Mn²⁺ accelerates ligand-binding site activation of α_IIbβ₃ integrin: Insight from all-atom simulation

Abstract

The activation of integrins by Mn²⁺ is a crucial area of research, yet the underlying mechanisms remain poorly understood. Previous studies have shown that substituting Mg²⁺ with Mn²⁺ at the metal ion-dependent adhesion site (MIDAS) enhances the affinities of high-affinity open and low-affinity closed integrins. However, the molecular effect of Mn²⁺ and how it compares to physiological activation mediated by Mg²⁺/Ca²⁺ remain unclear. This is partly due to the lack of experimental techniques capable of detecting these processes dynamically. In this study, we used equilibrium molecular dynamics simulations to examine the effects of Mn²⁺ on the binding site of platelet integrin α_IIbβ₃. Our findings show that Mn²⁺ accelerates conformational changes related to activation. Specifically, Mn²⁺ promotes an earlier displacement of M335 in the β6-α7 loop away from the ADMIDAS site (adjacent to the MIDAS site) and a rapid downward movement of the α7 helix in the βI domain. Additionally, Mn²⁺ leads to faster stabilization of the α1 helix, strengthening the interactions between the α_IIbβ₃ ligand-binding site and the RGD motif. These results suggest that Mn²⁺ accelerates high-affinity rearrangements at the ligand-binding site, resembling those seen in physiological activation, but occurring more rapidly than with Mg²⁺/Ca²⁺. Overall, our data suggest that Mn²⁺-induced affinity modulation proceeds through similar early activation steps, even without full integrin extension.

Significance

Integrin α_IIbβ₃ is crucial for platelet aggregation and blood clotting, yet the molecular mechanisms behind its activation remain largely unknown. Although Mn²⁺ enhances integrin’s ability to bind ligands, the exact process at the molecular level is still unclear. In this study, we used molecular simulations to compare how Mn²⁺ influences the molecular rearrangements of α_IIbβ₃, relative to Mg²⁺/Ca²⁺. Our results show that Mn²⁺ accelerates key changes at the ligand-binding site, enabling the integrin to reach a high-affinity state faster than Mg²⁺/Ca²⁺.

Introduction

Integrin α_IIbβ₃ is a crucial transmembrane receptor involved in physiological processes such as platelet aggregation and blood hemostasis, as well as in pathological conditions such as arterial thrombosis. Due to its central role, α_IIbβ₃ has become a major target for antithrombotic therapies, including small molecules, antibodies, and synthetic peptide mimics, many of which are currently in clinical trials (1,2). Understanding the molecular dynamics (MD) governing α_IIbβ₃ activation is essential for advancing these therapies (3).

Conformational activation of integrin involves the transition from a low-affinity, bent (closed) conformation to a high-affinity, extended (open) form (Fig. 1, A and B). In resting platelets, α_IIbβ₃ adopts a low-affinity conformation, with its extracellular headpiece bent over the legs (Fig. 1 A) (4,7). Activation by ligands, divalent cations, activating mutants, or antibodies can trigger a long-range conformational shift—extending the headpiece, opening the legs, and reorienting the ligand-binding site away from the legs (Fig. 1 B). This transition results in a nearly 200-fold increase in ligand-binding affinity (8,9). Studies using electron microscopy (EM), small-angle x-ray scattering, and PAC1 antibody binding have shown that Mn²⁺ induces this extended, high-affinity conformation (10,11,12,13,14). However, other reports based on cryoelectron tomography (cryo-EM), small-angle neutron scattering, and x-ray crystallography suggest that Mn²⁺ does not always lead to full integrin extension (15,16,17,18,19,20), and that ligand binding can occur even when the integrin remains in the bent conformation (21). Therefore, integrin extension and activation may be at least partially decoupled, and ligand-affinity modulation might not require complete integrin extension. However, the molecular mechanisms by which Mn²⁺ exactly modulates the structural rearrangements of integrin—and how these compare with physiological activation by Mg²⁺ and Ca²⁺—remain poorly understood.

Figure 1.

All-atom structure of integrin α_IIbβ₃. (A) Cartoon representation of full-length integrin α_IIbβ₃ in its bent, closed conformation and (B) extended, open conformation. The cryo-EM structures were resolved in (4), and previously analyzed in molecular simulation studies (5,6). The red box indicates the α_IIb β-propeller and β₃ βI domains containing the RGD binding site. Blue and orange indicate the α_IIb and β₃ chains, respectively. (C) Zoomed-in view of the ligand-binding site, showing state 1 (green) and state 8 (cyan). The β1-α1 loop (gray and black), α1 helix (yellow and orange), β6-α7 loop (magenta and green), and α7 helix (red and salmon) are shown in respective colors for the two states. The center of mass (COM) of the α7 helix is indicated by small red and salmon spheres for states 1 and 8, respectively. The cyan and deep olive structures represent the cap of the α_IIb β-propeller domain. The large purple spheres, located on the leftmost and rightmost sides of the MIDAS ion (yellow sphere), represent the SyMBS and ADMIDAS ions, respectively. (D) Change in ADMIDAS ion coordination from the closed to the open state. In the closed state, D251 does not directly coordinate the ADMIDAS ion, but in the open state, it does. M335 coordinates the ADMIDAS ion in the closed state but not in the open state. (E) Changes in RGD motif interactions between the closed and open states. D224 does not directly contact the R residue in the RGD motif in the closed state but does in the open state. S123 does not directly interact with the D residue of the RGD motif in the closed state but does in the open state. N215 and Y122 consistently contact the D residue of the RGD peptide across all known states. The RGD peptide and individual residues are shown in stick representation, with nitrogen atoms in blue, oxygen atoms in red, and sulfur atoms in yellow. Purple and pink spheres represent Ca²⁺ ions, and the yellow sphere represents a Mg²⁺ ion.

The ligand-binding site of integrin α_IIbβ₃ involves the β-propeller domain from the α_IIb subunit and the βI domain from the β₃ subunit (Fig. 1 B). The ligand’s RGD motif—composed of arginine (R), glycine (G), and aspartic acid (D)—interacts with a “cap” formed by four loops on the β-propeller domain (Fig. 1 C) and a “specificity-determining” loop in the βI domain, known as the β1-α1 loop (Fig. 1 C) (22). Upon transition to the high-affinity state, the ligand-binding site undergoes molecular rearrangements, especially in the βI domain, which increases its binding affinity (16,23,24,25,26). Divalent cations, including Mg²⁺, Ca²⁺, and Mn²⁺, play a role in these changes. Key rearrangements in the βI domain include an increase in the helicity of the α1 helix (Fig. 1 C), a shift of the metal ion in the adjacent to metal ion dependent adhesion site (ADMIDAS) (comprising residues D126, D127, and M335, see Fig. 1 C) toward the β-propeller domain (Fig. 1 D), a downward motion of the α7 helix (∼10 Å, Fig. 1 C), and movement of residue M335 in the β6-α7 loop away from the RGD motif (Fig. 1 D).

Under physiological conditions (∼1 mM Mg²⁺/∼1 mM Ca²⁺), Mg²⁺ occupies the metal ion-dependent adhesion site (MIDAS) (comprising residues E220, S121, S123, D119, and D251), whereas Ca²⁺ binds to the ADMIDAS and synergistic metal-binding site (SyMBS) (comprising residues D217, N215, D158, and P219) (19,22). Mg²⁺ at MIDAS directly contacts the ligand, while Ca²⁺ at SyMBS influences ligand binding indirectly (27). The ion at the ADMIDAS site coordinates with residues M335, D126, and D127, and as activation progresses, it shifts its coordination from M335 to D251 (Fig. 1 D) (19,22,28). Additionally, the ion at the MIDAS coordinate with the ligand’s D495 residue. This residue also forms hydrogen-bonds with residues Y122, N215, and in later stages of activation, with S123 in the β1-α1 loop (Fig. 1 E) (26). Meanwhile, the ligand’s residue R493 forms a water-mediated salt bridge with the β-propeller residue D224, and this interaction is replaced by a direct interaction as integrin activation progresses (Fig. 1 E) (26). However, the precise mechanisms by which divalent cations—such as Mn²⁺ compared with Ca²⁺/Mg²⁺—modulate the rearrangements of the ligand-binding site remain poorly understood.

In this study, we employed equilibrium MD simulations to investigate how Mn²⁺ influences the structural dynamics of the ligand-binding site of α_IIbβ₃ comparing its effects with those of physiological Mg²⁺/Ca²⁺ conditions. Our results reveal that Mn²⁺ accelerates the motions of the β6-α7 loop and the downward motion of the α7 helix in the βI domain, initiating these molecular rearrangements earlier than physiological Mg²⁺/Ca²⁺. These molecular rearrangements strengthen ligand interactions, thus establishing a clear link between integrin structural dynamics and ligand binding through divalent cations.

Materials and methods

Energy minimization and equilibration of α_IIbβ₃ headpiece conformations

Eight structures of the α_IIbβ₃ ligand-binding site were used in this study: PDB: 3ZDY, 3ZDZ, 3ZE0, 3ZE1, and 3ZE2 (26). The specific states derived from these structures are as follows: state 1 (3ZDY, molecule 2), state 2 (3ZDY, molecule 1), state 3 (3ZDZ, molecule 2), state 4 (3ZE0, molecule 2), state 5 (3ZE1, molecule 2), state 6 (3ZE2, molecule 1), state 7 (3ZDZ, molecule 1), and state 8 (3ZE2, molecule 2). State 1 and state 8 represent the completely closed and open ligand-binding sites, respectively. These eight states contain the β-propeller (residues 1–452), βI domain (residues 110–352), and ligand (RGD). For the Mg²⁺-containing systems, four Ca²⁺ ions in the β-propeller domain, Mg²⁺ at the MIDAS site and Ca²⁺ at the ADMIDAS and SyMBS sites were added. For Mn²⁺-containing systems, Mn²⁺ were added at the three metal-binding sites. All initial structures already included ions at the MIDAS, ADMIDAS, and SyMBS sites. When preparing Mn²⁺ systems from structures originally containing Mg²⁺ or Ca²⁺ (e.g., states 1 and 2 from 3ZDY), the ion residue names were modified to “MN2P” to represent Mn²⁺, while retaining their original coordinates. Conversely, for Mg²⁺/Ca²⁺ systems derived from Mn²⁺-bound structures, Mn²⁺ ions were replaced with Mg²⁺ or Ca²⁺ and renamed as “MG” or “CAL,” respectively. All simulations were carried out in GROMACS (29), using CHARMM36m force field (30). Each state was placed in a box with a 1.5 nm buffer from the structure edges (Table S1). CHARMM-modified TIP3P water model and 150 mM NaCl were used to solvate each structure. The total number of atoms of each state is reported in Table S1. The systems were first subjected to energy minimization using the steepest descent algorithm for 5000 steps or until the max force was <1000 kJ/mol/nm, whichever came first. The LINCS technique was used to confine the length of the covalent bonds involving hydrogen atoms. Then, equilibration simulations were performed using restraining potentials of protein’s heavy atoms and the divalent cations of 1000 kJ/mol/nm². First, equilibration simulations were carried out in the constant temperature (310 K), constant volume (NVT) ensemble with a time step of 2 fs for 100 ps. Then, equilibration simulations were continued in the constant temperature (310 K) and constant pressure (NPT) ensemble with a time step of 2 fs for 100 ps. Once the NVT and NPT equilibration runs were finished, restraining potentials were released. The Lennard-Jones interactions and short-range electrostatics were cut off at 1.0 nm. With a grid spacing of 0.16 nm and interpolation order of 4, the particle-mesh Ewald approach was used for the long-range electrostatic interactions. Temperature was controlled with the V-rescale thermostat under periodic boundary conditions.

Metal ion parameters

Default parameters from the CHARMM36m force field were used for all divalent ions. These parameters are based on the solvent boundary potential approach (31), which optimizes them to match hydration free energies in TIP3P water. However, these parameters do not explicitly account for coordination with protein residues and were not specifically optimized for particle mesh Ewald electrostatics. A single set of Lennard-Jones parameters for divalent cations cannot simultaneously reproduce both hydration free energies and ion-oxygen distances (32), meaning some accuracy trade-offs are inherent. Although we did not analyze metal coordination numbers or geometries during the simulation, crystal structures of α_IIbβ₃ integrin suggest typical octahedral geometry with five to six coordinating ligands. Given that the metal ions in our system are not expected to form persistent covalent bonds, the use of nonbonded models is appropriate for modeling general electrostatic behavior, despite their limitations in capturing coordination geometry.

Production MD simulations of α_IIbβ₃

Following equilibration, structures corresponding to each integrin structure (states 1 through 8) and ion conditions (Mg²⁺/Ca²⁺ or Mn²⁺) were duplicated to create between 10 and 24 replicas. For independent sampling, all replicas began production runs with newly generated velocities at 310 K and equilibrium MD simulations under the NPT ensemble were run for 300 ns. Temperature, pressure, and bond lengths were controlled by V-rescale thermostat, Parrinello-Rahman barostat, and LINCS algorithm, respectively.

Visualization of the simulation trajectory was performed using VMD (33) and PyMOL. GROMACS analysis tools and homemade scripts in Python, MATLAB 2024a, and VMD were used to analyze the MD trajectories quantitatively (34,35).

Replica selection

To exclude simulations with persistent ADMIDAS ion dissociation, we applied a distance-based criterion to all trajectories. Specifically, any replica in which the γ-carbon of residue D127 in the βI domain maintained an average distance greater than 5 Å from the ADMIDAS ion over the final 1.25 ns of the 300 ns simulation was excluded from further analysis. The 5 Å threshold was selected based on typical coordination distances for divalent cations (e.g., Mg²⁺, Ca²⁺), which typically fall in the 2–3 Å range. A sustained distance beyond 5 Å suggests loss of stable ion coordination. The final 1.25 ns was chosen to evaluate the state of the ion-binding site at the end of the simulation, capturing persistent dissociation rather than short-term fluctuations. If the ion re-established coordination before the end of the simulation, the replica was retained.

Fig. S4 shows the evolution of the D127-ADMIDAS ion distance for each simulation. Based on this 5 Å criterion, the number of valid replicas included for the Mg²⁺/Ca²⁺ condition was as follows: state 1, 10/14; state 2, 10/10; state 3, 7/10; state 4, 6/10; state 5, 9/10; state 6, 10/14; state 7, 9/10; state 8, 14/25. In each fraction, the first number indicates the number of replicas retained after applying the criterion, and the second number represents the total number of replicas simulated for that state. Notably, state 8 exhibited the highest rate of ADMIDAS ion loss, suggesting that the fully open conformation has a less stable ADMIDAS site.

In contrast, no Mn²⁺ replicas were excluded using this threshold, consistent with the smaller ionic radius and higher charge density of Mn²⁺ compared with Ca²⁺. These properties result in shorter bond lengths and stronger electrostatic interactions at the ADMIDAS site. This enhanced stability aligns with prior findings that Mn²⁺ promotes a more open integrin conformation and increases both bond lifetime and ligand-binding affinity (36). To ensure consistency across all integrin states and ion conditions, we selected the minimum number of valid replicas across all systems—six—as the basis for further analysis. Thus, each state (1 through 8) under both Mg²⁺/Ca²⁺ and Mn²⁺ conditions was analyzed using six representative replicas.

Quantitative analysis of conformational shifts in integrin α_IIbβ₃ across states

Three key conformational shifts previously identified in the α_IIbβ₃ ligand-binding site were analyzed to quantify structural changes across eight states (25). The first measurement examined the downward shift of the α7 helix compared with the open and closed state. All structures in each trajectory were aligned to the corresponding post minimization structures. To assess this shift, the reference closed (PDB: 3T3P) and open (PDB: 2VDR) structures were aligned to the initial frames. These two reference structures were chosen instead of state 1 and state 8 because they have been widely used in prior studies (37)—including the original work that defined the intermediate activation states (26)—to represent canonical closed and open integrin conformations. Their use allows for consistency with previously published analyses and facilitates comparisons across studies. The center of mass (COM) distance between the α7 helix (residues 338–352) at each time point of the trajectory and the crystal structures was then calculated. The difference in these distances was used to calculate how far away from the closed state and how much closer to the open state the structure moved. The second measurement focused on the coordination between the backbone oxygen atom of M335 in the β6-α7 loop of the βI domain and the ADMIDAS ion. An increase in the distance between M335 and the ADMIDAS ion correlates with more active integrin states (26). To analyze this conformational shift, we measured the distance between M335’s oxygen and the ADMIDAS ion. The third measurement assessed the helical content of the α1 helix, comprising residues 127–146 in the βI domain, as this helical structure increases in the open state. Details about analysis of the helical content of the α1 helix are provided in supporting material.

Hydrogen-bond analysis

Previous studies have reported that the backbone nitrogen atoms of residues 122 and 215 in the βI domain form hydrogen-bonds with the side chain of the ligand’s aspartate residue (D495) across all eight states of headpiece activation (26). Additionally, Ser-123 establishes a direct hydrogen-bond with the same residue in state 3, which remains through the fully open state (26). Here, to evaluate the effects of ions (Mg²⁺/Ca²⁺ versus Mn²⁺) on hydrogen-bonds, we evaluated their occupancy throughout the 300 ns of equilibrium MD simulations for the different conformational states. We first extracted from the simulation trajectories specific atoms: backbone nitrogen and associated hydrogen atoms for residues 122, 123, and 215 in the βI domain; terminal side chain oxygen atoms for residue 224 in the β-propeller domain; terminal side chain oxygen atoms for the ligand Asp residue (D495); and terminal nitrogen atoms with two associated hydrogen atoms for the ligand Arg residue (R493). Hydrogen-bond detection was performed with bond parameters set to a threshold distance of 3.475 Å, an angle of 30°, and pair-type were selected using the VMD HBonds Plugin.

Bootstrapping procedure

To quantify the variability and assess the robustness of the measured structural and interaction features—namely, the distance of the α7 helix relative to its positions in bent and open integrins, the distance between the ADMIDAS ion and residue M335, the helicity of helix α1, and the occupancy of hydrogen-bonds—we first computed the average values of each feature across time for each simulation replica. From the pool of selected replicas, a consistent subset of six was randomly selected per integrin state and ion condition. For each of these subsets, bootstrap resampling was performed with 1000 iterations. In each iteration, replicas were resampled with replacement, and the mean of the selected data was computed across time for each replica. These per-replica means were then used to calculate an overall mean for that bootstrap sample. After completing all iterations, the resulting bootstrap means were aggregated into a distribution representing the variability of the structural feature under that condition. From these bootstrap distributions, we computed the mean, standard deviation, and 95% confidence intervals (2.5th and 97.5th percentiles). The standard deviation of the bootstrap distribution served as the error bar in visualizations (Figs. 4 and 5). The median values were also calculated as a robustness check, but only the means and their associated standard deviations were used for statistical testing and plotting (Figs. 4 and 5). This approach provided a nonparametric estimate of variability, enabling reliable comparisons across states and metal ion conditions.

Figure 4.

Mn²⁺ shifts the α1 and α7 helices earlier than Mg²⁺/Ca²⁺. (A) Average distance between the ADMIDAS ion and the backbone oxygen atom of residue M335. On the right, two representative snapshots of the α7 helix from state 3 in Mg²⁺/Ca²⁺ (cyan) and Mn²⁺ (green) are shown in secondary structure representation. These are overlaid with the α7 helix from the closed (PDB: 3T3P, in yellow) and open (PDB: 2VDR, in purple) conformations. The three ions are represented as large spheres: cyan for Mg²⁺/Ca²⁺ and green for Mn²⁺. The distance between the ADMIDAS ion (rightmost large sphere) and M335 (small red sphere) is indicated by a horizontal black line. M335 from the closed (small yellow sphere) and open (small purple sphere) states are also shown to highlight their increased distance from the ADMIDAS ion as the conformation transitions from bent to open. Mn²⁺ (green) shifts the oxygen atom of M335 further from the ADMIDAS ion than Mg²⁺ (cyan). (B) Difference in the average distance of the COM of the α7 helix of each integrin state from the COM of the α7 helix in the closed state (PDB: 3T3P, yellow sphere), defined as d1, and from the open state (PDB: 2VDR, purple sphere), defined as d2. On the right, a representative snapshot of the α7 helix from state 3 in the Mg²⁺ (cyan) and Mn²⁺ (green) conditions is shown. Bars are normalized so that 0 nm represents the same position as the closed state. (C) Average percentage of helicity of the α1 helix across eight integrin states and two ion conditions, each simulated for 300 ns. On the right, a representative snapshot of the α1 helix from state 5 in Mg²⁺ (cyan) and Mn²⁺ (green) is overlaid with the α1 helices from the closed and open states to show changes in helical content. Bootstrap resampling with replacement was performed on six replicas for each integrin state and ion condition. Error bars indicate the standard error of the mean. An asterisk indicates a p value of <0.05 using the Wilcoxon rank-sum test on the means from bootstrap samples comparing Mg²⁺ versus Mn²⁺ for each integrin state.

Figure 5.

Mn²⁺ promotes more hydrogen-bonds between an RGD peptide and α_IIBβ₃ integrin and increases the occupancy of specific known interactions compared with Mg²⁺. (A) Average number of hydrogen-bonds formed between the RGD peptide and the α_IIBβ₃ integrin across simulations. (B) Representative snapshot from state 8 showing the four hydrogen-bonds previously identified in x-ray crystallography, corresponding to the data quantified in (C–F). (C) Average occupancy of the hydrogen-bond between the side chain of ligand residue R493 and the side chain of the β-propeller residue D224. (D) Average occupancy of the hydrogen-bond between the side chain of ligand residue D495 and the main chain of integrin β3 residue Y122. (E) Average occupancy of the hydrogen-bond between the side chain of ligand residue D495 and the main chain of integrin residue S123 in βI domain. (F) Average occupancy of the hydrogen-bond between the side chain of ligand residue D495 and the main chain of integrin β3 residue N215. Data were collected from all frames over 300 ns of simulation time, across eight integrin states simulated under two ion conditions (Mg²⁺ and Mn²⁺), with six replicas per state and condition. Bootstrap resampling was performed on replica data from each state and condition, treated as independent samples, to estimate the mean and standard error of the mean (error bars). Asterisks indicate statistically significant differences (p < 0.05) based on Wilcoxon rank-sum tests comparing Mn²⁺ and Mg²⁺ conditions within each state.

Results

Effects of Mn²⁺ and Mg²⁺/Ca²⁺ on the structural dynamics of α_IIbβ₃ integrin

To investigate how Mn²⁺ versus Mg²⁺/Ca²⁺ influences the structural dynamics of the α_IIbβ₃ ligand-binding site, we began by analyzing crystal structures of the β-propeller and βI domains bound to an RGD motif (26). These structures represent eight activation states, ranging from fully closed (state 1, black in Fig. 2 A) to fully open (state 8, white), with six intermediates (states 2–7, shades of gray). In Mg²⁺/Ca²⁺ systems, Mg²⁺ occupied the MIDAS site and Ca²⁺ the ADMIDAS and SyMBS, while in Mn²⁺ systems, Mn²⁺ was placed at all three sites (Figs. 1 D and 2 B); both setups included four distal Ca²⁺ ions in the β-propeller domain (Fig. 2 B). Initial cryo-EM structures revealed that the α-carbon (Cα) root mean-square displacement (RMSD) across eight conformational states remained low, averaging below 1.2 Å—indicating minimal structural variation in the starting models (Fig. 3 A). However, during equilibrium MD simulations, Cα-RMSD values increased rapidly before plateauing at approximately 2–3 Å by ∼50 ns (Fig. S1), with the standard deviation across sliding time windows stabilizing after ∼10 ns (Fig. S2), indicating that the systems entered stable conformational ensembles. Mg²⁺/Ca²⁺-bound integrins showed lower structural variability in early activation states (1,2,3,4,7,8), with Cα-RMSD values of 1.5–2.3 Å (Fig. 3 B) compared with 1.6–3.2 Å in Mn²⁺ systems (Figs. 3 C and S3). In states 7–8, RMSDs for both ion systems converged (∼1.8–2.7 Å), indicating attenuated ion-specific effects on the ligand binding site in later activation states (Figs. 3, B, C and S3). In summary, Mn²⁺ enhances structural rearrangements of the α_IIbβ₃ binding site during early activation but shows minimal influence in fully activated states, aligning with its inability to fully stabilize the high-affinity conformation (10,16,38).

Figure 2.

Structures and systems setup. (A) Zoom-in overlay of states 1 through 8 (light gray to black, respectively), with indication of residues 123, 126, and 127 in the β1-α1 loop, and residue 251 in the β4-α4 loop. The backbone oxygen of residue M335 in the β6-α7 loop is highlighted with two red circles connected by a red dotted line to show its change in position across states. (B and C) Space-filling representations of the β-propeller (blue) and βI (orange) domains of α_IIbβ₃ integrin in state 1 at the onset of the simulations, with the Mg²⁺/Ca²⁺-containing system on the left (B) and Mn²⁺ on the right (C). Pink spheres represent Ca²⁺, yellow spheres represent Mg²⁺, green spheres represent Mn²⁺, red spheres represent oxygen, and cyan color represents carbon.

Figure 3.

Evaluation of structural differences between the eight structural states of the α_IIBβ₃ ligand-binding site before and after MD simulations. (A) Comparison between RMSD of the Cα atoms of the β-propeller and βI domains of RGD-bound integrin across initial states. (B) Comparison between RMSD of the Cα atoms of the β-propeller and βI domains across representative conformations using Mg²⁺/Ca²⁺ and (C) Mn²⁺. Results are obtained from 5 to 14 independent simulations, using clustering analysis on all frames within the last 10 ns of all replicas based on identifying a representative integrin conformation with the smallest RMSD compared with the ensemble of conformations.

Distance between M335 and ADMIDAS ion in Mn²⁺ versus Mg²⁺/Ca²⁺

To examine how Mn²⁺ versus Mg²⁺/Ca²⁺ affects the β6-α7 loop—key to ligand binding and activation—we measured the distance between the ADMIDAS ion and the carbonyl oxygen of M335 in the β6-α7 loop across the eight α_IIbβ₃ states. In the initial cryo-EM structures, states 1–5 showed distances of 2–3 Å, state 6 around 6 Å, and states 7–8 approximately 14 Å (Fig. 4 A, horizontal black lines). After 300 ns of MD, Mn²⁺ induced a marked shift in states 1–4, increasing the distance between the ADMIDAS ion and the carbonyl oxygen of M335 to 5–7 Å. This change suggests reduced coordination of the ion with this site and a downward movement of the β6-α7 loop, consistent with Mn²⁺'s ability to disrupt the coordination of the ADMIDAS ion and promote integrin activation (20,39,40). In contrast, Mg²⁺/Ca²⁺ systems maintained shorter distances between the ADMIDAS ion and M335 in the β6–α7 loop (∼3 Å in states 1–4, ∼6 Å in states 5–6), reflecting a more stable, bent-like conformation of the ligand-binding site. In states 7–8, distances between the ADMIDAS ion and M335 in the β6-α7 loop were similar across both ion conditions, indicating convergence to a stable (∼14 Å), open conformation of the ligand-binding site. These results suggest Mn²⁺ enhances early-stage movements of the β6–α7 loop, while Mg²⁺/Ca²⁺ stabilize the bent state—aligning with Mn²⁺’s partial, and not complete, activation of α_IIbβ₃ (10,16,38).

Effect of divalent cations on α7 helix movement

To assess how divalent cations affect the movements of the α7 helix in the βI domain, we measured its COM distance in each of our eight conformations from the corresponding COM in closed (PDB: 3T3P (41)) and open (PDB: 2VDR (28)) α_IIbβ₃ structures. T The difference between the COM distances from the closed and open α_IIbβ₃ structures (denoted as d1−d2) serves as an indicator of progression toward the open conformation, with higher values reflecting greater displacement from the closed state and toward the open configuration. Mn²⁺ caused significantly larger displacements in early activation states (2,3,4,7,8), with d1−d2 ranging from 2.3 to 4.1 Å, compared with <2.5 Å for Mg²⁺/Ca²⁺. In later states (9,10), both cations showed similar displacement from the closed state (4–8 Å), indicating convergence toward the open conformation. Thus, Mn²⁺ promotes an earlier and more pronounced movement of the α7 helix during activation.

Impact of Mn²⁺ on the α1 helix transition during integrin activation

To further assess how Mn²⁺ affects βI domain structural rearrangements, we analyzed the helical content of the α1 helix across the eight activation states (Fig. 4 C). Mn²⁺ increased α1 helicity by up to 20% compared with Mg²⁺/Ca²⁺ in early states (1,2,3,4,7), whereas differences diminished in later states (9,10), suggesting Mn²⁺ stabilizes the α1 helix during early activation.

Impact of Mn²⁺ versus Mg²⁺/Ca²⁺ on the interactions between integrin and the RGD motif

To evaluate how Mn²⁺ versus Mg²⁺/Ca²⁺ affects α_IIbβ₃-RGD interactions, we measured the occupancy of key hydrogen-bonds identified in crystal structures (22,28,42). In all eight activation states, ligand-protein interactions were maintained in both systems, but the average hydrogen-bond count was generally slightly yet significantly higher with Mn²⁺ (Fig. 5 A). We then examined specific hydrogen-bonds: R493-D224 in the β-propeller, and D495 with Y122, S123, and N215 in the βI domain (Fig. 5 B). The R493-D224 bond showed consistently lower occupancy with Mn²⁺ compared with Mg²⁺/Ca²⁺ across most states, except states 1 and 8 (Fig. 5 C). In contrast, Mn²⁺ enhanced D495-Y122 occupancy with activation, surpassing 40%, while Mg²⁺/Ca²⁺ decreased from ∼20% to nearly 0% (Fig. 5 D). Similar trends were observed for D495-S123 and D495-N215, where Mn²⁺ increased occupancies to 50–60%, whereas Mg²⁺/Ca²⁺ remained near or below 20% across states (Fig. 5, D and E). In the Mn²⁺ systems, repositioning of the ADMIDAS ion toward the β-propeller domain during early activation stages (Fig. 4 A) reduced its electrostatic interaction with the RGD motif, while enhancing that of Mn²⁺ at the SyMBS (Fig. S5). In later activation stages, Mn²⁺ at the ADMIDAS site exhibits strengthened electrostatic interactions with the RGD motif (Fig. S5). A more detailed description of electrostatic interaction analyses is provided in the supporting material. In summary, Mn²⁺ strengthens critical RGD-integrin hydrogen-bonds, supporting its role in stabilizing integrin-RGD interactions more than Mg²⁺/Ca²⁺.

Discussion

The conformational activation of α_IIbβ₃ is tightly regulated and essential for proper platelet function in both physiological and pathological contexts. Divalent cations such as Mn²⁺ have long been known to enhance this activation by increasing the integrin’s affinity for extracellular ligands. (23,43,44,45). However, the precise molecular mechanisms underlying the increase in integrin affinity by Mn²⁺ remain incompletely understood (40). In this study, we employed equilibrium MD simulations to explore how Mn²⁺ binding at the canonical metal-binding sites—ADMIDAS, MIDAS, and SyMBS—affects the structural dynamics of the α_IIbβ₃ ligand-binding domains. We compared the influence of Mn²⁺ to that of Mg²⁺ and Ca²⁺, which occupy these sites in physiological conditions.

Our simulations reveal that Mn²⁺ accelerates the movement of the M335 backbone carbonyl oxygen of the β6-α7 loop (Fig. 4 A) and causes a more pronounced downward shift of the α7 helix in the βI domain relative to Mg²⁺ and Ca²⁺ (Fig. 4 B). Mn²⁺ also promotes earlier stabilization of the α1 helix (Fig. 4 C) and enhances the interactions of the integrin binding site with the RGD ligand (Fig. 5). Mn²⁺ is known to activate integrins by competing with Ca²⁺ at the ADMIDAS site (39), and recent studies further suggest that changes in metal ion coordination can alter loop movements within the βI domain, thereby stabilizing alternative conformational states (16). Our findings are consistent with known mechanisms of integrin activation, which involve localized rearrangements within the βI domain, including shifts in the β1-α1 loop, reorganization of the α1 helix, and changes in ion coordination at the ADMIDAS site. Prior studies have shown that such rearrangements help stabilize the ligand-bound, high-affinity state of the integrin (20), thereby enabling subsequent large-scale conformational changes (Fig. 1, A and B) (24,46).

Although several structural studies have captured integrins in specific conformational states (7,22,23,26,47), they have not fully explored the dynamic structural rearrangements that occur as integrins transition between these states. This study leverages MD simulations to investigate how Mn²⁺ ions modulate the structural dynamics of the ligand-binding interface. This approach provides new insights into the mechanisms by which Mn²⁺ modulates integrin behavior, in contrast to previous studies that focused on static conformations or limited activation stages (48,49,50,51,52).

Importantly, our results demonstrate that Mn²⁺ facilitates conformational transitions along the same pathway as physiological activation, with earlier and more pronounced molecular rearrangements within the ligand-binding interface. These observations also support the idea that Mn²⁺ enhances integrin activation by promoting intermediate, high-affinity states of the βI domain. However, even though these changes mirror those seen during normal integrin activation, our simulation setup does not consider the full ectodomain and therefore cannot address whether these local rearrangements propagate into global integrin extension. This distinction is particularly relevant given the conflicting literature on Mn²⁺-mediated integrin activation. Cryo-EM and small-angle x-ray scattering studies have suggested that Mn²⁺ stabilizes an extended, high-affinity conformation (10,11,12,13,14), but other structural techniques, such as small-angle neutron scattering and x-ray crystallography, have reported little to no evidence of extension in the presence of Mn²⁺ (15,16,17,18,19,20). Some studies further suggest that Mn²⁺ is a less potent activator than other agents (7,38,53,54). These discrepancies raise the possibility that high-affinity ligand binding and global integrin extension may be at least partially decoupled. Our findings are consistent with this hypothesis: Mn²⁺ appears to promote a locally open conformation of the ligand-binding site even in the absence of full integrin extension, suggesting that Mn²⁺ may activate integrins through localized changes.

In conclusion, although further work is needed to determine whether the effects of Mn²⁺ extend to the full protein structure, our study demonstrates that Mn²⁺ induces structural changes in the βI domain more readily than physiological conditions (Mg²⁺/Ca²⁺). These changes are consistent with the observed increase in ligand-binding affinity and occur along the same structural pathway as physiological activation. Collectively, these insights provide a deeper understanding of integrin activation and its dependence on metal ion coordination, with implications for therapeutic strategies targeting integrin-mediated processes.

Limitations

Our study provides new insights into the conformational events associated with Mn²⁺-mediated integrin activation, but several limitations must be acknowledged. First, our simulations focused exclusively on the β-propeller and βI domains of α_IIbβ₃, omitting the remainder of the ectodomain and transmembrane helices (Fig. 1, A and B). Therefore, we cannot determine whether the observed local changes would propagate to full integrin extension. Furthermore, the use of equilibrium MD simulations inherently limits our ability to observe rare or slow conformational transitions that may occur on longer timescales.

Another limitation involves the ligand used in our simulations. We modeled binding with a simple RGD peptide, whereas physiological integrin ligands are typically large, multidomain proteins that engage integrins through extended binding interfaces. The effects of Mn²⁺ on integrin activation may differ in the context of these native ligands. Moreover, integrin activation in vivo is influenced by mechanical forces exerted on the ligand-receptor complex (55,56,57), which were not included in our simulations. Such forces play a known role in mechanotransduction and could modulate the structural pathway of activation.

Our force field parameters for divalent cations present further limitations. We used the default CHARMM36m parameters (31), which are commonly applied in biomolecular simulations. However, these parameters were not specifically optimized for ion coordination with protein residues and the use of nonbonded ion models may not fully capture the subtleties of these interactions.

Conclusions

In summary, our study demonstrates that Mn²⁺ binding to α_IIbβ₃ facilitates earlier molecular conformational changes in the βI domain and enhances interactions with the RGD motif—key events critical for α_IIbβ₃ activation and platelet function. These conformational changes occur more readily than those induced by Mg²⁺ or Ca²⁺, leading to earlier stabilization of key secondary structural elements within the ligand-binding site (Fig. 4) and enhanced ligand engagement (Fig. 5), and align with the established pathway of physiological integrin activation. Specifically, these structural rearrangements are consistent with crystal structures showing that the open, ligand-bound βI domain adopts a nearly identical conformation whether Mn²⁺ is bound at all three metal-binding sites or Mg²⁺ and Ca²⁺ occupy their respective positions (22,26,28).

Although our simulations do not capture full integrin extension, they support the idea that Mn²⁺ enhances integrin function by promoting a high-affinity conformation at the ligand-binding interface. This observation is consistent with previous studies showing that Mn²⁺ increases ligand-binding affinity across multiple conformational states (40,58,59,60). Our results suggest that Mn²⁺ achieves this by favoring a locally open, high-affinity structure within the ligand-binding interface, without necessarily initiating large-scale conformational transitions. This interpretation also supports the broader understanding that Mn²⁺ enhances integrin affinity for extracellular ligands (40,53,58,61,62) by acting through early activation mechanisms such as those operating under physiological conditions.

Despite the localized focus of our simulations, these results offer meaningful mechanistic insights into how divalent cations regulate integrin structure and function. To determine whether Mn²⁺ can fully activate α_IIbβ₃, future studies should incorporate complete ectodomain models, more physiologically relevant ligands, and account for the mechanical forces present in the extracellular environment that act directly on the ligand binding interface. In addition, longer-timescale simulations or enhanced sampling methods will be necessary to observe potential large-scale structural transitions. Refinement of metal ion force fields and more accurate modeling of ion-protein coordination will also improve simulation accuracy. Together, these directions will help clarify the structural basis of integrin activation and inform therapeutic strategies targeting integrin-mediated signaling in thrombosis and other diseases.

Data and code availability

All MD simulations trajectories, analysis scripts, input configurations, MD parameter files, including topologies, force field, and simulation parameters, are provided in the following repository: https://github.com/tamarabidone/ManganeseVsMagnesium_EQ_MD.git.

Acknowledgments

NIH National Institute of General Medical Sciences grant R35GM14749 (to T.C.B.). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. We acknowledge the Center for High Performance Computing at the University of Utah. We also acknowledge Dr. Tomasz Skora and Onkar Joshi for useful discussion.

Author contributions

Supervision, T.C.B.; conceptualization, T.C.B. and R.E.C.; funding acquisition, T.C.B.; project administration, T.C.B.; resources, T.C.B.; resources, T.C.B.; writing – original draft, R.E.C. and R.K.; writing – review & editing, T.C.B., R.E.V., and R.K.; methodology, R.E.C.; software, R.E.C.; validation, R.E.C.; formal analysis, R.E.C.; investigation, R.E.C.; data curation, R.E.C.; visualization, R.E.C. and R.K.

Declaration of interests

The authors declare no competing interests.

Declaration of generative AI and AI-assisted technologies in the writing process

During the preparation of this work, the authors used GitHub Copilot to assist with syntax and grammar. The content generated by this tool was subsequently reviewed and edited by the authors, who take full responsibility for the final content of the manuscript.

Editor: Lalima Madan.