Protein adsorption on solid surfaces

Abstract

The research field of protein adsorption on surfaces appears to be as popular as ever. In the past year, several hundred published papers tackled problems ranging from fundamental aspects of protein surface interactions to applied problems of surface blood compatibility and protein surface immobilization. Although some parts of the protein adsorption process, such as kinetics and equilibrium interactions, can be accurately predicted, other aspects, such as the extent and the rate of protein conformational change, are still somewhat uncertain. The whole field is ripe for a comprehensive theory on protein adsorption.

Introduction

Proteins accumulate at interfaces, a property that can be both a practical asset and a problem. New biotechnological methods of protein production depend on protein interracial properties in downstream protein purification and separation. Furthermore, the adsorption of proteins at solid/liquid interfaces has enabled the development of diverse biomedical applications, such as biosensors, immunological tests, and drug-delivery schemes. In the biomaterial field, protein adsorption is much less desirable because it can elicit adverse host responses such as blood coagulation and complement activation. On the other hand, cell adhesion to surfaces depends on the availability of specific protein-binding sites.

The desire to control, predict, and manipulate protein adsorption has been the main driving force for past research in this field. This has led to a wealth of current research directed at gaining a better understanding of the behavior of proteins at interfaces, which is reflected in the large number (>300) of papers on the subject published during the past year. The advances in the field are summarized in a recent monograph [1••].

The present short review is not a comprehensive overview of this fertile and diverse field, rather, it reflects our bias toward certain areas of biomedical applications. We focus on the most recent research efforts that have led to the use of new experimental techniques and theoretical approaches for studying protein adsorption.

Proteins are surface-active macromolecules

It is important to recognize that most proteins are large amphiphatic molecules. This characteristic makes them intrinsically surface-active molecules; thus, the problem is not how to adsorb them to interfaces, but how to control their interfacial adsorption. A thermodynamic inventory of the various interactions that contribute to protein adsorption has been made by Haynes and Norde [2••]. The origins of these interactions are found in intermolecular forces, such as Coulombic forces, van der Waals forces, Lewis acid-base forces, and more entropically based effects such as hydrophobic interactions, conformational entropy and restricted mobilities. In addition, the adsorption process depends on intramolecular forces within the protein molecules that might lead to an alteration of protein conformation. As a result, one sometimes finds a large difference between protein adsorption and desorption behavior, leading to an apparent irreversibility of the adsorption process. It has been difficult, however, to determine exactly the extent of these conformational changes and even more difficult to predict them. One way of dealing with some of the protein adsorption paradoxes (we borrow this term from [3]) is to study the kinetics of the adsorption process, rather than to wait for the equilibrium of a protein/surface system. But, the kinetics approach often requires a model to which the experimental results can be successfully fitted, and any model needs to be proved in some independent way. The vicious circle of protein adsorption continues, echoing the words of Leo Vroman, “Many a protein changes its song after it has been caught, and it turns out to be a mockingbird” [4].

If one component is missing in the protein adsorption field, it is a rigorous protein adsorption theory. The protein adsorption field has been traditionally experimental and empirical. In addition, the adsorption of globular proteins has dominated research in the field. Times are changing: new protein molecular structures are being determined at a steady pace of several hundred per year [5•] and new rules governing protein (un)folding pathways are being discovered and re-defined [6••]. Knowing that so many interprotein homologies exist in nature, one wonders why a comprehensive protein adsorption theory isn’t already here.

New experimental approaches for protein adsorption studies

Everybody agrees that well defined and characterized model surfaces are needed in order to successfully study protein adsorption. The importance of such surfaces has been appreciated even more after the introduction of new high-resolution techniques, such as atomic force microscopy (AFM) [7], for studying surfaces. The surface ideal is now approached using the concept of self-assembly [8,9••]. A simple immersion of a surface, which is coated with a thin layer of gold, into a millimolar ethanolic solution of alkanethiol for few hours can result in a uniform monolayer [10]. The chemical design of terminal groups opens the possibility of tailoring these monolayers for different uses, such as protein adsorption or cell adhesion studies [11].

Surface chemistry can be used to manipulate protein–surface interaction by controlling the surface density of chemical groups or ligands. A one-dimensional surface density gradient of octadecyldimethylsilyl (ODS) chains attached to the flat silica plate is an example of the use of ‘gradient surfaces’ in protein adsorption. Using the gradient surface approach, adsorption of human serum albumin (HSA) [12] and other plasma proteins [13•] has been studied as a function of surface density of ODS chains. Other gradient surfaces used in protein adsorption include positive-negative charge gradients [14], and surface density gradients of end-attached poly(ethyleneoxide) [15].

The capability of AFM to resolve fine surface details in situ has led to its application in protein adsorption studies. The technique has been used to map the spatial distribution of proteins such as adsorbed and immobilized IgG [16]. A disadvantage of AFM is that the so-called contact imaging mode, often found in the older AFM instruments, is too destructive for the adsorbed protein layer; in this mode, the scanning forces imposed to the adsorbed protein lead to their re-distribution, typically outside, or at the edges of, the scanning area. One way to get around this problem in in situ protein kinetic experiments is to scan each surface region only once and at very light forces [17]. The problem of inadvertent manipulation of the adsorbed layer can be minimized using the newly introduced ‘tapping’ imaging AFM mode, in which the probe intermittently touches the adsorbed protein layer during the imaging. The ‘tapping’ mode is useful not only in imaging individual protein molecules [18], but also in monitoring the dimensional changes occurring in a single adsorbed lyzozyme molecule during its enzymatic cleavage of polysaccharide substrates [19••].

The AFM technique has also made it possible to measure the strength of individual ligand–receptor forces [20•]; biotinylated albumin adsorbed onto the AFM probe can be used as a means of immobilization for biotin ligands. In principle, this technique can lead to AFM-based molecular recognition devices. Attempts to use a monoclonal antibody as an element for recognizing an immobilized ligand were hampered, however, by non-specific IgG adsorption onto both the liganded probe and the cantilever-bead system used in the AFM analysis [21]. Measurements of specific and non-specific protein–protein interactions have also been performed using the surface force apparatus (SFA) [22,23••]. The advantages of SFA, compared with the AFM technique, lie in its ability to accurately determine the distance between two interacting surfaces and in the ‘tunability’ of the stiffness of the double-leafed SFA spring.

A relatively new method for obtaining more detailed information on some physical parameters of the adsorbed protein layer is neutron reflectivity. This approach provides information on both the thickness and chemical composition of adsorbed layers. It has been used to study the effect of the amphiphilic character of, for example, β-casein on self-assembled monolayers formed by octadecyltrichlorosilane (OTS) [9••] and β-casein adsorbed at the air/water interface [24].

Some spectroscopic techniques (e.g. circular dichroism [CD] [25•,26] and intrinsic fluorescence measurements [27]), which are conventionally used to study proteins in solution, have also been applied to the study of protein adsorbed on the surface of nanosized silica particles. Although the choice of nanoparticle surfaces is limited, typical results on silica indicate that some conformational changes take place upon adsorption and that the extent of conformational change is related to the structural stability of the protein.

Modeling protein surface interactions

A comprehensive, rigorous protein adsorption theory is still pending. In the meantime, some parts of the adsorption process are becoming more and more accurately modeled. For example, significant advances have been made in predicting the role of electrostatic interactions in protein adsorption, which is particularly important in ion-exchange chromatography [28]. Furthermore, a linearized finite-difference Poisson–Boltzmann equation algorithm, implemented in popular molecular graphics software packages, can easily be employed in the computation of adsorption electrostatic effects. For example, computation as analysis of the electrostatic interactions of cytochrome b₅ with an anion-exchange surface showed an energetically preferred orientation of the molecule [29]. The importance of electrostatic interactions in adsorption is demonstrated by the profound effects of a change of a single charge in cytochrome b₅ on the adsorption kinetics of the protein on a variety of different surfaces [30].

A somewhat more comprehensive approach in electrostatic modeling was employed by Noinville et al. [31•], who used an AMBER force-field to compute interaction energies for all atoms in a protein/polymer surface system composed of α-lactalbumin (or lyzozyme) and poly(vinylimidazole). Similar to the case of cytochrome b₅, the computations predicted a preferred orientation of α-lactalbumin and lyzozyme at the interface. A classic titration study of the adsorption of the same two proteins on the silver iodide/water interface showed that protein carboxyl groups in particular are affected by adsorption, and it was assumed that these groups are located close to the surface [32]. It would be interesting to see if the computations correlate with experimental results.

The modeling of experimental results from protein adsorption studies often requires the adaptation of different adsorption isotherm models. For example, Johnson and Arnold [33] found that the Temkin isotherm could describe protein adsorption to metal-chelating ligands [33]. A somewhat different model of continuous energetic heterogeneity for a protein/surface interaction has been proposed to explain the logarithmic kinetics of IgG and HSA adsorption on a quartz surface and the Freundlich character of HSA adsorption [34]. The random sequential adsorption model has been used to predict both the partial irreversibility of protein adsorption and the clustering of proteins at surfaces [35] and to explain antibody binding to antigen-coated surfaces [36]. On other hand, the HSA and low-density lipoprotein adsorption kinetics onto the ODS-silica gradient surfaces can be accurately modeled using simple Langmuirian kinetics [12,13•]. It is also interesting that the low-density lipoprotein adsorption kinetics in the ODS-silica gradient region can be simply predicted from a weighted sum of two Langmuirian kinetic processes occurring at the ODS-binding sites and the silica-binding sites [13•]. Besides modeling basic interactions and equilibrium situations, the description of the diffusive-convective transport of proteins from a laminar flowing solution to the sorbent surface has also been improved [37].

Interfacial protein orientation, conformation and dynamics

To exploit the biological function of an adsorbed protein molecule, one requires some control of the orientation of the protein at the interface. Recently, the determination of protein orientation has been shown possible in the case of a single fluorophore–protein molecule, a porphyrin–cytochrome c complex, using polarized total internal reflection fluorescence spectroscopy [38•]. Although the method can be used to determine the average orientation of adsorbed protein, its accuracy was very sensitive to surface roughness and the scattering of the light from the interface.

Many experimental studies of protein adsorption find that some conformational changes in protein molecules have taken place upon protein surface adsorption [39–41]. Two rarely answered questions in these measurements is how large these changes are and where in the protein molecule they have occurred? This is particularly puzzling in the case of CD measurements, which are often interpreted as percentage change of a given secondary structure [25•,26,42]. For example, what is the physical picture of a 12% decrease of α-helix structure that is recorded by CD on hundreds of thousands of protein molecules in the observation volume? Does it mean that all protein molecules have lost the same degree of their α-helix structure or are some molecules completely denatured and others retain their native conformation? Clearly, until single-molecule spectroscopy is developed and applied to adsorbed proteins, one always has to keep in mind that each technique performs some kind of averaging over the observed protein population.

What about adsorbed protein dynamics? Proteins are known to undergo a continuous conformational fluctuation, which enables them to perform their biological functions. There are very few reports on intramolecular dynamics of adsorbed protein because of the limited choice of techniques that can measure fast processes (i.e. from a few picoseconds to a few nanoseconds) at interfaces. One possibility is to use time-resolved fluorescence spectroscopy that utilizes an evanescent surface wave to excite the adsorbed species. This method has been used to observe the nanosecond dynamics of the local environment of cysteine at residue 34 of bovine serum albumin molecules [43].

Selected applications: biosensors, biomaterials, plasma proteins, and cell adhesion

Biosensors, along with immunoassays and immunochromatography systems, are often based on the ability of IgG to bind a large variety of molecules in a highly specific way. These applications require a control of the above mentioned orientation and conformation of IgG in order to retain its binding capacity in the adsorbed state. The layer thickness of adsorbed IgG determined using ellipsometry indicates that under most adsorption conditions, an end-on orientation of IgG is obtained [44]. With respect to the conformation, it has been shown that the Fab′ portion, which contains the binding sites, exhibits a relatively high structural stability on adsorption compared with the Fc portion [45•]. This observed difference in structural stability also explains why the Fc portion is more readily adsorbed than the Fab portion under similar adsorption conditions, which, in turn, might affect the orientation of adsorbed IgG [46]. Another problem with these applications is related to the non-specific interactions of the protein with the sorbent surface, which can be misinterpreted as a specifically detected analyte in biosensing devices. Several studies have focused on protein immobilization and suppression of non-specific protein binding [47–49] and on the deposition of proteins to electrodes [50,51], In two very interesting studies, a novel way has been developed for preparing multicomponent protein [52•], and macromolecular films [53•]. The approach is based on the layer-by-layer assembly of alternate polyelectrolyte/protein components. The multiprotein film is prepared using charged proteins with linear polyelectrolyte molecules acting as a glue or filler. Such films are expected to carry a coupled enzymatic reaction resulting in the detection of an analyte by coupling to an electrode.

Biomaterial surfaces are often exposed to concentrated solutions of proteins. The adsorption of three plasma proteins, namely human serum IgG, albumin (e.g. HSA) and fibrinogen, has been studied in more detail than the adsorption of any other plasma proteins. This is perhaps justified by the relatively large concentration of these three proteins in plasma. A dynamic protein exchange is typically observed when two or more proteins are allowed to adsorb simultaneously to a surface. The exchange also occurs between proteins at the interface and proteins in solution [54,55]. For plasma proteins, the exchange process, which is usually governed by mass transport and the respective affinity of a given protein for the surface, is often referred to as the ‘Vroman’ effect. In another report, the adsorption of three proteins on a variety of phospholipid surfaces has been studied [56]. Plasma protein dynamic exchange has also been observed at wettability gradient surfaces [57]. In this case, albumin has an inhibitory effect on IgG adsorption on hydrophobic surfaces, but enhances IgG deposition on hydrophilic surfaces. One way of making biomaterial surfaces biocompatible is to make them resistant to protein adsorption. Poly(ethyleneoxide) [15,58], modified dextrans [59] and other hydrophilic polymer coatings [60,61] have been investigated for this purpose. A different biocompatibility hypothesis predicts that materials, such as low-temperature isotrophic carbon, that preferentially and strongly adsorb albumin will be biocompatible [62].

Control over the adhesion of cells to biomaterials often means control over the composition of the surface protein layer onto which the cells attach themselves [11]. In the case of biomaterials in contact with blood, the concentration of fibrinogen in the adsorbed plasma protein layer determines the extent of platelet adhesion [63,64]. The density of adsorbed proteins controls the adhesion of bacteria to implant surfaces exposed to circulating blood [65]. A new simple method for inducing differential Staphylococcus aureus adhesion is to use the so-called ‘lens-on-the-surface’ method to control the Vroman effect on protein adsorption [66].

Outlook

The field of protein adsorption is as exciting as ever. It is becoming increasingly appreciated that the surfaces commonly used in protein adsorption experiments are heterogeneous, dynamic and often contaminated. The traditional empirical ‘rules of thumb’ [67] are slowly being replaced with more accurate predictions based on protein molecular properties and the availability of well defined surfaces. Further advances in the field would be greatly assisted by the availability of a comprehensive adsorption theory that could be used to check and evaluate any new experimental data.

Abbreviations

AFM

atomic force microscopy

CD

circular dichroism

HSA

human serum albumin

ODS

octadecyldimethylsilyl

OTS

octadecyltrichlorosilane

SFA

surface force apparatus