1 INTRODUCTION
Plasmids are an essential tool for molecular biology and industrial biotechnology. In addition to applications like large-scale transient transfection, plasmid DNA (pDNA) is gaining relevance as a therapeutic agent. According to a recent survey, pDNA is used as the active pharmaceutical ingredient in most of all the gene therapy drugs (36.36 %) approved worldwide from 1998 to 2019 (Ma et al., 2020). A pDNA vaccine against COVID-19 that express the SARS-CoV-2 spike is currently being evaluated in pre-clinical trials (Ahn et al., 2020). Therefore, efficient culture processes are needed for large-scale pDNA production. Efficient large-scale processes require the availability of cell factories less sensitive to environmental fluctuations (Wehrs, Tanjore, Eng, Lievense, Pray, & Mukhopadhyay, 2019). For instance, pDNA is usually produced in fed-batch, high cell-density cultures ofEscherichia coli (E. coli ) (Rozkov, Gillström, Björnestedt, & Schmidt, 2008). In high cell-density cultures, the oxygen demand by cells easily surpasses the oxygen transfer capacity of the bioreactor, which leads to local or general microaerobic conditions in the vessel (Palomares, Lara, & Ramírez, 2012). O2limitation in E. coli cultures is highly undesirable, since it triggers metabolic shifts that result in the production of organic acids, as well as loss of energy and biomass yields. Nevertheless, cultures under O2 limitation may allow to reach the maximum oxygen transfer rate (OTRmax) capacity of the bioreactor at the operating power input, which is an economic advantage. In addition, it has been demonstrated that O2 limitation can increase the pDNA yields (YpDNA/X) (Passarinha et al., 2006; Lara, Jaén, Folarin, Keshavarz-Moore, & Büchs, 2019). Therefore, microaerobic pDNA production can be an attractive technology, provided that strains that can better cope with O2limitation are available. There have been recent efforts to developE. coli strains with a superior performance under microaerobic conditions (Veeravalli et al., 2018; Jaén, Velazquez, Delvigne, Sigala, & Lara 2019). A K-12-derived E. coli strain expressingVitreoscilla hemoglobin (VHb) has been proposed as an attractive host to produce pDNA (Jaén, Velazquez, Delvigne, Sigala, & Lara 2019). VHb is a well-known globin with a remarkable affinity for O2 and a fast delivery of it to the terminal oxidases. The application of VHb to improve microaerobic processes is well documented (Stark, Pagilla, & Dikshit, 2015). However, its expression is usually plasmid-based, which is undesirable for pDNA production. Jaén and coworkers (2019) inserted the vgb gene, coding for VHb, under transcriptional control of Ptrc in the chromosome of E. coli W3110recA -. The construction was integrated into the chromosome between lacI and lacZby homologous recombination, making vgb expression constitutive. The recA gene was inactivated to increase plasmid stability (Roca & Cox, 1997). A positive effect on pDNA supercoiled fraction was also observed in the recA mutant, compared with its wild-type (Jaén, Velazquez, Delvigne, Sigala, & Lara 2019). In the present contribution, both strains, W3110recA - and W3110recA - vgb + were cultured in mineral medium with glucose as carbon source to produce plasmid pUC18 in fed-batch mode. To evaluate the suitability of the engineered strain to produce pDNA under O2 limitations, microaerobic conditions were created in the fed-batch phase by decreasing the stirrer speed. Figure 1 shows the profiles of the cultures. Both strains depleted the initial glucose in ca. 3.5 hr. The fed-batch phase started at that time and the glucose concentration remained close to zero during the rest of the process (Figures 1a and 1e). The culture was maintained aerobic until 6 hr, then the dissolved oxygen (DOT) was set to 2 % air sat. (Figures 1d and 1h) and controlled via the stirrer speed. During the aerobic part of the fed-batch, both strains grew at rates approximately equal to the programmed dilution rate (0.15 ± 0.01 and 0.13 ± 0.01 hr-1 for W3110recA - and W3110recA -vgb +, respectively). During the microaerobic phase, strain W3110recA - accumulated more biomass than W3110recA -vgb +, reaching 29.5 ± 4.0 and 22.5 ± 0.2 g/L, respectively (Figures 1a and 1e). These results contrast with many studies that report an increased growth rate under microaerobic conditions when VHb is expressed (see Stark, Pagilla, & Dikshit, 2015). However, the effect of VHb has been more frequently studied in batch cultures, where glucose was in excess. Enayati and co-workers (1999) reported higher biomass accumulation ofE. coli JM103 expressing VHb from a high copy-number plasmid in fed-batch cultures in M9 medium, compared with the non-expressing cells. However, during the fed-batch cultures, those authors showed that the intracellular content of VHb reached more than 50 nmol/gWCW, while in the strain used here, the amount of active VHb is only 1 nmol/gWCW in batch cultures (unpublished data). The lower biomass accumulated by W3110recA -vgb +, compared with W3110recA - could be originated by a higher burden due to the increase of plasmid replication and the consequent increased expression of bla (the gene conferring ampicillin resistance) (Figures 1c and 1g). During the batch phase, acetate was the most important by-product, the presence of VHb resulted in ca . 25 % less acetate produced (Figures 1b and 1f). This is consistent with previous observations on the effect of VHb on overflow metabolism (Pablos, Sigala, Le Borgne, & Lara, 2014). At the end of the process, the most highly accumulated by-product was lactate for both strains, although the amount accumulated by strain W3110recA -vgb + was 5-fold higher than that of W3110recA - (Figures 1b and 1f) may be due an increased demand of NAD+. In fact, strain W3110recA -vgb +accumulated more by-products than W3110recA -. pDNA production profile is depicted in Figures 1c and 1g. The pDNA titer increased through the culture of W3110recA -. The pDNA yield from biomass (YpDNA/X) remained around 3.5 mg/g during the batch phase and then increased to a maximum of 6.2 ± 0.5 mg/g after 8 h of culture, to decrease to 5.4 ± 1.1 mg/g at the end of the process (Fig. 1c). Although the supercoiled pDNA fraction (SCF) was higher than 85 % during the whole culture, there was a trend to decrease after the transition to microaerobic conditions. At approximately 10 hr, the stirrer reached its maximum speed (2000 rpm) and the DOT fell to 0 % air sat. The OTRmax attained was 117 mmol/L hr (Figure 1d). The maximum pDNA concentration was obtained at the end of the process, reaching 156.2 ± 10.5 mg/L (Figure 1d). The performance of strain W3110recA -vgb + for pDNA production was superior to that of W3110recA -. The YpDNA/X in cultures of this VHb expressing strain remained around 4.5 mg/g during the batch phase and steadily increased, even after the transition to microaerobic conditions. The YpDNA/X reached a maximum of 12.1 ± 0.3 mg/g at the end of the process (Figure 1g), more than the double of the yield reached by the non-expressing strain. Accordingly, the pDNA concentration reached a maximum of 271.6 ± 4 mg/L at the end of the culture, which is 70 % higher than that of the strain W3110recA -. Similar to cultures of W3110recA -, DOT fell to 0 % air sat. around 10 hr, and the OTRmax was 125 mmol/L h (Figure 1h). Moreover, the SCF remained above 95 % regardless the O2availability.
The performance of the strains was also investigated at molecular and metabolic levels. Samples were taken at 4, 6 and 12 hr of culture to determine the plasmid copy number by chromosome (PCN, relative to theihfB gene) (Figure 2A). The results are in close agreement with the observed YpDNA/X. For strain W3110recA -, the PCN was ca . 90 at 4 and 10 hr, and increased to 178 ± 11 at 12 hr. For the strain W3110recA -vgb +, the PCN increased from 182 ± 69 at 4 hr to 556 ± 126 at the end of the process. The PCN of pUC plasmids is mainly controlled by two RNA molecules. Namely, RNAII serves as a primer to bind to the DNA and start the replication, while RNAI interacts with RNAII to inhibit the replication (del Solar, & Espinosa, 2000). Therefore, increasing the intracellular RNAII copies/RNAI copies ratio is a strategy to increase the PCN (Freudenau, et al., 2015; Jaén, Velázquez, Sigala, & Lara, 2019). Figure 2b depicts the RNAII copies/RNAI copies ratio from the different strains and sampling times, normalized to the sample at 2 hr. The RNAII copies/RNAI copies ratio increased in strain W3110recA -, at 10 and 12 hr, relative to 4 hr, which coincides with the measured YpDNA/X (Figure 1c). In turn, the RNAII copies/RNAI copies ratio decreased to one-half at 10 hr and increased to the same value than that at 2 hr in strain W3110recA - vgb +. This contrast with the PCN and measured YpDNA/X (Figure 1g and 2a), and implies that the PCN is also determined by the synthetic capacity of the host, as previously studied (Jaén, Velázquez, Sigala, & Lara, 2019). The copy number of vgb mRNA was also measured in W3110recA - vgb +, and results relative to ihfB copies per chromosome normalized to the sample at 4 hr are shown in Figure 2c. The vgb copies decreasedca . 75 % at 10 hr, but increased to ca. 9.1 ± 1.5 copies at the end of the culture. The latter increase could be attributed to a slower biomass formation rate during the last hours of the culture, in combination with the constitutive nature of the promoter. That results suggest that the VHb-expressing strain can address more energy and building blocks to synthesize pDNA. This was further investigated by flux balance analysis (FBA). Figure 3 shows the estimated carbon fluxes in three pseudo steady-states: the exponential growth phase during the batch phase, and the aerobic and microaerobic phase during the fed-batch. During the batch phase, the specific glucose uptake rate (qS ) was 13 % higher for W3110recA - vgb +, compared with W3110recA -. This was accompanied by a 13 % increase on the specific O2 consumption rate (qO2 ) and an 18 % increase of the specific CO2 production rate. This implies that the carbon source can be oxidized faster when VHb is present under aerobic conditions. VHb expression increased the flux to the pentose phosphate pathway (PPP) and to ribose 5-phosphate (R5P) by 15 and 16 %, respectively, in comparison with the non-expressing strain. This may be a reason for the higher YpDNA/X observed in the batch phase in W3110recA - vgb +, compared with W3110recA - (Figure 1). The flux from acetyl Co-A (AcCoA) to citrate (CIT) and from phosphoenol pyruvate (PEP) to oxaloacetate (OAA) were 20 and 10 % higher in W3110recA - vgb +, compared with W3110recA -. The presence of VHb increased the flux from ICIT to α-ketoglutarate (AKG) ca . 8 %, and the flux from malate (MAL) to oxaloacetate (OAA) by 26 %. In both reactions, NADH is produced. This cofactor is key for aerobic respiration and oxidative phosphorylation, and its estimated production rate is in agreement with the observed qO2increase when VHb is expressed. During the aerobic glucose limitation phase, qS , qO2 andqCO2 were similar, regardless the presence of VHb while during the batch phase, around 36 % and 38 % of glucose 6-phosphate was used in the PPP in W3110recA -and W3110recA - vgb +, respectively. This fraction increased to ca . 59 % in both strains during the aerobic part of the fed-batch, which may explain the YpDNA/X increase during this part of the process, compared to the batch phase (Figure 1). The estimated glycolytic fluxes were similar for both strains and the rates of by-products accumulation was negligible. Still, the flux from PEP to OAA was higher in W3110recA - vgb +, compared with W3110recA . The increased flux from ICIT to MAL due to VHb presence during the batch phase were decreased by 30 % in W3110recA - vgb +compared to W3110recA - during the aerobic part of the fed-batch. During the microaerobic fed-batch phase,qS increased by 27 % in both strains with respect to the previous aerobic conditions. Both,qO2 and qCO2 wereca . 16 % higher when VHb was expressed, consistent with the proposed action of this globin. However, the flux of G6P to the PP wasca. 11 % higher in W3110recA -vgb +, compared with W3110recA -. DNA supercoiling requires the action of ATP-consuming gyrases (Cullis, Maxwell, & Weiner; 1992). Therefore, the increased ATP generation expected from a higherqO2 may explain the higher SCF observed, particularly during the microaerobic phase in strain W3110recA -vgb + (Figure 1). The glycolytic fluxes were between 5-12 % higher when VHb was expressed. Such glycolytic activity requires a higher NAD+ supply, which may partially explain the considerably higher lactate synthesis by W3110recA - vgb +, compared with W3110recA -. This hypothesis implies that the increased qO2 in W3110recA - vgb + is not enough to provide the necessary NAD+. In general, W3110recA - vgb +produced more organic acids than W3110recA -. Frey and co-workers (2001) also reported an increase of lactate and succinate production when VHb was present in microaerobic cultures of MG1655. However, Tsai and co-workers (1996) found that when VHb was expressed in W3110, the synthesis of all the mixed-acid fermentation products was decreased. It is possible that such contradictory results may be related not only to the genetic background, but also to the amount of active VHb present. In the present experiments, the flux from AcCoA to CIT was only 5 % higher when VHb was present, but the flux from PEP to OAA was 5-fold higher, compared with W3110recA -. The flux from ICIT to AKG was 50 % lower, and the flux from MAL to OAA was only 9 % higher for the strain expressing VHb, compared with the non-expressing strain, which contrasts with the fluxes during the aerobic part of the fed-batch. The flux from PEP to OAA were consistently higher when VHb was present in the three experimental conditions tested. This flux may be influenced by a higher AKG demand for the synthesis of heme group (Kwon, de Boer, Petri, & Dannert, 2003).
Taken together, the presented results demonstrate that VHb expression is an efficient way to enhance highly supercoiled pDNA production in fed-batch cultures under both, aerobic and microaerobic regimes. Although less biomass was attained, more pDNA per biomass unit was obtained when VHb was expressed, which is advantageous for downstream operations. It was found that more pDNA can be synthesized with less RNAII copies/RNAI copies per cell when VHb is expressed. The high lactate accumulation by VHb-expressing cells may be partially explained by insufficient NAD+ regeneration in relation with the amount necessary for the increased glycolytic fluxes. Such effects may be overcome by increasing the amount of active VHb or by increasing the flux over the lower part of the TCA.