loop) contact loops on the ␤-propeller domain that participate in ligand recognition. Hence, ␣1 helix movement is likely to affect the ␣ subunit/␤ subunit interface, potentially leading to changes in exposure of the ligand binding loops. In support of this hypothesis, we have shown that divalent cations affect the binding of inhibitory mAbs on the ␣ subunit (11); the epitopes of these mAbs include residues in the same loops that are important for ligand recognition (12,38,39). In integrins with an A-domain in the ␣ subunit, there is evidence that the MIDAS face of the ␤Adomain is in contact with the lower face of the ␣A-domain (40); hence, conformational changes in the ␤A-domain could affect the position of the ␣7 helix in the ␣A-domain and thereby alter the activation state of this domain.
In summary, we have shown that a conformational shift in the ␣1 helix of the ␤A-domain is involved the regulation of integrin activity. Integrins are important therapeutic targets in many inflammatory and vascular disorders (41), and our findings suggest a novel way in which highly specific regulators of integrin activity could be developed. A more complete understanding of the activation mechanism will require crystallization of an integrin in both active and inactive states.