1. Introduction
When a biomaterial is inserted into the body, a protein film consisting of plasma proteins from the blood is immediately formed on the implant surface. The 340 kDa glycoprotein fibrinogen is an abundant protein in the blood of vertebrates and is primarily involved in blood clotting and readily adsorbs to biomaterial surfaces upon implantation. In a wound, thrombin polymerizes fibrinogen into fibrin that, together with platelets, form a clot over the wound. When adsorbed to the surface of a biomaterial, fibrinogen can trigger inflammatory responses and subsequent formation of a fibrous capsule that can lead to failure or loss of function of the biomaterial. The fibrinogen molecule harbors epitopes that are recognized by both human and bacterial cells through specific integrins [
1,
2,
3,
4]. An infection on an implant can have severe consequences and is one of the most common causes of implant failure and need for revision surgery. Surface modifications at the nanoscale have previously been suggested as a promising approach to prevent adhesion of bacteria and development of biomaterial-associated infections on implants [
5]. A leading cause for biomaterial-associated infections is the bacterium
Staphylococcus epidermidis [
6], which is normally present on human skin. A recent study found that the ability of
S. epidermidis to adhere on adsorbed fibrinogen is highly dependent on the nanostructure of the underlying substrate, possibly by affecting the orientation or conformation of the adsorbed fibrinogen molecule [
7].
In the present study, we used coarse-grained Monte Carlo simulations to investigate how fibrinogen adsorbs to smooth and nanostructured surfaces (with attached nanoparticles). This enabled us to analyze in more detail how fibrinogen adsorbs and explain the experimental findings described above. We investigated the effect of surface charge, the effect of nanoparticle size, the angles between bound fibrinogen and the surface, and the effect of the disordered fragments (the C-chains) of fibrinogen.
Human fibrinogen consists of two symmetrical halves that each contain three polypeptide chains called A
(610 amino acid residues), B
(461 amino acid residues), and
(411 amino acid residues). The chains are connected to each other by several disulfide bonds. Fibrinogen folds into a 45-nm-long rod with two thicker nodules (the D domains) at the ends, and one nodule in the middle of the rod (the E domain). The C-terminal 410 amino acid residues of the A
-chains are not visible in the crystal structure [
8] and are here counted as the disordered
C-chains, which extend out from the two D domains (sometimes only the C-terminal 400 residues are included). NMR studies have shown that the
C-regions of bovine fibrinogen each contain a structured domain of approximately 60 amino acid residues [
9]. However, the stability of this structure is low. Human fibrinogen seems to have
C-domains with similar structure but even less stable than the bovine ones [
10].
In one of the simplest models that has been used for simulations of fibrinogen, the protein is approximated as an elongated ellipsoid [
11] (see
Figure 1a). In another study of fibrinogen adsorption, the protein was described as three connected squares [
12] (see
Figure 1b). Zhdanov et al. described fibrinogen as a linear pentamer with a monomer diameter of 7.5 nm [
13] (see
Figure 1c). A somewhat more detailed model that has been used to describe fibrinogen in simulations is a linear chain consisting of 23 touching spheres of different diameters, mimicking the differences in thickness at different parts of the fibrinogen rod [
14,
15] (see
Figure 1d). The spheres at the ends had a diameter of 6.7 nm, the one in the middle 5.3 nm, and the remaining ones 1.5 nm. This model was later extended by adding two side arms representing the
C-chains [
16] (see
Figure 1f). The side arms were also linear but the angle
between the side arms and the body of the protein was varied.
Atomistic molecular dynamics simulations of the main body of fibrinogen (excluding the
C-chains and other flexible parts not visible in the crystal structure) in solution have also been performed, revealing bending motions of the protein [
17]. An atomistic representation of the fibrinogen crystal structure is shown in
Figure 1h. The simulations showed that two hinges are responsible for the flexibility of the protein while the rest of the fibrinogen main body does not undergo large conformational changes. Based on these results, a simplified coarse-grained model of fibrinogen was developed (see
Figure 1e). In that model, fibrinogen is represented by a stiff central rod connected to two other rods that can pivot around the hinges. The ends of the protein and the central domain are represented as spheres.
The atomistic simulations of fibrinogen in solution were followed by a study of fibrinogen adsorption on mica and graphite, where one half of the symmetric fibrinogen rod was modeled atomistically [
18]. The binding of fibrinogen to gold nanoparticles has been studied using a coarse-grained model where fibrinogen was modeled from the crystal structure determined by Kollman et al. [
8] and each amino acid residue was represented as a sphere (see
Figure 1g). The amino acids interacted via a bonded potential (a sum of potentials for bonds, angles and dihedrals) and a nonbonded potential (two Lennard-Jones-type potentials—one local and one nonlocal) [
19].
We used the model depicted in
Figure 1g, where each amino acid of the crystal structure is coarse-grained into a sphere. However, we considered the molecule as completely rigid and thus the spheres did not move relative to each other. Such a model was also used by Lopez and Lobaskin [
20]. This model has an intermediate level of detail and takes the charge distribution over the molecule into account. This enables us to study the electrostatic effects behind the adsorption of fibrinogen.
We found that the main body of fibrinogen protrudes more into solution the larger the curvature of the surface. The fibrinogen main body is anchored to the surface by one of its D domains in the same way regardless of the curvature, thus the effect on the protein orientation seems to be solely a function of the curvature. Our model included only electrostatic interactions (and excluded volume), but the results are still similar to those of Lopez and Lobaskin, who studied fibrinogen adsorption onto hydrophobic surfaces [
20].
It has been suggested that fibrinogen may adsorb to negatively charged surfaces by means of electrostatic interactions between the net positively charged
C-domains and the surface [
21,
22,
23,
24]. However, in our present study, the
C-chain was found to have a net negative charge. This difference is likely to be due to the treatment of the histidines. While we treated them as having zero charge, Doolittle et al., for example, used a histidine p
of 6.7 to calculate the charge of the A
-chain at pH 7.3, giving a non-negligible charge on the histidines [
25].) Our comparison of the free energies of adsorption of the main body and the disordered fragments reveals that there is a stronger attraction between one disordered fragment and the surface than between the fibrinogen main body and the surface.
4. Conclusions
We used coarse-grained simulations to study the adsorption of the fibrinogen main body, as well as the fibrinogen C-chain, on negatively charged smooth surfaces, nanospheres, and flat surfaces with attached nanospheres.
Our study confirms that the disordered C-chains of fibrinogen are important for adsorption on negatively charged surfaces. A single C-chain attaches more firmly than the fibrinogen main body.
When fibrinogen adsorbs to surfaces with attached nanospheres, the available surface area is generally what determines the proportion of fibrinogen that is adsorbed. However, when the different parts of the fibrinogen molecule are adsorbed to spherical nanoparticles in solution, the adsorbed amount per surface area increases as the size of the nanoparticle decreases. The reason could be that that negatively charged amino acid residues feel less repulsion from the negatively charged nanoparticle the smaller it is, since the surface area is smaller.
We found that the main body of fibrinogen protrudes more into solution as the curvature of the surface increases, and that the main body of fibrinogen is anchored to the surface through one of is D-domains. This was found true regardless of the surface curvature and the orientation therefore seems to be solely a function of the surface curvature. Thus, we hypothesize that fibrinogen adsorbed on surfaces with attached nanoparticles makes cell-binding epitopes less available to attaching cells. This could explain previous findings where the adhesion of S. epidermidis was hampered on fibrinogen adsorbed to nanostructured compared to smooth substrates.