|Home | About | Journals | Submit | Contact Us | Français|
Numerous strategies exist to prevent biological fouling of surfaces in physiological environments; our strategy focuses on the modification of surfaces with poly-N-substituted glycine oligomers (polypeptoids). We previously reported the synthesis and characterization of three novel polypeptoid polymers that can be used to modify titanium oxide surfaces, rendering the surfaces resistant to adsorption of proteins, to adhesion of mammalian and bacterial cells and to degradation by common protease enzymes. In this study, we investigated the effect of polypeptoid chain length on the antifouling properties of the modified surfaces. For these experiments we used poly(N-methoxyethyl) glycines with lengths between ten and fifty repeat units and determined the influence of chain length on coating thickness and density as well as resistance to protein adsorption and cellular adhesion. Short-term protein resistance remained low for all polymers, as measured by optical waveguide lightmode spectroscopy, while fibroblast adhesion after several weeks indicated reduced fouling resistance for the polypeptoid-modified surfaces with the shortest chain length polymer. Experimental observations were compared to predictions obtained from a molecular theory of polymer and protein adsorption. Good agreement was found between experiment and theory for the chain length dependence of peptoid grafting density, and for protein adsorption as a function of peptoid grafting density. The theoretical predictions provide specific guidelines for the surface coverage for each molecular weight for optimal antifouling. The predictions show the relationship between polymer layer structure and fouling.
Inhibition of protein adsorption and cellular adhesion to surfaces is critical for proper functioning of medical devices such as cardiovascular implants,1–3 stents,4 and catheters.5, 6 While a variety of approaches exist for rendering surfaces resistant to fouling, immobilization of poly(ethylene glycol) (PEG) polymers is the most commonly explored method.7–17 In order to avoid some of the reported problems associated with PEG, such as chemical or enzymatic degradation which limit long-term use in physiological environments,18–22 researchers have developed alternative antifouling coatings including zwitterionic polymers,23–29 polysaccharides,30–34 and peptide-mimetics.35–38 Our strategy focuses on the use of poly-N-substituted glycine oligomers or polypeptoids anchored onto surfaces through a mussel adhesive-inspired peptide.39, 40 A systematic investigation of antifouling polypeptoid chain length has not previously been reported; this work should allow for interesting comparisons to be drawn between polypeptoids and numerous PEG systems.7–17 Grafted PEG chains have been shown to exhibit increased protein resistance with increasing chain lengths. The increased fouling resistance of longer chains is explained as being due to greater excluded volumes, entropy and steric repulsion.15 It is not clear whether these results reflect the ability of the PEG layers to thermodynamically or kinetically control protein adsorption. As it has been shown the role of molecular weight is very important in kinetic control, however, it plays only a secondary role in the thermodynamic control41–43. A similar trend is expected for polypeptoid polymers, although the chain length threshold for fouling resistance is unknown.
In this work, we investigated the influence of polypeptoid chain length on antifouling properties and further characterized the modified surfaces. The polymer thickness and surface density were compared to previous results on PEG-based systems in an attempt to better understand the packing of polypeptoid chains on surfaces. Although short-term protein fouling was found to be independent of peptoid chain length, the shortest chain length peptoid became fouled with fibroblast cells within days whereas the longer peptoid chains remained cell-free for several weeks. These findings are explained with the help of theoretical predictions for the adsorption isotherms of proteins on peptoid modified surfaces. The molecular theory provides a link between the adsorption reduction and the structure of the polypeptoid layer and suggest that in the case of very short chain length the fouling by cells is due to a kinetic effect on protein adsorption, while the surface coverage for longer peptoids is enough for the full thermodynamic protection of the surface from protein adsorption.
Methoxyethylamine, triisopropylsilane (TIS), dimethylformamide (DMF), acetonitrile (ACN), N-morpholinopropanesufonic acid (MOPS) buffer salt, fibrinogen, 4-(2-hydroxyethyl)piperazine-1-ethanesulfonic acid (HEPES) buffer salt, Tris buffer salt, sodium tetraborate, 2-propanol, 1,1’-dioctadecyl-3,3,3’,3’tetramethylindocarbocyanine perchlorate (DiI) and alpha-cyano-4-hydroxycinnamic acid (CHCA) matrix were purchased from Aldrich (Milwaukee, WI). Rink amide-MBHA resin LL, Fmoc-Lys(Boc)-OH, Fmoc-Tyr(tBu)-OH and Fmoc-DOPA(acetonide)-OH were purchased from Novabiochem (San Diego, CA). Acetic anhydride and N-methylpyrrolidone (NMP) were purchased from Applied Biosystems (Foster City, CA). Trifluoroacetic acid (TFA) was obtained from Fisher Scientific (Pittsburgh, PA). Silicon wafers were purchased from University Wafer (South Boston, MA). Lyophilized whole human serum (Control Serum N) was purchased from Roche Diagnostics (Indianapolis, IN). Calcein-AM was purchased from Molecular Probes (Eugene, OR). 3T3-Swiss albino fibroblasts, Dulbecco’s modified Eagle’s medium (DMEM), fetal bovine serum (FBS) and penicillin/streptomycin were obtained from American Type Culture Collection (Manassas, VA). Ultrapure water (U.P. H2O) used for all experiments was purified (resistivity ≥ 18.2 MΩ-cm, total organic content ≤ 5 ppb) with a NANOpure Infinity System from Barnstead/Thermolyne Corp. (Dubuque, IA).
The peptidomimetic polymers were synthesized as described previously,36, 37 using a C S Bio 036 (C S Bio Co., Menlo Park, CA) automated peptide synthesizer. The five polymers of varying chain lengths were synthesized in one batch by removing portions of the resin from the reaction vessel at the appropriate coupling step. The C-terminal DOPA-Lys-DOPA-Lys-DOPA peptide anchor was first synthesized on a low loading rink amide resin using conventional Fmoc strategy of solid-phase peptide synthesis; the polypeptoid portion was then synthesized using a submonomer protocol.44 The Tyr polymer analog was synthesized according to the same procedure but using Fmoc-Tyr(tBu)-OH instead of Fmoc-DOPA(acetonide)-OH. Acetic anhydride was used to acetylate the N-terminus of the polypeptoid chain upon removal from the vessel; cleavage of the polymers from the resin and deprotection of the amino acid side chains was accomplished by treating the resin with 95% (v/v) TFA, 2.5% H2O and 2.5% TIS for 20 minutes. The cleaved polymer was then removed by filtering and rinsing several times with TFA, and solvent was removed using a rotary evaporator; the oily product was dissolved in 50/50% water/acetonitrile, frozen and lyophilized. The crude products were purified by preparative reversed-phase high performance liquid chromatography (RP-HPLC) (Waters, Milford, MA) using a Vydac C18 column, and purified fractions were frozen and lyophilized. The purity of each final product was confirmed by RP-HPLC and matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS) (Voyager DE-Pro, Perspective Biosystem, MA).
Silicon wafers were coated with a 20 nm-thick layer of TiO2 by electron beam evaporation (Edwards Auto306; <10−5 Torr), and the coated wafers were cut into 8 mm by 8 mm pieces. The substrates were cleaned ultrasonically for ten minutes in 2-propanol and dried under N2. Surfaces were then exposed to O2 plasma (Harrick Scientific, Ossining, NY) at ≤150 Torr and 100 W for three minutes. OWLS waveguides were purchased from Micro Vacuum Ltd. (Budapest, Hungary) and coated with a 10 nm-thick layer of TiO2 by electron beam evaporation as described above. Sensors were cleaned following the same procedure as TiO2 substrates. After use, OWLS waveguides were regenerated for subsequent use by 10-minute sonication cycles in 0.1 M HCl, U.P. H2O and 2-propanol followed by exposure to O2 plasma to remove adsorbates.
Unless otherwise noted, the general approach used for surface modification involved immersion of clean substrates and sensors in a 0.3 mM solution of peptidomimetic polymer in Buffer A (3 M NaCl buffered with 0.1 M MOPS, pH = 6) at 50°C for 24 hours. After modification, substrates were extensively rinsed with U.P. H2O to remove any unbound polymer, and then dried in a stream of filtered N2.
Prior to modification, substrates were cleaned as described above and measured using a M-2000 spectroscopic ellipsometer (J.A. Woollam, Lincoln, NE). Measurements were made at 65°, 70° and 75° using wavelengths from 193–1000 nm. After modification, substrates were rinsed and dried as described above and measured again. The spectra were fit with multilayer models in the WVASE32 software (J.A. Woollam). Optical properties of the substrate were fit using a standard TiO2 model, while properties of the polymer layer were fit using a Cauchy model (An = 1.45, Bn = 0.01, Cn = 0).45 The obtained ellipsometric thicknesses represent the “dry” thickness of the polymer under ambient conditions. The average thickness and standard deviation of three or more substrates is reported for each polymer.
For in situ polymer adsorption experiments, TiO2 coated waveguide sensors were cleaned and inserted into the measurement head of an OWLS110 (Micro Vacuum Ltd.) and exposed under static conditions to Buffer A through the flow-through cell (16 µL volume) for at least 24 hours to allow for equilibration. The measurement head was mounted in the sample chamber and heated to 50°C; the signal was recorded to ensure a stable baseline. Polymer solution (1 mL total volume) was injected into the flow-through cell in stop-flow mode. The waveguide sensor was exposed to the polymer solution for 4 hours, subsequently rinsed with Buffer A (2 mL), and allowed to equilibrate for another 30 minutes. Adsorption experiments with varying polymer concentrations were conducted in U.P. H2O at 25°C using the same procedure.
The measured incoupling angles, αTM and αTE were converted to refractive indices NTM and NTE by the Micro Vacuum software, and changes in refractive index at the sensor surface were converted to adsorbed mass using de Feijter’s formula.46 The refractive indices of solutions were measured using a refractometer (J157 Automatic Refractometer, Rudolph Research) under identical experimental conditions. A refractive index value of 1.35616 was used for Buffer A, and a value of 0.129 cm3/g was used for dn/dc in the polymer-adsorption calculations.
AFM measurements were performed on an Asylum MFP-3D instrument (Asylum Research, Santa Barbara, CA) installed on a Nikon TE2000 microscope. Silicon cantilevers (VISTAprobes, T300) were used for tapping-mode measurements in air and silicon nitride cantilevers (Veecoprobes, DNP-S20) were used for tapping-mode measurements in U.P. H2O.
Survey and high resolution XPS spectra were collected on an Omicron ESCALAB (Omicron, Taunusstein, Germany) configured with a monochromated Al Kα (1486.8 eV) 300-W X-ray source, 1.5 mm circular spot size, a flood gun to counter charging effects, and an ultrahigh vacuum (<10−8 Torr). The takeoff angle was fixed at 45°. Substrates were mounted on standard sample studs using double-sided Cu adhesive tape. Spectra were fitted using CasaXPS software; specifically a Shirley background subtraction and the sum of 90% Gaussian and 10% Lorentzian function were used. Atomic sensitivity factors were used to normalize peak areas from high-resolution spectra to intensity values, which were then used to calculate atomic compositions.47
Lyophilized human serum was reconstituted in water to reach the typical concentration in blood; fibrinogen from human plasma was dissolved at 3 mg/mL concentrations in Buffer B (10 mM HEPES, 150 mM NaCl, pH = 7.4). For in situ protein adsorption experiments, TiO2 coated waveguide sensors were modified with peptidomimetic polymers as explained previously. After equilibration of the OWLS baseline in Buffer B, protein solution was injected and allowed to adsorb for 20 minutes at 37°C before rinsing with Buffer B. A refractive index value of 1.33127 was used for the Buffer B, and a standard value of 0.182 cm3/g was used for dn/dc in the protein-adsorption calculations.15
3T3-Swiss albino fibroblasts were maintained at 37°C and 5% CO2 in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal bovine serum (FBS) and 100 U/ml of penicillin/streptomycin. Immediately before use, fibroblasts of passage 12–16 were harvested using 0.25% trypsin-EDTA, resuspended in DMEM with 10% FBS and counted using a hemacytometer. Modified and unmodified substrates were placed in a 12-well tissue culture polystyrene (TCPS) plate and sterilized by exposure to UV light for 10 minutes, after which 1 ml of DMEM containing FBS was added to each well and incubated for 30 minutes at 37°C and 5% CO2. For the long-term experiment, the fibroblast cell suspension was diluted, and cells were seeded on each substrate at a density of 2.9 × 103 cells/cm2 and substrates were reseeded twice per week. For live cell staining, the medium was aspirated from each well to remove any non-adherent cells and PBS was used to rinse the substrates and wells. Fibroblasts were stained with 2.5 µM calcein-AM in complete PBS for 1 hour at 37°C; substrates were transferred to new culture plates with fresh media and imaged weekly. After imaging, substrates were reseeded and placed back into the incubator; media was changed every three days. For short-term 4 hour assays on varying surface chemistries, cells were plated at a density of 29 × 103 cells/cm2; adherent cells were fixed in 3.7% paraformaldehyde for 5 minutes and stained with 5 µM 1,1’-dioctadecyl-3,3,3’,3’tetramethylindocarbocyanine perchlorate (DiI) for epifluorescent microscope counting.
Quantitative cell attachment data was obtained by acquiring nine images (10× magnification) from random locations on each substrate using a Leica epifluorescent microscope (W. Nuhsbaum Inc., McHenry, IL) equipped with a SPOT RT digital camera (Diagnostics Instruments, Sterling Heights, MI). Three identical substrates for each experiment were analyzed for statistical purposes, resulting in a total of 27 images per time point for each modification. The microscopy images were quantified using thresholding in Metamorph (Molecular Devices, Downingtown, PA).
The theoretical approach that we apply is a molecular theory that has been shown to provide accurate information as compared with experimental observations for the structure and thermodynamics of tethered polymer layers48 as well as in the determination of the amount of protein adsorption on surfaces with grafted PEG. The agreement with experimental observations is for oligomeric chains42 as well as long polymers49 and at all polymer-grafting densities. Thus, we believe that the approach is very appropriate to predict the amount of grafted polypeptoid end-adsorbed as well as the ability of the peptoid to prevent protein adsorption.
We present a simple version of the theory to highlight the main points and refer the reader to references43, 50, 51 for a more detailed discussion and for technical details. The basic idea of the theory is to consider conformations of the chain molecules (the end tethered polypeptoids and the proteins) and by minimization of the system’s free energy determine the probability of each of those conformations depending upon the solution conditions. In this way, the theory enables the study of both structural and thermodynamic properties for the explicit molecular system studied. We first derive the theory to study the amount of peptoid that adsorbs to the surface and then generalize it for the understanding of protein adsorption with surfaces with tethered polypeptoids.
Consider first a surface of total area A, with Ng end-tethered polypeptoids in contact with water. The surface coverage of polymers is defined by σ=Ng/A. The total Helmholtz free energy per unit area of the surface is given by
with β=1/kT where k is the Boltzmann constant and T the absolute temperature. The first term in the free energy represents the conformational entropy of the peptoids, with P(α) being the probability of finding the peptoid in conformation α. The second term is the two–dimensional (surface) translational entropy of the peptoids, with L = (h2 / 2pmkT)1/2 representing the peptoids de-Broglie wavelength, where h is the Planck constant and m the peptoid’s mass. The third term represents the position dependent translational (mixing) entropy of the water (solvent) molecules, with ρw(z) being the position dependent water density and vw is the molecular volume of a water molecule. The fourth and fifth terms represent the attractive and repulsive intermolecular interactions with χ measuring the strength of the peptoid-peptoid attractions and π(z) being the self-consistently determined strength of the repulsions, arising from excluded volume interactions; ϕi(z) represent the position dependent volume fraction of species i=p, h, w for peptoid tail, head and water respectively. The last term in the free energy expression represents the bare adsorption free energy, where Fbinding is the total strength of the free energy of binding between the pentapeptide (DOPA-Lys-DOPA-Lys-DOPA sequence) and the surface. The probability of chain conformations as well as the position dependent solvent density is determined by the minimization of the free energy. This yields
where q is the normalization constant, the first (second) integral in the Boltzmann factor arises from the repulsive (attractive) interactions between the polymer in conformation α and the other molecules in the system. n(α,z)dz represents the number of peptoid segments that the molecule in conformation α has in between z and z+dz, while vh(z) represents the volume that the head-group (pentapeptide) occupies at z. For the solvent volume fraction the minimization yields
The only unknowns are the position dependent repulsive interactions, π(z), which are determined by the constraint that at all distances from the surface, z, polymer or solvent must occupy the total volume. (See references43, 50, 51 for the technical details of how the chain conformations are generated and the equations solved).
To determine the amount of peptoid that end-tethers to the surface we need to determine the surface coverage that corresponds to the same chemical potential as the peptoid in the solution. The chemical potential of the end-tethered molecules is determined by taking the appropriate derivative of the free energy, Eq. (1.1), with respect to the number of peptoids51. This yields
which is readily calculated once the π(z) are determined. The bulk chemical potential is given by
where Mb is the total number of generated chain conformations in bulk and the other quantities have been defined. Note that in Eq. (1.5) ϕi represents the bulk volume fraction of species i. The amount of polymer adsorbed, through the peptapeptide-surface binding attraction Fbinding, is determined for each molecular weight of polypeptoid using the following procedure. The chemical potential of the polymer on the surface is calculated as a function of the surface coverage and the σ for which the chemical potential has the same value as that of the bulk, determined through Eq. (1.5), corresponds to the bound amount.
The adsorption of the protein as a function of the amount of tethered polymers is determined using a generalization of the theory presented above. The details can be found in references43, 50, 51, here we present the basic ideas. We consider a surface with grafted polymers at a surface coverage σ, with σpro adsorbed proteins. The free energy of the system is given by
where the only difference with Eq. (1.1) is the addition of the third and fourth terms which correspond to the protein contributions, including the conformational entropy; the adsorption energy of proteins in conformation γ and the translational entropy of the adsorbed proteins respectively. Ppro(γ) is the probability of finding the adsorbed protein in conformation γ and εads(γ) is the bare adsorption energy of the protein in that conformation. We minimize the free energy and obtain the expressions for the probability distribution function of the polymer and the protein and the volume fraction distribution, f i (z), for each molecular species i. From the minimized free energy the probability of the adsorbed protein conformation is given by
and the adsorbed proteins chemical potential is related to the amount of proteins adsorbed by
The adsorption isotherms, i.e. the amount of proteins adsorbed as a function of the surface coverage of grafted polypeptoids, is obtained by finding at each grafted polymer surface coverage, the amount of adsorbed proteins that has a chemical potential equal to that of the bulk solution, which is given by
The details of how the calculations are carried out can be found in ref. 43, 50, 51. The model for fibrinogen is identical to the one we used in earlier work, which provided good agreement for the adsorption of the proteins on PEG layers41, 43, 49, and which properly accounts for the size and shape of this complex protein. We assume that the protein can adsorb in either on-flat or edge configurations and the bare free energy with the TiO2 surface is −305kBT=−732kJ/mol and −762.5kBT=−1,830kJ/mol, respectively. These values were chosen so as to get the right amount of protein adsorbed in the bare TiO2 surface. The peptoids conformations are generated by using a coarse grained model in which each peptoid is considered within a united atom model and the volume of each peptoid is given by 0.09nm3 while that of the penptapeptide is 0.27nm3. The value of the interaction parameter is χ=−70/T, where T is the absolute temperature. At room temperature, water is a good solvent for the polymer. The peptapeptide is assumed to be on the surface and therefore has no conformational degrees of freedom. For each polypeptoid chain length we generate 106 independent conformations. Each conformation provides a set of volume distributions, n(α,z)vp for all z, that are used in all the calculations. The equations that we solve, correspond to the packing constraints, e.g. for the protein adsorption case this is
The polypeptoid polymers investigated in this study were designed based on the superior performance of PMP1, a peptidomimetic polymer consisting of a DOPA and lysine pentapeptide anchor coupled to 20 N-methoxyethyl glycine residues.36, 37 The chemical structures of the polymers with varying chain lengths (PMP1n) are shown in Fig 1. The importance of the DOPA residues in the pentapeptide was revealed through a polymer analog that contained tyrosine residues in place of the DOPA residues in the pentapeptide anchor coupled to 10 N-methoxyethyl glycine residues (Tyr-PMP110); the structure of this polymer is also shown in Fig. 1. Characterization data (RP-HPLC and MALDI-MS) for the purified polymers is shown in the Supporting Information, Fig. S1–S2.
The polymers were adsorbed onto TiO2 substrates from Buffer A at 50°C, corresponding to the previously reported marginal solvation conditions for PMP120, which allow for greater grafted polymer density by reducing chain repulsion.36, 37, 52, 53 While these adsorption conditions may not be optimal for all other chain lengths in the study, for experimental consistency this modification condition was used for all experiments. The thicknesses of the polymer coatings adsorbed onto the TiO2 surfaces as measured by spectroscopic ellipsometry are reported in Table I. Average thickness values for the polymer-modified surfaces appear to increase with number of peptoid residues, but statistically significant differences (p < 0.02) are detected only between PMP150 substrates and the shorter polymers, suggesting that small increases in chain length do not have a measurable effect on polymer thickness. The Tyr-PMP110 modified substrates had a significantly (p < 0.02) thinner polymer coating compared to all DOPA-containing polymers, suggesting that either the polymer was not able to adsorb to the substrate or the adhesive strength was weak and the polymer was easily displaced by subsequent rinsing and drying steps. Reduction in interaction strength is expected for the tyrosine-containing polymers compared to the DOPA-containing polymers based on AFM experiments which demonstrated that the strong interactions between DOPA and TiO2 surfaces were much higher than the interaction strength of tyrosine and TiO2 surfaces.54
Polymer adsorption was also characterized by in situ OWLS adsorption. This technique allows for highly sensitive (< 0.5 ng/cm2) measurement of mass adsorption per area in a flow-through cell device.55, 56 Average mass results for 3 independent experiments for adsorption of each polymer on TiO2-coated waveguides are shown in Table II. As suggested by ellipsometry results, the mass of the polymer coating increases with increasing chain length and is significantly lower for Tyr-PMP110. These mass adsorption values were used to determine the surface density of polymer chains (σ); using the following equation:
where NA is Avogadro’s number, mA is the measured mass of adsorbed polymer (in g/nm2) and Mw is the molecular weight of the polymer. While longer chain length polymers result in greater mass adsorption, this correlates to decreased surface density of polymer chains on the surface. A decrease in polymer density does not appear to decrease fouling resistance, most likely because the longer chains allow more peptoid units to pack in each area. As the polymer surface coverage increases, a decrease in protein adsorption should be seen due to the steric barrier from the polymer layer that the adsorbing proteins encounter.51
The polypeptoid backbone structure varies significantly from that of PEG; therefore similar calculations for radius of gyration and molecular spacing cannot be made using the same parameters that are reported for PEG in the literature.10, 12, 14, 15 But the polymer surface densities can be compared among different polymer systems in order to attempt to better understand fouling resistance of such coatings. PEG-containing surface coatings investigated by Malmsten et al. ranged in layer thickness from 2 nm to 22 nm and chain density from 0.004 chains/nm2 to 0.12 chains/nm2 for either covalently grafted or strongly adsorbed polymers. 14 The PLL-g-PEG system used on Nb2O5 substrates achieved surface densities of 0.9, 0.5 and 0.3 chains/nm2 for 1, 2, and 5 kDa PEGs.15 Comparisons of PMP1 polymer densities to equivalent mass PEGs suggest that the PMP1 polymers are able to adsorb at nearly twice the density of the PLL-g-PEG system. The surface density of mPEG-DOPA3 neared 50 EG/nm2 (~0.44 chains/nm2 for PEG 5k), with a threshold for protein adsorption demonstrated around 15–20 EG/nm2 (~0.13–0.18 chains/nm2).10 Thus the poor fouling-resistance of the Tyr-PMP110 substrates should be expected with a polymer density of 0.16 chains/nm2. Molecular spacing calculations indicated that the surface-bound mPEG-DOPA3 chains were in a brushlike structure at these high surface densities;10 the structural conformation of the polypeptoid polymers has not been determined, but a similar brush-like structure is expected, as it is shown by the predictions of the theory (see Fig. 5). The surface density results for the polypeptoid polymers are relatively high compared to PEG systems, suggesting that the MAP-inspired anchoring strategy allows for strong attachment to TiO2 and formation of densely packed polymer layers.
To more closely study the adsorption of the polypeptoid polymers to TiO2 substrates, experiments were conducted using a range of polymer concentrations for PMP110 in H2O. Surface density is plotted as a function of polymer concentration in Fig. 2. Results indicate that the density increases with concentration but appears to level off at higher concentrations.
The next step is to compare the predictions from the molecular theory with the measured amount of polypeptoid on the surface, Table II. To this end, Fig. 3 shows the amount of polymer adsorbed as a function of the length of the peptoid. The predictions of the theory are in good agreement with the experimental observations. The trends are identical in both cases, i.e. the amount of polypeptoid adsorbed decreases as the peptoid chain length increases. This result is not unexpected since the anchoring group, the DOPA-Lys pentapeptide, is identical in all cases and therefore the driving force for binding to the surface should be the same. However, as the chain length increases the repulsions between the peptoids increases and thus less polymer can be attached with the same driving force. Interestingly the theory predictions are worst at the shortest chain length indicating that in these cases a more detailed model for the peptoids may be necessary. We are working in that direction, however it is clear that the main trends of the adsorption are nicely captured by the theory.
To better understand the surface architecture of polymer-modified surfaces, tapping mode AFM was used to image PMP150 substrates under dry and aqueous conditions (Fig. 4). The dry substrate had an average RMS roughness of 161 ± 58 pm, and the roughness of the wet substrate was similar (RMS 193 ± 8 pm). These low roughness values and images demonstrate that homogeneous polymer layers can be created using these polypeptoid polymers. Similar roughness values were reported for PEG surfaces on glass.11 To determine the hydrated thickness of the polymer coatings, a scratch was made in the coating using a clean surgical blade, and then the area was imaged using tapping-mode. The change in height between the scratched surface and the polymer coating was assumed to be the thickness of the polymer coating. The experiments were repeated for the same surfaces after drying with N2. The average dry thickness for PMP150 is 5.2 ± 0.3 nm, and the average thickness under aqueous conditions is 10.4 ± 0.8 nm, suggesting a doubling of polymer thickness when hydrated. Using similar AFM imaging methods, Harbers et al. reported a dry film thickness of 10–20 nm and a hydrated film thickness of 50–100 nm for PEG coatings on glass.11 Using the contour length for amino acids (0.34 nm per residue), the calculated maximum thickness for PMP150 is 17 nm. Taking into account this calculated thickness, a hydrated surface thickness of 10 nm is quite plausible.
We can get more detailed information on the structure of the grafted peptoids by looking at the predictions of the molecular theory for the volume fraction profile of the peptoid for the measured surface coverage of polymers. This is shown in Fig. 5 for the five different polymers at the experimentally measured surface coverages. The peptoid segment profile shows a depletion region very close to the surface due to the presence of the adsorbed pentapeptide. The profiles show a stretched configuration with high local densities, due to the bulky nature of the peptoid side groups. Interestingly, the local volume fraction in the region of maximal density is higher for the short polypeptoids. However, the films are thicker as the chain length increases. One can understand these two features, by recalling that the strength of the binding to the surface is the same for all the different polymers. Therefore, the total repulsion between the peptoids at equilibrium, needs to be the same. This is achieved by getting more polymers of short chain length with higher local segmental density. Note that this is a simple explanation, since the full calculation also accounts for the loss of conformational entropy, which is a strong function of chain length. Interestingly, the structure of the PMP150 predicted by the theory corresponds to a thickness slightly larger than 10nm, in excellent agreement with the AFM observations. As we will discuss below, the structure of the different polymer layers is directly responsible for the non-fouling capabilities of the film.
OWLS was also used for short-term protein adsorption experiments with fibrinogen and human serum; results are shown in Table III. Protein adsorption values on unmodified TiO2 sensors were comparable to published OWLS data for serum12 and fibrinogen;57 protein adsorption results on all PMP1-modified sensors were significantly lower (p < 0.05) than unmodified TiO2, but no statistically significant differences were observed between the PMP1-modified surfaces. The adsorbed serum masses for all PMP1-modified surfaces are similar to values for PEG coatings8, 10, 12, 17 and other PMP coatings.36 Fibrinogen adsorption on the polymer-modified substrates is near to the ~5 ng/cm2 threshold below which activation of the pathways for blood coagulation does not occur;12 suggesting possible use of the coatings for blood contacting applications where prevention of thrombosis is desired. Serum and fibrinogen adsorption values on the Tyr-PMP110-modified sensors were significantly greater than the values for PMP1-modified sensors, as was expected based on the thinner polymer coating. PMP1 chain lengths of 10 to 50 repeats are equally suited for short-term protein resistant surfaces, but further experiments are necessary to predict long-term resistance. According to Szleifer and coworkers, increasing the polymer chain length slows the adsorption kinetics for proteins;41 thus protein adsorption on the different polymers may not be seen until much longer time scales (months to years).
The experimental observations for each polypeptoid length correspond to one value of the polymer surface coverage. To obtain a better understanding of how the polymers prevent protein adsorption we present calculations for the whole adsorption isotherms for each polypeptoid length. Fig. 6 displays the amount of fibrinogen adsorbed as a function of the amount of polypeptoid on the surface. For each chain length of polypeptoid, increasing the polymer surface coverage reduces the amount of protein adsorbed. Furthermore, for each polypeptoid length there is an amount of grafted polymer above which there is no more protein adsorption. The critical surface coverage for non-fouling decreases as the polymer chain length increases in the range of chain length used in the experimental observations, i.e. up to PMP150. We have also included in the figure the predictions for PMP1100, which has not been studied experimentally. The ability of this long polymer to prevent protein adsorption is almost identical to that of PMP150. This is very much in line with previous predictions for model PEG that predict that once the polymer film thickness is larger than the size of the protein the polymer chain length has no effect on the equilibrium adsorption isotherm50, however it may have a very strong effect on the kinetics. As can be seen in Figure 5, PMP130 forms a layer of thickness similar to the largest domain of fibrinogen, 6.5nm. Thus, we expect that for polypeptoids with more than around 30 to 40 peptoids, the adsorption isotherm will be only weakly dependent on polymer size, as shown in Figure 6.
In order to show the differences between polypeptoids and PEG, Figure 6 shows the adsorption of fibrinogen as a function of surface coverage also for PEG-2000 and PEG-5000. In both cases the ability of PEG to prevent protein adsorption at all surface coverages is less efficient than the polypetoids with more than 20 units. Moreover, the predicted threshold for complete prevention of protein adsorption is above the surface coverage achieved with PLL-g-PEG of 0.5 and 0.3 chains/nm2 for PEG-2K and 5K respectively. Thus, we expect the non-fouling capabilities of these PEG layers to be time dependent.
All these results can be explained in terms of the ability of the flexible polypeptoid to reduce and reject protein adsorption due to the effective steric interactions that result from the excluded volume repulsions combined with the reduction of the available number of polymer conformations when proteins adsorb. This also explains the more effective capabilities of the polypeptoids as compared to PEG, for the same number of units, since the peptoids are bulkier than the ethylene oxide units, and therefore offer more steric repulsion on a per unit basis.
It is important to emphasize that the calculations represent equilibrium predictions. Namely, the theory assumes that the system has reached the thermodynamic preferred state of the system. However, it is not clear that the experimental observations, which are carried out over a relatively short time reach the equilibrium state. Having this consideration, we see that for all the experimental polypeptoids, except PMP110, the amount of polymer bound to the surface is larger than what is necessary for the complete thermodynamic prevention of protein adsorption, and indeed the measured amount of protein adsorption (shown as symbols in Fig. 6) is close to zero.
The case of PMP110 is clearly different. The predicted isotherm shows that at the surface density of the experiment there will be a finite amount of adsorbed fibrinogen at equilibrium. However, the experimental observations show no adsorption. We return to this point later when we discuss cell adhesion on PMP110 covered surfaces.
An important prediction from the theory is the chain length dependence of the threshold surface coverage to completely prevent protein adsorption. Figure 7 shows the threshold surface coverage as a function of the film thickness, with the inset showing the film thickness as a function of polymer length. The thickness is calculated as H = 2 ∫ ϕpol(z)z dz/∫ϕpol(z)dz, which corresponds to the first moment of the polymer volume fraction profile. The thickness at the threshold surface coverage is linear with the polypeptoid chain length. Furthermore, the surface coverage threshold decreases very sharply for small thickness, but it levels off once the film thickness of the order of the protein size, in agreement with earlier predictions for PEG; 50. These results show that molecular weight dependence for the equilibrium amount of protein adsorption is not very large once the polymer thickness is larger than the size of the protein, however one needs to keep in mind that this weak dependence does not hold for the kinetics of protein adsorption, where both surface coverage and molecular weight play very important roles41.
For cell adhesion studies, bare TiO2 and polymer-modified TiO2 substrates were seeded twice weekly with fresh 3T3 fibroblasts suspended in serum containing media. Cell attachment was quantified weekly for up to seven weeks by live cell staining, fluorescence microscopy and image analysis (Fig. 8). At the initial 4-hour time point, fibroblasts attached readily to the unmodified TiO2 substrates and the Tyr-PMP110-modified substrates, while all PMP1n-modified substrates were highly resistant to adhesion. Fibroblasts formed confluent monolayers on the unmodified TiO2 substrates and the Tyr-PMP110-modified substrates by day 14. The PMP110-modified substrates initially exhibited low levels of cell attachment but demonstrated a steady increase in adhesion, reaching confluent monolayers by week 7. All other PMP1n-modified substrates remained highly resistant to fibroblast adhesion throughout the experiment, which was expected for the 20mer and longer chain lengths based on our previously reported 5-month study.37
The apparent threshold for long-term (7 weeks) fouling resistance is around 15 peptoid repeat units, which correlates to a dry polymer thickness of 32.5 ± 4.3 Å from ellipsometry experiments and an adsorbed polymer mass of 412.0 ± 67.6 ng/cm2 from OWLS experiments. We can infer from these results that protein adsorption on polypeptoid polymers is increasing on time scales of months to years; therefore if the 10mer substrate was fouled by cells (and assumingly proteins) within 7 weeks in in vitro culture, we could predict that the 15mer polymer would become fouled next, followed subsequently by the longer chain lengths. However, the theoretical predictions, see below, show that for the experimental surface coverages the non-fouling capability of the surfaces with 15mer polymer (and longer) are time independent, i.e. equilibrium. Importantly, our cell adhesion results also support the previous claim that short-term (minutes to hours) protein adsorption experiments may not always be the most reliable predictors of long-term (days, months, years) biofouling events.36, 41
We can make a more accurate prediction of protein (and cell) surface fouling by using the theoretical results presented in Fig. 6. According to the predictions only PMP110 is tethered at a surface coverage lower than the minimal needed for thermodynamic prevention of protein adsorption. Therefore, that surface should foul, as demonstrated in the experiments. The time scale for fouling can be determined as shown in earlier work41, and the time scale of days is not surprising. The predictions in Fig. 6 also show that none of the other surfaces should foul within the timeframe of the current experiments since they are in the regime where thermodynamic control of protein adsorption is achieved. We thus believe that in those cases biofouling will not occur even in very long time scales.
Polypeptoid polymers, composed of a peptide anchor coupled to N-methoxyethyl glycines of varying repeat lengths, were determined to have significant antifouling properties when immobilized onto TiO2 substrates. A minimum chain length of 15 peptoids was demonstrated to be necessary for long-term (7-week) cell fouling resistance, suggesting that shorter polymer chain lengths do not provide sufficient coatings to prevent adsorption of proteins to the underlying substrates for longer time scales.
The use of a predictive molecular theory can be a very important component in the design of surface modifiers for nonfouling applications. We have shown that the theory enables the prediction of the amount of polypeptoid bound to the surface and the ability of that surface to prevent protein adsorption. The optimal conditions for modification are such that the bound surface coverage is larger than the minimal coverage needed for thermodynamic prevention of protein adsorption. If the tethered polymers are at lower surface densities, long-term fouling will occur. The compromise between surface coverage and molecular weight in determining fouling capabilities depends on the chemical nature of the polymers and can be determined a priori with the theoretical tools.
This research was supported by NIH grants EB005772 and DE014193. XPS surface analysis was performed at KeckII/NUANCE at Northwestern University. Mass spectrometry was completed at the Northwestern University Analytical Services Laboratory. The authors thank Dr. Haeshin Lee for assistance with AFM and XPS experiments and many discussions.
Andrea R. Statz, Biomedical Engineering Department, Northwestern University, 2145 Sheridan Rd, Rm E310, Evanston, IL 60208.
Jinghao Kuang, Biomedical Engineering Department, Northwestern University, 2145 Sheridan Rd, Rm E310, Evanston, IL 60208.
Chunlai Ren, Biomedical Engineering Department, Northwestern University, 2145 Sheridan Rd, Rm E310, Evanston, IL 60208.
Annelise E. Barron, Bioengineering Department, Stanford University, W300B James H. Clark Center, 318 Campus Drive, Stanford University, Stanford, CA 94305.
Igal Szleifer, Biomedical Engineering Department, Northwestern University, 2145 Sheridan Rd, Rm E310, Evanston, IL 60208.
Phillip B. Messersmith, Biomedical Engineering Department, Northwestern University, 2145 Sheridan Rd, Rm E310, Evanston, IL 60208.