USTAT: Unified Sequential Template Amplification and Transcription—a fully synthetic mRNA manufacturing platform

Abstract

mRNA vaccines have emerged as a transformative modality for infectious disease prevention. In response to the emergence of SARS-CoV-2, large-scale in vitro transcription (IVT) of mRNA vaccines was developed. Large-scale IVT currently relies on linearized plasmid DNA (pDNA) as a template for mRNA production. Linearized pDNA production presents several challenges at manufacturing scale, including removal of residual host-cell DNA, protein, endotoxins, and antibiotics. Additionally, pDNA-derived sequences irrelevant to mRNA production must be removed from the final product. Finally, the generation of linear pDNA template is laborious, which reduces mRNA production speed, a renowned advantage of this technology. Enzymatic DNA amplification strategies such as rolling circle amplification (RCA) of a synthetic circular DNA molecule offer a rapid, isothermal reaction as an alternative to pDNA. Therefore, we have developed a fully synthetic, single-vessel mRNA manufacturing platform. Beginning with a chemically synthesized circular DNA template, we amplify via a fit-for-purpose RCA, linearize with a TypeIIS restriction enzyme (RE), and perform IVT in a single vessel. The entire process—from circular template to mRNA—can be completed in as little as two days. This method, termed Unified Sequential Template Amplification and Transcription (USTAT), eliminates bacterial components, large volume pDNA production, and enables rapid, modular mRNA production.

Introduction

Large-scale IVT relies on linearized pDNA as the template for mRNA production. While widely used, pDNA poses regulatory, biochemical, and timeline challenges. These include eliminating bacterial contaminants such as genomic DNA, RNA, and residual proteins¹. Extensive control strategies must be in place to ensure the removal of pDNA from the mRNA drug substance (DS). E. coli-based pDNA fermentation is subject to topological variability, including supercoiling, dimerization, and truncation of homopolymers (e.g., polyA), thus requiring laborious screening². As the field matures, more agile manufacturing processes are being considered, including alternative DNA template formats that are fully synthetic and cell-free.

One such strategy is the use of phi29-driven RCA, a high-yield, isothermal enzymatic process that generates linear DNA from a small quantity of circular DNA template^3–5. Previous studies have demonstrated the use of RCA-derived DNA in IVT, producing mRNA with reasonable quality compared to pDNA-derived mRNA⁶. Given these studies and the ability of RCA to produce milligram to gram scale DNA in a simple isothermal format, it is attractive for the development of large-scale manufacturing⁷.

Here, we have developed a fit-for-purpose, fully synthetic platform, beginning with DNA fragments—designed in silico, synthesized chemically, and circularized in vitro—as the starting material for RCA. These synthetic circular templates are designed to include only transcriptionally relevant elements (Fig. 1A), including a T7 RNA polymerase (T7RNAP) promoter, untranslated regions (UTRs), open reading frame (ORF), and a defined polyA tail. The synthetic template strategy removes any unnecessary sequences and enables a higher molar template concentration in IVT reactions.

Fig. 1. Schematic representation of Unified Sequential Template Amplification and Transcription (USTAT) process with a minimal circular template.

A Golden Gate Assembly (GGA) of a minimal circular DNA template using a synthesized gene block containing transcriptional elements and a duplexed ultramer oligo containing polyA tail, is followed by exonuclease hydrolysis of all non-circularized DNA. B The assembled minimal circle DNA is used in an RCA reaction to produce concatemeric DNA, which is enzymatically linearized and used as a transcription template in an IVT reaction, generating 5’capped mRNA in a single vessel.

Critically, we developed a novel, single-vessel formulation of an RCA reaction such that the resulting DNA can be linearized and fed into IVT without purification (Fig. 1B). This method, termed Unified Sequential Template Amplification and Transcription (USTAT), includes buffer and component modifications to the RCA, designed to ensure quality amplification, complete linearization of the branched concatemeric DNA to prevent read-through transcription, and preserve template integrity to support production of high-quality mRNA.

Results

Minimal template design

To evaluate whether a fully synthetic DNA template could be used in a single-vessel mRNA production system, we designed two minimal double-stranded DNA fragments for assembly: one containing a T7 promoter, 5′ UTR, ORF, and 3′ UTR, and a second fragment containing a polyA tail (80A) flanked by a TypeIIS RE site (BspQI) for linearization (Fig. 1A). These fragments were synthesized commercially using high-fidelity gBlocks (fragment 1) and ultramer duplexed DNA (fragment 2) from Integrated DNA technologies (IDT). Standard assembly protocols like Golden Gate Assembly (GGA) are designed for low input E.coli transformations relying on bacterial amplification⁸. In contrast, our optimized GGA system generates improved closed circular DNA yield, which is then exonuclease-purified and used as a template for RCA (Fig. 1A). Several modifications were introduced into our optimized GGA to increase ligation efficiency and maximize the final mass yield of circular DNA.

Assembly optimization

First, we substituted T7 DNA ligase for T4 DNA ligase to improve sticky-end ligation and increase circularization⁹. Synthetic DNA overhangs were designed to be GC-rich to promote more efficient overhang hybridization (Fig. 2A)¹⁰. These overhangs are efficiently cut with BsaI (TypeIIS) flanking the ends of each fragment, supporting our assembly strategy.

Fig. 2. Systematic optimization of Golden Gate Assembly parameters for synthetic circular RCA template.

A Normalized ligation counts for overhangs used in T7 Ligase-mediated GGA generated through NEBridge Ligase Fidelity Viewer®. B Agarose gel electrophoresis of systematic GGA optimization through titration of small fragment (LF) to large fragment (SF) ratios, decreasing titration of T7 DNA Ligase, and increasing titration of ligation enhancer 1,2 PrD with desired covalently closed circular monomer indicated by box. C Densitometry analysis of AGE in (B) showing GGA reaction efficiency through percent circularization of covalently closed circular DNA. Optimal fragment ratio of SF:LF for the highest circular DNA product was determined and was used in subsequent testing of T7 DNAL. Both SF:LF of 1, and 25U T7 DNAL were used in testing with 1,2 PRD. Statistical analysis was performed using a one-way ANOVA, with Dunnett’s post-hoc test used for condition comparison between enhancement strategies. P = 0.0001 fragment ratio plot, P = 0.0016 Ligase plot; error bars represent mean ± s.d. (n = 3, biological replicates). D Densitometry of 1 ug GGA reaction showing various topological products along with 48% covalently closed circular DNA. E Confirmation of covalently closed circular DNA from GGA with T5 Exonuclease digest of non-circularized material, such as multimers and linear unligated fragments.

We next performed optimizations that included varying molar ratios of large to small fragments, downward titration of T7 ligase to minimize enzyme usage and linear multimer formation, and 1,2 PrD to lower the melting temperature (Figs. 2B, 2C), in order to identify our best case assembly^11–13. We improved ligation efficiency 4-fold from 8–10% (standard protocol, data not shown) to as high as 39% in our small-scale optimization studies. A 1:1 stoichiometry, 25-150 U/uL T7 concentration and inclusion of 1,2 PrD yielded the most optimal ligation efficiency of 48% at larger scale (1ug DNA yield), as confirmed by agarose gel electrophoresis (AGE) densitometry (Fig. 2D) and subsequently after exonuclease digestion of linear fragments (Fig. 2E). Our improved ligation efficiency of closed circular DNA is sufficient to support downstream RCA for integration into a fully synthetic mRNA manufacturing platform.

Using RCA DNA as a template for IVT

We sought to determine whether RCA DNA could serve as a suitable template for IVT. The resulting mRNA should have high yield, integrity, and 5’ capping with low double-stranded RNA (dsRNA) content. RCA is commonly used in molecular diagnostics and genomics for signal amplifications such as in situ detection, padlock probe-based diagnostics, and nucleic acid quantification platforms^14–16. RCA is less frequently used to produce IVT templates¹⁷. It is unclear whether RCA DNA, in its hyperbranched form, would inhibit T7RNAP or linearize sufficiently to serve as an IVT template, as insufficient linearization leads to poor quality mRNA¹⁸.

To evaluate this, we used a commercial RCA kit (NEB phi29 DNA Polymerase) to generate the IVT template. Figure 3A illustrates the functional features of the primary synthetic construct (linearized) used for this study, together with its linearized plasmid counterpart, used as an IVT control. RCA of the minimal circular template was followed by DNA purification, linearization by BspQI, and a second purification step prior to IVT (Fig. 3B, right gel, column 1, +/+). Using an IVT protocol adapted from a method described by Popova et al. (2024)⁴⁹, mRNA yield and integrity were assessed by CGE (Bioanalyzer)⁴⁹. We found that RCA DNA could function as an IVT template; however, mRNA integrity is poor compared to conventional pDNA, with significant levels of lower-than-expected molecule-weight RNA species (Fig. 3D, left electropherogram +/+). To understand why the RCA DNA template led to lower quality mRNA, we assessed the RCA DNA template quality using AGE, specifically looking for undesirable amplification products (Fig. 3B, boxed bands). We defined desirable DNA (Fig. 3B, blue arrow) as the template size equivalent species in AGE after linearization. All other species were defined as non-specific material, either immobile in the gel well or appearing as higher/lower than expected molecular weight, which we interpret as undigested, truncated, sheared, and branched species, or non-templated background amplification (Fig. 3B, boxed bands). Thus, we evaluated the total non-specific amplification by AGE densitometry on each lane (Fig. 3C), and this metric was used as a quality checkpoint to guide further RCA optimizations. Additionally, where possible, we sought to reduce non-specific amplification and increase DNA yield as a key pharmaceutical manufacturing consideration.

Fig. 3. Product quality improvements through removal of intermediate purification steps.

A Schematic overview of a 2.3 kb linearized minimal circle DNA template for RCA compared to a 4.3 kb linearized pDNA template. Both templates transcribe a 2.1 kb luciferase mRNA. B AGE of linearized pDNA and unlinearized RCA DNA, followed by systematic demonstration of the removal of purification steps in RCA for the IVT process. Highlighted are the non-specific DNA products present in the linearized RCA reactions (boxed) and desirable DNA (blue arrow). C Non-specific DNA is calculated as the percentage of DNA that is not target DNA according to AGE densitometry. Statistical analysis was performed using a one-way ANOVA with Dunnett’s post-hoc test for comparison to std. double purification. P values are indicated on the plot; error bars represent mean ± s.d. (n = 3, biological replicates). D mRNA electropherograms showing percent mRNA integrity of various RCA-IVT purification workflows and pDNA controls.

We hypothesized that intermediate purification of RCA DNA, such as column-based cleanup or alcohol precipitation, may damage the branched RCA DNA concatemers through handling, shear forces, or buffer incompatibilities, leading to a loss of template quality^19,20. Therefore, we developed a single-vessel workflow that eliminates purification between RCA, linearization, and IVT (Fig. 3B, −/+ and −/−) in which the mini-circle templated RCA reaction is allowed to incubate to completion, followed by a batch addition of RE directly into the reaction vessel without heat inactivation, and a final batch addition of IVT reaction mix for mRNA synthesis. This process preserves template structure (Fig. 3B), reduces non-specific DNA products (Fig. 3C), improves mRNA quality (Fig. 3D, 3rd electropherogram), and importantly reduces laborious unit operations in the context of a scalable manufacturing process.

USTAT development and optimization

Knowing that our single-vessel system improved overall mRNA quality, we next focused on understanding the source of non-specific DNA amplification, a known phenomenon in extended, random primed RCA reactions without a template^21,22. Non-specific amplification could reduce final mRNA quality and present an impurity concern¹⁸. We conducted a time-course of RCA reactions over 3.5 h, with and without template DNA (Fig. 4A) and found that background DNA synthesis became detectable in the no-template control (NTC) after 2 h while template-containing reactions showed robust amplification at 2 h (Fig. 4A). Therefore, we chose to limit RCA reactions to 2 h to reduce any non-specific amplification and improve product quality.

Fig. 4. Optimization of USTAT.

A RCA incubation time shows non-specific amplification in the no template control (NTC) at >2 h. B Removal of Ppase shows no impact on target DNA yields or non-specific amplification. C Titration of random hexamer (RH) primers to determine optimal primer density for amplification without resulting in high non-specific amplification. D Titration of RH primers in RCA templated IVT to determine optimal primer density in relation to mRNA quality metrics. E Adjustment of pH and inclusion of monovalent salt with variable anion shows variable effects on target RCA DNA quality. *pH of RCA used in (D). F mRNA yield and integrity from monovalent salt anion and pH adjustment. G Further refinement of RCA reaction using reaction enhancers A—5% DMSO, B—40 mM TMAC, C—40 mM (NH₄)₂SO₄, versus (−) no enhancers reaction (A–F), shown are the target DNA amount and percent non-specific DNA. H mRNA yield and integrity are shown compared to the pDNA control. Statistical analysis was performed using a one-way ANOVA with Dunnett’s post-hoc test for comparison to a standard condition or pDNA. For direct comparison between enhancement strategies, an unpaired t-test was used. P values are indicated on the plot; error bars represent mean ± s.d. (n = 3, biological replicates).

To simplify our RCA reaction, we sought to remove inorganic pyrophosphatase (PPase), a common additive in amplification reactions for pyrophosphate (PPi) hydrolysis²³. We found that RCA reactions conducted without pyrophosphate produced no significant difference in DNA yield or non-specific amplification (Fig. 4B), suggesting that PPi accumulation did not limit shorter duration RCA reactions. As a result, we excluded PPase from the final RCA formulation to remove a potential source for contaminating DNA and reduce cost. This was added back during the IVT process to prevent Mg-PPi accumulation during the much faster mRNA synthesis.

To continue USTAT development, we examined conditions supporting robust RCA activity, RE function, and efficient transcription in a single vessel. Key optimization parameters included: (1) random hexamer (RH) primer concentration, (2) buffer components (e.g., pH, KCl, (NH₄)₂SO₄) compatible with both BspQI and T7RNAP, and (3) linearization temperature. We evaluated the reaction components’ impact on RCA yield, non-specific amplification, as well as mRNA yield and integrity. Each reagent was formulated based on its biochemical role in RCA, linearization, and transcription to achieve a high-quality or high-yield mRNA product. In the absence of a significant impact on the mRNA product, we choose to include reaction parameters which resulted in lower non-specific DNA or resulted in higher RCA DNA yield to maximize process robustness.

phi29 initiates amplification from primed sites on circular DNA and addition of RH primers increases initiation sites, enhancing amplification yield²⁴. However, high priming density can negatively impact both RCA efficiency and DNA quality by resulting in non-specific amplification, steric interference, sequestering, or premature displacement of phi29²⁵. Conversely, low priming density can decrease complementary strand synthesis resulting in a high accumulation of ssDNA, which can also negatively impact RCA efficiency through phi29 sequestration or steric interference²⁶. phi29 relies on strand displacement synthesis, making optimal primer concentration essential for continuous amplification and quality DNA production. phi29 also possesses a potent 3′ → 5′ exonuclease activity, thus RH primers were 3’ phosphorothioate-modified to preserve primer integrity and prolong reaction availability^27–29. The impacts of primer density on downstream linearization and transcription are currently unknown.

To optimize priming of the RCA in USTAT, we titrated RH primers downward from 50uM to 10uM (Fig. 4C). We determined that 25 µM RH provides the lowest percentage of non-specific DNA in a templated RCA while achieving adequate desired DNA yield (Fig. 4C). RCA DNA from each RH input was then normalized to 10 nM for IVT and RNA yield and integrity remained consistent across tested conditions (Fig. 4D), thus 25uM RH concentration was selected to sustain exponential amplification without producing excess non-specific DNA.

Monovalent cations (e.g., Potassium and Sodium) are essential for nucleic acid amplification/polymerization reactions as modulators of the electrostatic environment around DNA and RNA^30–32. In RCA, monovalent ions help maintain the conformational stability of the DNA and phi29 active sites for high fidelity and continuous strand displacement across long concatemers³³. Potassium also enhances polymerase activity by promoting correct primer alignment and suppressing non-specific initiation events³⁴.

Tris-HCl was selected as the primary buffering agent in USTAT formulation to maintain a stable pH compatible with phi29, BspQI, and T7RNAP enzymatic functions. Concentration ranges of 20 mM–50 mM Tris (effective buffering ranges of pH 7.5–8.5) are often used in independent DNA polymerization, linearization, and mRNA synthesis reactions, with pH reported to affect both polymerase and nucleotide incorporation rates^35–38; we sought to optimize Tris-HCl buffering of our single-vessel reaction together with added salts.

To evaluate how pH and salt influence USTAT, we tested 50 mM Tris-HCl at two different pH values (8.0 and 8.8 at 25 °C), each paired with 50 mM KCl (Fig. 4E), versus the stock manufacturer’s buffer (NEB) used thus far to evaluate RCA. Tris-HCl at pH 8.0 consistently produced the highest yield and quality of target DNA (Fig. 4E). Tris-HCl at pH 8.8 yielded lower amplification and was not developed further (Fig. 4E). These reactions were carried forward into IVT without purification, producing consistent mRNA yield and significantly improved mRNA integrity upon inclusion of KCl (from ~60% to ~81%) (Fig. 4F). Together, these findings support the use of 50 mM Tris-HCl at pH 8.0 with 50 mM KCl as the optimal buffering system for USTAT.

Next, we explored the potential of small molecule enhancers and other salts for their ability to promote amplification efficiency, specificity, and positively impact USTAT. We tested 5% dimethyl sulfoxide (DMSO), 40 mM tetramethylammonium chloride (TMAC), and 40 mM ammonium sulfate ((NH₄)₂SO₄) as additives in the RCA reaction. Each of these reagents has been previously reported to improve amplification outcomes through distinct mechanisms^39–42. The addition of DMSO reduced both DNA amplification (Fig. 4G) and RNA integrity (Fig. 4H), while the addition of either 40 mM TMAC or 40 mM ammonium sulfate showed reduced non-specific DNA products (Fig. 4G). These data support that ammonium ions may provide a beneficial balance of ionic strength and primer annealing specificity, especially in the context of RCA reactions where multiple primer-template interactions occur simultaneously⁴¹. Due to the toxicity and hazards of TMAC, we selected ammonium sulfate for inclusion in the final USTAT formulation.

To further increase USTAT mRNA quality, we hypothesized that higher RE incubation temperatures might improve accessibility to structured or partially double-stranded regions in concatemeric DNA, potentially enhancing linearization efficiency or T7RNAP accessibility, leading to an increase in RNA integrity. To assess whether linearization temperature could influence downstream mRNA quality attributes, we evaluated BspQI digestion of RCA product at two temperatures: 37 °C (Data in Figs. 1–4) and 50 °C. RCA products are known to contain extensive secondary and tertiary structures due to repetitive, branched architecture and incomplete strand displacement⁴³. At elevated temperatures, these conformations are more likely to transiently melt or destabilize, reducing steric hindrance and making recognition sites more accessible to RE’s such as BspQI^44,45.

Following RCA and restriction digestion at 37 °C and 50 °C, the resulting DNA was resolved on an agarose gel (Fig. 5A) to visually inspect amplification and linearization. These reactions were used directly in single-vessel IVT reactions, and we observe no statistically significant differences in RNA yield compared to pDNA (1.4-1.5 µg/µL USTAT versus 2.0 µg/µL pDNA), (Fig. 5B). Importantly, both temperatures produced acceptable mRNA integrity levels of 86%, nearing that of pDNA-derived mRNA (Fig. 5C). While DNA yield, DNA target specificity, and mRNA integrity were the primary metrics used throughout the optimization of the USTAT workflow, additional critical quality attributes (CQAs) are required to evaluate therapeutic-grade mRNA. These CQAs include dsRNA content, 5′ cap structure integrity, in vitro expression, and RNA sequence fidelity, all factors that impact mRNA efficacy and safety.

Fig. 5. Exploiting BspQI tolerance to high temperature incubation on final USTAT mRNA quality.

A Triplicate linearizations of minimal circle templated RCA at 37 °C and 50 °C. pDNA is shown as a control. B DNA quality metrics, including target DNA yield and percent non-specific DNA from elevated linearization temperature. C Resulting mRNA yield and integrity from RCA performed at 37 °C and 50 °C. Statistical analysis was performed using a one-way ANOVA with Dunnett’s post-hoc test for comparison to pDNA. P values are indicated on the plot; error bars represent mean ± s.d. (n = 3, biological replicates).

Additional USTAT quality metrics

dsRNA is a well-known innate immune activator and potent trigger of type I interferon responses when introduced in vivo⁴⁶. Therefore, regulatory guidance (Regulation EC No. 726/2004, Regulation EC No. 1394/2007, EMA/CHMP/BWP/82416/2025) for mRNA therapeutics requires stringent control of dsRNA impurities. Prior studies have shown that synthetic or cell-free template systems often produce higher levels of dsRNA than traditional pDNA-derived IVT workflows⁴⁷. This is thought to result from more frequent template slippage, self-priming, or aberrant extension events during transcription when template structure or quality is suboptimal⁴⁸. Therefore, temperature changes during linearization may impact template DNA quality, folding, or reaction dynamics and result in variable dsRNA content post-IVT. We examined dsRNA content in our RNA samples using a J2 ELISA dsRNA detection assay. dsRNA content was indeed higher in our cell-free system (Fig. 6A). dsRNA content was reduced in samples linearized at 50 °C compared to 37 °C; however, dsRNA content remained more than two-fold higher than that observed with pDNA templates (Fig. 6A), suggesting that additional reaction refinements or mRNA purification strategies may be necessary for comparability to pDNA workflows. These findings do suggest that linearization temperature has a significant and measurable impact on dsRNA content and support the use of elevated temperatures as a potential strategy to mitigate dsRNA generation in synthetic workflows.

Fig. 6. Comprehensive quality assessment of final USTAT mRNA product encompassing 5' capping, dsRNA burden, integrity, sequence fidelity and functional expression.

A Assessment of 5’ capping and dsRNA from elevated RCA linearization temperature (USTAT mRNA) in comparison to mRNA from pDNA templated IVT. B Comparison of mRNA from optimized USTAT reaction and pDNA templated IVT. C Transcript sequence analysis of mRNA generated by USTAT and pDNA processes. D Cellular expression of mRNA generated by USTAT and pDNA process measured by luminescent output at various time points post-transfection in HEK293T cells. Statistical analysis was performed using a one-way ANOVA with Šidák’s correction applied to plot A-dsRNA, and Dunnett’s post-hoc test for comparison to pDNA applied to plot A-5’ Capping. For direct comparison between USTAT samples or to pDNA in (D), an unpaired t-test was used. P values are indicated on the plot; error bars represent mean ± s.d. (n = 3, biological replicates).

Important to overall mRNA expression and therapeutic efficacy is the inclusion of a 5’ cap. Both our pDNA and USTAT processes are co-transcriptionally capped and therefore potentially sensitive to reaction conditions. To compare USTAT (at 37 °C and 50 °C) to pDNA, we examined 5’ capping efficiency using cap-specific fragment LC-MS. USTAT capping efficiency was equivalent between both temperatures and consistently 5–6% lower than mRNA derived from pDNA templates (Fig. 6A).

We next scaled up our USTAT reaction 10× to 1.5 mL final volume (IVT stage) to generate material for RNA-sequencing and cell-based expression analysis. mRNA yield and integrity were comparable to the pDNA process (Fig. 6B). Since USTAT relies on enzymatic DNA amplification, the fidelity of the template and resulting mRNA are of potential concern. Using RNA sequencing, we examined RNA produced from both USTAT and the pDNA workflow to evaluate sequence fidelity. Sequencing reads were aligned to the luciferase reference sequence, and both datasets demonstrated near-complete alignment without insertions, deletions (indels), or nucleotide mismatches (Fig. 6C), confirming that the USTAT process does not introduce detectable sequence errors that compromise transcript integrity.

Cell-based expression of USTAT mRNA

Given that we found USTAT-produced RNA to have increased dsRNA content and slightly reduced capping efficiency with comparable integrity (Fig. 6A), we wanted to determine any impact on expression. To evaluate the translational efficiency of mRNA generated from USTAT, we compared cell-based expression of luciferase-encoding mRNA produced from USTAT and pDNA template following transfection into HEK293T cells. Luminescence activity was measured using a Firefly luciferase (fLuc) luminometer assay, and RLU values were normalized with a co-transfected control plasmid (Nanoluc) to account for transfection variability (Fig. 6D). At early timepoints (6.5-h and 24-h post-transfection), luciferase expression from pDNA-derived mRNA showed ~35% higher luminescence. Interestingly, by 48 h post-transfection, luciferase expression from USTAT mRNA surpassed that of the pDNA control by 2-fold (Fig. 6D). We speculate that low expression of USTAT-mRNA at early time points may be due to the higher dsRNA content, reduced capping, or an unknown factor. Further examination of mRNA folding from USTAT may help explain the slower rate of decline in luciferase activity compared to pDNA. Collectively, these results demonstrate we were able to develop a feasible, fully synthetic mRNA production platform (USTAT), capable of generating functional mRNA. Furthermore, we were able to demonstrate consistency of mRNA integrity and comparable to pDNA fidelity (Fig. 6B, C) with a 10× scaling of the initial USTAT reaction, suggesting greater scalability is possible in a manufacturing setting.

Discussion

This study introduces a fully synthetic, single-vessel mRNA manufacturing platform that eliminates the need for pDNA fermentation and downstream processing steps. By ligating synthetic DNA fragments into a minimal circular DNA with a robust assembly reaction and amplifying with an optimized RCA reaction, we demonstrate a streamlined method to generate IVT DNA template. This template does not contain non-essential DNA elements and removes the need for E. coli fermentation at any stage of the process. The resulting workflow, USTAT, enables mRNA synthesis in two days– a dramatic acceleration over traditional pDNA workflows that can take several weeks. Notably, we show that omitting intermediate purification steps between RCA, linearization, and IVT not only simplifies manufacturing and quality control, but also improves final mRNA quality. The proposed single vessel workflow is analogous to continuous-unit operations increasingly adopted in pharmaceutical manufacturing, where discrete steps are collapsed into one controlled process to reduce handling, reduce suite footprint, minimize variability, and shorten cycle times—like a one-pot chemical synthesis or an integrated perfusion bioreactor. Through systematic RCA condition optimization, we identified reaction parameters that minimize non-specific DNA species and improve downstream mRNA yield & integrity. The resulting mRNA from USTAT shows quality comparable to pDNA-template-derived mRNA across multiple metrics, including dsRNA, 5’ capping, and sequence fidelity.

While these results establish the feasibility of this workflow at bench scale, further studies are required to demonstrate performance and reproducibility at manufacturing scale (both clinical and commercial), including careful evaluation of enzyme stability, reaction homogeneity, and process controls. Risk mitigation from contaminating host cell DNA found in enzymes or low-fidelity chemically synthesized gene blocks used in template assembly, remain a key consideration for implementation. The sourcing of high-quality raw materials and a strong analytical release package will be critical to the success of the manufacturing process. While we have observed mRNA quality metrics similar to pDNA processes, further impurity characterization and development of new CQAs are required. For example, USTAT exhibits elevated dsRNA, lower peak cellular expression, yet longer sustained expression when compared to pDNA (Fig. 6B, D). The cause of this observation remains unclear and requires further investigation. Finally, it is unknown how reaction parameters would need to be changed to accommodate longer transcripts or how efficient it would be to produce large synthetic templates at scale. Looking ahead, USTAT offers a faster mRNA production platform to rapidly address unmet and evolving medical needs in biotherapeutics, including personalized medicine. The proposed single-vessel workflow has advantages similar to continuous manufacturing platforms, where discrete unit operations are collapsed into one controlled process. This limits handling, minimizes variability, improves product quality, shortens cycle times, and reduces manufacturing footprint. USTAT also offers a unique opportunity to potentially mobilize biotherapeutics production, an area of interest for the pharmaceutical industry.

Methods

Fragment design

A 2.1 kb HiFi gene block (gBlock) was ordered from IDT containing a 175 bp stuffer sequence (GTTAACGGGACTGAATCCGGTGAGAATGGCACGGTTGGACTCAAGACGATAGTTACTTGTTCAACAGGCCAGCCATTACGCTCGTCAGCAAAATCACTCGCATCAACCAAACCACTTACTATAGAGGACCGAACTGAGATACCTACAGCGTGAGCCTCGAGGTACTGAAGATATC), followed by T7RNAP promoter, AG start site, 5’UTR, ORF coding for firefly luciferase gene, 3’UTR. Flanking on both 5’ and 3’ sides were a GGA cassette containing a 4 bp high GC-rich overhang, BsaI recognition site, and a 35 bp stuffer sequence for primer annealing and PCR amplification. Additionally, a duplexed ultramer sequence was ordered from IDT containing identical GGA flanking cassettes with an 80 A homopolymer sequence and BspQI recognition site. Overhang sequence was determined using NEBridge Ligase Fidelity Viewer® and selecting the top two overhangs that yielded the most ligation events.

Fragment PCR, digest, and cleanup

Each fragment was PCR amplified with Q5® High-Fidelity 2× Master Mix from NEB for 28 cycles following the provided protocol with primers (CAATACATGCGAGTGAGCG and ACATTAAAGATATCGTGTTACGGG). 0.4 uM of each primer was used in the large fragment PCR, while 2 uM of each primer was used in the short fragment PCR, with an additional 10 uM of dATP/dTTP mix added to the latter reaction. PCR cleanup was performed using the QIAquick PCR Purification Kit using the provided protocol. Fragments were then digested with 5 U BsaI-HF (NEB)/1ug DNA in 1× Cutsmart buffer (NEB) at a final DNA concentration of 100 ng/uL for 1 h at 37 °C followed by cleanup using QIAquick PCR Purification Kit. Each fragment was run on a 1% agarose gel for 1 h at 100 V before and after digest to check for proper overhang digestion.

Assembly

The starting point of GGA optimization was a standard T7 DNA ligase protocol (NEB) in manufacturer provided buffer at 37 °C for 18 h. The time and temperature of the reactions were unaltered in optimizations. Included in all ligations were 0.2 U/uL BsaI-HF in the event there were partially undigested fragment ends. Following ligation, T5 Exonuclease was added at 0.1 U/uL at 37 °C for 2 h to remove non-ligated fragments. 23.4 uL of reaction before and after exonuclease treatment was run on 1% agarose gel with EtBr for 1 h at 100 V for visual inspection. The post-exonuclease product was purified using the QIAquick PCR Purification Kit. Circularization efficiency of the target DNA band, and again post-exonuclease treatment, was measured using densitometry with BioRad ImageLab software (50% sensitivity).

RCA

Standard RCA conditions unchanged in optimization reactions include 20 pM of purified synthetic template or pDNA, RH primers (MCLAB), 0.25 mM of each dNTP (Invitrogen), and 0.5 U/uL phi29 (NEB) in a 50 μL reaction, incubated at 30 °C. Initial RCA reactions also included NEB 10X phi29 buffer and 200 U/mL Ppase (NEB), and 50 uM RH primers. Purifications of RCA were done using QIAquick PCR Purification Kit (Qiagen) using the standard protocol. RCA was analyzed on a 1% agarose gel with EtBr for 1 h at 100 V.

Linearization

The 100 U/mL of BspQI (NEB) was used to digest hyperbranched RCA in either 1× NEB r3.1 or directly in RCA reaction at either 37 °C or 50 °C for 1 h. Purifications were performed using the QIAquick PCR Purification Kit using the standard protocol. RCA DNA yield was only calculated after linearization with Qubit dsDNA Assay standard protocols. Linearized RCA was run on a 1% agarose gel for 1 h at 100 V, and the target product percentage was measured using densitometry on BioRad ImageLab software set at 50% sensitivity. Non-specific DNA percentage was calculated by subtracting the percentage of the target product from 100%.

IVT

IVT protocol was adapted from Popova et al. (2024)⁴⁹, low dsRNA process with the addition of linearized RCA DNA or pDNA input and co-transcriptional capping reagents (CleanCapAG obtained from TriLink). IVT volume was adjusted to keep DNA template concentration consistent at 10 nM. mRNA was precipitated using standard 2.5 M LiCl (Invitrogen) precipitation followed by EtOH washes and resuspended in 250 μL of RNase-free water. Total mRNA was quantified using a NanoDrop UV-Vis Spectrophotometer, and mRNA yield per reaction was calculated by dividing total mRNA by IVT reaction volume. mRNA was analyzed using the Agilent RNA 6000 Nano Kit with standard protocols on the 2100 Bioanalyzer. mRNA integrity percentage was determined by calculating the area under the expected mRNA peak relative to total mRNA on 2100Expert software.

dsRNA ELISA and 5’ capping evaluation

dsRNA was evaluated using SCICONS J2-based anti-dsRNA ELISA kit (exalpha) using the standard protocol, and absorbance at 450 nm was measured using a UV-Vis Spectrophotometer. poly(I:C) was used for the standard curve. Quantitation of 5’-Capping efficiency was determined by first annealing the produced mRNA to a biotinylated probe designed to complement the 5’ end. Then, treating the material with RNase H, the resulting duplex is separated from the larger RNA remnant using biotin-streptavidin affinity purification. The 5’ end related species are then chromatographically separated by reversed-phase high-performance liquid chromatography (RP-HPLC) using an Agilent 1260 system with quaternary pumps and DAD detector—while monitoring UV absorbance at 260 nm. An ACQUITY UPLC Oligonucleotide BEH C18 Column (130 Å, 1.7 µm, 2.1 mm × 150 mm, 1 K–30 K) (Waters) held at 75 °C was used. Relative quantification of the 5’ end cap level (5’-Cap) was determined from relative peak areas. The percentage of species that are capped was determined by dividing the percent area for the capped species by the total area of all 5’ end-related species.

Illumina NGS RNAseq

Illumina libraries for the IVT transcripts were generated using Illumina’s TruSeq Stranded mRNA kit, following the manufacturer’s protocol, with the exception that poly-T oligo attached magnetic beads were not used. This modification allows for the detection of RNA molecules without poly-A tails. The library preparation entails RNA that is randomly fragmented into short fragments using divalent cations under high temperature, followed by cDNA generation using random primers. The newly synthesized cDNA is adenylated at its 3’ ends, then ligated to sequencing-indexed adapters. The ligated molecules are then enriched by limited-cycle PCR amplification. The prepared libraries were quantified by qPCR using an NEBNext Library Quant Kit for Illumina on a QuantStudio 5 Real-Time PCR system, and the fragment size distribution was determined by TapeStation using a D1000 ScreenTape assay. High-throughput Illumina sequencing was performed using a 76-bp paired-end format on an Illumina MiSeq instrument. The demultiplexed sequencing reads were quality trimmed, and reads passing filters were mapped to the reference sequences using a custom bioinformatics pipeline.

Analysis of cellular expression

Capped mRNA containing the fLuc CDS was generated through both the USTAT process and pDNA templated IVT. A NanoLuc® Luciferase vector (Promega) was also transcribed, and mRNA was used as a normalization control. mRNA from either USTAT or pDNA templated IVT was co-transfected with NanoLuc mRNA in HEK293T cells using Lipofectamine 3000 Transfection Reagent following standard protocols. Luminescence was measured at multiple time points post-transfection using the GloMax Navigator luminometer (Promega), and fLuc signal was normalized to NanoLuc signal.

Author contributions

S.G, C.L.S., S.D.R., C.H.C., J.W.S., K.N., M.J.S., and J.R. conceptualized ideas and experiments presented in this work. S.G, C.L.S., S.D.R., C.H.C., J.W.S., K.N., M.J.S., and J.R. performed experiments presented or related to this work to develop conclusions. S.G, C.L.S., S.D.R., C.H.C., J.W.S., K.N., M.J.S., and J.R. wrote the main manuscript. S.G, C.L.S., S.D.R., C.H.C., J.W.S., K.N., M.J.S., and J.R. prepared figures. S.G, C.L.S., S.D.R., C.H.C., J.W.S., K.N., M.J.S., and J.R. reviewed the manuscript.

Data availability

The data supporting the findings are contained within the manuscript or supplementary files. Genetic data is present in the supplementary data file. Additional raw data are available from the corresponding author upon request.

Competing interests

The authors declare no competing interests.