Functional and structural correlations of individual αIIbβ3 molecules

The activity of circulating platelets is tightly regulated to prevent the spontaneous formation of platelet aggregates.¹  Thus, circulating platelets are inactive until they adhere to exposed subendothelial matrix or are stimulated by soluble agonists such as adenosine diphosphate (ADP) and thrombin. Each activating event is associated with a change in platelet shape, reorganization of the platelet cytoskeleton, secretion of platelet granules, and an increase in the affinity of the integrin αIIbβ3 for soluble ligands such as fibrinogen and von Willebrand factor. The latter is responsible for platelet aggregation when the macromolecular ligands bind to activated αIIbβ3 and bridge adjacent platelets.² 

Although the activation state of αIIbβ3 is normally regulated by agonist-generated “inside-out” signaling,¹  αIIbβ3 can also be activated experimentally by perturbing the conformation of its extracellular domain using Mn²⁺ ions^3,4  or the reducing agent dithiothreitol (DTT).^5,6  Thus, in platelets, Mn²⁺ has been reported to activate αvβ3 and αIIbβ3, thereby promoting their interaction with ligands such as fibrinogen, von Willebrand factor, vitronectin, and osteopontin, mimicking the consequences of conventional inside-out signaling.^3,4  Mn²⁺ also stabilizes platelet-fibrinogen interactions.⁷  Moreover, in experiments using purified integrins, Mn²⁺ affects the binding kinetics, affinity, and specificity toward synthetic and natural ligands.^8-11  Mn²⁺-induced changes in integrin function have been attributed to specific conformational rearrangements in the integrin ectodomain,¹²  a suggestion supported by electron microscope studies showing that Mn²⁺ promotes the opening of integrin molecules into extended structures.^13-15 

Millimolar concentrations of DTT also induce platelet aggregation by directly stimulating ligand binding to αIIbβ3.^5,6  How DTT activates αIIbβ3 is unclear. β3 has been reported to contain an extracellular redox site that is associated with the presence of 2 unpaired cysteines in inactive αIIbβ3 and 6 unpaired cysteines following exposure of αIIbβ3 to DTT.¹⁶  However, the identity of the putatively unpaired cysteines in either inactive or active forms of αIIbβ3 has not been determined, and it has been proposed that αIIbβ3 activation by DTT may involve disulfide bond rearrangement of the originally unpaired cysteines, as well as overall bond reduction.^16,17 

The ability of Mn²⁺ and DTT to enhance integrin function, as well as perturb the conformation of integrin ectodomains, provides an opportunity to test the hypothesis that there is an equilibrium between inactive and active integrin activation states that is a consequence of a reversible structural rearrangement of the entire integrin molecule. To address this hypothesis, we used transmission electron microscopy to probe for structural differences between αIIbβ3 in the presence of Ca²⁺, Mn²⁺, and DTT and laser tweezers-based force spectroscopy to measure the fibrinogen-binding function of the integrin in the presence of each at the single molecule level. We found that both Mn²⁺ and DTT increase the probability of specific interactions between αIIbβ3 and fibrinogen, but they do so without changing the average yield strength of fibrinogen binding. Both agents also induce a change in the shape of the αIIbβ3 headpiece, shift αIIbβ3 from a closed conformation with stalks touching to an open conformation with stalks separated, and stimulate the formation of clusters of the open αIIbβ3 conformer. These results indicate that there is a direct correlation between the activation of αIIbβ3 and the overall conformation of the molecule. Further, they are consistent with the presence of a linked equilibrium between inactive and active αIIbβ3 molecules and αIIbβ3 clusters.

Using laser tweezers to measure integrin function on platelets has been described previously in detail.¹⁸  Briefly, we used a custom-built laser tweezers setup assembled from a Nikon Diaphot 300 inverted microscope (Nikon, Mellville, NY), 100 × 1.3NA Fluor lens and a Spectra-Physics (Mountain View, CA) FCBar Nd:YAG laser to measure the strength of fibrinogen binding to human platelets or purified αIIbβ3 in the presence of either Ca²⁺ or Mn²⁺. For these measurements, human fibrinogen (American Diagnostica, Stamford, CT) was covalently bound to 0.93 μm carboxylate-modified latex beads using N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride as a cross-linking agent in a 2-step procedure described in the TechNote no. 205 issued by Bangs Laboratories (Fishers, IN). Before use, the fibrinogen-coated beads were disaggregated by mild sonication and used at a concentration of approximately 10⁷/mL. For studies using purified human αIIβ3, the purified integrin (Enzyme Research Laboratories, South Bend, IN) at a concentration of 1 mg/mL in 0.01 M N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid (HEPES) buffer, pH 7.4, containing 60 mM octyl-glucoside was bound covalently to polyacrylamide-coated 1.40 μm silica pedestals using glutaraldehyde as previously described.¹⁸  Prior to immobilization, the αIIbβ3 was preincubated with either 1 mM CaCl₂, 1 mM MnCl₂, or 5 mM DTT/1 mM CaCl₂ in the same binding buffer at 37°C for 30 minutes. Interactions between fibrinogen and purified αIIbβ3 were studied in 0.1 M HEPES buffer, pH 7.4, containing 2 mg/mL bovine serum albumin, 0.1% Triton X-100, and either 1 mM CaCl₂, 1 mM MnCl₂, or 5 mM DTT/1 mM CaCl₂. To measure fibrinogen binding to αIIbβ3 on living platelets, an individual platelet was trapped from a suspension of gel-filtered human platelets containing approximately 5 × 10⁶ platelets/mL and approximately 10⁵/mL fibrinogen-coated beads and manually attached to a 5-μm diameter silica pedestal coated with polylysine.¹⁸  All experiments with unstimulated platelets were performed in a 4 mM HEPES gel-filtration buffer, pH 7.4, containing 135 mM NaCl, 2.7 mM KCl, 5.6 mM glucose, 1 mM CaCl₂, 3.3 mM NaH₂PO₄, and 0.35 mg/mL bovine serum albumin.

To measure the rupture force between fibrinogen and either purified αIIbβ3 or αIIbβ3 on platelets, a fibrinogen-coated bead, trapped by the laser light, was brought to a distance of 2 to 3 μm from the αIIbβ3-coated pedestal or immobilized platelet. After oscillation of the bead was initiated at 5 Hz or 50 Hz with 0.8 μm peak-to-peak amplitude, the bead was brought into contact with the platelet or the pedestal by micromanipulation using a keyboard-controlled piezoelectric stage. Data collection was initiated at the first contact between the bead and the platelet or pedestal. Rupture forces following repeated contacts between the platelet or the pedestal and the bead were collected for periods of several seconds to 1 minute and were displayed as normalized force histograms for each experimental condition. Because only a small percentage of contact/detachment cycles result in effective receptor-ligand binding/unbinding, data from 10 to 22 experiments, representing 3 × 10³ to 10⁵ individual measurements, were combined. Individual forces measured during each contact-detachment cycle were collected into 10 pN-wide bins. The number of events in each bin was plotted against the average force for that bin after normalizing for the total number of interaction cycles. The percentage of events in a particular force range (bin) represents the probability of rupture events at that tension. Optical artifacts observed with or without trapped latex beads produce signals that appeared as forces below 10 pN.¹⁸  Accordingly, rupture forces in this range were not considered when these data were analyzed.

Rotary-shadowed samples were prepared using a modification of standard procedures^19,20  by spraying a dilute solution of molecules in a volatile buffer (0.05 M ammonium formate) and glycerol (30%-50%) onto freshly cleaved mica and shadowing with tungsten in a vacuum evaporator (Denton Vacuum, Cherry Hill, NJ).²¹  All specimens were examined in a FEI/Philips 400 electron microscope (Philips Electronic Instruments, Mahwah, NJ), operating at 80 kV and at a magnification of × 60 000. The molecular dimensions for approximately 100 individual images from the various groups of αIIbβ3 molecules were measured after scanning negative prints using Adobe Photoshop 7.0.1. Measurements were made with a Photoshop tool calibrated to have 1.6 nm resolution. Molecular dimensions of the digitized images were corrected for a shell of tungsten by subtracting 1 nm on each side of a measured molecule.

When probed by laser tweezers-based force spectroscopy, the vast majority of the interactions between surface-bound fibrinogen and unstimulated platelets in the presence of Ca²⁺ were characterized by rupture forces that ranged from 10 pN to several tens of pN. Moreover, the probability of detecting these rupture forces decreased exponentially as the rupture force increased (Figure 1A). However, the same platelets incubated with 1 mM MnCl₂ for 5 minutes at 25°C were highly reactive with fibrinogen-coated beads, producing a peak in the histogram of rupture forces that ranged from 60 to 110 pN (Figure 1B). Thus, the cumulative probability of detecting rupture forces greater than 60 pN, which we previously found to be specific for fibrinogen binding to αIIbβ3,¹⁸  increased nearly 20-fold. To confirm that the Mn²⁺-induced peak of rupture forces resulted from fibrinogen binding to αIIbβ3, measurements were repeated in the presence of either of the αIIbβ3 antagonists, tirofiban, or abciximab.¹⁸  As shown in Figure 1C-D, each antagonist abrogated the effect of Mn²⁺ and returned the force histograms to that of unstimulated platelets incubated with Ca²⁺.

To verify that the increase in rupture force we observed between platelets and fibrinogen-coated beads in the presence of Mn²⁺ resulted from an increase in the affinity of individual αIIbβ3 molecules and was independent of possible Mn²⁺-induced changes in the platelet membrane, we measured rupture forces between fibrinogen-coated beads and purified αIIbβ3 that had been preincubated with 1 mM Mn²⁺ or 1 mM Ca²⁺. Consistent with our previously reported laser tweezers measurements using purified αIIbβ3,¹⁸  we found that the cumulative probability of detecting rupture forces greater than 60 pN was 2.1% in the presence of 1 mM Ca²⁺, indicating that some of the purified αIIbβ3 was in an active conformation (Figure 2A). Others have observed that approximately 10% of the αIIbβ3 isolated from platelets is in an active conformation.²²  Nonetheless, as shown in Figure 2B, the Mn²⁺-treated preparations were much more reactive with fibrinogen-coated surfaces, such that the cumulative probability of detecting rupture forces greater than 60 pN increased to 4.5%. The ability of the αIIbβ3 antagonists tirofiban and abciximab to decrease this probability to 0.4% confirmed that rupture forces greater than 60 pN resulted from fibrinogen bound to αIIbβ3 (Figure 2C-D). It is noteworthy that, despite the presence of Mn²⁺, the yield strength of αIIbβ3-fibrinogen binding was not changed substantially, whereas the cumulative probability of specific rupture forces greater than 60 pN increased more than 2-fold. Thus, these results are consistent with the hypothesis that Mn²⁺ binding directly induces an increase in the affinity of individual αIIbβ3 molecules for fibrinogen.

Purified αIIbβ3 molecules from the same preparation used for the laser tweezers experiments were visualized by transmission electron microscopy after rotary shadowing with tungsten. Previous electron microscopy studies indicated that αIIbβ3 molecules are composed of a dense “headpiece” and 2 “stalks,” one from each subunit of the heterodimer.²³  The headpiece, composed of the amino-terminal portions of the extracellular domains of αIIb and β3, contains the ligand-binding site of the integrin, whereas the stalks contain the transmembrane and cytoplasmic domains of each subunit. From the observations of more than 2000 individual images, we were able to distribute αIIbβ3 molecules into 3 groups. Group 1 consists of relatively compact molecules in which the stalks appear to touch (“closed” images) (Figure 3A). Group 2 consists of extended molecules in which the stalks are separated (“open” images) (Figure 3B). Group 3 consists of globular heads in which the stalks are not visualized.

Although each group was present regardless whether αIIbβ3 had been incubated with Ca²⁺ or Mn²⁺, the distribution was significantly different. As shown in Table 1, there was a large increase in the proportion of open (group 2) structures in the presence of Mn²⁺ (60% ± 4% in the presence of 1 mM Mn²⁺ versus 11% ± 4% in presence of Ca²⁺, P < .001), with corresponding decreases in the proportion of group 1 (24% ± 3% in the presence of 1 mM Mn²⁺ versus 51% ± 12% in the presence of Ca²⁺, P < .001) and group 3 structures (17% ± 3% in the presence of 1 mM Mn²⁺ versus 38% ± 10% in presence of Ca²⁺, P < .001).

We also compared the molecular dimensions of individual αIIbβ3 molecules in the presence of Ca²⁺ or Mn²⁺ using digitized images of the electron micrographs. In the presence of Ca²⁺, there were no significant differences in the dimensions of the αIIbβ3 headpiece when the molecules were in either the open or closed conformation (Table 2). Moreover, the distance separating the αIIb and β3 stalks of open molecules did not change when αIIbβ3 was bound to fibrinogen (13.4 ± 4.1 nm versus 12.6 ± 2.7 nm, respectively). The latter measurements suggest that stalk separation of approximately 13 nm is sufficient to identify active αIIbβ3 molecules. By contrast, in the presence of Mn²⁺, there was an increase in both the length (10.9 ± 1.5 nm versus 12.6 ± 1.5 nm, P < 10^-8) and width (7.3 ± 1.5 nm versus 8.4 ± 1.1 nm, P < 10^-4) of the αIIbβ3 headpiece when αIIbβ3 shifted from a closed to an open conformation. In addition, the distance separating the stalks of open αIIbβ3 was significantly greater in the presence of Mn²⁺ than in the presence of Ca²⁺ (17.8 ± 3.6 nm versus 13.7 ± 4.1 nm, P < 10^-8). These measurements indicate that not only does Mn²⁺ shift αIIbβ3 from a closed to an open conformation by inducing the separation of αIIb and β3 stalks, but it also alters the overall size and shape of the headpiece.

Besides altering the size and configuration of individual αIIbβ3 molecules, Mn²⁺ increased their tendency to oligomerize. Whereas oligomers were uncommon in the presence of Ca²⁺, 38% ± 10% of the open forms of αIIbβ3 in the presence of Mn²⁺ consisted of dimers, trimers, and higher-order oligomers (Figure 4A-C). Moreover, as shown in Figure 4D, the distribution of monomers, dimers, trimers, and higher-order oligomers could be fit to an exponential function (Figure 4D), as previously described for equilibria involving molecules that self-assemble into oligomers such as actin,^24,25  hemoglobin S,²⁶  and prions.²⁷  It is also noteworthy that the oligomers resulted from interactions that exclusively involved the distal ends of the stalks and that oligomers composed of 3 or 5 molecules did not form closed rosettes as one would expect if the stalks underwent homomeric interactions.

Previously, we observed that αIIbβ3 bound to fibrinogen tended to have an open conformation with separated stalks.²³  To quantify these observations and to measure the distance separating the stalks when αIIbβ3 was bound to fibrinogen, we incubated fibrinogen with αIIbβ3 in the presence of 1 mM Mn²⁺ at 37°C for 30 minutes before the mixture was sprayed onto mica and rotary-shadowed with tungsten. As illustrated by Figure 5A, fibrinogen displayed its typical trinodular structure with 2 lateral D nodules and a central E nodule, and αIIbβ3 complexes were present in the 3 different conformations described in the previous section. Although the majority of individual fibrinogen and αIIbβ3 molecules were separated from each other, a minor fraction formed bimolecular and trimolecular complexes of 2 possible stoichiometric ratios, either 1:1 or 1:2 of αIIbβ3 to fibrinogen (Figure 5B-D). As we reported previously,²³  αIIbβ3 was always spatially oriented so that its headpiece was attached to the end of a fibrinogen molecule, and the stalks of 2 integrins attached to 1 fibrinogen were oriented in opposite directions. The orientation was such that the entire complex had a 2-fold axis of symmetry through the center of the fibrinogen, ie, a rotation of 180° about this axis brings one αIIbβ3 into the other one. All of the αIIbβ3 molecules interacting with fibrinogen were in the open conformation, implying that this conformation represents its activated state.

Because αIIbβ3 was in excess, quantitative analysis of complex formation was based on the relative proportion of fibrinogen molecules participating in the complexes versus those remaining free. Comparison of the fractions of free fibrinogen molecules in the presence and in the absence of Mn²⁺ clearly showed the promoting effect of manganese ions on fibrinogen binding to αIIbβ3. Thus, in the absence of Mn²⁺, 23% ± 8% of the fibrinogen molecules were bound to αIIbβ3. In the presence of Mn²⁺, the percentage increased significantly to 66% ± 6%. Further, we found that the distance separating the stalks of open αIIbβ3 increased from 17.8 ± 3.6 nm in the presence of Mn²⁺ alone to 19.9 ± 4.7 nm (P = 10^-3) when Mn²⁺-treated αIIbβ3 was bound to fibrinogen (Table 2).

αIIbβ3 on platelets and transfected tissue culture cells is activated by incubating the cells with DTT.^5,6,28  Therefore, to determine whether the changes in the conformation of αIIbβ3 induced by Mn²⁺ are unique to this cation, we repeated the measurements described in the previous section using DTT as the stimulus for αIIbβ3 activation. First, we used laser tweezers to measure rupture forces between fibrinogen-coated beads and purified αIIbβ3 that had been preincubated with 5 mM DTT for 30 minutes in the presence of 1 mM Ca²⁺. Similar to Mn²⁺, DTT increased the cumulative probability of detecting rupture forces greater than 60 pN in the presence of Ca²⁺ from 1.9% ± 0.5% to 3.2% ± 0.6% (Table 3). Moreover, there was no difference between the average yield strength of fibrinogen binding to αIIbβ3 in the presence of Mn²⁺ or DTT, suggesting that the αIIbβ3 activation state was similar under both sets of conditions.

DTT-treated αIIbβ3 molecules were then visualized by transmission electron microscopy after rotary shadowing with tungsten. Like Mn²⁺, DTT treatment resulted in a nearly 3-fold increase in the number of αIIbβ3 molecules in the open, rather than closed, conformation and in the formation of αIIbβ3 clusters. Further, DTT induced changes in the molecular dimensions of individual αIIbβ3 molecules that were similar to the changes induced by Mn²⁺. Thus, the distance separating the αIIb and β3 stalks of open molecules significantly increased from 13.1 ± 2.8 nm in the presence of Ca²⁺ to 18.1 ± 3.9 nm in the presence of DTT and Ca²⁺ (P = 4.8 × 10^-23) (Table 4). Moreover, there were significant increases in both the head length (11.9 ± 1.9 nm versus 13.4 ± 2.0 nm, P < 10^-6) and width (8.7 ± 1.4 nm versus 9.2 ± 1.1 nm, P < .034) when αIIbβ3 was treated with DTT. Thus, these measurements confirm that agents that activate αIIbβ3 by perturbing its extracellular domain induce separation of the αIIb and β3 stalks. They also indicate that these agents can alter the overall size and shape of the αIIbβ3 headpiece as well.

Integrins can be activated in vitro by cleavage with specific proteases,²⁹  stabilization of their activated states using monoclonal antibodies,³⁰  exposure to reducing agents such as DTT,²⁸  and incubation with the divalent cation Mn²⁺.³¹  How each of these treatments alters integrin activation states is not entirely clear. For example, although Ca²⁺ or Mg²⁺ are required for ligand binding to integrins, neither by themselves activate integrins, whereas Mn²⁺, at millimolar concentrations, induces integrin-mediated cell adhesion,^4,9,31,32  binding of isolated integrins to immobilized ligands,^12,33,34  and binding of soluble ligands to integrins on cell surfaces.^35-37 

The crystal structure of the extracellular portion of the integrin αvβ3 revealed that it contains 8 divalent cation-binding sites.¹⁴  Four sites were located in the β-propeller domain of αv, 1 at the α subunit genu (knee), and 3 in the β3 βA domain.³⁸  The number of divalent cations bound to the βA domain appears to be directly related to the presence or absence of ligand. Thus, in the absence of ligand, only the cation binding site in the βA ADMIDAS (adjacent to the metal ion-dependent adhesion site) motif is occupied,¹⁴  whereas in the presence of Mn²⁺ and a cyclic Arg-Gly-Asp (RGD) ligand, each of the remaining 2 βA domain binding sites contain a cation. One of these sites is located in the βA MIDAS (metal ion-dependent adhesion site), and Mn²⁺ at this site contacts one of the ligand Asp carboxylate oxygens. A second Mn²⁺ is located 0.6 nm away from the MIDAS at a site designated as the ligand-induced metal binding site (LIMBS), but does not interact with ligand. Although it has been postulated that Mn²⁺ affects integrin activation states by antagonizing inhibitory effects of Ca²⁺,⁹  analysis of the crystal structure of the αvβ3 extracellular domain suggests that by occupying sites in the MIDAS and LIMBS motifs of the βA domain, Mn²⁺ stabilizes its ligand-occupied conformation.¹⁴ 

It is currently thought that integrins such as αIIbβ3 reside on cell surfaces in a thermodynamic equilibrium between inactive and active conformations that can be perturbed by altering the relative position of an integrin α and β stalks.^13,34  An essential element of this hypothesis is that the equilibrium can also be perturbed by altering the conformation of the integrin extracellular domain. To test this premise, we used laser tweezers-based force spectroscopy and electron microscopy to correlate the functional and ultrastructural consequences of exposing the platelet integrin αIIbβ3 to either Mn²⁺ or DTT. Laser tweezers are an optical system in which external forces applied to a spherical particle trapped by a laser can be accurately measured because the angular deflection of the laser beam is directly proportional to the lateral force applied to the particle and are sensitive and accurate at the lower end of the force spectrum (0-150 pN).^39,40  Previously, we found that specific binding of fibrinogen to αIIbβ3 resulted in rupture forces ranging from 60 to 150 pN and an average yield strength of 80 to 100 pN.¹⁸  Because the specific rupture forces occurred as a single well-defined peak, they likely represent the interaction of individual αIIbβ3 and fibrinogen molecules. Using living platelets and isolated αIIbβ3 molecules, we found that Mn²⁺ and DTT increased the affinity of αIIbβ3 for fibrinogen and that the rupture forces between fibrinogen and Mn²⁺- or DTT-stimulated αIIbβ3 were essentially the same as those we measured previously using ADP- and thrombin-related activation peptide (TRAP)-stimulated platelets.¹⁸  Thus, like physiologic platelet agonists, Mn²⁺ and DTT shift αIIbβ3 from an inactive to an active conformation and do so in the absence of αIIbβ3 clustering. However, unlike Mn²⁺- and DTT-activated α4β1 that was found to have an affinity for vascular cell adhesion molecule-1 (VCAM-1) or a ligand peptide intermediate between its inactive and fully active state,^41-43  we detected only 2 αIIbβ3 activation states. Thus, we found essentially no difference in the spectrum of rupture forces between fibrinogen and αIIbβ3 regardless whether we measured it in the presence of Ca²⁺, Mn²⁺, or DTT or whether platelets were stimulated with ADP or TRAP. One might conclude erroneously that there are intermediate activation states when large ensembles of αIIbβ3 molecules are studied because time-averaged mixtures of low- and high-affinity αIIbβ3 molecules are being measured. Moreover, because the activation of integrin by Mn²⁺ is presumably reversible,¹²  it is reasonable to assume that there is an equilibrium between the closed and open forms that is shifted toward the open form by Mn²⁺. Thus, it is not necessary to hypothesize an intermediate conformer since such a conformer can be attributed to the mixture of 2 forms of the integrin.

The possibility that there are only 2 αIIbβ3 activation states is supported by the electron microscope images of single αIIbβ3 molecules. Although we detected 3 basic αIIbβ3 structures: open, closed, and globular, it is likely that the latter 2 are related because a comparable fraction of each was converted to the open form by Mn²⁺ or DTT. Similar images were obtained in earlier electron microscope studies of αIIbβ3²³  and α5β1,^13,36  regardless of the molecular staining technique. These structures were present in the absence of Mn²⁺ or DTT, but each agent converted most of the αIIbβ3 molecules to the open form. In electron microscope images of negatively stained αvβ3, Takagi et al¹³  observed inactive molecules that had a bent conformation, similar to the bent conformation of αvβ3 in crystals,³⁸  and that αvβ3 was both extended and active in the presence of Mn²⁺. We did not observe bent forms of αIIbβ3, even though our αIIbβ3 preparations clearly contained inactive and active molecules. However, Takagi et al¹³  studied recombinant αvβ3 molecules containing a carboxyl-terminal clasp, whereas we studied αIIbβ3 molecules isolated from platelets. Accordingly, our studies and those of Takagi et al¹³  may not be comparable.

Takagi et al¹³  also observed that the Stokes (hydrodynamic) radius of Mn²⁺-treated αvβ3 in the absence of ligand was intermediate between that of αvβ3 in the presence of Ca²⁺ and that of αvβ3 complexed with an RGD-containing ligand. However, Mould et al⁴⁴  observed no gross differences in the conformation of the α5β1 headpiece in the presence of Ca²⁺ and Mn²⁺ by solution x-ray scattering, but their data were also consistent with an opening of the headpiece that involved an outward movement of the β1 hybrid domain and downward swing of the α7 helix in the presence of Mn²⁺. We did not detect a difference in the dimensions of the closed and open conformations of αIIbβ3 or in the distance separating the αIIb and β3 stalks of open and fibrinogen-bound αIIbβ3 in the presence of Ca²⁺. Thus, these dimensions are at least sufficient to identify an active conformation of αIIbβ3. Moreover, like Takagi et al,¹³  we found an increase in the size of the αIIbβ3 headpiece and a further increase in the distance separating the αIIb and β3 stalks in the presence of Mn²⁺ and DTT. Nonetheless, x-ray crystallography revealed no changes in the structure of the extracellular portion of αvβ3 when the crystals were soaked with buffer containing MnCl₂.¹⁴  There are at least 2 possibilities to reconcile the difference between these results. First, contact forces in preexisting crystals may prevent the structural change normally induced by Mn²⁺. Second, transmembrane and cytoplasmic segments absent in the crystal of the ectodomain may be critical to integrin activation because they were shown to have a major impact on the ligand-binding activity and the shape of the integrin.⁴⁵ 

We found a clear correlation between the ability of αIIbβ3 to bind fibrinogen and the presence of open αIIbβ3 molecules. Thus, all αIIbβ3 molecules bound to fibrinogen, whether in the presence of Ca²⁺ or Mn²⁺, had separated stalks. Moreover, we found that fibrinogen binding to αIIbβ3 was enhanced by Mn²⁺ and DTT in parallel with the increased fraction of open αIIbβ3 conformers. Because there is an equilibrium involving the inactive and active conformations of αIIbβ3, these observations suggest that the differential effects of Ca²⁺ and Mn²⁺ on the αIIbβ3 activation state are a function of the ability of each cation to stabilize one conformational state or the other. In the presence of Ca²⁺, the probability of encountering an active open αlIbβ3 conformation was approximately 10%, but, in the presence of Mn²⁺, the probability increased to approximately 60%. It is also noteworthy that, as would be expected in a chemical equilibrium, active and inactive molecules coexisted in the presence of either cation. Thus, Mn²⁺ appears to induce αIIbβ3 activity by stabilizing the active conformation of the αIIbβ3 headpiece, thereby shifting the chemical equilibrium in the active direction and transmitting the conformational change to the stalks. The changes in the dimensions of the αIIbβ3 headpiece are likely a consequence of the differences in size and electronegativity of Ca²⁺ and Mn²⁺ ions and of the additional Mn²⁺ bound to the headpiece in the presence of ligand.

Although αIIbβ3 activation by DTT likely involves overall disulfide bond reduction, as well disulfide bond rearrangement,¹⁶  the identity of the cysteines involved is not clear. Because αIIbβ3 activation appears to involve changes in the conformation of βA and hybrid domains,^44,46,47  it would be logical to assume that the relevant cysteines are located in these domains, an assumption consistent with the changes in the dimensions of the αIIbβ3 headpiece that we detect in the presence of DTT. Nonetheless, the free cysteines identified when αIIbβ3 is exposed to mild reducing conditions are located in the epidermal growth factor-like repeats that constitute the β3 stalk,^16,38  as is an activating Cys583→Tyr mutation.¹⁷  Thus, it is likely that the perturbed disulfide bonds that are responsible for αIIbβ3 activation by DTT remain to be identified.

Clusters of αIIbβ3 molecules have been observed on the surface of thrombin-stimulated platelets,⁴⁸  and αIIbβ3 clustering has been induced in vitro as well.^49-51  Using Mn²⁺-activated αIIbβ3 and electron microscopy, we observed that in the absence of membrane or cytoskeletal constraints, isolated αIIbβ3 formed dimers, trimers, and higher-order oligomers that involved the distal ends of αIIb and β3 stalks. Thus, these images indicate that directly perturbing the activation state of the αIIbβ3 extracellular domain also results in the formation of αIIbβ3 oligomers in the absence of ligand, as would be predicted from the equilibrium model of integrin regulation. It is also noteworthy that the trimers and pentamers were always open, as would be predicted if the ends of the stalks only undergo homo-oligomerization, ie, the formation of α-α or β-β subunit oligomers. Although it is possible that this observation could be the result of steric interference, it is consistent with the hypothesis that homomeric associations involving transmembrane domains are associated with integrin activation and clustering.⁴⁵  It also suggests that a component of Mn²⁺-induced modulation of integrin function in cell membranes may result in increased integrin avidity arising from Mn²⁺-induced aggregation or clustering of integrin molecules.

In conclusion, we used laser tweezers and electron microscopy to demonstrate a direct correlation between conformational changes in individual αIIbβ3 molecules and their ligand-binding activity. Using Mn²⁺ and DTT as activating tools, we found that separation of the αIIb and β3 stalks is an integral part of the mechanism leading to the exposure of the fibrinogen binding site in the αIIbβ3 ectodomain. Moreover, because the open conformation induced by Mn²⁺ and DTT occurred in the absence of ligand binding, it likely represents a primary activating event, perhaps mimicking the consequences of agonist-induced stimulation of αIIbβ3 in platelet membranes.