Recombinant T7 RNA polymerase production using ClearColi BL21(DE3) and animal-free media for in vitro transcription

Abstract

Abstract

In vitro transcription (IVT) using T7 RNA polymerase (RNAP) is integral to RNA research, yet producing this enzyme in E. coli presents challenges regarding endotoxins and animal-sourced toxins. This study demonstrates the viable production and characterization of T7 RNAP using ClearColi BL21(DE3) (an endotoxin-free E. coli strain) and animal-free media. Compared to BL21(DE3) with animal-free medium, soluble T7 RNAP expression is ~50% lower in ClearColi BL21(DE3). Optimal soluble T7 RNAP expression in flask fermentation is achieved through the design of experiments (DoE). Specification and functional testing showed that the endotoxin-free T7 RNAP has comparable activity to conventional T7 RNAP. After Ni-NTA purification, endotoxin levels were approximately 109-fold lower than T7 RNAP from BL21(DE3) with animal-free medium. Furthermore, a full factorial DoE created an optimal IVT system that maximized mRNA yield from the endotoxin-free and animal-free T7 RNAP. This work addresses critical challenges in recombinant T7 RNAP production through innovative host and medium combinations, avoided endotoxin risks and animal-derived toxins. Together with an optimized IVT reaction system, this study represents a significant advance for safe and reliable reagent manufacturing and RNA therapeutics.

Key points

• Optimized IVT system maximizes mRNA yields, enabling the synthesis of long RNAs.

• Novel production method yields endotoxin-free and animal-free T7 RNAP.

• The T7 RNAP has equivalent specifications and function to conventional T7 RNAP.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00253-023-12939-w.

Introduction

The advancement of messenger RNA (mRNA) drugs has revolutionized the field of therapeutics, offering a novel approach to treating a myriad of diseases. The mechanism of mRNA drugs can either compensate for a missing or faulty protein, as seen in genetic diseases, or stimulate an immune response, as harnessed in vaccines (Sahin et al. 2014). The advantages of mRNA drugs over traditional therapeutics are manifold, including their rapid design and production capabilities (Rohner et al. 2022). These are particularly beneficial in responding to emerging health threats like new viral strains (Pardi et al. 2018). The potential of mRNA drugs was vividly demonstrated in the fight against COVID-19, where mRNA vaccines were developed quickly and proved highly effective. This success has sparked further interest in mRNA technology, with ongoing research exploring its use in treating genetic disorders, cancers, and other diseases (Puy et al. 2019).

The promising therapeutic potential of mRNA drugs highlights the critical need to develop methods for reliable, safe, cost-effective mRNA production, typically achieved through IVT. In vitro transcription driven by T7 RNAP allows template-directed synthesis of RNA molecules ranging from short oligonucleotides up to thousands of bases in length, achieving milligram-scale RNA yields (Beckert and Masquida 2011). T7 RNAP is a highly promoter-specific and DNA-dependent RNA polymerase from the T7 bacteriophage (Tabor and Richardson 1985). Conventional T7 RNAP production uses E. coli B strain BL21(DE3) as a host. But the outer membrane of BL21(DE3) contains lipopolysaccharides (LPS) (Galanos and Freudenberg 1993), known as endotoxin, which can trigger uncontrolled acute inflammatory responses, inducing pathological processes that can ultimately lead to multiple organ failure, shock, and death (Śmiechowicz 2022). The U.S. Food and Drug Administration (FDA) emphasizes endotoxin-free production processes due to the potential dangers associated with endotoxin (Affairs 2018). In the case of the T7 RNAP application, endotoxin-free is a prerequisite for its usage in mRNA drug production. T7 RNAP also traditionally produced using animal-derived media, which contains tryptone, an animal-origin (AO) pancreatic digest of casein. Using AO ingredients in the manufacturing process of therapeutic proteins can introduce a potential risk of transmitting animal-derived contaminants, such as bovine spongiform encephalopathy (BSE) and porcine circovirus (PCV) (Victoria et al. 2010). The potential for them to be found in animal products (e.g., the serum or peptones used in culture media) is one reason for frequently prohibiting their use in bioproduction (Klug et al. 2016). Moreover, the FDA maintains stringent regulations regarding animal-derived components in pharmaceuticals destined for human use (Research 2020). Therefore, the presence of these components during protein production constitutes a significant regulatory hurdle.

The current method of T7 RNAP production poses the risk of introducing endotoxin and potential animal toxins into mRNA drugs. Additional processing steps to remove these toxins are essential before the T7 RNAP can be used for mRNA production. Those removal procedures increase the cost, decrease the T7 RNAP yield, and necessitate verification assays, adding another layer of complexity and cost. Furthermore, traditional T7 RNAP-based IVT kits primarily focus on synthesizing small RNA, falling short of current demands for long mRNA sequences required in mRNA drugs, particularly vaccines like the Pfizer-BioNTech and Moderna COVID-19 vaccines.

Given these challenges, this paper practiced a new approach for T7 RNAP production that is both animal-free and endotoxin-free and capable of producing long RNA sequences in large quantities. Animal-free medium were combined with ClearColi BL21(DE3) (Mamat et al. 2013), an endotoxin-free strain, to achieve endotoxin-free and animal-free T7 RNAP production. Our design of experiment data showed that the primary factors affecting endotoxin-free bacterial growth in animal-free medium are similar to conventional media and bacteria combination. We also used a four-factor full factorial design to find an optimal IVT system. This approach led us to an IVT recipe that produces higher RNA yields (about a threefold increase compared to the conventional IVT system) and can synthesize long RNA sequences (8.7 Kb), which is required for self-amplification RNA synthesis. This innovative approach aligns with FDA guidelines and paves the way for making more effective, safer, and longer RNA, thereby contributing to the broader effort to realize the potential of mRNA therapeutics fully.

Material and methods

Bacteria and growth media

E coli host strains DH5α and BL21(DE3) were procured from NEB (Ipswich, MA, USA), while ClearColi BL21(DE3) was sourced from Lucigen (Middleton, WI, USA). Unless otherwise stated, E. coli cultures were grown in Luria-Bertani (LB) broth or LB agar from Sangon Biotech (Shanghai, China). Bacteria containing the pLQY199 plasmid were cultured and supplemented with 100 μg/ml of ampicillin. The study also utilized the following animal-free media, with each volume adjusted to final pH 7.0 before autoclaving at 121 °C for 20 min: HEP Medium (1 L), 10 g of Hy-Express™ Peptone I (Kerry Group, Ireland), 5 g of yeast extract (Sangon Biotech), and 10 g of sodium chloride (Sangon Biotech). HES Medium (1 L), 15 g of Hy-Express™ System II (Kerry Group) and 10 g of sodium chloride. HS Medium (1 L), 10 g of Hy-Soy™ (Kerry Group), 5 g of yeast extract, and 10 g of sodium chloride. BL4 Medium (1 L), 10 g of N-Z Soy™ BL 4 (Kerry Group), 5 g of yeast extract, and 10 g of sodium chloride.

Plasmid and IVT templates

Gibson assembly was used to construct the pLQY175 and pLQY199 plasmids, featuring an N-His-T7 RNAP gene under the control of the lacUV5 promoter and regulated by lac operator (Gibson et al. 2009). The backbones were linearized from the pACYC-lacUV5 (Supplementary sequence 1) and pETDuet-lacUV5 (Supplementary sequence 2) plasmids (using primers LQY01 and LQY02) for pLQY175 and pLQY199, respectively (Wladyka et al. 2005). The T7 RNAP fragment (NC_001604.1:3139-5887, 2692 bp) was amplified from E. coli BL21(DE3) genomic DNA using primers LQY03 and LQY04. Similarly, the pLQY200 plasmid, which functions as an in vitro transcription template, was created using Gibson assembly. The backbone of pLQY200 was derived from pRSFDuet-1, with primers LQY01 and LQY05, while the T7-GFP insert fragment was amplified from pOSIP-KO-sulA-GFP using primers LQY06 and LQY07 (St-Pierre et al. 2013). Both pLQY199 and pLQY200 were confirmed to have correct gene sequences through Sanger sequencing. For IVT, templates were prepared by PCR. The 1-Kb GFP IVT template was derived from pLQY200 using primers LQY08 and LQY09. The 8.7-Kb saRNA GFP IVT template was sourced from pT7-VEE-GFP (Addgene plasmid: 58970) using primers LQY10 and LQY11. The sequence information of used primers is listed in Table S1.

Bacterial growth curve

E. coli growth curve experiment was performed according to Stevenson et al. (2016). Briefly, a toothpick was dipped into the glycerol stock containing the pLQY199 plasmid in either BL21(DE3) or ClearColi BL21(DE3). It was then introduced into a 15-ml Falcon centrifuge tube containing 3 ml of media, incubated overnight at 37 °C. The following day, the overnight cultures were diluted to OD600 2.0 by adding the corresponding medium to balance the cell numbers. Next, 198 μl of the medium was dispensed into each well of a 96-well plate. Each well received 2 μl of the balanced overnight cultures. The 96-well plate was placed in a Tecan Infinite^® 200 PRO Microplate Reader for incubation and absorbance measurement. The program was set to record absorbance at 600 nm at an interval of 5 min over 22 h at 37 °C. Once the OD600 of most wells approximated 0.2, Isopropyl-β-D-thiogalactopyranoside (IPTG) was added to the induced groups to achieve IPTG 0.5 mM. All groups were performed in triplicate.

T7 RNAP expression

All small-scale experiments were conducted using a 500-ml Erlenmeyer flask filled with 50 ml of media. Overnight cultures (3 ml in a 15-ml Falcon centrifuge tube) in media were used to inoculate the expression cultures at an inoculation ratio of 1:100. The expression cultures were incubated at 37 °C with agitation at 180 rpm. When the cultures reached OD600 ~0.5–0.6, expression of T7 RNAP was induced by the addition of IPTG to a final concentration of either 0.1 mM or 0.5 mM. Post-induction, the cultures were incubated at the temperature as required, with continuous agitation. For larger-scale flask fermentation, 1 L of media was cultured in a 5-L Erlenmeyer flask. The induced cells were harvested 6-h or 18-h post-induction.

T7 RNA polymerase purification

One liter of induced cells was pelleted by centrifugation at 4000 g for 10 min at 4 °C. Then, the pellets were resuspended in 16 ml lysis buffer (50 mM Tris-HCl, pH 8.0, 5 mM β-mercaptoethanol, 0.5 M NaCl, and 1 mM imidazole). Following complete resuspension, lysozyme (Sangon Biotech) and a 100× protease inhibitor cocktail (Beyotime, Shanghai, China) were added to the lysate and incubated at 4 °C for 30 min with gentle agitation. Subsequently, 0.0025% (v/v) Benzonase (Vazyme, Nanjing, China) was added, followed by incubation at 20 °C for 20 min with continuous agitation. The lysate was then sonicated (2 s ON, 2 s OFF) at 100 W for 5 min, followed by centrifugation at 22,000 g for 30 min at 4 °C. The supernatant was filtered through a 0.2-μm membrane filter.

For purification, a 5 mL chromatography column was packed with 5 mL of Ni-NTA agarose resins (Ni Sepharose Excel, Cytiva, MA, USA) and equilibrated with 8 column volumes (CVs) of lysis buffer. The supernatant was then loaded onto the column. Then, the column was washed with 8 CVs of lysis buffer and 8 CVs of washing buffer (lysis buffer containing 10 mM imidazole). The protein was eluted with a buffer containing 500 mM imidazole. The eluted protein was dialyzed in 2× dialysis buffer (100 mM Tris-HCl, 200 mM NaCl, 40 mM β-mercaptoethanol, 2 mM Ethylenediaminetetraacetic acid (EDTA), 0.2% (w/v) Triton® X-100). Following dialysis, the protein was mixed with an equal volume of 100% glycerol and stored at −80 °C. For the endotoxin Limulus amebocyte lysate (LAL) assay, the eluted protein was dialyzed in endotoxin buffer (Table S2). All purification procedures were carried out using an AKTA Explorer 100 system from GE Healthcare (IL, USA).

Bradford assay

Protein sample concentrations were determined using the Detergent Compatible Bradford Protein Assay Kit (Vazyme), following the manufacturer’s guidelines. Briefly, set up a standard curve using bovine serum albumin (BSA) at known concentrations, ranging from 0.025 to 0.125 mg/ml. Protein samples were mixed with the Bradford reagent, and then 200 μl of each mixture was transferred into wells in a transparent 96-well plate. Absorbance was measured at 595 nm using a Tecan Infinite® 200 PRO Microplate Reader. The protein concentration of the unknown samples was deduced by comparing their absorbance values to those of the BSA standard curve.

Sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and coomassie staining

When sampling cells for protein extraction, the OD600 value was multiplied by the volume in milliliters (ml) for each sample and adjusted to 13.75 to normalize the number of cells across all samples. For total protein sample preparation, samples were pelleted, then resuspended with 500 μl of lysis mixture containing 2× Laemmli Sample Buffer (Bio-Rad, CA, USA), β-mercaptoethanol, and water at a 10:1:9 ratio. For soluble protein sample preparation, the pellet was resuspended in 500 μl of lysis buffer containing 1 × proteinase inhibitor, 1 mg/ml lysozyme, and 0.0025% Benzonase. The lysate was then sonicated (2 s ON, 2 s OFF) at 30 W for 1 min, followed by centrifugation at 22,000 g for 30 min at 4 °C. The supernatant was mixed with 2× Laemmli Sample Buffer and β-mercaptoethanol at a 9:10:1 ratio. Samples were incubated for 5 min at 95 °C before loading into 12% (w/v) Tris-Glycine SDS precast polyacrylamide gels (Vazyme). The gel electrophoresis was run in a mini gel electrophoresis system (Bio-Rad) with a constant voltage set at 80 V for the stacking gel and 120 V for the separating gel. Blue prestained protein standards (Vazyme) with a 10 to 180 kDa range were utilized to estimate protein size. Following electrophoresis, a standard Coomassie stain procedure was used to stain gel, which was imaged using the Gel Doc system (Bio-Rad). The soluble T7 RNAP band, with a molecular weight of 99.9 kDa, was analyzed on the SDS-PAGE gel image by using ImageJ software. The density of this band was then normalized using the 100 kDa band from a protein marker of known concentration. The loading volume of the Blue prestained protein standards is 5 μl (equivalent to 1 μg total protein mass) for each gel (Kasli et al. 2019).

RNase detection

The RNase activity was quantified using the Fluorometric RNase Assay Kit (Beyotime) per the manufacturer’s instructions. This kit used fluorescence resonance energy transfer (FRET) and a modified ssRNA with FAM and TAMRA in two ends to measure RNase activity. As RNase cleaves the ssRNA, fluorescein is released, and the resulting FAM fluorescence can be detected. An RNase A standard was briefly serially diluted to concentrations ranging from 0 to 0.5 pg/μl to establish a standard curve. Following sample preparation, 100 μl of each solution was transferred into individual wells of a black 96-well plate. Fluorescence was subsequently measured using Tecan Infinite^® 200 PRO without the lid covered. An excitation wavelength of 490 nm and an emission wavelength of 520 nm was used to measure the FAM fluorescence.

Benzonase detection

Benzonase activity was quantified using the Fluorometric Benzonase Assay Kit (Beyotime). The mechanism of this assay is the same as the Fluorometric RNase Assay Kit but uses a modified DNA sequence with VIC and BHQ1 at two ends. Initially, a standard curve was established using serial dilutions of Benzonase standard, with concentrations ranging from 0 to 50 mU/μl. The samples were prepared according to the manufacturer’s protocol. One hundred microliters of each sample was transferred into a black 96-well plate. The subsequent fluorescence measurements were carried out with Tecan Infinite^® 200 PRO set at an excitation wavelength of 514 nm and an emission wavelength of 552 nm.

Endotoxin detection

Endotoxin levels were determined using the ToxinSensor™ Endotoxin Detection Kit (GenScript, USA), an LAL assay method utilizing a quantitative colorimetric method to detect endotoxin concentrations. Samples were prepared according to the manual from the manufacturer. Set up a standard from a series of endotoxin standard dilutions, ranging from 0.01 to 0.1 EU/ml. Two hundredmicroliters of samples were transferred to a transparent 96-well plate and then measured at a wavelength of 545 nm using Tecan Infinite^® 200 PRO.

In vitro transcription

In this study, we designated the T7 RNAP derived from BL21(DE3) with BL4 media as conventional T7 RNAP (C-T7 RNAP) and designated the T7 RNAP derived from ClearColi BL21(DE3) with BL4 media as endotoxin-free and animal-free T7 RNAP (EFAF-T7 RNAP). The functionality of the C-T7 RNAP and EFAF-T7 RNAP in IVT reactions (Table S3) was assessed using the components listed in Table S4. Table S5 shows the components used in the IVT DoE; the corresponding DoE table is Table S6. Table S7 shows the final optimized IVT recipe. All IVT reactions were incubated at 37 °C for 2 h. Afterward, DNase I and DNase I reaction buffer were added, followed by incubation at 37 °C for 15 min to digest the DNA template. Subsequently, EDTA was added to inactivate the T7 RNAP. RNA yield was quantified immediately after inactivation using the Qubit RNA Broad Range Assay kit with the Qubit 4 Fluorometer (Thermo Fisher Scientific, Waltham, MA, USA).

RNA agarose gel electrophoresis

RNA quality was assessed with gel electrophoresis. RNA samples and the ssRNA Ladder (NEB) were combined with 2× RNA loading dye and incubated at 70 °C for 10 min to denature the RNA. The denatured RNA was then subjected to electrophoresis on a 1.2% agarose gel using 1× TBE Running Buffer (90 mM Tris base, 90 mM Boric acid, and 2 mM EDTA, pH 8.3) at 80 V for 60 min. The gel was imaged using the GelDoc system (Bio-Rad).

Design of experiments and data treatment

All design of experiment analyses were conducted using JMP software, version 16.1, primarily employing full factorial design. The response of T7 RNAP DoE is soluble T7 RNAP expression level; the response of IVT DoE is the RNA yield from IVT reactions. Data from each DoE were analyzed by fitting a standard least squares model for effect screening and effect leverage, allowing for the elimination of non-significant factors to enhance the models and refine optimal conditions for T7 RNAP expression and maximal RNA yield. The resultant graphical representations were exported from JMP 16.1 software.

Results

ClearColi BL21(DE3) with animal-free media grows slower and has less soluble T7 RNAP expression

The endotoxin-free strain ClearColi BL21(DE3) has seven genetic deletions (ΔgutQ, ΔkdsD, ΔlpxL, ΔlpxM, ΔpagP, ΔlpxP, ΔeptA) that modify lipopolysaccharide (LPS) to lipid IV_A and a compensating mutation (msbA148) to keep the strain viable (Mamat et al. 2013). The pLQY199 medium copy plasmid was used for T7 RNAP expression because of its twofold higher protein expression compared to that from the pLQY175 low copy plasmid (Fig. S1). Compared to the growth of BL21(DE3) and ClearColi BL21(DE3) (Fig. S2), the latter strain grows slower than BL21(DE3) by approximately 50% in the LB medium. We checked the soluble expression of T7 RNAP in the ClearColi BL21(DE3) strain (Fig. S3 Lanes 1 and 2). The target protein expression level is about 71.4% of that in BL21(DE3).

This study sought to address the media issue by using animal-free media. Previous study shows that animal-free hydrolysates can support E. coli growth and viability (Vaze et al. 2015). Four animal-free hydrolysates (Hy-Express™ Peptone I, Hy-Express™ System II, Hy-Soy™, N-Z Soy™ BL4) were used to prepare four animal-free media (recipes are detailed in Material and methods), which were tested subsequently (Fang et al. 2009). In uninduced conditions, the growth of the BL21(DE3) strain in the four animal-free media was similar (Fig. 1a), except cells in the HS medium grew slower in the stationary phase. When induced, the bacteria growth in LB and BL4 are similar. But, the growth was inhibited for the other three animal-free recipes (HEP, HES, and HS) (~28% on average). However, the SDS-PAGE results of protein samples from flask fermentation (Fig. 1c, Fig. S3) showed that the soluble expression of T7 RNAP in these four different media is almost the same (Fig. S3).

Fig. 1.

BL21(DE3) and ClearColi BL21(DE3) growth curve and protein expression with different media. a Growth curves of BL21(DE3) with different media. Cells with pLQY199 plasmid were induced by IPTG (0.5 mM) when the OD600 of most wells reached ~0.2 in microplates. Cells were incubated for 22 h. b Growth curves (OD600) of ClearColi BL21(DE3) with different media. c OD600 value of flask fermentation with BL21(DE3) and ClearColi BL21(DE3) in different media. The cells grow at 37 °C, then move to 30 °C for 6 h or 18 h after being induced with 0.5 mM IPTG (OD600 0.5–0.6). d The soluble T7 RNAP expression in BL21(DE3) and ClearColi BL21(DE3) with different media. Expressed proteins were sampled 6 h after being induced with 0.5 mM IPTG at 30 °C. Bar values at c and d are average ± standard deviations for all experiments shown for n = 3 repeats

Based on the above tests, this study further tested the combination of ClearColi BL21(DE3) and animal-free media for soluble T7 RNAP expression, which will enable the production of EFAF-T7 RNAP. In 96-well microplates, ClearColi BL21(DE3) growth in HES and HS animal-free media is about 37% slower than in LB during the log phase (Fig. 1b). The bacteria growth in BL4 is the same as HES and HS but was strongly inhibited by about 45% in the stationary phase. The HEP and ClearColi BL21(DE3) combination was the worst. It grows extremely slowly and has no noticeable log phase growth (Fig. 1b). After induction, bacteria with all the animal-free media were strongly inhibited, about 32% of that with LB medium.

In the flask fermentation experiment (initially, cells grow at 37 °C, then induced with 0.5 mM IPTG at OD600 0.5–0.6, after induction, cells grow at 30 °C), the growth difference at different time points (6 h versus 18 h) is even more significant between the BL21(DE3) and ClearColi BL21(DE3) batches (Fig. 1c). For BL21(DE3) strain, the OD600 at 18-h time point is 2–3 times higher than that at 6-h time point (Fig. 1c). And it does not matter if it is LB broth or animal-free media. However, in the ClearColi BL21(DE3) strain, the OD600 value after 6 h of induction is only half that in the BL21(DE3) strain. Moreover, after 18 h of bacterial growth, the OD600 value goes even lower, about 60–70% of that at 6 h in ClearColi BL21(DE3) strain (Fig. 1c). We did not observe growth differences in different growth media (LB broth or animal-free media) in ClearColi BL21(DE3). Checked the soluble expression of T7 RNAP in those combined growth conditions, the expression level in ClearColi BL21(DE3) with animal-free media is lower than that in BL21(DE3), and it is also lower than that in ClearColi BL21(DE3) with LB broth (Fig. 1d). The soluble expression of T7 RNAP in ClearColi BL21(DE3) with HEP medium, HES medium, HS medium, and BL4 medium is 52.2%, 59.9%, 52.8%, and 52.6%, respectively, compared to that in BL21(DE3) strain with corresponding media. Information from Fig. S4 shows that time and temperature affect the expression of T7 RNAP in BL21(DE3) with LB broth. It is worth investigating factors that affect T7 RNAP expression in ClearColi BL21(DE3) with animal-free media.

Time and temperature factors affect T7 RNAP expression in ClearColi BL21(DE3) with an animal-free media setting

The above study showed that using ClearColi BL21(DE3) with animal-free media can express soluble T7 RNAP. However, the bacteria grow slowly, and the T7 RNAP expression is low. We focus on the ClearColi BL21(DE3) and BL4 medium combination for the rest of this study because of its availability and low price. We used the DoE to systematically investigate how temperature, time, and inducer concentration affect T7 RNAP expression in ClearColi BL21(DE3) with BL4 medium during flask fermentation. We utilized this methodology in our study to maximize soluble T7 RNAP expression by a full factorial design. JMP 16.1 software (JMP 2020) was used to design a multiple-level 3-factor experiment (Table S8). Three temperature levels (16 °C, 30 °C, and 37 °C), two time points (6 h and 18 h), and two inducer concentrations (0.1 mM and 0.5 mM) are factors for testing. Fig. S5a and Fig. S5b are one set of the SDS-PAGE gel images of soluble T7 RNAP expression. After inputting all the T7 RNAP yield values in the table created from JMP 16.1 software, a least square model was fitted and produced an expression equation, which was shown in Python code format (Supplementary code 1). The T7 RNAP concentration used as input for the JMP software was quantified from SDS-PAGE gel using ImageJ software (Rueden et al. 2017) instead of using Bradford assay. ImageJ-based gel quantification enables fast and specific relative T7 RNAP concentration quantification using a protein marker as a reference. While Bradford is a regular method for protein quantification, it is mainly used for total protein quantification, or protein purification has to be done before target protein quantification.

The normal probability plot confirms that the model residuals follow a normal distribution, satisfying assumptions for regression adequacy (Fig. 2a). Furthermore, the high Rsq (0.99427) and low p value (<0.001) demonstrate that the fitted standard least squares model accurately captures the relationship between predicted and actual soluble T7 RNAP expression (Fig. 2b). The normal residual distribution and significant correlation between measured and predicted T7 RNAP levels validate the fitted model adequately describing the system. Among these tested factors, temperature and time interaction were statistically significant, with a log worth of 4.678 (p value, 0.00002). Temperature and time are also significant (log worth = 4.573, 4.084, respectively). The inducer is statistically significant on the T7 RNAP yield (Log worth = 2.657). (Table S9). The coefficient estimates of each factor (Fig. 2c) showed that only temperature (30 °C) has the strongest positive relationship with the T7 RNAP expression. The time, individually, has an inverse relationship with output, but it positively affects the relationship when combined with temperature (16 °C or 30 °C). The factors leverage plots (Fig. S6) show that all the confidence curves cross the horizontal line, which means all the test factors (temperature, inducer, time, temperature*time) are significant. The prediction profile (Fig. 2d) shows that the effects of the three tested factors varied on the Soluble T7 RNAP expression. These results showed that 30 °C, 0.1 mM inducer, and 18 h are optimal for soluble T7 RNAP expression, which has a 15% improvement compared to the previous expression condition (30 °C, 6 h, 0.5 mM IPTG). These results also showed that 0.1 mM of inducer would be better for a higher T7 RNAP yield, which is economically better for production. This flask fermentation-based DoE experiment unveils pivotal findings regarding the parameters influencing optimal soluble T7 RNAP production on a smaller scale. The central players here are temperature and time, which have surfaced as critical elements to fine-tune for optimal yield.

Fig. 2.

A three-factor full factorial DoE analysis of soluble T7 RNAP expression. a Normal quantile plot of residuals for the model. b The actual values of soluble T7 RNAP expression versus the predicted values. c The coefficient estimates plot of each component and their secondary interactions. d Prediction profiler plot of each component effect varied on the soluble T7 RNAP expression and the relationship between each element to optimal soluble T7 RNAP expression

T7 RNAP purification and characterization of RNase, Benzonase, and endotoxin levels

One liter of BL4 medium was used for lab-scale flask fermentation of C-T7 RNAP and EFAF-T7 RNAP production. The Ni-NTA column was used to purify both versions of T7 RNAP. SDS PAGE gel (Fig. 3a) confirmed the protein’s size and showed their purity, about 76% and 71% for C-T7 RNAP and EFAF-T7 RNAP, respectively. The 80 kDa and 20 kDa bands that are proteolytic cleavage products of the T7 RNAP also can be observed, a common phenomenon during T7 RNAP purification (Davanloo et al. 1984; Muller et al. 1988).

Fig. 3.

Specification tests for comparing the C-T7 RNAP and EFAF-T7 RNAP. a SDS-PAGE gel image of C-T7 RNAP and EFAF-T7 RNAP after Ni-NTA purification. The protein size is 99.9 KDa. The molecular weight of the protein marker (M) is given on the left (kDa). b RNase quantification of commercial T7 RNAPs and in-house T7 RNAPs. c DNase quantification of commercial T7 RNAPs and in-house T7 RNAPs. d Interference experiments of commercial T7 RNAPs, EFAF-T7 RNAP, and endotoxin buffer. e Total endotoxin of in-house T7 RNAPs before and after purification with Ni-NTA. Bar values at d and e are average ± standard deviations for all experiments shown for n = 3 repeats. The FD means fold changes.

Then, we used RNase detection assay, Benzonase detection assay, and chromogenic LAL endotoxin assay to quantify the level of RNase, DNase, and endotoxin in both versions of T7 RNAP. Quantifying RNase and Benzonase are essential because they degrade RNA products and DNA templates, respectively. Compared with two commercial T7 RNAPs (from NEB and Vazyme, respectively), our C-T7 RNAP and EFAF-T7 RNAP have almost the same RNase level as commercial enzymes (range from 0.12 to 0.16 pg/μl, Fig. 3b). The Benzonase residue (Fig. 3c) in these four tested samples is different. Their values are 16.1 mU/μl (NEB), 26.0 mU/μl (Vazyme), 20.3 mU/μl (C-T7 RNAP), and 32.3 mU/μl (EFAF-T7 RNAP). The Benzonase level in EFAF-T7 RNAP is about twice that in NEB T7 RNAP, but it is still under the industrial requirement (50 mU/μl). Overall, the RNase and DNase levels in C-T7 RNAP and EFAF-T7 RNAP are almost identical. This is mainly because ClearColi BL21(DE3) is a mutation derivative of BL21(DE3), and the mutated genes do not cause more RNase and DNase accumulation.

When first performing the LAL assay, the LAL assay signals of commercial enzymes are even below the negative control. We reviewed components in the NEB T7 RNAP and Vazyme T7 RNAP storage buffers. There are reducing reagents and EDTA, which are known to interfere with the LAL assay. LAL enzymes involve disulfide bonds and can be inhibited by reductants that can cause decreased LAL assay sensitivity. EDTA can chelate divalent cations like Mg²⁺ and Ca²⁺ that are required for LAL enzyme function, leading to false negatives or artificially reduced LAL assay signals. Then, we performed an interfering experiment according to the Zhong Guo Yao Dian Volume III, version 2015, 1143 bacterial endotoxin test method (National Pharmacopoeia Committee 2015). Our study tested available commercial T7 RNAP, EFAF-T7 RNAP, and a buffer without reducing agent and EDTA (Table S2). The endotoxin recovery rates were calculated from Table S10 using a formula (under the table). The averaged endotoxin recovery rate (Fig. 3d) showed that both NEB (R=39.0%) and Vazyme (R = 8.4%) storage buffers have an interfering effect (R ≤ 50% or R ≥ 200%). The endotoxin buffer, which only has Tris HCl and NaCl, has no interfering effect. Hence, we only used endotoxin buffer for the following LAL assay for protein storage.

Our study quantified the endotoxin before and after Ni-NTA purification for BL21(DE3) and ClearColi BL21(DE3). For balancing the downstream purification, the same amount of cell well was taken for lysis and purification. Depending on the OD difference, we took 290 ml cells from BL21(DE3) with BL4 medium and 390 ml cells from ClearColi BL21(DE3) with BL4 medium. Their total endotoxin is 2,787,298 EU and 8333 EU, respectively (Fig. 3e). Endotoxin in BL21(DE3) is 335 times that in ClearColi BL21(DE3). LAL assay kit can not discriminate between endo toxically active hexa-acylated LPS and endo toxically inactive tetra-acylated lipid IV_A (ClearColi 2013). So even if ClearColi BL21(DE3) virtually has no endotoxin toxic effect, which has lipid IV_A instead of LPS, the LAL assay can still be tested to have substantial LAL value (Mamat et al. 2013). After Ni-NTA and dialysis, the endotoxin level decreased by 99.2% and 97.5% (Fig. 3e), respectively. The endotoxin in C-T7 RNAP (22711 EU) is still approximately 109-fold of that in EFAF-T7 RNAP (208 EU). For endotoxin concentration after Ni-NTA, the concentration is 7570.5 EU/ml (20879.6 EU/mg) and 69.6 EU/ml (252.2 EU/mg) for C-T7 RNAP and EFAF-T7 RNAP, respectively. This level of endotoxin level after a Ni-NTA purification is consistent with the producer’s data (ClearColi 2013). An extra purification process will decrease the endotoxin level from the current 69.6 EU/ml of EFAF-T7 RNAP to 2.5 EU/ml, which is the standard undetectable level of endotoxin in medical applications (Kondratova et al. 2019). However, C-T7 RNAP, which has a high endotoxin concentration (7570.5 EU/ml), will require more complicated or specialized endotoxin-removing processes to meet downstream medical applications (Schwarz et al. 2014; Koziel et al. 2018).

Optimized IVT reaction system for endotoxin-free and animal-free T7 RNAP shows high RNA yield and can produce long RNA

C-T7 RNAP and EFAF-T7 RNAP, together with commercial T7 RNAPs from NEB (Ipswich, MA, USA) and Vazyme (Nanjing, China) were used to do IVT test. The IVT starts with a standard reaction system (Tables S3 and S4) and with the same amount of enzyme input (50 U for NEB and Vazyme T7 RNAP and 0.26 μg for C-T7 RNAP and EFAF-T7 RNAP; the 0.26 μg input is the same mass as 50 U Vazyme T7 RNAP measured by Bradford Assay). The IVT of C-T7 RNAP and EFAF-T7 RNAP yield 44.73 μg and 37.23 μg of RNA, respectively (Fig. 4a and Fig. S7). Even if the yield is slightly different, it is not statistically significant.

Fig. 4.

A four-factor full factorial DoE analysis of RNA synthesis by IVT and IVT function tests. a The RNA yield of primary IVT tests with a standard IVT reaction system. b Actual values of RNA yield versus the predicted values. c Log worth of each significant factor or secondary interaction (value ≥ 2, p < 0.01). d Prediction profiler plot of each component effect varied on the RNA yield and the relationship between each element to produce a high RNA yield. e Comparison of the RNA yield with standard IVT reaction system and the final optimized IVT reaction system using C-T7 and EFAF-T7 with 1-Kb template. The “B.O.” and “A.O.” annotations mean before optimization and after optimization, respectively. f A native agarose gel of synthesized RNA using the final optimized IVT reaction system for quality control. The molecular size of the ssRNA ladder (M) is given on the left. Bar values at a and e are average ± standard deviations for all experiments shown for n = 3 repeats

Next, this study uses DoE to maximize IVT RNA yield using EFAF-T7 RNAP. A typical IVT reaction includes a linearized DNA template, RNA polymerase, nucleoside triphosphates (NTPs), RNase inhibitor, pyrophosphatase, Mg²⁺, antioxidant, polyamine, and PH buffer (Geall et al. 2013). Based on previous studies (Young et al. 1997; van de Berg et al. 2021; Rosa et al. 2022), we identified several variables, Mg²⁺, NTP, enzyme input, and template input, which can affect RNA yield. A four-factor full factorial DoE was designed to find an optimized IVT reaction system. Table S6 is the table for this four-factor full factorial design, where we included three levels of NTP concentration, two levels of MgCl₂ concentration, two levels of template concentration, and two levels of enzyme concentration. The RNA yield was quantified by Qubit 4 flurometer with RNA BR reagent immediately after IVT reactions stopped. The Qubit results were input into the table created by JMP 16.1 software to fit a standard least squares model. The fitted model can give reliable prediction with an RSq of 0.96 and a p value less than 0.0001 (Fig. 4b). The Log worth data (Fig. 4c) showed that NTP, NTP*MgCl₂, MgCl₂, and T7-RNAP*NTP are the four critical factors of statistical significance. MgCl₂ and NTP are the most significant factors, and both have a positive relationship with RNA yield. The interaction between NTP and MgCl₂ showed vital importance, which may imply an optimal ratio between them. Even if the enzyme itself is not statistically significant, the interaction between the enzyme and NTP is substantial. This indicates that the enzyme input should also be high enough to maximize RNA yield if a high NTP concentration is used. The prediction profiler plot (Fig. 4d) showed that the optimal NTP concentration is 10 mM. For the other three factors (MgCl₂, T7-RNAP, and template input), the higher will result in higher RNA yield.

Compared with the previous recipe, our optimized formula has improved the RNA yield by about threefold for both C-T7 RNAP and EFAF-T7 RNAP (Fig. 4e, Fig. S8), which indicates that the optimized formula has a general enhancement effect on the yield of in vitro transcription. Here, we also tested our final IVT recipe to produce long RNA with EFAF-T7 RNAP. As self-amplification RNA (saRNA) emerges, which can induce a higher antigen-specific immune response with less RNA injected (Bloom et al. 2021), the ability to produce long RNA becomes necessary. Fig. 4e shows that our EFAF-T7 RNAP can produce long RNA without compromising the RNA yield with our optimized IVT recipe. The purity of produced long RNA is lower than that of shorter RNA. This has been observed in previous studies (Xia et al. 2022).

Discussion

This study individually tested the expression of T7 RNAP in ClearColi BL21(DE3) with LB medium and BL21(DE3) with animal-free media. When ClearColi BL21(DE3) was tested in the LB medium in the microplate, cell growth in the log and the stationary phases was inhibited by about 50% (Fig. S2). In flask fermentation, the T7 RNAP expression was lower, about 71.4% (Fig. 1d) of that in BL21(DE3) with LB medium combination. Then, four animal-free media were tested with BL21(DE3) strain. Krefft et al. reported that using plant-based media with a lac operator-controlled promoter, such as T7-Lac promoter, lacUV5-Lac promoter, can cause spontaneous depression of the host cell or even plasmid mutation (Krefft et al. 2022). In the microplate growth condition, we observed apparent inhibition after induction, but the cells grew back to a similar OD600 as LB in stationary phase. In flask fermentation, no growth defect was observed at 6 h and 18 h after induction, and the soluble expression of T7 RNAP was the same as that in the LB medium. Furthermore, Sanger sequencing data showed no mutation on the T7 RNAP gene in the pLQY199 plasmid in animal-free media. Davanloo et al. cloned a similar construct (pAR1219) as pLQY199 plasmid (a medium copy number plasmid) and reported cloning difficulty and growth inhibition upon induction (Davanloo et al. 1984). The cloning of the lower copy number plasmid pLQY175 for T7 RNA expression is also not easy. We only got one sequence correct clone out of 24 screened colonies for pLQY175 and got one sequence valid clone out of 30 screened colonies for pLQY199. The T7 RNAP expression in pLQY175 is about half of that in pLQY199 (Fig. S1). One area worth exploring to enhance T7 RNAP expression is using alternative induction systems. Some systems place a high induction burden on the host cell, potentially causing stress and affecting overall expression levels. In this context, the work by Matthews et al. explored using pBAD vectors for T7 RNAP expression (Wycuff and Matthews 2000). However, their study failed to achieve high levels of T7 RNAP expression, suggesting that the induction system is not the sole factor affecting performance. We posit that achieving optimal T7 RNAP expression may require a more holistic approach involving carefully balancing multiple variables. These variables include plasmid copy number, gene regulation mechanisms, and culturing conditions (Balzer et al. 2013; Niu et al. 2014).

Then, we combined ClearColi BL21(DE3) with animal-free media, which showed impaired growth versus LB, with 37% slower log phase growth in HES/HS/BL4 microplate cultures. After induction, additional inhibition occurred across all media compared to LB. In flasks, overnight ClearColi BL21(DE3) growth in animal-free media decreased to OD600 ~0.6, whereas BL21(DE3) reached OD600 ~5. Soluble T7 RNAP expression positively correlated with growth rate, only 50–60% of that in BL21(DE3). The lower T7 RNAP expression in the ClearColi BL21(DE3) strain and animal-free media combination can be explained by two reasons: first, the ClearColi BL21(DE3) strain grows slower, especially in induced conditions. Second, the animal-free media is not optimal for the growth of the ClearColi BL21(DE3) strain. The switch to ClearColi BL21(DE3) and animal-free media resulted in slower bacterial growth and a lower T7 RNAP yield. In this case, the slower growth rate of the ClearColi BL21(DE3) strain and its suboptimal performance in animal-free media could be viewed as a trade-off for achieving endotoxin-free and animal-free protein production. Supplementing specific depleted amino acids could significantly improve ClearColi BL21(DE3) growth in an animal-free N-Z-Soy medium (Wilmes et al. 2010). Serine, threonine, asparagine, and aspartic acid decrease 50–90% during the log phase, likely causing growth limitation. Adding these four amino acids may help to boost the growth rate of ClearColi BL21(DE3) with animal-free media. More broadly, strategies successful for promoting microbial growth in standard media can be systematically tested. Adding inorganic nitrogen sources could compensate for bacteria growth partially. Extra carbon, like glycerol, enhances growth by increasing the carbon source pool (Kumar et al. 2021). Finally, micronutrients, including metals, phosphates, and vitamins, are critical for robust metabolism (Sezonov et al. 2007). In summary, rational supplementation with amino acids, nitrogen, carbon, and minerals will likely enable optimized ClearColi BL21(DE3) fermentation in animal-free media.

Our study employed DoE to investigate how temperature, time, and the level of inducer affect the soluble T7 RNAP expression in ClearColi BL21(DE3) and BL4 medium. Knowing very little about the three factors that influence the soluble expression of T7 RNAP in this combined host and media, we sought to use a multi-level full factorial DoE design to find the best culturing condition for optimal soluble T7 RNAP expression (Papaneophytou and Kontopidis 2014). This full factorial design provides a complete set of experiments (Table S8) bearing on a limited set of factors. Wet experiment results were then combined with the table to fit the model and generate information about the main effects and interactions of the three tested factors. Our findings, presented in Fig. 2a–d and Table S9, ensure that the residuals of the fitted model for soluble T7 RNAP expression satisfy a normal distribution, with a high Rsq (0.99427) and a low p value (0.0001), validating model adequacy. The interaction between temperature and time was statistically significant among the tested factors, showing that temperature was the most impactful factor. The model finds that a temperature of 30 °C, 18 h, and 0.1 mM IPTG is optimal for soluble T7 RNAP expression in ClearColi BL21(DE3) and BL4 medium. In both 0.1 mM and 0.5 mM induction batches, the highest soluble T7 RNAP expression is at 30 °C and 18-h time point, but the 0.1 mM inducer concentration results in about 10% higher soluble T7 RNAP expression. In ClearColi BL21(DE3), T7 RNAP transcription is regulated by the lacUV5 promoter in the host genome, a configuration also seen in the pLQY199 plasmid. While it has been reported that T7 RNAP expression from only the host chromosome is non-toxic, our experimental setup involves IPTG-induced simultaneous transcription of the T7 RNAP gene from both the chromosome and the plasmid. This results in an overproduction of T7 RNAP, which competes with the host’s native RNA polymerase for resources. This competition not only diminishes the synthesis of host mRNA but also serves as a negative feedback mechanism, limiting the further accumulation of T7 RNAP (Studier and Moffatt 1986). This likely explains why 0.1 mM IPTG is optimal for ClearColi BL21(DE3) compared to higher 0.5 mM concentrations. IPTG exacerbates toxicity caused by the expression burden of the recombinant protein (in this case, T7 RNAP) already borne by the ClearColi BL21(DE3) host (Dvorak et al. 2015). IPTG can act to damage recombinant protein expression host, especially at inappropriately high concentrations. Consequently, the IPTG concentration must be carefully tuned to mitigate adverse effects while still inducing T7 RNAP production. The moderate 0.1 mM IPTG level balances induction of the positive feedback loop for T7 RNAP expression against the metabolic stress imposed on the ClearColi host. The nuanced understanding from this research helps our current grasp of T7 RNAP production and lays a solid foundation for future upscale operations in bioreactor fermentations. While this study utilized flask fermentations, further optimization could explore bioreactor cultures that may support higher T7 RNAP yields. Additionally, performing a chemical screen to identify supplemental nutrients synergizing with the animal-free media in ClearColi BL21(DE3) could enable higher growth rates or fermentation cell densities and increased soluble T7 RNAP expression. A combinatorial media supplement screen would allow the systematic identification of minerals, vitamins, lipids, or other components that address nutritional gaps in the chemically defined animal-free medium. It is also crucial to understand how physical factors like time, temperature, shaking speed, and media components, as well as transcriptional factors such as promoter strength and plasmid copy number, influence recombinant protein production in animal-free setup (Balzer et al. 2013; Niu et al. 2014; Lozano Terol et al. 2021), so that adjustments can be made in a coordinated way to improve soluble protein expression while retaining the desired endotoxin-free and animal-free characteristics. Furthermore, bioreactor cultivation provides enhanced control over parameters like dissolved oxygen and pH that influence protein solubility and potentially offers avenues to increase yield in this novel expression system.

Relative to commercial enzymes, the purified in-house T7 RNAPs have reduced purity at 76.3% (C-T7 RNAP) and 71.2% (EFAF-T7 RNAP) but satisfy industrial requirements for RNase and Benzonase activity. Our study measured the endotoxin level in a growing culture without concentration. The BL21(DE3) with BL4 medium has 2787298 EU endotoxin, while the ClearColi BL21(DE3) with BL4 medium has 8333 EU endotoxin. ClearColi BL21(DE3) has inactive tetra-acylated lipid IV_A, which does not induce a toxic endotoxin effect but can still be quantified to have certain endotoxin units by LAL assay. Post-purification, there is 208 EU endotoxin in 0.828 mg (252.2 EU/mg, 69.6 EU/ml) EFAF-T7 RNAP (this is on par with the producer’s data: after Ni-NTA, the remaining endotoxin is 450 EU/mg) (ClearColi 2013; Mamat et al. 2013). The measured endotoxin unit in EFAF-T7 RNAP is from IV_A, which does not induce an endotoxin effect. But for C-T7 RNAP, the endotoxin level (20879.6 EU/mg, 7570.5 EU/ml) is still very high after a single Ni-NTA. An extra cleaning step can make EFAF-T7 RNAP safe for medical use by lowering its endotoxin level to 2.5 EU/ml, but C-T7 RNAP, with its much higher endotoxin level, would need more advanced endotoxin-removing methods to meet medical standards. Current methods for removing endotoxin from recombinant proteins include ultra-filtration, activated carbon, surfactants, anion exchange chromatography, and immobilized Sepharose (Magalhães et al. 2007; Zhang et al. 2013; Mamat et al. 2015). Each of these strategies involves adverse effects, such as significant yield loss, high cost, and loss of protein bioactivity. However, efficiently eliminating endotoxin is challenging, and none of the downstream applications remove endotoxin entirely. Our solution provided a solution for making endotoxin-free T7 RNAP, the critical enzyme for mRNA drug manufacture.

Compared with commercial T7 RNAP, the RNA yield of C-T7 RNAP and EFAF-T7 with standard reaction buffer is about half of NEB IVT yield and two-thirds of Vazyme IVT yield. The RNA yield of C-T7 RNAP and EFAF-T7 is less than that of NEB and Vazyme T7 RNAP. This may be due to the low purity of T7 RNAP (C-T7 RNAP, 76.3%, EFAF-T7 RNAP, 71.2%), which means the actual enzyme input of in-house T7 RNAP is lower than commercial counterparts. The DoE result shows that NTP, NTP*Mg²⁺, and Mg²⁺ are the three most significant factors for maximizing RNA production. Magnesium is the co-factor for T7 RNAP (Gunderson et al. 1987); it also forms a complex with NTP in a 1:1 ratio, which is the building block of RNA (Nanninga 1961). The formed magnesium-NTP complex may serve as a substrate for IVT reaction (Osumi-Davis et al. 1992). On the other hand, magnesium forms Mg₂PPi (which can precipitate) with the inhibitory pyrophosphate (PPi) produced during the IVT reaction (Akama et al. 2012). It is reported that too little or too much NTP can decrease RNA yield (Milligan et al. 1987). Our DoE result highlights the importance of NTP*Mg²⁺; it may imply an optimal ratio between Mg²⁺ and NTP in specific IVT reaction conditions. T7 RNAP input is not an essential factor, but T7 RNAP*NTP is a statistically significant factor, which may imply that more enzyme input is necessary if NTP concentration is high. Our four-factor full factorial design has found an optimal IVT system, which increased the RNA yield by about threefold in 1 kb template in vitro transcription (Fig. 4e). The major difference of this optimized IVT system with the standard IVT system is that more Mg²⁺(60 mM), NTP (10 mM each), and T7 RNAP (0.052 μg/μl) were used. A usage of 60 mM Mg²⁺ and 10 mM NTP in IVT was also recommended by Young et al. (1997). This optimal IVT reaction system was used to test the ability to produce long saRNA, the next generation of mRNA pharmaceutics. Our optimal IVT reaction can make an 8.7 Kb saRNA sequence without obvious yield compromise. This study lays the groundwork for further strengthening the IVT reaction for maximizing RNA yield. An important next step would be optimizing co-transcriptional capping using cap analogs, the preferred method for generating capped RNA transcripts (Henderson et al. 2021). Machine learning and directed evolution approaches could be leveraged to engineer enhanced T7 RNAP mutants and improve reaction conditions. A multifaceted optimization strategy of coupling protein engineering with systematic buffer screening and process optimization represents a promising avenue for maximizing RNA product yield while minimizing costs.

In conclusion, developing an endotoxin-free and animal-derivative-free production method for T7 RNAP presents a significant advance towards safer, more ethical bioproduction practices. We found that the T7 RNAP produced under these novel animal-free, endotoxin-free conditions are functionally equivalent to conventionally produced T7 RNAP. The optimized IVT system’s capacity to produce high yields of RNA emphasizes its potential for scaling up the manufacturing of RNA therapies. This area has gained significant interest, with over 500 mRNA drug development pipelines reported as of February 2023 (Staff 2023). In addition, this optimized IVT system also enabled the synthesis of long self-amplifying RNAs, which is now the primary developing focus of mRNA drugs for rare genetic diseases as it has extended expression duration (Bloom et al. 2021).

Author contribution

QYL and LC designed the study. QYL and LC performed the experiments and collected all the data. BWT obtained biological resources and analyzed data for this study. QYL and LC analyzed the data and wrote the manuscript. All authors read and approved the manuscript.

Funding

This work was supported by the Changzhou University Research start-up funding to LUN CUI (Grant No. ZMF21020037). This research is also supported by the Changzhou Science and Technology Bureau’s 2022 Science and Technology Support Plan (Grant No. CE20225012). This research is supported by the Changzhou Science and Technology Bureau’s 2021 International Collaboration Projects (Grant No. CZ20210016). This work is also supported by Top Talent of Changzhou, “The 14th Five-Year Plan" High-Level Health Talents Training Project (2022CZBJ096), and the Innovation Project of Changzhou Medical Center (CMCB202217).

Data availability

The datasets generated and analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval

This study does not include any experiment procedures related to or require ethical approval.

Competing interests

The authors declare no competing interests.