Bioreactor perfusion membrane

Fig. 1. Schematic illustration of a hepatocyte bioreactor with microcarrier-attachted hepatocytes. The capillary membranes are perfused with medium. (Modified from Dixit et al. [29])...

Several innovative membrane-based bioreactor designs have recently been proposed, including that by Sussman et al. [10], which involves the cultivation of hepatoma cells on the exterior surfaces of semipermeable capillary hollow fiber membranes which are bundled together with an enclosing plastic shell (Fig. 2). Required nutrient medium is circulated within the capillaries. After the hepa-tocytes have attached and formed a mass of liver tissue, the capillary membranes are perfused with the media. 

Combining these techniques, they carried out cultivations for 250-350 h, and were able to repeatedly use the same cartridge (four times at least) without measurable deterioration in filtration efficiency. However, when perfusion rate and cell concentration in the bioreactor increased, fouling eventually occurred. Van Reis et al. [92] provided backpressure on the filtrate line to control filtrate rates and so to avoid too high initial filtration rates, which can cause rapid fouling. De la Broise et al. [99] compared the filter performance using membranes of different pore sizes (2 and 10 pm). In both cases partial retention of the produced IgM was observed and membranes had to be changed every 5 days, the... 

The core of double membrane stirrer perfusion bioreactors is a stirrer on which two microporous hollow fiber membranes are mounted, one of them being hydrophobic and used for bubble-free aeration, the second of them being hydrophilic and used for cell-free medium exchange [15]. This system has been reported to provide viable cell densities of 20 million cells per miUiliter for more than two months [106]. Although Lehmann et al. [15] have described the scale-up of this system to the 20-L and 150-L scale, it has been most commonly employed at the bench-scale. 

Fermentation is typically conducted in dilute suspension culture. The low concentration in such systems limits reaction efficiency, and the presence of particulate and colloidal solids poses problems for product recovery and purification. By circulating the fermentation broth through an ultrafiltration system, it is possible to recover product continuously as they are generated while minimizing loss of enzyme or cells and keeping product concentration in the bioreactor below the self-inhibition level for the biocatalyst. This process is referred to as perfusion. As the ultrafiltration unit is part of the production process, the entire system is often considered a membrane reactor. 

Until now, bioreactors of various types have been developed. These include loop-fluidized bed [14], spin filter, continuously stirred turbine, hollow fiber, stirred tank, airlift, rotating drum, and photo bioreactors [1]. Bioreactor modifications include the substitution of a marine impeller in place of a flat-bladed turbine, and the use of a single, large, flat paddle or blade, and a newly designed membrane stirrer for bubble-free aeration [13, 15-18]. Kim et al. [19] developed a hybrid reactor with a cell-lift impeller and a sintered stainless steel sparger for Thalictrum rugosum cell cultures, and cell densities of up to 31 g L1 were obtained by perfusion without any problems with mixing or loss of cell viability the specific berberine productivity was comparable to that in shake flasks. Su and Humphrey [20] conducted a perfusion cultivation in a stirred tank bio-... 

Some of the efforts, so far, to model such membrane bioreactors seem to not have considered the complications that may result from the presence of the biomass. Tharakan and Chau [5.101], for example, developed a model and carried out numerical simulations to describe a radial flow, hollow fiber membrane bioreactor, in which the biocatalyst consisted of a mammalian cell culture placed in the annular volume between the reactor cell and the hollow fibers. Their model utilizes the appropriate non-linear kinetics to describe the substrate consumption however, the flow patterns assumed for the model were based on those obtained with an empty reactor, and would probably be inappropriate, when the annular volume is substantially filled with microorganisms. A model to describe a hollow-fiber perfusion system utilizing mouse adrenal tumor cells as biocatalysts was developed by Cima et al [5.102]. In contrast, to the model of Tharakan and Chau [5.101], this model took into account the effect of the biomass, and the flow pattern distribution in the annular volume. These effects are of key importance for conditions encountered in long-term cell cultures, when the cell mass is very dense and small voids can completely distort the flow patterns. However, the model calculations of Cima et al. [5.102] did not take into account the dynamic evolution of the cell culture due to growth, and its influence on the permeate flow rate. Their model is appropriate for constant biocatalyst concentration. 

Operation of various types of bioreactors in a perfusion mode (21, 22) enables the design engineer to combine several advantages of traditional modes of operation of well-stirred bioreactors (e.g., semibatch operation in a fed batch mode or use of recycle with a chemostat). Perfusion consists of operation in a mode in which a fresh growth medium (possibly together with a recycled growth medium) is fed to a bioreactor containing viable cells that are retained within the bioreactor by permselective membranes, microfllters, immobilization, or by partial separation and recovery from the reactor effluent followed by recycle to the entrance of the reactor. 

A similar approach to mimic capillary network and overcome diffusion limitations within large-tissue construct has been proposed by Narashima [23]. Hollow fiber membrane integrated within tissue-engineered scaffolds is supposed to supply required nutrients and gas through the walls of the fibers during in vitro perfusion bioreactor culture and then, being connected to the vasculature of the host, become a functional perfused network of the implant. 

Staining of the specimen removed from the HFMBs revealed that in the nonperfiised group the overall cell number was lower compared with the perfusion group and there were many dead cells too. In contrast, in the perfusion group, the cells were viable, and there were no detectable nonviable cells in this group. There was clear evidence of bonelike tissue formation in the hollow-fiber membrane bioreactors as well (Ye et al., 2(X)7). 

---