Understanding surface-adsorption of proteins: the Vroman effect

Pol Vilaseca,1, 2 Kenneth A. Dawson,3 and Giancarlo Franzese1

1
Departament de Fı́sica Fonamental, Facultat de Fı́sica,
Universitat de Barcelona, Diagonal 645, 08028, Barcelona, Spain
2
School of Mathematics, Trinity College Dublin, Ireland
3
Center for BioNano Interactions (CBNI), University College of Dublin, Ireland∗
(Dated: February 20, 2012)
It is now well accepted that cellular responses to materials in a biological medium reflect greatly
the adsorbed biomolecular layer, rather than the material itself. Here, we study by molecular
arXiv:1202.3796v1 [cond-mat.soft] 16 Feb 2012

dynamic simulations the competitive protein adsorption on a surface (Vroman-like effect), i.e. the
non-monotonic behavior of the amount of protein adsorbed on a surface in contact with plasma as
a function of contact time and plasma concentration. We show how the effect can be understood,
controlled and inverted.

PACS numbers: 87.10.Tf, 81.16.Fg, 87.85.J-

When nanoparticles are in contact with blood plasma, (b)

(a)
or other biological fluids, biomolecules rapidly coat the
bare surface in a relatively selective manner [1]. It is in-
creasingly accepted that the early biological responses
to nanoparticles will be determined by the adsorbed
(c) (d)
biomolecules rather than the pristine surface alone [2, 3].
Because of their size [2, 4] nanoparticles are trafficked by
active transport processes throughout the organism, us-
ing the information from the protein sequences associated
with the surface of nanoparticles. Unlike the situation
for flat macroscopic surfaces say of medical implants, for FIG. 1. Schematic representations of different proteins ad-
nanoparticles the protein environment changes in differ- sorbed on the surface. (a) Alb is a globular protein charac-
ent compartments of cells and organs, as the nanoparticle terized by the radius RA . In all the panels the continuous
line represents the surface profile. IgG or Fib are represented
traffics. This has lent urgency to the modern interest in as ellipsoids that interact with each other along the long axis
understanding the phenomenon at a more fundamental RS,i (b), or along the short axis Ri (c). In the last case they
level [2]. Still, we can learn a lot from an understanding are partially deformed (denatured) by the surface adsorption,
of the process for flat surfaces [5]. Studying the adsorp- represented here by a partial overlap of the ellipsoids with the
tion of Fibrinogen on a surface in contact with blood surface. (d) At a random moment, the adsorbed ellipsoids can
plasma, Vroman found that the surface concentration of have different arrangements.
Fibrinogen shows a maximum at an intermediate contact
time, indicating that Fibrinogen is replaced with time by
one or more families of different proteins [6]. The phe- competitive adsorption has been observed [11]. We show
nomenon is not specific to Fibrinogen, but is a general that the sequence of adsorption can be explained in terms
effect for many other proteins [7]. The plasma proteins of the relative bulk concentrations, difusivities and sur-
compete for the occupation of the surface, resulting in a face affinities of the proteins and that by thermal or
sequential competitive adsorption, known as the Vroman chemical energy-depletion is possible to control and in-
effect. vert the effect.
Alb is a globular protein, with an almost spherical
The effects depends on numerous factors such as the shape. We model Alb-Alb interaction as
plasma dilution, the temperature, and the specific surface
24
chemistry [8]. In highly concentrated plasma, the sequen-

2RA
tial adsorption takes place in seconds, but it takes sev- VA (r) = (1)
r
eral minutes when the plasma concentrations has phys-
iological values [9]. The more hydrophobic the surface, where RA is the radius of Alb and r the protein-protein
the stronger is the protein adsorption, eventually induc- distance (Fig. 1a). Attraction among protein is not in-
ing irreversible adsorption [10]. Here we study the ef- cluded at this level of description, as it is small compared
fect by Molecular Dynamics (MD) of a model protein to protein-surface interaction [11]. Despite this rough
solution in contact with glass. We consider the three zero-order approximation, our results support a posteri-
most abundant proteins in human blood: Albumin (Alb), ori this assumption within the approximations of our ap-
Immunoglobulin-γ (IgG) and Fibrinogen (Fib), for which proach. IgG and Fib in their folded conformation have
 2

non-spherical shapes. In particular, the IgG structure 6.08 ǫA . We express all the results in terms of the Alb
Mi
resembles a γ and the Fib resembles an elongated ellip- units: Mi∗ ≡ M A
for the mass, Ri∗ ≡ RRAi for lengths
soid. They both can be approximated by ellipsoids with  2 1
RA MA 2
two principal axes of rotation along which they interact and ǫ∗i ≡ ǫǫAi for the energies, t∗ = ǫA for the
with each other (Fig. 1b-d). This is encoded through the time, T ∗ = kǫBAT for the temperature, and ∆E/ǫA for
protein-protein potential, within the same protein family, the energy changes of the buffer. In experiments ∆E is
 24 controlled by adding sodium azide, or other depletion-
2Ri 3 energy chemical agents, to the protein solution [12]. In
Vi (r) = + (2)
r 1 + exp (30 (r − 2RS,i ) /2RA ) the following we drop the ∗ for sake of clarity.
We perform MD simulations at constant T , constant
where the index i = I, F stands for IgG (I) and Fib (F ), volume V and constant number of proteins Ni , in a par-
with a hard-core radius Ri and a soft-core radius RS,i allelepiped with two square faces and four rectangular
[13]. Interaction between pairs of proteins of different faces. A square face is occupied by the attractive sur-
families are given by Eq.(2) with parameters Ri and RS,i face, the other by a wall interacting with the proteins
equal to the averages of the corresponding parameters for through the repulsive part of the V24,12 potential. We
each family, and with RA = RS,A for Alb. apply periodic boundary conditions (pbc) along the four
The protein-surface interaction is given by rectangular faces. The volume concentrations of proteins
   is taken to match the average concentrations of the hu-
σi 24  σi 12
V24,12 (r) = 4ǫi − (3) man plasma, with cA = 4.25 g/dl, cI = 1.25 g/dl and
r r cF = 0.325 g/dl, at XP = 100% plasma concentration
where ǫi is the attractive energy between the surface and in blood. When a protein is adsorbed on (released by)
a protein of the family i, and σi = Ri /21/6 , with i = the surface, we keep its bulk concentrations constant by
A, I, F , is the maximum approach distance between each inserting (deleting) a protein of the same family in a
protein and the surface. randomly-chosen empty (occupied) space of the box.
For each family of proteins i, we set the soft-core ra- Experiments are usually carried out for highly diluted
dius RS,i = RH the hydrodynamic radius, determined plasma, at concentration as small as XP = 0.1%, to slow
experimentally from the diffusion coefficient D through down the adsorption rate to minutes or hours, allowing
the Einstein-Stokes equation D = 6πηR kB T
, where η is the precise measurements. However, such low rates would
H
viscosity of the medium, under the assumption that the decrease the statistics of our MD simulations. We, there-
proteins can be approximated by a sphere. The hard- fore, perform our simulations in conditions that are closer
core radiuses Ri are set by imposing for each protein to those of practical interest, with XP as high as 100%,
that the experimental surface concentration corresponds 50% and 25%, by considering different sizes of the sim-
to the close packing configuration [11]. These conditions ulation box while keeping constant the initial number of
give RA = 3.55 nm, RI = 4.9 nm, RF = 6.58 nm proteins, their relative proportions, and the size of the
RS,I = 5.51 nm and RS,F = 11 nm. Protein masses adsorption surface.
MA = 67 KDa, MI = 150 KDa, MF = 340 KDa, neces- For each XP we average the results over fourteen in-
sary to determine the time scales, are known from exper- dependent runs, starting from independent initial config-
iments [10]. Protein-surface attraction energy ǫi can be urations that have been equilibrated by applying pbc in
calculated from the adsorption rate constants [11]. These any direction. We find that protein surface concentra-
rates are proportional to the probability for a protein i tions CiS are non-monotonic in time (Fig. 2). For any
to attach to the nearby surface considered XP , Alb is the first protein that reaches the
S
surface, inducing an increase of CA . When the second
fastest and second most affine protein, IgG, diffuses to
 
ǫi
Pi ∝ exp . (4) the surface, it displaces Alb, leading to a decrease of CAS
kB T
S
and an increase of CI . Finally Fib, which is the slowest
However, the ǫi in physical units are not known a priori. and most affine protein to the surface, takes over decreas-
Hence, we consider the  relative
 probabilities for differ- ing CIS and increasing CFS . Each CiS saturates toward
ǫ −ǫ an equilibrium value at long times, while the total sur-
ent proteins PPji ∝ exp kiB Tj , from which is possible to
determine the values of the different energies as face concentration of proteins is saturated at early times.
This behavior qualitatively reproduces the Vroman ef-
fect, apart from the behavior of Fib that here is mono-
 
ǫj kB T PA
=1− ln (5) tonic, while in experiments has a maximum due to the
ǫA ǫA Pj
competitive adsorption with heavier and more surface-
adopting ǫA for Alb as the energy units. We set ǫA by affine plasma proteins not included in our model [14].
comparing our simulations results with experiments at The only effect of reducing XP is a slowing down in the
ambient temperature, and get ǫI = 2.79 ǫA and ǫF = dynamics of the process, as observed in experiments [15].
 3

(a)
0.25 0.25 (a) XP=100%
Surface Concentration Ci (µg/cm )

Surface Concentration Ci (µg/cm )

0.2 XP=100% 0.2
2

2
0.15 0.15
0.1 0.1
S

S
0.05 0.05
0 0

0.25
(b) 0.25
(b)
XP=25%
0.2 XP=25% 0.2
0.15 0.15
0.1 0.1
0.05 0.05
0 0
0 0.005 0.01 0.015 0 0.005 0.01 0.015
Time t (s) Time t (s)
FIG. 2. Simulations at T = 300 K and (a) XP = 100% and (b) FIG. 3. Surface concentrations CiS as function of time for
S
XP = 25% show that surface concentration CA of Alb (#), T = 120 K at (a) XP = 100% and (b) XP = 25%. At
CIS of IgG (2) and CFS of Fib (∆) are non-monotonic with long time, CIS > CFS , with an inversion with respect to the
time, while their sum is (∇). At XP = 50% (not shown) we standard conditions in Fig. 2 where CFS > CIS . We find the
find the same behavior with time-scales intermediate between same qualitative behavior at XP = 50%, not shown. Errors
those in (a) and (b). Bulk concentrations are as indicated in and symbols are as in Fig. 2.
the text. Errors are smaller than symbol sizes. Surface Concentration CiS (µg/cm )
0.2
2

Albumin
T=120K T=300K Inmuno γ
The model allows us to understand the sequence of 0.15 Fibrinogen

surface occupation as a consequence of the competition

between the smaller, but less affine, proteins with the 0.1
more affine, but bigger, proteins. For example, we test
that by increasing the Alb affinity, or artificially setting 0.05
to the same value all the diffusion constants, the effect
disappears. Therefore, affinity and hydrodynamic radius 0
0 0.01 t 0.02 0.03 0.04 0.05 0.06
are the relevant protein parameters for the effect. 0
Next, we study how energy depletion of the protein so- Time t (s)
lution affects the sequence of adsorption. Here, for sake of
simplicity, we decrease T , reducing the kinetic energy of FIG. 4. The surface concentrations CiS , as a function of time
for XP = 100%, is drastically affected when the system un-
the solution, but neglecting possible effects of the protein
dergoes a sudden change from en energy-depleted condition
stability. We find (i) that, although the surface affinity to a normal condition. The vertical dashed line marks the
of Fib is stronger than that for IgG, the latter becomes time t0 of the change. We control the energy of the solution
the dominant protein adsorbed on the surface for long by changing the external parameter T from T = 120 K to
time scales; (ii) that, by changing XP , the time scale T = 300 K. Errors and symbols are as in Fig. 2.
of the process becomes longer, but the inversion of the
protein concentration is always present (Fig. 3). Hence,
the energy depletion leads to an inversion of the Vroman tion of this layer, determining different biomimetics sur-
effect. face properties. This situation could occur, for example,
By comparing the results at different energies, kB T , when a medical device is manipulated in a bioenviron-
and same XP (Fig. 2-3), we observe only a week energy- ment whose composition is externally controlled during a
dependence of the times at which each CiS reaches its surgery [18]. In particular, we study the case in which the
maximum. Hence, the time-scales of the process are system is first equilibrated under energy-depleted condi-
mainly controlled by the total plasma concentration XP . tions and subsequently undergoes a sudden change that
Once we have understood that the protein layer cov- reestablishes the normal conditions (Fig. 4).
ering the surface is controlled by the energy depletion of At short times the energy-depleted system evolves until
the system, it is interesting to ask if a sudden change the equilibrium concentrations are reached. Under these
of external conditions could induce a different composi- conditions, as discussed (Fig. 3), the dominant protein
 4

is IgG instead of Fib. At time t0 we switch to normal on a surface, in which the different families of proteins
conditions, forcing the system out of equilibrium. As a occupy sequentially the surface, replacing each other, un-
consequence, the system re-enters a transitory situation til an equilibrium situation is reached. By decreasing the
in which the concentrations CiS evolve until they reach total concentration of protein in the solution, keeping the
their new equilibrium values at long times. In the spe- relative concentrations fixed, the time scales of the pro-
cific case considered here, we observe a fast change in the cess increase and the maxima of surface concentration for
surface concentrations, with CFS of Fib overcoming CIS each family of proteins occur at longer times.
of IgG, being the first, under normal conditions, more We find that the protein surface concentrations at equi-
stable on the surface than the second. The final equi- librium depend on external control parameters. In par-
librium concentrations are reached at large times. We ticular, we find that energy depletion induces a drastic
S
observe also a sudden change in CA of Alb, between the change in the composition of the covering protein-layer,
two equilibrium concentrations characteristics of the two leading to an inversion of the Vroman effect. Our re-
S
values of the external parameters T . However, CA always sults show that the inversion can be used to quantify
equilibrates to a value that is smaller then CI and CFS ,
S
how strongly irreversible is the process of surface adsorp-
consistent with its long-time values in Fig. 2-3. By de- tion of the proteins, an information useful in studies of
creasing XP , we find the same qualitative behavior for a thromboembolic events [17]. Furthermore, these results
sudden energy-change, but with the transient regime ex- suggest the possibility of engineering the composition of
tending to longer times, consistent with Fig. 3. Hence, at the protein layer covering a surface in a controlled way,
experimental values of XP the switching behavior would a feature particular relevant in biomimetic applications.
occur on time scales that are comparable to those char- We thank C. Åberg, F. Baldelli Bombelli, and M. P.
acteristic of the Vroman effect. Monopoli for discussions. We acknowledge the support
We remark that our predictions about inverting the of EU FP7 grant NMP4-SL-2011-266737; PV and GF of
Vroman effect by changing the experimental control pa- Spanish MEC grant FIS2009-10210 co-financed FEDER.
rameters should hold only if the protein adsorption on
the surface is reversible. If the adsorption is, instead, ir-
reversible the change of external parameters should not
lead to a new composition of the protein layer. Indeed, ∗
[email protected], [email protected]
under many practical conditions of interest for blood
[1] T. Cedervall et al., Proc. Natl Acad. Sci. USA 104,
plasma, it would appear that the binding is indeed mostly 2050 (2007); A. E. Nel et al., Nat. Mater. 8, 543 (2009);
irreversible [2, 16]. Hence, the switching protocol pro- C. Rocker et al., Nat. Nanotechnol. 4, 577 (2009).
posed above represents a possible experimental way to [2] M. P. Monopoli et al., J. Am. Chem. Soc. 133 2525-
evaluate how strongly irreversible is the adsorption pro- 2534 (2011); A. Salvati et al., Nature Nanotechnol.
cess on a specific surface. doi:10.1038/nnano.2011.191 (2011).
For an irreversible adsorption process, our findings pre- [3] D. Walczyk et al., J. Am. Chem. Soc. 132, 5761 (2010);
Z. J. Deng et al., Nat. Nanotechnol. 6, 39 (2011).
dict that by appropriately controlling the parameters of [4] G. Oberdorster et al., J. Nanosci. Nanotechnol. 9, 4996
the protein solution, such as the amount of depleted en- (2009).
ergy, it is possible to engineering a specific biomimetic [5] P. Tengvall et al., Biomat. 19, 407 (1998).
covering of a surface. Due to the irreversibility, the pro- [6] L. Vroman, Nature 196, 476 (1962).
teins, once adsorbed, cannot desorb from the surface, [7] M. Holmberg and X. Hou, Langmuir 26, 938 (2009).
even if the external conditions are modified. Therefore, [8] D. R. Jackson et al., Langmuir 16, 5449 (2000).
it is feasible to cover a device surface with any desired [9] S.-Y. Jung et al., J. Am. Chem. Soc. 125, 12782 (2003).
[10] H. Noh and E. A. Vogler, Biomat. 28,405 (2007).
protein composition, targeted to a specific biomimetic [11] C. F. Lu et al., J. Coll. Int. Sci. 168, 152 (1994).
property, by selecting an appropriate initial condition. [12] K. Shapero et al., Mol. BioSyst. 7, 371 (2011).
Subsequently, the device could be used under physiolog- [13] G. Franzese, J. Mol. Liq. 136 267 (2007); A. B.
ical conditions with no further changes of the protein de Oliveira et al., J. Chem. Phys. 128 064901 (2008);
cover and its biomimetic properties. P. Vilaseca and G. Franzese ibid. 133, 084507 (2010); J.
In conclusion, we study, by MD simulations of a coarse- Non-Cryst. Sol. 357, 419 (2011).
[14] P. Turbill et al., Biomat. 17,1279 (1996).
grained model, the Vroman effect for three kinds of pro-
[15] J. L. Brash and P. ten Hove, Thromb. Haemost. 51, 326
teins: Alb, IgG and Fib. We show that the effect is the (1984).
consequence of the differences among the proteins proper- [16] M. Lundqvist et al. ACS Nano, 5, 7503 (2011).
ties (mass, size and affinity) which regulate their diffusive [17] S. Uniyal and J. L. Brash, Thromb. Haemost. 47,285
behavior and their interaction with the surface. These (1982).
differences lead to a process of competitive adsorption [18] A. N. Cloud et al., Surf. Coat. Tech. 203, 913 (2008).