Search

Besides significantly broadening the scope of available data on adhesion of proteins on solid substrates, we demonstrate for the first time that all seven proteins (tested here) behave similarly with respect to adhesion exhibiting a step increase in adhesion as wettability of the solid substrate decreases. Also, quantitative measures of like-protein−protein and like-self-assembled-monolayer (SAM)−SAM adhesive energies are provided. New correlations, not previously reported, suggest that the helix and random content (as measures of secondary structure) normalized by the molecular weight of a protein are significant for predicting protein adhesion and are likely related to protein stability at interfaces. Atomic force microscopy (AFM) was used to directly measure the normalized adhesion or pull-off forces between a set of seven globular proteins and a series of eight well-defined model surfaces (SAMs), between like-SAM-immobilized surfaces and between like-protein-immobilized surfaces in phosphate buffer solution (pH 7.4). Normalized force−distance curves between SAMs (alkanethiolates deposited on gold terminated with functional uncharged groups −CH3, −OPh, −CF3, −CN, −OCH3, −OH, −CONH2, and −EG3OH) covalently attached to an AFM cantilever tip modified with a sphere and covalently immobilized proteins (ribonuclease A, lysozyme, bovine serum albumin, immunoglobulin, γ-globulins, pyruvate kinase, and fibrinogen) clearly illustrate the differences in adhesion between these surfaces and proteins. The adhesion of proteins with uncharged SAMs showed a general “step” dependence on the wettability of the surface as determined by the water contact angle under cyclooctane (θco). Thus, for SAMs with θco < ~66°, (−OH, −CONH2, and −EG3OH), weak adhesion was observed (>−4 ± 1 mN/m), while for ~66 < θco < ~104°, (−CH3, −OPh, −CF3, −CN, −OCH3), strong adhesion was observed (≤8 ± 3 mN/m) that increases (more negative) with the molecular weight of the protein. Large proteins (170−340 kDa), in contrast to small proteins (14 kDa), exhibit characteristic stepwise decompression curves extending to large separation distances (hundreds of nanometers). With respect to like-SAM surfaces, there exists a very strong adhesive (attractive) interaction between the apolar SAM surfaces and weak interactive energy between the polar SAM surfaces. Because the polar surfaces can form hydrogen bonds with water molecules and the apolar surfaces cannot, these measurements provide a quantitative measure of the so-called mean hydrophobic interaction (~−206 ± 8 mN/m) in phosphate-buffered saline at 296 ± 1 K. Regarding protein−protein interactions, small globular proteins (lysozyme and ribonuclease A) have the least self-adhesion force, indicating robust conformation of the proteins on the surface. Intermediate to large proteins (BSA and pyruvate kinase-tetramer) show measurable adhesion and suggest unfolding (mechanical denaturation) during retraction of the protein-covered substrate from the protein-covered AFM tip. Fibrinogen shows the greatest adhesion of 20.4 ± 2 mN/m. Unexpectedly, immunoglobulin G (IgG) and γ-globulins exhibited very little adhesion for intermediate size proteins. However, using a new composite index, n (the product of the percent helix plus random content times relative molecular weight as a fraction of the largest protein in the set, Fib), to correlate the normalized adhesion force, IgG and γ-globulins do not behave abnormally as a result of their relatively low helix and random (or high sheet) content.

Neither the adsorbed amount per unit actual surface area nor the structural stability of hen egg lysozyme is influenced by increasing the nanometer-scale roughness (5 < Rave < 60 nm) of a series of model substrates. Seven poly(ether sulfone) (PES) ultrafiltration membranes of increasing mean pore size with the same surface chemistry were chosen as model rough surface substrates. Topographical images, using atomic force microscopy, combined with attenuated total reflection Fourier transform infrared spectroscopy (ATR/FTIR) and sessile captive bubble contact angle measurements were used to characterize the surface properties of the substrates. ATR/FTIR spectroscopy together with a newly developed optimization algorithm for predicting the content of secondary structure motifs is used to correlate the secondary structure and amount of adsorbed lysozyme with the substrate surface roughness. From the adsorption measurements, the net adsorbed amount (total minus nonspecific adsorbed amount) of lysozyme corresponded to approximately one monolayer of coverage for all the substrates independent of the roughness. Although lysozyme was structurally disturbed through adsorption to PES substrates, no significant changes in its secondary structure were observed with the increasing roughness.

Mechanistic studies were conducted on UV-assisted graft polymerization of N-vinyl-2-pyrrolidinone (NVP) onto poly(ether sulfone) (PES) ultrafiltration (UF) membranes using a dip process. The molecular weight cutoff (MWCO) values of the UF membranes were 50, 70, and 100 kDa and the range of NVP monomer concentrations was 2−10 wt %. Our standard photooxidation protocol with 300-nm lamps was used. By operating below an irradiation energy, E < 4 kJ/m², both homopolymerization and main-chain scission can be minimized with up to CNVP = 5 wt % NVP concentration for all three UF membranes. A universal linear plot of the net amount grafted versus monomer concentration times irradiation energy (CNVP × E) for E < 4 kJ/m² was obtained for four NVP concentrations and three membranes. Resulting from graft polymerization, there appears to be a qualitative relationship between the thickness of the dense skin layer and the membrane surface roughness as measured by atomic force microscopy. Post-modification washing with ethanol was able to effectively remove entrapped homopolymer and other fragments. Also, UV-assisted graft polymerization of NVP on PES UF membranes effectively reduced irreversible membrane fouling. A protocol for optimizing graft-assisted photopolymerization of PES UF membranes is provided.

A novel two-dimensional surface molecular imprinting method using water-in-oil emulsion photopolymerization on a microporous polypropylene substrate was developed and used to separate the bronchodilator, theophylline, from the mild stimulant, caffeine, both of similar chemical structure. Surface molecular recognition sites were generated that could discriminate between these two molecules with a concentration-dependent separation factor of 4.9 ± 0.8 for theophylline over caffeine at 0.2 mM (1:1 v/v) solution mixture. Besides demonstrating its proof of concept, the attractive features of this new imprinting method are the following:  fewer imprint molecules are needed, it can be used to imprint from aqueous environments, postcrushing of the solid matrix is not needed, recognition sites are all near or at the surface reducing mass transfer limitations, and imprinting on synthetic microporous membranes (with convection and diffusion) instead of beads (with only diffusion) speeds up the process.

Many industrial feeds behave as non-Newtonian fluids, and little understanding exists as to their influence on cross-flow microfiltration (CMF) performance. The viscosity effects of a model non-Newtonian shear-thickening fluid were investigated in CMF with and without suspended silica particles in the feed. The goals were to compare the performance of Dean vortex filtration for non-Newtonian fluids with that for Newtonian fluids and, using a shear-thickening fluid, to establish, in a negative control experiment for equivalent axial flow rates, that the mean wall shear rates is higher in helical tube flow than in linear tube flow. Hence, the effect of viscosity on the formation of controlled centrifugal instabilities (Dean vortices) was studied. Non-Newtonian behavior was imparted to the fluid through the addition of a cationic surfactant, tetradecyltrimethylammonium salicylate (TTASal), that could freely permeate the membrane and that exhibited little noticeable fouling. Because the membrane did not retain the micelles and freely passed the surfactant monomer, fouling behavior could be separately studied by adding silica particles to the micellar solution. Results were compared with a previously studied Newtonian system for both linear and helical CMF. Two new mass transfer correlations were obtained for a shear-thickening surfactant solution in the presence of a 0.1 w/w % silica concentration for laminar flow in linear (no vortices) and helical (with vortices) membrane pipes. The viscosity was described as a function of surfactant concentration and velocity. The correlation for Dean vortex laminar flow incorporated the effect of increasing viscosity on the stability of Dean vortices by adjusting the exponent of the Reynolds number in the expression Shhel = 0.19(a/rc)^0.07Re^[0.55(η(v=0)^/η(v=v)^)]Sc^0.33 for Re < 800.

Low protein fouling 50 kDa poly(ether sulfone) ultrafiltration membranes were produced by UV-assisted graft polymerization of the monomer 1-vinyl-2-pyrrolidinone by using a dip modification technique without loss of protein rejection. UV lamps with an emission wavelength maximum of 300 nm and two specially selected UV light filters, benzene and an aromatic polyester film, were used to filter out 254 nm wavelength light which was found to be responsible for severe loss of protein rejection. The modified membranes that performed the best were prepared by using 300 nm lamps and the benzene filter. With increasing degree of grafting, these membranes exhibited up to 20% higher surface wettability (cos Θ = 0.83−0.89) than the base membrane (cos Θ = 0.74), which translated into lower irreversible flux loss (−0.05 to 0.03 compared with 0.42). While the protein rejection remained unchanged after modification, the permeability of the membranes decreased from 6.5 to 0.3 l m^-2 h^-1 (kPa)^-1 with increasing degree of grafting as a result of pore plugging by the grafted chains. These membranes exhibited similar irreversible flux loss (0.03 vs ~0), similar permeability (1.2 vs 1.8 l m^-2 h^-1 (kPa)^-1), and higher rejection (97.4 vs 93.4%) as compared with a 50 kDa regenerated cellulose membrane.

Contact angle measurements have been widely used to estimate the surface energy of various materials. Such measurements are severely limited with substrate surfaces that exhibit surface restructuring, are contaminated, and/or are porous. Although the captive bubble/drop method addresses the capillarity problem, surface undulations have not previously been accounted for in a quantitative way. We do so here with a series of 8 different pore size synthetic polymer membranes, all fabricated from poly(ether sulfone), as model rough porous surfaces of the same surface chemistry. Also, 3 of the 8 different pore size membranes were rendered hydrophilic through photoinduced graft polymerization producing 17 different modified membranes that are similarly tested. By incorporation of roughness parameters obtained from AFM measurements, corrections to the captive bubble/drop constant angle measurements were successfully made using a simple model of the surface. The predicted average value for the sessile drop contact angle of poly(ether sulfone) accounting for undulations (44.5 ± 1.3°) was, within error, equal to that value estimated for a smooth relatively nonporous PES nanofiltration membrane (42.9 ± 2.5°).

Designing low-adhesive surfaces is a challenging task made difficult in part by a lack of understanding of the fundamental intermolecular forces underlying adhesion. Atomic force microscopy was used to measure the adhesion energy between a well-known marine mussel adhesive protein, Mytilus edulis foot protein 1 (Mefp-1), and seven polymeric surfaces characterized by their critical surface tension (γc) and wettability (cos θw) in simulated and natural seawater. On the basis of their attractive binding (adhesive) energy during separation, the surfaces were divided in two groups, those that exhibited γc-values greater than and less than that of the model protein. The γc-value of the adsorbing protein determines the thermodynamic criteria for which substrates exhibit low and invariant adhesive energy or high and increased adhesive energy. The dominant interactions appear to be polar in nature (and possibly chemically specific π−π interactions for one of the surfaces) and not dispersive. Unusually close predictions between the Johnson−Kendall−Roberts theory and measurements were obtained.

Correlations between intermolecular forces and ultrafiltration measurements for a thin polysulfone film and membranes modified for increased hydrophilicity by graft polymerization of 2-hydroxyethyl methacrylate and a model protein (hen egg-white lysozyme, Lz) suggest that altering either the chemistry of the polymer surface or the solution conditions should lead to a minimization of protein adhesion and hence fouling for a specific protein/polymer combination. Using the surface forces apparatus, normalized adhesion forces were measured below, at and above the pI of Lz, and compared with corresponding permeation flux ratios from ultrafiltration experiments. Simple exponential correlations were obtained relating the normalized adhesion forces to several different permeation flux ratios. Also, the amount of protein adsorbed onto the membrane from solution during filtration was linearly related to the adhesion force through the choice of solution pH. The correlations imply that protein−polymer adhesive interactions are important during ultrafiltration. The results obtained for both a hydrophilic and a hydrophobic surface were compared. The hydrophilic surface exhibited lower contact angles, reduced adhesion forces, reduced adsorbed amount, and most importantly, reduced protein fouling. Long range attraction between adsorbed protein and hydrophobic polysulfone films was absent with the hydrophilic films. The results provide a fundamental molecular basis to the widely reported and observed phenomenon that hydrophilic membranes are known to foul less than hydrophobic ones during membrane filtration of protein solutions.

Two different techniques were used to photochemically modify 50 kDa poly(ether sulfone) (PES) membranes with the monomer N-vinyl-2-pyrrolidinone (NVP) to increase surface wettability and decrease adsorptive fouling during the constant volume diafiltration of 0.1 wt % bovine serum albumin (BSA). The filtration performance of the modified membranes was compared to that of a commercially available PES membrane and a regenerated cellulose membrane. Both the dip and immersion modification techniques produced membranes with essentially the same wettability as regenerated cellulose, a wettability increase of 30% over the base PES membrane. There was a substantial decrease in the irreversible adsorptive fouling of the base membrane as measured by the permanent flux drop after water cleaning with respect to the initial buffer flux using either modification (from 0.42 to 0−0.09). The immersion-modified membrane with the best performance exhibited no adsorptive fouling, similar permeability, and higher rejection than the regenerated cellulose membrane. Both modification techniques sharply decreased membrane permeability at high monomer concentrations due to pore blockage by grafted polymer chains. The dip-modified membranes exhibited simultaneous loss of BSA rejection and permeability, which suggested that although radiation cleaved PES bonds and enlarged the pores, the high degree of grafted polymer chains on the surface blocked the pores and decreased the permeability. The immersion- modified membranes retained their rejection because the monomer NVP solution was found to absorb up to 88% of the emitted energy, depending on its concentration, thereby protecting the pore structure from intense irradiation. Thus, the dip and immersion techniques are useful for applications where high protein transmission and retention are desired, respectively.

To reduce the number of recovery steps during downstream processing and to overcome the limitations of present fusion-based affinity separations, a controllable self-splicing protein element in the form of a mini-intein was used to optimize the recovery of proteins for both batch and flow purification strategies. The ability to recover purified proteins was demonstrated using a tripartite fusion consisting of a maltose binding domain, a truncated intein as a controllable linker molecule, and a protein of interest. To characterize expression level, solubility, cleavage rates, pH and temperature controllability, and protein activity, recombinant human acidic fibroblast growth factor (aFGF) was used as a model protein. A simple mass transport model, based on cleavage reaction-limited mass transfer and constant dispersion, was successfully used to predict product concentration and peak shape in relation to critical process parameters (with no fitting parameters). Insight into the nature of the cleavage reaction and its regulation was obtained via temperature- and pH-dependent kinetic data.

Monolayers of bifunctional bolaamphiphiles were self-assembled on a polymeric substrate (spin-coated film of poly(allylamine hydrochloride)). Various features of the novel bifunctional, well-ordered, and relatively stiff monolayer and the subsequent reactions of the exposed benzoyl azide groups were investigated. The characterization of the monolayers and the kinetics of self-assembly was determined via reflection−absorption infrared spectroscopy and ellipsometry. Atomic force microscopy and ellipsometry were used to measure roughness and thickness, respectively. From molecular modeling and ellipsometry, the bolaaphiphiles in the monolayers were found to be inclined at an angle of about 68° to the normal. In addition, the monomolecular layers appeared to be fairly crystalline. Serial functionalization of the exposed benzoyl azide groups with dansyl-cadaverine in the liquid phase followed by propylamine in the gas phase illustrated a stable, dense, well-ordered, two-dimensional monomolecular layer of bolaamphiphiles. Photopatterning, via UV irradiation of exposed benzoyl azide groups, was also demonstrated. The novelty of the surface modification procedure described here is the extraordinary ease with which self-assembly took place and the stability of the resulting monolayers which formed a highly reactive new surface suitable for further modification. Arranging molecules in a tidy ordered array on a polymeric surface is a main result of this work.

Multiwalled carbon nanotubes were purified and size-separated by a multistep microfiltration process through a sequence of track-etched polycarbonate membranes of various pore sizes in both dead-ended and cross-flow mode. For this cascade microfiltration, the electric arc derived raw multiwalled samples were suspended in an aqueous solution of sodium dodecyl sulfate in deionized water. By examining the deposits on the membrane surfaces and in the permeate suspensions with scanning electron microscopy and atomic force microscopy, the nanotube fractionation was confirmed and analyzed. These scanning techniques showed that the components of the crude sample, which included carbon nanotubes, polyhedral nanoparticles, and large aggregates, were separated from each other during the filtration. In addition, fractionation of the multiwalled carbon nanotubes according to length was possible.

High-frequency reverse filtration is an effective method of improving membrane performance. Models based on blocking filtration laws have been successfully used for flux prediction during discontinuous semibatch processes. In this paper, we present two semiempirical models to predict the performance of a continuous cross-flow filtration system with back-pulsing. Parameters evaluated from batch cell experiments are shown to be useful in scale-up and design of such systems. Both models are applied to results obtained for this work and to published results and are shown to be effective tools for prediction of process variables. Methods of process optimization are also discussed.

The activity of ribonuclease A (RNase A) during adsorption onto molecular smooth mica increases from 16% to 78% in a period of 24 h when compared to its activity in free solution at pH 5 and 20 +/- 0.5 degrees C. From electropotential plots, the tertiary structure of RNase A, the characteristics of the mica surface, and direct measurements of the intermolecular forces between two adsorbed enzyme layers, a molecular explanation is offered for the changing activity with time. Initially, the RNase A molecules lie flat-on the mica with their smallest axis perpendicular to and their active site facing the surface. As adsorption proceeds, the molecules slowly reorient until at long times they lie end-on with their largest axis perpendicular to the surface and their active site partially exposed to the free solution. A translational diffusion process is indicated for the phase transition and molecular reorientation of the RNase A molecules.

To gain a better understanding of marine fouling, we have measured the intermolecular forces between two layers of adsorbed extracellular polysaccharides (EPS) on silica. The measurements were made using an atomic force microscope (AFM) with a modified cantilever. Polymer bridging occurred between negatively charged EPS and was absent with neutral EPS. It was also observed for an anionic model dextran. Light-scattering measurements have confirmed a contraction of the EPS in free solution from large to small dimensions with an increase in ionic strength. As the ionic strength was increased to marine salt concentrations (about 0.5 M), adsorbed anionic EPS has changed from an extended to a contracted configuration and collapsed onto the solid substrate surface, forming a strongly adhered layer. Presumably, this EPS layer is an important component of the conditioning surface onto which bacteria and ultimately higher organisms eventually adhere.

In order to understand the effects of protein fouling during ultrafiltration of biological fluids, we have investigated the molecular interactions between a thin polysulfone film and hen egg lysozyme with the surface forces apparatus (SFA). The normalized forces between the adsorbed protein layers and polymer films were measured below, at and above the pI of lysozyme, and compared with four different permeation fluxes obtained from ultrafiltration experiments. The intermolecular forces between two protein layers were also measured at the different pH values. Adsorption kinetics of lysozyme onto mica were also obtained. Buffer and lysozyme solutions at similar pH values and concentrations were filtered with 6 kD polysulfone membranes to obtain flux decline and hence fouling measurements. Hydrophobic membranes, such as polysulfone, exhibit extremely long-range attractive interactions (on the order of 1500−2000 Å) with proteins such as lysozyme. Even in the presence of electrostatic repulsion at pH values above the isoelectric point of lysozyme (when both lysozyme and polysulfone were negatively charged), a long-range attractive interaction of around 210 μN/m was observed. Such interactions were absent with measurements between adsorbed lysozyme−lysozyme layers. From these measurements, simple linear correlations were found relating the normalized forces to the fluxes from the ultrafiltration experiments. With respect to fouling, protein−protein and protein−polymer interactions are about equally important during ultrafiltration. This suggests that both the surface chemistry of the membrane and the solution conditions could be chosen to minimize fouling for specific protein solutions. Hence, as a result of this study, fouling of polysulfone membranes with lysozyme solutions can be reduced if (i) filtration is conducted at pH values above the pI of lysozyme (approximately 10.8) and (ii) the membranes are modified such that the long-range attractive interactions are reduced. These results support those from previous phenomenological studies on membrane filtration of protein solutions and are the first evidence relating intermolecular force interactions with macroscopic events in membrane fouling.