Duarte Miguel Prazeres

Preparative purification of supercoiled plasmid DNA using anion-exchange chromatography

Large scale manufacturing of gene vectors such as plasmid DNA is an important issue in gene therapy. Anion-exchange chromatography is fundamental in the downstream processing of plasmids both as a process and analytical technique. This work reports the use of Q-Sepharose columns (1, 10 and 40 ml) for the preparative purification of plasmid pUC18. NaCl gradient elution enabled the isolation of supercoiled plasmid from low-M(r) RNA, cDNA and plasmid variants. A compact covalently closed, supercoiled form of denatured plasmid carrying large stretches of single-stranded DNA was identified as one of the major contaminants. Anion-exchange HPLC on a Poros QE 20 column was used to quantify plasmid yield. Supercoiled plasmid was recovered in a single fraction with a 62 +/- 8% yield. Loadings higher than 40 micrograms/ml gel could be used but at the expense of a loss of resolution between open circular and supercoiled forms. Plasmid quality was evaluated by gel electrophoresis, restriction analysis, transformation experiments and protein assays.

De novo creation of MG1655-derived E. coli strains specifically designed for plasmid DNA production

The interest in plasmid DNA (pDNA) as a biopharmaceutical has been increasing over the last several years, especially after the approval of the first DNA vaccines. New pDNA production strains have been created by rationally mutating genes selected on the basis of Escherichia coli central metabolism and plasmid properties. Nevertheless, the highly mutagenized genetic background of the strains used makes it difficult to ascertain the exact impact of those mutations. To explore the effect of strain genetic background, we investigated single and double knockouts of two genes, pykF and pykA, which were known to enhance pDNA synthesis in two different E. coli strains: MG1655 (wild-type genetic background) and DH5α (highly mutagenized genetic background). The knockouts were only effective in the wild-type strain MG1655, demonstrating the relevance of strain genetic background and the importance of designing new strains specifically for pDNA production. Based on the obtained results, we created a new pDNA production strain starting from MG1655 by knocking out the pgi gene in order to redirect carbon flux to the pentose phosphate pathway, enhance nucleotide synthesis, and, consequently, increase pDNA production. GALG20 (MG1655ΔendAΔrecAΔpgi) produced 25-fold more pDNA (19.1xa0mg/g dry cell weight, DCW) than its parental strain, MG1655ΔendAΔrecA (0.8xa0mg/g DCW), in glucose. For the first time, pgi was identified as an important target for constructing a high-yielding pDNA production strain.

Plasmid DNA production with Escherichia coli GALG20, a pgi-gene knockout strain: Fermentation strategies and impact on downstream processing

The market development of plasmid biopharmaceuticals for gene therapy and DNA vaccination applications is critically dependent on the availability of cost-effective manufacturing processes capable of delivering large amounts of high-quality plasmid DNA (pDNA) for clinical trials and commercialization. The producer host strain used in these processes must be designed to meet the upstream and downstream processing challenges characteristic of large scale pDNA production. The goal of the present study was to investigate the effect of different glucose feeding strategies (batch and fed-batch) on the pDNA productivity of GALG20, a pgi Escherichia coli strain potentially useful in industrial fermentations, which uses the pentose phosphate pathway (PPP) as the main route for glucose metabolism. The parental strain, MG1655ΔendAΔrecA, and the common laboratory strain, DH5α, were used for comparison purposes and pVAX1GFP, a ColE1-type plasmid, was tested as a model. GALG20 produced 3-fold more pDNA (∼141 mg/L) than MG1655ΔendAΔrecA (∼48 mg/L) and DH5α (∼40 mg/L) in glucose-based fed-batch fermentations. The amount of pDNA in lysates obtained from these cells was also larger for GALG20 (41%) when compared with MG1655ΔendAΔrecA (31%) and DH5α (26%). However, the final quality of pDNA preparations obtained with a process that explores precipitation, hydrophobic interaction chromatography and size exclusion was not significantly affected by strain genotype. Finally, high cell density fed-batch cultures were performed with GALG20, this time using another ColE1-type plasmid, NTC7482-41H-HA, in pre-industrial facilities using glucose and glycerol. These experiments demonstrated the ability of GALG20 to produce high pDNA yields of the order of 2100-2200 mg/L.

Towards effective non-viral gene delivery vector.

Despite very good safety records, clinical trials using plasmid DNA failed due to low transfection efficiency and brief transgene expression. Although this failure is both due to poor plasmid design and to inefficient delivery methods, here we will focus on the former. The DNA elements like CpG motifs, selection markers, origins of replication, cryptic eukaryotic signals or nuclease-susceptible regions and inverted repeats showed detrimental effects on plasmids’ performance as biopharmaceuticals. On the other hand, careful selection of promoter, polyadenylation signal, codon optimization and/or insertion of introns or nuclear-targeting sequences for therapeutic protein expression can enhance the clinical efficacy. Minimal vectors, which are devoid of the bacterial backbone and consist exclusively of the eukaryotic expression cassette, demonstrate better performance in terms of expression levels, bioavailability, transfection rates and increased therapeutic effects. Although the results are promising, minimal vectors have not taken over the conventional plasmids in clinical trials due to challenging manufacturing issues.

Engineering of Escherichia coli strains for plasmid biopharmaceutical production: Scale-up challenges

Plasmid-based vaccines and therapeutics have been making their way into the clinic in the last years. The existence of cost-effective manufacturing processes capable of delivering high amounts of high-quality plasmid DNA (pDNA) is essential to generate enough material for trials and support future commercialization. However, the development of pDNA manufacturing processes is often hampered by difficulties in predicting process scale performance of Escherichia coli cultivation on the basis of results obtained at lab scale. This paper reports on the differences observed in pDNA production when using shake flask and bench-scale bioreactor cultivation of E. coli strains MG1655ΔendAΔrecA and DH5α in complex media with 20 g/L of glucose. MG1655ΔendAΔrecA produced 5-fold more pDNA (9.8 mg/g DCW) in bioreactor than in shake flask (1.9 mg/g DCW) and DH5α produced 4-fold more pDNA (8 mg/g DCW) in bioreactor than in shake flask (2 mg/g DCW). Accumulation of acetate was also significant in shake flasks but not in bioreactors, a fact that was attributed to a lack of control of pH.

On the dual effect of glucose during production of pBAD/AraC-based minicircles

Minicircles are promising vectors for DNA vaccination, gene or cell therapies due to their increased transfection efficacy and transgene expression. The in vivo production of these novel vectors involves the arabinose inducible excision of a parental molecule into a minicircircle and a miniplasmid bacterial backbone. Tight control of recombination is crucial to maximize minicircle yields and purity. In this work, a minicircle production system was constructed that relies on the enzymatic activity of ParA resolvase, a recombinase that is expressed under the transcription control of the arabinose inducible expression system pBAD/AraC, and on Escherichia coli BWAA, a strain improved for arabinose uptake. Undesired recombination already after 4h of incubation in Luria-Bertani broth at 37 °C was observed due to the leaky expression from pBAD/AraC. While addition of glucose to the growth media repressed this leaky expression, it triggered a pH drop to 4.5 during exponential phase in shake flasks, which suppressed growth and plasmid production. The quantitative PCR analysis confirmed only few copies of high-copy number plasmid inside of the E. coli cells. To ensure the stability of minicircle-producing system, seed cultures should be grown at 30 °C with glucose overnight whereas cells for minicircle production should be grown in shake flasks at 37 °C without glucose up to early stationary phase when the recombination is induced by addition of arabinose.

Evidence that the insertion events of IS2 transposition are biased towards abrupt compositional shifts in target DNA and modulated by a diverse set of culture parameters

Insertion specificity of mobile genetic elements is a rather complex aspect of DNA transposition, which, despite much progress towards its elucidation, still remains incompletely understood. We report here the results of a meta-analysis of IS2 target sites from genomic, phage, and plasmid DNA and find that newly acquired IS2 elements are consistently inserted around abrupt DNA compositional shifts, particularly in the form of switch sites of GC skew. The results presented in this study not only corroborate our previous observations that both the insertion sequence (IS) minicircle junction and target region adopt intrinsically bent conformations in IS2, but most interestingly, extend this requirement to other families of IS elements. Using this information, we were able to pinpoint regions with high propensity for transposition and to predict and detect, de novo, a novel IS2 insertion event in the 3′ region of the gfp gene of a reporter plasmid. We also found that during amplification of this plasmid, process parameters such as scale, culture growth phase, and medium composition exacerbate IS2 transposition, leading to contamination levels with potentially detrimental clinical effects. Overall, our findings provide new insights into the role of target DNA structure in the mechanism of transposition of IS elements and extend our understanding of how culture conditions are a relevant factor in the induction of genetic instability.

Improvement of DNA minicircle production by optimization of the secondary structure of the 5′-UTR of ParA resolvase

The use of minicircles in gene therapy applications is dependent on the availability of high-producer cell systems. In order to improve the performance of minicircle production in Escherichia coli by ParA resolvase-mediated in vivo recombination, we focus on the 5′ untranslated region (5′-UTR) of parA messenger RNA (mRNA). The arabinose-inducible PBAD/araC promoter controls ParA expression and strains with improved arabinose uptake are used. The 27-nucleotide-long 5′-UTR of parA mRNA was optimized using a predictive thermodynamic model. An analysis of original and optimized mRNA subsequences predicted a decrease of 8.6–14.9xa0kcal/mol in the change in Gibbs free energy upon assembly of the 30S ribosome complex with the mRNA subsequences, indicating a more stable mRNA-rRNA complex and enabling a higher (48–817-fold) translation initiation rate. No effect of the 5′-UTR was detected when ParA was expressed from a low-copy number plasmid (∼14 copies/cell), with full recombination obtained within 2xa0h. However, when the parA gene was inserted in the bacterial chromosome, a faster and more effective recombination was obtained with the optimized 5′-UTR. Interestingly, the amount of this transcript was 2.6–3-fold higher when compared with the transcript generated from the original sequence, highlighting that 5′-UTR affects the level of the transcript. A Western blot analysis confirmed that E. coli synthesized higher amounts of ParA with the new 5′-UTR (∼1.8xa0±xa00.7-fold). Overall, these results show that the improvements made in the 5′-UTR can lead to a more efficient translation and hence to faster and more efficient minicircle generation.