A predictive aggregate transport model for microfiltration of combined macromolecular solutions and poly-disperse suspensions: testing model with transgenic goat milk

To meet the technical challenge of recovering human IgG fusion protein from transgenic whole goat milk at reasonable cost with high purity and yield, a predictive aggregate transport model for microfiltration has been developed (Baruah and Belfort, 2003). Here, to test the model's predictability of permeate flux and mass transport, a comprehensive series of experiments with varying wall shear rate, feed temperature, feed concentration, and module design are presented. A very good fit was obtained between the model predictions and measurements for a wide variety of experimental conditions. For microfiltration module design comparison, a linear hollow fiber module (representing current commercial technologies) gave lower permeation flux and higher yield than a helical hollow fiber module (representing the latest self-cleaning methodology). These results are easily explained with the model that is now being used to define operating conditions for maximizing performance. The procedure described by the model is generalizable and can be used to obtain optimal filtration performance for applications other than milk.     

Optimized recovery of monoclonal antibodies from transgenic goat milk by microfiltration

The Predictive Aggregate Transport Model for microfiltration is used in combination with optimum fluid mechanics and electrostatics to maximize recovery of a heterologous immunoglobulin (IgG) from transgenic goat milk. The optimization algorithm involved varying pH (6.8-9), transmembrane pressure (2-4.5 psi), milk feed concentration (1-2X), membrane module type (linear vs. helical design), and axial velocity (Reynolds number: 830-1170). Operation in the pressure-dependent regime at low uniform transmembrane pressures (approximately 2 psi) using permeate circulation in co-flow, at the pI of the protein (9 in this case) was used to increase IgG recovery from less than 1% to over 95%. Sodium dodecyl sulfate polyacrylamide gel electrophoresis and attenuated total reflection Fourier transform infrared spectroscopy of the microfiltration permeate samples confirmed that all the fat globules and most of the casein micelles were retained in the MF membrane whereas a large amount of the target IgG was transported through the membrane. Transmembrane pressure and hence permeation flux was kept low (approximately 15 lmh) to maximize IgG membrane transport and thus recovery, due to a sparse deposit on the membrane which facilitated high solute transport. Next, an analytical method was used to optimize the diafiltration process using the aggregate transport model, experimental target protein sieving coefficients and permeation flux (Baruah and Belfort, 2003). The methodology reported here should be generalizable to the recovery of target proteins found in other complex suspensions of biological origin using the microfiltration process.     

Controlled shear filtration: A novel technique for animal cell separation.

A novel rotary microfiltration technique specifically suited for the separation of animal cells has been developed. The concept allows the independent adjustment of wall shear stress, transmembrane pressure, and residence time, allowing straightforward optimization of the microfiltration process. By using a smooth, conically shaped rotor, it is possible to establish a controlled shear field in which animal cells experience a significant hydrodynamic lift away from the membrane surface. It is shown in preliminary experiments that shear-induced cell-rupture speeds up membrane clogging and that cell debris poses the most significant problem in harvesting of BHK cell cultures by dynamic microfiltration. However, a threshold value of shear stability exists which depends on the frequency of passing the shear field, the residence time in the shear field, as well as on cell status. By operating close to this threshold value, cell viability can be maintained while concentration polarization is efficiently minimized. By applying this concept, it is possible to attain flux rates several times higher compared to conventional crossflow filtration. Controlled shear filtration (CSF) can be used for batch harvesting as well as for cell retention in high cell density systems. In batch harvesting of hIL-2 from rBHK cell culture, a constant flux rate of 290 L h-1 m-2 has been adjusted without indication of membrane clogging or fouling.     

Transient and stationary operating conditions on performance of lactic acid bacteria crossflow microfiltration

A filtration rig equipped with a tubular alumina membrane was used to study the performance of crossflow microfiltration of Lactobacillus helveticus. Experiments were performed at constant permeation flux. High cell concentrations and fast transient conditions to the stationary J adversely affected permeability. Membrane fouling was due to a fast irreversible layer formation and to a reversible cell cake. This microbial deposit characteristics were dependent on the ratio permeation flux/wall shear stress, J/tau(w). Fouling was faster and more severe when J/tau(w) was greater than a critical value of 1.15 L(-1) . h(-1) . m(-2) . Pa(-1). The disordered structure of this cell cake seemed to lead to a macromolecule deposit between the cells which adversely affected the membrane permeability. (c) 1996 John Wiley & Sons, Inc.     

Crossflow microfiltration of animal cells

Laminar shear is the primary mechanism of cell damage, limiting flow rate (and hence flux) in crossflow microfiltration of animal cells. Sensitivity to hydrodynamic and interfacial stress is reduced by the addition of 0.1% Pluronic polyol. A critical average wall shear rate of 3000 s(-1) (above which damage occurs) is found for several cell types, including mammalian and insect cells. Hydrodynamic stress also limits the maximum tip speed in a rotary lobe pump to less than 350 cm/s. Turbulent flow in the recirculation loop piping at Reynolds numbers of up to 71,000 does not cause cell damage. Maximum sustainable flux decreases with cell concentration and increases with cell size (in qualitative agreement with the hydrodynamic lift model). A flux of 30 to 75 L/m(2) h (depending on cell size) can be sustained during 20-fold concentration from 2.5 x 10(6) cells/ml, while maintaining high cell viability.     

Effect of membrane morphology on system capacity during normal flow microfiltration

Understanding the effects of membrane fouling on system capacity is critical for the successful design and scale-up of microfiltration systems. The underlying morphology and structure of the microfiltration membrane can have a significant effect on system capacity by altering the rate and extent of fouling. Experimental data were obtained for system capacity during protein microfiltration using several model membranes with both homogeneous and composite structures. Data were compared with predictions of a new model that can account for both pore blockage and cake formation, and also includes the effects of membrane morphology on internal flow profiles within the membrane. Membranes with highly interconnected pores have a significantly higher capacity due to the reduction in flux decline arising from the fluid flow under and around any surface blockage. The model calculations are in good agreement with the flux decline data, allowing far more accurate predictions of system capacity than for the commonly used V(max) analysis.     

A new coiled hollow-fiber module design for enhanced microfiltration performance in biotechnology.

The microfiltration performance of a novel membrane module design with helically wound hollow fibers is compared with that obtained with a standard commercial-type crossflow module containing linear hollow fibers. Cell suspensions (yeast, E. coli, and mammalian cell cultures) commonly clarified in the biotechnology industry are used for this comparison. The effect of variables such as transmembrane pressure, particle suspension concentration, and feed flow rate on membrane performance is evaluated. Normalized permeation fluxes versus flow rate or Dean number behave according to a heat transfer correlation obtained with centrifugal instabilities of the Taylor type. The microfiltration performance of this new module design, which uses secondary flows in helical tubes, is significantly better than an equivalent current commercial crossflow module when filtering suspensions relevant to the biotechnology industry. Flux and capacity improvements of up to 3.2-fold (constant transmembrane pressure operation) and 3.9-fold (constant flux operation), respectively, were obtained with the helical module over those for the linear module.     

Microfiltration of recombinant yeast cells using a rotating disk dynamic filtration system

To develop a highly efficient cell harvest step under time constraint, a novel rotating disk dynamic filtration system was studied on the laboratory scale (0.147-ft.(2) nylon membrane) for concentrating recombinant yeast cells containing an intracellular product. The existing cross-flow microfiltration method yielded pseudo-steady state flux values below 25 LMH (L/m(2). h) even at low membrane loadings (10 L/ft.(2)). By creating high shear rates (up to 120,000(-1)) on the membrane surface using a rotating solid disk, this dynamic filter has demonstrated dramatically improved performance, presumably due to minimal cake buildup and reduced membrane fouling. Among the many factors investigated, disk rotating speed, which determines shear rates and flow patterns, was found to be the most important adjustable parameter. Our experimental results have shown that the flux increases with disk rotating speed, increases with transmembrane pressure at higher cell concentrations, and can be sustained at high levels under constant flux mode. At a certain membrane loading level, there was a critical speed below which it behaved similarly to a flat sheet system with equivalent shear. Average flux greater than 200 LMH has been demonstrated at 37-L/ft.(2) loading at maximum speed to complete sixfold concentration and 15-volume diafiltration for less than 100 min. An order of magnitude improvement over the crossflow microfiltration control was projected for large scale production. This superior performance, however, would be achieved at the expense of additional power input and heat dissipation, especially when cell concentration reaches above 80 g dry cell weight (DCW)/L. Although a positive linear relationship between power input and dynamic flux at a certain concentration factor has been established, high cell density associated with high viscosity impacted adversely on effective average shear rates and, eventually, severe membrane fouling, rather than cake formation, would limit the performance of this novel system. (c) 1995 John Wiley & Sons, Inc.     

Effect of Lipophilicity and Drug Ionization on Permeation Across Porcine Sublingual Mucosa

Sublingual route is one of the oldest alternative routes studied for the administration of drugs. However, the effect of physical-chemical properties on drug permeation via this route has not been systemically investigated. The objective of this study was to determine the effect of two key physicochemical properties, lipophilicity and ionization, on the transport of drugs across porcine sublingual mucosa. A series of β-blockers were used to study the effect of lipophilicity on drug permeation across the sublingual mucosa, while nimesulide (pKa 6.5) was used as a model drug to study the effect of degree of ionization on sublingual mucosa permeation of ionized and unionized species. Permeation of β-blockers increased linearly with an increase in the lipophilicity for the range of compounds studied. The permeability of nimesulide across sublingual mucosa decreased with an increase of pH. The flux of ionized and unionized forms of nimesulide was determined to delineate the contribution of ionized and unionized species to the total flux. At low pH, the apparent flux was primarily contributed by unionized species; however, when the pH is increased beyond its pKa, the primary contributor to the apparent flux, nimesulide, is ionized species. The contribution of each species to the apparent flux was shown to be determined by the thermodynamic activity of ionized or unionized species. This study identified the roles of lipophilicity and thermodynamic activity in drug permeation across the sublingual mucosa. The findings can help guide the design of sublingual drug delivery systems with optimal pH and solubility.     

Development of crossflow filtration processes for the commercial-scale isolation of a bacterial lipase

Isolation of a lipase produced by Pseudomonas aeruginosa (MW 29,000) employed crossflow microfiltration for production of a cell-free enzyme solution and crossflow ultrafiltration for concentration of the enzyme and removal of low molecular weight impurities. Poor flux and enzyme permeation were measured during initial screening of various microfiltration membrane types for isolation of the enzyme from a peptonized-milk-based broth; the results suggested that a soluble broth component was forming a gel layer which controlled both hydraulic flux and enzyme permeation. Reformulation of the fermentation medium resulted in enhanced performance, obtaining fluxes of 40 l/h m² and enzyme permeation of 50% on hydrophilically-modified PVDF membranes and resulted in a feasible clarification process. Enzyme permeation remained constant with respect to activity in the feed, rather than being proportional to activity in the retentate; it was hypothesized that this resulted from a concomitant concentration of the gel-forming components with cell concentration. Concentration of the clarified enzyme solution was performed using 30 000 MWCO regenerated cellulose membranes. Complete enzyme retention and high flux (57 l/h m²) were maintained through a 130-fold concentration of the microfiltrate. As both systems were taken to the 100 and 1000 l scales, similar filtration performances were obtained with system hold-up volume and pump cavitation becoming important considerations at the larger scales. Excellent reproducibility was observed in a series of eight large-scale experiments.     

Membrane separation of protein precipitates: Studies with cross flow in hollow fibers

Hollow fiber ultrafiltration and microfiltration membranes are examined for the processing of isoelectric soya protein precipitate suspensions. A model based on the various resistances to permeate flux is used to describe membrane performance. The main resistance to permeate flux is due to the interaction between the active membrane and the soluble and precipitated protein; that is, as compared with resistances due to the active membrane itself or the membrane support structure, or arising from concentrated soluble or precipitated protein layers over the membrane surface. Soluble protein rejection and precipitate mean particle diameter are correlated with observed values of this main resistance.In contract to the ultrafiltration of soluble proteins, the flux rates observed when processing protein precipitate suspensions under a similar range of operating conditions do not approach a limiting value with increased transmembrane pressure. At high protein concentrations, greater flux rates may be achieved for precipitated as compared with soluble proteins. The use of a microfiltration membrane does not give further improvement in flux rate; this may be attributed to problems of pore fouling with precipitate particles.     

Microfiltration of yeast suspensions with self-cleaning spiral vortices: possibilities for a new membrane module design

A novel method of producing controlled vortices was used to reduce both concentration polarization and membrane fouling during microfiltration of Saccharomyces cerevisiae broth suspensions. The method involves flow around a curved channel at a sufficient rate so as to produce centrifugal instabilities (called Dean vortices). These vortices depolarize the build-up of suspended particles such as yeast cells at the membrane-solution interface and allow for increased membrane permeation rates. Various operating conditions under which such vortices effectively reduced cake build-up of suspended particles such as yeast cells at the membrane-solution interface and allow for increased membrane permeation rates. Various operating conditions under which such vortices effectively reduced cake build-up during microfiltration of 0 to 0.55 dry wt% yeast broth were investigated. Flux improvements of over 60% for 0.25 dry wt% yeast broth for flow with over that without Dean vortices were observed. This beneficial effect increased with increasing retentate flow rate and increasing transmembrane pressure and decreased with increasing concentration of suspended matter. Similar behavior was observed whether the cells were viable of killed. the improvement in flux in the presence over that in the absence of vortices correlated well with centrifugal force or azimuthal velocity squared. The relative cake resistances increased with reservoir yeast concentration. These values with vortices increased from 62% to 75% of that without vortices with increasing yeast concentration. The ratio of the cake thicknesses in the limiting case (at high feed concentration) was 3.25. These results suggest that self-cleaning spiral vortices could be effective in maintaining good and steady microfiltration performance with cell suspensions other than those tested. (c) 1995 John Wiley & Sons, Inc.     

Comparison of different options for harvest of a therapeutic protein product from high cell density yeast fermentation broth

Recovery of therapeutic protein from high cell density yeast fermentations at commercial scale is a challenging task. In this study, we investigate and compare three different harvest approaches, namely centrifugation followed by depth filtration, centrifugation followed by filter-aid enhanced depth filtration, and microfiltration. This is achieved by presenting a case study involving recovery of a therapeutic protein from Pichia pastoris fermentation broth. The focus of this study is on performance of the depth filtration and the microfiltration steps. The experimental data has been fitted to the conventional models for cake filtration to evaluate specific cake resistance and cake compressibility. In the case of microfiltration, the experimental data agrees well with flux predicted by shear induced diffusion model. It is shown that, under optimal conditions, all three options can deliver the desired product recovery ( >80%), harvest time ( <15 h including sequential concentration/diafiltration step), and clarification ( <6 NTU). However, the three options differ in terms of process development time required, capital cost, consumable cost, ease of scale-ability and process robustness. It is recommended that these be kept under consideration when making a final decision on a harvesting approach.     

Recovery of insect cells using hollow fiber microfiltration

An efficient method was developed for media separation and cell collection for eukaryotic cells growing in suspension. The method is based on tangential flow microfiltration using an open channel arrangement in a hollow fiber configuration. Best results (highest processing flux rate) for polysulfone hollow fibers were obtained using fibers with internal diameter of 0.75 mm, 0.45 mum pore size, and a cell suspension flow at a shear rate of 14000 s(-1) (0.032 L/min per fiber). A flux rate of 500 L/m(2) h can be obtained by maintaining the surface area/cell ratio at 0.05 m(2)/10 L of cells at a concentration of 2.5 x 10(6) cells/mL. Forty liters of infected insect cells can be concentrated 10 times in 20 min without affecting cell viability. (c) 1995 John Wiley & Sons, Inc.     

An evaluation of the potential triggers of photoinactivation of photosystem II in the context of a Stern-Volmer model for downregulation and the reversible radical pair equilibrium model

Photoinactivation of photosystem II (PS II) is a light-dependent process that frequently leads to break-down and replacement of the D1 polypeptide. Photoinhibition occurs when the rate of photoinactivation is greater than the rate at which D1 is replaced and results in a decrease in the maximum efficiency of PS II photochemistry. Downregulation, which increases non-radiative decay within PS II, also decreases the maximum efficiency of PS II photochemistry and plays an important role in protecting against photoinhibition by reducing the yield of photoinactivation. The yield of photoinactivation has been shown to be relatively insensitive to photosynthetically active photon flux density (PPFD). Formation of the P680 radical (P680+), through charge separation at PS II, generation of triplet-state P680 (3P680*), through intersystem crossing and charge recombination, and double reduction of the primary stable electron acceptor of PS II (the plastoquinone, Q(A)) are all potentially critical steps in the triggering of photoinactivation. In this paper, these processes are assessed using fluorescence data from attached leaves of higher plant species, in the context of a Stern-Volmer model for downregulation and the reversible radical pair equilibrium model. It is shown that the yield of P680+ is very sensitive to PPFD and that downregulation has very little effect on its production. Consequently, it is unlikely to be the trigger for photoinactivation. The yields of 3P680* generated through charge recombination or intersystem crossing are both less sensitive to PPFD than the yield of P680+ and are both decreased by down regulation. The yield of doubly reduced Q(A) increases with incident photon flux density at low levels, but is relatively insensitive at moderate to high levels, and is greatly decreased by downregulation. Consequently, 3P680* and doubly reduced Q(A) are both viable as triggers of photoinactivation.     

Non-Additive Increases in Sediment Stability Are Generated by Macroinvertebrate Species Interactions in Laboratory Streams

Previous studies have shown that biological structures such as plant roots can have large impacts on landscape morphodynamics, and that physical models that do not incorporate biology can generate qualitatively incorrect predictions of sediment transport. However, work to date has focused almost entirely on the impacts of single, usually dominant, species. Here we ask whether multiple, coexisting species of hydropsychid caddisfly larvae have different impacts on sediment mobility compared to single-species systems due to competitive interactions and niche differences. We manipulated the presence of two common species of net-spinning caddisfly (Ceratopsyche oslari, Arctopsyche californica) in laboratory mesocosms and measured how their silk filtration nets influence the critical shear stress required to initiate sediment grain motion when they were in monoculture versus polyculture. We found that critical shear stress increases non-additively in polycultures where species were allowed to interact. Critical shear stress was 26% higher in multi-species assemblages compared to the average single-species monoculture, and 21% greater than levels of stability achieved by the species having the largest impact on sediment motion in monoculture. Supplementary behavioral experiments suggest the non-additive increase in critical shear stress may have occurred as competition among species led to shifts in the spatial distribution of the two populations and complementary habitat use. To explore the implications of these results for field conditions, we used results from the laboratory study to parameterize a common model of sediment transport. We then used this model to estimate potential bed movement in a natural stream for which we had measurements of channel geometry, grain size, and daily discharge. Although this extrapolation is speculative, it illustrates that multi-species impacts could be sufficiently large to reduce bedload sediment flux over annual time scales in streams where multiple species of caddisfly are present.     

Effects of intermolecular thiol-disulfide interchange reactions on bsa fouling during microfiltration

Several studies have shown that one of the critical factors governing protein fouling of microfiltration membranes is the presence of denaturedand/or aggregated protein in the bulk solutions. Experiments were performed to evaluate the role of intermolecular disulfide interchange reactionson protein aggregation and membrane fouling during stirred cell microfiltration of bovine serum albumin (BSA). The flux decline during BSA filtration was quite dramatic due to the formation of a protein deposit thatfully covered the membrane pores. This flux decline could be completely eliminated by capping the free sulfhydryl group present on the BSA with eithera carboxymethyl or cysteinyl group, demonstrating the critical importance of this free thiol in the intermolecular aggregation reactions and, in turn, protein fouling. BSA aggregation during storage could be reduced by the addition of metal chelators (EDTA and citrate) or dithiothreitol, orby storage at lower pH (7.0) these solutions all had a significantly lower rate of fouling upon subsequent filtration. This behavior is completely consistent with the known chemistry of the thiol-disulfide interchange reaction, demonstrating that an understanding of these intermolecular (aggregation) reactions can provide a rational framework for the analysis and control of protein fouling in these membrane systems. (c) 1994 John Wiley & Sons, Inc.     

Protein recovery from cell debris using rotary and tangential crossflow microfiltration

Protein recovery from a bacterial lysate was accomplished using microfiltration membranes in a flat crossflow filter and in a cylindrical rotary filter. Severe membrane fouling yielded relatively low long-term permeate flux values of 10(-4)-10(-3) cm/s (where I cm/s = 3.6 x 10(4) L/m(2) - h). The permeate flux was found to be nearly independent of transmembrane pressure and to increase with increasing shear rate and decreasing solids concentration. The flux increased with shear to approximately the one-third power or greater for the flat filter and the one-half power or greater for the rotary filter; the stronger dependence for the rotary filter is thought to result from Taylor vortices enhancing the back transport of debris carried to the membrane surface by the permeate flow. The average protein transmission or sieving coefficient was measured at approximately 0.6, but considerable scatter in the transmission data was observed. The largest sieving coefficients were obtained for dilute suspensions at high shear rate. The rotary filter provided higher fluxes than did the flat filter for dilute suspensions, but not for concentrated suspensions. (c) 1995 John Wiley & Sons, Inc.     

Performance of a tubular microporous membrane filter used for in situ sampling of mammalian cell culture medium

Summary The transmembrane permeation rate of the medium and the membrane transmission of low- and high-molecular medium components were evaluated for a microfiltration sampling system as a function of the operation time.