Boyd, J.W.R. and Varley, J.

Biotechnology and Biochemical Engineering Group

Reading University, Whiteknights, Reading, Berks, RG6 2AP

Gas-liquid contacting is used extensively in the biotechnology industry. The design of bioreactors, such as in impeller agitated tanks and bubble columns, is influenced by the gas-liquid interfacial area and bubble size distributions. Existing bubble size measurement techniques such as photography, suction probes and ultrasonics each have their own limitations ^(1-3) . In this study, the acoustics of a laboratory scale bubble column are investigated, using a Bruel and Kjaer 8301 hydrophone and Hewlett-Packard Dynamic Signal Analyser. The objective of these studies is to ultimately develop an acoustic method for bubble size measurement. This technique could be used for looking at gas-liquid interfacial area where other methods are inappropriate and inaccurate in mass transfer studies. The potential advantages of using acoustics would be that it could be used on-line and in opaque liquids.

When excited by a change in internal or external pressure, bubbles oscillate^(4,5) . This oscillation produces a sound pulse at a frequency dependent on the bubble radius (figure 1). Many researchers^(6-8) have investigated the contribution of sound due to the entrainment of bubbles at sea to oceanic ambient noise. Glasgow et al.⁽⁹⁾ studied the acoustics of airlift reactors and measured sound believed to be due to the formation of bubbles.

A simple model used by Loewen and Melville⁽¹⁰⁾ for sound production due to a breaking wave has been applied to the case of bubble formation in a bubble column. Sound is assumed to be due to the summation of the individual bubble pressure pulses. An example of a modelled bubble pressure pulse is shown in figure 1.

Experimental sound spectra (sound pressure-frequency traces) measured close to a sintered glass plate sparger are presented and compared to idealised model spectra generated from photographically measured bubble size distributions (see figure 2). The model does not account for low frequency sound (due to tank resonance) and any reduction of the natural bubble frequencies due to the proximity of solid surfaces.

Loewen and Melville⁽¹⁰⁾ suggest a method of estimating the bubble size distribution from a sound spectrum and this technique has been demonstrated using the ideal modelled sound spectra.

Presenting author: Boyd, Jonathan

Corresponding author: Varley, Julie

E-mail: J.W.R.BOYD@reading.ac.uk and J.VARLEY@afnovell.reading.ac.uk

http://www.fst.rdg.ac.uk./people/pboyd/index.htm

1. GREAVES, M. and BARIGOU, M. (1988) 6th European Conference on Mixing , Pavia, Italy 24-26 May
1. PARATHASARATHY, R., JAMESON, G.T. and AHMED, N. (1991) Transactions of the Institute of Chemical Engineers, 69A, 295-300
1. CHAPELON, J.Y., SHANKAR, P.M. and NEWHOUSE, V.L. (1985) Journal of the Acoustical Society of America, 78(1), 196-201
1. MINNAERT, M. (1933) Philosophical Magazine, 16, 235
1. STRASBERG, M. (1956) Journal of the Acoustical Society of America, 16, 20-26
1. KERMAN, B.R. (1984) Journal of the Acoustical Society of America, 75(1), 149-165
1. KOLAINI, A.L. and CRUM, L.A. (1994) Journal of the Acoustical Society of America, 96(3), 1755-1765
1. LONGUET-HIGGINS, M.S. (1990) Journal of the Acoustical Society of America, 87(2), 652-661
1. GLASGOW, L.A., HUA, J., YIIN, T.Y. and ERICKSON, L.E (1992) Chemical Engineering Communications, 113, 155-181
1. LOEWEN, M.R. and MELVILLE, W.K. (1991) Journal of the Acoustical Society of America, 90(4), 2075-2080

Figure 1 - An example of a modelled sound pulse from a bubble.

Figure 2 - Comparison between model and experimental sound spectra for gas flow rate of 300cm³min^-1 at a sintered, glass sparger in a bubble column. Unbroken line - model result. Broken line - experimental results. Low frequency sound has not been filtered from the experimental spectra and is not accounted for in the model.

Brown, A.K and Varley, J.

Biotechnology and Biochemical Engineering Group

Reading University, Whiteknights, Reading, Berks, RG6 2AP

Foam fractionation is a well documented protein purification technique (1, 2) which has potential use in the preliminary stages of the downstream processing of recombinant and other proteins. The advantages of the process include, ease of operation, mechanical simplicity and therefore low cost compared to existing purification methods.

Foam fractionation exploits the fact that proteins which are surface active can adsorb to the gas/liquid interface of a bubble. This phenomena is related to the general physico-chemical properties of individual proteins such as, size, charge and hydrophobicity.

Much research has been conducted using single protein solutions (3, 4) but limited work exists for protein mixtures (5). In order for the successful purification of one protein component from any mixture of proteins, the surface activities of proteins must differ. This results in a greater surface active species, separating into the foam phase while the unwanted proteins remain in the residual (retentate) liquid.

The work presented here highlights studies on the foaming of binary protein mixtures using BSA/lysozyme, b-casein/lysozyme and BSA/b-casein. For this study, a continuous system was employed which may offer greater industrial application than a batch system. The operating parameters were chosen from a previous single protein study (6) that indicated the best possible conditions for enrichment of b-casein into the foam phase.

Figure 1 shows the apparatus used for continuous foam fractionation of protein mixtures.

Protein concentration in foam, initial feed and residual liquid samples was determined using anion exchange chromatography. Subsequently, the enrichment ratio ([protein]_foam/[protein]_initial), protein recovery ((mass _foam/mass _initial)*100) and purification factor (((mass_foam/total mass_foam)*100) / ((mass_initial/total mass_initial)*100)) were calculated to ascertain the efficiency of foam fractionation.

Table 1 shows selected results for binary protein mixtures.

Table 1 shows that separation of BSA/lysozyme and b-casein/lysozyme mixtures is successful. In these mixtures, the BSA and b-casein enrichments are significantly greater than corresponding lysozyme enrichment, indicating that BSA and b-casein adsorb to the gas/liquid interface to a greater extent than lysozyme. The physico-chemical properties of BSA and b-casein, in particular hydrophobicity and conformational flexibility, promote greater surface activity to BSA and b-casein. The likely reason for low lysozyme enrichments in these mixtures is that the physical properties of lysozyme prevent this protein from rapidly adsorbing to the interface i.e. disulphide linkages reduce protein flexibilty; low hydrophobicity. Enrichments reduce as the foam flow rate increases (due to greater foam stability at higher initial feed concentrations).

The b-casein/BSA mixtures display reduced separation efficiency and this is a consequence of b-casein and BSA simultaneously adsorbing to the gas/liquid interface. Both proteins are very surface active and therefore compete for interfacial area. Despite the absence of adequate purification for these mixtures, both proteins in the foam phase are significantly enriched. At higher initial feed concentrations of BSA (> 0.1 mg/ml), b-casein enrichment increases. On the other hand, at lower initial feed concentration of BSA (< 0.1 mg/ml), BSA enriches to a greater extent than b-casein. A possible explanation is that b-casein may be forming aggregates, which may reduce surface activity, explaining why BSA enriches more than b-casein at relatively lower BSA concentrations. However, at higher BSA concentrations, possible aggregation is reduced therefore allowing b-casein to preferentially adsorb. In experiments were aggregation is reduced, b-casein enrichment is consistently higher than BSA, supporting our experimental evidence that b-casein has greater surface hydrophobicity relative to BSA.

Presenting author: Brown, Alistair

Corresponding author: Varley, Julie

E-mail: A.K.Brown@reading.ac.uk

References

4. Sarkar, P., Bhattacharya, P., Mukherjea, R.N., Mukherjea, M., Biotech Bioeng, 29, 1987,

pp 934-940.

5. Varley, J., Kaul, A., Ball, S., Biotech Techniques, 10, 1996, 2, pp 133-140.

6. Brown, A.K., Kaul, A., and Varley, J., IChemE Research Event, 1, 1996, pp 73-75.

T. Tucker Norton and Erik J. Fernandez

Department of Chemical Engineering

University of Virginia

Charlottesville VA 22903-2442

The high solute concentrations employed in preparative chromatography of proteins and oligonucleotides can lead to elevated sample densities and viscosities and flow instabilities which can compromise chromatographic resolution. In previous studies using in situ magnetic resonance imaging and numerical simulation, we have found that packing uniformity and sample properties have the most profound effect on the viscous fingering flow instability and chromatographic resolution in size exclusion chromatography.

In this work, we have explored the effect of additional operating parameters on viscous fingering such as media particle diameter and solute velocity. Possible strategies which allow the use of highly concentrated samples are then evaluated. Through the use of a hybrid particle-tracking/finite-difference numerical simulation, the effectiveness of several sample loading techniques are examined, including modified sample band shapes and the use of viscous "protective slugs" using solutes such as glycerol. Sample loading procedures analogous to the water-alternate-gas approach used in enhanced oil recovery have also been investigated.

The results are compared to MRI experiments using size-exclusion chromatography of model solutes, and guidelines for the implementation of these strategies will be proposed. Finally, the effects of the viscous fingering instability on step gradient elution adsorption operations are being examined via numerical simulation.

Presenting author: Dr. Erik Fernandez

e-mail:erik@virginia.edu

Hiroyuki Honda, Akemi Katayama, Taizo Hanai and Takeshi Kobayashi

Dept. of Biotechnology, Graduate School of Engineering, Nagoya University

Furo-cho, Chikusa-ku, Nagoya 464-01, Japan

Sake brewing process has been conducted by the expert person called as Toji. Recently, the aging of Toji and the shortage of young successors have become serious problems for sake industry. The Ginjo sake is one of the highest-quality in all Japanese sake, and has characteristic flavors and tastes. Therefore, the Ginjo brewing process shows higher dependency on the experiences of Toji than the ordinary sake brewing process. As solutions for these problems, the automation and mechanization for sake brewing process have been proposed. Among them, the fuzzy control of a fermentation process is one of the useful strategy. We have also constructed the fuzzy rules to control temperature as a process variable (1,2). When the fuzzy rules constructed by us were applied to Ginjo sake fermentation process in a commercial scale, the Ginjo sake with high quality was produced (3). However, it took relatively long time to tune membership functions by the try-and-error method or the simulation and a lot of Toji's experiences were necessary to identify the rules.

Recently, fuzzy neural networks (FNNs) were studied as a tool for the fuzzy modeling (4). The FNNs have neural network structures of which connection weights have particular meanings for acquiring fuzzy production rules and for tuning membership functions. In our previous study (5), the FNN model was applied to the estimation of total evaluation from sensory evaluation of Ginjo sake, and the model constructed was found to be more precise than the model constructed by neural network for the estimation.

In the present paper, fuzzy modeling was studied for Ginjo sake brewing process using FNN. From the analysis of data for 25 Ginjo sake brewings, the control period was separated into 4 regions. We constructed 4 FNN models for fuzzy control in each control region. From time course data, 10 input variables were picked up as important variables for temperature control. Parameter increasing method was applied for selection of input variables in each control regions. As shown in Table 1, 2 to 5 variables were selected. The choice coincided well with Toji experience. Mean values of absolute errors as a real scale were less than 0.27 C.

 Table 1 Results of the FNN modeling in all regions.

 Table 2 Production rule for the set temperature identified after learning in control region 2.

Acquired models could estimate the set temperature precisely. In FNN, the acquired knowledge can be described in the form of IF-THEN rule by analysis of the connection weight wf in the FNN after the learning. Example of acquired rule is shown in Table 2. From the Table, the IF-THEN rule, that is "If today's temperature is small and BaumÈ is medium, the set temperature of the next day is 9.0 ∞C" is described. In all control regions, IF-THEN rules obtained were found to coincide well with the experiences of Toji. The suitability of acquired models was confirmed by the simulation proposed by us. The model was applied to the actual fermentation, and the sake produced showed high quality as Ginjo sake.

REFERENCES

Presenting author: Hiroyuki Honda

Corresponding author: Takeshi Kobayashi

e-mail : honda@proc.nubio.nagoya-u.ac.jp

Paula Jauregi, Matthew Noble, and Julie Varley

Biotechnology and Biochemical Engineering group,

The University of Reading, PO Box 226, Whiteknights, Reading RG6 6AP

Downstream processing of bioproducts consists of a number of unit operations to achieve the recovery of the desired product. The recovery of proteins is currently carried out using several techniques such as: microfiltration, chromatography etc. In this paper the potential use of CGA for protein recovery is explored as a novel technique.

CGA are micro sized (10-100 m m) gas bubbles created by intense stirring of a surfactant solution. These dispersions were first reported by Sebba¹ , who proposed that the microbubbles were surrounded by a soapy shell (see Fig. 1). There is no conclusive evidence to prove this but some of the properties derived from their size and structure (charged or polar groups on the interface) indicate that these dispersions may have potential for protein recovery. These properties are: a high interfacial area, a relatively high stability, the ease of transportation and the possibility of different mechanisms (electrostatic and hydrophobic interactions) for protein adsorption.

Fig. 1. Structure of CGA proposed by Sebba¹.

CGA have been used for various applications (dye separation², harvesting of microorganisms³, etc.) but there have been no reports related directly to protein recovery.

In the present study results of the characterisation of CGA, created by an anionic surfactant (AOT), and their application in protein recovery are described.

Protein-surfactant interactions could cause protein denaturation; therefore, the stability of aphrons needs to be limited, so that they will collapse in a short period of time for subsequent recovery of the protein. At the same time, small bubbles and uniform bubble size distributions will be required in order to maximise interfacial area for the protein adsorption. The CGA were contacted with a solution of lysozyme; high recoveries of lysozyme, together with high enrichment and separation ratios, were achieved at optimum conditions in the aphron phase (see Fig. 2). However, a significant amount of precipitate, which is solubilised with urea, was observed in all the experiments. Furthermore, model systems, using different mixes of protein and surfactant have been investigated in order to minimise the precipitation effect, and any denaturation of the proteins. Some of the preliminary results are presented for a number of different ionic and non-ionic surfactants and proteins.

Fig. 2. Results of separation of lysozyme ( · Recovery %,

+ Separation ratio,u Enrichment ratio) as a function of C_po

at variable (ns/np)_o and C_so= 0.14 mg/ml.

Presenting author: Jauregi, Paula

E-mail: afrjaure@reading.ac.uk

Michael S. Kallos^*, Brent A Reynolds⁺, Angelo L. Vescovi⁺, and Leo A. Behie^*

*Pharmaceutical Production Research Facility (PPRF), Faculty of Engineering

The University of Calgary, Calgary, Alberta, Canada, T2N 1N4

Jay D. Keasling

Abstract not available.

Dong-Hoon Hyun¹, Tai-Wook Yoon¹, Ohoak Kwon²,

Yong-Hwan Kim, and Chan-Wha Kim

Graduate School of Biotechnology, Korea University, Seoul 136-701, Korea

Sang Yong Kim, Deok Kun Oh, and Jung Hoe Kim

Department of Biological Sciences

Korea Advanced Science and Technology, Taejon, Korea

Tel. 82-42-869-2614 FAX : 82-42-869-2610

Effect of oxygen on xylitol production from xylose was investigated by using Candida parapsilosis ATCC 21019 mutant. Xylitol production in shake flask was sensitive to agitation speed and maximum under microaerobic condition such as 150rpm. Fermentor experiments were carried out by varying the oxygen transfer coefficient (k_La) in order to examine more quantitatively the effect of oxygen on xylitol production. When the k_La was increased from 20 to 85h^-1, the cell concentration was increased. However, the maximum production of xylitol from 50 g/l xylose was achieved 35.8 g/l for 62 hours at 30 h^-1 of k_La. Cultures for xylitol production were performed by controlling the dissolved oxygen (DO) concentration in range of 0.0 to 5.0% of air saturation. Under strict unaerobic condition, no fermentation was observed. Cell concentration was increased with increasing the DO concentration. Especially, above 3.0% of DO concentration, the cell concentration was remarkably increased. However, xylitol production from 50 g/l of xylose was maximum as 36 g/l for 54 hours at 0.7% of DO concentration. Activities of xylitol dehydtrogenase were increased by increasing the DO concentration. At high activity of xylitol dehydrogenase, the mose part of xylitol was converted into xylulose which changed to cell material. This results indicate that the DO concentration must be carefully controlled for efficient xylitol production from xylose. By controlling the Do concentration in the range of 0.8 - 1.2 with a cell concentration of 20 g/l, final xylitol concentration of 210g/l which corresponding to 70% of xylitol yield and 2.2 g/l-hr of productivity was obtained from 300 g/l xylose for 66hrs. This high volumetric productivity is very promising for an industrial application.

Presenting author: Kim, Jung H.

Joseph E. Machamer and Philip E. MacKenzie

Bio-Rad Laboratories, Hercules, CA, 94547

Pete Gagnon

Validated Biosystems, Tuscon, AZ 85750-0912

Monoclonal antibodies are the largest class of proteins currently proceeding through therapeutic and diagnostic clinical trials. Hydroxyapatite chromatography offers a unique selectivity for the separation of monoclonal antibodies. Historically, the poor reproducibility and mechanical stability of the crystalline forms of hydroxyapatite presented serious problems for utilization at process scale. A new form of hydroxyapatite has been developed which solves the problems posed by the use of crystalline hydroxyapatite in process applications. The new form is the result of a process in which crystalline hydroxyapatite is fused into ceramic hydroxyapatite(CHT) through a propriatary sintering process. This new chromatographic support offers a a new tool for scaleable monoclonal antibody purifications and it will continue to play an increasingly important role in the design of modern purification strategies. This poster reviews recent theories and supporting data on the retention mechanism of monoclonal antibodies on hydroxyapatite, applications of ceramic hydroxyapatite chromatography to MAb purification strategies, and typical IgG dynamic capacities as well as DNA, endotoxin, and virus clearance rates.

Presenting author: Machamer, Joseph E.

Carl-Fredrik Mandenius, Ingemar Lundström and Thomas Bachinger

Linköping University, Department of Physics and Measurement Technology

S-581 83 Linköping, Sweden

The aim of the presented work is to provide bioprocess engineering with a new type of on-line sensors that are able to extract useful and relevant process information from the composition of volatile compounds in the bioreactor headspace effluent. The emission of volatiles from the cells reflects in a variety of ways their metabolic and physiological state. The formation of odourants and other gases from a fermentation may for example, be characteristic for the course of the fermentation and may undergo changes that are reflecting key events during the fermentation cycle.

The methodology used in our work is based on sensor integration. Multisensor arrays, containing configurations of different sensors with varying sensitivities to volatile compounds typical for the culture, generate informative response patterns. Gas sensitive field effect transisitors, semiconducting metal oxide devices, piezoelectric crystal microbalances with selective coatings and optical gas sensors have been integrated in portable units connected to the bioreactor. By employing pattern recognition methods, the signals from these multisensor arrays can be related to bioprocess variables such as biomass, primary metabolites as well as state transition parameters of the cultures.

Examples will be given from industrial processes with recombinant Escherichia coli and CHO-cells as well as from Saccharomyces cerevisiae fed-batch fermentation. Estimation of metabolites and cellmass will be examplified using artificial neural networks and multivariate analysis. Visualization of continuous state transitions of bioprocesses from sensor fusion data also integrating other on-line process sensors, such as in-situ probes and in-line analyzers with the mentioned computation techniques, will further demonstrate the applicability of the multisensor array concept.

Presenting author: Mandenius, Carl-Fredrik

E-mail: cfm@ifm.liu.se

http://www.ifm.liu.se/Applphys/BioMeas/

Heng-Ho Wong, Brian K. O’Neill, Anton P.J. Middelberg

Department of Chemical Engineering

The University of Adelaide, SA 5005, Australia

A solid inclusion body (IB) often results when a protein is overexpressed in Escherichia coli. Recovery of active protein at laboratory scale requires several steps including cell disruption and IB recovery by low-speed differential centrifugation. Research at laboratory scale usually focuses on protein refolding and chromatographic recovery, where the greatest yield increases can generally be obtained. Differential fractionation of the IBs and cellular debris (CD) receives scant research interest. CD can be reduced by using chemical washes and by high-speed batch centrifugation after dissolution. CD removal at laboratory scale presents few worthy research challenges.

In contrast, scale-up of a laboratory process for manufacturing recombinant protein from IBs presents biochemical engineering challenges. The use of wash chemicals and protease inhibitors increases direct consumable and waste-treatment costs, and complicates process validation. Differential IB and CD centrifugation using disc-stack machines is poorly characterised. CD removal following IB dissolution is also difficult due to the low Stokes velocity of the cell-wall particles following exposure to highly-concentrated denaturant. Nevertheless, CD removal remains important. Residual CD following IB solubilization can significantly reduce chromatographic performance if not removed prior to resin contact. It also leads to high pyrogen loads in downstream units. In short, improved IB and CD classification can significantly enhance the economic viability of a large-scale manufacturing process.

We have examined the optimal large-scale classification of CD and IBs by experimental, modelling, and simulation studies. A new model that accurately describes the size reduction of E. coli cell debris during repeated homogenization has been developed. It has been validated using debris-size data obtained with a new analytical method that we have named Cumulative Sedimentation Analysis (CSA). CSA overcomes many of the limitations of existing debris-sizing methods, and is particularly suited to sizing mixtures of CD and IBs. The simulation also uses experimentally-determined grade efficiency curves for a disc-stack centrifuge that, for the first time, have been determined using actual E. coli CD. Simulation results conclusively show that multiple centrifuge and homogenizer passes improve the efficiency of IB and CD separation. There are clear benefits to further homogenization after cell disruption is complete. We demonstrate the practical benefits of improved CD and IB separation using a protease-sensitive analog of insulin-like growth factor-II. Product proteolysis during IB dissolution was significantly reduced through improved cell debris removal. The new model, data, and analytical technique provide an excellent means to optimise large-scale mid-stream processing operations in industrial biotechnology.

Presenting author: Anton Middelberg

email: antonm@chemeng.adelaide.edu.au

RECOVERY OF RECOMBINANT BETA-GLUCURONIDASE FROM TRANSGENIC SEEDS

Zivko L. Nikolov, Ann Kusnadi, and Roque Evangelista

Department of Agricultural and Biosystems Engineering

Center for Crops Utilization Research, Iowa State University, Ames, IA 50011

and

Elizabeth E. Hood and John A. Howard

ProdiGene, Inc, 1500 Research Parkway, Suite 220, College Station, TX 77845

Recent advances in plant transformation technology make transgenic plants an attractive alternative to fermentation-based systems for production of recombinant proteins. The use of transgenic plants as bioreactors for enzyme production was investigated by using E. coli beta-glucuronidase (GUS) expression in transgenic maize, soybeans, and canola as model system. The objectives of this work were to evaluate the stability of the recombinant enzyme during seed processing and to address some of the protein extraction and purification issues pertaining to downstream processing.

The effect of pH extraction conditions (mixing speed, mixing time, particle size, solid-to-liquid ratio, pH, and ionic strength) on GUS recovery was investigated. Particle size and presence of oil in the ground maize and soybean samples apparently did not affect the extraction yield. Full-fat canola samples were more difficult to grind and extract due to higher oil content in the seed. The pH of the extraction buffer played a greater role in transgenic soybean and canola extraction than that in maize. The optimum pH range for the extraction of GUS was between pH 5 and 7. The extraction of transgenic maize samples was less effected by pH than that of oilseeds.

The maize germ was fractionated by a dry-milling process and the recovery of GUS from flaked kernels, full-fat germ, and defatted germ was compared. The removal of oil from the full-fat germ by hexane extraction did not significantly affect GUS activity. Depending on the starting material, GUS extraction yields ranged between 0.2 and 0.4% of the total soluble protein.

The hexane extraction of soybean and canola oil reduced the extractable GUS activity by as much as 30%. The extraction yield of GUS from soy flour ranged between 0.002 and 0.004% and that from canola between 0.1 and 0.2% of the total soluble protein.

Recombinant GUS was purified by a three-step chromatography. Purification yields and final purity of GUS were primarily a function of the GUS-to-total protein ratio in the extract.

Presenting author: Nikolov, Zivko

E-mail: zivko@iastate.edu

Patwardhan A. V. and Ataai M. M.

Department of Chemical Engineering and the Center for Biotechnology & Bioengineering, University of Pittsburgh, 300 Technology Drive, Pittsburgh, PA 15219

Immobilized copper affinity chromatography experiments with E. coli cell extract have demonstrated that the cellular proteins of this most commonly used host organism elute from the column in a discontinuous fashion characterized by two major peaks. Thus, the elution behavior of the E. coli cell extract can be approximately simulated by simply two representative proteins that have a similar elution behavior as that of the peaks. We have identified the mixture of chicken egg white lysozyme and tuna heart cytochrome c to be excellent representatives of the E. coli cell extract. The equilibrium parameters of these proteins were determined over a wide pH range. A detailed mathematical model incorporating multicomponent Langmuir isotherm, axial dispersion, intraparticle diffusion and film mass transfer was developed to simulate the loading and elution of E. coli cell extract on a Cu-IDA column. Target proteins of varying affinities were also included in the model. Simulation results provided an excellent framework for identifying effective strategies for optimal column utilization as well as for attaining high resolution of the target protein.

Presenting author: Patwardhan, Anshuman V.

Mark W. Vaughn, Lih Kuo, and James C. Liao, Department of Chemical

Engineering, Texas A&M University, College Station, Texas 77843-3122

Endothelial release of nitric oxide (NO) has been documented to play an important role in the regulation of vascular tone and permeability, platelet adhesion and aggregation, smooth muscle proliferation, and endothelial cell-leukocyte interactions. NO in the microcirculation (arterioles < 150 mm in diameter) is of particular interest since the majority of vascular resistance is found in there. The transfer of NO from the producing cell to the target cell is poorly understood, since NO, as a free radical, can be degraded in a variety of reactions. Two major hypotheses have been proposed for the transfer of NO. The first states that NO is transferred from endothelial cells through diffusion and that cell membranes are readily permeable to NO. Under this hypothesis, the diffusion and reaction of NO determine the effective range of NO. The second hypothesis states that NO is packaged and delivered by thiol-containing compounds. Although preliminary data exist to support this argument, the detailed mechanism remains unknown. This article addresses the first hypothesis and examines factors that affect the diffusion distance of NO. We will use mathematical models based on known mechanisms and physiologically relevant rate constants to investigate the simultaneous diffusion and reaction of NO.

Under the free diffusion hypothesis, NO diffuses from the endothelium into the surrounding smooth muscle or into the vascular lumen. The NO that diffuses into the vascular smooth muscle cells targets the enzyme guanylyl cyclase to exert its vasoregulatory function. The mechanisms and kinetics of the reactions of NO in the vascular smooth muscle and adventitia are still largely unknown. In aqueous solution, the situation is well defined; the reaction of NO with O2 is second order in NO. The NO-O2 system only provides a lower limit to the reaction rate, however, because NO may undergo many more reactions in the vascular smooth muscle than with free O2 alone. NO may also react with superoxide, O2-, bind with heme containing proteins, interact with enzymes containing iron-sulfur centers, or be degraded in several other reactions. If these reactions cause the concentration of NO to fall significantly below the concentration needed to activate guanylyl cyclase, which has an equilibrium dissociation constant (kdis) equal to 0.25 mM, then the biological function of NO will be diminished. Therefore, the effective diffusion distance of NO, which is defined as the distance within which NO concentration is greater than the kdis of guanylyl cyclase, determines the functional range of NO action.

When NO diffuses into the vascular lumen, it interacts with the blood. The reaction rate of hemoglobin (Hb) with NO in vitro is very high. However, there is uncertainty about the reaction rate between NO and Hb or oxyHb in vivo. Furthermore, evidence exists to suggest that the reaction of NO with the blood has additional complexities. It has been shown that NO, or its reaction product with O2, can react with sulfhydryl groups such as those found in serum albumin to form S-nitrosothiol. Some investigators suggest that these S-nitrosothiols may serve as a NO carrier as part of the dynamic control of blood pressure. Such NO-thiol interaction is the basis of the packaged NO transfer hypothesis.

In this study, we modeled the reaction of NO from stimulated endothelial cells as a surface reaction. Reactions that degrade NO abluminally in the smooth muscle cells and adventitia were lumped into a single term, the rate constant for which was determined from experimental data. Similarly, NO binding with Hb or reacting with the oxygen of oxyHb in the lumen is considered as a lumped reaction. Several cases are examined, each using physiologically based estimates for the rate constant. Our objective in the present study is to determine factors that affect the NO concentration and the effective diffusion distance, especially in the microcirculation.

We found that the size of the vessel is an important factor determining the effective diffusion distance of NO. Microvessels (about 30-100 µm i.d.) are in the range where the luminal NO concentrations and the abluminal effective diffusion distance is maximal. Furthermore, the model suggests that if the NO-hemoglobin reaction rate is as fast as that reported for the in vitro rate, the NO concentration in the vascular smooth muscle will be insufficient to stimulate smooth muscle guanylyl cyclase effectively. In addition, the existence of an erythrocyte-free layer near the vascular wall and the reaction order of NO decomposition in the abluminal region are important in determining the effective NO diffusion distance. These results suggest that (i) the range of NO action may exhibit significant spatial heterogeneity in vivo, depending on the vessel size and the local chemistry of NO degradation, (ii) the NO binding/reaction constant with hemoglobin in the red blood cell may be much smaller than that with free hemoglobin, and (iii) the microcirculation is the optimal site for NO to exert its regulatory function. Since NO exhibits both vasodilatory function and anti-atherogenic activity, the high NO concentration and its long effective range in the microcirculation may serve as an intrinsic factor to prevent the development of systemic hypertension and of atherosclerotic pathology in microvessels.

Presenting author: James C. Liao

e-mail: j-liao@tamu.edu

T.Scheper, J. Hagedorn, R.Ulber¹, W. Schäfer, H.R. Howind², W. Flachsbarth³

1Institut für Technische Chemie, Callinstr. 3, 30167 Hannover

S.B. Noronha^a, L.B. Trinh^a, D.C. Kaslow^b, and J. Shiloach^a

aBiotechnology Unit NIDDK, NIH

C. Tengborg¹, K. Stenberg¹, M. Galbe¹, E. Palmqvist², G. Zacchi¹ and B. Hahn-Hägerdal^2
1Dept. of Chemical Engineering I, ²Dept. of Applied Microbiology

Lund University, P.O.B. 124,S-221 00 Lund, Sweden

In an environmentally sustainable process for production of ethanol from biomass, the use of fresh water as well as waste-water produced must be minimised. One way to do this is to recycle process streams. However, this leads to an accumulation of inhibitors in the process. The aim of this study was to investigate the effects of recirculation of different process streams on the hydrolysis step and the fermentation step, respectively. Figure 1 shows a schematic flowsheet of the process including the recirculation alternatives. The fresh water stream (1) in the hydrolysis step can be replaced by the stillage stream (2) or by a part of the dilute ethanol stream from the fermenter (3). The latter alternative would also increase the ethanol concentration in the feed to the distillation unit and thereby decrease the energy demand and operating cost.

Figure 1. Process configurations

The recycling was investigated in a bench-scale unit. A base case where fresh water was used in the hydrolysis step was performed as follows. Spruce was impregnated with 3.6% wt SO₂ /wt dry matter (DM) and steam pretreated at 215°C for 5 minutes. The pretreated material was diluted with fresh water to 7.5% wt DM/wt and supplemented with Celluclast and Novozym (Novo Industri A/S, Denmark) corresponding to 11 FPU/g DM and 14 Cellobiase Units/g DM. The enzymatic hydrolysis was performed at 40°C and a pH of 4.8 for about 100 hours. Prior to the fermentation the solid residue was separated from the liquid. The hydrolysate was supplemented with nutrients to a concentration of 0.5 g/l (NH₄)₂HPO₄ and 0.025 g/l MgSO_4× 7H₂O, and fermented using Saccharomyces cerevisiae with a cell mass of 10 g DM/l. The pH was maintained at 4.8 and the fermentation was performed at 30°C.

The base case was compared to four different recirculations options. When recirculating the stillage stream, the non-volatile components in the hydrolysis were concentrated 1.6 (R1) and 2.2 (R2a) times, respectively, compared with the base case. When recirculating a part of the dilute ethanol stream, the non-volatile components in the hydrolysis were concentrated 2.2 (R2b) and 3.1 (R3) times, respectively, compared with the base case.

The study shows that the sugar formed in the enzymatic hydrolysis step is not affected by the recirculation. However, the yield of fermentable sugars, i.e. glucose and mannose, is low already in the base case, 47% of the theoretical yield. The ethanol yield in the fermentation step decreases with increasing concentration of non-volatile components in the hydrolysis step, from 85% of the theoretical in the base case to 53% in R3. The ethanol has no negative effect on the fermentation when comparing R2a and R2b. The ethanol production rate is the same for R1 as for the base case, but for R2a, R2b and R3 it is much lower, Figure 2.

Figure 2. Ethanol production rates for the different recirculation options

Fresh water is mainly added in the pretreatment, as steam, and in the enzymatic hydrolysis, as liquid to adjust the dry matter content. In the base case the total amount of fresh water used is about 6 kg/kg raw material (which contains 50% dry matter). In R1 the amount of fresh water is reduced to 2.8 kg/kg, Table 1. For case R1 no negative effect of recycling was observed. This case corresponds to replacing all fresh water in the hydrolysis when the dry matter content after the pretreatment is 12%, Table 1. A further decrease in amount of fresh water, which requires higher dry matter content after the pretreatment (R2a, R2b and R3 in Table 1), results in a decrease in the ethanol production rate due to higher amount of inhibitors. This increases the capital cost in the fermentation step. This has to be weighed towards the lower operating costs obtained, due to reduced costs for fresh water and waste water treatment. Furthermore, for the cases R2b and R3 the energy cost is also reduced, since the energy demand will be less in the distillation step

Table 1.

Presenting author: Tengborg, Charlotte

Corresponding author: Zacchi, Guido

E-mail: Guido.Zacchi@kat.lth.se

http://chemeng1.kat.lth.se

B.A. Andrews, M. Salamanca, J.C. Merchuk* and J.A. Asenjo

Centre for Biochemical Engineering and Biotechnology, Department of Chemical Engineering, University of Chile, Beauchef 861, Santiago, Chile

*Department of Chemical Engineering, Ben Gurion University, of the Negev, Beer-Sheva, Israel

Aqueous Two-Phase Systems have started to be used industrially for the isolation and separation of recombinant microbial, mammalian and other proteins. One of the main reasons to use liquid-liquid extraction is, on the one hand the relatively simple scale-up possibilities, but also the potential of continuous steady-state operation. Until now most studies have concentrated on investigations of factors affecting the partition coefficient of molecules and even the pilot plant and scale-up developments have lacked in depth considerations and analysis of phase behaviour and phase separation. Whether the mecanism of phase separation is controlled by droplet aggregation or by droplet sedimentation and which of the phases is the continuous one are some of the key issues important in the selection and design of appropriate batch and continuous settlers.

M.E. Lienqueo, J.C. Salgado, E.W. Leser and J.A. Asenjo

Centre for Biochemical Engineering and Biotechnology, Department of Chemical Engineering, University of Chile

Beauchef 861, Santiago, Chile

Until now it has been virtually impossible to select and design a protein separation and purification process in a rational manner due to a lack of fundamental knowledge on the molecular properties of the materials to be separated and the lack of an efficient system to organize such information. This paper describes fundamental advances in both these areas. A hybrid Expert System that combines expert rules and mathematical correlations to manipulate databases for selecting the sequence of operations for the downstream purification of proteins was built. This paper will describe how the physicochemical data on the protein product and the other proteins present (contaminants) was used to select a sequence of operations with a minimum number of steps to achieve a defined level of purity.

We have investigated the relationship between individual physico-chemical properties of a relatively large number of proteins with a wide range of molecular weights, pIs and hydrophobicities, and their performance in separation systems in order to predict behaviour. The proteins have been characterized in detail in terms of their m.w., pI, charge as a function of pH (electrophoretic titration curves) and surface hydrophobicity. Correlations have been developed to relate their molecular/physico-chemical properties to their behaviour in separation systems, mainly chromatography based on anion and cation exchange, molecular weight and hydrophobicity. Affinity chromatography is considered on a case by case basis. The prediction of performance of individual proteins allows simulation of chromatographic and other processes and thus selection (by calculation of a separation coefficient, SC, for each protein) and prediction of purification performance in individual chromatographic columns . We have developed simplified simulation tools to carry out this task through geometrical analysis of the peaks.

To calculate the SC the system will read a database containing the information on the properties of the main contaminant proteins present in the specific expression system used. (e.g. 13 predominant proteins present in a soluble protein homogenate of E. coli). An algorithm was developed to model the amount of each protein contaminant eliminated after each step. A consultation was carried out using the Expert System to find all the steps necessary to achieve the desired level of purity (e.g. 98%) for the purification of the protein somatotropin produced in E. coli. Once a process was found the original databases were randomly varied at the levels of 10% and 20% to see the effect on the process proposed by the system. The differences suggested by the expert system with those of a "good" industrial process that has been published will be analyzed and discussed. The system was tested in a practical example and it was found that process synthesis was sufficiently sensitive to substantial changes in the physicochemical parameters of the product and protein contaminants and rather robust to small variations or errors in the measurement of such parameters (e.g. 10%).

Presenting author: Asenjo, Juan A.

CYCLOSPORIN A PRODUCTION USING IMMOBILIZED FUNGUS, Tolypocladium inflatum, IN TWO-STAGE CONTINUOUS PROCESS

T.H. Lee¹, Y.K. Chang¹, G.-T. Chun^{2 1}Department of Chemical Engineering and BioProcess Engineering Research Center, Korea Advanced Institute of Science and Technology, Taejon, 305-701, Korea ²Division of Life Science, Kangwon National University, Chunchon, 200-701, Korea

A two-stage continuous process was investigated, which consisted of an immobilized- cell reactor and a suspended-cell reactor. The major objective of this study was to identify the optimum reactor system configuration and operation conditions for the production of cyclosporin A(CyA) using a fungus, Tolypocladium inflatum. The two- stage process was found to be superior in terms of CyA productivity to the one-stage continuous immobilized-cell culture and simple batch suspended-cell culture tested also. The CyA productivity of two-stage culture was about 1.3 and 11 fold higher than those of one-stage and batch cultures, respectively. A comprehensive mathematical modelling was carried out also.

Introduction : Cyclosporin A, a powerful immunosuppressive agent, is very effective in organ transplants as a key agent used to prevent foreign tissue rejection, and has considerable promise in the control of autoimmune and parasitic diseases[1]. It is produced intracellularly as a secondary metabolite by T. inflatum. It was reported by our group previously that one-stage continuous immobilized-cell culture was very efficient for CyA production[2]. In this system, T. inflatum was immobilized in celite beads. The medium composition was selected to promote cell growth as well as CyA production. In this situation, product-containing free cells or fragments are produced to leave the reactor. The product was recovered by cell disruption. We thought that the CyA productivity could be much increased if a suspended-cell reactor was added to the system to provide the cells from the immobilized-cell reactor with an optimal condition and enough time to produce CyA. For this reason, an experimental study on two-stage process carried out, in which the feed composition in the second stage was optimized.

Materials and Methods : Tolypocladium inflatum ATCC 34921 was adapted to grow on fructose. Celite R-560(Manville Co., U.S.A.) was used as the biosupport. The methods and procedures for seed culture and immobilization of spores into celite beads are described elsewhere[2]. The growth/production medium composition used for the first stage was : fructose 30 g/L, L-valine 6 g/L, (NH₄)₂SO₄ 10 g/L, KH₂PO₄ 0.75 g/L, MgSO₄ 0.5 g/L, CaCl 0.1 g/L, and trace element 0.1 %. The medium composition for the secondary feed in the second stage was identical to the above except for L-valine concentration(10 g/L) and fructose concentration(30/10g/L). A schematic diagram for two stage continuous process is shown in Fig. 1. The working volumes of the first and the second stages were 2 and 4 L, respectively. The dilution rates for the first and the second stages were 0.1 and 0.065 hr^-1, respectively. The temperature was 27⁰C and the pH was adjusted at 5.7. The analytical methods for cell concentration, fructose, and CyA are described elsewhere[2].

Results and Discussion : Experimental results of batch suspended-cell culture, one-stage continuous immobilized-cell culture, and two-stage continuous culture are summerized and compared in Table 1. The CyA productivity of the two stage culture was strongly influenced by the fructose concentration(S₀₂) in the secondary feed(F₂). When the same fructose concentration as in the primary feed(F₁), that is, 30 g/L was used for the secondary feed, the CyA productivity of two-stage culture(1.43 mg/L/hr) was very low, even lower than that of one-stage culture(2.50 mg/L/hr). The cellular productivity(0.858 mg/L/hr), however, was higher than that of one-stage culture(0.740 mg/L/hr). Such a high level of fructose concentration in the secondary feed was considered to make the residual fructose concentration in the reactor high enough to inhibit CyA production, the major role of the second stage. As the fructose concentration in the secondary feed was decreased to 10 g/L, the CyA productivity significantly increased to 3.25 mg/L/hr, which is 1.3 and 11 fold higher than those of one-stage and batch culture, respectively. The medium composition and the operating conditions such as dilution rates for the two-stage process need to be further optimized.

Figure 1. Schematic diagram of two stage continuous process

Table 1. Continuous immobilized-cell cultures vs. batch suspended culture

References

1. Borel, J.F.. Ciclosporin and its future. In Ciclosporin, Progress in Allergy, Vol. 38 (J.F. Borel, ed.), pp. 9-18, Karger, Basel, 1986.

2. T.H. Lee, G.-T. Chun and Y.K. Chang, J. Lee, and S.N. Agathos, "Bioprocess engineering considerations in cyclosporin A fermentation by immobilized fungus Tolypocladium inflatum", Immobilized Cells: Basics and Applications (R.H. Wijffels, R.M. Buitelaar, C. Bucke, and J. Tramper, eds.), pp. 402-409, Elsevier Science B.V., 1996.

Presenting author: Chang, Y.K.

E-mail: ychang@sorak.kaist.sc.kr

James M. Kubiak, Todd A. Battistoni, Kurt P. Gilson* and David W. Jayme.

As manufacturers of biopharmaceuticals implement large-scale production of approved biologicals, cost constraints mandate focusing facility and personnel commitment to core competencies associated with biological end product manufacture. Ancillary support functions to this primary activity (such as batch production of nutrient medium and purification buffers) are candidates for capital preservation and improved operating efficiencies. We have previously reported an integrated continuous process design as a possible alternative production method, both applied to our internal large-scale fluids manufacturing operation and potential use by collaborators for their on-site application. Fabrication, design prove-out, and preliminary validation have now been completed and a production-scale system has been implemented within our manufacturing facility. The system was designed to produce 1000-3000 liters per hour of single-strength liquid medium or buffered salt solution from concentrated intermediates meeting specific performance tolerances and to deliver lot sizes in excess of 10,000 liters to bulk containers. Key elements of the process validation master plan pertaining to product homogeneity and consistency, including monitoring and maintenance of pH, conductivity and solubility in a dynamically-buffered solution, will be described. We will also discuss operational management of the system, focusing upon sanitization, sterilization, and in-process logistics. Beyond meeting our internal manufacturing requirements, this scaleable system provides flexibility to meet increased volume demand for nutrient medium and chromatography buffers for large-scale biological production and offers cost-saving design options for feed-stream management in continuous bioproduction and purification processes.

Presenting author: Battistoni, Todd

Corresponding author: Kubiak, James

email: jkubiak@lifetech.com

Claus Emborg

No abstract available.

Antonio A. García*, Jaime Ramirez-Vick*, James J. Lee**

*Arizona State University, Tempe, Arizona

**Mayo Clinic Scottsdale, 13400 E. Shea Blvd., Scottsdale, Arizona

Non-radioactive labeling of biological molecules has become quite popular during the past 16 years mostly due to advances in molecular biology and more specifically due to the development of biotin-avidin methods and the increasing sophistication of immunological methods. These non-radioactive labeling techniques also provide opportunities for purifying complex, protein or biological fluids for clinical diagnostics and commercial production.

Immobilized Metal Affinity Chromatography (IMAC) is a highly versatile method whose selectivity can be varied by a judicious choice of ligand and metal ion as well as by varying the modes of elution. Porath introduced the concept of metal chelate chromatography [1] which was later renamed Immobilized Metal Affinity Chromatography [2]. Primarily, this method was developed for protein fractionation on the basis of the relative content of surface accessible imidazole, tryptophan, and cysteine residues as well as terminal amino groups, but in recent years, there has been increased interest in separating nucleotides [3]. In our work we are interested in combining labeling techniques with metal ion affinity interactions in order to selectively bind and recover biological molecules and cells. A promising way of doing this is through the use of soft metal ions and biotin labeling.

IMAC columns are generally packed with derivatives of a hydrophilic cross-linked biopolymer. Some of the insoluble matrix supports used include cellulose, polystyrene gels, polyacrylamide gels, porous glass, and agarose. The most commonly used metal ions are Zn(II), Ni(II), Co(II), Cu(II) which are classified as borderline Lewis acids by Pearson [4] and tend to interact preferentially with borderline hard bases such as the nitrogen bases found in histidine residues. On the other hand, Ag(I) is a soft acid and has an affinity for soft bases such as thioethers [4]. The purpose of this work is to extend the application of metal affinity interactions to the binding and recovery of biotin labeled biomolecules and cells using the soft metal ion Ag(I).

The thioether group on biotin is a particularly useful target ligand since it is a non-radioactive marker molecule used as a probe for the specific detection of target nucleic acid sequences [5]. Biotin is currently widely used in DNA/RNA detection [6] due to the extremely high binding constants of biotin-streptavidin complexes. However, the downside of this technique is the difficulty in reversing the complex under reasonably mild conditions (Kd ~ 1 x 10^-15 M). Recovery of the biotin labeled oligonucleotide molecule is crucial since oligonucleotides need to be released for subsequent use in recombinant DNA methods.

One facet of our studies is aimed at developing a useful, selective method employing soft metal affinity interactions for binding and recovering biotin labeled oligonucleotides. Some of the intended applications are: (1) the isolation of full length oligonucleotides from failure sequences that result during automated solid support synthesis; (2) study of DNA/protein complexes; and (3) the isolation of PCR products resulting from incorporation of biotinylated primers. Our research has led to the development of chromatographic supports as well as paramagnetic particles with immobilized silver ions for selective binding of biotin labeled oligonucleotides. We have been able to produce chromatgraphic supports with immobilized silver ions using Biogel P-2, P-200, and Sepharose 4B. Work thus far has shown that by adjusting the chloride concentration in the mobile phase buffer (phosphate buffer pH 7, 0.001 M NaCl), conditions can be obtained for affinity binding of biotin labeled dUTP (B-11-dUTP) while no interaction takes place with its unbiotinylated counterpart. We have also shown that at least 90% of the B-11-dUTP can be recovered by increasing the mobile phase NaCl concentration to 0.1 M. We are currently testing the selectivity of this technique using paramagnetic particles for recovering biotinylated oligonucelotides containing 20 mers when the base sequence and degree of biotinylation is varied.

Another venue for exploring the use of soft metal affinity interactions is in the selective recovery of biotin labeled cells. Paramagnetic particles with immobilized soft metal ions are being used to sequentially pan for specific cell types in the context of complex mixed populations. Magnetic cell sorting is currently available using streptavidin coated paramagnetic particles [7-9], however, this process can not be conveniently performed in a repetitive fashion. Three impediments exist in this process: (1) because of the high affinity of the biotin-streptavidin interaction it is all but impossible to separate the cell from the magnetic bead at this juncture; (2) although methodologies exist to disrupt the antibody-cell surface protein interaction, these methods are inefficient and expensive; and (3) current methods using paramagnetic particles can also lead to cell membrane rupture during particle detachment from the cell. In addition, the methods are antibody specific, and in most cases reagents do not exist to disrupt many of the available antibodies. In using soft metal ions for mouse white cell recovery, we have demonstrated that a low chloride ion buffer with higher concentration of phosphate in order to match the ionic strength of phosphate buffered saline can be used at 4 ^oC without loss of cell viability. We have also demonstrated that white cells recovered by these paramagentic particles can be easily detached from the particles by addition of phosphate buffered saline. Observation of cell viability after binding and release using trypan blue has also confirmed that the targeted cells remain viable during this procedure.

References

[1] J. Porath, J. Carlsson., Nature (London), 258 (1975) 598.

[2] J. Porath and B. Olin, Biochemistry, 22 (1983) 1621.

[3] J. Porath, G. Dobrowolska, and G. Muszynska, Journal of chromatography, 541 (1991) 333.

[4] R. G. Pearson, Chemistry in Britain, 3 (1967) 103.

[5] T. Takahashi, T. Mitsuda, and K. Okuda, Analytical biochemistry, 179, (1989) 77.

[6] X. Li, and W. M. James, Biotechnic and histochemistry, 70 (1995) 234.

[7] Leclercq, G., De Smedt M., and Plum, J., Interleukin-2 stimulated T cell receptor V gamma 3 positive thymocytes do not migrate to the skin, Immunology Letters , 28(2), (1991) 135-142.

[8] Elbe, A., Kilgus, O., Strohal, R., Payer, E., Schreiber, S., and Stingl, G., Fetal skin: a site of dendritic epidermal T cell development, J. Immunol., 149(5), (1992) 1694-1701.

[9] Qin S; Cobbold SP; Pope H; Elliott J; Kioussis D; Davies J; Waldmann H. "'Infectious' transplantation tolerance", (1993) Science 253, 974-977.

Presenting Author: Antonio A. García

email: tony.garcia@asu.edu

Kunthala Jayaraman

No abstract available.

Karen A. McDonald, N.-J. Remi Shih, and Alan P. Jackman University of California, Davis, Department of Chemical Engineering and Materials Science, Davis, CA 95616

We are studying the production of plant defense proteins from plant cell suspension cultures in shake flasks and bioreactors. Of particular interest are the ribosome-inactivating proteins (RIPs) (see reviews by Barbieri et al. (1993), Girbes et al. (1996), Hartley et al. (1996) and Stirpe et al (1992)), specific rRNA glycosidases, which remove adenine at a well conserved site on ribosomal RNA causing the shut down of protein synthesis in both eucaryotic and procaryotic cells. RIPs have potential applications as antiviral therapeutics (McGrath et al., 1989), immunotoxins for cancer therapies, and genetic engineering of crops for improved disease resistance (Lodgeman et al., 1992).

In our laboratory we are studying the production of RIPs from plant cell suspension cultures of Trichosanthes kirilowii, a member of the Cucurbitaceae family found in Japan, Korea and China. We have purified and characterized 4 novel RIPs, ranging from 15 kDa to 35 kDa in size, from the plant cell culture broths; these proteins possess the specific rRNA N-glycosidase activity which is characteristic of RIPs and inhibit protein synthesis in cell free translation systems.

We have also investigated the kinetics of growth and RIP production of T. kirilowii plant cell suspension cultures in 5 L agitatated, sparged bioreactors under well controlled conditions of temperature (27 C), dissolved oxygen concentration (70% sat), and agitation (50-100 rpm using a pitched blade impeller).

We have found that the cultures reached final biomass concentrations as high as 20 g dw/L and a doubling time of less than 2 days during the exponential growth phase. The lag phase for these cultures was variable, ranging from 4-11 days, depending on the inoculum source and inoculation density.

Extracellular RIP production was observed in the culture broths during the exponential growth phase and reaches a level of 50-58 units of activity (1 unit = protein concentration causing 50% inhibition of protein synthesis in a cell-free translation system using rabbit reticulocyte lysate), however, RIP activity declined during the stationary phase presumably due to proteolytic degradation. Sample results for extracellular RIP production for one of our batch cultures are shown in the figure below

Work is currently underway to purify and characterize the extracellular RIPs from these cultures. We are also investigated alternative bioprocessing strategies for enhanced production of RIPs from plant cell cultures.

REFERENCES

Presenting author: McDonald, Karen
e-mail: kamcdonald@ucdavis.edu
http: www.engr.ucdavis.edu/~pcdhome/karen/KMHOME.htm

Clint B. Pepper

No abstract available.

A. Amanullah, R. Blair, A. Davies, G.R. Riley, C.R. Thomas and A.W. Nienow

Centre for Biochemical Engineering, School of Chemical Engineering,

The University of Birmingham, Edgbaston, Birmingham B15 2TT, UK.

Mechanical forces due to agitation can influence the morphology of submerged cultures of filamentous fungi. These changes in turn can affect rheology, mixing and mass transfer and hence perhaps ultimately productivity. Therefore, agitation conditions need to be selected with great care. This paper reports on the effects of agitation speed on mycelial fragmentation and its effect consequently on protein production using a recombinant strain of Aspergillus oryzae. Constant mass chemostat cultures at controlled dissolved oxygen levels (using gas blending) have been conducted at 550, 700 and 1000 rpm at a dilution rate of 0.05h^-1. Such a set up allows the variation of agitation speed and hence gas hold-up without affecting the dilution rate or the level of dissolved oxygen. Thus the effect of agitation speed alone can be investigated without any simultaneous influence of mass transfer. The morphology of both freely dispersed mycelia and complex clumps (and hence the whole of the biomass) have been analysed using image analysis. Statistical analysis showed that it was possible to obtain steady states with respect to morphology. The mean projected area representing the total biomass at each steady state under growing conditions correlated well with a "energy dissipation/circulation" function.

The effects of agitation conditions in a 1.4 litre off-line mixing vessel using different impeller types on the fragmentation of culture taken from the chemostat have also been investigated. Radial flow (Rushton turbine) and axial flow (pitched blade and Prochem Maxflo T) impellers, with specific power inputs in the range 0.2 - 4 W/kg were employed. The changes in mycelial morphology could also be correlated with a "energy dissipation/circulation" function and the dependency of the mean projected area representing the biomass on this parameter has thus been determined under both growing and non-growing conditions.

The effects of agitation conditions on the mean number of tips per hypha is discussed and a possible explanation of increased tip formation at high agitation intensities is suggested. Transients of morphological parameters for the freely dispersed mycelia in response to a speed change from 1000 to 550 rpm is reported and agglomeration effects are proposed to account for the rapid changes in the freely dispersed morphology following the speed change. Protein production as a function of agitation speed has been measured and the problems of heterologous DNA loss in encountered in prolonged continuous cultures are discussed. An understanding of how agitation affects mycelial morphology and productivity ought to be valuable in optimising the design and operation of large scale fungal fermentations for the production of recombinant proteins.

Corresponding author: Amanullah, A.

e-mail: A.Amanullah@bham.ac.uk

Eva Ståhl Wernersson and Christian Trägårdh

Dept. of Food Engineering, Lund University, Box 124, 221 00 Lund, Sweden

Scale-up is important in the design of bioreactors for full scale production. It is also significant in the prediction of the performance for a certain process, when parameters are altered in an existing reactor. The fluid dynamics of the reactor is closely linked to scale-up and mixing, since it is the convective and turbulent flow pattern which initially distributes heat and mass in the reactor. The turbulent transport rate is orders of magnitude larger than the molecular transfer rate and turbulence is therefore the dominant and process controlling factor. Fluid dynamics is a powerful approach to analyze the flow and to relate local flow conditions with global parameters.

A measuring and data analysis system for turbulence measurements was developed for agitated tanks of sizes and geometries ranging from pilot scale to industrial production scale, Fig.1. The operating conditions were those that normally exist in fermentation processes. The measuring method used was Constant Temperature Anemometry (CTA) with a two-split film probe, where velocity signals were sampled simultaneously in two flow directions. Sampling times were up to 7 minutes and sampling rates were several kHz. This enabled the detection of long-term periodicities in the flow. Microfilatrated water was used as liquid and measurements were done both in the impeller zone and in the bulk zone.

Global scale-up parameters such as the impeller tip speed and power input per unit mass, showed correlations for the flow at several positions in the impeller zone and in the bulk zone for the three tanks, Fig. 2. Local turbulent parameters, the turbulent kinetic energy, q, and the local energy dissipation rate, ε, were scaled with the local parameter, the convective velocity U_conv, which confirmed the concistency of the flow in the high intensity turbulent flow of turbine-agitated tanks, Figs. 3 and 4. The convective velocities were estimated from an expanded Taylor hypothesis which consider both the mean velocity and the fluctuations of the flow.

Fig. 1: View of tanks A (0.75 m³), B (12 m³) expressed in dimensionless coordinates. and C (30 m³).The tank diameters were 0.8 m, The baffles in tanks A and C were

1.88 m, and 2.09 m. The tank to impeller positioned at 2r/d_imp = 2.4 and in tank B diameter ratios were 3.00, 2.47 and 3.00. at 2r/d_imp=1.97.

Fig. 3a and b: Turbulent kinetic energy, q, normalized to a) the convective velocity squarred, U_conv² (left) b) the impeller tip speed V_tip² (right).

Fig. 4a and b: Local energy dissipation rate ε normalized to the convective velocity over the impeller diameter U_conv³/d_imp (left). Local energy dissipation rates scaled with the power input per unit mass for tanks A, B and C, compared with literature data (right).

* The similarity of the local flow at high local Reynolds numbers was confirmed in the tanks of different geomety and size.

* Correlations between local turbulent parameters and global parameters were independent of tank geometry, with approriate chosen dimensional lengths.

* The inhomogeniety of the flow conditions were confirmed by the different magnitudes in the bulk zone and the impeller zone of the local turbulent parameters.

* Long-term periodicities were detected both in the mean flow and in the fluctuations of the flow.

Reference: Ståhl Wernersson E., Some Fluid Dynamic Characteristics in the Scale-up of Rushton Turbine-Agitated Tanks, thesis, 1997, Lund.

Presenting author: Ståhl Wernersson Eva

e-mail: eva.stahl_wernersson@livstek.lth.se

O.K. Lyngberg¹, V.S. Thiagarajan², J. Liu², M.L. Cruz³, J.L. Schottel³, and M.C. Flickinger^2,3

1Dept. of Chemical Engineering and Materials Science, ²Biological Process Technology Institute, ³Dept. of Biochemistry University of Minnesota, St. Paul, MN 55108

An improved method¤ has been developed for study of immobilized cells in a well defined environment of latex dots of known thickness and volume. This method generates dots of immobilized cells in an aerobic environment with good cell retention properties and high viability after rehydration of the film. The dots have a thickness of 35 um ± 1um and a total volume of approximately 4.4 ul, and they hold about 109 cells. Cells are immobilized in latex dots created by patch coating a flat substrate with good adhesion properties. First, a template is

made by cutting holes into a lined pressure- sensitive polyester sheet. The template is stuck to the substrate to be coated by removing the liner and rolling it on to the substrate. A paste of cells, acrylic/vinyl acetate latex and 10% glycerol is coated on top, and the template is removed just as the coating becomes flow stable. A second template slightly higher than the cell patches is laid down beside the patches. One or more sealing topcoats are coated on top of the patches. After drying, the film can be rehydrated. After overnight rehydration, b-galactosidase activity can be induced in the immobilized cells. The LacZ protein as well as RNA can be extracted from the dots for further analysis. This system is being used to investigate gene expression in non-growing E. coli using various promoter/lacZ fusion constructs.

• Swope, K.L. and M.C. Flickinger. 1996. The use of confocal scanning laser microscopy and other tools to characterize Escherichia coli in a high cell-density synthetic biofilm. Biotech. Bioeng. 52:340-356.

Presenting author: Schottel, Janet

V.S. Thiagarajan¹, Z. Huang², M.C. Flickinger^1,3, J.L. Schottel³ and L.E. Scriven²

1Biological Process Technology Institute, ²Department of Chemical Engineering

and Materials Science, and ³Department of Biochemistry

University of Minnesota, St. Paul, MN 55108

Cryogenic scanning electron microscopy (Cryo-SEM)¤ has been employed to elucidate the microstructure of a synthetic biofilm made for use as an immobilized whole cell biocatalyst. This biocatalyst consists of Escherichia coli cells entrapped between latex polymer particles. Latices are stabilized colloidal suspensions of polymer particles which, when coated as a thin layer and let dry, form a continuous film of polymer. Our synthetic biofilms are 20 to 40 microns thick and consist of a cell coat and a top coat. The cell coat is made by coating a mixture of E. coli, glycerol and latex. The top coat is made from a mixture of latex and glycerol and seals the cells in the matrix. These coats have been examined separately and together. Topographical images (surface) and fracture images (cross section) show E. coli cells (1-1.5 microns) surrounded by the polymer particles of monodisperse (~0.28 microns) and polydisperse (0.05-1.75 microns) latices. The images also show the effects of drying and rehydration on these films. The micrographs reveal that the latex polymer particles weld around the bacterial cells thus physically entrapping them. Latex polymer particles develop contact, deform and coalesce as drying progresses. Glycerol protects the cells during the stressful drying and film formation process of the cell coat. Glycerol retards contact development between particles and perhaps the rate of coalescence (welding) of the particles as well. It enables a more porous structure of the top coat, probably by forming a film around particles and/or by being occluded between particles. There is a gradient in the degree of coalescence through the thickness of the film. The particles toward the top coalesce to a greater degree than those near the bottom. Rehydration causes swelling of the film and appears to partially reverse particle deformation and perhaps coalescence as well. This process might, on hydration, result in increased porosity in some latex films without glycerol. The micrographs also reveal empty spaces within the film structure when the bacteria are washed out of the film without a top coat. Bacteria apparently occupied the spaces when the film was first formed. The matrix surrounding the cells is completely microporous with no pores larger than the size of the cells. Cryo-SEM is being used to evaluate latices for permanent entrapment of E. coli in multi-layer biofilms for whole cell biocatalysis.

• Thiagarajan, V., Ming, Y., Scriven, L.E. and Flickinger, M.C. 1996. Cryo-electron microscopy of polymer particles in a high cell density synthetic biofilm. In "Immobilized Cells: Basics and Applications", (R.H. Wijffels, R.M. Buitelaar, H.S. Wessels, C. Bucke and J. Tramper, eds.) Elsevier Science B.V.

Presenting author: Schottel, Janet

Guido Greco jr., Domenico Pirozzi

Dipartimento di Ingegneria Chimica, Università Federico II

P.le Tecchio, 80 - 80125 Napoli - Italia

In nonaqueous environments, enzymes exist as aggregates due to their poor dispersion in organic media. Obviously, aggregates enhance the irreversible self-association of proteins. In turn, the latter leads to conformational and covalent changes and, eventually, to complete deactivation. Only few studies are available on this subject, that refer to room temperature conditions. The influence of physico-chemical characteristics of nonaqueous media on interproteic interactions at high temperatures still is still largely unknown.

In the present study, the long-term stability of -chymotrypsin (EC 3.4.21.1, from bovine pancreas) is characterised with reference to protein-protein interactions. The effect of both covalent and noncovalent interactions on the irreversible intermolecular aggregation of solid-state proteins has been examined.

The experimental results indicate that the formation of disulfide bonds is the dominating mechanism implicated in protein aggregation and that, in the course of deactivation, the amount of enzyme molecules involved in the formation of noncovalent linkages reaches a maximum, subsequently decreasing. Thus, although a relevant fraction of proteins is initially insolubilised by noncovalent interactions, intermolecular, covalent bonds tend to involve all aggregated molecules.

Presenting author: Greco, Guido jr

E-mail: grecog@ds.unina.it

Kristin A. Yarema and Douglas S. Clark

Department of Chemical Engineering, University of California, Berkeley, CA 94720

Growth factors are small proteins that act as signaling molecules to initiate cell growth. These proteins are secreted by cells, only to later bind to specific receptor proteins located in the outer plasma membrane of the same or neighboring cells. Growth factors and their receptors are the subject of intense interest, as they are linked to cell growth and repair processes as well as some cancers in ways which remain poorly understood. Of the large array of presently discovered growth factors and specific receptors, the standard model still remains the epidermal growth factor (EGF) protein family and their corresponding receptors, including the epidermal growth factor receptor (EGFR). Important questions yet unanswered include the exact mechanism of the binding/activation event, what steps and other proteins are involved in the downstream signaling process, what defects occur in cancerous cells, and what functional differences exist among the different proteins in the EGF and EGFR families. We focus on the binding and activation phenomena normally taking place in the cell plasma membrane, an environment amenable to study using electron spin resonance (ESR) spectroscopy.

We present the results of experiments using spin-labeled EGF and EGFR species, either singly or in combination, in both solution and whole-cell environments. Spin labels have been covalently attached to hEGF at the N-terminus and the methionine-21 residue, and specific affinity labels have been synthesized to label hEGFR. These affinity labels are based upon the 4-(3-bromophenylamino) quinazoline family of tyrosine kinase inhibitors. Our ultimate aim is to develop a system for using such spin labels to observe changes in receptor and ligand populations in cells using ESR spectroscopy.

Corresponding author: Clark, Douglas S.

Robert Clyde

Clyde Engineering, POB 740644, New Orleans, La. 70174

Oxygen transport is an important factor in plant and mammalian cells that produce drugs. Yarmush (1) describes for hepatocytes and Thilly (2) for mammalian cells. Blanch (3) says it is a major limitation for hybridomas and in a recent article (4),Heijnen discusses plant cells. Cephalosporin is strongly affected by the availability of oxygen, says Karaffa(5). HeLa, and CHO cells also need oxygen.

Patent 5,256,570 tells how to get oxygen into cells quickly. Claim 4 mentions diatoms. We have a method of fluffing up the fibers to entrap Celite. Agathos produces cyclosporine with Celite,(6).In the patent,corrugated fibers with holes in the valleys are rotated in a half full horizontal unit. Liquid is carried up into the vapor space and falls down through the holes. Mass transfer to drops is much faster (7). Maas (8) says it's hard to get dissolved oxygen above one ppm.

White rot fungus also needs oxygen. It degrades many chlorine compounds and PAH's (polycyclic aromatic hydrocarbons). There are over 2,000 sites contaminated with PAH's from old coal burning plants. When coated on old cardboard boxes and buried in soil, oxygen is entrapped in the corrugations for growth of the fungus. Bogan and Lamar describe PAH destruction (9) and say that Phanerochaete laevis is better than chrysosporium (10). Azo dyes are degraded with the fungus (11).PCB's (12), pentachlorophenol (13), and TNT (14) are degraded with the fungus. We have found that the fungus grows twice as fast using corrugations. Fahy decolors molasses spent wash from an ethanol fermentation (15).

Bacteria on fibers remove toxic metals such as lead, chromium, selenium, strontium and uranium. Lead can be removed in 2 sec. with Zymomonas mobilis on Tyvek fiber. Some people think that all bacteria are poisonous, but Morais found (16) that Zymomonas is therapeutic for stomach ailments. Patent 4,530,763 describes bacteria on fibers to remove metals.

Cells also grow on foam. When a steel screen is put around it, it is made heavier so it can be fluidized as in patent 4,333,893. Kargi used this method to remove COD and nitrogen (17).

REFERENCES

Presenting author: Clyde, Robert A.

The effects of coexpressing other endoplasmic reticulum localized proteins that include thioredoxin-like domains will also be discussed.

 

In addition to PDI, we have cloned and expressed the chaperone, immunoglobulin heavy chain binding protein (BiP) and observed that the secreted IgG levels increased by 90% after several days baculovirus infection (Tsu et al., Prot. Exp. & Purif., 1994, Tsu and Betenbaugh, Biotech. Prog., 1997). However, assembled antibodies also accumulate intracellularly at later times in the infection to indicate that a post-assembly step may inhibit protein secretion. In order identify the limiting step in the secretion pathway and to elucidate the oligosaccharide processing pathways in insect cells, we have, in collaboration with Y. C. Lee and Noriko Takahashi, examined the N-linked oligosaccaharides attached to intracellular and secreted antibodies using 2-dimensional chromatographic techniques (Tsu et al., J. Biol. Chem., 1997). Substantial differences have been observed in the oligosaccharide content of intracellular and secreted glycoproteins. The intracellular glycoforms appear to include 50% high mannose type structures (Fig. 2a). In contrast, the secreted glycoproteins include principally complex and hybrid glycoforms (Fig. 2b and 2c):

These studies indicate that significant similarities and some differences exist between mammalian and insect-derived processing pathways. Insect cell processing appears to diverge from mammalian processing in pathways leading to low mannose structures, in the presence of a(1,3) fucose linkages, and in the absence of sialylation. Methods are now under consideration to manipulate the oligosaccharide processing and transport pathways in order to make insect cell pathways even more similar to those from mammalian cells.

Edgard Carvalho¹, Colleen Merritt¹, Joe Chappell² and Wayne Curtis¹

1The Pennsylvania State University, Department of Chemical Engineering

108 Fenske Lab, University Park, PA 16802

Sarah W. Harcum¹ and William E. Bentley²

1New Mexico State University, Department of Chemical Engineering, PO Box 30001, SC 3805 Las Cruces, New Mexico, 88003

and

James T. Hsu

Biopharmaceutical Technology Institute, Department of Chemical Engineering,

Lehigh University, Bethlehem, PA 18015

Polymerase chain reaction (PCR) is an in vitro method to carry out amplification of selected nucleotide (DNA or RNA) sequences through a succession of incubation at different temperatures. The double stranded DNA is heat denatured, cooled to promote primer annealing, and finally, heated to an intermediate temperature to facilitate the extension reaction. This process is repeated several times to generate a large number of copies of the specific DNA target segment to carry out further analysis. The key factor in this reaction is the thermostable Taq DNA polymerase enzyme which has a higher optimum temperature of activity (72E-80E C) for DNA synthesis and can withstand repeated exposure to high temperatures (94E-96EC). The discovery of this enzyme has enhanced many areas of biological research including molecular biology, biotechnology, and medicine.

Theoretically, in a PCR process, the DNA copy number should increase in logarithmic fashion with respect to the cycle number, because the amount of DNA doubles after each discrete cycle of amplification. This theoretical amount is difficult to achieve because a large number of variables control the reaction rate and the fidelity of the PCR process. A small difference in any of the variables affects the yield and the purity of the product dramatically. So, the concentrations of individual reaction components, time and temperatures of various steps, and the number of cycles must be optimized for efficient amplification as well as for maximum yield.

A mathematical model for polymerase chain reaction (PCR) is developed taking into account the three steps in this process, i.e. melting of DNA, primer annealing, and DNA synthesis (polymerization). Activity and deactivation of the polymerase enzyme as a function of temperature is incorporated in the kinetic model to get a better understanding of the amplification of DNA. Computer simulation of the model is carried out to determine the effects of various parameters such as the cycle number, initial DNA concentration (copy number), initial enzyme concentration, extension time, temperature ramp, and enzyme deactivation on the DNA generation.

Presenting author: Hsu, James

E-mail: jth0@lehigh.edu

Shravan Nagarajan and Randy S. Lewis

School of Chemical Engineering, 423 Engineering North, Oklahoma State University

Stillwater, OK 74078

Nitric oxide (NO) is a recently discovered biochemical messenger and cytotoxic agent synthesized by many cells including macrophages, neutrophils, and endothelial cells. Several physiological functions of NO include blood pressure regulation, neurotransmission, and inhibition of platelet aggregation and adhesion. Several pathological effects of NO and the reaction products of NO with oxygen or superoxide include DNA damage, lipid peroxidation, and enzymatic inactivation. In order to quantify the physiological and pathological effects of NO and its reaction products on cell systems, it is important to control the NO concentrations to which cell systems are exposed.

A system for exposing a specified NO concentration to a cell system has been developed and modeled. A specified NO concentration to which cells are exposed is maintained by the controlled delivery of gaseous NO to solution via a semi-permeable membrane (see figure) following which the aqueous NO solution continuously flows through a well stirred chamber (20 ml). Cells attached to a circular polymeric plate are placed in the bottom of the chamber. The delivery rate of NO to solution and the subsequent NO concentrations prior to and following the chamber have been modeled at both 23 ° C and 37 ° C for a flow rate of 3 ml/min and agree very well with experimental data.

An advantage of the NO delivery system described above is that the NO concentration to which cells are exposed is predictable based upon system parameters such as the fluid flow rate and the gaseous NO concentration. In addition, the concentration can be varied with time or maintained at steady state. However, the model predictions only include the reaction of NO with oxygen. Therefore, if other significant reactions with NO occur they must be included in the model. Another advantage of this system is that the reaction products of NO, such as nitrite, are continuously removed from the chamber. Thus, the reaction product concentrations do not increase with time such that their effect(s) upon cells is minimized. Compounds which release NO have been utilized in several cell studies in closed chambers. However, the continual increase of reaction products and the non-constant release of NO leading to non-steady state NO concentrations illustrates that the NO delivery system described above is advantageous to using NO-releasing compounds for in vitro cell studies.

Presenting author: Lewis, Randy S.

e-mail: randyl@okway.okstate.edu

M.B.F. Martins ⁽¹⁾ , S.I.D. Simões⁽¹⁾, A. Supico⁽¹⁾, M.E.M. Cruz⁽¹⁾, R.Gaspar⁽²⁾.

(1) Grupo de Bioquímica I - Departamento de Biotecnologia-IBQTA-INETI, Estrada do Paço do Lumiar, 1699 LISBOA Codex, PORTUGAL.

(2) Fac. Farmácia e Centro de Neurociências da Universidade de Coimbra, R. Norte, 3000 Coimbra. PORTUGAL.

An increased number of enzymes, or other proteins, with potential for medical therapy are produced nowadays. Meanwhile their effectiveness is limited by their low stability and unwanted reactions after in vivo administration. One of the approaches to overcome those problems is the incorporation in colloidal carriers (Torchilin, 1991; Kreuter, 1994).

Due to the complex structural characteristics of enzymes and the need for preservation of their native conformation, the construction of appropriate carriers requires the optimization of several process parameters. Polyisobutylcyanoacrylate nanoparticles, produced by an emulsion polymerization process, are one of the carriers extensively studied for the incorporation of small molecules, although only a scarce number of studies were devoted to the incorporation of enzymes (Kreuter, 1994; Martins et al., 1996 and 1997). The absence of organic solvents in the medium of polymerization is the more obvious advantage of this process of construction of polymeric colloidal carriers. Although the low pH, the presence of stabilizers and the high reactivity of the monomer, can limit the retention of activity of incorporated enzymes.

The performances of the incorporation of L-asparaginase and superoxide dismutase, enzymes with very different 3D structure, in this type of colloidal carriers are evaluated in this work. Optimization criteria used was the maximization of enzyme load and the minimization of the lost of catalytic activity.

The efficacy of incorporation was obtained by the ratio: (E_f/E_i) x100, being E_f the enzyme content in nanoparticles and E_i the enzyme quantified in the polymerization medium before addition of the monomer. The retention of catalytic activity was obtained by the ratio: (A_i /A_t ) x 100, being A_i the catalytic activity on intact nanoparticles and A_t the catalytic activity of an enzyme solution (in polymerization medium containing all the components except the monomer) at a concentration equal to the initial enzyme concentration present in the polymerization medium before the addition of the monomer.

An increase of incorporation of enzyme and a decrease in the efficiency of incorporation as a function of the increase of initial enzyme concentration was observed. The retention of activity after disruption of nanoparticles points to a decrease on the efficiency of the disruption process with the increase of enzyme load.

The analysis of results of incorporation of enzymes in nanoparticles must consider both the interaction of the enzyme with the emulsion polymerization system and the effect of process of polymerization on the characteristics of the enzyme. The identification of critical parameters of the process was performed considering the mechanism of emulsion polymerization in order to identity critical parameters on process scale-up.

Basic groups of enzymes can act as polymerization starters, with the correspondent covalent linkage of the enzyme to the polymer. The extend of the capture of enzymes during the enlargement of the small oligomeric subunits that form the nanoparticles (Couvreur and Vauthier, 1991) can be dependent on interactions of the enzyme with the emulsifiers or with micellar structures. When the emulsifier concentration in solution falls below the critical micelle concentration the micelles become unstable and dissolve; polymerization continues in polymer particles with emulsifier molecules adsorbed on them (Heller, 1987). Also in this step the interactions between enzyme and emulsifier can interfere in the incorporation of the enzyme in nanoparticles. At high degree of conversion monomer droplets disappear completely and all the unreacted monomer or growing unterminated polymer molecules are contained in primary polymer particles (Heller, 1987). The termination frequency in this environment is reduced due to the reduced H⁺ ions concentration, responsible by termination. The presence of unreacted monomer can affect the characteristics of the enzyme incorporated in nanoparticles.

The performances of the incorporation of enzymes in PIBCA nanoparticles can be compared with results considering other systems (Tanswell and Freeman, 1987; Cruz et al., 1994). The influence of the process parameters and the number of constraints in selecting the process parameters was evidenced by the results of the effect of pH, enzyme molecular weight, enzyme concentration and presence of molecules like enzyme substrate, on the incorporation of enzymes.

Couvreur, P. and Vauthier, C., Polyalkylcyanoacrylate nanoparticles as drug carriers: present state and perspectives. J. Cont. Release, 17 (1991) 187-198.

Cruz, M.E.M., Martins, M.B., Corvo, M.L., Jorge, J.C.S. and Gaspar, M.M., Native and lipophilic derivatives of asparaginase and SOD and respective liposomal forms. Proceed. Intern. Symp. Control. Rel. Bioact. Mater., 21 (1994) 346-347.

Heller, J., Fundamentals of polymer science. In: Robison, J. and Lee, V. (Eds), Drugs and the Pharmaceutical Sciences, Vol. 29, Marcel Dekker, Inc., New York, 1987, pp. 139-177.

Kreuter, J., Nanoparticles. In: Kreuter, J. (Ed.), Colloidal Drug Delivery Systems. Marcel Dekker, Inc., New York, 1994, pp. 219-342.

Martins, M.B.F., Simões, S.I.D., Supico, A., Cruz, M.E.M. and Gaspar R., Enzyme-loaded PIBCA nanoparticles (SOD and L-ASNase): optimization and characterization Int. J. Pharm., 142 (1996) 75-84.

Martins, M.B.F., Supico, A., Simões, S.I.D., Gaspar R. and Cruz., M.E.M., An analytical methodology to quantify the incorporation of enzymes in polyalkylcyanoacrylate nanoparticles. J. Pharm. Biomed. Anal., (1997) in press.

Tanswell, A.K. and Freeman, B.A., Liposome-entrapped antioxidant enzymes prevent lethal O2 toxicity in the newborn rat. J. Appl. Physiol. 63 (1987) 347-352.

Torchilin, V.P., Therapeutic immobilized enzymes for parenteral application. Immobilized Enzymes in Medicine. Progress in Clinical Biochemistry and Medicine, Vol. 11, Springer-Verlag, Berlin, 1991, pp. 29-124.

Presenting author: Martins, Bárbara Figueira

e-mail: barbara.martins@ ibqta.ineti.pt

Alison J. Mastrangelo^*, J. Marie Hardwick⁺, Michael J. Betenbaugh^*

* Department of Chemical Engineering, The Johns Hopkins University, 3400 North Charles Street, Baltimore, Maryland 21218; ⁺Department of Molecular Microbiology and Immunology, The Johns Hopkins School of Hygiene and Public Health, Baltimore, Maryland 21205

Apoptosis, the process by which a cell effects its own demise, is a tightly regulated cascade under the control of many different gene products. In particular, Bcl-2 and Bcl-x_L, two members of the Bcl-2 family, have been shown to limit or completely prevent the apoptotic process in response to a variety of external insults, such as nutrient limitation, toxin accumulation, and viral infections. In this work, stable transfectants of mammalian cell lines have been generated which express bcl-2 or bcl-x_L. These cells have been exposed to recombinant viral infections in an effort to inhibit viral induced cell death and improve heterologous protein production.

During initial studies, AT3-bcl2 cells infected with a Sindbis viral vector containing the gene for chloramphenicol acetyltransferase (CAT) consistently maintained viabilities close to 100% (Figure 1) and a growth rate equivalent to that of uninfected cells, 0.040 h^-1. However, the same vector rapidly induced apoptosis in the AT3-neo cells, which were all dead by three days post-infection (Figure 1). CAT levels were also measured in both cell types (Figure 2), and although infected AT3-neo cells initially generated higher levels of heterologous protein, CAT activity in these cells fell to zero by two days post-infection. In contrast, CAT was present in AT3-bcl2 cells for almost a week, reaching a maximum level of 580 mUnits per well. Furthermore, recombinant protein production in AT3-bcl2 cells was extended and amplified by the regular addition of virus to the culture medium, a process which resulted in expression for the duration of the cell culture process (Figure 3). We are currently investigating the abilities of bcl-2 and bcl-x_L to protect other cell lines, such as baby hamster kidney (BHK) and Chinese hamster ovary (CHO) from viral-induced apoptosis.

Using anti-death genes, however, may not always be the best way to block apoptosis. The protective effects of certain genes may be cell line specific and therefore, may not be as effective in preventing cell death as other treatments. For instance, it has recently been demonstrated that certain chemicals are effective inhibitors of the apoptotic process. In particular, N-acetylcysteine (NAC), bongkrekic acid (BA), and Z-VAD have been shown to delay death in a variety of cell lines. Preliminary studies in our laboratory indicate that both NAC and BA are capable of retarding apoptosis in AT3-neo cells infected with a Sindbis viral vector (Figure 4). Further development of these methods of inhibiting cell death, both genetic and chemical, will be a valuable technology to extend the productive lifetimes of cells in culture.

Figure 4: Survival of AT3-neo cells upon dsSV-CAT infection. The protective chemicals N-acetylcysteine and bongkrekic acid were added as indicated.

Presenting author: Alison J. Mastrangelo

e-mail: alisonm@jhuvms.hcf.jhu.edu

Lee Mermelstein*, Kelly Wong, and Diane Blumenthal

Department of Fermentation Process Development, Manufacturing and Engineering, Scios Inc.

2450 Bayshore Parkway, Mountain View, CA 94043

The ability to determine the number of plasmids or integrated recombinant genes that exist within a genetically engineered clone is fundamental to the evaluation and optimization of microbial expression systems. Furthermore, from a regulatory standpoint, copy number quantification is important, especially for processes using chromosomally integrated expression systems. Previously described methods for quantification of copy number based on hybridization of nucleic acid probes to chromosomal DNA utilize dot blots, Southern blots, or quantitative PCR. These methods are time consuming and labor intensive or suffer from problems of poor reproducibility, accuracy, or precision. Previously described methods for the determination of copy number which do not utilize hybridization rely on the quantification of electrophoretically or chromatographically separated DNAs or densitometric scanning of images. These methods suffer from problems of poor reproducibility, accuracy, linearity, specificity, or precision. Furthermore, these methods cannot be used for quantification of integrated gene copy number.

A rapid method for the determination of copy number in plasmid based or chromosomally integrated systems which does not suffer from any of the shortcomings described above has been developed using the Molecular Devices Threshold system and Immunoligand Assay Reagents. The method relies on the hybridization of pairs of oligonucleotide probes complementary to proximally located segments of either host chromosomal DNA or the recombinant DNA of interest (located chromosomally or within a plasmid). One of each pair of probes is biotinylated while the other is fluoresceinated. Restriction enzyme digested total DNA is thermally denatured then annealed to a pair of probes in aqueous reactions. The hybridization reaction is subsequently saturated with streptavidin which, due to its polyvalent biotin binding ability, allows the hybridized DNA fragments to be bound to biotinylated nitrocellulose membranes. An anti-fluorescein urease conjugate is then bound to the immobilized hybridization fragments. Up to 48 test reactions may be simultaneously immobilized to discrete regions of membranes. A Light Addressable Potentiometric Sensor (LAPS) is used to measure the rate of increase in pH due to the reaction of the membrane associated urease with urea. This rate is proportional to the quantity of DNA which has hybridized to the pair of probes. Separate reactions are carried out using probes specific for host chromosome and the recombinant DNA of interest and the results are compared in order to quantify the copy number. The assay can be completed in less than 2 hours using restriction enzyme digested DNA.

Ribosomal RNA (rRNA) gene sequences were used as chromosomal probe targets because they exist in multiple copies and therefore give a larger signal than single copy targets. rRNA genes are also more highly conserved than multi-copy translated genes. Since no sequence data was available for Pichia pastoris rRNA genes, we cloned a fragment of the 18s rRNA gene to provide target sequence data for use in probe design. For Escherichia coli plasmid quantification probe targets specific for the ColE1 origin of replication were used. These probes should be useful in the copy number determination of any ColE1 based plasmid, which accounts for the majority of the expression vectors used for recombinant protein expression in E. coli. In all cases probes were 20 to 40 nucleotides long. The size of the restriction endonuclease fragment target to which the pairs of probes hybridized was relatively small (100-200 nucleotides) to minimize displacement of the probes, during annealing, by the strand of DNA complementary to the target.

The assays were optimized with respect to probe concentration, annealing temperature, and annealing time. Specificity of the assay is high because the likelihood that both probes would hybridize to the same incorrect fragment (as would be necessary to produce non-specific signal from DNA) is very low. Precision, accuracy, and reproducibility data are presented and they compare favorably to published copy number quantification methods. The E. coli chromosomal quantification portion of the assay is sensitive enough to detect the effect of growth rate on the number of probe targets. This effect occurs because, in rapidly growing cells, sequences near the origin of chromosomal replication are more prevalent than those near the terminus of replication. When this effect has to be taken into consideration and how it is corrected for is discussed.

Copy number determinations made over the course of a high cell-density fed-batch fermentation of E. coli demonstrated a greater than 20-fold increase in plasmid copy number over the course of the fermentation. The copy number determination method presented should be applicable to the quantification of episomal or chromosomally integrated genes in many other procaryotic and eucaryotic expression systems.

Presenting author: Lee Mermelstein

e-mail: mermelstein@sciosinc.com

John W.M. Mulders, Ton Swolfs and Hans Wijn

N.V. Organon Scientific Development Group,

Room RE 3136 P.O. Box 20, 5340 BH Oss, The Netherlands

Currently, the in vivo biopotency of commercial preparations containing FSH is declared on the basis of the ovarium weight augmentation assay: e.g. Ph. Eur., second edition, 958, (1994). The availability of highly purified human recFSH has allowed a detailed analysis of the hormone by physico-chemical analysis methods. Human (rec)FSH contains approximately 35% (w/w) carbohydrate which is attached to asparagine residues 52 and 78 of the a -subunit and 7 and 24 of the ß-subunit. These oligosaccharides introduce considerable micro-heterogeneity (isohormones) which can be easily visualised by charge-based separation methods.

Isohormones of recFSH, depending on the extent of terminal sialylation and the oligosaccharide branching pattern, have largely differing in vivo bioactivities viz. the specific in vivo bioactivity of isoforms with a pI of 3.5 and 5.5 are approximately 30-40,000 and 200-500 IU/mg protein, respectively. In contrast, acidic isohormones (pI 3.5) are about 2-3 fold less potent than basic isohormones (pI 5.5, de Leeuw et al., 1996). These differences explain why in vitro bioassays or receptor binding assays can not replace the in vivo bioassay.

The relation between pI and in vivo bioactivity of isohormones prompted us to investigate whether knowledge of the quantitative distribution of the isohormones would allow to predict/calculate the in vivo bioactivity. Here we show that there is a linear relation between the amount of isohormones focusing at or below pI 4.9, as determined by densitometry ofIsoelectric Focusing (IEF) patterns, and the in vivo bioactivity (r=0.962, n=25, Figure 1A). Similarly, there is a linear relation between the median migration time (time point at which 50% of the recFSH peak area has passed the UV detector) as determined by Capillary Zone Electrophoresis (CZE) and the in vivo bioactivity (r=0.962, n=25, Figure 1B).

These two models were subsequently used to predict the in vivo bioactivity of 10 other samples of recFSH. Results were compared as described by Bland and Altman (1986) and Garthoff et al. (1995). The mean difference between the predicted values and the mean of the experimental and predicted values (which is regarded as the closest approximation of the true, but unknown, value) was 1.6% ± 4.0% and 1.2% ± 3.1% using the IEF and CZE models, respectively (Figure 2).

Quantitative charge profiling (IEF, CZE) of recFSH offers an easy to perform method which can be used next to/instead of the currently used in vivo bioassay for the potency determination of recFSH. The method will probably also be applicable to the other glycoprotein hormones (LH, CG and TSH) as well as to a number of other glycoproteins (e.g. erythropoietin). These methods may also be used as in process control tests during the development of up stream and down stream processes of recombinant glycoproteins.

A B

Figure 1. Relation between IEF scanning data (panel A) or CZE migration data (panel B) and in vivo bioactivity. For each data set n=25 and r=0.962.

A B

References

• Ph. Eur. (1994) 958
• Leeuw, R. de, Mulders, J., Voortman, G., Rombout, F., Damm, J. and Kloosterboer, L. (1996) Mol. Hum. Reprod. 2: 361-369
• Mulders, J.W.M., Derksen, M., Swolfs, A. and Maris, F. (1997) Biologicals in press
• Bland, J.M. and Altman, D.G. (1986) Lancet 1, 8 Februari, 307-310
• Garthoff, B., Hendriksen, C., Bayol, A. et al., (1995) ATLA 23: 699-712

Presenting author: Mulders, John

e-mail: j.mulders@organon.akzonobel.nl

Torben Lauesgaard Nissen*^#, John Villadsen^#, Morten Kielland-Brandt* & Jens Nielsen^#

# Department of Biotechnology, Technical University of Denmark, DK-2800 Lyngby, Denmark

*Carlsberg Laboratory, Department of Yeast Genetics, DK-2500 Copenhagen Valby, Denmark

During ethanol production in anaerobic cultivations of Saccharomyces cerevisiae, yield is limited by significant amounts of carbon going to biomass and glycerol. Biomass synthesis involves a surplus production of NADH. Under anaerobic conditions this compound cannot be reoxidised in the mitochondria and gives rise to formation of glycerol. The goal of the present project is to shift carbon flux from glycerol formation to ethanol formation by engineering alternative steps to oxidise NADH.

Yeast has two isoenzymes of glutamate dehydrogenase, catalysing the conversion of ammonia and 2-oxoglutarate to glutamate. In vitro measurements have shown that the enzyme activity of the NADPH-dependent glutamate dehydrogenase (Gdh1p) is 50 times higher than the activity of the NADH-dependent glutamate dehydrogenase (Gdh2p). In this study it is investigated whether an increase in the expression level of GDH2 can lead to reoxidation of NADH formed in the biosynthesis and hence, a decrease in glycerol formation.

In a strain derived from the industrial Saccharomyces cerevisiae CBS8066, overexpression of GDH2 will be achieved by substituting the original promoter with the promoter of CHA1, encoding the catabolic L-serine dehydratase, in a strain with a functional Gdh1p and in a strain with a deletion in GDH1. The CHA1 promoter is induced by the presence of serine or threonine in the medium.

The growth characteristics of the strains will be examined under anaerobic conditions in steady state continuous cultivations. Formation of ethanol and secondary products, e.g. glycerol and organic acids, and consumption of glucose, ammonia and serine will be measured. Furthermore, the content of macromolecular components in the biomass, e.g. proteins, carbohydrates and RNA, will be measured.

Finally, a metabolic flux model describing the anaerobic metabolism and product formation in S. cerevisiae will be applied in order to quantify the effect of the genetic changes on the intracellular fluxes.

Presenting author: Nissen, Torben L.

Sun Ho Park^*, Yeon Su Yu^* and JongWon Lee^**

*Department of Chemical Engineering, Keimyung University, Daegu 704-701, Korea

**Department of Biochemistry, School of Medicine, Taegu Catholic University,

Daegu 705-716, Korea

In eukaryotic cells, expression of several genes coding for DNA systhesis and precursor-producing enzymes is induced at the boundary of the G1 to S-phase of the cell cycle. One of the intensively studied enzymes of this class is thymidine kinase(TK), that is indirectly involved in the DNA replication as well as in the salvage pathway of thymidine triphosphate(TTP) formation. The level of TK activity is almost undetectable in quiescent cells, but increasing at least 30-fold as cells enter S-phase to meet the demand for TTP in DNA replication[1]. This regulation takes place on a transcriptional[2] as well as translational level[3]. In this work, we have examined the effect of overexpression of TK on animal cell growth and have studied the growth advantage of TK cloned animal cells with respect to the same cell lines lacking the TK gene.

As shown in Fig.1, retroviral vector(pLtkSN) was used to construct recombinant virus encoding Herpes Simplex Virus Type 1(HSV1tk). The pLtkSN with neomycin resistance gene was prepared by insertion of HSVtk cDNA into pLXSN retroviral vector. To get a high-titer clone, the retroviral vectors were transferred into the packaging cell line(PA317). The recombinant retrovirus was then transduced into target cell line NIH3T3(ATCC CRL 1658). The recombinant cell line was selected with G418 and was tested with virus titer and ganciclovir(GCV) sensitivity assay.

We have found that TK cloned cell(NIH3T3:pLtkSN) outgrew 2.3 times than standard cell(NIH3T3) in DMEM medium containing 10% horse serum. The recombinant cell line also showed a significant enhancement on maximum cell concentration and specific growth rate even at low serum concentrations(Fig.2). Meanwhile, the specific uptake rate of oxygen by the NIH3T3:pLtkSN was considerably reduced compared to that by the NIH3T3. It is evident from our studies that the expression of intracellular TK plays an important role for enhancing the growth characteristics of animal cells, especially in low serum medium.

References

1. Wawra, E., Pöckl, E., Müllner, E. and Wintersberger, E. 1981, J. Virol. 38: 973-981.
2. Liberman, H. B., Lin. P., Yeh, D. and Ruddle, F. H. 1988, Mol. Cell biol. 8: 5280-5291.
3. Knöfler, M., Waltner, C., Wintersberger, E. and Müllner, E. W. 1993, J. Biol. Chem.

268: 11409-11416.

Fig.1. Overview of thymidine kinase cloning and expression using retroviral vector.

Fig.2. Growth curves of NIH3T3(--), NIH3T3:pLtkSN(--) at virious serum concentrations.

Microcarrier concentraion was 10g/L with an average diameter of 122m

Presenting author : Park, Sun Ho

e-mail : park@kmucc.keimyung.ac.kr

Limin Qu

Protein Design Lab, 3955 Annapolis Lane, Plymouth, MN 55447

Fed-batch fermentation has been widely used to improve product yield in cell culture. In an attempt to improve the robustness of a fed-batch fermentation process, we found that trace elements in the medium significantly affect fermentation performance and the final product yield. In this paper, we report the effect of one trace element CuSO₄ on cell growth, antibody production and cell metabolism in 750L fed-batch fermentation. It was shown that increasing the concentration of CuSO₄ in both basal and feed media not only improved the robustness of the fermentation process, but also increased the peak cell density and product titer. As shown in Figure 1, addition of CuSO₄ resulted in almost doubled peak cell density and final antibody titer. Changes in cellular metabolism were also observed when CuSO₄ concentration was increased. As summarized in Table I, increasing the concentration of CuSO₄ reduced the glucose consumption and lactate production rates, and increased glutamine consumption rate. Lactate yield was also reduced and OUR was increased at higher CuSO₄ concentration, suggesting that cell switched to a more oxidative metabolism at higher CuSO₄ concentration. Increase in CuSO₄ concentration also resulted in changes in the consumption rates of certain amino acids. Effect of additional CuSO₄ on the characteristics of antibody produced was also investigated. The purified product was characterized by N-terminal sequencing, peptide and oligosaccharide mappings, SDS-PAGE, IEF, and cation exchange chromatography. It was shown that addition of CuSO₄ did not significantly affect the characteristics of final product.

Table I. Consumption(production) rate and yield in 750L fed-batch fermentation.

Presenting author: Qu, Limin

E-mail: Lqu@PDL.com

Guozheng Wang1, Wenying Zhang1, Alastair J. Strain2, Lee Eppstein1, Yinlian Chen1, David Freedman1, Jan Zaleski3, and Frederick C. Kauffman3

1R & D Lab, New Brunswick Scientific Co. Inc., 44 Talmadge Road, Edison, NJ 08818, USA

2Liver research Labs., University of Birmingham/University Hospital Birmingham, Edgbaston, Birmingham, B15 2TT, UK

3Lab. for Cellular and Biochemical Toxicology, Rutgers University, 41 Gordon Road, Piscataway, NJ 08854, USA

Major metabolic functions of rat hepatocytes cultured in a packed-bed bioreactor (NBS Spinner Basket) for more than 90 days were investigated. Fibra-Cel, a three dimensional polyester fiber carrier was used as a growth support material. The 500 mL working volume reactor contained a 100 mL fixed-bed basket which was loaded with 10 grams of Fibra-Cel discs. Serum containing and serum-free Williams E media were used. Almost 100% of hepatocytes isolated from male Spargue-Dawley rats became immobilized on the discs after being inoculated into the bioreactor. A four-gas system for D.O. control and a re-circulating perfusion strategy were used in the process control.

At 20 hours post-inoculation and after injection of a 10 mM NH4Cl solution into the bioreactor, the urea synthesis rate reached 22 pg/cell/hr. The rate then decreased to 2.4 pg/cell/hr at 18 days after injection of a 1.0 mM NH4Cl solution. The albumin secretion rate reached a maximum value of 5.0 pg/cell/hr at day 1 and decreased to 2.2 pg/cell/hr at day 18. In contrast, in serum-containing medium, urea synthesis and albumin secretion rates were more stable, and by 90 days were 1.1 pg/cell/hr and 2.1 pg/cell/hr, respectively. Cell viabilities using the MTT cell count method were 68.6% at day 18 in serum-free medium and 68.7% at day 90 in serum containing medium.

Further analysis of hepatic functions were carried out on discs withdrawn from the bioreactor at various intervals. Hepatic metabolic function was determined by measuring rates of ureagenesis, ketogenesis, glycolysis, and 7-ethoxycoumarin O-de-ethylation as well as ATP/ADP and NAD+/NADH ratios.

Results indicated that differentiated hepatocyte functions were able to be maintained in the bioreactor over a prolonged period of time. This approach using a three-dimensional Fibra-Cel packed bed reactor has numerous potential applications in tissue engineering including the development of a bioartificial liver device as well as system for maintaining hepatocytes for long-term drug metabolism and toxicity studies.

Presenting author: Wang, Guozheng

e-mail: wangg@nbsc.com

Jianyong WU¹, Qian RUAN¹ and Peter H.Y. LAM²

1 Department of Applied Biology and Chemical Technology, Hong Kong Polytechnic University, ² Department of Biochemistry, Hong Kong University of Science and Technology Kowloon, Hong Kong

The insect cell-baculovirus host-vector system has been increasingly used for the expression of a variety of heterologous proteins, including many high-value biologicals such as hormones, growth factors and vaccines for human and animals. The production of recombinant proteins using the insect cell-baculovirus system is a two-step process, the first is to grow the cells to certain stage, and the next, to infect the cells with a recombinant virus to express the desired product protein. The final protein yield may depend on numerous process and environmental parameters, among which the state and density of the insect cells and the availability of nutrients at the time of infection are of particular influence. Optimal protein yields and productivities have been mostly achieved with the cells infected at exponential growth phase and cell densities ranging from 1H10⁶ to 3H10⁶ cells/ml and with the medium enriched with key nutrients or replaced by fresh medium (Lindsay and Betenbaugh, 1992; Reuveny et al., 1993; Wang et al., 1993; Nguyen et al., 1993).

Therefore, it is envisaged from previous studies that medium renewal at the time of virus infection is one of the most effective ways to maintain high product yields and enhance protein productivity in the insect cell-baculovirus process. In this process mode, the spent medium used in cell growth stage, especially when the medium has only been used to grow the cells to a density far below the maximum obtainable, at least 4H10⁶ to 6H10⁶ cells/ml. Fortified with some key nutrients, the spent medium may still be reused for another growth batch. Spent medium recycle has been practised for mammalian cell (hybridoma) cultures to reduce medium costs (Kempken et al., 1991; Büntemeyer et al, 1992), but no reports have been seen for insect cell culture. Considering the positive impacts of medium renewal prior to virus infection on the protein yield, spent medium recycle may be of greater interest for the insect-baculovirus system. In addition, since the spent medium is only used in the growth stage and the protein production is in fresh medium, the effects of waste cell products and lysates on product formation and purification are minimized.

In this study, our main objective was to assess the feasibility of reusing the spent medium for insect cell growth in the production of recombinant proteins with the insect cell-baculovirus system. The experiments were conducted with two insect cell lines, Sf-9 and High-FiveJ on a serum-supplemented medium (IPL-41) in shake-flasks, producing the β-galactosidase from E. coli. First of all we conducted a screening test to identify the key nutrients and supplements which could restore cell growth in the recycled spent medium. Then we compared the process incorporating spent medium recycle with a few other process strategies, including batch and fed-batch, based on protein yield, medium cost and process complexity.

Our results showed that the spent insect cell medium IPL-41 with 10% FBS can be reused at least once for cell growth without apparent negative effect (Fig.1); Fortified with 15-20% fresh medium and a few key supplements including yeastolate, glucose and glutamine, the recycled spent medium could support cell growth similar to or even better the original fresh medium; Among various nutrient supplements, yeastolate played the key role in restoring insect cell growth in the spent medium. Both the cell growth and protein yields from the spent medium recycle process compared favourably with the other process and nutrient feeding strategies, batch and fed-batch (Table 1).

Table 1. Growth parameters of insect cells and protein yields^a in different processes.

Büntemeyer H, Wallerius C and Lehmann J (1992) Optimal medium use for continuous high density perfusion processes. Cytotechnol. 9, 59-67.

Kempken R, Büntemeyer H and Lehmann J (1991)The medium cycle bioreactor (MCB): monoclonal antibody production in a new economic production system. Cytotechnol. 7: 1991.

Lindsay DA, Betenbaugh MJ (1992) Quantification of cell culture factors affecting recombinant yields in baculovirus-infected insect cells. Biotechnol. Bioeng., 39, 614-618.

Nguyen B, Jarnagin K, Williams S, Chan H, Barnett J (1993) Fed-batch culture of insect cells: a method to increase the yield of the recombinant human nerve growth factor (rhNGF) in the baculovirus expression system. J. Biotechnol., 31, 205-217.

Reuveny S, Kim YJ, Kemp CW, Shiloach J (1993) Production of recombinant proteins in high-density insect cell cultures. Biotechnol. Bioeng., 42, 235-239.

Wang MY, Kwong S, Bentley (1993) Effects of oxygen/glucose/glutamine feeding on insect cell baculovirus protein expression: a study on epoxide hydrolase production. Biotechnol. Prog. 9, 355-361.

Presenting author: Wu, Jianyong

E-mail: bcjywu@hkpucc.polyu.edu.hk

Jian-Jiang Zhong and Yi-Heng Zhang

State Key Laboratory of Bioreactor Engineering,

East China University of Science and Technology,

130 Meilong Road, Shanghai 200237, China

A major problem in the commercialization of plant cell culture processes for useful metabolite production is low productivity, and hence, high cost. Panax notoginseng, which belongs to the Araliaceae family, is one of the most famous Chinese traditional medicinal plants. Recent studies indicate that its two metabolites, ginseng saponin (secondary metabolite) and ginseng polysaccharide (primary metabolite), possess antitumor and immunological activities. Cell culture of P. notoginseng is viewed as a potential alternative for simultaneous production of ginseng saponin and polysaccharide. This work addresses a bioprocess engineering approach for controlling the physiological and metabolic aspects of cultured notoginseng cells and optimizing the biosyntheses of ginseng saponin and polysaccharide. We first investigated the effects of medium phosphate, nitrogen, sugar, copper, and inoculum size on the cell growth, consumption of major medium nutrients, as well as accumulation of these two metabolites in P. notoginseng cell cultures. A nutrient feeding strategy was further studied for enhanced metabolite production by the cultures.

Suspension cells of P. notoginseng were subcultured every 14 days by transferring 2 g fresh cells to 50 mL Murashige and Skoog (MS) medium supplemented with 2 mg/L of 2,4-dichlorophenoxyacetic acid, 0.7 mg/L of kinetin and 30 g/L of sucrose. The medium pH was adjusted to 5.7 before autoclaving. The cultures were grown in 250 mL Erlenmeyer flasks with working volume of 50 mL on a gyratory shaker (110 rpm) at 25oC in the darkness. Analyses of fresh and dry cell weights, residual major medium nutrients, and intracellular ginseng saponin and polysaccharide were carried out as reported previously.^1-5

It was found that the optimized medium for both the cell growth and metabolite production was MS medium with 1 mM copper (Table 1), 3.75 mM phosphate, and 50 g/L of sucrose. Inoculum size had a significant effect on the cell growth rate but little on the yield of cells; and 50 g fresh cells/L was the best for inoculation. By investigating the kinetics and optimizing the environmental conditions of the cell cultures, a cell density of 24.1 g dry cells/L was obtained in a batch culture.

Table 1 Effects of initial Cu²⁺ concentration on notoginseng cell cultures in MS medium²

___________________________________________________________________________

Initial conc. Specific growth Content (mg/100 mg DW) Production (g/L)__

0 0.035 8.06 8.86 0.54 0.49

0.1 0.067 8.97 9.77 0.77 0.88

1.0 0.112 9.60 12.95 0.92 1.26

6.0 0.098 11.50 11.82 1.08 1.07

A fed-batch cultivation was conducted to further enhance the cell density and metabolite accumulation. As shown in Table 2, feeding of the nitrogen source (especially ammonium) was harmful to the saponin biosynthesis, which agreed with our previous results.⁴ A high cell density of 35.1 g dry cells/L and high production of ginseng saponin (i.e., 1.57 g/L) and polysaccharide (i.e., 5.22 g/L) were successfully achieved by intermittent feeding of sucrose to keep its medium level below 20 g/L at the later exponential phase of cultivation. The cell sedimentation volume was nearly 100% of the culture suspension, and the cell density reached was estimated to be the upper limit of the cell cultures.⁵ The volumetric productivity of ginseng saponin and polysaccharide was 2.8 and 3.4-fold higher than that obtained in a conventional batch culture, respectively. The production and productivity of the polysaccharide achieved in this work were the highest as ever reported in plant cell cultures.

Table 2 Effect of carbon and nitrogen source feeding on cell growth and the content (C), production (Pn), and productivity (Pv) of ginseng saponin and polysaccharide

________________________________________________________________________

Case Dry Saponin Polysaccharide

cells C Pn Pv C Pn Pv

Control 24.1 5.68 1.37 0.061 14.2 3.42 0.152

1 35.1 4.48 1.57 0.055 14.9 5.22 0.186

2 30.6 3.61 1.10 0.037 14.7 4.50 0.158

3 31.3 3.36 1.05 0.035 14.6 4.57 0.161

4 28.9 4.71 1.36 0.047 13.2 3.81 0.132

5 30.3 2.90 0.88 0.028 11.4 3.44 0.118

The cells were cultured in a modified MS medium. Case 1, 2 and 3: Sucrose feed starting from day 15, 18 and 20, while its level was kept below 20, 10 and 20 g/L, respectively. Case 4, 5 and 6: Feeding of 20 mM nitrate, 10 mM nitrate plus 10 mM ammonium, and 20 mM ammonium on day 15, respectively, while sugar addition was the same as in Case 1.

Presenting and corresponding author: Zhong, Jian-Jiang

e-mail: jjzhong@npc.haplink.co.cn

Scott Frykman and Friedrich Srienc

Department of Chemical Engineering and Materials Science

and

Biological Process Technology Institute

University of Minnesota

Little information is available regarding the secretion of proteins by single cells. Such information would be useful for quantitatively assessing the accumulation of a secreted protein in a bioreactor, for determining the dependence of protein secretion on the cell cycle position, and for isolating individual cells with increased secretion potential.

To observe events occurring in the microenvironment surrounding individual cells, a mathematical framework has been developed describing the behavior of a compound following its release or prior to its uptake by a cell. This description is based on the diffusional and binding processes taking place in the vicinity of the surface of a cell. It allows prediction of the rate of capture and accumulation of a secreted compound around a single cell. This concept provides the basis for the design of two experimental assays for measuring single-cell secretion rates: i) cells are immobilized in hydrogel microbeads which contain capture sites for the secreted compound; and ii) artificial receptors are bound directly to the cell surface which are capable of binding molecules secreted by individual cells. This general methodology is demonstrated in the specific case of the model organism Saccharomyces cerevisiae secreting a heterologous protein, but can be applied to any cell/secreted protein combination. Binding studies have shown that approximately 2 x 10⁵ of these artificial receptors can be bound to the surface of a single yeast cell. At this surface density of a putative artificial receptor, which likely can be further increased, it is predicted that single-cell secretion rates of 47 molecules/cell/second of a 150 kDa protein can be detected. The developed concept can be applied also to products other than protein that are synthesized and secreted by cells and should be useful for rapid screening of desirable, productive cell lines.

Presenting author: Srienc, Friedrich

e-mail: fried@cbs.umn.edu

Julian B Chaudhuri¹, Helen Connaris¹, David W Hough², Michael J Danson² & John A Howell¹

1Centre for Extremophile Research and School of Chemical Engineering

The gene for Hf volcanii DHlipDH has already been cloned in our Centre (Jolley et al, 1996). This gene was inserted into a range of E coli expression vectors and the appropriate strains were transformed. The effect of incubation temperature and promotor have been studied. A variety of results were obtained, ranging from no expression of DHlipDH to the formation of inclusion bodies. A maximum DHlipDH expression of 50% of the total cell protein was obtained where the inclusion bodies were produced. We did not obtain any soluble, active protein in any of the expression studies. Thus we are currently attempting to renature the inclusion bodies by solubilisation in denaturant and controlled dilution into high salt to reactivate the enzyme. These results will be described in detail in the poster paper.

Jolley, K A, Rapaport, E, Hough, D W, Danson, M J, Woods, W G and Dyall-Smith, M L (1996). J Bacteriol, 178, 3044-3048.

Ryu, K, Kim, J & Dordick, J S (1994). Enz. Microb. Tech. 16, 266-275.

Ventosa, A and Nieto, J J (1995). World J Microbiol Biotechnol, 11, 85-94.

Presenting author: Chaudhuri, J B

Email: j.b.chaudhuri@bath.ac.uk

Web address: http://www.bath.ac.uk/~cesjbc/home.htm

Man Bock Gu*, Gautam G. Banik, Frank W. Lee, Paul Todd, and Dhinakar S. Kompala

*Kwangju Institute of Science and Technology, Korea

Department of Chemical Engineering

University of Colorado, Boulder, CO 80309-0424

Foreign protein production can be maximized in fed-batch or high cell density perfusion cultures of recombinant mammalian cells to the fullest extent if the protein synthesis rate is inversely proportional to the cell growth rate. The ideal promoter and enhancer elements for overexpressing heterologous proteins at low growth rates are those that are most active in the G1 phase of the cell cycle. Previous results from our laboratory as well as others indicate that the most commonly used SV40 and CMV promoters are predominantly active in only the S (DNA synthesis) phase of the cell cycle. The only indirect indication available in the literature of a G1 phase specific expression is from the adenovirus major late promoter.

The current work is aimed at confirming this unique G1 phase specific expression from this AMLP promoter from the most direct two parameter (DNA and foreign protein) analysis with a dual laser flow cytometer. A new recombinant plasmid containing the E. coli lac z gene under the control of AMLP and neomycin resistance gene is stably transfected into Chinese hamster ovary cells. The cellular DNA content is measured by staining the cells with Hoechst 33342 and the intracellular-galactosidase content is measured with the fluorogenic substrate ImageneTM. The two parameter histograms will be analysed with the Mcycle and Elite software to determine the cell cycle phase(s) in the foreign gene expression is obtained for the AMLP promoter.

Presenting Author: Gu, Man Bock

Corresponding Author: Dhinakar S. Kompala

E-mail: Dhinakar.Kompala@Colorado.edu

Chun Zhang, Ryan P. Taber, and Wei-Shou. Hu

Department of Chemical Engineering and Materials Science,

University of Minnesota, Minneapolis, MN 55455

Increasing the forest productivity on limited lands has been the major challenge of today=s forestry management. Traditional breeding and tree improvement program can only partially fulfill this challenge. Much greater genetic gains can be achieved through cloning genetically superior trees. Micropropagation through somatic embryos has been proven to be an efficient propagating method for the long life-cycle species.

Somatic embryos of Douglas-fir, one of the most important softwood species, were grown in batch and perfusion cultures. The cultivation contains three phases characterized by the osmolarity change and ABA concentration. Growth kinetics of these three phases were described. The cells have no significant preference in utilizing glucose and fructose as carbon sources. The cells metabolized ammonia slightly faster than nitrate in the suspension cultures. However, in the maturation stage, the cells consumed little nitrate before the ammonia was depleted at day 20. Mature embryos were observed in four to six weeks.

An image analysis and neural network based pattern recognition system was also developed to quantify the morphological changes during the embryo development. By using a hierarchical decision tree, this pattern recognition system was able to classify the embryos into three normal embryo classes (stage 1, 2, and 3) and one abnormal embryo class. The classification accuracy of normal embryos were all above 80%. In order to describe the embryo development as a continuous process, Fisher discriminant analysis was used to find a maturation index. This maturation index was based on morphological features extracted by the image analysis and correlated with the quality of the somatic embryos. Quantitative comparison of experimental results can be achieved by using this image analysis and pattern recognition system. The results indicate that this system has great potential uses in studying conifer somatic embryogenesis.

Presenting author: Zhang, Chun

Gautam G. Banik, Frank W. Lee, Paul Todd, and Dhinakar S. Kompala

University of Colordao, Boulder, CO 80309-0424

Protein synthesis in mammalian cells can be observed in two strikingly different patterns: 1. production of monoclonal antibodies in hybridoma cultures is typically non-growth-associated, and 2. production of most glycoproteins in recombinant mammalian cell cultures is found to be growth-associated. Production of monoclonal antibodies has been easily maximized by culturing hybridoma cells at very low growth rates in high cell density fed-batch or perfusion bioreactors. Applying the same bioreactor techniques to recombinant mammalian cell cultures results in drastically reduced production rates due to their growth-associated production. Optimization of such growth associated production requires high cell growth conditions, such as in repeated batch cultures or chemostat cultures. Unfortunately, the growth-associated production pattern requires maximization of cell mass production to accomplish our real goal of maximizing the product formation.

Our recent research has shown that this growth-assoicated production in recombinant CHO cells is related to the S-phase specific production from the SV40 promoter commonly used for driving the foreign gene expression. Using two stably transfected CHO cell lines, one producing a secreted glycoprotein and the other producing an intracellular reproter protein under the control of SV40 promoters, we have shown in continuous cultures that the product formation is strongly growth-associated. We have now replaced this strong S-phase promoter in newly constructed expression vectors with a weak AMLP promoter, which is suspected to be G1-phase specific and expected to yield non-growth-associated product formation patterns. Protein synthesis results from these new CHO cell lines in continuous and high cell density perfusion culture experiments will be presented.

Presenting author: Kompala, Dhinakar

E-mail: Dhinakar.Kompala@Colorado.edu

http: spot.colorado.edu/~kompala

Shao-Yi Hou and James C. Liao, Department of Chemical Engineering, Texas

A&M University, College Station, TX 77843-2122

The assimilation of nitrogen is regulated through a complex mechanism in Escherichia coli and other enteric bacteria. Under nitrogen (ammonia)-rich conditions, glutamine synthetase (GS), the major enzyme for nitrogen assimilation under nitrogen limitation, is expressed at a low level. Upon nitrogen limitation, this enzyme is induced several fold. The gene for GS, glnA, is located within the glnALG operon, which codes for GS, NRII (product of glnL/ntrB), and NRI (product of glnG/ntrC). NRI is the response regulator and NRII is the protein kinase/phosphatase in the Ntr regulon. The activation of transcription of the glnA gene in response to nitrogen limitation requires the phosphorylated form of NR-I. NRI phosphorylation and NRI-P dephosphorylation controls the transcription level of GS. It has been reported that NRI uses two sources of phosphoryl groups, the phosphorylated form of the bifunctional kinase/phosphatase NRII and acetyl phosphate.

The glnA gene is expressed from two promoters: glnAp1 and glnAp2. Promoter glnAp2 is the major promoter upstream of glnA and is recognized by s54, whereas promoter glnAp1 is recognized by s70, and is regulated by the cAMP-CRP complex. The regulation of the operon involves at least two more proteins, PII (product of glnB) and a urydylyltransferase/urydylyl-removing enzyme (UT/UR, product of glnD). When ammonia concentration is low, the UT activity is activated, causing the conversion of PII to PII-UMP. Unmodified PII can interact with NRII and stimulate the phosphatase activity of NRII. When PII is converted to PII-UMP, NRII becomes a NRI protein kinase and converts NRI to NRI-P, which can bind to the enhancer sites upstream of glnA and activate the s54-dependent transcription from glnAp2 through a DNA-looping mechanism . The sensors of ammonium availability are suggested to be PII and UT/UR, whose activities are regulated by a-ketoglutarate and glutamine. A recent study has led to the hypothesis that PII is the main protein that senses a-ketoglutarate, whereas UT/UR is the one that senses glutamine.

In view of the complexity of glnA regulation, we hypothesized that the Ntr regulon is also affected by other metabolites in central metabolism. In this study we overexpressed phosphoenolpyruvate (PEP) carboxykinase (Pck) to perturb the metabolite levels and detected the changes in the regulation of GS.

Results in this study demonstrates that overexpressed Pck in a wild-type strain does not affect nitrogen regulation significantly. However, a overexpression of Pck in a pta strain abolishes the induction of glnA under nitrogen limitation. We also showed that the observed effect is specific to Pck, rather than a general protein overexpression effect. It has been demonstrated that high-level overexpression of lacZ or truncated tufB genes in E. coli leads to growth inhibition and rRNA degradation. The levels of LacZ or truncated tufB proteins in these reported experiments reached 30% of the total cellular proteins. In contrast, the overexpressed Pck protein in this study is less than 6% of the total cellular proteins. At this level of protein expression, we did not observe significant growth inhibition, rRNA destruction, or heat shock response.

The measurement of b-galactosidase activity from the glnA-lacZ operon fusion suggests that the effect of Pck is exerted at the transcriptional level. The lgln101 phage contains the upstream control region of glnA. The b-galactosidase activity of the lgln101 lysogen paralleled the GS activity of the non-lysogen strains. The transcriptional regulation of glnA may involve several metabolites. It has been postulated that a-ketoglutarate and glutamine serve as signals for the availability of nitrogen. During nitrogen (ammonia) limitation, the intracellular glutamine concentration drops and the a-ketoglutarate concentration increases. However, the level of a-KG does not change with and without the presence of IPTG in the pta/pCK601 strain. In contrast, pyruvate and FDP levels show about 4 and 2 fold difference, respectively, with and without IPTG. The increases in pyruvate and FDP pools correlate with the reduced inducibility of the glnA gene under nitrogen limitation. _These metabolites may play a role in signal transduction under nitrogen limiting conditions in pta strain which can not produce acetyl phosphate in glucose medium. This result highlights the possible roles of central metabolites in the regulation of global physiology.

Presenting author: James C. Liao

e-mail: j-liao@tamu.edu

Anshuman V. Patwardhan, Gaddam Narsa Goud, Richard R. Koepsel and Mohammad M. Ataai.

Department of Chemical Engineering and the Center for Biotechnology and Bioengineering, University of Pittsburgh, 300 Technology Drive, Pittsburgh, PA 15219.

Immobilized Metal Affinity Chromatography (IMAC) has shown promise for isolating desired proteins on the basis of their affinity for chelated metal ions. We suggest that how much affinity a target protein should have for chelated metal ion should be based on relative affinity of the

cellular proteins of the host cell expressing the target protein. For example the elution pattern of E. coli proteins from IDA-Cu column is striking in that protein is absent over a major portion of the elution profile. We have developed a screening procedure for a combinatorial peptide library for identifying potential affinity tags which are capable of directing the elution of fusion products to the contaminant-free region in the elution profile of E. coli proteins from immobilized copper and nickel columns. As compared to the most commonly used affinity tag comprised of six adjacent histidine residues (His6), these peptides could significantly increase the efficiency of IMAC for separation of fusion proteins.

Presenting author: Patwardhan, Anshuman V.

E.M. Sipkema¹, W. de Koning², J.E.T. van Hylckama Vlieg², K.J. Ganzeveld¹,

A.A.C.M. Beenackers¹ and D.B. Janssen²

1Chemical Engineering and ²Biochemistry Department, University of Groningen, Nijenborgh 4, 9747 AG Groningen, The Netherlands

Chlorinated ethenes (Ces), such as trichloroethene (TCE), are important pollutants that can successfully be treated in bioremediation processes. Aerobically many of these compounds can only be converted by cometabolism. Cometabo-lism can be defined as 'the inadvertent conversion of a non-growth substrate as a result of the non-specificity of a cellular enzyme'. Aspects of concern in this process include: 1) need for an additional growth substrate; 2) competitive inhibition; 3) formation of toxic intermediates; 4) use of NADH. Because of the negative effects, cometabolism based processes are intrinsically unstable and therefore constitute more of a technological challenge than growth-based processes.

The aim of our research is to develop a process for cometabolic CE removal from groundwater. To increase the understanding of the cometabolic conversion process and as a tool for the design, a metabolic model has been developed. This model describes the growth and cometabolic TCE conversion of the methanotroph Methylosinus trichosporium OB3b.

 

The model incorporates each of the aspects of concern in the TCE conversion: 1) competitive inhibition between the growth substrate methane and the cometabolic substrate TCE; 2) the toxicity of the intermediates that leads to enzyme inactivation and cell death; and 3) the use and regeneration of the cosubstrate NADH. This latter aspect implies that each of the processes of catabolism, biosynthesis and PHB (Poly-ß-Hydroxybutyrate, a storage polymer) metabolism are included.

Methane breakdown occurs via methanol, formaldehyde and formate (Fig. 1). Biomass formation is incorporated as formaldehyde assimilation. PHB - which is believed to act as a source of endogenous electron donor, and as a sink for transient excess reducing power - is formed from formaldehyde via acetyl-coenzyme A (serine pathway) and broken down to carbon dioxide. The enzyme responsible for methane conversion (Methane MonoOxygenase or MMO) also converts TCE. As a result of TCE conversion, MMO is inactivated and biomass formation decreases. NADH acts as an integrating and controlling factor. For each of the conversion steps Michaelis-Menten type equations are used, and stoichiometry (mass balances) is based on the metabolic pathways. The energy carrier ATP is assumed to be in equilibrium with NADH via the respiratory chain. The model is combined with a reactor model incorporating flows, volumes, mass transfer etc. to describe a continuous culture. Most parameter values could be determined fairly easily (either measured or derived from literature); the remainder was derived from pulse experiments in continuous cultures (methane, methanol, formaldehyde and formate pulses).

Comparisons of simulations with experi-mental data lead to the conclusion that the model can describe dynamic experiments. From studies with different sets of para-meter values of methanol pulses it is concluded that NADH can be used as a growth regulating substance. Simulations show that formate pulses lead to a rapid increase in NADH levels. This additional NADH may be either oxidized in the respiratory chain and used for biomass synthesis or used for the reduction of formaldehyde to biomass. Comparison with experimental data (Fig. 2) lead to the conclusion that the latter is the case. The predicted high NADH levels, which are in agreement with NADH measurements, are expected to be responsible for the improved TCE conversion observed as a result of formate addition to resting cells. Steady state literature data indicate that increasing TCE loads lead to 1) decreasing yields of biomass on methane, 2) continuously high TCE conversions, and 3) wash-out between 50 and 300 nmol of TCE converted per mg of cells produced. Simulations of these situations (Fig. 3) agree with these data. In the future, futher experimental verification of the model will be carried out.

Presenting author: Sipkema, E. Marijn

e-mail: e.m.sipkema@chem.rug.nl

Pil Kim, Sang Yong Kim and Jung Hoe Kim

Department of Biological Sciences

Korea Advanced Institute of Science and Technology, Taejon 305-701, Korea

To characterize the relationship between the polysaccharide structure and the emulsifying activity of the anionic lipopolysaccharide, emulsan, the effect of conformation change of emulsan derived by adding charged polypeptides and degree of branch of polysaccharide backbone chain were studied. When a positively-charged polylysine was introduced to emulsan, its emulsifying activity of emulsan was markedly decreased, whereas no significant change was observed by adding a negatively- or un-charged polypeptide. It was found that emulsan conformation was withdrawn because the hydrodynamic volume (intrinsic viscosity) was decreased by the ionic interaction between the anionic emulsan and the positive charge of the polylysine. This result suggests that the structural conformation of the polysaccharide backbone in an aqueous solution is an important factor in determining the emulsifying activity of emulsan. The relationship between the branching degree of polysaccharide backbone as a conformation index and the emulsifying activity of emulsan, was further investigated. Many kinds of emulsan with the different branching degrees of carbohydrate backbone were obtained from different culture conditions such as carbon sources, culture temperature, culture times, dilution rates, and inhibitors. As the branching degree of emulsan was increased, the emulsifying activity of emulsan increased with a linear relation even though emulsan had the same content of fatty acid. Thus, it was concluded that the conformation of polysaccharide backbone was the most important factor influencing the emulsifying activity in emulsan. The possible interaction of emulsan and oil droplet is indicated as following figures in both cases of highly branched and less branched emulsan.

1. Zajic, J. E. and Panchal, C. J., (1976) CRC Critical Reviews in Microbiol., November 39-66
1. Kosaric, N., Gray, N. C. C., and Cairns, W. R., (1987) in Biosurfactant & Biotechnology : Surfactant Science Series Vol.25 p1
1. Desai, A. J., Patel, R. M., and Desai, J. D., (1994) J. Sci. Indus. Res. 53, 616-629
1. Gutnick, D. L. (1987) Biopolymers, 26, Suppl. S, S223-S240
1. Hayes, M. E., Nestaas, E., Hrebenar, K. R., (1986)Chemtech, April 239-243
1. Zosim, Z., Gutnick, D. L., and Rosenberg, E., (1982) Biotechnol. Bioeng., 24, 281-292
1. Gutnick, D. L., Allon, R., Levy, C., Petter, R., and Minas, W., (1991) in The Biology of Acinetobacter, p411-441
1. Sar, N. and Rosenberg, E., (1983) Current Microbiol. 9, 309-314
1. Shabtai, Y. and Wang, D. I. C., (1990) Biotechnol. Bioeng. 35, 753-765
1. Kaflan, N and Rosenberg, E., (1982) Appl. Environ. Microbiol., 44, 1335-1341
1. Belsky, I., Gutnick, D. L., and Rosenberg, E., (1979) FEBS Letts., 101, 175-179
1. Prehm, P., (1980) Carbohydrate Res., 78, 372-374
1. Stellner, K., Saito, H., and Hakomori, S., (1973) Arch. Biochem. Biophys. , 155, 464-472
1. Zuckerberg, A., Diver, A., Peeri, Z., Gutnick, D. L., and Rosenberg, E., (1979) Appl. Environ. Microbiol. 37, 414-420
1. Yakimov, M. M., Fredrickson, H. L., and Timmis, K. N., (1996) Biotechnol. Appl. Biochem. 23, 13-18
1. Zhang, Y. and Miller, R. M., (1995) Appl. Environ. Microbiol. 61, 2247-2251
1. Neu, T. R., Dengler, T., Jahn, B., and Poralla, K., (1992) J. Gen. Microbiol., 138, 2531-2537
1. Zosim, Z., Fleminger, G., Gutnick, D. L., and Rosenberg, E., (1989) J. Dispersion Sci. Technol., 10, 307-317

Presenting author : Pil Kim

Corresponding author : Jung Hoe Kim

E-mail : kimjh@sorak.kaist.ac.kr or pil@bioneer.kaist.ac.kr

C. Shene, A. Olivera, B.A. Andrews and J.A. Asenjo

Centre for Biochemical Engineering and Biotechnology, Department of Chemical Engineering, University of Chile

Beauchef 861, Santiago, Chile

A model to describe the behaviour of continuous culture and fed-batch fermentations of Bacillus subtilis DN 1885 containing the plasmid pCH7 has been developed. pCH7, derived from the pUB110 family, does not have genetic functions for effective propagation. This is a common feature of most available vectors and gives rise to unstable cultures in which the fraction of plasmid containing cells decreases over time. The highly structured model accounts for biomass, segregational culture stability, recombinant (β-1,4-glucanase) and non-recombinant (protease) enzymes and substrate.

A notable feature is the inclusion of a distribution of the plasmid-containing cells in the fraction of recombinant biomass which has a critical effect on the rate of formation of plasmid-free cells as compared to average values of plasmid content. This feature is of key importance as it explains culture instability which cannot be simulated properly with non-distributed models. It is known that using values of average plasmid content gives rise to unrealistic rates of plasmid-free cell formation. The use of plasmid content distribution also has a key influence on the recombinant enzyme synthesis rate. Parameter estimation and model validation have been carried out using continuous and fed-batch cultures. The model has also been successfully used for optimization of enzyme production in an unstable recombinant culture.

Currently we are using a Genetic Programming Algorithm and the mechanistic data from our system to find better approximations for the parts of the model where the mechanism is uncertain.

Presenting author: Andrews, Barbara A.

Florence Wu, Twana Howard, Noushin Dunkelman, Al Peterson, Rebecca de la Torre, Ronda Schreiber, Dawn Applegate

Cartilage Research and Product Development, Advanced Tissue Sciences -

Smith & Nephew, La Jolla, CA 92037-1005

By seeding primary ovine articular chondrocytes onto poly-glycolic acid (PGA) biodegradable scaffolds within a convective-flow bioreactor, the synthesis of tissue-engineered articular cartilage has been recently demonstrated. The ability to cultivate and manipulate this cell-polymer construct to possess specific biochemical and biomechanical properties is critical for potential application as an in vivo therapy of damaged articular surfaces. Characterization of the kinetics of construct growth, metabolism and matrix deposition was performed. During bioreactor cultivation, cells attached to the PGA scaffold and then proliferated into a three-dimensional construct. Analysis of the kinetics of DNA content over culture time suggested that chondrocyte cell number plateaued to a stationary level, after an initial exponential increase. This kinetic behavior appeared to indicate classical contact-inhibited growth. As illustrated by safranin-O staining, sulfated-proteoglycans were deposited uniformly throughout the thickness of the construct. Biochemical analysis of glycosaminoglycans and total collagens showed significant increases in deposition throughout cultivation, and approached levels observed in native tissue. Evaluation of the kinetics of glycosaminoglycan content during cultivation indicated that deposition appeared mixed-growth associated, whereas total collagen deposition was predominantly non-growth associated. Measurement of oxygen uptake and glucose utilization rates indicated that the metabolic activities of cells in constructs were comparable to those exhibited by native explant tissue. Confocal microscopic analysis with dual fluorescence markers indicated high viability at the end of culture thoughout the cross-section of the construct. These studies illustrate the ability to synthesize tissue-engineered cartilage under convective-flow for potential human tissue repair.

1. INTRODUCTION Analysis of metabolic flux distribution and activity of metabolic loop inside living cell is of critical importance in understanding the relationship between metabolic regulation and expression of foreign gene in recombinant cell. In previous report, we developed a signal metabolic flow model to analyze the activity of metabolic loop in recombinant Saccharomyces cerevisiae with a Gal10 promoter (1). In this paper, a stoichiometric flux balancing model was used to identify the coherent relationship between a heterologous protein expression and host cellular metabolism.

2. MATERIALS AND METHODS The strain used was Saccharomyces cerevisiae D452-2a: YEp51-lacZ containing GAL10 promoter, LEU2 marker and 2m replication origin and lac Z (Fig. 1).

The detailed construction procedures can be found elsewhere (2). The expression of recombinant b-galactosidase under control of Gal10 was repressed by glucose and derepressed by galactose. Galactose was served as both a carbon source and an inducer in the production phase. Two different feeding policies were used in the fed-batch cultivation, namely, 1) only galactose was added to the medium; 2) glucose was first used to obtain a high growth rate and galactose was added when glucose was depleted in the medium. All experiments were performed in the 2-L fermentor, equipped with on-line laser turbidimeter, CO2/O2 analyzer, ethanol sensor, and DO, pH, temperature controllers

3. MODELING OF METABOLIC FLUX DISTRIBUTION IN THE FED-BATCH CULTIVATION OF RECOMBINANT Saccharomyces cerevisiae As mentioned above, galactose served as both carbon source and inducer. A typical metabolic pathway used in this work is presented in Fig. 2. Part of galactose is considered to convert to glucose-6-phosphate by the sequential action of four enzymes, galatokinase, a-D-galactose-1-phosphate uridyltransferase, uridine diphosphoglucose-4-epimerase and glucose-1-phosphate phosphoglucomutase. Whereas part of intercellular galactose is accumulated directly at specific site of Gal10 promoter and bound to the Gal80, causing the synthesis of large amount of mRNA. We found that galactose and ethanol were metabolized together when they coexist in the medium (2). Based on these considerations, a stoichiometric flux model was developed as follows: S R=B, where S, R, B are an 24x24 of stoichiometric coefficients vector, an 24x1 of flux vector, and an 24x1 of metabolic accumulation rate vector, respectively. There are 24 linear equations and 24 unknown fluxes. S was found to be of full rank. The system is, therefore, determined and an unique solution can be found, such R=S-1 B. The analysis results were illustrated in Fig. 2. We can see that the ratio of the carbon flux through PP pathway to that of the EMP pathway and the ratio of flux through the EMP pathway to that of the TCA cycle were the biggest in the expression phase. The carbon source used for biosynthesis from TCA cycle was 2.54% which was 2-fold higher than that in the exponential phase and 5-fold higher than that in the stationery phase. High ratio of PP/EMP suggests that much more carbon source was directed to the production of recombinant protein. We also found that utilization of ethanol together with galactose reduced the flux of EMP pathway, but increased the fluxes from AcCoA through TCA cycle and led to the amplification of the fluxes directing towards the synthesis of various kinds of precursors for recombinant protein

5. CONCLUSION In conclusion, the metabolic model developed can be used to analyze the carbon flux distribution of recombinant cell, which helps us to make clear on the coherent relationship between metabolism, expression and cultivation condition. We found that the expression of foreign gene channeled much more carbon flux through the PP pathway. When ethanol was used as a co-carbon source with galactose in the medium, the production of carbon dioxide was significantly decreased and biomass yield increased. The metabolism of ethanol appears to adjust the flux from AcCoA to TCA cycle.

2. Jin, S., Ye, K., and Shimizu, K.:" Application of artificial neural network and fuzzy control for fed-batch cultivation of recombinant Saccharomyces cerevisiae", J. Ferment. Bioeng. 81, 412-421 (1996).

HEPATOCYTE SPHEROID ASSEMBLY AS A MODEL SYSTEM OF IN VITRO TISSUE FORMATION
¹Manolis S. Tzanakakis, ¹Julie R. Friend, ¹Chang-Chun Hsiao, ¹Florence J. Wu, ²Linda Hansen, ¹Wei-Shou Hu
¹Department of Chemical Engineering and Materials Science, ²Laboratory of Medicine and Pathology, University of Minnesota, MN 55455

Guardia Alba, M.J.¹, Europa, A.¹, Gambhir, A.¹, Ramkrishna, D.² and Hu, W-S¹.