Introduction

Over the past few years, numerous studies have aimed at understanding which driving forces govern protein adsorption at liquid-solid interfaces (Norde 1998, 2000). Spontaneous adsorption of protein molecules onto a solid sorbent will occur if the Gibbs energy of the system decreases. The enthalpic energy results from an electrostatic mechanism and H-bonding interactions, while the entropic contribution arises from the change of the initial ordered structure of the adsorbed protein as well

Sylvie Noinville, Madeleine Revault: Laboratoire de Dynamique, Interactions et Réactivité, CNRS-Université Pierre et Marie Curie, UMR 7075, 2 rue Henri Dunant, 94320 Thiais, France, E-mail: [email protected]

Principles and Practice Proteins at Solid-Liquid Interfaces Philippe Déjardin (Ed.) © Springer-Verlag Berlin Heidelberg 2006

as the rearrangement of water molecules and counterions at the interface. The protein folding is well known to be a highly cooperative process in solution. The denaturation of the globular proteins corresponds to a Gibbs energy in the 20-100 kJ mol-1 range, which is equivalent to the energy required for the disruption of 1-8 hydrogen bonds. Cooperative interactions do not involve the entire protein molecule, so that some partially folded conformations exist in solution as a result of local rather than global unfolding (Radford et al. 1992). When a denaturation process occurs in solution, the conformational entropy is offset by the enthalpy gain and the entropy loss due to the rearrangement of water molecules around the de novo external amino-acid distribution of the protein. When adsorbed onto a solid support, one external side of the protein molecules previously surrounded by water molecules is then in contact with the sorbent surface. Consequently, the mechanism of surface-induced conformational changes may be quite different from that causing protein unfolding in solution induced by denaturant addition or by heating. Recently, a high rate of partial unfolding of a-lactalbumin adsorbed on polystyrene (around 74 s-1) was determined by nuclear magnetic resonance (NMR) experiments, compared to the lower rate of pH-induced global unfolding (Engel et al. 2004). Circular dichroism (CD) studies have shown the difference between the heat-induced unfolding of immunoglobulin and the surface-induced process when the protein is adsorbed on Teflon particles (Vermeer etal. 1998).

One common feature of the structure of the globular proteins is that the polar residues are located on the external surface and that most of the apolar amino acids are buried in the protein core. However, a significant fraction of the external surface of the protein molecule is composed of hydrophobic patches, which account for up to 40-50% (Branden and Tooze 1991; Dill et al. 1995; Richards 1977). If the external surface of the protein is polar as for the sorbent phase, some hydration water molecules are retained to solvate the protein charged residues and the surface charges. Ifthe protein surface is rather apolar and the sorbent phase polar, or vice-versa, a dehydration of the protein-solid interface could occur. In the case of a hydrophobic character for both the protein surface and the solid support, structural changes of the proteins as well as variations in solvation states modify the entropy of the overall system (Boulkanz et al. 1997; Dorsey and Dill 1989; Gilpin 1993; Lu et al. 1998; Norde 1998). In some cases, the adsorption on hydrophobic supports induces a conformational change large enough to cause an average increase of the protein hydration (Boulkanz et al. 1995; McNay and Fernandez 1999). For both hydrophobic and hy-drophilic supports, protein solvation and structural changes also depend on the pH and the ionic strength of the liquid phase, as these parameters indirectly influence the entropic forces (Baron et al. 1999; Giacomelli et al.

1999b; Haynes and Norde 1995;Kondo etal. 1991; Quiquampoix et al. 1993; Servagent-Noinville et al. 2000; Su et al. 1998b).

Of course, the adsorption process depends strongly on the chemical and electrical properties of the sorbent phase. Many studies concerning protein adsorption onto a polar charged support show that the major driving force is of coulombic nature (Gill et al. 1994; Lesins and Ruckenstein 1989; Quiquampoix et al. 1995; Servagent-Noinville et al. 2000; Su et al. 1998b). The design of electrostatically neutral surfaces is sought for their ability to reduce protein adsorption. In that case, the different hydration properties lead to repulsive hydration forces at close approach responsible for the inhibited adsorption (Herrwerth et al. 2003; Jeon et al. 1991). On hydrophobic supports, adsorption is driven by the reduction of interfacial energetics concomitant with the replacement of water molecules by the adsorbed proteins (Vogler 1998).

Besides these interactions between a single protein molecule and the solid phase, protein-protein interactions of either electrostatic or hydrophobic origin may also influence the accumulation of proteins at the liquid-solid interface. For both hydrophobic and hydrophilic supports, all of these parameters govern the specific orientation and the structure of the protein molecules inside the adsorbed layer (Malmsten 1998; Wahlgren et al. 1998). One approach with which to gain insight into the impact of hydrophobic or electrostatic interactions on the adsorption mechanism is to study protein adsorption on molecularly well-defined surfaces such as self-assembled monolayers (SAMs) on gold or silicon (Mrksich et al. 1995; Silin et al. 1997). Surface plasmon resonance (SPR) studies have shown that the propensity of SAMs of gradient wettabilities to adsorb proteins is related to the interfacial free energy of these surfaces, despite important exceptions (Sigal et al. 1998). Non-charged hydrophilic surfaces such as OH-terminated SAMs are generally correlated with low protein retention (Silin et al. 1997; Tengvall et al. 1998). Although ethylene-glycol-terminated SAMs are known to be highly resistant to protein adsorption, these supports are characterised by intermediate wettabilities (Herrwerth et al. 2003; Ostuni et al. 2001). The correlation between a macroscopic surface property such as wettability and protein adsorption seems to be restrictive to the choice of model proteins under investigation. However, a means with which to study the influence of one single isolated hydrophobic end-group on protein adsorption is provided by mixing two types of end-groups, one of which is a known repelling moiety, to build well-controlled hydrophobic patches inside the SAMs (Ostuni et al. 2003).

In general, the adsorption of proteins from aqueous solutions onto a solid support occurs via three major steps: (1) transport of protein molecules from the bulk to the interface, (2) contact of the protein molecules to the solid surface and (3) conformational changes of the protein molecules during adsorption. Because the structural changes of the protein are ki-netically controlled, accurate knowledge of the adsorption kinetics is of great importance. Most of the general patterns governing the kinetics of protein adsorption have been described in previous reviews (Brash 1996; Malmsten 1998; Ramsden 1998).

The structural rearrangement caused by the direct contact with the sorbent phase enables the existence of different conformers, which result from the way local unfolding could propagate along the entire molecule, depending on the intrinsic stability of the protein. In the case of strong protein-sorbent interactions, non-equilibrium states could be retained at the liquid-solid interface. Hence, protein adsorption could appear like an irreversible process, depending on different time scales. Moreover, structural changes of the protein involving more than 10% of the polypeptide backbone are generally associated with a spreading phenomenon, allowing a greater number of contacts with the sorbent phase (Jeon et al. 1994; Noinville et al. 2002; Su et al. 1998b). Hence, the desorption of even partially unfolded proteins becomes difficult. Besides, by spreading over the sorbent phase, the protein molecules delay or prevent the attachment of additional protein molecules. The adsorption rate and the conformational changes could depend on the surface coverage insofar as the adsorption takes place at a more and more modified interface. As a consequence, the layer of adsorbed proteins could be composed of a heterogeneous distribution of adsorbed conformers (Pitt and Cooper 1986; Zoungrana et al. 1997). A general trend in the adsorption of globular proteins is to distinguish between two classes associated with distinct adsorption behaviour in relation to their structural stabilities (Arai and Norde 1990; Kondo et al. 1991; Norde 1998). On one hand, the so-called "soft" proteins with low structural stability depict protein molecules with a high tendency to denature upon adsorption, even if the sorbent phase is weakly attractive (Table 1). On the other hand, "hard" proteins tend to adsorb less strongly on hydrophilic supports unless attractive electrostatic interactions prevail (Table 2). These hard proteins generally undergo minor conformational changes except when adsorbed onto hydrophobic supports. Recently, dynamic Monte Carlo simulations have corroborated the relationship between the native-state structural stability of the protein and its adsorption behaviour by computing the changes in the system entropy (Liu and Haynes 2005). One strategy to change the protein structural stability is to perform site-directed mutagenesis. The substitution of one single residue in the primary sequence of the protein provides mutants of different intrinsic stability and adsorption behaviour (Brash 1996; Malmsten et al. 2003). McGuire et al. (1995) studied protein adsorption of the wild type of bacteriophage T4 lysozyme and various mutants on flat silica surfaces. The variant with the lower structural stability displays the slowest adsorption kinetics, while it undergoes the largest

Table 1. Properties of "soft" proteins. HAS Human serum albumin, BSA bovine serum albumin, p~Lg ¿8-lactoglobulin, a-La a-lactalbumin

Property

HSA

BSA

ß-Lg

a-La

Molar mass (kD)

67

65

18.4

14

£Ni

586

582

162

123

Dimension" (Ä3)

30X80X80

30X80X80

36X36X36

37X32X25

Hydrophobicityb (J g-1)

-3.8

-5.8

1010Compressibilityc (m2

N-1) 0.712

1.05

0.845

0.827

Isoelectric point

4.7

4.8

5.2

4.3

Secondary structure3

%a-helix

67

67

10

29

%^-sheet

0

0

43

8

Reference

(Carter and

(Carter and

(Qin et al.

(Pike et al.

Ho 1994)

Ho 1994)

1998)

1996)

PDB entryd

1BMO

3BLG

1HFX

aX-ray data if available or sequence homology bPrivalov 1979

cGekko and Hasegawa 1986

dBerman et al. 2000

aX-ray data if available or sequence homology bPrivalov 1979

cGekko and Hasegawa 1986

dBerman et al. 2000

Table 2. Properties of "hard" proteins. Lys Lysozyme, Chym chymotrypsin, Cyt c cytochrome c, RNase ribonuclease, PDB protein data bank

Property

Lys

Chym

Lipase

Cyt c

RNase

Molar mass (kD)

14

25

22

12

14

Dimension11 (Ä3)

45X30X30

30X80X80

37X32X25

45X35X30

38X28X22

£Ni

129

245

269

104

124

Hydrophobicityb (J g-1)

-7.6

-8.7

1010 compressibilityc

0.467

0.415

0.339 (e)

0.112

(m2N-1)

Isoelectric point

11

8

4.4

10.7

9.4

Secondary structurea

%a-helix

29

11

33

39

11.5

%^-sheet

6

51

26

0

33

Reference

(Wilson

(Birktoft and

(Lawson

(Banci

(Leonidas

et al. 1992)

Blow 1972)

etal. 1994)

etal. 1997)

etal. 1997)

PDB entryd

1HEL

2CHA

1TIB

1AKK

1AFU

aX-ray data if available or sequence homology bPrivalov 1979

cGekko and Hasegawa 1986

dBerman et al. 2000

eChalikian et al. 1995

aX-ray data if available or sequence homology bPrivalov 1979

cGekko and Hasegawa 1986

dBerman et al. 2000

eChalikian et al. 1995

unfolding of a-helices, as shown by CD studies on silica nanoparticles (Billsten et al. 1998). The structural stability of the protein is a key factor in determining the rate of conformational changes for different mutants of human carbonic anhydrase I and II (Lundqvist et al. 2004; Malmsten et al. 2003). Recently, the rate of the surface-induced conformational change of the human carbonic anhydrase variants is indeed affected by the stability of the variant, but the average conformation of the final adsorbed state is not (Karlsson et al. 2005). Wild-type recombinant interferon y and its Asp25-mutant have different adsorption behaviours, as studied on hydrophobic supports by Fourier transform infrared spectroscopy (FTIR) analysis (deCollongue-Poyet et al. 1996). The amount of the mutant adsorbed is lower than that of the wild-type protein. The contact with the hydrophobic support entails a different increase in the amount of polar helices and bend domains in mutant and wild types with respect to the conformation in solution. However, this mutation of one single residue does not change the protonation state of the interferon, but induces a local change in the packing of the a-helices. One drawback of site-directed genetic engineering or chemical modification used to modulate protein hydrophobicity or protein net charge is that it also alters the structure of the protein under study (van der Veen et al. 2004).

Due to the peculiar behaviour of protein molecules, the adsorption of protein to solid surfaces is the result of a large number of parallel and consecutive reaction steps, which render the study of the kinetics and thermodynamics of the protein-solid system difficult (Fang and Szleifer 2001). Models concerning the conformational changes of adsorbed proteins have been proposed for fitting experimental kinetics studies (Kridhasima et al. 1993; Pantazaki et al. 1998; Soderquist and Walton 1980). However, the main difficulty encountered in most theories of protein adsorption is to choose at which level to depict the protein-solid interface. As a starting point, the macroscopic description requires consideration of the globular protein as a rigid sphere with a net electrostatic charge on a plane surface (Roth and Lenhoff 1993). Some colloidal models account for conformational changes upon adsorption by considering the existence of different adsorbed con-formers such as a folded state and a fully denatured state of the adsorbed protein (Van Tassel et al. 1998). A more sophisticated model is to depict the map of the protein surface, the distribution of positive and negative charges and the surface polarity. This methodology yields better predictions of interactions between like-charged proteins and surfaces (Asthagiri and Lenhoff 1997). The more sophisticated approach is to depict the protein using its three-dimensional structure. Early simulation studies of protein adsorption based on its full atomistic description have reported calculated energetics as a function of the orientation and the separation distance of the protein towards the surface in the case of uniformly charged support

(Yoon and Lenhoff 1992) or a polymer surface represented atomistically (Noinville et al. 1995; Roush et al. 1994). Most recent studies are focussed not only on the interaction energy and adsorbed amount, but also on the orientation of adsorbed proteins or peptides (Noinville et al. 2003; Zhou et al. 2003). Monte Carlo simulations of adsorption of immunoglobulins onto charged surfaces lay emphasis on the dependence of the protein orientation versus surface charge density, as previously shown experimentally (Zhou et al. 2003). A preferred orientation for the antibodies on the charged surface is determined when electrostatic interactions prevail over the hydrophobic ones. In all of these studies, the aqueous medium is described as a continuous dielectric medium. A simulation study of a small peptide of five residues is reported with explicit water molecules, demonstrating their orientational structure at a peptide-non-polar solid interface (Bujnowski and Pitt 1998).

Was this article helpful?

0 0

Post a comment