Fabrication and bioseparation studies of adsorptive membranes/felts made from electrospun cellulose acetate nanofibers

Abstract

Adsorptive membranes have shown great promise for bioseparations as an alternative to packed bed chromatography. Here we evaluate an adsorptive membrane (felt) made from electrospun cellulose acetate nanofibers as an ion-exchange medium for protein separations. The as-electrospun cellulose acetate nanofibers had diameters ranging from tens of nanometers to microns, and the pore sizes within the nanofiber felts were in the range from sub-microns to microns. The cellulose acetate nanofibers were hydrolyzed/deacetylated to yield regenerated cellulose nanofibers, which were then surface functionalized with diethylaminoethyl (DEAE) anion-exchange ligand. SEM imaging, along with FT-IR spectroscopy, and nitrogen content analysis were used to follow each stage of the process. For comparison, a regenerated cellulose microfiber adsorption medium, a commercially available regenerated cellulose adsorptive membrane, and bleached absorbent cotton balls, were all similarly treated and evaluated. The results indicated that the functionalized nanofiber felt had the highest static binding capacity of 40.0 mg/g for bovine serum albumin (BSA), compared to 33.5 mg/g, 14.5 mg/g, and 15.5 mg/g for the functionalized commercial membrane, cellulose microfiber medium, and cotton balls, respectively; and the DEAE nanofiber felt had over five times higher permeability of 10 mM Tris buffer, pH 8.0 than the DEAE commercial membrane. It was also found that detrimental system dispersion, as determined by calculation of the Peclet number, could be limited by increasing the number of felt layers used for the adsorption bed, without compromising pressure limitations. Dynamic adsorption of BSA, at 10% breakthrough, was higher for the DEAE nanofiber felt (26.9 mg/g) compared to the DEAE commercial membrane (20.9 mg/g).

Introduction

Biopharmaceutical therapeutics (e.g. recombinant proteins, monoclonal antibodies, viral vaccines, and plasmid DNA) have become major contributors in the fights against life threatening and debilitating disorders. Prior to 1997 only 6% of approvals by the U.S. Food and Drug Administration (FDA) for new therapies were for biopharmaceuticals. However, between 1997 and 2003 that number grew to 26%, and there are currently thousands of potential drug products of biological origin in clinical development [1]. As encouraging as these figures are, the biopharmaceutical industry is facing enormous pressures from the government and the public to improve the quality of therapeutics and increase the speed to market, while at the same time reducing the production costs [2], [3]. These demands are particularly relevant to the downstream purification of biological therapies because not only are the separation operations responsible for producing a safe product that meets purity guidelines established by the FDA, but economic modeling of processes have shown that a significant percentage (up to 80%) of the overall manufacturing costs are incurred during downstream purification [4], [5].

The traditional purification workhorse for many industrial bioseparation processes is the selective adsorption and elution of the target molecule within a packed bed of porous resin beads. The operation provides excellent purification factors, is reliably scaled between development and manufacturing sizes, and can be readily validated for commercial production [6]. Unfortunately, this process also suffers from several major limitations. First, the operational flow rates used during processing must be kept relatively low to maintain acceptable pressure limits, and it often requires reducing the flow rate during processing due to pressure increases. Similarly, in order to achieve high binding levels of the target molecule, in terms of bound product per volume of resin, very long residence times are required (again necessitating slow flow rates or cumbersome packing arrangements). These limitations are primarily due to the slow intra-particle diffusion of the relatively large biomolecules to access available binding sites deeply within the porous resin beads. Finally, concerns persist regarding the potential for flow channeling and poor dispersion within the packed bed, which leads to inefficient use of the expensive resin [7], [8].

A promising alternative to packed bed chromatography, which is gaining support from not only academic researchers but also industrial practitioners in the biopharmaceutical arena, is the use of adsorptive membranes (also referred to here as felts) [9], [10]. This style of adsorption utilizes the fibers to act as the support for ligands used during the selective adsorption process. The most important characteristics of this operation are that first, flow is through micro and macropores of the felt (as opposed to tightly packed resin beads), and second adsorption takes place on the surface of the fibers, where no internal diffusion is required. These factors reduce concerns of high-pressure drops with elevated flow rates, and eliminate the slow intra-particle diffusion required for adsorption within resin beads. It has been shown that the binding capacity of biomolecules to currently available adsorptive felts are similar in magnitude to resin beads, but can operate at processing flow rates over 10 times faster than packed beds [7], [8]. These factors allow for much faster processing times and potentially higher binding levels for purifying valuable biological products. This is highly desirable, especially for large biomolecules (molecular weights greater than 250 kDa, hydrodynamic diameters of 20–300 nm) because they are extremely difficult to purify using packed beds due to the severe mass transfer limitations within the small pores of resin beads.

Numerous examples of applying adsorptive membranes/felts to the purification of large biomolecules (e.g. plasmid DNA [11], [12], supercoiled DNA [13], and large proteins [14], [15]) have appeared recently. The results have been promising, and have repeatedly demonstrated superior performance compared to packed bed chromatography by providing higher binding capacities and elevated operating flow rates. Current industrial applications using membrane/felt adsorbers for monoclonal antibody purification have focused on employing them for polishing steps to remove nucleic acids, viruses, trace amounts of host cell protein, and product impurities (e.g. large aggregates). To maximize the advantages of membrane/felt adsorption operations, a flow-through mode (negative binding) has often been chosen where the high molecular weight impurities (DNA, virus, etc.) are captured by the membrane/felt while the targeted product emerges without binding [16], [17]. Bind and elute separations using membrane/felt adsorbers have shown less promise due to poor peak resolution, but have had some success at small scale [18].

Several important felt properties and operating conditions have emerged from the literature that suggests careful considerations should be made during membrane/felt adsorption. First, to maximize binding capacity, the optimum pore size among fibers should be less than ∼15 μm due to external (liquid phase) convective mass transfer limitations of transporting the target to the surface of the fibers [19], [20]. Many current adsorptive membranes/felts utilize a pore size in the range of 0.5–5 μm. Second, much like packed bed chromatography, even inlet flow dispersion and tight pore size distribution are required for efficient utilization of all available binding sites within the bed. This potential concern can be mitigated by using fibers oriented in random overlay to discourage unimpeded flow (channeling) of the feed solution (i.e. producing the adsorptive felt with a controlled thickness) [21]. Third, the static and dynamic binding capacities are generally equivalent for adsorption. This indicates that the binding capacity is independent of feed flow rate (and corresponding residence time within the adsorptive felt) and fast load times should not affect the performance; this is in sharp contrast to packed bed chromatography, which often requires slow feed rates or large columns to realize high binding levels [14], [15]. However, it should be noted that in some cases with large target molecules, flow rates must still be considered for felt adsorption [13]. Finally, and most importantly, the binding capacity of large molecules is often limited by available surface area of the fibers [22], [23]. This has been observed through mathematical modeling of experimental data [22] and confocal microscopic visualization of protein adsorption on fibers [24], [25].

Currently, the commercially available adsorptive felts are composed of fibers with diameters in the range of 5–25 μm. This relatively large diameter reduces the specific surface area within a bed, and significantly limits the potential binding capacity of the module. Thus, by utilizing electrospun nanofibers with diameters in the sub-micron to nanometer range, the available surface area within a given bed volume for potential binding will be greatly increased, by as much as two orders of magnitude. By controlling the pore size of nanofiber felts, the pressure drop and hydrodynamic flow characteristics can also be controlled and made to be as efficient as microfiber felts. Research efforts have indicated that the binding capacity of adsorptive felts made from so-called ultra-fine fibers (with diameters of ∼1–2 μm) were substantially higher to capture bovine serum albumin (BSA, a small to moderate sized protein with the molecular weight of ∼66 kDa) [26], [27], [28]. Nonetheless, the ultra-fine fibers are relatively expensive, inconvenient to produce, and still significantly thicker than electrospun nanofibers.

Electrospinning is a technique that utilizes the electric force alone to drive the spinning process and to produce polymer fibers from solutions or melts [29], [30], [31]. Unlike conventional spinning techniques (e.g. solution- and melt-spinning), which are capable of producing fibers with diameters in the micrometer range (ca. 5–25 μm), electrospinning is capable of producing fibers with diameters in the nanometer range (ca. 10–1000 nm). Electrospun polymer nanofibers possess many extraordinary properties including the small fiber diameter and the concomitant large specific surface area, the high degree of macromolecular orientation and the resultant superior mechanical properties. Additionally, felts made of electrospun polymer nanofibers offer a unique capability to control the pore sizes among nanofibers. Unlike nanorods, nanotubes, and nanowires that are produced mostly by synthetic, bottom-up methods, electrospun nanofibers are produced through a top-down nano-manufacturing process, which results in low-cost nanofibers that are also relatively easy to assemble and process into applications.

Cellulose-based materials are widely used in the biopharmaceutical processing industry as the base matrix for adsorbent beads and membranes. Cellulose is a naturally occurring polymer of particular interest due to its abundant availability, biodegradability, compatibility with biological systems, and most importantly, very low non-specific binding when used during purification of biopharmaceuticals from complex solutions. Unfortunately, the processing of cellulose is restricted by its limited solubility in common solvents and its inability to melt. The dissolution of cellulose requires the use of special solvent mixtures such as N-methylmorpholine-N-oxide (NMMO) and water [32] or lithium chloride and N,N-dimethylacetamide [33]. Conventionally, cellulose fibers are produced via wet spinning and involved derivatization of the polymer. In wet spinning, a mixture of cellulose (such as wood pulps harvested from tree farms) is fed to the spinneret, which is submerged in a chemical bath; and the fibers are collected as the solvent is removed and the polymer is precipitated/solidified. In this process, cellulose is derivatized into its xanthate form; and a sulfuric acid/zinc mixture is used as a coagulant to regenerate cellulose fibers. Alternatively, cellulose fibers can be produced from cellulose acetate. Unlike cellulose, cellulose acetate is soluble in many common solvents such as acetone. After cellulose acetate fibers undergo hydrolysis/deacetylation, the regenerated cellulose fibers are produced. It has been reported that cellulose acetate can also be electrospun into nanofibers; and the cellulose acetate nanofibers can be readily converted into the regenerated cellulose nanofibers by simply immersed in a NaOH aqueous solution at the room temperature [34], [35].

In this research, cellulose acetate was first electrospun into nanofibers, and the nanofibers were collected as randomly overlaid felts. The diameters of the cellulose nanofibers were in the range from tens of nanometers to microns, and the pore sizes within the nanofiber felts were in the range from sub-microns to microns. The cellulose acetate nanofiber felts were then converted into the regenerated cellulose nanofiber felts by hydrolysis/deacetylation in a NaOH aqueous solution. Subsequently, for the introduction of diethylaminoethyl (DEAE) groups used as anion-exchange ligands, the regenerated cellulose nanofibers, containing exposed hydroxyl groups, were treated with an aqueous solution containing 2-(diethylamino) ethyl chloride hydrochloride (DAECH) in an alkali condition at 90 °C. The surface functionalized nanofiber felts were then rinsed with dilute NaOH aqueous solution, dilute acetic acid aqueous solution, and finally water at room temperature [36], [37]. The morphologies of nanofibers (including as-electrospun cellulose acetate nanofibers, regenerated cellulose nanofibers, and DEAE surface functionalized cellulose nanofibers) were examined by scanning electron microscopy (SEM); the hydrolysis/deacetylation was studied by Fourier transform infrared (FT-IR); and the degree of surface functionalization was analyzed by measuring the samples’ nitrogen contents. Three alternative cellulose-based adsorption mediums were likewise surface functionalized with DEAE and used for comparison purposes. These were: (1) felts composed of micron-sized regenerated cellulose fibers with diameters of ∼15 μm, (2) bleached absorbent cotton balls, and (3) commercially available regenerated cellulose adsorptive membranes. Subsequently, performance of each felt was evaluated as a medium for bioseparations. Static adsorption isotherms for binding BSA were determined first. Then performance was measured in dynamic mode by determining permeability, system dispersion (axial diffusion), and breakthrough of BSA with the different felts. Experimental results indicated that the DEAE functionalized cellulose nanofiber felts significantly outperformed the control samples for binding capacity of BSA, by as much as 176%. Additionally, the nanofiber felt displayed elevated permeability, complete desorption of product, and by increasing the number of felt layers axial dispersion could be controlled.