Protein Adsorption Modalities on Polyelectrolyte Multilayers

Kin Man Au and Steven P. Armes . Heterocoagulation ..... Reine Nehmé, Catherine Perrin, Hervé Cottet, Marie-Dominique Blanchin, Huguette Fabre. Stab...
0 downloads 0 Views 258KB Size
Biomacromolecules 2004, 5, 1089-1096

1089

Protein Adsorption Modalities on Polyelectrolyte Multilayers David S. Salloum and Joseph B. Schlenoff* Department of Chemistry and Biochemistry, Center for Materials Research and Technology (MARTECH), The Florida State University, Tallahassee, Florida 32306-4390 Received December 11, 2003; Revised Manuscript Received January 16, 2004

Protein adsorption on polyelectrolyte multilayers (PEMUs) was evaluated using a combination of synthetic polyelectrolytes and proteins, including serum albumin, fibrinogen, and lysozyme. Variables such as surface and protein charge, polymer hydrophobicity, and hydrophilic repulsion were introduced to probe interaction mechanisms. Quantitative analysis with reflectance Fourier transform infrared spectroscopy, optical waveguiding, and UV-vis absorption, together with qualitative information from atomic force microscopy, provided a coordinated picture for what drives protein adsorption and how the molecules are disposed on the multilayer surface. It was found that multilayers bearing a particular surface charge sorbed biomolecules if they were of opposite charge, yielding significant loadings within the bulk PEMU. Adsorption of likecharged proteins, as surface aggregates, occurred to a much lower extent, driven by nonelectrostatic forces. A diblock copolymer comprising a hydrophilic poly(ethylene oxide) block was capable of further minimizing protein adsorption as a result of hydrophilic repulsion, although none of the surfaces tested defeated protein adsorption completely. However, poly(acrylic acid) homopolymer was quite effective in this respect. A composition gradient, formed during multilayer buildup, induced a gradient in hydrophilicity through the PEMU, which is an efficient and economical method of creating a protein-resistant surface. Introduction Biofunctional thin films with controlled bulk or surface properties have been made by the alternating adsorption of synthetic or natural macromolecules to various substrates.1,2 The wide variety of charged nanocomponents incorporated into these “polyelectrolyte multilayers” (PEMUs) includes biomolecules3 such as DNA,4 immunoglobulin,5 glucose oxidase,6 and cytochrome c.7 These biomolecular thin films offer applications as biosensors,8 nano-filtration,9 bioreactors,10 and protein capsules.11 Although multiple “electrostatic” interactions between synthetic and natural charged polymers might be presumed to modify protein conformations, individual interactions are actually quite weak,12 and under certain conditions, proteins embedded into PEMUs during the buildup process maintain a secondary structure close to their native form,13 which renders such PEMUs bioactive. A recent area of interest addresses protein interactions with multilayer-coated surfaces. Fine-tuning of protein adsorption at the solid/liquid interface is critical in certain areas of materials science and biomedical engineering.14 Systems for delivery or biosensors, for example, bear modified surfaces designed to enhance or minimize protein adsorption.15,16 The latter goal is generally desirable for blood-contacting devices, chromatographic supports, contact lenses, and immunoassays, to name a few. As a result of their ease of use and water compatibility, PEMUs have been investigated as surfacemodifying agents for protein interactions.17-22 Because * To whom correspondence should be addressed. E-mail: schlen@ chem.fsu.edu.

protein adsorption triggers further cellular or tissue responses, it is important to study the mechanism by which proteins adsorb onto these thin films.20 Interactions of more biocomplex systems, such as cells, with PEMUs, a topic of more recent interest, were studied,23 and different designs for cytophilic/cytophobic PEMUs have been proposed.24 Protein adsorption is driven by the net influence of various interdependent interactions between and within surface and biopolymer. Possible protein-polyelectrolyte interactions25 can arise from (1) van der Waals forces, (2) dipolar or hydrogen bonds, (3) electrostatic forces, and (4) hydrophobic effects. Given the apparent range and strength of electrostatic forces, it is generally accepted that the surface charge plays a major role in adsorption. However, proteins are remarkably tenacious adsorbers because of the other interaction mechanisms at their disposal. In this paper, we present a systematic comparison of protein adsorption mechanisms onto PEMUs, with an emphasis on quantitative analysis, and we show how surfaces may be selected to encourage or discourage the adsorption of proteins. Different model proteins with different sizes and charges were selected to represent variety from the adsorbate perspective. Similarly, the variety of surface types and potential adsorption mechanisms was represented by different surfaces. Factors such as surface charge, ionic strength, and thickness of the multilayer were investigated. Our analytical techniques for adsorbed amounts included optical labeling, surface-sensitive Fourier transform infrared (FTIR) spectroscopy, and optical waveguiding, permitting cross-comparisons between data as well as a wide experimental range (from 0.01 to 100 mg m-2). Qualitative surface information was obtained with atomic force microscopy

10.1021/bm034522t CCC: $27.50 © 2004 American Chemical Society Published on Web 03/19/2004

1090

Biomacromolecules, Vol. 5, No. 3, 2004

Salloum and Schlenoff

Figure 1. Structures of polyelectrolytes used in this study. Table 1. Proteins and Polyelectrolytes Used in This Study protein/ polyelectrolyte

molecular weight, kDa

pIa

net charge at pH 7.4

BSA fibrinogen lysozyme PDADMAC PSS PM2VP PM2VP-b-PEO PAA PAH

66.4 340 14.3 400-500 70 50 56.5:5.9 240 70

4.9 5.5 11.4 NA NA NA NA 5.332 8.833

+ + + +/neutral +

a

pI ) Isoelectric point for proteins and pKa for polyelectrolytes.

(AFM), which allowed us to discern between possible adsorption models. We demonstrate that surface properties, rather than the bulk of the multilayer, play the crucial role in the adsorption process, and we show that, by exposing them to films of opposite charge, proteins can be incorporated into PEMUs, where the amount “sorbed” depends on the thickness of the underlying PEMU. Materials and Methods Materials. The polyelectrolytes used in this study are shown in Figure 1. Poly(styrenesulfonic acid), PSS (molecular weight, MW ∼ 7 × 104); poly(diallyldimethylammonium chloride), PDADMAC (MW ∼ 3 × 105); poly(allylamine hydrochloride), PAH (MW ∼ 7 × 104); and poly(acrylic acid), PAA (MW ∼ 24 × 104) were used as received from Aldrich. Poly(N-methyl-2-vinyl pyridinium bromide), PM2VP (MW ∼ 5 × 104) was from Polysciences, Inc. Poly(N-methyl-2-vinyl pyridinium iodide-block-ethylene oxide), PM2VP-b-PEO (PM2VP block 86% quarternized, respective block molecular weights 56 500:5900 Mw/Mn ) 1.08), was from Polymer Source, Inc. All polymer and buffer solutions were prepared using 18 MΩ water. Polymer solution concentrations were 1 mM or 10 mM (with respect to the monomer repeat unit) in sodium chloride salt (NaCl, Fisher). Proteins used for this study are summarized in Table 1. Bovine serum albumin (BSA) (fraction V powder, approximately 99%), fibrinogen (fraction I, type I-S from bovine plasma), and lysozyme (from chicken egg white) were

obtained from Sigma. For experiments requiring fluorescent probes, fluorescein isothiocyanate (FITC)-albumin from Sigma was used. Proteins were dissolved in TRIS buffer (pH 7.4) with a protein concentration of 1.0 mg/mL. This protein bulk concentration lies in the plateau region of the adsorption isotherm.16 The buffer ionic strength was adjusted by addition of sodium chloride. Protein solutions were filtered prior to use through a 0.45-µm nylon filter disk. PEMUs on the Si Wafer. Polymers were deposited on silicon wafers (Si〈100〉, 0.5-mm thick, 1-in. diameter, undoped, polished on one side, Topsil, Inc.) that were cleaned in 70% H2SO4 (concentrated)/30% H2O2(aq) (“piranha”: caution, piranha is a strong oxidizer and should not be stored in closed containers) and then in hot H2O2/ammonia/water (1:1:7 v/v), rinsed in water, and blown dry with a stream of nitrogen. Sequential adsorption of polyelectrolytes was performed by hand dipping or with the aid of a robot (nanoStrata, Inc.), where the exposure time for the two polymers was 5 min with three rinses of fresh distilled water, 1 min each, between. Film Thickness. The film thickness was determined on PEMUs that had been rinsed in water and dried, using a Gaertner Scientific L116B autogain ellipsometer with 632.8nm radiation at a 70° incident angle. A refractive index of 1.54 was employed for the multilayer film. Thicknesses are quoted as “dry” thickness unless otherwise stated. When immersed in water, PEMUs swell. For example, PDADMAC/PSS in water approximately doubles in thickness.12 Attenuated Total Reflection (ATR)-FTIR Spectroscopy. FTIR spectroscopy (Nicolet Nexus 470 FTIR) was used in attenuated total internal reflection mode (ATR, Specac, Inc., flow cell of volume 0.49 mL) to monitor protein adsorption onto PEMUs assembled on an ATR cell housing a 70 mm × 10 mm × 6 mm 45° germanium (Ge) crystal. Multilayer buildup was done by alternately filling the ATR cell with polymers (1 mM in 0.25 M NaCl), with intervening rinses in water. Solution pH for buildup, including rinse, with PAH was stabilized with TRIS buffer (pH 7.4). The exposure time for each solution was 10 min. Multilayers for ATR were not dried before protein adsorption. A multilayer spectral background in buffer was taken prior to protein adsorption. Layer-by-layer buildup and protein adsorption were monitored using areas of characteristic bands of interest (sulfonate stretch for PSS, ν(SO3-), at ∼1033 cm-1 and amide II band at ∼1540 cm-1). All spectra were recorded using 32 scans and a 4-cm-1 resolution. After addition of the protein solution to the ATR cell, the protein spectrum was monitored with time until there were no further significant changes in the spectra, the cell was rinsed with buffer, and amide II peaks were integrated. H2O spectra were subtracted from raw infrared spectra according to Chittur.26 The amounts of proteins were calculated on the basis of calibration curves for each protein (see Supporting Information). UV-Vis Spectroscopy. UV-vis absorption spectra were recorded on quartz-supported multilayers using a PerkinElmer UV/vis/NIR spectrometer (Lambda 900). Fused quartz plates (2-mm thick, 1-in. diameter, GM Associates) were pretreated with “piranha” and then in H2O2/ammonia/water (1:1:7) and then rinsed. The film thickness was estimated

Protein Adsorption Modalities

from multilayers deposited on the native SiO2 layer (about 20-Å thick) on silicon wafers using the same conditions. AFM. AFM images were performed in air using a Dimension 3001 unit (Digital Instruments, Inc., Santa Barbara, CA) with a type RTESP silicon tip with a 125-µm length and 300-kHz resonance in the tapping mode. 27 AFM was also used to measure the thickness on quartz disks and Ge crystal. As a check, a profilometer (KLA-Tencor P15) provided the same thickness information. Optical Waveguide Lightmode Spectroscopy (OWLS). The technique and instrument have previously been employed for multilayer buildup studies, where the experimental setup and mathematical treatment of raw data are presented in detail.28-30 From R (incoupling angle) and N (effective refractive index), an apparent (or “optical”) thickness tf and refractive index nc of a uniform film contacting the waveguide may be estimated. OWLS, having a somewhat lower limit of detection than ATR, was used here to determine surface coverage for minimally adsorbing systems. The waveguide was cleaned with hot sulfuric acid, rinsed with water, and then stored in buffer. All solutions were made up in TRIS buffer of pH 7.4 and ionic strength 154 mM. Polymers were PAH and PAA (1 mM each). Polyelectrolyte solutions were alternately pumped through the cell for 4 min each. Pure buffer, instead of water, was used as the rinse because of the extreme sensitivity of OWLS to solution refractive index changes. The flow rate of the solutions through the cell was 8.3 µL s-1. Measurements of R were taken every 13.4 s. The temperature was maintained at 25 ( 0.5 °C. After six layers of PAH/PAA, the protein was injected for 2 h followed by a buffer rinse. Using a bilayer model identical to that described in ref 31, protein adsorbed amounts were calculated. In our hands, the limit of detection was 0.2 mg/m2 for ATR and 0.017 mg/m2 for OWLS. PEMU Nomenclature. For clarity, we employ the following shorthand for multilayers: (A/B)x where A is the starting polyelectrolyte contacting the substrate, B is the terminating polyelectrolyte in contact with subsequent protein solutions, and x is the number of layer pairs. In (A/B)xA, A would be the terminating polymer. Salt, MY (cation M and anion Y), has an important role in the buildup process and is represented by (A/B)x @c MY, where c is the molarity of the salt (MY) in the polymer solution. The pH can be included in the nomenclature especially when using pHdependent PEMUs. For example, (PAH/PAA)2 @0.25 M NaCl @pH 7.4 represents two layers pairs of PAH/PAA built at 0.25 M NaCl and a pH of 7.4. Results and Discussion Protein Adsorption Kinetics using ATR-FTIR Spectroscopy. The PEMU/protein combinations were selected to represent various modes of surface/biomolecule interaction and adsorption. While the number of possible combinations is immense, our goal was to make some broad deductions concerning the role of each interaction type. Clearly, from electrostatic arguments, protein adsorption should be minimized on surfaces with like charge. This philosophy was behind our use of positively charged PEMUs for the highly

Biomacromolecules, Vol. 5, No. 3, 2004 1091

Figure 2. Adsorption kinetics of BSA onto (A) (PM2VP/PSS)2PM2VP @0.25 M NaCl (thickness 65 ( 15 Å) and (B) (PM2VP/PSS)2 @0.25 M NaCl (thickness 44 ( 15 Å). BSA (1 mg/mL) was deposited from TRIS buffer (pH 7.4) with 1 mM ionic strength at room temperature (25 ( 1 °C). Inset shows the increase in amide I and amide II bands with time.

efficient capillary electrophoretic separations of basic proteins. The efficiency of separations was evidence for minimal protein adsorption.34 Using ATR-FTIR spectroscopy, layer-by-layer deposition of PEMUs was monitored in situ using the strong sulfonate stretch for PSS, ν(SO3-), at ∼1033 cm-1 and the ν(COO-) for PAA at ∼1554 cm-1. As observed previously,17 (see also Supporting Information), multilayer buildup was characterized by an increase in the SO3- peak intensity for each layer, followed by a leveling off of this signal as the multilayer thickness exceeded the range of penetration of the evanescent wave from the ATR crystal. The intensity of the sulfonate band oscillated depending on which polymer was the outer layer. This phenomenon is due to the differential water content of the multilayer35 (there is less water and the film is denser when PSS is on top) and the resulting shrinking or swelling of the PEMU. To establish the time required for adsorbed amounts to reach their limiting values, the adsorption kinetics was followed. Figure 2 depicts kinetics of adsorption for BSA onto (PM2VP/PSS)2PM2VP @0.25 M NaCl (curve A), where the outer surface charge is positive, and (PM2VP/ PSS)2 @0.25 M NaCl (curve B), where the surface charge is negative. It can be seen from both systems that adsorption is almost 100% complete within 30 min. The inset in Figure 2 shows the growth of the amide II band at ∼1540 cm-1 used to follow adsorption. For accurate quantities, the area was converted to the adsorbed amount using calibration curves for proteins.36 Electrostatic Contributions. While Figure 2 suggests the major role of electrostatic interactions in BSA adsorption, it is clear that the protein still adsorbs on like-charged surfaces. Because the protein charge is the net sum of all positive and negative charges on the biomolecule, it could be argued that patches of positive charge are responsible for adsorption on the negative surface.25 On the other hand, it may be that nonelectrostatic interactions are responsible for adsorption on like-charged surfaces. To distinguish between these two possible modes of adsorption onto like-charged surfaces, the ionic strength of

1092

Biomacromolecules, Vol. 5, No. 3, 2004

Salloum and Schlenoff

Table 2. Ionic Strength Effect for BSA Adsorption Amount, Γ, onto PM2VP/PSS @0.25 M NaCl multilayer

TRIS ionic strength, mM

surface charge

Γ, mg/m2

(PM2VP/PSS)2 (PM2VP/PSS)2PM2VP (PM2VP/PSS)2 (PM2VP/PSS)2PM2VP

154 154 1 1

negative positive negative positive

0.80 ( 0.14 1.6 ( 0.2 1.1 ( 0.1 7.5 ( 0.2

the buffer was changed. Salt concentration controls the range and strength of electrostatic interactions, which are reduced at a higher ionic strength. BSA adsorbed amounts on PM2VP/PSS from buffers of different ionic strengths are summarized in Table 2. It is seen that the adsorption of BSA is almost five times less on the oppositely charged surface because of enhanced screening of the surface charge by salt, while it is almost independent of the ionic strength for the like-charged surface. It can be concluded that adsorption on the like-charged surface is due to nonelectrostatic interactions. Bulk versus Surface Sorption. After two or three layers, the film thickness for all multilayers studied here was proportional to the number of layers. The thickness of the multilayers is often many equivalent monolayers worth of material. Even a “thin” PEMU of 100-Å thickness can be larger than a protein molecule. The question arises as to whether the protein remains on the surface or is absorbed into the bulk of a PEMU, spongelike. The term “sorption” is used to denote a combination of (surface) adsorption or bulk absorption. While surface adsorption of proteins should not depend on the thickness of the PEMU, bulk absorption should scale with the multilayer thickness. The experiment to measure the sorbed amount as a function of the multilayer thickness employed optically labeled BSA (FITC-BSA) because the OWLS and ATRFTIR methods were limited to films of thinner dimension (because of the limited penetration of evanescent waves). While FITC-labeled proteins are commonly used in the fluorescence mode, we found some variability in the quantum efficiency of the FITC-BSA material that depended on the environment (i.e., whether it was incorporated into a multilayer). We, therefore, relied on UV-vis absorbance of the label at 503 nm. Adsorption kinetics of BSA onto PEMUs were measured to establish a saturation adsorption time. The absorbance of FITC-BSA on a range of thicknesses and surface charges established that about 90 min was enough for the system to reach steady-state adsorption (see Supporting Information). Sorption of proteins onto PDADMA/PSS, presented as an apparent surface coverage, as a function of the number of layers is plotted in Figure 3. For PEMUs terminating with a positive charge, the sorbed amount is substantial and is a linear function of the PEMU thickness (number of layers), while sorption on negative PEMUs is low and relatively independent of the multilayer thickness. From our previous work,37 the surface charge density (φ) stays constant during the buildup process of the multilayers. The most apparent interpretation of Figure 3 is that proteins are drawn into the multilayer when the surface is

Figure 3. Dependence of BSA adsorption on the number of layers of PDADMA/PSS multilayers @0.25 M NaCl. (A) Adsorption amounts, Γ, in mg/m2, on positively charged surfaces where the fitted line shows a linear dependence. (B) Adsorption amounts on negatively charged surfaces, which is almost independent of the thickness. The 28-layer PEMU had a thickness of approximately 930 Å.

of opposite charge, whereas they remain on the surface when it is like-charged. These very different modes of behavior can be overlooked if the surface charge is not an experimental variable. For example, Caruso et al.5 showed that the amount of IgG adsorbed on two and five bilayers of PAH/PSS (negative surface charge in both cases) is almost the same (consistent with our results), leading to the conclusion that proteins are unable to penetrate the multilayer. The bulk composition of the multilayer is independent of the surface charge. Therefore, if bulk PEMU is able to absorb proteins like a sponge, there must be significant kinetic limitations preventing a protein from soaking into a PEMU with a like-charged surface. The finding that large amounts of protein spontaneously absorb into a multilayer has important implications for biomolecule/PEMU composites and applications: instead of alternately depositing protein and polymer, which is rather laborious, biomolecules can be loaded into a PEMU simply by ensuring that its surface has the opposite charge and soaking it in a protein solution. It should be possible to load mixed biomolecules, such as coupled enzyme systems,6,10,38 by sequential or simultaneous soaking in solutions of the desired biomolecule. Loadings can be appreciable: for example, in the 900-Å film (27 layers) in Figure 3, the protein amounts to about 26 wt % of the total dry material (polyelectrolytes plus protein) in the PEMU. Surface Characterization using AFM. There are several possible modalities for protein sorption into/on PEMUs. Some of these, previously considered,20 are shown in Figure 4. Cases a and b represent roughly a monolayer adsorbing to the surface either uniformly (a) or islandlike (b). Cases c and d are different ways of adsorbing several monolayers, either aggregated on the surface (c) or sorbed throughout the bulk (d). Measurements of sorbed amounts cannot distinguish case a from case b and case c from case d. Therefore, AFM was utilized to characterize the surface before and after protein adsorption. Figure 5 shows the difference between the positively charged surface (A) and the negatively charged surface (B) after the adsorption of BSA onto PDADMA/

Protein Adsorption Modalities

Biomacromolecules, Vol. 5, No. 3, 2004 1093

Figure 4. Schematic of protein adsorption onto/into PEMUs. (a and b) Apparent monolayers. (c and d) Apparent multilayers.

Figure 5. AFM images, 5 µm × 5 µm, of BSA adsorbed on (A) (PDADMA/PSS)9PDADMA @0.25 M NaCl (thickness 625 ( 50 Å) and (B) (PDADMA/PSS)10 @0.25 M NaCl (thickness 660 ( 50 Å). Section analysis shows the features on the PEMU surface.

PSS. This Figure depicts 19- and 20-layer PEMUs with BSA sorbed. The positively charged PEMU, though it contains much more protein, has a much smoother surface. The negative PEMU shows islands of protein aggregates. AFM images of the starting multilayers, not shown, resemble that for Figure 5A, with a root-mean-square surface roughness of about 20 Å. These results are consistent with modes b and d from Figure 4 for sorption on negative and positive surfaces, respectively. Aggregates in Figure 5B are much larger than the original size of the protein molecule (4 nm × 14 nm).39 It is not

possible to tell from AFM whether the areas between aggregates have a monolayer of BSA, such as that in Figure 4a, but the aggregation itself is consistent with the low affinity of protein with the surface. It may be that certain areas for adsorption are nucleated, perhaps by hydrophobic interactions, but because the aggregates are negatively charged, they repel each other. Also, because the polyelectrolyte constituents in the PEMU are intimately mixed to the surface, a small fraction of surface PDADMA may reach out and participate in immobilizing the BSA via ion-pairing interactions.

1094

Biomacromolecules, Vol. 5, No. 3, 2004

Salloum and Schlenoff

Figure 6. Evanescent wave electric field amplitude (E) falls off exponentially away from the surface of the crystal. (A) BSA adsorption on the negatively charged surface. (B) BSA adsorption on the positively charged surface. dp represents the penetration depth of the evanescent wave normal to the surface of the crystal, using Ge (n ) 4.0) in contact with the medium of n ) 1.42.

AFM images for BSA adsorption on positively charged surfaces, though consistent with model 4d and the results in Figure 3, did not provide definitive evidence that proteins are distributed throughout the multilayer. To eliminate the possibility that BSA sorbs as multiple monolayers (as in 4c), we again employed ATR-FTIR, but this time with PEMUs that were thick enough to contain entirely the evanescent wave. In this case, both the solution and the surface protein are not observed, and only material that is “soaked” into the multilayer, closer to the ATR crystal, is seen. Figure 6 provides a rough picture of the probing length scales. The evanescent wave electric field strength, E, decays exponentially away from the crystal with the penetration depth (dp). For the present system, which uses a Ge ATR crystal, dp is about 0.6 µm at 1033 cm-1 (sulfonate stretch ν(SO3-),40 which corresponds to about half the wet thickness of (PDADMA/PSS)15 @1 M NaCl (as measured by AFM).12 For this thickness of multilayer, no further sulfonate is observed as additional PSS layers are added (see Supporting Information). Protein adsorption experiments were conducted using BSA as the model protein and “thick” PDADMA/PSS multilayers. In the case of (PDADMA/PSS)14PDADMA, where the surface charge is positive, the characteristic bands for BSA were detected. In (PDADMA/PSS)15, where the multilayer surface charge is negative, no protein peaks were detected (Supporting Information). For both (PDADMA/PSS)14PDADMA and (PDADMA/PSS)15, the wet film thickness was about 1.2 µm. Also relevant is the observation that on “thin” negatively charged PEMUs (e.g., four layers), we can clearly detect the protein peaks as compared to the thick negatively charged surfaces. This experiment provides further support for model d over c in Figure 4. The rate that protein bands increased for thick, positive PEMUs allowed us to estimate diffusion coefficients. Protein diffusion coefficients have been reported for proteins adsorbed on, or embedded in, multilayers.41 These lateral diffusion coefficients, D, were close to our findings for diffusion of BSA through PDADMA/PSS (D ) 2.1 × 10-10 cm2/s, Supporting Information), which is unequivocally diffusion through the bulk polyelectrolyte complex. Salt-Induced Desorption of Proteins. The amount of protein sorbed is, to a great extent, under the reversible control of salt concentration. Not only does the ionic strength limit the initial amounts of sorbed protein (Table 2) but also additional salt actually removes protein from PEMUs loaded

Figure 7. Steady-state desorption of sorbed amount, Γ, of BSA from (PDADMA/PSS)13PDADMA @0.25 M NaCl following exposure to salt with different concentrations. The initial coverage was 34 mg/m2.

with protein at a low salt concentration. This is demonstrated in Figure 7, which depicts the release of BSA from a (PDADMA/PSS)13PDADMA @0.25 M NaCl multilayer that had initially been loaded with BSA at 1.0 mg/mL from TRIS buffer of 1 mM ionic strength. Increasingly higher salt concentrations lead to progressive release of BSA, until 85% of the original sorbed protein is removed (with much of the residual perhaps adsorbed by nonelectrostatic forces). Release of proteins, under the influence of salt, from monolayers of polyelectrolyte, such as PM2VP,42 has been rationalized in terms of screening of surface-segment attraction. For protein adsorbed into the bulk of a multilayer, as seen here, the salt effect can be thought of as an anion exchange process, where anions from solution displace (net) negative BSA from the multilayer, as depicted in Figure 8, with the process being reversible. Considering BSA as a polyelectrolyte, reversible complexation of polyelectrolytes is not typical behavior, unless the interaction energy between oppositely charged segments is weak.43 Hydrophilic Repelling Diblock. As shown previously, PEMUs can be designed to minimize protein adsorption by proper selection of surface charge, but no system was completely effective at defeating adsorption altogether (as is generally observed for all protein “resistant” technologies). To further reduce adsorption, it is necessary to introduce mechanisms for minimizing nonelectrostatic interactions. Poly(ethylene oxide) (PEO) or poly(ethylene glycol) is extensively promoted for decreasing protein adsorption.44-49 The mechanism has variously been explained with arguments based on hydrophilicity, steric repulsion, and excluded-

Biomacromolecules, Vol. 5, No. 3, 2004 1095

Protein Adsorption Modalities

Figure 8. Ion-exchange-type displacement, where salt counterions replace adsorbed protein in high ionic strength buffers. Table 3. Hydrophilic Repulsion Effect on Proteins Adsorbed from TRIS Buffer, Amounts in mg/m2 multilayer

fibrinogena

lysozymea

BSAa

BSAb

(PM2VP/PSS)2PM2VP (PM2VP/PSS)2PM2VP-b-PEO

3.7 ( 0.5

0.26 ( 0.2

1.6 ( 0.2

7.5 ( 0.2

1.9 ( 0.5

0.23 ( 0.2

1.1 ( 0.2

5.7 ( 0.2

a

Adsorption from 154 mM TRIS. b Adsorption from 1 mM TRIS.

Table 4. Protein Adsorption onto (PAH/PAA)3 Multilayers @0.25 M NaCl (thickness 175 ( 15 Å) from 154 mM TRIS multilayer

surface charge

protein

Γ, mg/m2

(PAH/PAA)3 (PAH/PAA)3 (PAH/PAA)3

-ve -ve -ve

fibrinogen (-) BSA (-) lysozyme (+)

0.34 ( 0.1a 0.08 ( 0.02b 0.82 ( 0.1a

a

Measured with ATR-FTIR spectoscopy. b Measured with OWLS.

volume effects of PEO segments.50-53 Here, we used a copolymer that comprises a positively charged block and a PEO block (Figure 1).The PEO block is neutral and hydrophilic and will exhibit “hydrophilic repulsion”54 while the PM2VP block provides a charged “handle” for multilayer associations. A comparison was made between PM2VP homopolymer and PM2VP-b-PEO incorporated into multilayers, with PSS serving as the polyanion in both cases. The three comparison proteins were allowed to adsorb onto thin PM2VP/PSS @0.25 M multilayers, terminated with either PM2VP or PM2VP-b-PEO. As seen in Table 3, for the negatively charged proteins, BSA, and fibrinogen, there is a decrease in proteins adsorbed onto the PEO surfaces although the effect is not dramatic. Electrostatic repulsion dominates for lysozyme, which is positively charged. The BSA adsorption results in Table 3 for different ionic strengths demonstrate that electrostatic and hydrophilic repulsion forces can be designed into a multilayer to be somewhat additive. The copolymer can actually be used in multilayer buildup to produce thicker multilayers (44% thicker) than PM2VP/PSS (see Supporting Information). Hydrophilic Homopolyelectrolytes. It is generally accepted and recently confirmed that more hydrophobic surfaces induce greater protein adsorption.55,56 We have estimated relative hydrophobicities by comparing the relative degrees of swelling of PEMUs in water.12 PDADMA/PSS is believed to be of “intermediate” hydrophobicity, whereas PDADMA/PAA is hydrophilic when ionized. PAH/PAA multilayers constructed at low pH have been shown to be bioinert toward cell attachment.24 As can be seen from Table 4, PAA as an outer layer in (PAH/PAA)x is particularly effective at suppressing both BSA and fibrinogen adsorption and, surprisingly, lysozyme adsorption (protein has the

Table 5. Hydrophilicity/Hydrophobicity Gradient for Fibrinogen Adsorption surface charge

Γ, mg/m2

(PAH/PAA)3

-ve

0.34 ( 0.1a 0.26 ( 0.05b

PM2VP/PSS/PM2VP PAA/PAH/PAA (PM2VP/PSS)3

-ve

0.41 ( 0.1a

-ve

3.4 ( 0.5a

multilayer

a

Measured with ATR-FTIR spectroscopy. b Measured with OWLS.

opposite charge to PAA). For comparison, the adsorption of lysozyme on (PM2VP/PSS)3 @0.25 M NaCl @pH 7.4 under the same conditions was 8.8 mg m-2. The effectiveness of PAA in suppressing the adsorption of an oppositely charged protein is possibly due to excluded volume effects of the highly hydrated PAA segments, a property which makes the “binding energy” between a PAA segment and a positive protein segment minimal. We were interested in seeing whether the effectiveness of a protein-repelling multilayer comprising PAA could be accomplished by limiting the PAA to the surface. Thus, multilayers having a composition gradient and, therefore, a hydrophilicity gradient were prepared starting with PM2VP/ PSS (relatively hydrophobic) and ending with PAA (relatively hydrophilic). Fibrinogen was used in this experiment because it is relatively large and used as a model for “sticky” serum proteins.57 A composition gradient would be desirable where PEMU durability is an issue because hydrophobic polyelectrolytes form less swollen and more resilient films. One would preserve durability but maintain surface repellency, with the hydrophilic polyelectrolyte limited to the outer, or outer few, layers. Also, using a specialized polymer as the outer layer only would help conserve a potentially costly material. Table 5 shows that a PEMU with a hydrophobic bulk (PM2VP/PSS), capped with a hydrophilic surface (PAH/ PAA), shows no significant change in fibrinogen adsorption compared to a PEMU of uniform hydrophilic composition (PAH/PAA). A surface layer is, thus, able to “mask” protein adsorption properties of bulk PEMU material. This property is also seen in other experiments comparing protein adsorption on PEMUs made from PM2VP-b-PEO and PSS, with those made from PM2VP and PSS and capped with one layer of PM2VP-b-PEO. The economical use of one layer only of diblock copolymer proved as effective as making the entire PEMU from the diblock. Also shown for comparison are the fibrinogen adsorption results obtained from OWLS. The results were obtained using a bilayer model described by Picart et al.,31 and they were close to protein adsorption amounts using the ATR-FTIR method.

1096

Biomacromolecules, Vol. 5, No. 3, 2004

Conclusion In condensing the rather large number of possible combinations of PEMU, protein, and conditions, we focused on a manageable selection that represented “typical”, or readily accessible, materials. All proteins are rather “sticky”: capable of adsorbing via electrostatic (including “patch charge” type adsorption), hydrogen bonding, and “hydrophobic” interactions. Surfaces of opposite charge to that of the protein were found to be more effective at promoting protein adsorption. Where such electrostatic interaction dominates, increased ionic strength was shown generally to decrease protein adsorption. Also evident was the partial effectiveness of a hydrophilic neutral block (PEO) at preventing access to the charged surface beneath it. For multilayer surfaces of opposite charge, the thickness of the PEMU was shown to play an important role in the adsorption process because the PEMU can act as a “sponge” or a matrix to load proteins, an alternative to protein/polyelectrolyte layer-by-layer assembly. Direct ATR-FTIR measurement on thicker PEMUs confirmed protein penetration into the multilayer matrix for oppositely charged surfaces, whereas AFM revealed islands of surface aggregate for like-charged multilayers. On exposure to a solution of higher ionic strength, sorbed protein could be released by an ion-exchange-type mechanism, which could prove useful for protein separations and purification. Acknowledgment. This work was supported in part by a grant from the National Science Foundation (DMR 9727717). Acknowledgement is made to the Donors of the American Chemical Society Petroleum Research Fund for partial support of this research. We are grateful to C. Picart for help with waveguide experiment calculations. Supporting Information Available. Calibration curves for proteins, protein adsorption and desorption kinetics, layerby-layer buildup curves, ATR-FTIR spectra. This material is available free of charge via the Internet at http:// pubs.acs.org. References and Notes (1) Decher, G., Schlenoff, J. B. Multilayer Thin Films - Sequential Assembly of Nanocomposite Materials; Wiley-VCH: Weinheim, Germany, 2003. (2) Decher, G. Science 1997, 277, 1232-1237. (3) Lvov, Y.; Mo¨hwald, H. Protein Architecture: Interfacial Molecular Assembly and Immobilization Biotechnology; Marcel Dekker: New York, 2000. (4) Pei, R. J.; Cui, X. Q.; Yang, X. R.; Wang, E. K. Biomacromolecules 2001, 2, 463-468. (5) Caruso, F.; Niikura, K.; Furlong, D. N.; Okahata, Y. Langmuir 1997, 13, 3427-3433. (6) Lvov, Y.; Ariga, K.; Ichinose, I.; Kunitake, T. J. Am. Chem. Soc. 1995, 117, 6117-6123. (7) Ichinose, I.; Takaki, R.; Kuroiwa, K.; Kunitake, T. Langmuir 2003, 19, 3883-3888. (8) Lahav, M.; Kharitonov, A. B.; Katz, O.; Kunitake, T.; Willner, I. Anal. Chem. 2001, 73, 720-723. (9) Tieke, B.; van Ackern, F.; Krasemann, L.; Toutianoush, A. Eur. Phys. J. E 2001, 5, 29-39. (10) Onda, M.; Lvov, Y.; Ariga, K.; Kunitake, T. Biotechnol. Bioeng. 1996, 51, 163-167. (11) Lvov, Y.; Caruso, F. Anal. Chem. 2001, 73, 4212-4217. (12) Dubas, S. T.; Schlenoff, J. B. Langmuir 2001, 17, 7725-7727. (13) Schwinte, P.; Voegel, J. C.; Picart, C.; Haikel, Y.; Schaaf, P.; Szalontai, B. J. Phys. Chem. B 2001, 105, 11906-11916. (14) Horbett, T. A. ACS Symp. Ser. 1995, 602, 1-23. (15) Elbert, D. L.; Hubbell, J. A. Annu. ReV. Mater. Sci. 1996, 26, 365394.

Salloum and Schlenoff (16) Szleifer, I. Curr. Opin. Solid State Mater. Sci. 1997, 2, 337-344. (17) Mu¨ller, M.; Rieser, T.; Ko¨the, M.; Kessler, B.; Brissova, M.; Lunkwitz, K. Macromol. Symp. 1999, 145, 149-159. (18) Mu¨ller, M.; Brissova, M.; Rieser, T.; Powers, A. C.; Lunkwitz, K. Mater. Sci. Eng., C 1999, 8-9, 163-169. (19) Mu¨ller, M.; Rieser, T.; Lunkwitz, K.; Meier-Haack, J. Macromol. Rapid Commun. 1999, 20, 607-611. (20) Ladam, G.; Gergely, C.; Senger, B.; Decher, G.; Voegel, J. C.; Schaaf, P.; Cuisinier, F. J. G. Biomacromolecules 2000, 1, 674-687. (21) Ladam, G.; Schaaf, P.; Cuisinier, F. J. G.; Decher, G.; Voegel, J. C. Langmuir 2001, 17, 878-882. (22) Mu¨ller, M.; Rieser, T.; Dubin, P. L.; Lunkwitz, K. Macromol. Rapid Commun. 2001, 22, 390-395. (23) Richert, L.; Lavalle, P.; Vautier, D.; Senger, B.; Stoltz, J. F.; Schaaf, P.; Voegel, J. C.; Picart, C. Biomacromolecules 2002, 3, 1170-1178. (24) Mendelsohn, J. D.; Yang, S. Y.; Hiller, J.; Hochbaum, A. I.; Rubner, M. F. Biomacromolecules 2003, 4, 96-106. (25) Haynes, C. A.; Norde, W. Colloids Surf., B 1994, 2, 517-603. (26) Chittur, K. K. Biomaterials 1998, 19, 357-369. (27) Fritz, M.; Radmacher, M.; Cleveland, J. P.; Allersma, M. W.; Stewart, R. J.; Gieselmann, R.; Janmey, P.; Schmidt, C. F.; Hansma, P. K. Langmuir 1995, 11, 3529-3535. (28) Ramsden, J. J.; Prenosil, J. E. J. Phys. Chem. 1994, 98, 5376-5381. (29) Ramsden, J. J.; Lvov, Y. M.; Decher, G. Thin Solid Films 1995, 254, 246-251. (30) Chluba, J.; Voegel, J. C.; Decher, G.; Erbacher, P.; Schaaf, P.; Ogier, J. Biomacromolecules 2001, 2, 800-805. (31) Picart, C.; Ladam, G.; Senger, B.; Voegel, J. C.; Schaaf, P.; Cuisinier, F. J. G.; Gergely, C. J. Chem. Phys. 2001, 115, 1086-1094. (32) Blaakmeer, J.; Bohmer, M. R.; Stuart, M. A. C.; Fleer, G. J. Macromolecules 1990, 23, 2301-2309. (33) Ochiai, H.; Anabuki, Y.; Kojima, O.; Tominaga, K.; Murakami, I. J. Polym. Sci., Part B: Polym. Phys. 1990, 28, 233-240. (34) Graul, T. W.; Schlenoff, J. B. Anal. Chem. 1999, 71, 4007-4013. (35) McCormick, M.; Smith, R. N.; Graf, R.; Barrett, C. J.; Reven, L.; Spiess, H. W. Macromolecules 2003, 36, 3616-3625. (36) Fink, D. J.; Hutson, T. B.; Chittur, K. K.; Gendreau, R. M. Anal. Biochem. 1987, 26, 147-154. (37) Dubas, S. T.; Schlenoff, J. B. Macromolecules 1999, 32, 8153-8160. (38) Sun, Y. P.; Zhang, X.; Sun, C. Q.; Wang, B.; Shen, J. C. Macromol. Chem. Phys. 1996, 197, 147-153. (39) Putnam, F. W. The Plasma Proteins: Structure, Function and Genetic Control, 2nd ed.; Academic Press: New York, 1975; Vol. 1. (40) The dry thickness of (PDADMAC/PSS)15 @1 M NaCl was determined to be 0.63 µm. The thickness in water doubles. See ref 12 for further details. (41) Szyk, L.; Schaaf, P.; Gergely, C.; Voegel, J. C.; Tinland, B. Langmuir 2001, 17, 6248-6253. (42) Sukhishvili, S. A.; Granick, S. J. Chem. Phys. 1999, 110, 1015310161. (43) Dubas, S. T.; Schlenoff, J. B. Macromolecules 2001, 34, 3736-3740. (44) Morra, M. J. Biomat. Sci., Polym. Ed. 2000, 11, 547-569. (45) Llanos, G. R.; Sefton, M. V. J. Biomat. Sci., Polym. E. 1993, 4, 381400. (46) Leckband, D.; Sheth, S.; Halperin, A. J. Biomat. Sci., Polym. E. 1999, 10, 1125-1147. (47) Lee, J. H.; Lee, H. B.; Andrade, J. D. Prog. Polym. Sci. 1995, 20, 1043-1079. (48) Hoffman, A. S. J. Biomat. Sci., Polym. Ed. 1999, 10, 1011-1014. (49) Harris, J. M. Poly(ethylene glycol) Chemistry: Biotechnical and Biomedical Applications; Plenum Press: New York, 1992. (50) Jeon, S. I.; Lee, J. H.; Andrade, J. D.; Degennes, P. G. J. Colloid Interface Sci. 1991, 142, 149-158. (51) Israelachvili, J. N. Intermolecular and Surface Forces, 2nd ed.; Academic Press: London, 1991. (52) Fleer, G. J.; et al. Polymers at Interfaces, 1st ed.; Chapman & Hall: London, 1993. (53) Malmsten, M.; Van Alstine, J. M. J. Colloid. Interface Sci. 1996, 177, 502-512. (54) van Oss, C. J. J. Mol. Recognit. 2003, 16, 177-190. (55) Elwing, H.; Askenda, A.; Ivarsson, B.; Nilsson, U.; Welin, S.; Lundstrom, I. ACS Symp. Ser. 1987, 343, 468-488. (56) Prime, K. L.; Whitesides, G. M. Science 1991, 252, 1164-1167. (57) Mrksich, M.; Sigal, G. B.; Whitesides, G. M. Langmuir 1995, 11, 4383-4385.

BM034522T