A radial flow hollow fiber bioreactor for the large-scale culture of mammalian cells

A radial flow hollow fiber bioreactor has been developed that maximizes the utilization of fiber surface for cell growth while eliminating nutrient and metabolic gradients inherent in conventional hollow fiber cartridges. The reactor consists of a central flow distributor tube surrounded by an annular bed of hollow fibers. The central flow distributor tube ensures an axially uniform radial convective flow of nutrients across the fiber bed. Cells attach and proliferate on the outer surface of the fibers. The fibers are pretreated with polylysine to facilitate cell attachment and long-term maintenance of tissuelike densities of cell mass. A mixture of air and CO(2) is fed through the tube side of the hollow fibers, ensuring direct oxygenation of the cells and maintenance of pH. Spent medium diffuses across the cell layer into the tube side of the fibers and is convected away along with the spent gas stream. The bioreactor was run as a recycle reactor to permit maximum utilization of nutrient medium. A bioreactor with a membrane surface area of 1150 cm(2) was developed and H1 cells were grown to a density of 7.3 x 10(6) cells/cm(2).     

Ethanol production in a microporous hollow-fiber-based extractive fermentor with immobilized yeast

Microporous-membrane-based extractive product recovery in product-inhibited fermentations allows in situ recovery of inhibitory products in a nondispersive fashion. A tubular bioreactor with continuous strands of hydrophobic microporous hollow fibers having extracting solvent flowing in fiber lumen was utilized for yeast fermentation of glucose to ethanol. Yeast was effectively immobilized on the shell side in small lengths of chopped microporous hyrophilic hollow fibers. The beneficial effects of in situ dispersion-free solvent ex (oleyl alcohol and dibutyl phthalate) were demonstrated for a 300 g/L glucose substrate feed. Outlet glucose concentration dropped drastically from 123 to 41 g/L as solvent/ substrate flow ratio was increased from 0 to 3 at 9 mL/h of substrate flow rate with oleyl alcohol as extracting solvent. The significant productivity increase with in situ solvent extraction became more evident as solvent/ substrate flow ratio increased. A model of the locally integrated extractive bioreactor describes the observed fermentor performance quite well.     

Effect of harvesting protocol on performance of a hollow fiber bioreactor.

In this study, a bioreactor subject to Starling flow in closed shell batch harvest mode was compared to two forms of additional forced extracapillary (EC) space convection including EC circulation and EC cycling. Despite the presence of Starling flow as the dominant EC convection mechanism in the batch harvest system, the bioreactor start up was fairly good. However, the antibody productivity of the batch harvest system fell off rapidly after day 20 resulting in only 4.5 g of antibody produced. EC circulation with flow parallel to the fibers had a slightly better start up than the batch harvest. However, the antibody productivity also dropped after day 20 with EC circulation, resulting in only 7.5 g of antibody produced. EC cycling with flow both parallel and perpendicular to the fibers resulted in a start up similar to that of EC circulation. However, in contrast to the other two systems, antibody productivity in the EC cycling system was stable over the 60-day experiment resulting in the production of 23 g of antibody. These results demonstrate the importance of inducing the proper flow distribution in the EC space to allow consistent and stable production in hollow fiber bioreactors.     

Model of oxygen transport limitations in hollow fiber bioreactors

Axial and radial oxygen depletion are believed to be critical scale-limiting factors in the design of cell culture hollow fiber bioreactors. A mathematical analysis of oxygen depletion has been performed in order to develop effectiveness factor plots to aid in the scaling of hollow fiber bioreactors with cells immobilized in the shell-side. Considerations of the lumen mass transport resistances and the axial gradients were added to previous analyses of this immobilization geometry. An order of magnitude analysis was used to evaluate the impact of the shell-side convective fluxes on the oxygen transport. A modified Thiele modulus and a lumen and membrane resistance factor have been derived from the model. Use of these terms in the effectiveness factor plots results in a considerable simplification of the presentation and use of the model. Design criteria such as fiber dimensions and spacing, reactor lengths, and recycle flow rates can be selected using these plots. Model predictions of the oxygen limitations were compared to experimental measurements of the axial cell distributions in a severely oxygen limited hollow fiber bioreactor. Despite considerable uncertainty in our parameters and nonidealities in hollow fiber geometry, the cell distribution correlated well with the modeling results.     

Enhancing Cell Viability with Pulsating Flow in a Hollow Fiber Bioartificial Liver

A pulsating flow of medium was used to alleviate diffusion and transport limitations in a hollow fiber bioreactor containing a human hepatoblastoma cell line. The strategy is easy to implement but effective. The pulsating flow is introduced by a solenoid pinch valve at the outlet of the bioreactor and regulated by a timing circuit. In a permeability test, the system with pulsating flow had much less membrane fouling as compared to the control, a conventional hollow fiber unit. In hepatocyte culture test runs, the pulsating-flow bioreactor demonstrated the ability to maintain a higher cell viability. Histological sections indicated significantly smaller necrotic regions in the pulsating-flow bioreactor as compared to the conventional unit.     

Influence of domain orientation on the mechanical properties of regenerated cellulose fibers

The determination of the crystal orientation of regenerated cellulose fibers produced under different drawing regimes is presented. Orientation is determined by using wide-angle X-ray diffraction from a synchrotron source and by measuring the azimuthal width of equatorial reflections. The orientation parameter theta is then determined to compare fiber samples. By using a 500 nm beam size, clear differences between the crystal orientations of the skin and the core of the fibers are reported for a range of differently processed fibers for the first time. These results are shown to have implications for the mechanical properties of regenerated cellulose fibers. By applying tensile deformation to fiber bundles it is shown that the most misoriented samples undergo rapid decreases in the orientation parameter, which is an indication of crystal reorientation. However, the more highly oriented fibers undergo little reorientation. An average shear modulus for these fibers is determined by placing the data on a master curve and fitting with a model equation. By using another model for the fibers of low orientation and the shear modulus from the master curve analysis, it is shown that the deformation of less oriented fibers is dominated by shear between crystals, whereas the more oriented filaments are likely to undergo more significant chain deformation. By using a new model for fibers of low orientation, a parameter ksigma is introduced that gives the proportion of the fiber stress that is due to crystal shear. Systematic differences between this parameter for fibers of increasing initial orientation are reported. Moreover it is shown that the fibers of initially lower average orientation are governed by uniform strain, in agreement with the new model, whereas more highly oriented fibers deform under uniform stress. Furthermore, the model that we propose for misoriented domains and the use of a new factor dictating the proportion of shear stress may have general applications in materials engineering.     

A novel multi-coaxial hollow fiber bioreactor for adherent cell types. Part 1: hydrodynamic studies.

A novel multi-coaxial bioreactor for three-dimensional cultures of adherent cell types, such as liver, is described. It is composed of four tubes of increasing diameter placed one inside the other, creating four spatially isolated compartments. Liver acinar structure and physiological parameters are mimicked by sandwiching cells in the space between the two innermost semi-permeable tubes, or hollows fibers, and creating a radial flow of media from an outer compartment (ECC), through the cell mass compartment, and to an inner compartment (ICC). The outermost compartment is created by gas-permeable tubing, and the housing is used to oxygenate the perfusion media to periportal levels in the ECC. Experiments were performed using distilled water to correlate the radial flow rate (Q(r)) with (1) the pressure drop (DeltaP) between the media compartments that sandwich the cell compartment and (2) the pressure in the cell compartment (P(c)). These results were compared with the theoretical profile calculated based on the hydraulic permeability of the two innermost fibers. Phase-contrast velocity-encoded magnetic resonance imaging was used to visualize directly the axial velocities inside the bioreactor and confirm the assumptions of laminar flow and zero axial velocity at the boundaries of each compartment in the bioreactor. Axial flow rates were calculated from the magnetic resonance imaging results and were similar to the measured axial flow rates for the previously described experiments.     

Endothelial cell capture of heparin-binding growth factors under flow

Circulation is an important delivery method for both natural and synthetic molecules, but microenvironment interactions, regulated by endothelial cells and critical to the molecule's fate, are difficult to interpret using traditional approaches. In this work, we analyzed and predicted growth factor capture under flow using computer modeling and a three-dimensional experimental approach that includes pertinent circulation characteristics such as pulsatile flow, competing binding interactions, and limited bioavailability. An understanding of the controlling features of this process was desired. The experimental module consisted of a bioreactor with synthetic endothelial-lined hollow fibers under flow. The physical design of the system was incorporated into the model parameters. The heparin-binding growth factor fibroblast growth factor-2 (FGF-2) was used for both the experiments and simulations. Our computational model was composed of three parts: (1) media flow equations, (2) mass transport equations and (3) cell surface reaction equations. The model is based on the flow and reactions within a single hollow fiber and was scaled linearly by the total number of fibers for comparison with experimental results. Our model predicted, and experiments confirmed, that removal of heparan sulfate (HS) from the system would result in a dramatic loss of binding by heparin-binding proteins, but not by proteins that do not bind heparin. The model further predicted a significant loss of bound protein at flow rates only slightly higher than average capillary flow rates, corroborated experimentally, suggesting that the probability of capture in a single pass at high flow rates is extremely low. Several other key parameters were investigated with the coupling between receptors and proteoglycans shown to have a critical impact on successful capture. The combined system offers opportunities to examine circulation capture in a straightforward quantitative manner that should prove advantageous for biologicals or drug delivery investigations.     

Solvent removal during synthetic and Nephila fiber spinning

The process by which spiders make their mechanically superior fiber involves removal of solvent (water) from a concentrated protein solution while the solution flows through a progressively narrowing spinning canal. Our aim was to determine a possible mechanism of spider water removal by using a computational model. To develop appropriate computational techniques for modeling of solvent removal during fiber spinning, a study was first performed using a synthetic solution. In particular, the effect of solvent removal during elongational flow (also exhibited in the spinning canal of the spider) on fiber mechanical properties was examined. The study establishes a model for solvent removal during dry spinning of synthetic fibers, assuming that internal diffusion governs solvent removal and that convective resistance is small. A variable internal solvent diffusion coefficient, dependent on solvent concentration, is also taken into account in the model. An experimental setup for dry (air) spinning was used to make fibers whose diameter was on the order of those made by spiders (approximately 1 microm). Two fibers of different thickness, corresponding to different spinning conditions, were numerically modeled for solvent removal and then mechanically tested. These tests showed that the thinner fiber, which lost more solvent under elongational flow, had 5-fold better mechanical properties (elastic modulus of 100 MPa and toughness of 15 MJ/m3) than the thicker fiber. Even though the mechanical properties were far from those of dragline spider silk (modulus of 10 GPa and toughness of 150 MJ/m3), the experimental methodology and numerical principles developed for the synthetic case proved to be valuable when establishing a model for the Nephila spinning process. In this model, an assumption of rapid convective water removal at the spinning canal wall was made, with internal diffusion of water through the fiber as the governing process. Then the diffusion coefficient of water through the initial spinning solution, obtained ex vivo from the Nephila clavipes major ampullate gland, was determined and incorporated into the numerical procedure, along with the wall boundary conditions and canal geometry. Also, a typical fiber reeling speed during web making, as well as the assumption of a dry exiting fiber, were included in the model. The results show that a cross-section of spinning solution (dope), which is initially 70% water, spends 19 s in the spinning canal in order to emerge dry. While the dope cross-section traverses the canal, its velocity increases from 0.37 mm/s at the entrance to 12.5 mm/s at the canal exit. The obtained results thus indicate that simple diffusion, along with the dry wall boundary condition, is a viable mechanism for water removal during typical Nephila fiber spinning.     

Protein-free human-human hybridoma cultures in an intercalated-spiral alternate-dead-ended hollow fiber bioreactor

Batch cell cultures of a human-human hybridoma line in a convective flow dominant intercalated-spiral altetnate-dead-ended hollow fiber are compared with those using conventional axial-flow hollow fiber bioreactors and a stirred-tank bioreactor. Relatively short-term fed-batch and perfusion cell cultures were also employed for the intercalated-spiral bioreactor. When operating conditions of a batch intercalated-spiral bioreactor were properly chosen, the cell growth and substrate consumption paralleled that of a batch stirred-tank culture. The results verified the premise of the intercalated-spiral hollow fiber bioreactor that nutrient transport limitations can be eliminated when the convective flux through the extracapillary space is sufficiently high.(c) John Wiley & Sons, Inc.     

Evidence that αC region is origin of low modulus, high extensibility, and strain stiffening in fibrin fibers

Fibrin fibers form the structural scaffold of blood clots and perform the mechanical task of stemming blood flow. Several decades of investigation of fibrin fiber networks using macroscopic techniques have revealed remarkable mechanical properties. More recently, the microscopic origins of fibrin's mechanics have been probed through direct measurements on single fibrin fibers and individual fibrinogen molecules. Using a nanomanipulation system, we investigated the mechanical properties of individual fibrin fibers. The fibers were stretched with the atomic force microscope, and stress-versus-strain data was collected for fibers formed with and without ligation by the activated transglutaminase factor XIII (FXIIIa). We observed that ligation with FXIIIa nearly doubled the stiffness of the fibers. The stress-versus-strain behavior indicates that fibrin fibers exhibit properties similar to other elastomeric biopolymers. We propose a mechanical model that fits our observed force extension data, is consistent with the results of the ligation data, and suggests that the large observed extensibility in fibrin fibers is mediated by the natively unfolded regions of the molecule. Although some models attribute fibrin's force-versus-extension behavior to unfolding of structured regions within the monomer, our analysis argues that these models are inconsistent with the measured extensibility and elastic modulus.     

Bioartificial kidney. II. A convective flow model of a hollow fiber bioartificial renal tubule

An existing model for volume transport in the rat proximal tubule was modified and applied to a bioartificial renal tubule. The predicted volume transport agrees well with experimental data. The volume transport model was coupled to the analytic solutions of flow in the bioartificial renal tubule bioreactor, operated in the open-shell mode with perfusion in both the lumen and surrounding shell. The results suggest that the performance of a multifiber bioreactor can be improved by controlling shell inlet conditions and fiber spacing. These results have important implications for the design and analysis not only for the bioartificial renal tubule bioreactor but also for the general case of hollow fiber bioreactors operated in the open-shell mode with perfusion in both the lumen and surrounding shell.     

A failure model for ligaments

We propose a failure model for ligament which assumes that sequential uncrimping and stretching of collagen fibers is responsible for the mechanical response of ligament. We further assume that the fibers rupture sequentially and in a brittle, strain-limited manner. The model was fit to stress strain curves obtained from medial collateral ligaments of New Zealand White rabbits from two age groups (4 and 7 months). The model indicated that collagen modulus values ranged from 300 to 680 MPa and that fiber failure strains ranged from 6 to 22%. The model provides a convenient means of describing the elastic and failure response of ligament using four structurally based parameters.     

Detoxification of organophosphate nerve agents by immobilized dual functional biocatalysts in a cellulose hollow fiber bioreactor

A whole-cell technology for detoxification of organophosphates based on genetically engineered Escherichia coli cell expressing both cellulose-binding domain (CBD) and organophosphorus hydrolase (OPH) onto cell surface was reported recently (Wang et al., 2002). This study reports the application of these biocatalysts when immobilized in a cellulose hollow fiber bioreactor (HFB) for the biodetoxification of a model organophosphate, paraoxon, in a continuous flow mode. In 24 h, 0.79 mg wet cell/cm2 fiber surface were immobilized onto cellulose fibers specifically and strongly through the cellulose binding domain, forming a monolayer demonstrated by Scanning Electronic Micrograph, and essentially no cell was washed away by washing buffer. The immobilized biocatalyst had a high performance of detoxifying paraoxon solution of 5,220 mumol/h x L reactor or 990 mumol/h x m2 reactor. The immobilized biocatalysts maintained a stable degradation capacity for 15 uses over a period of 48 days with only 10% decline in degradation efficiency under operating and storage conditions. In addition, the bioreactor was easily regenerated by washing with 1% sodium dodecyl sulfate (SDS), with 86.7% immobilization capacity and 93.9% degradation efficiency recovery. This is the first report using the HFB in a non-traditional way, immobilizing whole-cell biocatalysts by specific adhesion thus rendering the catalysis operation the advantages of low pressure drop, low shear force, and low energy requirement. The successful application of this genetically engineered dual functional E. coli strain in a model bioreactor shows its promise in large-scale detoxification of organophosphate nerve agents in bulk liquid phase.     

Flow modeling in a novel non-perfusion conical bioreactor

We have developed a bioreactor vessel design which has the advantages of simplicity and ease of assembly and disassembly, and with the appropriately determined flow rate, even allows for a scaffold to be suspended freely regardless of its weight. This article reports our experimental and numerical investigations to evaluate the performance of a newly developed non-perfusion conical bioreactor by visualizing the flow through scaffolds with 45 degrees and 90 degrees fiber lay down patterns. The experiments were conducted at the Reynolds numbers (Re) 121, 170, and 218 based on the local velocity and width of scaffolds. The flow fields were captured using short-time exposures of 60 microm particles suspended in the bioreactor and illuminated using a thin laser sheet. The effects of scaffold fiber lay down pattern and Reynolds number were obtained and correspondingly compared to results obtained from a computational fluid dynamics (CFD) software package. The objectives of this article are twofold: to investigate the hypothesis that there may be an insufficient exchange of medium within the interior of the scaffold when using our non-perfusion bioreactor, and second, to compare the flows within and around scaffolds of 45 degrees and 90 degrees fiber lay down patterns. Scaffold porosity was also found to influence flow patterns. It was therefore shown that fluidic transport could be achieved within scaffolds with our bioreactor design, being a non-perfusion vessel. Fluid velocities were generally same of the same or one order lower in magnitude as compared to the inlet flow velocity. Additionally, the 90 degrees fiber lay down pattern scaffold was found to allow for slightly higher fluid velocities within, as compared to the 45 degrees fiber lay down pattern scaffold. This was due to the architecture and pore arrangement of the 90 degrees fiber lay down pattern scaffold, which allows for fluid to flow directly through (channel-like flow).     

Effects of oxidative stress-induced changes in the actin cytoskeletal structure on myoblast damage under compressive stress: confocal-based cell-specific finite element analysis

Muscle cells are frequently subjected to both mechanical and oxidative stresses in various physiological and pathological situations. To explore the mechanical mechanism of muscle cell damage under loading and oxidative stresses, we experimentally studied the effects of extrinsic hydrogen peroxides on the actin cytoskeletal structure in C2C12 myoblasts and presented a finite element (FE) analysis of how such changes in the actin cytoskeletal structure affected a myoblast’s capability to resist damage under compression. A confocal-based cell-specific FE model was built to parametrically study the effects of stress fiber density, fiber cross-sectional area, fiber tensile prestrain, as well as the elastic moduli of the stress fibers, actin cortex, nucleus and cytoplasm. The results showed that a decrease in the elastic moduli of both the stress fibers and actin cortex could increase the average tensile strain on the actin cortex–membrane structure and reduce the apparent cell elastic modulus. Assuming the cell would die when a certain percentage of membrane elements were strained beyond a threshold, a lower elastic modulus of actin cytoskeleton would compromise the compressive resistance of a myoblast and lead to cell death more readily. This model was used with a Weibull distribution function to successfully describe the extent of myoblasts damaged in a monolayer under compression.     

Development of a novel membrane aerated hollow-fiber microbioreactor

A new challenge in biotechnological processes is the development of flexible bioprocessing platforms, allowing strain selection, facilitating scale-up and integrating separation steps. Miniaturization of such a cultivation system allows parallel use and the saving of resources but makes the supply of oxygen to the cells difficult. In this work we present a membrane aerated hollow-fiber microbioreactor (HFMBR) which consists of an acrylic glass module equipped with two different types of membrane fibers. Fibers of polyethersulfone and polyvinyldifluoride were used for substrate and oxygen supply, respectively. Cultivation of E. coli as model organism and production of His-tagged GFP were carried out in the extracapillary space of the membrane aerated HFMBR and compared with cultivations in shaking flask which are commonly used for screening experiments. The measurement of the oxygen transfer capacity and the online monitoring of the dissolved oxygen during the cultivation were performed using a fiber optic oxygen sensor. Online measurement of the optical density was also integrated to the bioreactor. Due to efficient oxygen transfer, a better cell growth than in the shaking flask experiments was achieved, while no negative influence on the GFP productivity was observed in the membrane aerated bioreactor. Thus the feasibility of a future integrated downstreaming could also be demonstrated.     

Methodology for predicting oxygen transport on an intravenous membrane oxygenator combining computational and analytical models

A computational methodology for accurately predicting flow and oxygen-transport characteristics and performance of an intravenous membrane oxygenator (IMO) device is developed, tested, and validated. This methodology uses extensive numerical simulations of three-dimensional computational models to determine flow-mixing characteristics and oxygen-transfer performance, and analytical models to indirectly validate numerical predictions with experimental data, using both blood and water as working fluids. Direct numerical simulations for IMO stationary and pulsating balloons predict flow field and oxygen transport performance in response to changes in the device length, number of and balloon pulsation frequency. Multifiber models are used to investigate interfiber interference and length effects for a stationary balloon whereas a single fiber model is used to analyze the effect of balloon pulsations on velocity and oxygen concentration fields and to evaluate oxygen transfer rates. An analytical lumped model is developed and validated by comparing its numerical predictions with experimental data. Numerical results demonstrate that oxygen transfer rates for a stationary balloon regime decrease with increasing number of fibers, independent of the fluid type. The oxygen transfer rate ratio obtained with blood and water is approximately two. Balloon pulsations show an effective and enhanced flow mixing, with time-dependent recirculating flows around the fibers regions which induce higher oxygen transfer rates. The mass transfer rates increase approximately 100% and 80%, with water and blood, respectively, compared with stationary balloon operation. Calculations with combinations of frequency, number of fibers, fiber length and diameter, and inlet volumetric flow rates, agree well with the reported experimental results, and provide a solid comparative base for analysis, predictions, and comparisons with numerical and experimental data.     

A fiber-reinforced composite model of the viscoelastic behavior of the brainstem in shear.

Brainstem trauma occurs frequently in severe head injury, often resulting in fatal lesions due to importance of brainstem in crucial neural functions. Structurally, the brainstem is composed of bundles of axonal fibers distinctly oriented in a longitudinal direction surrounded by an extracellular matrix. We hypothesize that the oriented structure and architecture of the brainstem dictates this mechanical response and results in its selective vulnerability in rotational loading. In order to understand the relationship between the biologic architecture and the mechanical response and provide further insight into the high vulnerability of this region, a structural and mathematical model was created. A fiber-reinforced composite model composed of viscoelastic fibers surrounded by a viscoelastic matrix was used to relate the biological architecture of the brainstem to its anisotropic mechanical response. Relevant model parameters measured include the brainstem's composite complex moduli and relative fraction of matrix and fiber. The model predicted that the fiber component is three times stiffer and more viscous than the matrix. The fiber modulus predictions were compared with experimental tissue measurements. The optic nerve, a bundle of tightly packed longitudinally arranged myelinated fibers with little matrix, served as a surrogate for the brainstem fiber component. Model predictions agreed with experimental measures, offering a validation of the model. This approach provided an understanding of the relationship between the specific biologic architecture of the brainstem and the anisotropic mechanical response and allowed insight into reasons for the selective vulnerability of this region in rotational head injury.     

The speed of sound in DNA

We have used Brillouin scattering to determine the speed of sound in (and hence longitudinal modulus of) A- and B-DNA fibers. The speed of sound is very sensitive to the degree of hydration of the fibers, and measurements have to be made at laser powers below 5 mW to avoid local heating and dehydration. Under those conditions, we obtain sound speed perpendicular to the fiber axis of about 2.2 and 1.9 km/s in A- and B-DNA fibers, respectively. A-DNA fibers show a small anisotropy with sound speeds along the fiber axis higher by up to 10% B-DNA fibers appear to be isotropic.