Kinetics of the Homogeneous Exchange of Lysozyme Adsorbed on a

des Industries Chimiques de Strasbourg, 1 rue Blaise Pascal, BP 296F, 67008 Strasbourg Cédex, France ... Publication Date (Web): February 19, 199...
0 downloads 0 Views 190KB Size
Langmuir 1997, 13, 729-735

729

Kinetics of the Homogeneous Exchange of Lysozyme Adsorbed on a Titanium Oxide Surface A. Bentaleb,† V. Ball,† Y. Haı¨kel,† J. C. Voegel,*,† and P. Schaaf‡,§ INSERM U-424, Fe´ de´ ration de Recherches Odontologiques, 11 Rue Humman, 67000 Strasbourg, France, Institut Charles Sadron, CNRS-ULP, 6 rue Boussingault, 67083 Strasbourg Ce´ dex, France, Ecole Europe´ enne des Hautes Etudes des Industries Chimiques de Strasbourg, 1 rue Blaise Pascal, BP 296F, 67008 Strasbourg Ce´ dex, France Received May 28, 1996. In Final Form: November 26, 1996X The desorption and exchange processes of the protein lysozyme adsorbed on TiO2 particles in the presence of lysozyme in solution are investigated by means of a 125I radio-labeling technique. Lysozyme is a compact and highly positively charged protein, making this study complementary to previous work devoted to the understanding of the exchange mechanisms of adsorbed proteins on solid surfaces. It is found that at least three populations of adsorbed molecules exist on the surface: (i) a rapidly exchangeable and/or desorbable population (The rate constants associated with the exchange and rapid desorption processes are equal to 6.28 × 104 L‚mol-1‚h-1 and 2.7 × 10-4 s-1, respectively), (ii) a population which does not exchange but can only desorb from the surface over a time scale of the order of 4 h and whose rate constant is equal to 6.8 × 10-5 s-1, and, (iii) finally, an irreversibly adsorbed population. It is also demonstrated that the homogeneous exchange process can be modeled by a reaction of order one with respect to molecules in solution. Moreover, it is suggested that both the desorption and exchange processes are of order one with respect to the adsorbed lysozyme molecules.

Introduction Adsorption processes of macromolecules on solid surfaces are complex due to the fact that large molecules can interact with a solid surface through many points of contact and, further, can change their conformations in a large number of ways. These general features often lead to the irreversibility of the adsorption processes of macromolecules, as has been observed in the case of synthetic polymers1 and of various protein systems.2,3 Irreversibility means that, once adsorbed, a macromolecule does not desorb spontaneously in the presence of pure solvent. However, it has been shown that, despite being irreversibly adsorbed, in the sense of classical thermodynamics, macromolecules can still desorb from the surface in the presence of macromolecules in solution by a so-called exchange mechanism. This mechanism has been demonstrated for both synthetic polymers and proteins on surfaces. It is viewed as a gradual replacement of an adsorbed molecule by a molecule from the solution. After the pioneering work of Varoqui and Pefferkorn4 for the case of synthetic polymers adsorbed on aluminosilicated glass bead surfaces, in which it was shown that the exchange process is of order one with respect to the adsorbed proteins and with respect to the molecules in solution, a systematic study of this process for synthetic polymers was performed by Granick.5,6 * Author to whom correspondence should be addressed. † Fe ´ de´ration de Recherches Odontologiques. ‡ Institut Charles Sadron. § Ecole Europe ´ enne des Hautes Etudes des Industries Chimiques de Strasbourg. X Abstract published in Advance ACS Abstracts, January 15, 1997. (1) Pefferkorn, E.; Haouam, A.; Varoqui, R. Macromolecules 1989, 22, 2677. (2) Soderquist, M. E.; Walton, A. G. J. Colloid Interface Sci. 1980, 75, 386. (3) Norde, W.; Macritchie, F.; Nowicka, G.; Lyklema, J. J. Colloid Interface Sci. 1986, 111, 447. (4) Pefferkorn, E.; Carroy, A.; Varoqui, R. J. Polym. Sci., Polym. Phys. Ed. 1985, 23, 1997. (5) Frantz, P.; Granick, S. Phys. Rev. Lett. 1991, 66, 899. (6) Douglas, J. F.; Johnson, H. E.; Granick, S. Science 1993, 262, 2010.

S0743-7463(96)00519-7 CCC: $14.00

It had already been recognized that exchange processes play a primary role when blood plasma is brought in contact with a solid surface. In this case, abundant proteins of relatively low binding affinity, which are preferentially adsorbed at short times, are later replaced by less abundant proteins of high binding affinity. This most likely results from the interplay of transport and binding, which favors initially adsorption of abundant proteins, and high-affinity proteins, are taken up after longer reaction times.7,8 This cascade of events is known as the “Vroman effect”, and its understanding is of primary importance in the domain of biocompatibility.7 Many studies have been devoted to the Vroman effect, but most of them remained at the level of qualitative descriptions of the phenomenon. 9-17 Thus, the effects of plasma dilution,9 the nature of the surface,10 and residence time11 were found to be important. Many models have been proposed to explain the Vroman effect. Slack and Horbett11 developed an empirical model that incorporated the process of adsorption, transition, and displacement of proteins. In this model plasma was considered as a twoprotein system in which fibrinogen was taken as one component and all the other molecules as a second component. An alternative model in which adsorption and desorption rate constants were assumed to depend on the total surface protein concentration was proposed (7) Vroman, L.; Adams, A. L. J. Colloid Interface Sci. 1986, 111, 391. (8) Wojcieckowski, P.; Brash, J. L. J. Biomater. Sci., Polym. Ed. 1991, 2, 203. (9) Slack, S. M.; Horbett, T. A. J. Colloid Interface Sci. 1988, 124, 535. (10) Santerre, J. P.; ten Hove, P.; Vanderkamp, N. H.; Brash, J. L. J. Biomed. Mater. Res. 1992, 26, 39. (11) Slack, S. M.; Horbett, T. A. J. Colloid Interface Sci. 1989, 133, 148. (12) Cuypers, P. A.; Willems, G. M.; Hemker, H. C.; Hermens, W. T. Ann. N. Y. Acad. Sci. 1987, 516, 244. (13) Lu, C. F.; Nadarajah, A.; Chittur, K. K. J. Colloid Interface Sci. 1994, 168, 152. (14) Brash, J. L.; Uniyal, S.; Pusineri, C.; Schmitt, A. J. Colloid Interface Sci. 1983, 95, 28. (15) Burghardt, T. P.; Axelford, D. Biophys. J. 1981, 31, 455. (16) Beissinger, R. L.; Leonard, E. F. Am. Soc. Int. Org. J. 1980, 3, 160. (17) Lok, B. K.; Cheng, Y. L.; Robertson, C. R. J. Colloid Interface Sci. 1983, 91, 104.

© 1997 American Chemical Society

730

Langmuir, Vol. 13, No. 4, 1997

by Cuypers et al.12 The model developed by Chittur et al.13 is based on solving coupled mass transport and kinetics equations and incorporates the diffusion and reversible adsorption of the proteins followed by their transformations to an irreversible bound form. Despite these numerous studies, the laws governing the exchange reaction, which is the basis of the Vroman effect, have not been precisely established. With this goal, we began a series of experiments aimed at determining (i) the order of the reaction of the exchange process and (ii) if all (or only a fraction) of the adsorbed proteins can be exchanged. In previous studies, we investigated homogeneous (the same protein both present in solution and adsorbed on the surface) and heterogeneous (the adsorbed protein of a different nature than the protein in solution) exchange processes with fibrinogen or γ-immunoglobulins (IgG) adsorbed on various surfaces.18-20 The systems studied all exhibit common features: (i) The exchange process is found in the short time range. (ii) The exchange process is of order one with respect to the proteins in solution and there is strong evidence that it is also of order one with respect to the adsorbed macromolecules. (iii) At least three different populations of adsorbed molecules generally exist on the surface: (a) rapidly exchangeable and desorbable macromolecules which we call the type 1 population; (b) a population of molecules which slowly desorbs but does not exchange (we call this the type 2 population); and, (c) finally, a population of totally irreversibly adsorbed proteins (type 3 population). Similar conclusions were drawn by Schmidt et al.,21 who investigated the adsorption of hen-egg lysozyme on highly hydrophobic alkylated silicon oxide surfaces by means of fluorescence and photobleaching experiments. (iv) Even if in first approximation the three populations of adsorbed proteins can be assumed to be independent, there is experimental evidence that the adsorbed proteins evolve slowly from a population of type 1 to a population of type 2 or 3. These results have been obtained for fibrinogen and IgG proteins which are relatively high molecular weight macromolecules; moreover, they can easily change their conformation once adsorbed.2,22 However, it is not clear if these results should also apply to the case of compact, stable, and relatively small molecular weight proteins which are adsorbed on hydrophilic surfaces. Lysozyme represents such a protein,23 and the aim of this article is to analyze the homogeneous exchange and desorption processes of lysozyme adsorbed on TiO2 particles. Titanium is a well-known biomaterial which has shown excellent performance as a dental and orthopedic implant material, forming a close contact with the surrounding bone tissue with low occurrence of adverse reactions.24 Titanium reacts immediately with biological liquid, forming a surface oxide layer consisting primarily of titanium oxide. Thus, the tissue-implant reaction concerns the oxide and not the pure titanium. Adsorption onto a pure titanium surface was studied through ellipsometry and capacitance measurements25 or by radio-labeling.26 The (18) Ball, V.; Huetz, Ph.; El Aissari, A.; Cazenave, J. P.; Voegel, J. C.; Schaaf, P. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 7330. (19) Huetz, Ph.; Ball, V.; Voegel, J. C.; Schaaf, P. Langmuir 1995, 11, 3145. (20) Ball, V.; Bentaleb, A.; Hemmerle, J.; Voegel, J. C.; Schaaf, P. Langmuir 1996, 12, 1614. (21) Schmidt, C. F.; Zimmermann, R. M.; Gaub, H. E. Biophys. J. 1990, 57, 577. (22) Buijs, J.; Norde, W.; Lichtenbelt, J. W. Th. Langmuir 1996, 12, 1605. (23) Blake, C. C. F.; Koenig, D. F.; Mair, G. A.; North, A. C. T.; Phillips, D. C.; Sarma, V. R. Nature 1965, 206, 757. (24) Okabe, T.; Hero, H. Cells Mater. 1995, 5, 211. (25) Ivarsson, B. A.; Hegg, P. O.; Lundstroˆm, I.; Joˆnsson, U. Colloids Surf. 1985, 13, 169.

Bentaleb et al.

incidence of surface treatment and of titanium oxide type was checked and quantified by means of immunoassay27 or immunoelectrophoresis.28 However, except our previous study on titanium,20 and in spite of the importance of the dynamical aspect of the adsorption process and the protein layer composition, no precise experimental data concerning the dynamical behavior of proteins on this biomaterial are available. The study will be performed using a 125I radiolabeling technique, which is sensitive to the amount of adsorbed proteins. Moreover, this technique allows discrimination between the adsorption process and the release of adsorbed proteins from the surface, by labeling alternatively the macromolecules present in the solution and the macromolecules adsorbed on the surface. Techniques which are sensitive to the optical properties of the adsorbed layer could be well adapted to study desorption processes but cannot be used to investigate exchange processes. Materials and Methods Chemicals and Sorbent. Lyophilized hen-egg lysozyme (Mw 14 600 g‚mol-1; dimensions, 4.6 × 3.0 × 3.0 nm3;23 pHi, 11.4) was purchased from Sigma (Sigma, St. Quentin, France) and was of 90% purity. Ovotransferrin (0.2%), ovalbumin (3.8%), lysozyme dimer (0.7%), and an unknown 18 kDa protein (1.38%) were identified in the sample of the same origin.29 The protein was dissolved in PBS buffer prepared with 50 mM Na2HPO4‚12H2O and 50 mM NaH2PO4‚2H2O in the presence of 150 mM NaCl, the pH being adjusted to 7.5 (Mettler Electrode DG111-SC, Greifensee, Switzerland). All chemicals were of analytical grade and used without further purification. Buffer solutions were prepared with deionized water and were systematically filtered through a membrane (Millex, Millipore) with a pore size of 0.22 µm, degassed, and stored at 4 °C under vacuum until use. Lysozyme was radio-labeled according to a modification of the McFarlane method30 using iodine monochloride as oxidizing agent and Na125I for labeling. The amount of Na125I (Amersham France, Les Ulis, France) was chosen so that about 1 in 3000 lysozyme molecules was labeled. The solution was then divided in 2 mL aliquots, stored at -20 °C, and diluted to the desired protein concentration in PBS buffer just before use. Concentrations of labeled protein solutions were determined by absorbance at 280 nm with a DL Model 640 Beckman spectrophotometer (Beckman, Fullerton, CA) assuming an extinction coefficient of 2.59 cm2‚mg-1,31 and by γ counting (Minaxi Series 5000, United Technologies, Packard, France). For diluted solutions, the concentration was checked only by γ counting. The sorbent was TiO2 (Sigma, St. Quentin, France), having a specific surface area of (2.3 ( 0.2) m2‚g-1 determined by the BET method (Sorpty 1750, Carlo Erba). The small-angle X-ray diffraction pattern (Siemens D500 X-ray diffractometer) using Co KR radiation was characteristic for the pure rutile form. The point of zero charge surface and the charge density determined by potentiometry (Mettler Electrode) were pH0 ) 5.7 and σ0 ) -3.66 µC/cm2, respectively. A monomodal particle size distribution with an average size of 1.2 µm and variance of 0.246 was found by using quasielastic light scattering (Zeta Master S, Version PCS, Malvern Instruments, Malvern, England). The ζ potential in PBS buffer was found to be -8.4 mV (Zeta Master S). The sieved powder was rinsed in 99.5% ethanol, dried, and stored in a nitrogen atmosphere. Experimental Procedures. All the experiments were performed in 2 mL polycarbonate Eppendorf tubes at room temperature (25 ( 1) °C. (26) Williams, D. F.; Askill, I. N.; Smith, M. J. Biomed. Mater. Res. 1985, 18, 313. (27) Walivaara, B.; Arronsson, B. O.; Rodahl, M.; Lausmaa, J.; Tengvall, P. Biomaterials 1994, 15, 827. (28) Ellingsen, J. E. Biomaterials 1991, 12, 593. (29) Thomas, B. R.; Veikilov, P. G.; Rosenberger, F. Acta Crystallogr. 1996, D52, 776. (30) McFarlane, A. S. J. Clin. Invest. 1963, 42, 346. (31) Sober, H. A. CRC Handbook of Biochemistry; CRC Press: Cleveland, OH, 1970; p C71.

Lysozyme Adsorbed on a Titanium Oxide Surface

Figure 1. Influence of the 125I label on the adsorption process. Adsorption was performed for 7 h at a constant initial concentration of (4.7 ( 0.1) × 10-2%(w/w), varying the fraction of radioactive molecules. Γ refers to only the labeled adsorbed lysozyme. The full line is a linear regression through the data. The dashed lines correspond to the limit of the confidence interval of 95%. Adsorption Experiments. Adsorption processes were performed by adding (2.00 ( 0.02) g of solution containing various amounts of proteins to (0.20 ( 0.02) g of TiO2 under gentle rotation: (13 ( 1.0) rpm (Agitest 34050, Bioblock Scientific, Illkirch, France). The surface to volume ratio was maintained constant in all experiments. After a given time, the tube was centrifuged at 1000g for about 5 min, and three precisely weighed 200 µL samples of supernatant solution were taken for γ counting. The activity difference before and after adsorption allowed a precise estimation of the protein uptake. When using labeled protein to study the adsorption, one must first check if the label itself induces modification of the adsorption properties of the protein under study. We performed a simple test to check the effect of labeling on lysozyme adsorption. Labeled and unlabeled proteins were mixed in different ratios varying from 40% to 100% of the labeled protein solution while the total protein concentration was kept constant and equal to (4.7 ( 0.1) × 10-2%(w/w). When the labeling process induces a modification of the adsorption properties of a protein, a modification of the labeled to unlabeled protein ratio will lead to different adsorbed amounts. Figure 1 shows that, within experimental precision, the apparent lysozyme adsorption varied linearly with specific activity. Thus, after correction no difference in surface concentration for the different ratios was observed. For the system under study, labeling had no apparent incidence on the adsorption process. For kinetics, experiments were stopped at different time intervals, whereas the adsorption isotherm was performed by varying the bulk concentration, the contact time between the protein solution and the surface being kept constant at t ) 7 hours. The reproducibility in the surface concentration measurements was about 6%. Lysozyme Release Experiments. In these experiments, we checked the effect of varying the lysozyme bulk solution concentration on the release process of the adsorbed lysozyme proteins. Thus, labeled molecules were first adsorbed onto TiO2 particles at an initial concentration Ca ) (4.6 ( 0.1) × 10-2%(w/ w) corresponding to the plateau value of the isotherm Γ ) 0.045 µg‚cm-2. After 7 h of contact, rotation was stopped, the tube was centrifuged, and (1.4 ( 0.1) mL of the supernatant was removed, from which three precisely weighed 200 µL samples were taken to estimate the adsorbed amount. This volume of 1.4 mL was replaced by the same precisely weighed amount of pure buffer solution. After powder redispersion, the particles were maintained in contact with buffer under gentle rotation (cycle 1). A full cycle consists of powder centrifugation, the withdrawal of the supernatant, the addition of a fresh solution, and the redispersion of the particles followed by contact, during a given time period, with the solution. The three initial cycles each lasted 15 min. At the start of cycle 4 (considered as t ) 0 for the desorption or the release experiments) the supernatant was replaced by pure buffer (desorption reaction) or by an unlabeled protein solution with concentration (denoted Cbulk) varying from (1.36 ( 0.05) × 10-2%(w/w) to (5.49 ( 0.10) × 10-2%(w/w) (protein release experiments). At the end of each cycle, and after sample

Langmuir, Vol. 13, No. 4, 1997 731

Figure 2. Adsorption isotherm after 7 h of contact of the sorbent with labeled lysozyme at different initial protein concentrations. The horizontal axis corresponds to the bulk concentration at the end of the adsorption experiment. centrifugation, the released protein amounts were again estimated from the comparison of the activity of the three precisely weighed 200 µL aliquots and the bulk concentration of the labeled lysozyme solution immediately after addition of the unlabeled solution. The time duration of each cycle, from cycle 4 until cycle 13, was 1 h except for cycle 12, for which it was 15 h (overnight). Two complementary experiments of the same type with concentrations Ca equal to (1.07 ( 0.05) × 10-2%(w/w) and (4.7 ( 0.1) × 10-2%(w/w) used rigorously the same procedure but with labeled adsorbed molecules, in the presence of labeled bulk molecules with Cbulk ) (0.47 ( 0.05) × 10-2%(w/w) and Cbulk ) (3.1 ( 0.1) × 10-2%(w/w), respectively. In this series of experiments we determined the kinetics of supplementary lysozyme adsorption. The exchange process is not detectable, since both the adsorbed and the bulk proteins are labeled. Finally, two other sets of experiments, with Ca ) (1.07 ( 0.05) × 10-2%(w/ w) and Ca ) (4.7 ( 0.1) × 10-2%(w/w), used unlabeled molecules adsorbed onto TiO2 particles, the release being induced by labeled bulk molecules at Cbulk ) (0.47 ( 0.05) × 10-2%(w/w) and Cbulk ) (3.1 ( 0.1) × 10-2%(w/w) which replaced unlabeled ones at the sorbent surface.

Results and Discussion We first determined the adsorption isotherm by bringing the TiO2 particles in contact with lysozyme solutions of various concentrations Ca for 7 h (see Figure 2). Prior to these experiments, we verified that the uptaken amount remained constant after 2 h of contact with the solution even at such low protein concentrations as (0.43 ( 0.05) × 10-2%(w/w). The adsorption isotherm performed under such conditions shows a low affinity of the lysozyme for this sorbent surface. This is indicated by the weak steepness of the slope of the isotherm and the low plateau value, which is of the order of 0.045 µg‚cm-2. The adsorption of all traces of proteins present in the solution would represent less than 5% of the adsorbed amount in the plateau domain. The plateau domain is reached for bulk concentrations above 4 × 10-2%(w/w). This saturation amount corresponds to a coverage of approximately 13% and 20%, assuming “end-on” and “side-on” configurations, respectively, for the adsorbed molecules, the characteristic dimensions of the lysozyme being 4.6 × 3.0 × 3.0 nm3. This result contrasts with adsorbed amounts measured on silicon surfaces, almost 10 times more than in our case, indicating the formation of multilayers.32 On the other hand, it is in agreement with the observation of Horseley et al.33 for hen lysozyme adsorbed on dichloromethylsilane silica. Our low saturation value could be (32) Ball, A.; Jones, R. A. Langmuir 1995, 11, 3542. (33) Horseley, D.; Herron, J.; Hlady, V.; Andrade, J. D. In Proteins at Interfaces: Physico-chemical and Biochemical Studies; Brash, J. L., Horbett, T. A., Eds.; ACS Symposium Series 343; American Chemical Society: Washington, DC, 1987; p 290.

732

Langmuir, Vol. 13, No. 4, 1997

Figure 3. Typical exchange kinetics of labeled adsorbed lysozyme with unlabeled species for various bulk concentrations Cbulk: O, in the presence of pure buffer solvent; 3, (1.37 ( 0.10) × 10-2%(w/w); 0, (2.56 ( 0.1) × 10-2%(w/w); 4, (3.13 ( 0.10) × 10-2% (w/w). Continuous lines correspond to the fit by a sum of two exponential functions given by eq 1.

due either to surface heterogeneity, adsorption on specific sites or onto particular crystallographic planes,34 or an overestimation of the surface accessibility for proteins evaluated by the BET method. We examined the release of labeled adsorbed lysozyme proteins in the presence of unlabeled lysozyme in solution at bulk concentrations Cbulk ranging from 0 to 5.5 × 10-2%(w/w). Typical results are shown in Figure 3, which represents the amount of labeled lysozyme adsorbed on the surface as a function of the contact time between the adsorbed layer and the unlabeled solution. Initially the release of proteins is rapid, whereas it becomes much slower after 3 h of contact with the solution. This strongly suggests that an unique mechanism is not sufficient to describe the whole process. This observation is supported by the fact that an unique exponential decay function does not describe accurately the release kinetics over the entire time domain (data not shown). We then made the hypothesis, based on previous results obtained on other protein systems,14,19 of the existence at the surface of three independent adsorbed populations, and we postulated the following expression for the evolution of the adsorbed amount Γ*(t) as a function of the contact time t:

Γ*(t) ) Γ1* exp(-K1‚t) + Γ2* exp(-K2‚t) + Γ∞* (1) where Γ1* and Γ2* correspond respectively to the initial surface concentration of type 1 or type 2 adsorbed proteins (fast and slowly released macromolecules, respectively), which are characterized by the rate constants K1 and K2, and Γ∞* corresponds to the amount of irreversibly adsorbed proteins. The experimental data for zero bulk concentration correspond to a desorption experiment. The experimental data were fit by expression 1 using a least squares fitting procedure, the adjustment parameters being Γ1*, Γ2*, Γ∞*, K1, and K2. Figure 4 shows the evolution of the rate constants K1 (circles) and K2 (squares) as a function of the unlabeled lysozyme proteins concentration in solution (Cbulk). The curves obtained from eq 1 using the best fitting parameters are compared to the experimental data in Figure 3 (solid lines). The good agreement between expression 1 and the experimental data over the whole experimental time is a strong indication that our assumption of the existence of three independent populations adequately describes the analyzed release process. Moreover, the evolution of the rate constant with Cbulk demonstrates that the rapid release process is of order one with respect to the proteins in solution. In addition, (34) Glueckauf, E.; Petterson, L. Biochim. Biophys. Acta 1974, 351, 57.

Bentaleb et al.

Figure 4. Variation of the kinetic constants obtained from the fit versus bulk concentrations (Cbulk): O, K1; 0, K2. The full line is a linear regression through the data corresponding to K1. The dashed lines correspond to the limit of the confidence interval of 95%.

the fact that this release process is slower for pure desorption (Cbulk ) 0) suggests that it does constitute an exchange process between the adsorbed molecules and the proteins in solution. In addition to this exchange process, the population of type 1 can also desorb, as indicated by the nonzero value of K1 for Cbulk ) 0. On the other hand, the independence of K2 from Cbulk is a signature of a pure desorption process, already observed for other systems.19,35 The characteristic times associated with these processes range from 0.25 to 1 h for type 1 population, depending on the bulk concentration, and are equal to 4 h for the desorption process of the population of type 2. The study of the homogeneous lysozyme/lysozyme or IgG/ IgG18,20 and the heterogeneous IgG/fibrinogen19 exchange reactions led to identical observations for the rapid process for which the apparent rate constants vary also linearly with the protein concentration in the bulk. It is surprising that this first-order reaction holds as well for proteins able to undergo important conformational changes (human IgG’s, fibrinogen) as for a more rigid molecule like lysozyme. The fact that the whole release process can be described by eq 1 seems to indicate that these three populations are each independent of each other. Furthermore, the fact that the release/desorption process can be modeled by a double exponential decay function suggests that the process can also be modeled by a first-order kinetic reaction with respect to adsorbed lysozyme. Thus, the evolution of the populations 1 and 2 should be given by

dΓ1 ) -K1Γ1 ) -ke1CbulkΓ1 - kd1Γ1 dt

(2)

dΓ2 ) -K2Γ2 ) -kd2Γ2 dt

(3)

where Γ1 and Γ2 correspond to the adsorbed amounts of proteins of type 1 and 2, and ke1, kd1, and kd2 represent the exchange and desorption constants for type 1 and 2 populations. The rate constant associated with the rapid and slow release process of adsorbed molecules can be written as

K1 ) ke1Cbulk + kd1

(4)

K2 ) kd2

(5)

These constants are equal to (6.28 ( 1.0) × 104 L‚mol-1‚h-1, (35) Cheng, Y. L.; Darst, S. A.; Robertson, C. R. J. Colloid Interface Sci. 1987, 118, 212.

Lysozyme Adsorbed on a Titanium Oxide Surface

Langmuir, Vol. 13, No. 4, 1997 733

Table 1. Kinetical Exchange (ke1) and Desorption (kd1 or kd2) Rate Constants for Various Sorbent/Protein Systems system IgG/IgG IgG/IgG IgG/Fib Fib/Fib Lys/Lys

surface type latex (SO4 titanium SiO2 SiO2 TiO2

-)

ke1 (10-4 L‚mol-1‚h-1) 0.58 0.95 38.40 2.97 6.28

2.7 × 10-4 s-1, and 6.8 × 10-5 s-1, respectively (ke1 is obtained from the slope of the straight line in Figure 4, and kd1, from the intercept). These values, together with those obtained from other systems, are gathered in Table 1. The rapid exchange mechanism on hydrophilic surfaces is characterized by the rate constants ke1 and kd1, which generally do not differ by more than one order of magnitude between the different proteins and surfaces we used in all the studies. The variation observed in the ke1 values for IgG’s on titanium and lysozyme on TiO2 could be a consequence of differences in conformational stability of both proteins. The conformational changes of monoclonal IgG’s adsorbed on hydrophilic silica have been studied by means of FTIRATR spectroscopy.22 It was observed that the amounts of β-sheet decrease with time. This observation is in line with other works which concluded that IgG molecules are more tightly bound to the surface, thereby undergoing further structural changes.36 In contrast, lysozyme, a more compact protein, shows less conformational changes as observed by micro-DSC.37 The existence of an exchange process which is of first order with respect to both the molecules in solution and those on the surface implies the existence of an adsorption flux which is equal to the flux of released proteins. In order to verify this point, we have performed experiments in which the adsorbed macromolecules were unlabeled, with labeled proteins in solution, and compared them with their counterparts where the unlabeled molecules were in the solution, the adsorbed molecules being labeled. These experiments were performed at two different concentrations Ca ) 1.07 × 10-2%(w/w) and 4.7 × 10-2%(w/ w) (which correspond respectively to an adsorbed amount of 0.015 and 0.043 µg‚cm-2) during the adsorption stage of the lysozyme proteins on the bare TiO2 surface. The higher concentration Ca corresponds to the same experimental conditions as for the preceding experiments. We performed the experiment at the lower bulk concentration Ca for which the isotherm plateau is not reached in order to verify if our conclusions can be extended to these adsorption conditions. This type of experiment is more difficult to perform than the preceding ones because the adsorbed amounts are determined by depletion of the bulk concentration, which should not change significantly in order to be suitable for a quantitative analysis. Moreover, after the adsorption of the unlabeled proteins and before the labeled protein solution is introduced in the cell, the TiO2 particles are rinsed with pure buffer. During this stage, desorption occurs from the surface. The amount of labeled proteins which arrive on the surface, after the exchange process has taken place, can thus come not only from the exchange process but also from readsorption on the sites that have become vacant due to the desorption process during the three rinsing steps. In order to evaluate the contribution from the exchange process, we must first determine the readsorption contribution by bringing a layer of labeled adsorbed lysozyme proteins in contact with a solution of labeled proteins, the procedure remaining otherwise identical to the one described in the (36) Elgerma, A. V.; Zsom, R. L. J.; Norde, W.; Lyklema, J. J. Colloid Interface Sci. 1992, 152, 410. (37) Haynes, C. A.; Norde, W. J. Colloid interface Sci. 1995, 169, 313.

kd1 (10-4 s-1) 0.25 1.10 0.58 2.70

kd2 (10-4 s-1)

ref

6.26

18 20 19 unpublished data present work

0.68

Figure 5. Lysozyme released or fixed at the sorbent surface during an exchange experiment performed with lysozymecoated TiO2 particles. The sorbent was saturated from Ca ) (1.07 ( 0.05) × 10-2%(w/w) lysozyme. The exchange was performed with Cbulk ) (0.47 ( 0.05) × 10-2%(w/w) lysozyme. Different experiments correspond to radio-labeling of different molecules (Ca and/or Cbulk). b, release kinetics of labeled adsorbed lysozyme Ca by an unlabeled lysozyme solution Cbulk; O, labeled lysozyme adsorption from Cbulk on a TiO2 surface initially coated by unlabeled lysozyme Ca; ∇, lysozyme readsorption onto a lysozyme coated surface and obtained by reacting a labeled lysozyme solution Cbulk with a TiO2 surface coated with labeled lysozyme solution Ca; 4, exchange kinetics of the labeled lysozyme reacting with the unlabeled lysozyme-covered surface. This was obtained by subtracting 3 from O.

Materials and Methods. We are thus insensitive to the pure exchange process. It is observed that the readsorption is a slower process than the direct adsorption onto a bare surface; it can last over more than 10 h (see 3 in Figures 5 and 6). Moreover, it leads, at equilibrium, to an increase of the adsorbed amount of the order of 0.0024 and 0.0089 µg‚cm-2 for the low and high bulk concentrations Ca, respectively. We then performed the experiments in which we determined the amount of proteins which reach the surface during the adsorption/exchange process. The amount of protein which does not correspond to the readsorption process is obtained from the experimental kinetic data (O in Figures 5 and 6) by subtracting from it the readsorption contribution (3 in Figures 5 and 6). The difference (4 in Figures 5 and 6) then corresponds to the amount that adsorbs on the surface through an exchange or desorption/ adsorption process. It can be directly compared to the amounts of adsorbed proteins which are released as determined by labeling the proteins on the surface, the solution being unlabeled (b in Figures 5 and 6). The good agreement between the quantities determined by the two kinds of experiments demonstrates that the flux of molecules released from the surface, measured previously, is indeed counterbalanced on the surface by an incoming flux of proteins. The processes corresponding to the release of type 1 proteins are thus exchange and desorption/adsorption processes whereas the release of adsorbed proteins of type 2 must correspond to a pure desorption/ adsorption mechanism. The additional uptake of lysozyme during the experiments in which both the adsorption layer and the solution were labeled can be attributed to a reorganization of the adsorption layer following the desorption processes during the three rinsing steps. It cannot result from a reorganization of the layer due to

734

Langmuir, Vol. 13, No. 4, 1997

Figure 6. Lysozyme released or fixed at the sorbent surface during an exchange experiment performed with lysozymecoated TiO2 particles. The sorbent was saturated from Ca ) (4.7 ( 0.1) × 10-2%(w/w) lysozyme, and the exchange was performed with Cbulk ) (3.1 ( 0.1) × 10-2%(w/w) lysozyme. Different experiments correspond to radio-labeling of different molecules (Ca and/or Cbulk): b, release kinetics of labeled adsorbed lysozyme Ca by an unlabeled lysozyme solution Cbulk; O, labeled lysozyme adsorption from Cbulk on a TiO2 surface initially coated by unlabeled lysozyme Ca; 3, lysozyme readsorption onto a lysozyme-coated surface and obtained by reacting a labeled lysozyme solution Cbulkwith a TiO2 surface coated with labeled lysozyme solution Ca; 4, exchange kinetics of the labeled lysozyme reacting with the unlabeled lysozymecovered surface. This was obtained by subtracting 3 from O.

Figure 7. Evolution with Cbulk of the relative population of 1, 2, and 3 (in %): b, population 1; 0, population 2; O, population 3.

desorption or exchange processes measured in the presence of the solution in the bulk because this would be in contradiction with the fact that one attains a steady state after only 2 h of adsorption, as shown previously. The comparison between the released and adsorbed quantities by the same mechanism also gives an idea of the precision of our experimental approach. Finally, from the experiments in which we have determined the amounts of proteins released from the surface, modeled by eq 1, one can also deduce the percentage of adsorbed lysozyme which corresponds to proteins of type 1, 2, and 3 as a function of Cbulk (Figure 7). We made the assumption that these populations are all independent. This, however, would imply that they are also independent of Cbulk because all the desorption and exchange experiments were run by a similar adsorption procedure. However this is not observed in Figure 7; on the contrary, one observes an increase (respectively, a decrease) of the percentage of the molecules of type 1 (respectively, type 3) when Cbulk increases. On the other hand the percentage of the population of type 2 seems independent of Cbulk. This proves that, in fact, the adsorbed proteins evolve with contact time on the sorbent surface, and the independence of the three types of adsorbed molecules is no more than a first and crude approximation

Bentaleb et al.

which, nevertheless, allows the description of many of the observed features. The decrease of population 3 (irreversibly adsorbed proteins) with Cbulk was never clearly described previously but occurs in the exchange kinetics of synthetic polymers adsorbed onto modified silica beads,4 for phosphorylase b adsorbed onto butylagarose gels,38 and for IgG adsorbed onto titanium particles.20 This observation seems to be a general phenomenon and could be attributed to a competition between the cooperative exchange process and the transformation of reversibly adsorbed proteins to proteins in an irreversible adsorbed state. In order to test the evolution of the different types of populations adsorbed on the surface with residence time, we have performed one experiment in which labeled lysozyme proteins from a solution of concentration Ca equal to 4.7 × 10-2%(w/w) were adsorbed during 24 h on bare TiO2 particles, instead of the 7 h in all the preceding experiments. The adsorbed layer was then brought in contact with an unlabeled lysozyme solution at a bulk concentration Cbulk of 2.56 × 10-2%(w/w). The release kinetics of the adsorbed proteins was again modeled by eq 1. We found that K1 remained unchanged (1.98 h-1 to be compared to 2.00 h-1) whereas K2 increased slightly (0.26 h-1 instead of 0.21 h-1). On the other hand, the percentage of proteins of type 1 decreased markedly (12.5% instead of 30%) while the percentage of molecules of type 2 and 3 increased from 30% to 41% and from 40% to 46.5%, respectively. These observations can only be explained by a more refined approach. One possible explanation is as follows: we will still make the assumption of the existence of three (and only three types of) populations of adsorbed proteins on the surface. In the experiments relative to Figure 7 one follows the evolution of the proteins which were adsorbed during the 7 h. These molecules could be initially adsorbed in one of the three states corresponding to the three adsorbed population types. However, the molecules adsorbed in a state of type 1 evolve with contact time by conformational changes to a state of type 3 whereas the molecules adsorbed in a state of type 2 remain in their adsorption state. On the other hand molecules of type 1 can be exchanged by proteins from the bulk, which mainly adsorb in a state of type 2. Thus, when an adsorbed layer is brought in contact with a protein solution, the proteins which adsorbed in a state of type 1 will be exchanged at a rate that is proportional to Cbulk. When the characteristic exchange time is smaller than the characteristic time associated with the conversion from a state of type 1 into a state of type 3, the percentage of molecules of type 1 will decrease whereas the percentage of molecules of type 3 will increase accordingly, as is observed in Figure 7. On the other hand if the adsorption of the proteins on the bare TiO2 surface takes place over 24 h, the exchange processes gradually replace molecules adsorbed in state 1 by new molecules which will adsorb in state 2. There is no doubt that this explanation constitutes no more than a first rough approach and that further studies are necessary to investigate precisely the evolution of the different adsorbed states with residence time. Conclusion The exchange and desorption processes have been studied for a small, stable, and highly charged protein, lysozyme, adsorbed on TiO2 particles. The existence of at least three adsorbed protein populations (a “rapidly” exchangeable and desorbable, a slowly desorbable, and an irreversibly fixed population) has been determined. (38) Jennissen, H. P. J. Colloid Interface Sci. 1986, 111, 570.

Lysozyme Adsorbed on a Titanium Oxide Surface

The value of ke1, the exchange rate constant, is estimated at 6.28 × 104 L‚mol-1‚h-1. We also confirm that this reaction is also of order one with respect to the molecules in the bulk. We demonstrate for the first time that the kinetics of the release process of molecules of type 2 is independent of the bulk concentration and should thus constitute a desorption process. Our experimental findings thus validate some of the assumptions generally made in the usual kinetic models (in particular the reaction orders) for competitive adsorption and which were never totally demonstrated up to now. This view of the release processes still constitutes a first approximation which accounts for many of the observed features and provides a general scheme that seems to apply to a large number of protein systems. However, it does not take into account

Langmuir, Vol. 13, No. 4, 1997 735

the evolution of the different adsorbed populations with adsorption time. Different more refined models could be developed to take these processes into account. New additional parameters could thus also be introduced. However the experimental data would not allow us to discriminate between the different models, and these could thus constitute pure speculation. Acknowledgment. One of the authors (A.B.) is indebted to the Faculte´ de Chirurgie Dentaire de Strasbourg for financial support. We thank E. K. Mann for critical reading and improvement of the English text. LA9605192