mRNA production using the ReadyToProcess WAVE™ 25 Rocker

Linearization

Linearization of plasmid DNA requires a restriction endonuclease and an enzyme-optimized salt containing buffer. In this example, we used a DNA construct that encoded enhanced green fluorescent protein (eGFP) and contained an EcoRI restriction site downstream of the gene of interest. We used a commercially available EcoRI alongside a 10× proprietary buffer solution provided by the enzyme vendor.

• Small-scale linearization reactions were performed in Eppendorf tubes placed in an incubator at 37°C which is the optimal temperature for EcoRI.
• Large-scale linearizations were performed in the ReadyToProcess WAVE™ 25 Rocker using a 2 L Cellbag™ single-use bioreactor container. Mixture was preheated to 37°C prior to enzyme addition.
• Linearized DNA was purified via tangential flow filtration (TFF) with a 100 kDa MWCO hollow-fiber membrane followed by 0.2 µm filtration.

IVT

In vitro transcription requires:

• RNA polymerase
• Nucleotide triphosphates (NTPs)
• Linear DNA template
• Mg²⁺ ions (as a cofactor)
• Reaction buffer
• Optional: RNase inhibitor (to reduce RNase degradation of the mRNA product) and inorganic pyrophosphatase (IPP)(to inhibit the formation of magnesium pyrophosphate)

Optimizating the IVT recipe has been extensively researched, but with limited consensus among practitioners (1, 2, 4, 5). In our example process, we chose a recipe to ensure a yield ≥ 4 g/L of the eGFP mRNA and is laid out in Table 1. Due to limited pDNA supply, we used a concentration of 50 µg/mL during scale-up studies, and 150 µg/mL in the 1 L and 2 L runs to maximize mRNA yield.

We carried out three IVT reactions:

• Small scale (0.1 to 5 mL): Performed in Eppendorf or conical tubes, placed in a shaking incubator at 37°C and 300 rpm for 2 h.
• Process development scale (20 mL): Performed in an Allegro™ enzyme micro reactor bag (Cytiva part no. 10190-2272F or 10190-2265D) on a WAVE™ 25 Rocker equipped with the enzyme micro reactor accessory (Cytiva part no. 29875667).
• Large scale (120 mL to 2 L): Performed on ReadyToProcess WAVE™ 25 Rockers in 2 L or 10 L Basic Cellbag™, chosen for simplicity due to their minimum required functionality.

For IVTs carried out on WAVE™ 25 Rockers, we allowed the reaction mixture to pre-heat to 37°C before adding the T7 RNA polymerase and starting the reaction.

After 2 h, products were treated with 3.5 µg/mL DNAse and 2.5 mM CaCl₂ for 40 min to digest the DNA template. The reaction was then quenched by adding EDTA to a final concentration of 80 mM and harvesting from the Allegro™ enzyme micro reactor bag or Cellbag™.

During the end-to-end production runs we rinsed each Cellbag™ twice with 40 mM Tris, using volumes equivalent to the final IVT reaction volume (i.e., 1.16 L or 2.32 L). This allowed for more complete recovery of the mRNA and resulted in a 3× dilution of the IVT solution.

Table 1. IVT reaction components

*enzymes were sourced from Aldevron LLC

Analytical methods

Linearization

Performed using gel electrophoresis. Circular DNA constructs produce two distinct bands representing the open circular and super coiled structures, whereas a completely linearized product should produce a single band at a size of 6000 nucleotides (nt).

Development IVTs

Performed on samples that were purified using LiCl precipitation, unless otherwise specified. The mRNA concentration was measured spectrophotometrically; percent integrity was measured by capillary electrophoresis (Agilent TapeStation) and is defined as the ratio of the main peak area to the total area; and, double stranded RNA (dsRNA) content was measured by J2 immunoblot assay using a 142 basepair (bp) control dsRNA.

Unpurified samples from the 1 L and 2 L production runs were used to assess mRNA concentration (RiboGreen fluorescence assay), percent integrity (SciEx PA800 Plus capillary electrophoresis system), residual DNA content (resDNASEQ kanamycin resistance qPCR kit), residual protein content (NanoOrange fluorescent quantitation assay), and RNase contamination (RNaseAlert fluorescence assay). dsRNA content was measured as described above on a LiCl purified sample.

To determine protein expression (potency), purified mRNA samples were post-transcriptionally capped with vaccinia capping enzyme and 2'-O-methyltransferase, purified a second time with LiCl precipitation and transfected to BHK 570 cells using jetMESSENGER transfection reagent; the percentage GFP positivity was determined 24 h after transfection using cell-based high throughput fluorescence microscopy using the Cytation 7 platform or flow cytometry using the Cytoflex.

We began by performing small-scale linearization reactions at 50 µL to determine effective enzyme loading. As shown in Figure 2, complete linearization was observed after 1 h for all enzyme concentrations tested. No non-specific activity was observed, with a single band at 6000 nt confirming complete digestion.

This reaction with 1 U/µg was then scaled up almost 1000-fold to 40 mg of DNA, maintaining complete digestion after 1 h (Fig 3). We then transferred the process to the ReadyToProcess WAVE™ 25 Rocker and scaled up further to 120 mg and 200 mg (Fig 4).

Enzyme concentration remained at 1 U/µg DNA in these reactions, although DNA concentration varied from 0.35 to 0.9 mg/mL to modify total volume. Samples taken every 15 min showed incomplete digestion for the 0.9 mg/mL run up to 30 min, although in both cases full digestion was achieved by 1 h (Fig 4).

Fig 2. 1% agarose gel images for 50 µL DNA linearization reactions performed at 37°C and 1 h for different EcoRI enzyme loading. Non-linearized plasmid DNA was run as a control in the lane marked “circDNA” and the two bands represent open circular and supercoiled forms.

Fig 3. 1% agarose gel images for DNA linearization reactions of different scales performed with an EcoRI concentration of 1 U/µg DNA at 37°C and 1 h. Non-linearized plasmid DNA was run as a control in the lane marked “circDNA” and the two bands represent open circular and supercoiled forms.

Fig 4. 1% agarose gel images for the linearization of (A) 120 mg of DNA at 0.35 mg/mL, and (B) 200 mg of DNA at 0.9 mg/mL. Reactions performed on the ReadyToProcess WAVE™ 25 Rocker with EcoRI concentration of 1 U/µg DNA at 37°C and 1 h. Non-linearized plasmid DNA was run as a control in the lanes marked “circDNA” and the two bands represent open circular and supercoiled forms.

IVT yield for a 1 kb mRNA construct was scale independent from 1.5 mL to 1000 mL with the total amount of mRNA produced directly proportional to the reaction volume (Fig 5). Reaction rate remained consistent across the reaction scales with mRNA concentration increasing linearly out to 2 h (Fig 6).

This suggests the reaction does not become reagent limited over 2 h, and yield could be increased by extending reaction duration, or by increasing DNA template and/or T7 enzyme concentrations.

Reaction scale was also found to have no impact on the mRNA quality. The dsRNA content and mRNA integrity remained consistent at all scales, and protein expression from large scale (ReadyToProcess WAVE™ 25 Rocker) and small scale (Eppendorf tube) mRNA was equivalent (Figs 7 and 8).

Fig 5. (A) mRNA yield and (B) total mass produced from IVT reactions performed at different scales in tubes (1.5 mL to 5 mL) or on the ReadyToProcess WAVE™ 25 Rocker (≥ 20 mL). Data points are average of three measurement replicates of a single reaction sample.

Fig 6. Yield of mRNA over time for IVT reactions performed at different scales in a tube (1.5 mL) or on the ReadyToProcess WAVE™ 25 Rocker (≥ 120 mL). Data points are mean of three measurement replicates of a single reaction sample.

Fig 7. (A) dsRNA content and (B) integrity of mRNA produced from IVTs of different scales in tubes (1.5 mL to 5 mL) or on the ReadyToProcess WAVE™ 25 Rocker (≥ 20 mL). (C) Electropherograms from capillary electrophoresis (CE) analysis of each sample and (D) example % integrity calculation for the 1000 mL sample, where % integrity is the ratio of the peak between 810 to 1400 bp to the total peak area. Data points in (A) are mean of three measurement replicates of a single sample. Data points in (B) are single measurement of a single sample.

Fig 8. GFP expression 24 h post transfection in BHK 570 cells of mRNA produced from IVTs performed at small-scale and on the ReadyToProcess WAVE™ 25 Rocker.

Using the ReadyToProcess WAVE™ 25 Rocker enabled us to capture process data throughout the IVT reaction. Pre-heating of the reaction solution took approximately 40 min regardless of scale (Fig 9). However, we found that reaction volume did impact the extent of temperature drop during T7 addition and subsequent reaction sampling (where applicable).

Temperature in the 20 mL and 120 mL reactions dropped by 2.0°C and 3.5°C respectively during T7 addition as compared to an associated drop of only 0.3°C at 1 L, while the temperature drop at each sampling point was 1.1°C at 120 mLand 0.1°C at 1 L. A similar effect was observed with the addition of DNase, which may need to be considered during process development to support optimal enzyme performance.

We used an external probe to measure temperature in the enzyme micro reactor bags during the run and found an offset of 1°C from the Rocker's temperature. This is due to the energy lost by the enzyme micro reactor accessory, and was accounted for by setting the Rocker temperature to 38°C. This strategy maintained a steady temperature of 37°C in the bag throughout the IVT reaction.

Fig 9. ReadyToProcess WAVE™ 25 Rocker data from (A) 20 mL (with enzyme micro reactor accessory), (B) 120 mL and (C) 1 L IVT reactions. (a) addition of T7 and reaction initiation; (b) sampling points; (c) addition of DNase and CaCl2; (d) reaction quench with EDTA; (e) harvest.

Although not used in this study, Cellbag™ biocontainers can include a probe for in-line monitoring of pH. We evaluated the utility of this functionality and performed a 150 mL IVT reaction for 4 h on the ReadyToProcess WAVE™ 25 Rocker, with samples taken every 30 min for mRNA content and off-line pH analysis.

Considering the reaction chemistry—where an H⁺ ion is released for every nucleotide addition to mRNA—we expected to see decreasing pH with increasing mRNA production, which we can see in Figure 10. We observed a linear relationship between mRNA and pH (Fig 11), suggesting that pH (if verified for specific constructs and reaction conditions) could act as a real-time, sample free, proxy metric for the progression of the IVT reaction, as has been previously proposed (6). This would be an additional benefit of performing IVTs in the ReadyToProcess WAVE™ 25 Rocker system.

Fig 10. mRNA yield and solution pH (measured off-line) from a 150 mL IVT reaction performed on the ReadyToProcess WAVE™ 25 Rocker.

Fig 11. mRNA yield as a function of solution pH (measured off-line) from a 150 mL IVT reaction performed on the ReadyToProcess WAVE™ 25 Rocker.

Following process development work, we performed an end-to-end production run to produce approximately 15 g of mRNA product for further downstream processing and encapsulation studies. As part of that workflow, we carried out a DNA linearization in ReadyToProcess WAVE™ 25 Rocker systems to produce 920 mg of linear DNA template, using the same conditions as development: Loading at 1 U/µg DNA and 0.9 mg/mL DNA.

Before starting the enzymatic digestion, the reaction solution was pre-heated to 37°C, taking approximately 40 min (Fig 12). EcoRI addition caused a decrease in temperature of 2.8°C, which re-stabilized within 15 min. Sampling time points minimum had less of an impact on temperature, decreasing it by only 0.5°C. Complete linearization of the DNA was achieved within 15 min of reaction initiation (Fig 13), with a process recovery of 96%.

Fig 12. ReadyToProcess WAVE™ 25 Rocker process data from large-scale linearization reaction. (a) addition of EcoRI and reaction initiation; (b) sampling; (c) Cellbag™ biocontainer harvest.

Fig 13. 1% agarose gel images for the linearization of 920 mg of DNA performed on the ReadyToProcess WAVE™ 25 Rocker with EcoRI concentration of 1 U/µg DNA at 37°C and 1 h. Non-linearized plasmid DNA was run as a control in the lane marked circDNA and the two bands represent open circular and supercoiled forms.

After TFF purification, linearized DNA template was used in two IVT reactions performed on the ReadyToProcess WAVE™ 25 Rocker at 1 and 2 L scales, using the same IVT reaction recipe from the development work—with the exception that the DNA concentration was increased to 150 µg/mL in order to maximize utilization of NTPs and mRNA yield.

Like the development work, pre-heating the reaction solution took approximately 40 min, after which the temperature was maintained at the set-point. Slight deviations (approximately 0.5°C) occurred when the T7 polymerase was added to initiate the reaction, and when the DNase and CaCl₂ were added after transcription to digest template DNA (Fig 14) and took approximately 12 min to restabilize.

Fig 14. ReadyToProcess WAVE™ 25 Rocker process data from large-scale (A) 1 L and (B) 2 L IVT reactions. (a) addition of T7 and reaction initiation; (b) addition of DNase and CaCl₂; (c) reaction quench with EDTA; (d) Cellbag™ biocontainer rinse and harvest.

The 1 L and 2 L production IVT runs were found to produce 7.10 and 6.92 g/L of mRNA respectively (Table 2). The increase in yield as compared to what was observed during scale-up activities was due to the increased DNA concentration that was used in the larger runs as previously discussed.

In both cases dsRNA content was < 1% and no RNase activity was detected. Moreover, residual DNA content was found to be 59 ng DNA/mg RNA—below regulatory specifications of 10 ng/dose for typical mRNA products, even prior to downstream processing (7).

Potency was found to be similar to process development work, with 57% GFP positivity. mRNA integrity was 100% in both scales (Fig 15). This differs from the measured integrity during development work which was likely due to differences in analytical equipment/methodology rather than a difference in product quality.

*performed on a sample taken after the 1 L and 2 L reaction products were pooled together prior to downstream processing

Table 2. Results from large-scale IVT process on ReadyToProcess WAVE™ 25 Rocker stud

Fig 15. Electropherograms (SciEx PA800 system) of mRNA from (A) 1 L and (B) 2 L IVT reactions performed on the ReadyToProcess WAVE™ 25 Rocker system. Analysis was performed on crude samples.

Hands-on support in the execution of reactions, analytical testing, and equipment set-up was provided by: Darius Menezes, Sree Gayathri Talluri, Divakara Uppu, Kyle Burrell, Chris Ivimey, Caroline Duffy, Marilyn Patterson, Stephanie Purkis, Mark Auger Jr., Ariel Zhang, Andrew Kondratowicz, Shawheen Fagan, Steve Turbayne, Matt J. Leprohon, Logan Molnar, and Stephen W. Standring. Additional technical support was provided by: Danylo Sirskyj, Daniel Chettiar, Max Hoang, and Hans Blom.

1. Daniel S, Kis Z, Kontoravdi C, Shah N. Quality by Design for enabling RNA platform production processes. Trends Biotechnol. 2022; 40(10):1213-1228. doi: 10.1016/j.tibtech.2022.03.012
2. Youssef M, Hitti C, Fulber JPC, Kamen AA. Enabling mRNA therapeutics: current landscape and challenges in manufacturing. 2023; 13:1497-1524. doi: 10.3390/biom13101497
3. American Society of Cell and Gene Therapy. Gene, Cell, & RNA Therapy Landscape; Q4 2023 Quarterly Report. American Society of Cell and Gene Therapy: Waukesha, WI, USA, 2024. https://asgct.org/publications/landscape-report.
4. Rosa SS, Nunes D, Antunes L, Prazeres DMF, Marques MPC, Azevedo AM. Maximizing mRNA vaccine production with Bayesian optimization. Biotech Bioeng. 2022; 119:3127-3139. doi: 10.1002/bit.28216
5. Samnuan K, Blakney AK, McKay PF and Shattock RJ. Design-of-experiments in vitro transcription yield optimization of self-amplifying RNA. F1000Research. 2022; 11:333. doi: 10.12688/f1000research.75677.1
6. Kern JA, Davis RH. Application of solution equilibrium analysis to in vitro RNA transcription. Prog. 1997; 13:747-756. doi: 10.1021/bp970094p
7. Leban M, Seravalli TV, Hauer M, Bohm E, Mencin N, Potusek S, Thompson A, Trontelj J, Strancar A and Sekirnik R. Determination of linearized pDNA template in mRNA production process using HPLC. Anal Bioanal Chem. 2024; 416: 2389-2398. doi: 10.1007/s00216-024-05204-0