Design-of-experiments <i>in vitro</i> transcription yield optimization of self-amplifying RNA

Karnyart Samnuan; Anna K Blakney; Paul F McKay; Robin J Shattock

doi:10.12688/f1000research.75677.1

Home Browse Design-of-experiments in vitro transcription yield optimization of...

ALL Metrics

-

Views

-

Downloads

Get PDF

Get XML

Export

▬

✚

Research Article

Design-of-experiments in vitro transcription yield optimization of self-amplifying RNA

[version 1; peer review: 2 approved with reservations]

Karnyart Samnuan¹, Anna K Blakney¹, Paul F McKay¹, Robin J Shattock¹

PUBLISHED 18 Mar 2022

Author details Author details

¹ Section of Immunology of Infection, Department of Infectious Disease, Faculty of Medicine, Imperial College London, London, NW8 9NU, UK

Karnyart Samnuan
Roles: Conceptualization, Formal Analysis, Investigation, Methodology, Writing – Original Draft Preparation, Writing – Review & Editing

Anna K Blakney
Roles: Conceptualization, Formal Analysis, Methodology, Writing – Review & Editing

Paul F McKay
Roles: Conceptualization, Writing – Review & Editing

Robin J Shattock
Roles: Conceptualization, Supervision, Writing – Review & Editing

OPEN PEER REVIEW

REVIEWER STATUS

This article is included in the Cell & Molecular Biology gateway.

This article is included in the Pathogens gateway.

Abstract

Background: Self-amplifying RNA (saRNA) vaccines are able to induce a higher antigen-specific immune response with a more cost-effective and rapid production process compared to plasmid DNA vaccines. saRNAs are synthesized through in vitro transcription (IVT); however, this process has mainly been optimized for relatively short mRNAs.
Methods: Here, we optimized the IVT process for long saRNAs, approximately 9.4 kb, through a design of experiment (DoE) approach to produce a maximal RNA yield and validated the optimal IVT method on various sizes of RNA.
Results: We found that magnesium has the highest impact on RNA yield with acetate ions enabling a higher yield than chloride ions. In addition, the interaction between magnesium and nucleoside triphosphates (NTPs) is highly essential for IVT. Further addition of sodium acetate (NaOAc) during IVT provided no added benefit in RNA yield. Moreover, pyrophosphatase was not essential for productive IVT. The optimal IVT method can be used to synthesize different lengths of RNA.
Conclusions: These findings emphasize the ability to synthesize high quality and quantity of saRNA through IVT and that the optimal amount of each component is essential for their interactions to produce a high RNA yield.

Keywords

self-amplifying RNA, saRNA, in vitro transcription, IVT, design of experiments, nucleic acid, messenger RNA, vaccines

Corresponding author: Robin J Shattock

Competing interests: No competing interests were disclosed.

Grant information: This work was supported by the Department of Health and Social Care using UK Aid funding and is managed by the Engineering and Physical Sciences Research Council (EPSRC, grant number: EP/R013764/1 (Note: the views expressed in this publication are those of the authors and not necessarily those of the Department of Health and Social Care). AKB is supported by a Marie Skłodowska Curie Individual Fellowship funded by the European Commission H2020 (No. 794059). The Dormeur Investment Service Ltd provided funds to purchase equipment used in this study.
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Copyright: © 2022 Samnuan K et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

How to cite: Samnuan K, Blakney AK, McKay PF and Shattock RJ. Design-of-experiments in vitro transcription yield optimization of self-amplifying RNA [version 1; peer review: 2 approved with reservations]. F1000Research 2022, 11:333 (https://doi.org/10.12688/f1000research.75677.1) First published: 18 Mar 2022, 11:333 (https://doi.org/10.12688/f1000research.75677.1) Latest published: 18 Mar 2022, 11:333 (https://doi.org/10.12688/f1000research.75677.1)

Abbreviations

DoE: design of experiment

fLuc: firefly luciferase

GOI: gene of interest

IVT: in vitro transcription

NTP: nucleoside triphosphate

pDNA: plasmid DNA

saRNA: self-amplifying RNA

VEEV: Venezuelan Equine Encephalitis Virus

Introduction

The use of RNA-based vaccines as a vaccine platform against infectious diseases have emerged over the past decade because of advances in RNA production and formulation. RNA vaccines can trigger stronger and more potent cellular and humoral immune response than plasmid DNA (pDNA) and avoid any potential risk of host cell genome integration.¹ Self-amplifying RNA (saRNA) vaccines in particular, are advantageous in vaccine research due to their ability to self-replicate, allowing exponential expression of antigen.² Compared to the conventional mRNA vaccination approach, saRNA can induce equivalent protection with lower doses³ and has also been shown to elicit longer antigen expression in vivo.⁴ Furthermore, stronger cellular and humoral responses are induced when mice are immunized with saRNA compared to regular mRNA.⁵ Over the past two decades, many in vivo studies have shown that saRNA is able to induce potent and robust protection against infectious diseases like HIV-1,⁶^,⁷ influenza virus,³^,⁸^,⁹ dengue virus,¹⁰^,¹¹ respiratory syncytial virus (RSV)⁹^,¹² and Ebola virus.¹³^,¹⁴

saRNA, or an RNA replicon, is a single-stranded RNA derived from a positive strand virus genome, such as alphaviruses.¹⁵ The non-structural proteins of the virus encode for a replicase, an enzyme complex that catalyzes the self-replication of RNA template, producing a high number of RNA copies.¹⁵ The viral structural proteins, which are downstream of the non-structural proteins, are replaced with a gene(s) of interest (GOI), preventing the formation of an infectious viral particle. RNA replicons are produced by in vitro transcription (IVT) of linearized DNA templates using a T7 RNA polymerase that binds to the T7 promoter of the template to read off the sequence and synthesize RNA molecules.¹⁶ A typical IVT reaction for the synthesis of RNA includes: i) a linearized pDNA with a T7 promoter; ii) nucleoside triphosphates (NTPs) for the four bases; iii) a ribonuclease inhibitor for inactivating RNase; iv) a pyrophosphatase for degrading accumulated pyrophosphate; v) magnesium anion which is a cofactor for the T7 polymerase; vi) a pH buffer that contains optimal concentrations of an antioxidant and a polyamine.¹⁷

Contemporary IVT protocols have been primarily developed and optimized for small RNAs (<100 nt),¹⁸^,¹⁹ but not for saRNAs. Because saRNAs are much longer and have a high degree of secondary structure, the optimal conditions for IVT may be different than for mRNA. Therefore, we aim to use a design of experiment (DoE) approach to optimize the production of long RNA replicons (approx. 9.4 kb) through IVT. DoE is a method used to determine the relationship between each factor or variable that are known to influence a particular process or an output of a process.²⁰ By using this approach, we sought to establish a mathematical relationship between each variable as well as determine the most influential factor in order to maximize the output.

Here, we used a variety of designs such as full factorial designs and definitive screening designs to understand the relationship between each component that are thought to be necessary in synthesizing RNA replicons through IVT as shown in Figure 1. Firstly, a full two-level factorial DoE was conducted to determine which components have the highest significance in IVT as well as their secondary interactions. A DoE with a low, middle and high point for all of the components would not be feasible as that would result in >8000 (20³ reactions). Hence, we used a factorial design to screen the initial components. A full three-level factorial design was then utilized to explore the relationship between sodium (Na⁺), magnesium (Mg⁺²) and NTPs as well as to determine the difference between acetate ions and chloride ions in order to yield high RNA concentrations. Next, we titrated the concentration of magnesium acetate (MgOAc₂) to generate a definitive optimal ratio between NTPs and Mg⁺². We also titrated the T7 RNA polymerase concentration and subsequently increased the sodium acetate (NaOAc) concentration at different timepoints during IVT to see whether this can increase RNA yield. Once the initial components were screened, we used a definitive screening design to optimize and analyze curvature of the design space. Here, a definitive screening design was then performed to understand the significance of incorporating pyrophosphatase, spermidine, Dimethyl sulfoxide (DMSO), betaine and a surfactant of either Triton X-100 or Tween 20 and in addition, to explore any significant interactions between each of these components that would be crucial in increasing RNA yield. Lastly, we made different sizes of RNA using the optimal IVT method and transfected these RNAs in vitro before performing flow cytometry to confirm protein expression.

Figure 1. A visual flow diagram of all the DoEs performed for this study in order to achieve our optimal condition for IVT.

Methods

Plasmid DNA synthesis and purification

The plasmid DNA (pDNA) construct used for the synthesis of RNA replicons encodes for the non-structural proteins of Venezuelan Equine Encephalitis Virus (VEEV), wherein firefly luciferase (fLuc) gene (GenBank: AB762768.1) (GeneArt, Germany) was cloned into the plasmid right after the sub-genomic promoter using the restriction sites NdeI and MluI-HF. The pDNA was transformed into Escherichia coli and grown in 100 mL cultures in lysogeny broth (LB) media with 100 μg/mL carbenicillin (Sigma-Aldrich, U.S). Isolation and purification of the pDNA was done using a Plasmid Plus maxi kit (QIAGEN, UK) and a NanoDrop™ One Microvolume UV-Vis Spectrophotometer (ThermoFisher, UK) was used to measure the concentration and purity of the pDNA. The pDNA sequence was confirmed with Sanger sequencing (GATC Biotech, Germany). Prior to RNA IVT, pDNA was linearized using MluI-HF (New England Biolabs, UK) for 3 h at 37 °C.

RNA synthesis and quantification

Each of the components listed in Table 1 were used for certain DoEs. 1 μg of linearized DNA template was used and kept constant across all IVT reactions with a final volume of 100 μL per reaction which were incubated at 37 °C for either 2, 4 or 6 h. RNA yield was measured right after IVT using the Qubit RNA Broad Range Assay kit with the Qubit Fluorometer (ThermoFisher, UK) according to the manufacturer’s protocol. Each variable of each DoE was performed in triplicate.

Table 1. The list of components used for the different DoEs.

Components used for IVT	Original stock concentration for each component	Manufacturer
1 μg DNA template	-	-
ATP, CTP, UTP and GTP	100 mM	Thermo Fisher, UK
Tris Hydrochloride (Tris HCl) pH= 8	1000 mM	Thermo Fisher, UK
HEPES pH = 7.0-7.6	1000 mM	Sigma-Aldrich, US
RNase Inhibitor	40 U/μL	Thermo Fisher, UK
T7 RNA polymerase	50 U/μL	Thermo Fisher, UK
Sodium Chloride (NaCl)	5000 mM	Thermo Fisher, UK
Magnesium Chloride (MgCl₂)	1000 mM	Thermo Fisher, UK
Spermidine	100 mM	Sigma-Aldrich, US
Dithiothreitol (DTT)	1000 mM	AppliChem, Germany
Pyrophosphatase	0.1 U/μL	Thermo Fisher, UK
Triton X-100	100 ×	Sigma-Aldrich, US
Sodium Acetate (NaOAc)	3000 mM	Sigma-Aldrich, US
Magnesium Acetate (MgOAc₂)	1000 mM	Sigma-Aldrich, US
Tween 20	10%	Thermo Fisher, UK
Betaine	5 M	Sigma-Aldrich, US
Dimethyl sulfoxide (DMSO)	100%	Sigma-Aldrich, US

Design of experiment and statistical analysis

JMP software, version 14.1 (RRID:SCR_014242) was used to carry out all the DoE analysis using mainly full factorial designs and one definitive screening design with RNA yield as the response for each experiment. A fit model of standard least squares for effect screening was used to analyze the data of each DoE, where factors that were found to be non-significant were removed in order to improve the models and refine the optimal factors for IVT.²¹ Graphs were prepared using GraphPad Prism, version 8.2 (RRID: SCR_002798). An open-access alternative for both DoE analysis and the preparation of graphs is R (RRID:SCR_001905).

RNA purification and RNA gel

After IVT, RNA was purified using MEGAclear™ Transcription Clean-up Kit (ThermoFisher, UK) according to the manufacturer’s protocol. In order to assess the quality of the RNA, purified RNAs and the RNA Millennium Marker Ladder (ThermoFisher, UK) were mixed with 2× RNA loading dye (ThermoFisher, UK) and incubated at 50 °C for 30 min to denature the RNA. A 1.2% agarose gel with 1× NorthernMax Running Buffer (ThermoFisher, UK) was prepared. After incubation, the denatured ladder and samples were added to the gel and the gel was run at 80 V for 45 min. The gel was then imaged on a GelDoc-It2 (UVP, UK).

Cells and in vitro transfections

HEK293T.17 cells (ATCC, USA) were cultured in complete Dulbecco’s Modified Eagle’s Medium (DMEM) (Gibco, ThermoFisher, UK) containing 10% fetal bovine serum (FBS), 1% L-glutamine and 1% penicillin-streptomycin (Thermo Fisher, UK). Cells were plated in a 6-well plate at a density of 1.08 × 10⁶ cells per well 48 h prior to transfection. Transfection of saRNAs encoding different GOIs and mRNA fLuc was performed using Lipofectamine MessengerMAX (ThermoFisher, UK) according to the manufacturer’s instructions.

Flow cytometry

Transfected cells were harvested and resuspended in 1mL of FACS buffer (PBS + 2.5% FBS) at a concentration of 1 × 10⁷ cells/mL. 100 μL of the resuspended cells was added to a FACS tube and stained with 50 μL of Live/Dead Fixable Aqua Dead Cell Stain (ThermoFisher, UK) at a 1:400 dilution on ice for 20 min. Cells were then washed with 2.5 mL of FACS buffer and centrifuged at 1750 rpm for 7 min. The cells that were transfected with the fLuc RNAs were permeabilized with Fixation/Permeabilization solution kit (BD Biosciences, UK) for 20 min before washing them with 2.5 mL of FACS buffer and centrifuging at 1750 rpm for 7 min. Cells transfected with the fLuc RNAs were stained with 5 μL of anti-Luciferase mouse monoclonal IgG_2a κ antibody (C-12) PE (Catalog #sc-74548 PE, Santa Cruz Biotechnology, US) while the MDR1 replicons were stained with 5 μL of PE anti-human CD243 (ABCB1) mouse monoclonal IgG_2a κ antibody clone 4E3.16 (Catalog #919405, Biolegend, US). After staining for 30 min on ice, cells were washed with 2.5 mL of FACS buffer, centrifuged at 1750 rpm for 7 min and resuspended with 250 μL of PBS. Cells were fixed with 250 μL of 3 % paraformaldehyde for a final concentration of 1.5 %. Samples were analyzed on a LSRForterssa (BD Biosciences, UK) with BD FACSDiva software version 8 (RRID:SCR_001456; BD Biosciences, UK). Data were analyzed using FlowJo version 10 (RRID: SCR_008520; FlowJo LLC, Ashland, OR, USA;). An open-access alternative for flow cytometry data analysis is Flowing Software 2.5.1 (RRID:SCR_015781).

Data analysis

All raw data and analyzed data can be found in the underlying data section.²²

Results

Magnesium is the most significant component for influencing IVT yield of saRNA

A full factorial design of experiment was designed with six factors, including NaCl, MgCl₂, spermidine, Dithiothreitol (DTT), pyrophosphatase and Triton X-100, which were varied at a low and high concentration (Table 2). The RNA yield quantified were entered to the JMP software and a fit model of standard least squares was generated.

Table 2. The components used for a full two-level factorial DoE.

Fixed components	Concentration of fixed components	Components that were varied	Concentrations of components varied
Linearized DNA Template	1 μg	Sodium Chloride (NaCl)	2.5 mM
Linearized DNA Template	1 μg	Sodium Chloride (NaCl)	25 mM
ATP, CTP, UTP and GTP	4 mM each	Magnesium Chloride (MgCl₂)	2.4 mM
ATP, CTP, UTP and GTP	4 mM each	Magnesium Chloride (MgCl₂)	24 mM
Tris Hydrochloride (Tris HCl)	40 mM	Spermidine	0.2 mM
Tris Hydrochloride (Tris HCl)	40 mM	Spermidine	2 mM
RNase Inhibitor	4 U	Dithiothreitol (DTT)	4 mM
RNase Inhibitor	4 U	Dithiothreitol (DTT)	40 mM
T7 RNA Polymerase	100 U	Pyrophosphatase	0.01 U
T7 RNA Polymerase	100 U	Pyrophosphatase	0.1 U
		Triton X-100	0%
		Triton X-100	0.01%

MgCl₂ had the highest significance in IVT with a log worth of 73.862. Spermidine was the second most significant component for IVT with a log worth of 4.070 and its interactions with NaCl and MgCl₂ were also statistically significant for enhancing high RNA yield. However, DTT, pyrophosphatase and Triton X-100 all showed negligible impact in IVT (p = 0.20622, 0.56417, and 0.66715, respectively) (Table 3 and Figure 2). Furthermore, MgCl₂ also had the highest coefficient estimates compared to other components and its secondary interactions (Figure 3).

Table 3. The significance of each component varied and their interactions.

A log worth > 2 or a p-value < 0.01 is considered significant.

Source	Log worth	P-value
MgCl₂ (2.4, 24)	73.862	0.00000
NaCl*Spermidine	8.144	0.00000
Spermidine (0.2, 2)	4.070	0.00009
MgCl₂*Spermidine	4.070	0.00009
NaCl*DTT	2.536	0.00291
NaCl (2.5, 25)	1.857	0.0139
NaCl* MgCl₂	1.857	0.0139
NaCl*Triton X-100	1.631	0.0234
DTT*Triton X-100	1.391	0.0406
Spermidine*DTT	1.233	0.0585
DTT (4, 40)	0.686	0.206
Pyrophosphatase (0.01, 0.1)	0.249	0.564
Triton X-100 (0, 0.01)	0.176	0.667

Figure 2. The most significant components varied based on a full two-level factorial DoE analysis.

A log worth > 2 is considered significant. Each variable of the DoE was done in triplicates.

Figure 3. The coefficient estimates of each component and their secondary interactions based on a full two-level factorial DoE analysis.

Each component’s coefficient estimate corresponds to the change in the mean response for each level and the average response across all levels.

The model equation showing the mathematical relationship between each variable in order to give a specific RNA yield is:

664.171875 + - 52.47395833 * (\frac{(NaCl - 13.75)}{11.25}) + 664.171875 * (\frac{({MgCl}_{2} - 13.2)}{10.8}) + 84.953125 * (\frac{(Spermidine - 1.1)}{0.9}) + 26.796875 * (\frac{(DTT - 22)}{18}) + 12.203125 * (\frac{(Pyrophosphatase - 0.055)}{0.045}) + 9.0989583333 * (\frac{(Triton X - 100 - 0.005)}{0.005}) + (\frac{(NaCl - 13.75)}{11.25}) * ((\frac{({MgCl}_{2} - 13.2)}{10.8}) * - 52.47395833) + (\frac{(NaCl - 13.75)}{11.25}) * ((\frac{(Spermidine - 1.1)}{0.9}) * - 128.4010417) + (\frac{({MgCl}_{2} - 13.2)}{10.8}) * ((\frac{(Spermidine - 1.1)}{0.9}) * 84.953125) + (\frac{(NaCl - 13.75)}{11.25}) * ((\frac{(DTT - 22)}{18}) * - 63.765625) + (\frac{(Spermidine - 1.1)}{0.9}) * ((\frac{(DTT - 22)}{18}) * 40.223958333) + (\frac{(NaCl - 13.75)}{11.25}) * ((\frac{(Triton X - 100 - 0.005)}{0.005}) * - 48.296875) + (\frac{(DTT - 22)}{18}) * ((\frac{(Triton X - 100 - 0.005)}{0.005}) * - 43.56770833)

Acetate ions are more effective than chloride ions in enhancing RNA yield

We then sought to examine the interactions between the concentrations of magnesium (Mg⁺²), sodium (Na⁺) and NTPs as well as the significance of varying the ion type between acetate and chloride. Hence, a full three-level factorial design of experiment was performed with three factors (Mg⁺², Na⁺, NTPs) varied at three different concentrations (Table 4). Because DTT showed no significance in the previous full two-level factorial DoE but had an estimate of more than zero, we decided to use 40 mM DTT as the fixed component for this DoE. Furthermore, the previous DoE showed that spermidine mainly had significant secondary interactions with NaCl and MgCl₂ but because the main focus of this DoE is to better understand the main interactions between Na⁺, Mg⁺² and NTPs, we decided to use 0.2 mM spermidine as the fixed component here.

Table 4. The components used for a full three-level factorial DoE.

Fixed components	Concentration of fixed components	Components that were varied	Concentrations of components varied
Linearized DNA Template	1 μg	Sodium Chloride (NaCl) or Sodium Acetate (NaOAc)	0 mM
HEPES	40 mM		10 mM
Dithiothreitol (DTT)	40 mM		100 mM
RNase Inhibitor	4 U	Magnesium Chloride (MgCl₂) or Magnesium Acetate (MgOAc₂)	10 mM
T7 RNA Polymerase	100 U		75 mM
Spermidine	0.2 mM		140 mM
		ATP, CTP, UTP and GTP	5 mM each
			10 mM each
			20 mM each

We found that the ion type plays a significant role in IVT (Table 5 and Figure 4A) with acetate ions being more effective than chloride ions (p = 0.00906) (Figure 4B). Further examination of interaction between Mg⁺² and NTPs indicated that the highest RNA yield was achieved with a Mg⁺² concentration of 75 mM and NTPs concentration of 10 mM each (Figure 4C). Further increasing the concentration of each NTP had a deleterious effect on IVT, causing RNA yield to plateau. The coefficient estimates graph also showed that acetate ions have the highest estimate (Figure 5). These data suggest the optimal balance of Mg⁺² and NTPs needed for effective IVT and maximal RNA quantity requires a molecular ratio of 1: 1.875 between total NTPs and Mg⁺².

Table 5. The significance of each component varied.

A log worth > 2 or a p-value < 0.01 is considered significant.

Source	Log worth	P-value
Ion Type	2.043	0.00906
NTPs	1.898	0.0127
Mg⁺²	1.207	0.0621
Na⁺	0.448	0.357

Figure 4. A full three-level factorial DoE analysis of saRNA synthesis through IVT.

(A) The significance of each component varied represented in a graph; (B) The effects of each component varied on the RNA yield is plotted next to each other to understand the relationship between each component in order to produce high RNA yield; (C) The relationship between NTPs and Mg⁺² at different concentrations is shown in relation to the RNA yield. Each variable of the DoE was done in triplicates. A log worth > 2 is considered significant.

Figure 5. The coefficient estimates of each component and their secondary interactions based on a full three-level factorial DoE analysis.

Each component’s coefficient estimate corresponds to the change in the mean response for each level and the average response across all levels.

75 mM MgOAc₂ is the optimal concentration for all incubation periods

Based on the observation that Mg⁺² is the most significant component for IVT, and that acetate ions result in higher yield than chloride ions, we wanted to examine the optimal concentration of MgOAc₂ and incubation time in order to produce maximal RNA yield. The concentration of NTPs were kept at 10 mM each while varying the concentrations of MgOAc₂ ranging from 25–125 mM. RNA yields were measured after 2, 4 and 6 h of incubation. As shown in Figure 6, the general trend shown for all timepoints is that the RNA yield reaches its highest at 75mM MgOAc₂ and then it slowly decreases with higher concentrations of MgOAc₂. In addition, the longer the incubation period, the higher the RNA yield.

Figure 6. The impact of MgOAc₂ on saRNA synthesis through IVT at different timepoints.

Various IVT reactions were set up with different concentrations of MgOAc₂ and the RNA yields were measured at 2, 4 and 6 h. Each MgOAc₂ concentration varied was done in triplicates.

RNA yield is proportional to T7 RNA polymerase concentration at all timepoints

Because many commercial kits recommend using 100U of T7 RNA polymerase for IVT, we next wanted to explore whether lowering the T7 RNA polymerase concentration would yield the same RNA activity compared to 100 U of T7 RNA polymerase while increasing the incubation time. Hence, the optimal condition for IVT was used while varying only the concentration of T7 RNA polymerase, ranging from 12.5 to 100 U. Again, RNA yields were measured at 2, 4 and 6 h. We observed that lowering the T7 RNA polymerase concentration hinders the IVT process for all timepoints, leading to lower yields of RNA (Figure 7).

Figure 7. The effects of T7 RNA polymerase concentration on RNA yield at three different timepoints.

Four different concentrations of T7 RNA polymerase were used for each IVT reaction and the RNA yields were measured at 2, 4 and 6 h. Each T7 RNA polymerase concentration varied was done in triplicates.

Increasing the salt concentration during IVT process at different timepoints does not affect final RNA concentration

Because the binding efficiency of T7 RNA polymerase is dependent on the ionic strength and the low-ionic-strength kinetics of RNA synthesis have been shown to plateau over time,²³ we then wanted to investigate whether increasing the salt concentration in the IVT reactions at different timepoints of the process might lead to higher RNA yields. Each reaction started off with 10 mM NaOAc and after incubating the reactions for 10, 30 and 60 min, various concentrations of NaOAc were added to the reaction to make a final concentration of 0.02, 0.1 and 0.5 M. Interestingly, the addition of different NaOAc concentrations yielded lower RNA concentrations than the reactions that had no increase in salt concentration (Figure 8), demonstrating that increasing the salt concentration during IVT process does not facilitate higher RNA yields.

Figure 8. The effects of adding various NaOAc concentrations at different timepoints after the initiation of IVT on RNA yield.

Different concentrations of NaOAc were added at either 10 min, 30 min or 1 h after IVT was initiated and the RNA yields of these reactions were measured at 2 h. The RNA yields of the reactions with no addition of sodium acetate were measured at 10 min, 30 min, 1 h and 2 h as a control. All conditions were done in triplicates.

An optimal combination of pyrophosphatase, spermidine, DMSO, betaine and tween 20 is needed in order to produce high RNA yields

Next, we wanted to incorporate pyrophosphatase, spermidine, DMSO, betaine and a surfactant, either Triton X-100 or Tween 20, in order to see the interactions between each of these components in IVT. However, titrating these components individually would not allow us to determine these interactions. Therefore, we used a definitive screening design (DSD) of experiment where we took three different concentrations (a low, medium, and high point for each variable) and perform a minimum number of reactions to screen for effects and interactions, including quadratic effects. This allowed us to determine any curvature in the relationships between effects and estimate the local maximum. The concentrations used for each component are shown in Table 6. All RNA yields were measured at 4 and 6 h. Because the previous full three-level factorial DoE showed that 85 mM MgOAc₂ worked best for a 4 h incubation period, we used 85 mM for this DoE. A Fit Definitive Screening platform was used to analyze the data in order to identify the main effects of IVT by using an Effective Model Selection for DSDs approach.

Table 6. The components used for a definitive screening DoE.

Fixed components	Concentration of fixed components	Components that were varied	Concentrations of components varied
Linearized DNA Template	1 μg	Pyrophosphatase	0.02 U
HEPES	40 mM		0.1 U
Dithiothreitol (DTT)	40 mM		0.5 U
RNase Inhibitor	4 U	Spermidine	0 mM
T7 RNA Polymerase	100 U		0.2 mM
Sodium Acetate (NaOAc)	10 mM		1 mM
Magnesium Acetate (MgOAc₂)	85 mM		5 mM
ATP, CTP, UTP and GTP	10 mM each	Surfactant (Either Triton X-100 or Tween 20)	0%
			0.01%
			0.05%
		Betaine	0 M
			0.5 M
			1.5 M
		Dimethyl sulfoxide (DMSO)	0%
			2%
			10%

We observed that high yields of RNA were caused by a variation of an optimal combination of each component. The major effect estimates showed that spermidine, betaine and surfactant have the most impact on IVT with the surfactant being most significant (p = 0.0327) (Table 7). Interestingly, Tween 20 was shown to be more effective for increasing yield than Triton X-100 (Figure 9). However, pyrophosphatase and DMSO were shown to have a non-significant impact on IVT yield. In addition, for most of the components, high RNA yield is achieved only by using the medium concentration of each component rather than using the highest concentration, which suggests that too much of each component can result in a level that hinders saRNA production during IVT (Figure 9).

Table 7. The components that had major effects on IVT in a definitive screening DoE.

Main effect estimates
Term	Estimate	Std error	t Ratio	Prob > ǀtǀ
Spermidine	-491.3	211.58	-2.322	0.0593
Surfactant	584.79	211.58	2.674	0.0327*
Betaine	439.98	211.58	2.0795	0.0828

* A Prob > ǀtǀ value of <0.05 is considered significant.

Figure 9. A definitive screening DoE analysis of saRNA production through IVT to observe the importance of pyrophosphatase, spermidine, surfactant (either Triton X-100 or Tween 20), betaine and DMSO on RNA yield.

Each component varied was plotted against RNA yield to see their individual impact on RNA yield.

A color map of correlations also showed large absolute correlations between each effect, indicating that not only one effect is responsible for achieving high RNA yields but rather a combination of all components at their optimal concentration (Figure 10).

Figure 10. A color map on correlations between each component based on a definitive screening DoE analysis.

The blue squares represent small correlations between each component. The gray squares represent correlations between components with quadratic effects. The red squares represent high correlations.|r|represents the absolute value of the correlation between components. Components with large absolute correlations indicate an increase in the standard errors of estimates.

Inclusion of pyrophosphatase does not enhance the quantity of RNA yield

Because many IVT protocols include pyrophosphatase for catalyzing the hydrolysis of any pyrophosphate byproduct which can limit polymerization rate and synthesize shorter RNA molecules,²⁴ we wanted to explore whether pyrophosphatase was necessary in generating high RNA concentrations. We compared our two current best conditions for IVT with and without pyrophosphatase and measured RNA concentrations after 2 and 4 h incubation. Both conditions yielded similar concentrations at each timepoint (Figure 11A). However, analysis of RNA quality by gel electrophoresis indicated that the condition without pyrophosphatase showed a slightly better quality of RNA for both the 2 and 4 h timepoints compared to the condition with pyrophosphatase (Figure 11B). This suggests that pyrophosphatase provided no benefit in generating high RNA yield under these optimized conditions.

Figure 11. Comparison of the current best conditions with and without pyrophosphatase.

(A) After conducting all DoEs, the optimal conditions with and without pyrophosphatase were set up for IVT to compare the RNA yield at 2 and 4 h timepoint. (B) The RNA replicons synthesized at 2 and 4 h were then purified and analzyed on a denaturing gel for quality control.

The optimal conditions of IVT can be used to synthesize saRNAs encoding for different genes of interest with good protein expression in vitro

We then wanted to confirm that our optimal IVT condition could be used to synthesize different sizes of RNA. Thus, we prepared four other pDNA templates for making saRNA encoding different GOIs that included enhanced green fluorescent protein (eGFP) (8.5 kb), multi-drug resistance-1 (MDR1) (11.5 kb) and multi-drug resistance-1 with parainfluenza virus 5 (MDR1-PIV5) (12.4 kb). In addition, we used a pDNA template encoding for fLuc to synthesize mRNA fLuc (1.8 kb). The sizes of each RNA species ranged from 1.8 to 12.4 kb. Our optimized IVT condition (Table 8) generated high yields of good quantity and quality RNA irrespective of size (Figure 12A and B). The total mass yielded for VEEV-fLuc, VEEV-eGFP, VEEV-MDR1, VEEV-MDR1-PIV5, and mRNA fLuc were 118, 150, 142, 169 and 41 μg, respectively; while the average moles of VEEV-fLuc, VEEV-eGFP, VEEV-MDR1, VEEV-MDR1-PIV5, and mRNA fLuc were 38.94, 54.77, 38.32, 42.48 and 70.28 pmol, respectively. Lastly, we transfected these RNAs into HEK293T.17 cells and harvested the cells 48 h post transfection in order to confirm the protein expression from each GOI for saRNAs and for the fLuc mRNA. We observed appropriate protein expression from each of the saRNA and the fLuc mRNA (Figure 12C).

Table 8. The final optmized components for IVT.

Components	Concentration of each component
Linearized DNA Template	1 μg
HEPES	40 mM
Dithiothreitol (DTT)	40 mM
RNase Inhibitor	4 U
T7 RNA Polymerase	100 U
Sodium Acetate (NaOAc)	10 mM
Magnesium Acetate (MgOAc₂)	75 mM
ATP, CTP, UTP and GTP	10 mM each
Spermidine	0.2 mM

Figure 12. Comparison of the current best condition without pyrophosphatase with different genes of interest.

(A) The current best condition without pyrophosphatase was used to synthesize RNA replicons containing different GOIs as well as mRNA fLuc through IVT for 2 h and the RNA yield was measured using the Qubit. Total mass and moles of each RNA were quantified. (B) RNAs were then purified and ran on a denaturing gel for quality control. (C) Each RNA was transfected into a 6-well plate of HEK293T.17 cells. At 48 h post transfection, cells were harvested and stained with appropriate antibodies to observe protein expression of each GOI by flow cytometry, in which the expression level is represented as the median fluorescence intensity (MFI).

Discussion

The aim of this study was to utilize a DoE approach to optimize the IVT process for long saRNAs so that maximal RNA yield can be achieved. Here, we found that magnesium plays the most important role in IVT, especially its interaction with NTPs where the right balance of these two components is needed to yield high RNA concentrations. In addition, acetate ions were more effective for IVT compared to chloride ions with IVT being most effective at 75 mM MgOAc₂ and 10 mM of each NTP. Maintenance of ionic strength through the addition of NaOAc at different timepoints of the IVT process did not enable the production of high RNA yields. We also found that lowering T7 RNA polymerase hinders IVT process and that pyrophosphatase is not significantly required for IVT in order to yield maximal RNA concentrations. Finally, we observed that the optimized condition for IVT could be used to produce a range of RNA, from 1.8 to 12.4 kb and that each type of RNA was functional and thus able to express proteins once transfected into cells.

Despite the T7 RNA polymerase being the most essential component in order to initiate IVT,²⁵ other factors influence optimal IVT productivity. Magnesium ions are known to influence the catalytic activity of T7 RNA polymerase.²⁶ However, the ratio of magnesium to NTPs is the critical parameter influencing efficient IVT. We observed that a combination of 10 mM of each NTP with 75 mM of magnesium anion produced optimal IVT RNA yield, where 5 mM of each NTP is half as good and 20 mM of each NTP being too much. This means the optimal molecular ratio between total NTPs and magnesium concentration is 1: 1.1875 (M/M). The relationship between NTPs and magnesium found in this study corresponds to previous research whereby if a low concentration of NTPs was used, it limits RNA production but high concentrations of NTPs become inhibitory.²⁷ One explanation for this phenomenon is that when hydrogen ions are released from the NTP during the formation of the magnesium-NTP complex, the pH of the reaction is reduced which blocks the binding between the T7 RNA polymerase and the DNA.²⁸ This indicates that the higher the NTP and magnesium concentration, the more hydrogen ions are released and the fall in pH is greater, subsequently inhibiting enzyme activity. Therefore, the optimal amount of NTPs must be used in relation to the amount of magnesium added initially in order to synthesize high yields of RNA.

Previous research has shown that acetate anions are more effective during IVT compared to chloride anions. Activity of the T7 RNA polymerase is more sensitive to the chloride anions when the polymerase binds to the T7 promoter, in which the chloride anion competes with the DNA to bind to specific protein sites.²⁹ This causes a decrease in transcription efficiency.³⁰ Hence, when we performed a DoE to explore the interactions between sodium, magnesium and NTPs, we also compared the differences between using chloride and acetate anions. Similar to previous studies of small RNA IVT, our data using long RNA also indicate that the ion type is a significant factor in IVT with acetate anions being more effective than chloride anions.

Interestingly, while many protocols use pyrophosphatase for IVT,²⁴^,³¹ we found that pyrophosphatase was not necessarily required to yield high RNA concentrations. When the T7 RNA polymerase binds to the DNA template, this complex binds to the magnesium-NTPs complex and pyrophosphate is released, in which this pyrophosphate byproduct has been shown to slow down IVT.²⁸ Hence pyrophosphatase is incorporated to eliminate any pyrophosphate byproducts. However, because the free magnesium ions precipitate pyrophosphate,²⁸ we hypothesize that pyrophosphatase had little impact in our DoEs because the magnesium levels were high enough to cope with the generated pyrophosphate levels.

Furthermore, previous research has found that RNA synthesis by DNA-dependent RNA polymerases like the T7 RNA polymerase is highly influenced by the ionic strength.²³^,³² Fuchs et al. found that increasing the ionic strength 10 min after the initiation of IVT to a high ionic strength allowed the IVT reaction to run maximally for 6 h.³² An explanation for this was that under low ionic strength, the T7 RNA polymerase is only able to make a single copy of the transcript and only has the ability to re-initiate the reaction to make additional copies when the ionic strength is high. However, too high of an ionic strength might interfere with the binding activity of the T7 RNA polymerase to the DNA template.³³ We observed no benefit from the addition of sodium acetate during our IVT reactions. This likely reflects that the ionic strength was already optimal in our IVT setup.

Conclusions

Since many IVT protocols have been optimized for shorter mRNAs, the application of the DoE approach proved instructive to understanding the interactions between each reaction component to produce maximal yields of long RNA. Further benefits could be gained by determining the optimal conditions for co-capping of saRNA during IVT. However, our optimized IVT reaction conditions that generate high yields of saRNA can be used to produce large quantities of saRNA for research and development purposes and have the potential to be scaled up for GMP production of saRNA vaccines and biotherapeutics.

Data availability

Underlying data

Zenodo: Underlying data for ‘Design-of-experiments in vitro transcription yield optimization of self-amplifying RNA’. https://doi.org/10.5281/zenodo.6149318²²

This project contains the following underlying data:

• Data file 1: Exp1 – Results IVT RNA FF DoE Updated (4 mM dNTPs) with predicted formula.jmp
• Data file 2: Exp1 – Results IVT RNA FF DoE Updated (4 mM dNTPs) with Adjusted Fit Model.pdf
• Data file 3: Exp2 – Results IVT RNA FF DoE (Na, Mg, NTPs, Ion Type).jmp
• Data file 4: Exp2 – Results IVT RNA FF DoE (Na, Mg, NTPs, Ion Type).xlsx
• Data file 5: Exp3 – Results Repeat IVT RNA FF Titrating MgOAc2.jmp
• Data file 6: Exp3 – Results Repeat IVT RNA FF Titrating MgOAc2.xlsx
• Data file 7: Exp4 – Results Titrating T7 polymerase.xlsx
• Data file 8: Exp5 – Results Addition of NaOAc.xlsx
• Data file 9: Exp6 – RNA IVT DSD Evaluate Design.pdf
• Data file 10: Exp6 – RNA IVT DSD.jmp
• Data file 11: Exp7 – Results compare best current recipes.xlsx
• Data file 12: Exp8 – Results Different GOIs.xlsx
• Data file 13: RNA gel – Current best condition without pyrophosphatase with different genes of interest. TIF
• Data file 14: RNA gel – Current best conditions with and without pyrophosphatase. TIF
• Data file 15: Table 1. The list of components used for the different DoEs.docx
• Data file 16: Table 2. The components used for a full two-level factorial DoE.docx
• Data file 17: Table 3. The significance of each component varied and their interactions.docx
• Data file 18: Table 4. The components used for a full three-level factorial DoE.docx
• Data file 19: Table 5. The significance of each component varied.docx
• Data file 20: Table 6. The components used for a definitive screening DoE.docx
• Data file 21: Table 7. The components that had major effects on IVT in a definitive screening DoE.docx
• Data file 22: Table 8. The final optimized components for IVT.docx

Data are available under the terms of the Creative Commons Attribution 4.0 International license (CC-BY 4.0).

Accession numbers

NCBI Gene: Synthetic construct DNA, codon optimized luciferase. Accession number AB762768.1, https://www.ncbi.nlm.nih.gov/nuccore/AB762768.1

References

1. Iavarone C, O'Hagan DT, Yu D, et al.: Mechanism of action of mRNA-based vaccines. Expert Rev. Vaccines 2017; 16(9): 871–881. Publisher Full Text
2. Zhang C, Maruggi G, Shan H, et al.: Advances in mRNA Vaccines for Infectious Diseases. Front. Immunol. 2019; 10(594). Publisher Full Text
3. Vogel AB, Lambert L, Kinnear E, et al.: Self-Amplifying RNA Vaccines Give Equivalent Protection against Influenza to mRNA Vaccines but at Much Lower Doses. Molecular Therapy: The Journal of the American Society of Gene Therapy 2018; 26(2): 446–455. PubMed Abstract | Publisher Full Text
4. Geall AJ, Verma A, Otten GR, et al.: Nonviral delivery of self-amplifying RNA vaccines. Proc. Natl. Acad. Sci. U. S. A. 2012; 109(36): 14604–14609. PubMed Abstract | Publisher Full Text
5. Johansson DX, Ljungberg K, Kakoulidou M, et al.: Intradermal electroporation of naked replicon RNA elicits strong immune responses. PLoS One 2012; 7(1): e29732. PubMed Abstract | Publisher Full Text
6. Bogers WM, Oostermeijer H, Mooij P, et al.: Potent immune responses in rhesus macaques induced by nonviral delivery of a self-amplifying RNA vaccine expressing HIV type 1 envelope with a cationic nanoemulsion. J. Infect. Dis. 2015; 211(6): 947–955. PubMed Abstract | Publisher Full Text
7. Moyo N, Vogel AB, Buus S, et al.: Efficient Induction of T Cells against Conserved HIV-1 Regions by Mosaic Vaccines Delivered as Self-Amplifying mRNA. Molecular Therapy. Methods & Clinical Development 2019; 12: 32–46. PubMed Abstract | Publisher Full Text
8. Brazzoli M, Magini D, Bonci A, et al.: Induction of Broad-Based Immunity and Protective Efficacy by Self-amplifying mRNA Vaccines Encoding Influenza Virus Hemagglutinin. J. Virol. 2016; 90(1): 332–344. PubMed Abstract | Publisher Full Text
9. Fleeton MN, Chen M, Berglund P, et al.: Self-replicative RNA vaccines elicit protection against influenza A virus, respiratory syncytial virus, and a tickborne encephalitis virus. J. Infect. Dis. 2001; 183(9): 1395–1398. PubMed Abstract | Publisher Full Text
10. Khalil SM, Tonkin DR, Mattocks MD, et al.: A tetravalent alphavirus-vector based dengue vaccine provides effective immunity in an early life mouse model. Vaccine. 2014; 32(32): 4068–4074. PubMed Abstract | Publisher Full Text
11. White LJ, Sariol CA, Mattocks MD, et al.: An alphavirus vector-based tetravalent dengue vaccine induces a rapid and protective immune response in macaques that differs qualitatively from immunity induced by live virus infection. J. Virol. 2013; 87(6): 3409–3424. PubMed Abstract | Publisher Full Text
12. Bates JT, Pickens JA, Schuster JE, et al.: Immunogenicity and efficacy of alphavirus-derived replicon vaccines for respiratory syncytial virus and human metapneumovirus in nonhuman primates. Vaccine. 2016; 34(7): 950–956. PubMed Abstract | Publisher Full Text
13. Pyankov OV, Bodnev SA, Pyankova OG, et al.: A Kunjin Replicon Virus-like Particle Vaccine Provides Protection Against Ebola Virus Infection in Nonhuman Primates. J. Infect. Dis. 2015; 212(Suppl 2): S368–S371. PubMed Abstract | Publisher Full Text
14. Reynard O, Mokhonov V, Mokhonova E, et al.: Kunjin virus replicon-based vaccines expressing Ebola virus glycoprotein GP protect the guinea pig against lethal Ebola virus infection. J. Infect. Dis. 2011; 204 Suppl 3(Suppl 3): S1060–S1065.
15. Herweijer H, Wolff JA: Self-amplifying vectors for gene delivery. Adv. Drug Deliv. Rev. 1997; 27(1): 5–16. PubMed Abstract | Publisher Full Text
16. Brito LA, Kommareddy S, Maione D, et al.: Self-amplifying mRNA vaccines. Adv. Genet. 2015; 89: 179–233. Publisher Full Text
17. Geall AJ, Mandl CW, Ulmer JB: RNA: the new revolution in nucleic acid vaccines. Semin. Immunol. 2013; 25(2): 152–159. PubMed Abstract | Publisher Full Text
18. Milligan JF, Uhlenbeck OC: Synthesis of small RNAs using T7 RNA polymerase. Methods Enzymol. 1989; 180: 51–62. Publisher Full Text
19. Daniel V, Sarid S, Beckmann JS, et al.: In vitro transcription of a transfer RNA gene. Proc. Natl. Acad. Sci. U. S. A. 1970; 66(4): 1260–1266. PubMed Abstract | Publisher Full Text | Free Full Text
20. Politis N, Colombo P, Colombo G, et al.: Design of experiments (DoE) in pharmaceutical development. Drug Dev. Ind. Pharm. 2017; 43(6): 889–901. Publisher Full Text
21. Inc., S. I: JMP® 14 Fitting Linear Models Cary, NC: SAS Institute Inc.; 2018.
22. Samnuan K, Blakney A, McKay P, et al.: Design-of-Experiments in vitro Transcription Yield Optimization of Self-Amplifying RNA [Data set]. Zenodo 2022.
23. Schafer R, Zillig W: The effects of ionic strength on termination of transcription of DNAs from bacteriophages T4, T5 and T7 by DNA-dependent RNA polymerase from Escherichia coli and the nature of termination by factor rho. Eur. J. Biochem. 1973; 33(2): 215–226. PubMed Abstract | Publisher Full Text
24. Cunningham PR, Ofengand J: Use of inorganic pyrophosphatase to improve the yield of in vitro transcription reactions catalyzed by T7 RNA polymerase. BioTechniques. 1990; 9(6): 713–714. PubMed Abstract
25. Slater RJ: Vitro Transcription. Techniques in Molecular Biology: Volume 2. Walker JM, Gaastra W, editors. Boston, MA: Springer US; 1987; pp 203–227.
26. Chamberlin M, Ring J: Characterization of T7-specific ribonucleic acid polymerase. 1. General properties of the enzymatic reaction and the template specificity of the enzyme. J. Biol. Chem. 1973; 248(6): 2235–2244. PubMed Abstract | Publisher Full Text
27. Milligan JF, Groebe DR, Witherell GW, et al.: Oligoribonucleotide synthesis using T7 RNA polymerase and synthetic DNA templates. Nucleic Acids Res. 1987; 15(21): 8783–8798. PubMed Abstract | Publisher Full Text | Free Full Text
28. Young JS, Ramirez WF, Davis RH: Modeling and optimization of a batch process for in vitro RNA production. Biotechnol. Bioeng. 1997; 56(2): 210–220. PubMed Abstract | Publisher Full Text
29. Maslak M, Martin CT: Effects of solution conditions on the steady-state kinetics of initiation of transcription by T7 RNA polymerase. Biochemistry. 1994; 33(22): 6918–6924. PubMed Abstract | Publisher Full Text
30. Kern JA, Davis RH: Application of solution equilibrium analysis to in vitro RNA transcription. Biotechnol. Prog. 1997; 13(6): 747–756. PubMed Abstract | Publisher Full Text
31. Nilsen TW, Rio DC, Ares M Jr: High-yield synthesis of RNA using T7 RNA polymerase and plasmid DNA or oligonucleotide templates. Cold Spring Harb. Protoc. 2013; 2013(11): pdb.prot078535. Publisher Full Text
32. Fuchse, Millette RL, Zillig W, et al.: Influence of salts on RNA synthesis by DNA-dependent RNA-polymerase from Escherichia coli. Eur. J. Biochem. 1967; 3(2): 183–193. PubMed Abstract | Publisher Full Text
33. Oakley JL, Strothkamp RE, Sarris AH, et al.: T7 RNA polymerase: promoter structure and polymerase binding. Biochemistry. 1979; 18(3): 528–537. Publisher Full Text

Comments on this article Comments (0)

Version 1

VERSION 1 PUBLISHED 18 Mar 2022

Author details Author details

¹ Section of Immunology of Infection, Department of Infectious Disease, Faculty of Medicine, Imperial College London, London, NW8 9NU, UK

Karnyart Samnuan
Roles: Conceptualization, Formal Analysis, Investigation, Methodology, Writing – Original Draft Preparation, Writing – Review & Editing

Anna K Blakney
Roles: Conceptualization, Formal Analysis, Methodology, Writing – Review & Editing

Paul F McKay
Roles: Conceptualization, Writing – Review & Editing

Robin J Shattock
Roles: Conceptualization, Supervision, Writing – Review & Editing

Competing interests

No competing interests were disclosed.

Grant information

This work was supported by the Department of Health and Social Care using UK Aid funding and is managed by the Engineering and Physical Sciences Research Council (EPSRC, grant number: EP/R013764/1 (Note: the views expressed in this publication are those of the authors and not necessarily those of the Department of Health and Social Care). AKB is supported by a Marie Skłodowska Curie Individual Fellowship funded by the European Commission H2020 (No. 794059). The Dormeur Investment Service Ltd provided funds to purchase equipment used in this study.
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Article Versions (1)

version 1

Published: 18 Mar 2022, 11:333

https://doi.org/10.12688/f1000research.75677.1

Copyright

© 2022 Samnuan K et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Download

Export To

metrics

	Views	Downloads
F1000Research	-	-
PubMed Central Data from PMC are received and updated monthly.	-	-

Citations

0

SEE MORE DETAILS

CITE

how to cite this article

Samnuan K, Blakney AK, McKay PF and Shattock RJ. Design-of-experiments in vitro transcription yield optimization of self-amplifying RNA [version 1; peer review: 2 approved with reservations]. F1000Research 2022, 11:333 (https://doi.org/10.12688/f1000research.75677.1)

NOTE: If applicable, it is important to ensure the information in square brackets after the title is included in all citations of this article.

track

receive updates on this article

Track an article to receive email alerts on any updates to this article.

Open Peer Review

Version 1

VERSION 1

PUBLISHED 18 Mar 2022

Views

29

Reviewer Report 20 Jul 2023

Emily Voigt, Access to Advanced Health Institute (AAHI), Seattle, WA, USA

Devin Brandt, Access to Advanced Health Institute (AAHI), Seattle, WA, USA

Approved with Reservations

https://doi.org/10.5256/f1000research.79579.r179546

The overall findings and experimental design of “Design-of-experiments in vitro transcription yield optimization of self-amplifying RNA” (Samnuan et al) represents a valuable first step towards understanding the optimal conditions for the in vitro transcription of long RNA molecules. These saRNAs ... Continue reading

The overall findings and experimental design of “Design-of-experiments in vitro transcription yield optimization of self-amplifying RNA” (Samnuan et al) represents a valuable first step towards understanding the optimal conditions for the in vitro transcription of long RNA molecules. These saRNAs are increasingly used in clinical trials, and published information on optimization of their manufacturing is certainly worth publishing. However, the unique biology of saRNAs makes them a somewhat difficult target for studies such as this unless readouts are carefully selected to accurately come to the correct conclusions, which is not adequately addressed in the current manuscript. Due to their self-replicating nature, the in-vivo behavior of these molecules (such as the expression of a reporter gene) is often difficult to parse-out and directly correlate to something as simple as RNA yield. The 5’ and 3’ UTRs of these molecules are critical to their efficacy in vivo, and the loss or alteration of even a few nucleotides is sufficient to alter or eliminate replication in vivo and dramatically affect a drug product. As such, even nearly-full-length saRNA molecules do not behave as anticipated in vivo. Unlike traditional mRNA transfections (which have a nice linear relationship between translation of a transgene and the amount of mRNA directly delivered to a cell) saRNA-based transfections often exhibit paradoxical inverse-dose dependance, where lower amounts of saRNA give higher transgene expression.

Specific comments:

Introduction: From a strictly functional perspective, the data presented are valuable for optimizing saRNA-based IVT reactions, but the authors do not provide sufficient motivation for this optimization within the introduction. This is instead handled within the discussion, which makes some of the reasoning a bit difficult to follow upon initial reading. Please edit appropriately.
Results: The authors ability to increase yields so dramatically is certainly worthy of publication, but some of the rationale behind certain selection criteria (such as screen various NaCl conditions) are left to the readers imagination. For example, the NaCl inclusion is ostensibly designed as a screen for various ionic strengths, but this isn’t stated explicitly. Please add language to make rationale for screening parameters, etc., more explicit.
The large spread within groups in datasets makes some of the optimization reasoning suspect, as it is difficult to gauge whether the average IVT reaction yield was higher at a given condition due to the purposeful variation of parameters, or due to the inherent inconsistencies of IVT reactions as a whole. Please address.
The most obvious omission in the DoE is there is no screen of temperature conditions, which would affect not just the activity of the transcriptase, but also the amount of secondary structure present in the nascent RNA molecule, which has strong effects on the likelihood of successful transcription of the whole saRNA. Please address why this parameter was omitted, and what effects this may or may not have on the optimization process.
The in vivo data is not thorough enough to warrant inclusion, and the results seem somewhat counterintuitive to the stated goal of optimizing IVT conditions. In vivo efficacy includes other parameters (like method of delivery and capping efficiency) which are outside of the scope of the manuscript as presented. This should be addressed. One option is to eliminate these plots, and the in vivo work entirely, and focus instead on a more rigorous study of the RNA in an in vitro setting. This may help better focus the manuscript and seems more in line with their stated goal. The other clear option is to better justify the inclusion of the in vivo data, address and control for such non-DOE-optimized parameters such as capping, formulation characteristics and delivery characteristics.
The selection of certain DOE parameters seems obvious for certain parameters (NTPs, T7, Mg++) but others are a bit more cryptic. The choice of NaCl is interesting, but more motivation should be given to the choice prior to the discussion.
The use of a ribogreen assay for RNA quantification is problematic as performed in the study. The assay cannot distinguish between abortive transcripts and full-length RNA molecules, and thus cannot provide an accurate measure of yield. saRNAs are notorious for giving abortive transcripts during transcription (most likely due to their extensive secondary structure) and thus a better metric is needed to measure concentration of the full length products. Gel-based densitometry, or RT-qPCR with probes designed to anneal to the 5’ and 3’ UTR would help refine these measurements and provide measures not only of RNA concentration but also % full-length transcripts, etc., as the key products.
In the presented RNA gel, there appears to be significant degradation in all RNA samples, and the smearing looks like non-specific hydrolysis as opposed to RNAse contamination. This may be due to the divalent counterions present in the transcription. Please address this apparent degradation.
For the RNA gel figure, more experimental detail needs to be included, including the amount of RNA run on the gel, the type of denaturant used (e.g. formamide vs formaldehyde) and the running conditions of the gel.
The in vivo data lacks any sort of statistical treatment, and there are no errors bars in the graph as presented. The reported MFI isn’t the best parameter to report, as the control is an mRNA with very different expression and delivery kinetics than any saRNA. The fact that the saRNA groups have similar MFI as the control seems to indicate that the saRNA is actually not replicating as intended, which may be due to the lack of capping for the saRNA (there was no mention of the capping step in the protocol – which also must be rectified). Because the authors are looking at the replication of a functional saRNA, the sequence of that saRNA should be reported, and the construct should be sequenced to ensure it is theoretically functional.
The discussion presented is quite short, and some of the text should be moved up into the introduction section as mentioned above. The section rationalizing the NaCl screen would be much better if it were included in the introduction, as it would provide a good motivation for pursuing those experiments. There is little discussion of the in vivo results, and I suspect this is due to the less-than-stellar response of the transfected cells. If the authors choose to remove these studies, and instead do a more in-depth analysis of the RNA integrity and stability, there would be a good amount to discuss here. As is, the discussion is a bit limited in scope, and outside of the final optimized protocol and recipe, offers little in terms of a satisfying conclusion overall.
Data availability: We are missing the cytometry data and gating parameters used for population selection. These should be added to supplemental information.

Is the work clearly and accurately presented and does it cite the current literature?

Yes
Is the study design appropriate and is the work technically sound?

Partly
Are sufficient details of methods and analysis provided to allow replication by others?

Yes
If applicable, is the statistical analysis and its interpretation appropriate?

Yes
Are all the source data underlying the results available to ensure full reproducibility?

Partly
Are the conclusions drawn adequately supported by the results?

Yes

Competing Interests: EAV is named inventor on patent applications regarding thermostable self-amplifying RNA vaccine formulations.

Reviewer Expertise: RNA Vaccines

We confirm that we have read this submission and believe that we have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however we have significant reservations, as outlined above.

CITE

Report a concern

Respond or Comment

Views

85

Reviewer Report 14 Apr 2022

Rok Sekirnik, Chemistry Research Laboratory, Department of Chemistry, University of Oxford, Oxford, UK

Approved with Reservations

https://doi.org/10.5256/f1000research.79579.r127937

In their research article 'Design-of-experiments in vitro transcription yield optimization of self-amplifying RNA', Samnuan et al. describe a comprehensive study of factors that influence the productivity of IVT reaction for production of self-amplifying RNA (saRNA). In the article, the authors ... Continue reading

In their research article 'Design-of-experiments in vitro transcription yield optimization of self-amplifying RNA', Samnuan et al. describe a comprehensive study of factors that influence the productivity of IVT reaction for production of self-amplifying RNA (saRNA). In the article, the authors use the DoE approach to reduce the number of factors that could potentially affect the productivity of the reaction down to a handful, which are then evaluated in detail. The authors report that the major positive findings of the study are the key role of magnesium for productivity, and the relatively positive effect of acetate versus chloride. Furthermore, they conclude that pyrophosphatase does not positively impact the productivity of the reaction, and that increasing salt concentration during the reaction does not increase, but rather decreases productivity of IVT. Authors achieve a significant increase in saRNA yield, from starting 100 ng/uL to 4500 ng/uL.

Published studies on factors that might impact productivity of IVT reaction are relatively limited, and the manuscript is a welcome addition to the knowledge base. In their introduction, authors cite IVT protocols optimized for small RNAs (<100 nt), but do not refer to any recent protocols optimized for larger RNAs, e.g. mRNA, which, although not as large as saRNA, should still be considered highly relevant for their study, e.g. Henderson, J. M., et al., Curr Protoc., 2021. ¹ This is an important point from the perspective of factor selection. In the introductory paragraph #4, authors list the factors selected for DoE study, without theoretical justification. This paragraph should be moved to the discussion, where the authors currently provide the rationale for the selection of individual factors with references to literature. Because of this inverted structure, the introduction lacks a theoretical basis underpinning the study. The introduction would benefit from a more thorough review of current literature on key factors known to impact the activity of T7 polymerase (particular attention should be paid to the literature on the role of Mg²⁺ for activity of T7).

Selection of NaCl as a factor is surprising, and not justified by the authors. Authors initially exclude plasmid concentration as a factor, keeping it constant throughout the study at 1 mg/100 mL. A reference to published mRNA protocols would justify the selection of this value in context of mRNA production protocols. Whether the same mass of plasmid encoding for production of saRNA can be used without optimization is questionable, as the number of plasmid copies will thus be correspondingly lower compared to a standard mRNA IVT reaction, and reaction kinetics therefore correspondingly slower. Authors do not address this issue anywhere in the manuscript. Furthermore, the Methods section lacks information on the size of plasmid, so its molarity in the IVT protocol cannot be determined.

The study relies heavily on Qubit RNA Broad Range Assay. Authors do not comment on the limit of quantification (LoQ), selectivity or reproducibility of this assay. RNA yields in DoE study vary from <100 ng/uL to 4500 ng/uL. Authors should provide evidence that the assay response is linear in this range, e.g. qualify the assay with purified saRNA). Additionally, they should comment on whether the Qubit content method, is selective for target molecule (nominal molecular weight) only, or whether degradation products also contribute to 'yield' (see e.g. Figure 11B). Assay sensitivity and reproducibility is particularly relevant for evaluating one of two major conclusions of the paper, namely the impact of ion type on productivity. Early in the study (step 2 of DoE) authors perform a comparative study between MgCl₂ and MgOAc₂ and conclude that acetate leads to higher productivity; MgOAc₂ is thus fixed and used throughout the remainder of the study. Methodologically this is problematic as the absolute differences in productivity between chloride and acetate are minimal (100 ng/uL vs 200 ng/uL; ref Figure 4B); and the effect is not confirmed once the productivity of IVT is increased to >4000 ng/uL, where the significance would be demonstrated more clearly. Authors claim that they performed each IVT measurement in triplicate. Error bars should be added to each data figure. Significance of model equation (p. 6) is not clear. This should be moved to supplementary materials.

In Table 1 authors list the original stock concentrations for each chemical used in the study, with identification of the manufacturer. Catalogue numbers should be added for reagents used; if authors prepared the stock concentrations from solid powders, they should also report the hydration form used. This is particularly relevant for comparison of MgCl2 and MgOAc2; the authors should verify that correct molecular formula was used (e.g. MgCl2 hexahydrate) when calculating molarity of stock solutions.
Table 2, Table 3, Figure 2 and Figure 3 describe a single experiment, i.e. the full two-level factorial DoE. For clarity, Table 3 and Figure 3 should be moved to supplementary materials to help the flow of the manuscript. Table 2 and Figure 2 should stay in main text, as they are critical for interpretation of the experiment. A similar change should be made for the three-level factorial DoE analysis, i.e. only Table 4 and Figure 4 help the reader understand the significance of results and should be presented in main text; Table 5 and Figure 5 should be moved to supplementary. Legend to Figure 4 should state the timepoint at which yield was measured.

Optimal NTP concentration is identified as 10 mM of each NTP. Authors do not further optimize this parameter. Since the duration of IVT is not stated, it is difficult to interpret the data (would productivity be higher if experiment was performed longer?). Did authors try to adjust NTP concentrations based on saRNA sequence? The authors should also address the question of what is the limiting factor for productivity, i.e. why 10 mM leads to higher production of saRNA than 5 mM and why 20 mM is inhibitory. This is only partly addressed in the Discussion with the explanation that Mg-NTP complex leads to release of a proton with a concomitant decrease in pH which inhibits T7 polymerase. However, this explanation seems partial and does not address the impact of NTP/Mg on T7 polymerase kinetics.

Figure 6 presents a key experiment of the study. The data is fitted, but the fitting function is not described. Did the authors consider plotting the yield as a function of time to evaluate the effect of Mg²⁺ on reaction kinetics / saRNA stability? How do they interpret the decrease in saRNA concentration >75 mM MgOAc2 after 4h? Did they observe precipitation or degradation? This is particularly relevant in light of their conclusion that pyrophosphatase is not required to promote the reaction. Mg pyrophosphate is highly insoluble; in absence of pyrophosphatase this precipitation could either inhibit the reaction or affect content determination; authors should address this in the discussion.

With optimization of MgOAc2 concentration, authors successfully increased the productivity from ~1000 ng/uL to ~3000 ng/uL, which is very impressive and deserves a more explicit mention in the text. In Figure 6 and related discussion, authors state that 75 mM MgOAc2 and longer (i.e. 6 h) is optimal for increasing productivity. In the final optimization experiment, however, authors select a different experimental set-up which they claim to be optimal on basis of previous results (85 mM MgOAc2 and 4h; p. 11), which does not seem to be in line with results presented in Figure 6 and related discussion. It is understood that a shorter, i.e. 4 h reaction is preferred for practical reasons, and to avoid degradation, but this is not explicitly mentioned. Discussions for both figures should therefore be unified in this respect, while the experiment does not need to be repeated. They then address the effect of T7 polymerase, which is shown to be concentration-dependent, as demonstrated by decreasing the concentration from the nominal value of 100 U/100 uL. Given that this concentration is significantly below the concentrations recommended by manufacturers (authors reference to 'many commercial kits using 100 U' should be referenced for clarity. The concentrations used in published protocols for production of mRNA suggest significantly higher concentrations, e.g. 8 U/uL ¹, corresponding to 800 U in case of saRNA reactions as performed by authors. It would seem obvious to try to increase the concentration of polymerase to increase the yield. Authors should discuss this in the text. Instead, they focus on a poorly justified experiment to test the effect of addition of salt concentration over time. It is not clear why salt should be added at different timepoints, rather than at t=0. The results appear to be inconsistent with the positive effect of NaCl they reported in Table 3. Similarly, NaOAc was used in full three-level factorial reported in Table 4/Figure 4, but the result is not referenced in context of Figure 8. The authors do not comment on the variability of control reaction; the control experiment reported in Figure 8 reaches ~25% higher yield compared to same experiment reported in Figure 7 (time-point 2h). Is this to be understood as inter-experimental error of the study? This question deserves a wider treatment and is intimately linked to the comment about the selection of analytical method.

The DSD experiment leads to a further increase productivity to ~4500 ng/uL. This is a very impressive result, but does not get an explicit mention in the text. Table 7 is not a helpful for interpretation of the data, and should be moved to Supplementary. Figure 9 suggests that high productivity (~4500 ng/uL) can be achieved with all additives tested (pyrophosphatase, surfactant, spermidine, betaine, DMSO); since the design of the experiment is not described, additional interpretation of data is needed. The heat-map (Figure 10) is helpful in this respect, but a reader would benefit from an explicit interpretation (e.g. a table of optimized reaction conditions, current Supplementary Table 8, which should be moved to main text) of which minimal combinations of additives are required to reach productivity of 4500 ng/uL.

Finally, authors test the impact of pyrophosphatase 'using two best conditions for IVT'. It is not specified what those conditions are (see also point above). Figure 11 indicates that productivity of ~3000 ng/uL was reached, which is significantly lower than best conditions shown in Figure 10 (~4500 ng/uL), this discrepancy should be justified in the text. Denaturing gel (Figure 11B) shows a high degree of smearing, while authors comment on a 'slightly better quality' in absence of pyrophosphatase, but this interpretation does not appear to be possible based on PAGE alone. Based on PAGE profile as presented in Figure 11, authors should also comment on whether quality of saRNA is considered acceptable, and what are the degradation products observed by PAGE (degradation or abortive transcripts. Furthermore, a comment on how quantification of IVT compares pre- and post-purification (e.g. material from Figure 11B vs material from Figure 12B is needed. Authors should comment on any observations of precipitation in presence/absence of pyrophosphatase.

Discussion

Authors should add an explicit evaluation of the absolute increase in productivity they have achieved. The discussion on the optimal NTP:Mg ratio is thin; there is a critical typo citing the ratio as '1:1.1875' which should be corrected to '1:1.875' (p. 15). High NTP concentrations are interpreted as inhibitory due to lowering of pH; this should be demonstrated experimentally. There is a highly problematic statement concerning magnesium and pyrophophate. Authors claim that magnesium precipitates pyrophosphate, therefore justifying the omission of pyrophosphatase. This implies that optimized IVT reaction for production of saRNA is expected to lead to precipitation. Authors should comment on this implication and discuss whether precipitation is selective (i.e. that saRNA does not co-precipitate), that precipitation does not impact yield determination and that precipitation is not problematic for down-stream purification. In the conclusion, it is stated that the process as optimized can be used to produce large quantities of saRNA with potential for GMP production, so this seems a critical point for commercial utilization of the process.

Overall, the manuscript presents a comprehensive study of factors contributing to productivity of IVT for production of saRNA, a novel RNA-based modality with high therapeutic relevance. The conclusions of the paper are therefore important for driving process improvements in saRNA programmes in particular for reducing the production costs due to higher productivity of the IVT process step. Some edits are required to improve readability, clarify missing details to enable reproducibility, and provide further justification where indicated in the review above. Once the concerns raised above have been addressed, I recommend accepting this manuscript for indexing.

Is the work clearly and accurately presented and does it cite the current literature?

Partly
Is the study design appropriate and is the work technically sound?

Yes
Are sufficient details of methods and analysis provided to allow replication by others?

Yes
If applicable, is the statistical analysis and its interpretation appropriate?

I cannot comment. A qualified statistician is required.
Are all the source data underlying the results available to ensure full reproducibility?

Yes
Are the conclusions drawn adequately supported by the results?

Partly

References

1. Henderson JM, Ujita A, Hill E, Yousif-Rosales S, et al.: Cap 1 Messenger RNA Synthesis with Co-transcriptional CleanCap® Analog by In Vitro Transcription.Curr Protoc. 2021; 1 (2): e39 PubMed Abstract | Publisher Full Text

Competing Interests: No competing interests were disclosed.

Reviewer Expertise: in vitro transcription, nucleic acid chromatography, nucleic acid analytics

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.

CITE

Report a concern

Respond or Comment

Comments on this article Comments (0)

Version 1

VERSION 1 PUBLISHED 18 Mar 2022

Open Peer Review

Reviewer Status

Reviewer Reports

	Invited Reviewers
	1	2
Version 1 18 Mar 22	read	read

Rok Sekirnik, University of Oxford, Oxford, UK
Emily Voigt, Access to Advanced Health Institute (AAHI), Seattle, USA

Devin Brandt, Access to Advanced Health Institute (AAHI), Seattle, USA

Comments on this article

All Comments(0)

Add a comment

Sign up for content alerts

Browse by related subjects

Back to all reports

Reviewer Report

29 Views

20 Jul 2023 | for Version 1

Emily Voigt, Access to Advanced Health Institute (AAHI), Seattle, WA, USA

Devin Brandt, Access to Advanced Health Institute (AAHI), Seattle, WA, USA

29 Views Cite this report Responses(0)

Approved With Reservations

The overall findings and experimental design of “Design-of-experiments in vitro transcription yield optimization of self-amplifying RNA” (Samnuan et al) represents a valuable first step towards understanding the optimal conditions for the in vitro transcription of long RNA molecules. These saRNAs are increasingly used in clinical trials, and published information on optimization of their manufacturing is certainly worth publishing. However, the unique biology of saRNAs makes them a somewhat difficult target for studies such as this unless readouts are carefully selected to accurately come to the correct conclusions, which is not adequately addressed in the current manuscript. Due to their self-replicating nature, the in-vivo behavior of these molecules (such as the expression of a reporter gene) is often difficult to parse-out and directly correlate to something as simple as RNA yield. The 5’ and 3’ UTRs of these molecules are critical to their efficacy in vivo, and the loss or alteration of even a few nucleotides is sufficient to alter or eliminate replication in vivo and dramatically affect a drug product. As such, even nearly-full-length saRNA molecules do not behave as anticipated in vivo. Unlike traditional mRNA transfections (which have a nice linear relationship between translation of a transgene and the amount of mRNA directly delivered to a cell) saRNA-based transfections often exhibit paradoxical inverse-dose dependance, where lower amounts of saRNA give higher transgene expression.

Specific comments:

Introduction: From a strictly functional perspective, the data presented are valuable for optimizing saRNA-based IVT reactions, but the authors do not provide sufficient motivation for this optimization within the introduction. This is instead handled within the discussion, which makes some of the reasoning a bit difficult to follow upon initial reading. Please edit appropriately.
Results: The authors ability to increase yields so dramatically is certainly worthy of publication, but some of the rationale behind certain selection criteria (such as screen various NaCl conditions) are left to the readers imagination. For example, the NaCl inclusion is ostensibly designed as a screen for various ionic strengths, but this isn’t stated explicitly. Please add language to make rationale for screening parameters, etc., more explicit.
The large spread within groups in datasets makes some of the optimization reasoning suspect, as it is difficult to gauge whether the average IVT reaction yield was higher at a given condition due to the purposeful variation of parameters, or due to the inherent inconsistencies of IVT reactions as a whole. Please address.
The most obvious omission in the DoE is there is no screen of temperature conditions, which would affect not just the activity of the transcriptase, but also the amount of secondary structure present in the nascent RNA molecule, which has strong effects on the likelihood of successful transcription of the whole saRNA. Please address why this parameter was omitted, and what effects this may or may not have on the optimization process.
The in vivo data is not thorough enough to warrant inclusion, and the results seem somewhat counterintuitive to the stated goal of optimizing IVT conditions. In vivo efficacy includes other parameters (like method of delivery and capping efficiency) which are outside of the scope of the manuscript as presented. This should be addressed. One option is to eliminate these plots, and the in vivo work entirely, and focus instead on a more rigorous study of the RNA in an in vitro setting. This may help better focus the manuscript and seems more in line with their stated goal. The other clear option is to better justify the inclusion of the in vivo data, address and control for such non-DOE-optimized parameters such as capping, formulation characteristics and delivery characteristics.
The selection of certain DOE parameters seems obvious for certain parameters (NTPs, T7, Mg++) but others are a bit more cryptic. The choice of NaCl is interesting, but more motivation should be given to the choice prior to the discussion.
The use of a ribogreen assay for RNA quantification is problematic as performed in the study. The assay cannot distinguish between abortive transcripts and full-length RNA molecules, and thus cannot provide an accurate measure of yield. saRNAs are notorious for giving abortive transcripts during transcription (most likely due to their extensive secondary structure) and thus a better metric is needed to measure concentration of the full length products. Gel-based densitometry, or RT-qPCR with probes designed to anneal to the 5’ and 3’ UTR would help refine these measurements and provide measures not only of RNA concentration but also % full-length transcripts, etc., as the key products.
In the presented RNA gel, there appears to be significant degradation in all RNA samples, and the smearing looks like non-specific hydrolysis as opposed to RNAse contamination. This may be due to the divalent counterions present in the transcription. Please address this apparent degradation.
For the RNA gel figure, more experimental detail needs to be included, including the amount of RNA run on the gel, the type of denaturant used (e.g. formamide vs formaldehyde) and the running conditions of the gel.
The in vivo data lacks any sort of statistical treatment, and there are no errors bars in the graph as presented. The reported MFI isn’t the best parameter to report, as the control is an mRNA with very different expression and delivery kinetics than any saRNA. The fact that the saRNA groups have similar MFI as the control seems to indicate that the saRNA is actually not replicating as intended, which may be due to the lack of capping for the saRNA (there was no mention of the capping step in the protocol – which also must be rectified). Because the authors are looking at the replication of a functional saRNA, the sequence of that saRNA should be reported, and the construct should be sequenced to ensure it is theoretically functional.
The discussion presented is quite short, and some of the text should be moved up into the introduction section as mentioned above. The section rationalizing the NaCl screen would be much better if it were included in the introduction, as it would provide a good motivation for pursuing those experiments. There is little discussion of the in vivo results, and I suspect this is due to the less-than-stellar response of the transfected cells. If the authors choose to remove these studies, and instead do a more in-depth analysis of the RNA integrity and stability, there would be a good amount to discuss here. As is, the discussion is a bit limited in scope, and outside of the final optimized protocol and recipe, offers little in terms of a satisfying conclusion overall.
Data availability: We are missing the cytometry data and gating parameters used for population selection. These should be added to supplemental information.

Is the work clearly and accurately presented and does it cite the current literature?

Yes
Is the study design appropriate and is the work technically sound?

Partly
Are sufficient details of methods and analysis provided to allow replication by others?

Yes
If applicable, is the statistical analysis and its interpretation appropriate?

Yes
Are all the source data underlying the results available to ensure full reproducibility?

Partly
Are the conclusions drawn adequately supported by the results?

Yes

Competing Interests

EAV is named inventor on patent applications regarding thermostable self-amplifying RNA vaccine formulations.

Reviewer Expertise

RNA Vaccines

We confirm that we have read this submission and believe that we have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however we have significant reservations, as outlined above.

Respond to this report

Responses (0)

Back to all reports

Reviewer Report

85 Views

14 Apr 2022 | for Version 1

Rok Sekirnik, Chemistry Research Laboratory, Department of Chemistry, University of Oxford, Oxford, UK

85 Views Cite this report Responses(0)

Approved With Reservations

In their research article 'Design-of-experiments in vitro transcription yield optimization of self-amplifying RNA', Samnuan et al. describe a comprehensive study of factors that influence the productivity of IVT reaction for production of self-amplifying RNA (saRNA). In the article, the authors use the DoE approach to reduce the number of factors that could potentially affect the productivity of the reaction down to a handful, which are then evaluated in detail. The authors report that the major positive findings of the study are the key role of magnesium for productivity, and the relatively positive effect of acetate versus chloride. Furthermore, they conclude that pyrophosphatase does not positively impact the productivity of the reaction, and that increasing salt concentration during the reaction does not increase, but rather decreases productivity of IVT. Authors achieve a significant increase in saRNA yield, from starting 100 ng/uL to 4500 ng/uL.

Published studies on factors that might impact productivity of IVT reaction are relatively limited, and the manuscript is a welcome addition to the knowledge base. In their introduction, authors cite IVT protocols optimized for small RNAs (<100 nt), but do not refer to any recent protocols optimized for larger RNAs, e.g. mRNA, which, although not as large as saRNA, should still be considered highly relevant for their study, e.g. Henderson, J. M., et al., Curr Protoc., 2021. ¹ This is an important point from the perspective of factor selection. In the introductory paragraph #4, authors list the factors selected for DoE study, without theoretical justification. This paragraph should be moved to the discussion, where the authors currently provide the rationale for the selection of individual factors with references to literature. Because of this inverted structure, the introduction lacks a theoretical basis underpinning the study. The introduction would benefit from a more thorough review of current literature on key factors known to impact the activity of T7 polymerase (particular attention should be paid to the literature on the role of Mg²⁺ for activity of T7).

Selection of NaCl as a factor is surprising, and not justified by the authors. Authors initially exclude plasmid concentration as a factor, keeping it constant throughout the study at 1 mg/100 mL. A reference to published mRNA protocols would justify the selection of this value in context of mRNA production protocols. Whether the same mass of plasmid encoding for production of saRNA can be used without optimization is questionable, as the number of plasmid copies will thus be correspondingly lower compared to a standard mRNA IVT reaction, and reaction kinetics therefore correspondingly slower. Authors do not address this issue anywhere in the manuscript. Furthermore, the Methods section lacks information on the size of plasmid, so its molarity in the IVT protocol cannot be determined.

The study relies heavily on Qubit RNA Broad Range Assay. Authors do not comment on the limit of quantification (LoQ), selectivity or reproducibility of this assay. RNA yields in DoE study vary from <100 ng/uL to 4500 ng/uL. Authors should provide evidence that the assay response is linear in this range, e.g. qualify the assay with purified saRNA). Additionally, they should comment on whether the Qubit content method, is selective for target molecule (nominal molecular weight) only, or whether degradation products also contribute to 'yield' (see e.g. Figure 11B). Assay sensitivity and reproducibility is particularly relevant for evaluating one of two major conclusions of the paper, namely the impact of ion type on productivity. Early in the study (step 2 of DoE) authors perform a comparative study between MgCl₂ and MgOAc₂ and conclude that acetate leads to higher productivity; MgOAc₂ is thus fixed and used throughout the remainder of the study. Methodologically this is problematic as the absolute differences in productivity between chloride and acetate are minimal (100 ng/uL vs 200 ng/uL; ref Figure 4B); and the effect is not confirmed once the productivity of IVT is increased to >4000 ng/uL, where the significance would be demonstrated more clearly. Authors claim that they performed each IVT measurement in triplicate. Error bars should be added to each data figure. Significance of model equation (p. 6) is not clear. This should be moved to supplementary materials.

In Table 1 authors list the original stock concentrations for each chemical used in the study, with identification of the manufacturer. Catalogue numbers should be added for reagents used; if authors prepared the stock concentrations from solid powders, they should also report the hydration form used. This is particularly relevant for comparison of MgCl2 and MgOAc2; the authors should verify that correct molecular formula was used (e.g. MgCl2 hexahydrate) when calculating molarity of stock solutions.
Table 2, Table 3, Figure 2 and Figure 3 describe a single experiment, i.e. the full two-level factorial DoE. For clarity, Table 3 and Figure 3 should be moved to supplementary materials to help the flow of the manuscript. Table 2 and Figure 2 should stay in main text, as they are critical for interpretation of the experiment. A similar change should be made for the three-level factorial DoE analysis, i.e. only Table 4 and Figure 4 help the reader understand the significance of results and should be presented in main text; Table 5 and Figure 5 should be moved to supplementary. Legend to Figure 4 should state the timepoint at which yield was measured.

Optimal NTP concentration is identified as 10 mM of each NTP. Authors do not further optimize this parameter. Since the duration of IVT is not stated, it is difficult to interpret the data (would productivity be higher if experiment was performed longer?). Did authors try to adjust NTP concentrations based on saRNA sequence? The authors should also address the question of what is the limiting factor for productivity, i.e. why 10 mM leads to higher production of saRNA than 5 mM and why 20 mM is inhibitory. This is only partly addressed in the Discussion with the explanation that Mg-NTP complex leads to release of a proton with a concomitant decrease in pH which inhibits T7 polymerase. However, this explanation seems partial and does not address the impact of NTP/Mg on T7 polymerase kinetics.

Figure 6 presents a key experiment of the study. The data is fitted, but the fitting function is not described. Did the authors consider plotting the yield as a function of time to evaluate the effect of Mg²⁺ on reaction kinetics / saRNA stability? How do they interpret the decrease in saRNA concentration >75 mM MgOAc2 after 4h? Did they observe precipitation or degradation? This is particularly relevant in light of their conclusion that pyrophosphatase is not required to promote the reaction. Mg pyrophosphate is highly insoluble; in absence of pyrophosphatase this precipitation could either inhibit the reaction or affect content determination; authors should address this in the discussion.

With optimization of MgOAc2 concentration, authors successfully increased the productivity from ~1000 ng/uL to ~3000 ng/uL, which is very impressive and deserves a more explicit mention in the text. In Figure 6 and related discussion, authors state that 75 mM MgOAc2 and longer (i.e. 6 h) is optimal for increasing productivity. In the final optimization experiment, however, authors select a different experimental set-up which they claim to be optimal on basis of previous results (85 mM MgOAc2 and 4h; p. 11), which does not seem to be in line with results presented in Figure 6 and related discussion. It is understood that a shorter, i.e. 4 h reaction is preferred for practical reasons, and to avoid degradation, but this is not explicitly mentioned. Discussions for both figures should therefore be unified in this respect, while the experiment does not need to be repeated. They then address the effect of T7 polymerase, which is shown to be concentration-dependent, as demonstrated by decreasing the concentration from the nominal value of 100 U/100 uL. Given that this concentration is significantly below the concentrations recommended by manufacturers (authors reference to 'many commercial kits using 100 U' should be referenced for clarity. The concentrations used in published protocols for production of mRNA suggest significantly higher concentrations, e.g. 8 U/uL ¹, corresponding to 800 U in case of saRNA reactions as performed by authors. It would seem obvious to try to increase the concentration of polymerase to increase the yield. Authors should discuss this in the text. Instead, they focus on a poorly justified experiment to test the effect of addition of salt concentration over time. It is not clear why salt should be added at different timepoints, rather than at t=0. The results appear to be inconsistent with the positive effect of NaCl they reported in Table 3. Similarly, NaOAc was used in full three-level factorial reported in Table 4/Figure 4, but the result is not referenced in context of Figure 8. The authors do not comment on the variability of control reaction; the control experiment reported in Figure 8 reaches ~25% higher yield compared to same experiment reported in Figure 7 (time-point 2h). Is this to be understood as inter-experimental error of the study? This question deserves a wider treatment and is intimately linked to the comment about the selection of analytical method.

The DSD experiment leads to a further increase productivity to ~4500 ng/uL. This is a very impressive result, but does not get an explicit mention in the text. Table 7 is not a helpful for interpretation of the data, and should be moved to Supplementary. Figure 9 suggests that high productivity (~4500 ng/uL) can be achieved with all additives tested (pyrophosphatase, surfactant, spermidine, betaine, DMSO); since the design of the experiment is not described, additional interpretation of data is needed. The heat-map (Figure 10) is helpful in this respect, but a reader would benefit from an explicit interpretation (e.g. a table of optimized reaction conditions, current Supplementary Table 8, which should be moved to main text) of which minimal combinations of additives are required to reach productivity of 4500 ng/uL.

Finally, authors test the impact of pyrophosphatase 'using two best conditions for IVT'. It is not specified what those conditions are (see also point above). Figure 11 indicates that productivity of ~3000 ng/uL was reached, which is significantly lower than best conditions shown in Figure 10 (~4500 ng/uL), this discrepancy should be justified in the text. Denaturing gel (Figure 11B) shows a high degree of smearing, while authors comment on a 'slightly better quality' in absence of pyrophosphatase, but this interpretation does not appear to be possible based on PAGE alone. Based on PAGE profile as presented in Figure 11, authors should also comment on whether quality of saRNA is considered acceptable, and what are the degradation products observed by PAGE (degradation or abortive transcripts. Furthermore, a comment on how quantification of IVT compares pre- and post-purification (e.g. material from Figure 11B vs material from Figure 12B is needed. Authors should comment on any observations of precipitation in presence/absence of pyrophosphatase.

Discussion

Authors should add an explicit evaluation of the absolute increase in productivity they have achieved. The discussion on the optimal NTP:Mg ratio is thin; there is a critical typo citing the ratio as '1:1.1875' which should be corrected to '1:1.875' (p. 15). High NTP concentrations are interpreted as inhibitory due to lowering of pH; this should be demonstrated experimentally. There is a highly problematic statement concerning magnesium and pyrophophate. Authors claim that magnesium precipitates pyrophosphate, therefore justifying the omission of pyrophosphatase. This implies that optimized IVT reaction for production of saRNA is expected to lead to precipitation. Authors should comment on this implication and discuss whether precipitation is selective (i.e. that saRNA does not co-precipitate), that precipitation does not impact yield determination and that precipitation is not problematic for down-stream purification. In the conclusion, it is stated that the process as optimized can be used to produce large quantities of saRNA with potential for GMP production, so this seems a critical point for commercial utilization of the process.

Overall, the manuscript presents a comprehensive study of factors contributing to productivity of IVT for production of saRNA, a novel RNA-based modality with high therapeutic relevance. The conclusions of the paper are therefore important for driving process improvements in saRNA programmes in particular for reducing the production costs due to higher productivity of the IVT process step. Some edits are required to improve readability, clarify missing details to enable reproducibility, and provide further justification where indicated in the review above. Once the concerns raised above have been addressed, I recommend accepting this manuscript for indexing.

Is the work clearly and accurately presented and does it cite the current literature?

Partly
Is the study design appropriate and is the work technically sound?

Yes
Are sufficient details of methods and analysis provided to allow replication by others?

Yes
If applicable, is the statistical analysis and its interpretation appropriate?

I cannot comment. A qualified statistician is required.
Are all the source data underlying the results available to ensure full reproducibility?

Yes
Are the conclusions drawn adequately supported by the results?

Partly

References

1. Henderson JM, Ujita A, Hill E, Yousif-Rosales S, et al.: Cap 1 Messenger RNA Synthesis with Co-transcriptional CleanCap® Analog by In Vitro Transcription.Curr Protoc. 2021; 1 (2): e39 PubMed Abstract | Publisher Full Text

Competing Interests

No competing interests were disclosed.

Reviewer Expertise

in vitro transcription, nucleic acid chromatography, nucleic acid analytics

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.

Respond to this report

Responses (0)

[1] 1. Iavarone C, O'Hagan DT, Yu D, et al.: Mechanism of action of mRNA-based vaccines. Expert Rev. Vaccines 2017; 16(9): 871–881. Publisher Full Text

[2] 2. Zhang C, Maruggi G, Shan H, et al.: Advances in mRNA Vaccines for Infectious Diseases. Front. Immunol. 2019; 10(594). Publisher Full Text

[3] 3. Vogel AB, Lambert L, Kinnear E, et al.: Self-Amplifying RNA Vaccines Give Equivalent Protection against Influenza to mRNA Vaccines but at Much Lower Doses. Molecular Therapy: The Journal of the American Society of Gene Therapy 2018; 26(2): 446–455. PubMed Abstract | Publisher Full Text

[4] 4. Geall AJ, Verma A, Otten GR, et al.: Nonviral delivery of self-amplifying RNA vaccines. Proc. Natl. Acad. Sci. U. S. A. 2012; 109(36): 14604–14609. PubMed Abstract | Publisher Full Text

[5] 5. Johansson DX, Ljungberg K, Kakoulidou M, et al.: Intradermal electroporation of naked replicon RNA elicits strong immune responses. PLoS One 2012; 7(1): e29732. PubMed Abstract | Publisher Full Text

[6] 6. Bogers WM, Oostermeijer H, Mooij P, et al.: Potent immune responses in rhesus macaques induced by nonviral delivery of a self-amplifying RNA vaccine expressing HIV type 1 envelope with a cationic nanoemulsion. J. Infect. Dis. 2015; 211(6): 947–955. PubMed Abstract | Publisher Full Text

[7] 7. Moyo N, Vogel AB, Buus S, et al.: Efficient Induction of T Cells against Conserved HIV-1 Regions by Mosaic Vaccines Delivered as Self-Amplifying mRNA. Molecular Therapy. Methods & Clinical Development 2019; 12: 32–46. PubMed Abstract | Publisher Full Text

[8] 8. Brazzoli M, Magini D, Bonci A, et al.: Induction of Broad-Based Immunity and Protective Efficacy by Self-amplifying mRNA Vaccines Encoding Influenza Virus Hemagglutinin. J. Virol. 2016; 90(1): 332–344. PubMed Abstract | Publisher Full Text

[9] 9. Fleeton MN, Chen M, Berglund P, et al.: Self-replicative RNA vaccines elicit protection against influenza A virus, respiratory syncytial virus, and a tickborne encephalitis virus. J. Infect. Dis. 2001; 183(9): 1395–1398. PubMed Abstract | Publisher Full Text

[10] 10. Khalil SM, Tonkin DR, Mattocks MD, et al.: A tetravalent alphavirus-vector based dengue vaccine provides effective immunity in an early life mouse model. Vaccine. 2014; 32(32): 4068–4074. PubMed Abstract | Publisher Full Text

[11] 11. White LJ, Sariol CA, Mattocks MD, et al.: An alphavirus vector-based tetravalent dengue vaccine induces a rapid and protective immune response in macaques that differs qualitatively from immunity induced by live virus infection. J. Virol. 2013; 87(6): 3409–3424. PubMed Abstract | Publisher Full Text

[12] 12. Bates JT, Pickens JA, Schuster JE, et al.: Immunogenicity and efficacy of alphavirus-derived replicon vaccines for respiratory syncytial virus and human metapneumovirus in nonhuman primates. Vaccine. 2016; 34(7): 950–956. PubMed Abstract | Publisher Full Text

[13] 13. Pyankov OV, Bodnev SA, Pyankova OG, et al.: A Kunjin Replicon Virus-like Particle Vaccine Provides Protection Against Ebola Virus Infection in Nonhuman Primates. J. Infect. Dis. 2015; 212(Suppl 2): S368–S371. PubMed Abstract | Publisher Full Text

[14] 14. Reynard O, Mokhonov V, Mokhonova E, et al.: Kunjin virus replicon-based vaccines expressing Ebola virus glycoprotein GP protect the guinea pig against lethal Ebola virus infection. J. Infect. Dis. 2011; 204 Suppl 3(Suppl 3): S1060–S1065.

[15] 15. Herweijer H, Wolff JA: Self-amplifying vectors for gene delivery. Adv. Drug Deliv. Rev. 1997; 27(1): 5–16. PubMed Abstract | Publisher Full Text

[16] 16. Brito LA, Kommareddy S, Maione D, et al.: Self-amplifying mRNA vaccines. Adv. Genet. 2015; 89: 179–233. Publisher Full Text

[17] 17. Geall AJ, Mandl CW, Ulmer JB: RNA: the new revolution in nucleic acid vaccines. Semin. Immunol. 2013; 25(2): 152–159. PubMed Abstract | Publisher Full Text

[18] 18. Milligan JF, Uhlenbeck OC: Synthesis of small RNAs using T7 RNA polymerase. Methods Enzymol. 1989; 180: 51–62. Publisher Full Text

[19] 19. Daniel V, Sarid S, Beckmann JS, et al.: In vitro transcription of a transfer RNA gene. Proc. Natl. Acad. Sci. U. S. A. 1970; 66(4): 1260–1266. PubMed Abstract | Publisher Full Text | Free Full Text

[20] 20. Politis N, Colombo P, Colombo G, et al.: Design of experiments (DoE) in pharmaceutical development. Drug Dev. Ind. Pharm. 2017; 43(6): 889–901. Publisher Full Text

[21] 21. Inc., S. I: JMP® 14 Fitting Linear Models Cary, NC: SAS Institute Inc.; 2018.

[22] 22. Samnuan K, Blakney A, McKay P, et al.: Design-of-Experiments in vitro Transcription Yield Optimization of Self-Amplifying RNA [Data set]. Zenodo 2022.

[23] 23. Schafer R, Zillig W: The effects of ionic strength on termination of transcription of DNAs from bacteriophages T4, T5 and T7 by DNA-dependent RNA polymerase from Escherichia coli and the nature of termination by factor rho. Eur. J. Biochem. 1973; 33(2): 215–226. PubMed Abstract | Publisher Full Text

[24] 24. Cunningham PR, Ofengand J: Use of inorganic pyrophosphatase to improve the yield of in vitro transcription reactions catalyzed by T7 RNA polymerase. BioTechniques. 1990; 9(6): 713–714. PubMed Abstract

[25] 25. Slater RJ: Vitro Transcription. Techniques in Molecular Biology: Volume 2. Walker JM, Gaastra W, editors. Boston, MA: Springer US; 1987; pp 203–227.

[26] 26. Chamberlin M, Ring J: Characterization of T7-specific ribonucleic acid polymerase. 1. General properties of the enzymatic reaction and the template specificity of the enzyme. J. Biol. Chem. 1973; 248(6): 2235–2244. PubMed Abstract | Publisher Full Text

[27] 27. Milligan JF, Groebe DR, Witherell GW, et al.: Oligoribonucleotide synthesis using T7 RNA polymerase and synthetic DNA templates. Nucleic Acids Res. 1987; 15(21): 8783–8798. PubMed Abstract | Publisher Full Text | Free Full Text

[28] 28. Young JS, Ramirez WF, Davis RH: Modeling and optimization of a batch process for in vitro RNA production. Biotechnol. Bioeng. 1997; 56(2): 210–220. PubMed Abstract | Publisher Full Text

[29] 29. Maslak M, Martin CT: Effects of solution conditions on the steady-state kinetics of initiation of transcription by T7 RNA polymerase. Biochemistry. 1994; 33(22): 6918–6924. PubMed Abstract | Publisher Full Text

[30] 30. Kern JA, Davis RH: Application of solution equilibrium analysis to in vitro RNA transcription. Biotechnol. Prog. 1997; 13(6): 747–756. PubMed Abstract | Publisher Full Text

[31] 31. Nilsen TW, Rio DC, Ares M Jr: High-yield synthesis of RNA using T7 RNA polymerase and plasmid DNA or oligonucleotide templates. Cold Spring Harb. Protoc. 2013; 2013(11): pdb.prot078535. Publisher Full Text

[32] 32. Fuchse, Millette RL, Zillig W, et al.: Influence of salts on RNA synthesis by DNA-dependent RNA-polymerase from Escherichia coli. Eur. J. Biochem. 1967; 3(2): 183–193. PubMed Abstract | Publisher Full Text

[33] 33. Oakley JL, Strothkamp RE, Sarris AH, et al.: T7 RNA polymerase: promoter structure and polymerase binding. Biochemistry. 1979; 18(3): 528–537. Publisher Full Text

Design-of-experiments in vitro transcription yield optimization of self-amplifying RNA

Abstract

Keywords

Abbreviations

Introduction

Figure 1. A visual flow diagram of all the DoEs performed for this study in order to achieve our optimal condition for IVT.

Methods

Plasmid DNA synthesis and purification

RNA synthesis and quantification

Table 1. The list of components used for the different DoEs.

Design of experiment and statistical analysis

RNA purification and RNA gel

Cells and in vitro transfections

Flow cytometry

Data analysis

Results

Magnesium is the most significant component for influencing IVT yield of saRNA

Table 2. The components used for a full two-level factorial DoE.

Table 3. The significance of each component varied and their interactions.

Figure 2. The most significant components varied based on a full two-level factorial DoE analysis.

Figure 3. The coefficient estimates of each component and their secondary interactions based on a full two-level factorial DoE analysis.

Acetate ions are more effective than chloride ions in enhancing RNA yield

Table 4. The components used for a full three-level factorial DoE.

Table 5. The significance of each component varied.

Figure 4. A full three-level factorial DoE analysis of saRNA synthesis through IVT.

Figure 5. The coefficient estimates of each component and their secondary interactions based on a full three-level factorial DoE analysis.

75 mM MgOAc2 is the optimal concentration for all incubation periods

Figure 6. The impact of MgOAc2 on saRNA synthesis through IVT at different timepoints.

RNA yield is proportional to T7 RNA polymerase concentration at all timepoints

Figure 7. The effects of T7 RNA polymerase concentration on RNA yield at three different timepoints.

Increasing the salt concentration during IVT process at different timepoints does not affect final RNA concentration

Figure 8. The effects of adding various NaOAc concentrations at different timepoints after the initiation of IVT on RNA yield.

An optimal combination of pyrophosphatase, spermidine, DMSO, betaine and tween 20 is needed in order to produce high RNA yields

Table 6. The components used for a definitive screening DoE.

Table 7. The components that had major effects on IVT in a definitive screening DoE.

Figure 9. A definitive screening DoE analysis of saRNA production through IVT to observe the importance of pyrophosphatase, spermidine, surfactant (either Triton X-100 or Tween 20), betaine and DMSO on RNA yield.

Figure 10. A color map on correlations between each component based on a definitive screening DoE analysis.

Inclusion of pyrophosphatase does not enhance the quantity of RNA yield

Figure 11. Comparison of the current best conditions with and without pyrophosphatase.

The optimal conditions of IVT can be used to synthesize saRNAs encoding for different genes of interest with good protein expression in vitro

Table 8. The final optmized components for IVT.

Figure 12. Comparison of the current best condition without pyrophosphatase with different genes of interest.

Discussion

Conclusions

Data availability

Underlying data

Accession numbers

References

Comments on this article Comments (0)

Open Peer Review

Comments on this article Comments (0)

Open Peer Review

Reviewer Status

Reviewer Reports

Comments on this article

Browse by related subjects

Competing Interests Policy

Stay Updated

75 mM MgOAc₂ is the optimal concentration for all incubation periods

Figure 6. The impact of MgOAc₂ on saRNA synthesis through IVT at different timepoints.