Optimization of Enterovirus-like Particle Production and Purification Using Design of Experiments

Figure 1. Enterovirus polyprotein and representation of the iterative process of Design of Experiments. (A) The enterovirus genome is composed of a single-stranded positive-sense RNA molecule that encodes a polyprotein of approximately 2200 amino acids. The polyprotein is subsequently cleaved by viral proteases into 11 mature proteins. Of these proteins, the P1 region encodes the structural proteins of the virion, while the P2 and P3 regions encode non-structural proteins that are involved in viral replication and assembly. Previous studies have shown that the expression of only the P1 capsid proteins and the 3CD protease (highlighted in green) is sufficient to produce enterovirus-like particles [14,15,16,17,18,19]. (B) The input factors for the DoE (represented along the X, Y, and Z-axes): multiplicity of infection (MOI), viable cell density at the time of infection (VCD), and infection period (tinf), respectively. These input factors are the culturing conditions that are subjected to optimization to maximize the output (viral protein production). The blue and green spheres represent the boundary conditions and triplicate experiments of the first iteration of DoE, respectively. Once…

Figure 2. Analysis of the two iterations of the DoE for CVA6. (A) First iteration of DoE; western blot against CVA6 VP1. (B) Second iteration of DoE; western blot against CVA6 VP1. (C) Normalization of CVA6 western blot band intensities of the first iteration of the DoE using GelAnalyzer software v23.1 (light blue). Normalized ELISA against CVA6 A-particle (dark blue). The triplicates indicated a low biological variance, and the most promising fractions were found to be 3, 4, 6, and 8–11. (D) Normalization of CVA6 western blot band intensities of the second iteration of the DoE using GelAnalyzer software. Experiments 9–11 are triplicates indicating highly similar bands.

Figure 3. Summary of Fit and contour plots. (A) Summary of the fit plot for the second iteration of DoE for CVA6. R2 is an indicator for goodness of fit. Q2 estimates the prediction power of the model. MV is the model validity and measures the robustness of the model. Repro. is the reproducibility and provides a measure of the variability of the replicates relative to the variability of the dataset. Performance parameters of the model for CVA6: R2 = 0.976 (>0.5) indicated significance. Q2 = 0.768 (>0.5) determined a good model. MV = 0.65 (>0.5) indicated an absence in Lack of Fit, outliers, and transformation problems. Repro. = 0.97 (>0.5) indicated marginal biological variance. (B) Contour plots for CVA6 at the three different measured MOIs.

Figure 4. Overview investigated EV71 VLP DSP strategies. Strategy A was obtained from literature [16], and the most successful protocol (strategy F) was obtained after stepwise optimization (A through F). Red boxes indicated changes from the preceding DSP strategy. (A) Strategy for the DSP of EV71 VLPs produced at a small scale (50 mL) from the extracellular environment protocol was adapted from [16]. (B) As in Strategy A, but with Triton X-100 treatment prior to the sucrose cushion. (C) As in strategy B with a scaled-up culture volume (400 mL). (D) As in Strategy C, but with an increased volume (from 10 mL to 30 mL final volume) of the sucrose density gradient. (E) As in Strategy D, but with additional clarification (filtration) steps prior to SDG. (F) As in strategy E, with additional clarification (filtration) step prior to the sucrose cushion. This was the most successful protocol and maintained the standard for our EV71 VLP purifications.

Figure 5. Investigation of the effects of low concentration of Triton X−100 on extracellular vesicles and VLPs. Since the majority of the extracellular vesicles (EVs) originate from membrane shedding, supplementation of detergent should break the polar and hydrophobic interactions. Triton X-100 was used to investigate its effects on extracellular vesicles and VLPs (scale bar = 200 nm). Inset of the image indicates the detectable presence of VLPs (scale bar insets = 30 nm). A Triton X-100 concentration below 1% does not inhibit viral protein assembly into VLPs, while significantly reducing or eliminating extracellular vesicles.

Figure 6. SDS PAGE gels and Western blots for all VP1-indicating samples of the SDGs of each DSP strategy. All DSP strategies (A–F) are described in Figure 4. Running conditions are described in Section 2.3 SDS PAGE & Western blot. The proteins found on SDS PAGE are hypothesized to be viral proteins VP0 (38 kDa), VP1 (35 kDa), and VP3 (27 kDa).

Figure 7. Electron microscopy images of the most promising fractions of the SDGs for strategies (A–F). All DSP strategies (A–F) are described in Figure 4. Experimental conditions are described in Section 2. White scale bars in the labeled images represent 200 nm, whereas in the insets, the scale bars represent 30 nm. For each strategy, two examples of VLPs are presented below the images.

Figure 8. CVA6 VLP purification process. (A) CVA6-specific DSP strategy for the purification of 400 mL cell cultures containing VLPs. (B) SDS PAGE gels and western blots for all VP1-indicating samples of the SDGs of both CVA6 VLP DSP strategies. *: A truncated version of VP1. (C) Electron microscopy image of the most promising fraction of the SDG obtained by employing strategy F on CVA6 VLPs. The white scale bar indicates 200 nm. (D) Electron microscopy image of the most promising fraction of the SDG obtained by employing strategy G on CVA6 VLPs. The white scale bar indicates 200 nm.