Computational protocol: The Internal Dynamics of Fibrinogen and Its Implications for Coagulation and Adsorption

Similar protocols

Protocol publication

[…] Fg undergoes large bending motions in all the simulations that we have performed. Principal component analysis (PCA) is used to quantify these motions. The dominant principal components of motions (PCA modes, ) of the Fg protomer are the same in all sampled trajectories as revealed by a large overlap (see section) between the three dominant modes ranging from 0.6 to 0.9 between simulation subsets. In particular, the large overlap observed between the dominant PCA modes in both mono- and di-glycosilated and unglycosilated Fg protomer trajectories shows that the carbohydrate clusters do not affect the large scale dynamics of Fg in solution. Similarly, the dimerization state does not induce any noticeable change in the large scale motion of Fg. Dimer and monomer simulations show consistent hinge bending and the PCA-mode overlap between isolated protomer and dimerised protomer simulations is large. The difference between the dimer and protomer simulations is limited to the dynamics of the dimerization interface, where the absence of the disulphide bridges with the other protomer results in expectedly larger root mean square fluctuations (RMSF) localized to the residues α27–44 β 58–75 and γ 14–18 (see in Supplementary Information). Because of the overlap of the largest PCA modes in the different simulation sets, the analysis presented here is done using all the available data merged together in a single set, which improves the statistical significance of the results. The first three PCA modes span the degrees of freedom associated with bending at a hinge point in the coiled-coil region (), while the 4th PCA mode is related to a pure torsion of the coiled coil along its axis (not shown). The motions are reversible as shown by the time series of the PCA projections () Lower ranking PCA modes provide smaller contributions to the overall variance so they will not be analyzed further.The program DynDom [], applied to the extremal structures observed along the first PCA mode () of the Fg protomer, has been used to identify the regions of the molecule which are more rigid in our simulations, as well as the connecting hinge regions. DynDom reports the presence of two relatively rigid regions, separated by a hinge. The E region and the N-terminal part of the coiled-coil region represent one of the two rigid domains, while the C-terminal part of the coiled-coil region along with the D region represent the second. The hinge region is located approximately in the middle of the coiled-coil region and includes the break in the α-helical structure of the γ chain, which gives rise to a flexible loop (residues γ70–78), along with the neighbor residues on the Aα and Bβ chains (). The break in the α-helical structure of the γ-chain is facilitated by two proline residues.The bending around the identified hinge can be described by a bending angle γ and a torsion angle φ defined using groups of atoms from the E region, the hinge region and the D region (). The γ and φ angles strongly correlate with the projections along the dominant PCA modes: γ shows a linear correlation coefficient of 0.96 with a linear combination of the first 3 PCA modes (1st-2nd-3rd/10, see section), while φ has a 0.73 linear correlation coefficient with a linear combination of the 1st, 2nd and 4th PCA modes (1st/3+2nd/2+4th/2). Our simulation data show a consistent and significant bending occurring at the hinge region and reaching bending angles below 90 degrees (). The time it takes for the Fg structure to reach a bending angle below 110 deg from conformations similar to the crystal structure (bending angle above 150 deg) is 19±1 ns along the trajectories, averaged over the 12 observed events (see and in Supplementary Information, for examples). The reverse process occurs twice in the simulations, taking 20 and 26 ns. The simulations of the full Fg dimer do not show significant correlations between the angle values observed at the two hinges.Comparison of the crystallographic structures of Fg coiled-coil regions from various organisms already hinted at the presence of a flexible hinge []. This hypothesis is also supported by hydrogen-deuterium exchange experiments []. The latter are in good agreement with our simulations: amino acids from the coiled-coil region with lower helical probability in the simulations () correspond to amino acids with low protection factors in the experiments. Our simulations help to clarify the fact that the presence of a flexible hinge is an intrinsic feature of the coiled-coil region, and not an artefact due to sequence or crystal-packing differences in the compared crystal structures. The hinge is positioned on the non-helical segment of the γ chain (γ70–78), most probably due to the resulting reduction in the stiffness of the coiled coil. This segment is non-helical also in the other crystallized Fg structures [, ]. In addition, this segment has markedly helix-breaking features in most of the available Fg sequences from vertebrates that we have analyzed, showing a large density of proline and glycine residues ().As can be seen in , the appearance of prolines characteristic for the hinge region happens in three stages during evolution. The lamprey (Petromyzon marinus) has no proline, fish (Danio rerio) have a single proline and tetrapods have two or more. These steps coincide with major changes in the clotting cascade, namely the appearance of the intrinsic pathway in jawed vertebrates and the addition of the contact pathway (linking blood clotting and immune response) in tetrapods [–]. We used the program DisEMBL [] to analyze the Fg sequences and identify disordered or loop segments in the coiled-coil region. The region corresponding to the non-helical segment of the γ chain is marked as a hot-loop with high probability in most of the sequences, with the exception of lamprey Fg which, as mentioned above, is known to have a simpler coagulation mechanisms than the other vertebrates [] (). This analysis supports the idea that the non-helical segment of the γ chain provides a function that is strongly conserved across vertebrates. Our simulations suggest that this conserved function is linked to the bending motion of the coiled-coil region.Besides providing flexibility to the individual Fg molecules as well as the fibrin fibers [], the bending at the hinge may help expose the plasmin cleavage sites located nearby on the coiled-coil region, an hypothesis already proposed in ref. []. Our simulations strongly support this hypothesis showing that the α-helical structure around the plasmin cleavage sites Aα104–105 and Bβ133–134 is partly disrupted by the bending motions, and the exposure to the solvent of the involved peptide bonds increases (). The α-helical structure lost during bending is generally replaced by coil structure in our simulations and not by extended β-sheet structure as observed in experiments [] and simulations [] of fibrin subject to tension. The observation of the transition to extended β-sheet may then be linked to the presence of tension along the molecule and/or require significantly larger times than those simulated here to occur spontaneously. It is known that the fibrin molecules straighten when they are integrated into protofibrils []. This does not mean that the flexibility of individual molecules is lost. The twisting of fibrin fibers [] compresses molecules in the center of the fiber and stretches them on the perimeter. If the hinge bending is necessary to accommodate such deformation, it is reasonable to believe that the bending motions at the hinge may actually be reduced by tension applied along the fiber axis. Thus, fibrinolysis assisted by the bending motions at the hinge may selectively take place on fibrin molecules subject to reduced tension. This hypothesis is supported by experimental evidence indicating reduced plasmin fibrinolytic effectiveness on fibrin fibers subject to mechanical tension [].Bending may also play an important role in fibrin polymerization. According to the recently proposed Y-ladder model of fibrin polymerization [], in the early stages of the process, the protofibril grows as a single strand where each monomer is connected to the next by a single A-knob-a-hole bond between the E domain of one molecule and the D domain of the next in the chain, giving rise to a Y-ladder structure. The latter is later transformed in a double-stranded protofibril by the formation of the other knob-hole bonds and D-D bonds. Hinge bending as observed in the present simulations can provide the necessary flexibility to accommodate new molecules in the growing fiber. In support of this role, Fig.6C of ref. [] seems to depict single stranded growth events involving bent fibrin molecules as indicated by arrows 5 and 6. More in general, an imperfect straightening of Fg during integration into a double-stranded protofibril would provide a mechanism for the formation of branch points in the fibrin network.In the present simulations, several parts of the molecule have been omitted because they are unresolved in the crystal structure, and too long (and disordered) to be meaningfully sampled in the course of our simulations. These include the the βN stretch (residues β1–57) and the αC stretch (residues alpha 201–562). Both parts are mostly disordered (apart from a tendency to form a compact conformation in the C terminal part of the αC domain) as revealed by several experimental techniques, including H/D exchange [], X-ray crystallography [, , ] and NMR []. In fibrinogen, the compact domains of the αC region from the two protomers are believed to interact with each other and with the fibrinopeptide B (part of the unresolved βB stretch) in proximity of the E region of the molecule, while in fibrin the alphaC domain are believed to move away from the E domain and take part in inter molecular interactions []. Although none of these unresolved stretches are known to interact with the coiled coil region of the protein, in principle they could interfere with the bending movements observed in our simulations. The lack of evidence for interactions with the coiled-coil region and the flexible nature of these stretches suggest that their presence will not directly contribute to a stiffening of the hinges of the coiled-coil region. On the other hand, we cannot exclude that their presence may bias bending along particular directions by hindering hinge movements due to excluded-volume effects. […]

Pipeline specifications

Software tools DynDom, DisEMBL
Applications Small-angle scattering, Protein structure analysis
Diseases Blood Coagulation Disorders, Hemorrhage, Hemostatic Disorders