3.1. Cloning of RBD Gene Sequence into PP Vector
The presented work started with the construction of a PP vector encoding the RBD of S protein from SARS-CoV-2. First, a PCR amplification was performed with specific primers designed to amplify the RBD insert and using the pcDNA3-SARS-CoV-2-S-RBD-8his plasmid as template, as described in the previous section. The XbaI and BamHI enzymes were then used to recognize the restriction sites present in both RBD insert and PP vector, specifically in the multiple cloning site (MCS), as schematically represented in
Figure 2, to acquire the same cohesive ends and facilitate the further ligation step. After the purification of digested products, resultant fragments were analyzed by 0.8% agarose electrophoresis. As shown in
Figure 3, both PP vector (7.0 kbp) and RBD insert (640 bp) display good integrity. Furthermore, the PP vector was dephosphorylated between digestions to prevent vector self-ligation during the cloning step. Afterward, DNA T4 ligase enzyme was used to perform the ligation of the RBD insert into the PP vector. Three different molar ratios of vector:insert were tested (1:3, 1:6, 1:10) and
E. coli TOP10 competent cells were transformed with previous cloning mixtures by heat shock. The isolated colonies that had grown on an LB-agar plate containing kanamycin (50 µg/mL) were used for PCR screening tests to confirm the RBD insert presence and one positive colony was selected to grow in liquid medium. After pDNA purification, the vector was used to transform
E. coli ZYCY10P3S2T, the minicircle producer strain, and an isolated colony was used for a PCR test to confirm the presence of the RBD insert. The positive colony was cultivated in liquid medium and the purified vector was used for DNA sequencing, to confirm the alignment between RBD insert and cloned PP vector (
Figure 4). The results revealed that the RBD insert was successfully cloned into the PP vector. The sequenced colony was used to create cryopreserved bacterial banks to explore the best conditions for PP-RBD production and its recombination into mcDNA-RBD.
3.2. PP Amplification and Recombination into mcDNA
Minicircles are small molecules devoid of bacterial sequences, presenting great therapeutic interest due to their low immunogenicity [
11]. Moreover, an mcDNA vector exhibits increased transfection efficacy and transgene expression when compared to its PP precursor [
15,
24]. However, the lack of total recombination of PP into mcDNA during the induction step and the similarity between these biomolecules has a negative impact on the subsequent purification stages. Thus, after PP-RBD vector construction, this work aimed to optimize PP recombination into mcDNA.
For this purpose, preliminary experiments were performed to establish the best temperature conditions (37 °C or 42 °C) of the fermentation step and its indirect effect on the induction step, as well as the best way to transfer the cells between these two steps (with or without centrifugation), evaluating their influence on both PP and mcDNA levels. After each fermentation or induction step, cells were recovered, PP or mcDNA molecules were extracted, and the results were analyzed by agarose gel electrophoresis. In lanes 1 and 2 of
Figure 5 are presented the influence of temperature in PP amplification during the fermentation step. It is clear that the PP integrity was maintained at both fermentation temperatures (lanes 1 and 2,
Figure 5). However, the experiment at 42 °C (lane 2) presents a higher band density for PP than the one performed at 37 °C (lane 1). In line with previous studies, different fermentation temperatures can influence the pDNA final yield. For instance, using a genetically modified
E. coli strain, fermentation at 42 °C increased specific plasmid yield in comparison with the yields obtained at 37 °C [
25,
26].
Afterward, the indirect influence of temperature on the PP-RBD recombination process in the induction step combined with the cell transference in TB medium (without centrifugation) or without TB medium (after cell centrifugation) was evaluated. Therefore, for both fermentation temperatures (37 °C and 42 °C), two 250 mL Erlenmeyers with 62.5 mL of TB medium followed different paths. In the first one, a total of 62.5 mL of TB medium was added directly to a 500 mL Erlenmeyer with 62.5 mL of LB medium. In the second one, the total 62.5 mL of TB medium was centrifuged and the supernatant was discarded to recover the cells, which were resuspended in 125 mL of LB medium and placed in a 500 mL Erlenmeyer. These results are present in
Figure 5 and demonstrate an increase in mcDNA content when the cell centrifugation was performed (lanes 5–73 mg/L and 6–76 mg/L) in comparison to experiments without centrifugation (lanes 3–60 mg/L and 4–64 mg/L). This behavior can be explained due to the glucose present in the TB medium that is transferred to the induction step, which could repress the pBAD/AraC promoter, and consequently, inhibit PP recombination into mcDNA [
9].
Thus, the best conditions for PP production and cell transference to the induction step should be conducting the fermentation step at 42 °C and removing glucose from the TB medium by cell centrifugation before starting the induction process in LB medium containing L-arabinose. Despite these achievements, the conditions for PP recombination into mcDNA during the induction step need to be explored and optimized. DoE is a suitable tool to quickly accomplish this aim using as few experiments as possible.
3.3. DoE Inputs for PP Recombination into mcDNA
Besides the PP amount reached in the fermentation step, several parameters can influence PP recombination and consequently the mcDNA biosynthesis during the induction step, such as inductor concentration, antibiotic concentration, induction time, and induction temperature. However, previous studies performed at a small scale using orbital shakers have explored different L-arabinose concentrations and concluded that an improvement in inductor concentration is not reflected in a mcDNA yield increase [
15]. Based on these assumptions, we decided to start this work with 0.01% of L-arabinose, which was the lowest concentration that revealed a very satisfactory mcDNA yield.
Considering that our goal was to improve mcDNA yields, the other three parameters were explored as the DoE inputs. To define the input ranges, data from previously published works were considered [
12,
15,
27]. The parameters were explored with a defined range of 0–50 µg/mL for kanamycin concentration, given that 50 µg/mL was already used in our research group [
12]; 1–5 h for induction time, since it is not recommended to extend the induction step more than 5.5 h in an orbital shaker [
27]; and 30 °C–38 °C for the induction temperature, because PhiC31 integrase and I-SceI endonuclease enzymes involved in the recombination process had different optimal activity temperatures—32 °C and 37 °C, respectively [
28]. As the output defined was the % of recombined mcDNA and we only intend to explore points within predetermined ranges, the CCF design was chosen for this work. This model proposed 17 experiments to be performed with different input conditions, with three replicates of the central point, as indicated in
Table 1.
To conduct each individual experiment, considering the induction conditions proposed by the DoE, the cells were recovered and the mcDNA was extracted and analyzed by agarose gel electrophoresis, as depicted in
Figure 6. Each electrophoresis lane was analyzed using the software Image Lab to evaluate the percentage of recombined mcDNA from each experiment to be included in the DoE as the respective output. This percentage was calculated through the ratio between the relative density of the band corresponding to mcDNA and the relative density of the band corresponding to PP (
Table 1). Through the analyses of all results, it seems the most promising conditions to obtain high mcDNA yield during the induction step are 30 °C, without the presence of kanamycin, for 1 h (run 2 of
Table 1 and
Figure 6), given that 92.75% of recombined mcDNA was obtained. When comparing conditions without kanamycin presence and with a concentration of 50 µg/mL (lanes 1 and 5), a decrease in the mcDNA amount is observed (81.43% to 47.85%). This behavior can be related to some toxicity induced by the antibiotic. As the recombination process advanced, the bacteria lose antibiotic resistance since the selection marker gene is degraded by I-SceI endonuclease upon the PP and mP recognition. Observing experiments without kanamycin, during 1 h of induction, at opposite temperatures (runs 2 and 14), some changes in mcDNA yields indicate that when the temperature increases, 30 °C to 38 °C, the recombined mcDNA decreases, respectively, from 92.75% to 38.84%. The satisfactory results at low temperatures can be explained by the fact that the ΦC31 integrase has optimal activity at low temperatures and the endonuclease I-SceI presents a minimal activity under these conditions [
28]. Thus, no degradation of PP will occur before its recombination. When the induction step is prolonged for 5 h, regardless of other established conditions, a gradual decrease in the mcDNA yield is noticed. This evidence indicates that there is an increase in metabolic stress that favors cell lysis and death during a prolonged induction phase, and the recombined mcDNA is degraded through time [
27].
3.4. Model Generation and Statistical Analysis
After the accomplishment of all experiments proposed by the CCF design and assessing the outputs, statistical analysis was performed by Design-Expert software. In
Table 2 are the statistical coefficients obtained for the % of recombined mcDNA, which are used to understand if the statistical model generated from these experiments is valid and fits the data. Thus, R
2 represents the coefficient of determination, providing information regarding the fitness of the output statistical model to the data [
17]. This value varies between 0 and 1, with close to 1 being desirable. As is perceivable in
Table 2, the R
2 of the output is 0.9972, suggesting the model fits the data. Adjusted R
2 represents the theoretical values being adjusted to the experimental data [
29]. The output presents a valid adjusted R
2 since it only decreased by 0.0035 compared to its R
2. The predicted R
2 provides information concerning the suitability of the model in predicting new data. The model presents a high predicted R
2 value (0.9762), thus highlighting the predictive power of this model. At last, adequate precision allows the measurement of the signal-to-noise ratio [
16]. This parameter must be higher than four to indicate an adequate signal. According to
Table 2, our ratio of 60.885 indicates an excellent signal and suggests this model can be used to navigate the design space.
Observing all these coefficients, the quadratic model was chosen to proceed with the statistical analysis of this output. To further prove the validity of the DoE, ANOVA analysis was performed. In
Table 3 is represented the model significance for the output % of recombined mcDNA, including all the parameters used in this model, coupled with the corresponding lack of fit. A good valid model must present a significant value for the model (
p-value < 0.05) and a non-significant value for the lack of fit (
p-value > 0.05), thus suggesting the model data are significant and fit [
16]. According to
Table 3, all the model values are significant and do not present a significant lack of fit. Consequently, it can be confirmed that a good and valid statistical model was achieved for this output.
3.7. Determination of PP and mcDNA Concentration
After optimizing the PP fermentation and establishing the optimal point of recombined mcDNA, the quantitative analysis of the total pDNA amount present in each sample was determined with the CIMacTM pDNA analytical column, as previously described [
22]. Thus, the PP concentration reached after the fermentation process optimization in an orbital shaker was 23.07 mg/L. Considering that large amounts of pDNA are required for biopharmaceutical applications, several studies have attempted to improve the pDNA production in shake flasks. For example, Galindo and co-workers used an enzyme-controlled glucose release system and the
E. coli DH5α strain that resulted in a pDNA production of 26.6 mg/L in shake flasks [
31]. In another study, a proteome-reduced
E. coli strain produced 74.8 mg/L of pDNA when cultured in a fed-batch mode in shake flasks [
32]. In addition, Kay and collaborators used the genetically modified
E. coli ZYCY10P3S2T strain that expresses a set of inducible minicircle-assembly enzymes. These authors were able to obtain PP concentrations between 5.45 and 5.84 mg/L using TB medium containing kanamycin (50 µg/mL) at 37 °C [
15]. In fact, the variability of these results can be strongly dependent on the
E. coli strain used, which can influence and limit the pDNA production process. Given that we are using the same strain as Kay and coworkers, our production conditions, especially the temperature of 42 °C, allowed a considerably superior PP concentration.
Regarding the recombined mcDNA obtained in this work with the optimal conditions, a concentration of 16.48 mg/L was achieved. In this particular case, Kay and collaborators were able to obtain mcDNA yields between 3.40–4.83 mg/L, mixing the fermentation medium and the minicircle induction mix in the same proportion (1:1) and incubating at 32 °C for 5 h [
15]. Remarkably, our process optimization performed with DoE allowed us to significantly improve the mcDNA yield, obtaining one almost 20 times higher than those reported in the literature, by recovering cells through centrifugation before the start of the induction process to eliminate glucose and the presence of kanamycin, decreasing the temperature from 32 °C to 30 °C, and only requiring 1 h of induction.
3.8. Scale up of the mcDNA Biosynthesis in Mini-Bioreactor
As previously mentioned, the bioreactor cultures of
E. coli ZYCY10P3S2T were divided into two main phases, fermentation and induction. The fermentation phase has the main goal of increasing PP yield associated with biomass production in these cultures. Therefore, we applied identical operational and environmental conditions as for orbital shaker cultures. In addition, we investigated the impact of different levels of dissolved oxygen on cell growth that could maximize PP production in this initial stage. Considering culture conditions explored by other research groups for pDNA fermentation in bioreactors, we choose 30% pO
2 as starting point for the scale-up of our PP production [
33,
34,
35]. In this preliminary assay, bacterial samples were collected every hour to monitor OD
600nm evolution, and the result revealed that cells attained a stationary phase after 6 h of fermentation (OD
600nm = 27.7 ± 0.1). Subsequently, a bioreactor culture using 60% pO
2 was performed and after 6 h of fermentation, the cells reached an OD
600nm = 42.0 ± 0.8. Thus, the bacterial amplification using 60% pO
2 resulted in higher cell levels than those obtained at lower oxygen pressure. Afterward, the bacterial growth of
E. coli ZYCY10P3S2T using 60% pO
2 was characterized to determine the evolution profile during fermentation, being observed in a stationary phase between 5 and 6 h. Considering that the induction phase should be performed at the end of the exponential growth phase, in the following bioreactor assays, the induction was performed after 5 h of fermentation. Additionally, the analysis of agarose gel (data not shown) by densitometry revealed that the production of PP was enhanced by a factor of 37.22 and 65.46 for 30 and 60% pO
2, respectively, when compared to orbital shaker experiments. In line with previous studies, higher plasmid productivity and faster cell growth were obtained at higher dissolved oxygen concentrations [
36,
37].
Since the biomass obtained in the bioreactor is eight times higher than the biomass obtained in the orbital shaker (OD
600nm of 42 and 5, respectively), the influence of the L-arabinose concentration (0.01 and 1%) was explored in bioreactor induction step to examine whether some improvement can be obtained. Thus, the optimized conditions for the recombination of PP into mcDNA obtained in the orbital shaker were the first ones applied to the bioreactor induction step (temperature of 30 °C, without antibiotic and 0.01% of L-arabinose), applying 30% pO
2. The same assay was performed, although with the L-arabinose concentration changed to 1%. The PP recombination into mcDNA was incremented by a factor of 60.86 and 61.19, respectively. In parallel, the L-arabinose consumption from the culture medium during these induction steps in the bioreactor was assessed by HPLC coupled with RID, using a cation-exchange analytical column Agilent Hi-Plex H [
22]. For this, a calibration curve with several L-arabinose standards (range from 0.001 to 1%) was constructed, as previously indicated in Equation (2). Results presented in
Table 5 demonstrated the method’s capacity to detect L-arabinose as well as its consumption is more evident during after 4 or 6 h of induction. Nevertheless, despite the amount of L-arabinose consumed in the medium, the mcDNA increment was minimal. Thus, an additional supplement of the L-arabinose inducer during the induction phase is not required [
38] and an L-arabinose concentration of 0.01% was used in the following assays. On the other hand, as the scale-up provides an increase in biomass, the induction time may not be sufficient for the maximum recombination of PP into mcDNA to occur. Therefore, to evaluate the time influence on the recombination process, 5, 10, and 24 h of induction were explored at 30% of pO
2. However, no significant differences were found in the recombination rate after 5 h of induction, and the enhancement of the amount of mcDNA was a factor of 69.97 when compared to orbital shaker, and after that time point, the amount of PP and mcDNA started to decline, suggesting that 5 h is enough to maximize the recombination process under the operating conditions tested.
The last parameter that was explored in the bioreactor induction step was the pO
2 influence, being studied at 30 and 60%. The results showed that PP recombination into mcDNA was incremented by a factor of 69.97 and 43.67, respectively, when compared to orbital shaker studies. This result suggests that the acceleration of the biomass growth/amplification at 60% pO
2 did not favor the induction step, as occurred in the fermentation step. Probably, during the induction step, the increase in cellular metabolism interferes with the time that cells need for the expression of ΦC31 integrase and endonuclease I-SceI and their action in PP recombination and miniplasmid degradation [
27]. In addition, these conditions will ensure the plasmids’ structural and segregation stability during the induction step. Thus, the most favorable conditions for each phase of the bioreactor cultures were 60% pO
2 for fermentation for 5 h and 30% pO
2 for induction, with 0.01% L-arabinose for recombination for 5 h.
As described previously, we determined the concentration of PP and mcDNA obtained in the orbital shaker. For the bioreactor, we estimated that the PP concentration was 1.51 g/L (
Table 6) at the end of the fermentation step. Other studies in bioreactors have demonstrated pDNA yields between 0.18–218 mg/L [
35,
39] and 26.59–229.8 mg/L [
33], which indicate that the bioreactor bioprocess implemented here could be used to increase the PP yield using the
E. coli ZYCY10P3S2T host and consequently will be a good starting point to improve the recombination levels of PP into mcDNA. For the mcDNA, we estimated a concentration of 1.15 g/L (
Table 6). Concerning the only study, to our knowledge, about the yield of mcDNA attained in a mini-bioreactor platform, Šimčíková and colleagues developed a strategy to improve mcDNA yields by optimizing the parA gene 5′ untranslated region of the BW2P
E. coli strain, achieving recombination efficiency of approximately 80% and a total plasmid concentration (PP, mcDNA, and mP) of 50 mg/L [
40]. However, these authors used different fermentation conditions, LB medium, and a temperature of 37 °C, while in the present work, TB was used with a fermentation temperature of 42 °C. These differences can influence the obtained initial biomass, which may explain the increase in our mcDNA yield.