Scale up of Fed-Batch Culture to Produce Plasmid DNA in Escherichia coli (Peer Reviewed)

Published on: 
, , , , , , , ,
BioPharm International, BioPharm International-07-01-2012, Volume 25, Issue 7

The authors present scale-up from a 5-L fermentor to a 50-L pilot-scale using the criterion of constant power consumption per unit liquid volume.


Scale-up from a 5-L fermentor to a 50-L pilot-scale fermentor was carried out successfully using the criterion of constant power consumption per unit liquid volume (P/V). Fed-batch mode of cultivation using constant feeding of glucose and yeast extract mixture was employed for the production of plasmid DNA in Escherichia coli. Results showed that final biomass concentration and specific plasmid DNA yield were similar between small and large bioreactors.

Gene therapy and DNA immunization are promising possibilities for the prevention, treatment, and cure of various diseases (1, 2). In general, DNA-based vaccines are considered safe due, in part, to the lack of genetic integration, and to the absence of a specific immune response to the plasmid itself (3). This process requires considerable amounts of plasmid DNA (pDNA) that should be homogeneous with respect to structural form and DNA sequence (4). In spite of considerable scientific effort over the past few years, no gene-therapy product has yet reached the market.

For the production of large quantities of pDNA, an efficient fermentation process needs to be established. Optimization of the fermentation conditions of Escherichia coli (E. coli) for pDNA production could be fundamental, however, experimental data are limited compared with the extensive recombinant protein production literature. However, some general rules and methodologies pertaining to the production of pDNA by cultivation of E. coli are beginning to emerge (1).

A number of recent reports discuss fermentation strategies used for production of pDNA, but have not addressed the effect of fermentation conditions on the quality of the resulting pDNA. Because the location of pDNA is intracellular, productivity is proportional to the final cell density and the specific productivity (i.e., the amount of pDNA per unit cell mass) (5).

A feature of fermentation technology for large-scale plasmid production is the performance of high-density culture to obtain large amounts of biomass. Fed-batch fermentation provides higher biomass yields than batch fermentation because substrate is supplied at a rate such that it is nearly completely consumed, so nutrients are delivered over an extended period of time (6).

When a molecular biologist thinks of large-scale pDNA production, the range of 10–100 mg of DNA usually comes to mind. However, at pharmaceutical production-scale, pDNA requirements may exceed 50 g per batch. In extreme cases, many kilograms of pDNA per year may be needed to fill the demand for DNA vaccines currently in clinical trials (1).

The transfer from research-scale technology to manufacturing-scale requires management of the scale-up process. Scale-up is not just a simple multiplication of relevant factors, but instead requires skill, time investment, and incurs cost (7).

Many widely used fermentation processes were successfully scaled up on the basis of a constant volumetric oxygen transfer coefficient (KLa) and power consumption per unit volume (P/V). The use of traditional empirical methods, such as P/V leads to an increase in mixing and circulation times at large scale. In addition, high oxygen demands and high viscosity can cause concentration gradients in oxygen, shear, and pH which can have a significant impact on fermentation yield (8). Therefore, the choice of scale-up criteria in not an easy task, given the potentially sensitive and diverse responses of cells to each of the transport phenomena influenced by impeller design, system geometry, scale, fluid properties, and operating parameters (9).

In this investigation, scale-up of plasmid DNA (pIDKE2) production from a 5-L fermentor to a 50-L fermentor was carried out using power consumption per unit liquid volume (P/V) constant in a fed-batch process design as demonstrated by Ruiz and collaborators in 2009 (10).


Strain and plasmid

The plasmid used in this work was pIDKE2 with a Kanamicin resistance marker. It was transformed into E. coli DH10B host cells as described (11). The transformed E. coli was grown on several LB agar plates supplemented with 50 μg Kanamicin/mL.

Culture media

The basic LB medium for seed cultivation contained (per L): 10 g tryptone, 5 g yeast extract, 10 g NaCl, supplemented with 50 μg Kanamicin per mL. A fed-batch fermentation was carried out in 5-L and 50-L bioreactors in a working volume of 4 L and 40 L respectively, with complex medium containing (per L): 5 g glucose, 1.4 g KH2PO4, 8.6 g (NH4)2SO4, 1 g MgSO4 ·7H2O, 10 g yeast extract, 1 mL trace metal solution 1000X, and 50 μg Kanamicin per mL. The trace metal solution 1X was prepared using the following (per L): 2.29 mg AlCl3·6H2O, 1.6 mg CoCl2·6H2O, 41.21 mg H3BO3, 4.98 mg MnSO4·H2O, 0.73 mg Na2 MoO4·2H2O, 13.7 mg CuSO4·H2O. The addition of feed medium (263 g glucose/L and 50g yeast extract/L) to fed-batch cultures at a constant flow rate of 7 mL/min was achieved using a peristaltic pump when the initial carbon source was deployed due to an increase of pH.

Culture conditions

Shake-flask cultures were grown in 500 mL of LB medium, and were agitated at 200 rpm in a rotary shaking incubator for 6 hours at 37 °C. Fed-batch fermentations were aerated at 25 L/min and stirred at 385 rpm to keep 20% saturation of oxygen levels. PH was maintained at 7.0 ± 0.2 by the automatic additions of 25% (v/v) NH3 and 85% (v/v) phosphoric acid. Foaming was controlled automatically by the addition of antifoaming agent.

Dry-cell-weight determination

For dry-cell-weight determination, 1 mL aliquots of fermentation culture were centrifuged at 10,000 rpm for 10 min in a preweighed plastic tube. After careful removal of the supernatant, cells were resuspended in an equal volume of sterile pure H2O and centrifuged. The supernatant was decanted and the cell pellets were dried to a constant weight at 105 °C.


Glucose and total protein assay

Glucose concentration in supernatants was determined by dinitrosalicylic acid reducing-sugar assay and total protein concentration was determined using the Lowry method (12).

Plasmid DNA isolation

Each replicate of 10 mg bacterial cell pellet was resuspended in 300 μl of 50 mM Tris-HCl, 10 mM EDTA, 100 μg RNase A/mL, pH 8.0, and purified by the alkaline method (13).

Gel electrophoresis

Agarose gels (0.8 %) were made up in TAE buffer (40 mM Tris-acetate, 1 mM EDTA, pH 8.0) containing 0.5 μg ethidium bromide/mL. Gels were run at 100 V (5 V/cm), for 1 h. Photographs were taken using a video camera (GelPrinter Plus, TDI, Madrid, Spain).

Analytical method for quantification

Absorbance of DNA sample at 260 nm and 280 nm was determined with an HP spectrophotometer. Concentration of plasmid DNA was calculated from A260. The purity of samples was checked by the ratio of absorbance at 260 nm and 280 nm.


High cell-density fermentation requires a balanced medium that supplies adequate amounts of nutrients needed for energy, biomass, and cell maintenance and that commonly contains carbon and nitrogen sources, various salts, and trace metals. By employing this approach, Ruiz and collaborators in 2009 designed a culture medium according to bacterial element composition (10). Using that methodolgy, fed-batch fermentation was scaled up to 50 L in this study.

The scale of the fermenter, together with the expected biomass yield and product content, are key parameters in designing a manufacturing process, if the amount of product required is already known. The primary scale-up criterion of the process should be selected based upon the transport property most critical to the performance of the process. If oxygen transfer is the limiting factor, then scale up by equal P/V will be essential. This method is adopted for many authors using larger-scale fermentors, such as these below 1000 L capacity (8).

Current scale-up methods assume that, as in a small-scale fermentor, the environmental conditions are homogeneously distributed within the large-scale fermentation,. However, there are many factors, such as hydrodynamics, height, and geometric configuration of the reactor (see Table I) that can affect the environment of the fluid in large-scale reactors (8). Moreover, the suitability of scale-up methods is usually confirmed by experimental results.

Table I: Reactor geometries for 5-L and 50-L fermentation process.

Physical and chemical parameters remain constant over the scale and in some cases are optimized on a laboratory scale. The physical properties of the culture medium are:

  • Temperature = 37 °C

  • Density (ρ) = 1050 kg m-3

  • Acidity pH) = 7

  • Viscosity (μ) = 10-3 Pa·s

Scale-up of plasmid DNA (pIDKE2) production in E. coli from laboratory scale to pilot scale was carried out keeping the scaling criterion (P/V) constant. Large-scale manufacturing was completed using a 50-L Marubishi fermentor; relevant reactor geometric data are shown in Table I. The height of the liquid was calculated ifor the Marubishi 5-L fermenter assuming a flat bottom using equation (1) and for a 50-L fermenter with an ellipsoidal bottom using equation (2).

The fermenters generally met the geometric relationships in equation 3:

The correction factor for each scale was calculated using the following expression:

When taking into account the geometrical parameters, leading to one base of a fermentor with standard dimensions, and including the amount and type of propellant and the correction to be agitated and aerated fermenter, whereas the Re ≥104 and the Re power number (Np)=6, the expression for the calculation of agitation on the upper scale would be (14):

When the values from Table I are substituted into equation 5, the following expression is obtained:

Assuming that the number of aeration is constant for both scales, the air flow to each fermenter can be calculated employing equations 7 and 8:

The values of the operating parameters resulting from the calculation for 50 L are shown in Table II. Pilot plant scale-up is dominated by empirical criteria requiring geometric similarity, which is rarely achievable in practice, but which is necessary for adequate correlation of the biological responses of cells to the effects of changing scale.

Table II: Calculated parameters for 5-L and 50-L fermentation process.

The results of final biomass concentration, pDNA concentration, specific pDNA yield, volumetric pDNA yield, and percent of plasmid supercoil show that there is no significant difference in fermentation between small and large bioreactors carried out in fed-batch condition (see Table III). Because several studies have revealed that supercoiling linking number would be an important factor to consider when processing pDNA for therapeutic use, we analyzed the percentage of super coils at 24 h of culture, and it was maintained around 90% of total plasmid DNA (see table III) in both scales (4).

Table III: Results of the final fermentations in 5-L and 50-L fed-batch cultures.

The process characterization at 5-L and 50-L scale (see Figure 1), have shown that cell density increased to the logarithmic phase between 1–21 h and to the stationary phase at around 22–24 h. Figure 1 shows that carbon source exhaustion in both scales occurred after 5 h of batch growth. This depletion was indicated by an increase in pH, which was caused by the consumption of alternative carbon sources, and was confirmed by a reducing-sugar assay. Precisely at this time, a mixture of glucose and yeast extract was fed at constant flow and glucose accumulation was observed only when feeding started.

Figure 1: Cell growth kinetics of recombinant Escherichia coli DH10B transformed with pIDKE2 in the design medium using fed-batch fermentation process at 5-L and 50-L scale. DCW is dry cell weight. (ALL FIGURES ARE COURTESY OF THE AUTHORS)

During the initial batch portion, a specific growth rate (μ) of 0.60/h was achieved while a lower μ= 0.062/h was maintained during the fed batch portion. It has been reported that the low growth rate leads to a high specific plasmid DNA yield then improved plasmid DNA yield if the cell growth was not inhibited (10).

Final biomass concentration and specific pDNA yield were increased in comparison with cultures grown on a standard laboratory medium (TB) in batch mode, as has been reported by some authors (15, 16) (see Figure 2).

Figure 2: Effects of culture strategy and scale on pDNA production and cell growth.

Maintaining a robust scale-up with a consistent impurity profile was important for implementation of this process. The fermentation provides the primary control for minimizing unwanted impurities, which greatly enhances the efficiency of the downstream separation of plasmid DNA. This fermentation process is very easy to scale up and has been used to provide plasmid yields that are becoming acceptable from a manufacturing viewpoint.


Large-scale plasmid production for gene therapy presents very specific problems with regards to the reproducibility of process. Solutions for these problems and others will undoubtedly have an impact on the economics, efficacy, and safety of non-viral approaches to gene therapy. As advances continue in the field of DNA vaccines, factories capable of producing kilograms of pDNA per year must be designed.

Scale-up of plasmid pIDKE2 fermentation from a 5-L fermentor to 50-L pilot-scale fermentor was carried out successfully using P/V constant in a fed-batch process. With this culture procedure, larger amounts of plasmid DNA can be obtained in DH10B cells. The results from this study may be beneficial to the development of techniques for the fed-batch cultivation of E. coli cells and for the efficient large-scale production of plasmid DNA for therapeutic use in humans.

Odalys Ruiz Hernández* is a principal researcher in upstream process development, Jorge Valdes Hernández is head of the Fermentation Development Department, Willy Frometa Planche is a specialist in upstream process development, Michel Diaz Martínez is a specialist in upstream process development, Daniel Alvares Almiñaque is a specialist in upstream process development, Marta Pupo Peña is a researcher in the Analytical Development Department, Miladys Limonta Fernández is a researcher in downstream development of biomolecules, Dinorah Torres Idahody is the head of the Analytical Development Department, and Eduardo Martínez is head of the Development Division, all at the Center for Genetic Engineering and Biotechnology, Havana, Cuba.* To whom correspondance should be addressed,


1. K.J. Prather et at., Enzyme Microb. Technol. 33, 865–883 (2003).

2. M. Limonta et al., BioPharm Intl. 21 (9), 38–47 (2008).

3. H. Robinson, Clin. Microbiol. Newsletter 23, 17–22 (2000).

4. R. O´Kennedy, J. Ward, and E. Keshavarz. Biotecnol. Appl. Biochem. 37, 83–90 (2003).

5. B. Yakhchali et al., Jrnl. of Sci. 18 (2), 129–133. (2007).

6. A. Carnes, BioProcess Technic. 2–7 (2005).

7. M. Schleef, Biotechnol. 5a, 445–469 (1999).

8. Ch. Kim et al., Biotechnology and Bioprocess Engin. 8, 303–305 (2003).

9. D. Pollard et al., Biotechnol. Bioeng. 96 (2), 307–307 (2007).

10. O. Ruiz et al., BioPharm Intl. 22 (7), 40–45 (2009).

11. Z. Xu et al., Bioproc. Eng. 23, 669–674 (2000).

12. G. Miller, Anal. Chem. 31, 426–428 (1959).

13. H. C. Birnboim and J. Doly, Nucleic Acids Res. 7, 1513–1523. (1979).

14. S. Aiba, A. Humphrey and F. Millis, Biochemical Engineering (Academic Press, NY, 2nd ed., 1973). pp. 17–28, 75–83, 107–113, 133–149, 163–170.

15. M. Diogo et al., J. Gene Med. 3, 577–584 (2001).

16. N. Horn et al., Hum. Gene Therapy. l6, 565–73. (1995).