# Media optimization for SHuffle T7 Escherichia coli expressing SUMO-Lispro proinsulin by response surface methodology – BMC Biotechnology

Jan 3, 2022

### Culture media optimization

#### Factor screening by Plackett–Burman design

Eleven factors were evaluated for their effectiveness on bacterial growth and 20 experiments were designed by Minitab18.1.0 Software. By the end of the experiments, final cell density was measured (g/L CDW) and reported in the response column of PBD (Table 1).

After data analysis, the model was significant with a p-value of 0.0 and an R2 of 92.96% (Table 2). Model terms including pH, Yeast extract, MgCl2, N source, and KCl concentration were effective factors with p-values less than 0.05. The higher F-value of a term corresponds to the higher association of the term and the response. Pareto chart (Fig. 1) is a graphical representation of the standardized effect of each variable on response. Reference line with the value of 2.228 denotes effectiveness of factors with larger values based on significance level (α = 0.05). According to this chart, the first 3 bars with larger values corresponding to pH, Yeast extract, and MgCl2 concentration were selected for optimization experiment design by the CCD method of RSM. Besides, to interpret the effect of each independent variable on the Response Mean, the Main effects plot was generated by Minitab software (Fig. 2). Nearly horizontal lines correspond to insignificant variables denoting that responses are affected by none of the factor’s levels. According to this plot, Tryptone was applied in optimization experiments as N source since there was no preference between Tryptone and Peptone. The media was supplemented by the center point level of Tryptone and KCl. Also, the central point concentration of NaCl and 0.89 mM phosphate buffer were added to the medium due to their slight refinement on the response mean. Glycerol, glucose, and MgSO4 were omitted from the model.

#### Optimization by response surface methodology central composite design

The Design-Expert software generated 20 experiments for RSM-based optimization of chosen model terms, including pH, the concentration of Yeast extract, and MgCl2. Experiment runs were carried out in 50 mL culture containing a constant concentration of 2.5% Tryptone, 8.5 mM NaCl, 5 mM KCl and 0.89 mM Phosphate buffer in addition to varied values of Yeast extract, MgCl2, and pH according to each design point. Corresponding results were reported in the response column of CCD as presented in Table 3.

After performing analyses by different models, the quadratic model was suggested to predict and validate the optimal condition. The model p-value was significant (0.0001), while its lack of fit was insignificant (0.1247) in proportion to the pure error, implying that error does not have any impact on the suggested model (Table 4). The R2 value of 0.9581, adjusted R2 of 0.9204, and predicted R2 of 0.7309 (Difference < 0.2) indicated a reasonable fitness of the model to the experimental data and can explain 95.8% of response variations. Besides, the adequate precision value (17.8198) indicates a sufficient signal, and a smaller value of PRESS (0.8345) than the total sum of squares (3.2) depicted that the model was fitted sufficiently.

The goodness of fit of the quadratic model was further evaluated by diagnostic analyses that indicated the normality of data. The Predicted vs. Actual diagnostic plots denote that the actual response values of experiment runs were in acceptable agreement with predicted response values (Fig. 3). The compliance of the residuals with predicted values is illustrated in the Normal probability plots (Fig. 4). The Normal probability plots were linear and revealed that responses followed normal probability distribution, such that the residuals were in accordance with predicted values, and the model provided acceptable analyses.

Terms with p-values less than 0.05 are considered significant, and thus, can affect the response parameters; therefore, A (pH) and quadratic effect of terms B (Yeast extract) and C (MgCl2), (B2 and C2) were significant model terms. Based on the quadratic model, the 3D and contour plots were generated (Fig. 5). According to Fig. 5, the highest response was accomplished when the media was supplemented by medium levels of Yeast extract (2.5%) and MgCl2 (10 mM) coupled with maximum pH (8).

The equation in terms of actual factors was achieved from the quadratic model depicting the mathematical model for biomass production with culture optimization process:

begin{aligned} CDW left( frac{g}{L} right) & = – 2.68893 + 0.67111;pH + 0.736996;Yeast;Extract \ & quad + 0.099864;MgCl_{2} + 0.003752;pH * Yeast;Extract \ & quad + 0.005074;pH * MgCl_{2} – 0.011486;Yeast;Extract*MgCl_{2} \ & quad – 0.028798;pH^{2} – 0.135958;Yeast;Extract^{2} – 0.00555;MgCl_{2}^{2} \ end{aligned}

The Design-Expert software utilizes the obtained equation for point prediction according to chosen circumstances for each model term and response. Optimization was validated by examining three of the software suggestions with the highest desirability (Table 5). All resulted in an approximately same cell density of 2.5 g/L.

The optimum condition for maximum growth determined to be 2.5% Tryptone, 2.5% yeast extract, 10 mM MgCl2, 5 mM KCl, 8.5 mM NaCl and pH 8. The OM-I media was compared to LB media that resulted in more than 2.3-fold higher biomass with OD600 of approximately 5.8 (corresponding to 2.5 g/L CDW) compared to LB media (OD600 of 2.5 or 1.08 g/L CDW). The growth curve of SHuffle T7 culture in OM-I media was graphed against basic conditions (Fig. 6).

OM-I was applicable for other E. coli strains including BL21 (DE3) and Rossetagami B holding similar gene construct (SU-INS). More than twofold biomass was obtained when cells were cultivated in OM-I media compared to LB media (Fig. 7).

#### Evaluation of optimal points for soluble expression in shake flask

The soluble expression of the POI was evaluated in OM-I media compared to LB media in triplicates to assess the effect of media ingredient optimization on the soluble expression of the fusion protein. The results of experiments were visualized by Coomassie-stained SDS-PAGE that revealed competitively higher soluble POI produced in OM-I media (Fig. 8a).

#### Final product identity and bioactivity assessment

To evaluate the feasibility of bioactive Lispro insulin production from expressed fusion protein, the POI was purified, modified, and undergone proteolytic cleavage. His-tagged POI was isolated by Immobilized metal affinity chromatography (IMAC) via Nickel sepharose resin (Fig. 8b) [13].

The Purified POI was successfully converted to bioactive insulin Lispro and retained its solubility after the tag and C-peptide removal. The produced Lispro was identical to its commercially available analog considering electrophoretic mobility, LC–MS/MS, Circular Dichroism (CD), HPLC, and bioactivity analyses (Data not shown) [13].

#### Evaluation of OM-I media in fermentor (Batch culture)

The large-scale applicability of optimal media was assessed in a 5 L volume fermentor vessel containing 3 L OM-I media. The final OD600 of 15 was achieved after 15 h of inoculation (8 h after induction), and bacterial culture went in the stationary phase at this point (Fig. 9a). Approximately 86 g bacterial wet weight corresponding to 6.45 g /L CDW was obtained after harvest. The bacteria pellet was resuspended in 35 mL of the Lysis buffer, and the soluble lysate was collected. SDS-PAGE results revealed a considerably high concentration of soluble POI obtained from fermentor culture (Fig. 9b) (Additional file 1; Fig. S1).