For manufacturing plasmid DNA, good manufacturing practice begins with prudent vector design (Figure 1). While the eukaryotic promoter, the gene sequence, and the poly-A site primarily affect the therapeutic efficacy, the remaining part of the vector is important for manufacturing.
All elements, features, and characteristics of the vector backbone must be evaluated with regard to process robustness and product quality.
The choice of the plasmid’s origin of replication (ori) is critical for the plasmid copy number and hence for the cultivation titer. Currently, the ColE1-derived, high copy-number pUC ori is widely established due to its reliably high copy number. The introduction of random or defined mutations into the pUC ori can further increase the plasmid yield.
Another important factor is the size of the plasmid. From a regulatory, manufacturing, and therapeutic point of view, it is evident that all unnecessary DNA sequences should be removed from the vector.
As a general rule, the plasmid should be designed to be as small as possible. Since a large plasmid exerts a metabolic load onto the E. coli host, it will reduce the cell’s resources for plasmid replication, leading to a decreased pDNA yield.
From a therapeutic point of view, a small plasmid penetrates cellular hurdles more efficiently and shows increased expression rates. The size of the plasmid, however, may be predetermined by the therapeutic approach. Plasmids coding for one gene typically have about 5 kbp (monocistronic), while such coding for multiple genes may equal 8 kbp or larger (polycistronic). In contrast, rather small plasmids are obtained when a gene-silencing approach—based on RNAi—is pursued. Plasmids coding for expressed interfering RNA (eiRNA) rarely exceed 3.5 kbp in size.
The selection marker of the plasmid is another factor to be considered. Most commonly, plasmids code for an antibiotic-resistance gene (e.g., against kanamycin) to allow selection of plasmid-carrying clones. The therapeutic coadministration of a prokaryotic resistance gene and the use of antibiotics during manufacturing are problematic from a regulatory point of view. Therefore, concepts for antibiotic-free plasmid selection based on operator-repressor titration or the supplementation of an auxotrophic gene have been developed.
Boehringer Ingelheim Austria has developed a host-vector system for antibiotic-free plasmid selection based on the natural ColE1/pUC ori.
The system utilizes the ori-encoded, copy-number controlling RNA I to silence a host-encoded repressor. This inhibition of the repressor (by antisense hybridization) allows the expression of an essential gene, which enables growth of the plasmid-carrying cells (Figure 2). The advantage of this concept is that antibiotics are not used and a resistance gene is not present on the vector. Moreover, it enables a significant size reduction of the plasmid (excision of resistance) without the need to add an alternative selection marker.
Selecting the Right Host Strain
Yield and quality of pDNA are significantly influenced by the E. coli production host. Careful selection of an efficient host strain is therefore a critical factor right at the beginning of a pDNA development project. Mostly, strains derived from E. coli K12, in particular DH1 or DH5-alpha, have been used for pDNA manufacturing. Some strains do not always fulfill the expectations regarding quantity (titer) and quality (supercoiled ratio).
Although many host features are generic, specific plasmid-host combinations may show unexpected effects, and in some cases, host screening is recommended. As an alternative strain, E. coli K12 JM108 has been selected and successfully applied for high cell-density cultivation (HCDC). It has been demonstrated that E. coli JM108 consistently shows superior performance in small- and large-scale cGMP manufacturing.
Maximizing Titer by Upstream Optimization
Cultivation of the plasmid-bearing host is the first essential process step in pDNA manufacturing. The simple and well-known batch-cultivation mode has been widely replaced by high cell-density cultivation. One key feature of HCDC is a well-balanced culture medium that supports predictable conversion of substrate into pDNA-containing biomass. This is best achieved by the application of a synthetically defined culture medium without using complex compounds like yeast extract or soy peptone. Those components have the drawback of varying quality, undefined chemical composition, and problematic handling properties (e.g., foaming, dusting, and clumping).
Another important aspect of fermentation is control of the specific growth rate µ. Fed-batch cultivation employing an exponential feeding regime is the most efficient mode for controlling µ in order to support enrichment of plasmid DNA in the biomass. Strict feed-rate control on a time basis, without plant operator input or a feedback-control mechanism, is the principal benefit of exponential and predefined nutrient addition (Figure 3).
Currently, a plasmid titer of about 2.4 g/L is the benchmark, yet it should be noted that the titer is not the only important fermentation criterion. Achievement of a high specific plasmid yield (i.e., pDNA per biomass unit) is especially important for the subsequent alkaline-lysis step. A specific pDNA yield of more than 40 mg/g of dry cell weight can already be achieved. Finally, the harvest time point is critical in order to prevent the deterioration of pDNA quality at the fermentation end. pDNA homogeneity of 90% supercoiled form at the end of the cultivation should be the target.
pDNA fermentation is not a bottleneck any more, and recent advances in process technology, plasmid-host combination, and media development have led to fermentations delivering a high pDNA yield at superior quality.
Scale-Up of Alkaline Lysis
Essentially, alkaline lysis is the first step in downstream processing of pDNA, followed by a series of chromatography and filtration steps. The purpose of alkaline lysis is the release of pDNA from the biomass.
The cells are lysed at alkaline conditions (with a pH of approximately 12), followed by neutralization. The consequence is a precipitation and flocculation of host proteins and genomic DNA, whereby plasmid DNA remains soluble in the supernatant. At lab to bench scale, this procedure is straight forward, from a process-engineering perspective. At the large scale, however, the gentle removal of the lysate precipitate represents a serious bottleneck. Mechanical stress leads to significant loss of supercoiled pDNA and to an increased level of host-related impurities, which results in reduced yield as well as in complex and cost-intensive purification schemes.
Our approach was to design an integrated alkaline cell-lysis system to minimize shear forces and to ensure efficient mixing (Figure 4). The lysis step takes place in a tube-shaped lysis reactor filled with glass beads. After neutralization in a loop reactor, the precipitate is separated in a clarification device, which is also filled with glass beads. This lysis system can be operated rapidly, thereby avoiding plasmid degradation during processing (no loss of supercoiled pDNA). Up to 95% of the initial pDNA can be recovered from the biomass, which leads to a clarified lysate (30–150 mg pDNA/L) that can be directly processed.
Our results demonstrate that debottlenecking of the lysis step is a prerequisite for lean and scalable GMP manufacturing processes, and scalability is a precondition for the emergence of pDNA as a therapeutic or DNA vaccine.
Purification processes for plasmid DNA usually consist of up to three different chromatographic steps, each with specific features and purposes. Hydrophobic interaction chromatography (HIC) is suitable for the first (capture) step. Particular focus has to be put on the optimization of the binding and elution conditions.
Once a robust gradient profile is established, baseline separation of the supercoiled plasmid from the open circle is achieved. HIC is also the main removal step for endotoxins, RNA, and genomic DNA (Figure 5). Challenges with HIC include a difficult packing procedure and limited pDNA-binding capacity, with rather large columns as a consequence.
Due to its negative charge, pDNA is predestinated for purification by anion-exchange chromatography (AEC), primarily as a second (intermediate) step. Further endotoxin clearance, volume reduction, and increase in pDNA concentration (by a factor of 10) are the main features of AEC. Monolithic AEC chromatography supports may be suitable alternatives to conventional, particle-based media.
For buffer exchange and to obtain the desired bulk formulation, size exclusion chromatography is well suited as a final column step. The resulting pDNA concentration of 1 mg/mL can be further increased by ultrafiltration, which results in a final pDNA concentration of up to 10 mg/mL. Plasmids, however, at such concentrations represent a process challenge due to the increased bulk viscosity.
The pDNA GMP production technology developed by Boehringer Ingelheim Austria was applied to different plasmids at various production scales (1–200 L fermentation aliquots) and for different clinical purposes including DNA vaccination and gene therapy. This technology can be considered as generically applicable for any plasmid. Minor process adaptations that address the features of the specific plasmid but leave the overall process concept unchanged are beneficial.