Measuring Live-Cell mRNA Translational Dynamics with Split Luminescent Tagging in HEK293 Cells

Summary

Here, we present a flexible, high-throughput-compatible method for measuring live mRNA translation dynamics in cellulo in HEK293s, using split luminescent tagging. This versatile assay distinguishes structural effects, supports in vitro transcription optimization, and streamlines mRNA therapeutic development.

Abstract

The swift vaccine development to combat COVID-19 illustrated the potential for messenger RNA (mRNA) therapeutics to transform drug development. Like mature mRNA, in vitro transcribed mRNA possesses the same elements including a 5’ cap, untranslated regions (UTRs), coding sequence and a poly(A) tail. Previous work studying the effects these components have on mRNA translation has primarily utilized highly engineered reporter proteins which exhibit efficient translation and protein stability. With the structural elements of each mRNA differentially affecting their translation, it is imperative to identify the optimal design relevant to the therapeutic protein of interest (POI). To enable POI translation characterization, a split luciferase complementation system was employed. A short peptide tag (HiBiT), which can be fused to either terminus of the POI, associates with its complementary heterodimer (LgBiT) to reconstitute enzymatic activity in the presence of a cell-permeable substrate. To date, split luminescent tagging has been primarily used for high-throughput protein turnover studies. We have previously demonstrated how split luminescent tagging can be employed to enable high-throughput quantification of mRNA translation temporally in cellulo in HEK293 cells constitutively expressing the complementary heterodimer. Here, we further demonstrate the versatility of the assay and detail how this assay can be employed for optimizing in vitro transcription to reduce costs. The assay system can uniquely distinguish alterations in structural components whilst highlighting the effects of coding sequence optimization using non-engineered genes. Additionally, we demonstrate that a 4-fold reduction in 5’ cap concentration for in vitro transcription results in equivalent translation in cellulo. These findings illustrate how split luminescent tagging can be easily integrated into the mRNA therapeutic workflow, enabling monitoring of real-time mRNA-driven protein expression dynamics in cellulo thereby offering a versatile method for the advancement of mRNA-based therapeutics.

Introduction

Messenger RNA (mRNA) therapeutics and research applications have rapidly expanded in recent years, driven by the success of mRNA vaccines^1,2 and the growing recognition of RNA’s versatility as a programmable biomolecule. A central challenge in this field is the ability to quantitatively and dynamically assess mRNA translation efficiency in diverse contexts. The method described here – leveraging a split luminescent tagging system – provides a flexible, scalable platform for studying live mRNA translation in cellulo³. The overall goal of this method is to enable researchers to measure how sequence features, chemical modifications, and structural elements influence translation kinetics and subsequent protein turnover, thereby offering a powerful tool for optimizing mRNA constructs for therapeutic and basic science applications.

In vitro transcribed mRNA mirrors the essential architecture of mature transcripts, incorporating a 5′ cap, untranslated regions (UTRs), a coding sequence, and a poly(A) tail, while often including modified nucleotides to reduce immunogenicity^4,5. Extensive work has shown that each of these elements – alone and in combination – can be engineered to strongly influence translation efficiency and protein yield^6,7. Yet most prior studies have relied on highly optimized reporter proteins such as Green Fluorescent Protein (GFP)⁷, NanoLuc⁸, or the SARS-CoV-2 spike protein^8,9, which are engineered to translate with high efficiency and to produce very stable proteins, and therefore do not reflect the challenges faced in therapeutic contexts. Because therapeutic mRNAs encode diverse proteins of interest (POIs), each subject to unique structural constraints and cell-type-specific regulation, the design principles that maximize translation for one construct may not generalize to another, as previously reported³. Identifying the optimal configuration for a given therapeutic POI is therefore essential and requires methods that can sensitively and systematically evaluate how individual features of the transcript contribute to protein output.

Ribosome-focussed methods such as polysome profiling distinguish translation fractions through gradient centrifugation¹⁰, while ribosome profiling (Ribo-seq) provides nucleotide-level resolution of ribosome occupancy, initiation sites, and pausing¹¹. Like Ribo-seq, other sequencing-based approaches like High-Throughput Sequencing of RNA Isolated by Crosslinking Immunoprecipitation (HITS-CLIP) and Translating Ribosome Affinity Purification (TRAP) further characterize translation of RNA-binding protein–associated or ribosome-associated RNAs¹². Although powerful, these techniques require complex preparation, specialized equipment, and computational expertise, making them unsuitable for high-throughput therapeutic mRNA development. Nascent chain immunoprecipitation¹³ measures translation efficiency using FLAG-tagged proteins immunoprecipitated from ribosomes and quantified by Quantitative Reverse Transcription Polymerase Chain Reaction (qRT-PCR). While adaptable to therapeutic mRNAs, the assay is labor-intensive, lacks temporal resolution, and is unsuitable for high-throughput applications. The SunTag system¹⁴ offers live-cell measurements of protein synthesis output and translational regulation from single mRNA molecules and works through the interaction of a single-chain variable fragment (scFv) fused to GFP with tandem repeats of the peptide epitope. This system enables dynamic measurements of translation and localization with single-mRNA resolution, and it has therefore been exploited to develop numerous nascent chain tracking systems^15,16. However, this approach is low throughput and requires the use of high-power microscopes.

While tagging of the nascent protein is unavoidable, the potential use of a small peptide tag was identified, derived from NanoLuc luciferase, to report on mRNA translation in live cells. The NanoLuciferase was engineered to be partitioned into a small, 11-amino acid (HiBiT) and a larger, 158-amino acid (LgBiT) subunit, which can rapidly dimerize and, in the presence of substrate, reconstitute a functional luciferase^17,18. There are several key advantages of the split luminescent tagging system for studies of mRNA dynamics. Firstly, the minimal size of the small subunit fused to the POI allows the encoded construct to remain as close as possible to the native protein of interest. As coding sequence length can directly affect translation ratios¹⁹, minimizing the tag length improves our prediction of translation dynamics of synthetic mRNAs. Crucially, this minimal tagging strategy enables the detection of differences in translation driven by synonymous coding sequence variation, allowing translational efficiency to be interrogated independently of protein sequence or stability³. For this reason, the use of larger tags, such as fluorescent proteins or full-length luciferases, is suboptimal. Secondly, the split luminescent tagging system employed here enables immediate signal generation. Its rapid and efficient in cellulo complementation reaction has a reported high affinity (K_D = 700 pM) and has been used to measure protein turnover modulation with readings as short as every 30 s¹⁸. While the original split luminescent tagging system primarily benchmarked against other split luciferase systems^17,18, the key advantage in this context is that signal generation is not limited by protein maturation. In contrast, fluorescent proteins require chromophore maturation, which can occur in the scale of minutes to hours²⁰ thereby limiting temporal resolution. Finally, the established split luminescent tagging system here employed overcomes limitations of split fluorescent protein systems, which typically require chromophore formation after complementation²¹, introducing an additional delay and limiting their suitability for capturing rapid, dynamic processes. Previous studies have demonstrated the utility of split luminescent tagging in protein turnover assays and degradation studies^22,23,24. Extending this established technology to translation kinetics represents a significant methodological advance. For example, while ribosome profiling can reveal ribosome density along transcripts, it cannot easily capture dynamic changes in translation initiation or elongation rates under different conditions. As previously illustrated³, this split luminescent tagging system can detect differences when altering cap structure, 5′ untranslated region (UTR) composition, codon optimization, nucleotide modifications, and poly(A) tail length, all of which are critical determinants of translation efficiency. This breadth of applicability makes the method uniquely powerful for dissecting the contributions of individual mRNA features to overall protein output. It is worth noting that, depending on the instrument used, both single-cell and whole-well data can be captured with this technology. Imaging platforms capable of acquiring luminescence-based images can empower this technology to be used to study single-cell translation dynamics. Here, we focus on a plate-reader, well-based application due to its medium-to-high-throughput nature and minimal requirement for data analysis.

Researchers considering this method should evaluate its suitability in light of their specific experimental objectives. The assay is particularly valuable in contexts where high-throughput screening of synthetic mRNA constructs is required, such as in mRNA-based therapeutic design pipelines. It is well-suited to comparative analyses of synthetic mRNA sequence features – including untranslated regions, codon usage, or nucleotide modifications – where subtle differences in translation efficiency can be detected³. Because the system is compatible with cell-free translation platforms, live-cell assays, and a timepoint-specific lytic assay, it offers flexibility for investigators who wish to tailor their experimental design to either mechanistic studies or applied research. At the same time, it is important to recognize the boundaries of the technique. Critically, the split luminescent tagging assay does not provide resolution of ribosome occupancy, a capability that remains unique to ribosome footprinting approaches, nor does it directly measure RNA stability or degradation, which may need to be assessed using complementary methods. However, due to the limitations of existing ribo-centric techniques, researchers are reliant on tag-based assays like the split luminescence system to investigate time-resolved mRNA-driven protein expression as a proxy for translational dynamics. Thus, while this system is a powerful tool for monitoring synthetic mRNA-driven protein expression, its appropriateness depends on whether the research question focuses on dynamic protein output rather than fine-grained ribosome positioning or RNA decay.

In summary, the split luminescent tagging system represents a powerful and versatile method for studying mRNA translation. Its overall goal is to provide researchers with a flexible, system-scalable, high-throughput, and capable of detecting differences in protein output arising from synthetic mRNA with varying structural and sequence elements³. The rationale for its development lies in overcoming the limitations of existing techniques, offering advantages in sensitivity, scalability, and temporal resolution. Positioned within the wider literature on RNA biology and therapeutics, this method provides a valuable addition to the experimental repertoire, enabling both fundamental insights and translational advances. By carefully considering the strengths and limitations of the assay, researchers can determine its appropriateness for their specific applications, thereby maximizing its impact in the rapidly evolving field of RNA science.

Protocol

A graphical depiction of the entire protocol is shown in Figure 1.

1. Template preparation for in vitro transcription (IVT)

NOTE: To prepare the DNA template for IVT, ensure the plasmid DNA contains an RNA polymerase promoter (generally T7 or SP6), 5’ and 3’ UTRs, and the open reading frame of the POI (Figure 1).

1. Linearize plasmid DNA using a restriction enzyme with a single-cutting site located after the 3’ UTR and assemble the reaction according to the manufacturer’s instructions. After incubation, clean up the restriction digest reaction using a spin column-based DNA purification kit. Quantify the DNA using a spectrophotometer.
2. To confirm linearization, prepare a 1% agarose TAE gel containing DNA stain Ethidium Bromide at 0.1 µg/mL. Mix 150 ng of digested DNA with water and loading dye. Load the agarose gel with the sample, uncut plasmid, and a DNA ladder, and run the agarose gel at 100 V for 40 min in a 1× Tris-acetic acid-EDTA (TAE) buffer. Visualize with a gel imager.
   CAUTION: Ethidium bromide solution is a potent mutagen and requires careful handling using appropriate personal protective equipment (PPE).
3. If the plasmid DNA does not have a poly(A) tail immediately after the 3’ UTR, introduce it using a PCR. Complete PCR according to the manufacturer’s instructions using a high-fidelity polymerase with appropriately designed primers. Ensure the amplicon includes the T7 promoter, open reading frame, and a poly(A) tail.
   NOTE: Capping technologies can have specific sequence requirements at the T7 promoter; therefore, these requirements can be satisfied by appropriate forward primer design. The reverse primer binds the end of the 3’ UTR and introduces a poly(A) tail (recommended length of at least 80 T). For high GC-containing constructs, step-up PCR and/or 3% DMSO are recommended.
4. To confirm successful PCR, prepare an agarose TAE gel containing Ethidium Bromide (see step 1.2). If the construct length is ≥1 kb or <1 kb, prepare a 1% or 2% gel, respectively. Mix 5 µL of the PCR reaction with water and loading dye. Load the agarose gel with both the sample and a DNA ladder, and run it at 100 V for 40 min in TAE buffer. Visualize with a gel imager.
5. After PCR size confirmation, clean up the PCR reaction using a spin column-based DNA purification kit. Elute in 15–20 µL of elution buffer (often provided with the DNA purification kit) or nuclease-free water to concentrate the DNA template for use in IVT. Quantify the DNA using a spectrophotometer.

2. In vitro transcription

1. Before starting, spray and wipe down the bench, gloves, and pipettes with an RNase decontamination solution. To improve yield and quality, perform this procedure in a sterile hood environment.
2. Thaw and assemble the components of a T7-based in vitro transcription (IVT) kit according to the manufacturer’s instructions.
   NOTE: Some manufacturers recommend the use of dithiothreitol (DTT, 0.1 M) to limit reaction oxidation and improve efficiency. This step is recommended when compatible with the kit. If using modified dNTPs, replace the equivalent unmodified nucleotide with the same amount of modified nucleotide. Evaluate the yield from a half reaction, as no substantial improvement in yield was observed from a full reaction. When synthesizing more than one mRNA, it is recommended to create a master mix to reduce pipetting errors and variability.
3. Pipette thoroughly and briefly mix, then incubate at 37 °C for 5 h with a heated lid to avoid condensation. Extend incubation to overnight if needed, as this has minimal effect on mRNA yield or integrity.
4. Add 1 µL of DNase I and pipette thoroughly. Incubate at 37 °C for 30 min.
5. Purification of in vitro transcribed (IVT) RNA
   1. Clean up the IVT reaction using a column-based RNA Cleanup kit as per the manufacturer’s instructions (different kits may lead to different levels of recovery and have varying binding capacities). For eluting, add 60 µL of nuclease-free water and incubate at room temperature for 5 min before centrifuging.
   2. Alternatively, use lithium chloride precipitation as a cheaper and more flexible approach.
      1. Add an equal volume of 7.5 M lithium chloride to the reaction, mix, and incubate at -20 °C overnight.
      2. Centrifuge at 18,300 × g for 20 min at 4 °C, then carefully remove supernatant without disrupting the RNA pellet.
      3. Wash with 950 µL of 70% ethanol and centrifuge at 18,300 × g for 5 min at 4 °C. Remove the supernatant, let the pellet dry for approximately 5 min at room temperature, and resuspend in nuclease-free water (60 µL is standard; reduce the volume depending on pellet size to increase concentration).
         CAUTION: Lithium Chloride may cause skin or eye irritation if proper PPE is not used. Wear gloves and eye protection when handling. And, ethanol is a highly flammable liquid and can cause eye irritation. Keep away from naked flames, and wear appropriate PPE, including eye protection.
6. Quantify the mRNA using a spectrophotometer. Ensure that the A260/A280 ratio for mRNA is ~2.0, and confirm that the A260/A230 ratio falls between 1.8 and 2.2.
7. To confirm successful IVT, prepare a 1% agarose TAE (non-denaturing) gel containing Ethidium Bromide. Mix 500 ng of mRNA with water and loading dye, and follow the manufacturer’s instructions for RNA ladder preparation and denaturing instructions. Load the agarose gel and run it at 80 V for 40 min in TAE buffer. Visualize with a gel imager.
8. For further confirmation of correct mRNA size, run the samples on a Bioanalyser.
9. To avoid freeze-thawing, aliquot mRNA into the volume required for the experiment. For example, use 1.2 pmol of mRNA per well in a 96-well plate. Therefore, aliquot the appropriate volume for 1.2 pmol × n replicates, then store at -80 °C.

3. HEK293 reverse transfection

NOTE: To achieve temporal measurements, a cell line constitutively expressing the complementary luciferase fragment is required. This can be achieved by creating a stable cell line through transfection of a plasmid expressing the required luciferase fragment and subsequent clonal selection. This allows the user to measure temporally resolved mRNA translational dynamics in any cellular background of choice. Alternatively, cell lines expressing the required luciferase fragment can be purchased. These experiments were conducted using HEK293 cells constitutively expressing the complementary heterodimer.

1. Before starting, thaw the mRNA on ice to avoid degradation.
2. Prepare transfection complexes. For a 96-well plate, 1.2 pmol of mRNA is required per well. If running in duplicates, make enough for 2.5 duplicates to allow for pipetting error. Therefore, combine 3 pmol of mRNA with Opti-MEM Reduced Serum Medium to a final volume of 50 µL (20 µL/well).
3. Add 0.5 µL of cationic lipid-based transfection reagent (0.2 µL/well), mix by pipetting up and down, and incubate at room temperature. After incubating for 5 min, mix by pipetting up and down, then add 20 µL per well to a 96-well white flat-bottom microplate. It is suggested that the transfection wells are placed in the center of the plate, leaving the edge wells clear.
4. Prepare HEK293 cells by removing media and washing with 2 mL of phosphate-buffered saline (PBS) without Ca²⁺ and Mg²⁺. Remove the wash and add 500 µL of Trypsin-EDTA, then incubate at 37 °C for 3 min.
5. During cell dissociation, fill the edge wells of the 96-well plate with PBS to decrease media evaporation during cell culture. If all wells are required, the plate can be flooded with PBS in between the wells.
6. Add 4.5 mL of Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 1× L-glutamine and 10% fetal bovine serum (FBS) to collect cells. Perform cell count using a hemocytometer or automatic cell counter.
7. Transfer the required amount of cells (enough for 1 x 10⁵ cells to be seeded per well) plus 3 extra wells (for pipetting error) to a fresh tube and centrifuge at 500 × g for 2 min.
8. Remove supernatant and resuspend cells in the required volume of Leibovitz’s L-15 medium (phenol free) supplemented with 10% FBS and Endurazine Live Cell Substrate (1:200) to ensure a final concentration of 1 × 10⁶ cells/mL. Mix with a serological pipette to ensure thorough mixing of Endurazine. Pipette 100 µL (1 × 10⁵ cells) of resuspended cells per well.
   NOTE: If performing time course experiments running for 72 h, increase the volume to 200 µL to avoid evaporation.
9. Immediately post-transfection, place cells into a plate reader (luminometer) set at 37 °C. Take recordings at regular intervals with an integration time of 1 s. Due to the system's efficient complementation, the signal is detectable within minutes. Therefore, to get an accurate baseline reading, this step has to be executed promptly. These intervals are experiment- and instrument-dependent; however, in these experiments, measuring every 12 min for 18 h time course experiments, but only every 48 min for 72 h experiments, was found to be optimal. Luminescence does not require a filter for measurement; a better signal-to-noise ratio was observed when using a filter of 530–540 nm.
   NOTE: A Bioluminescence Resonance Energy Transfer (BRET) signal is produced from split-luciferase-tagged eGFP thus a filter at 530–540 nm will capture both a luminescent and fluorescent signal. To avoid capture of BRET, use a filter at 460–480 nm or alternatively, use a non-fluorescent protein such as MYL3 as a reference.

4. Data analysis

1. Generate fold changes in relative luminescence units (RLUs), by normalizing each condition’s signal to its first time point (t0, normalized to the average of the technical replicates), since the signal at the beginning of the assay is the background level.
2. Reduce variability between biological replicates by normalizing to a control condition.
3. Choose the statistical analysis for the assay, either single time point or area under the curve (AUC)—depending on the question being asked.
   1. For a single time point, use the time points where peak expression occurs (which is POI-dependent) along with the endpoint time. Plot the RLU calculated in step 4.1 at the specific time point (e.g., 240 min) and analyze it using an appropriate statistical test.
   2. Measure dynamics of protein translation and degradation by calculating the AUC for either a specific timeframe (i.e., t0 to 240 min) or for the length of the assay (i.e., t0 to endpoint). Normalize the AUC value for each biological replicate to the average of the reference mRNA, thereby yielding a measure of fold change in AUC that can be analyzed using appropriate statistical tests.

Results

There are several ways in which the cost-effectiveness of this protocol can be optimized. For example, the IVT reaction was miniaturized to half the volume recommended by the manufacturer, and it was found that the resulting yield satisfied the amount required for in cellulo experiments. It should be noted that sequence length and complexity impact IVT yield, so successful protocol optimization should yield at least 50 µg of mRNA per IVT reaction. Like IVT reaction kits, capping reagents also often cost a great deal. Considering only one cap is required per mRNA molecule, we hypothesize that the IVT reaction contains a surplus of cap molecules to the resulting number of mRNA molecules. Thus, it was investigated whether the protocol could be further optimized by titrating the amount of capping reagent added to the IVT reaction, the results of which are shown in Figure 2. 100% (10 mM cap final), 50%, 25% or 0% (uncapped) of the capping reagent was added to 10 µL IVT reactions and the resulting eGFP mRNA was transfected into HEK293 cells constitutively expressing the complementary heterodimer. As a normalization control, MYL3 (100% capped) was also transfected as this endogenous gene, encoding for a muscle-enriched structural protein, exhibits similar translation dynamics to eGFP, however, with a signal closer to that of other endogenous genes³. Consistent with previous data³, uncapped eGFP mRNA resulted in a lack of translation analogous to mock control (Figure 2A). eGFP mRNA synthesized with 25% or 50% of the recommended capping reagent showed comparable levels of expression with 100% (Figure 2A), resulting in no differences in area under the curve (AUC) values over the 18 h time course (Figure 2B). These results indicate that, under these conditions, the capping reagent is present in excess and that reducing its concentration does not impair translational output, consistent with sufficient cap incorporation. The findings presented in Figure 2 are specific to the brand of capping reagent used, and should be optimized for different reagents.

While advances in cap technology and the incorporation of modified nucleotides have enhanced mRNA expression, the development of mRNA design algorithms has further improved expression. These algorithms focus on optimizing mRNA stability and codon usage by improving the minimum free energy (MFE) and codon adaptability index (CAI), respectively. To test the effect of optimization on mRNA expression, LinearDesign⁹ was utilized. This algorithm relies on the input of a lambda value, which is a scaling parameter used in optimization formulae that determines how strongly an algorithm favors MFE or CAI. The same optimization settings were applied (i.e., the same lambda value) to four transcription factors – HAND2, GATA4, MEF2C, and TBX5. Figure 3 explores the effects of optimization on translation rates. Optimization improved values for both MFE and CAI for all transcription factors, with MEF2C exhibiting the greatest changes in both parameters (Figure 3A). A split luminescent tag was fused to the C-terminus of each construct, preceded by a GS linker, and mRNA was synthesized for all four transcription factors for both unoptimized and optimized forms. To ensure that each mRNA can be translated and produce a protein of the expected molecular weight, the rabbit reticulocyte lysate cell-free translation system³ was employed. Resulting proteins were resolved by western blot employing the enzymatic subunit of the split luciferase protein and appropriate substrate to detect the split luminescent tag. Western blotting showed that optimization still produced proteins that matched their expected molecular weight (Figure 3B). HEK293 cells were transfected with the unoptimized or optimized forms of each transcription factor, and live cell recordings were taken for 18 h. For all 18 h experiments, measurements from three biological replicates were taken, and three technical replicates per sample were used in each repeat experiment. Optimization of HAND2 (Figure 3C) and GATA4 (Figure 3D) had no effect on expression (as determined by an unpaired two-sample t-test), with a 1.07-fold increase and 1.25-fold decrease, respectively, when compared to their unoptimized counterparts. Meanwhile, optimization of MEF2C (Figure 3E) and TBX5 (Figure 3F) led to a statistically significant increase in expression with a 2.25 (P = 0.0486, n = 3, unpaired two-sample t-test) and 1.72-fold (P = 0.0379, n = 3, unpaired two-sample t-test) elevation in AUC over their respective unoptimized version. Consistent with this, differences are observed not only in endpoint expression but also in the shape and rate of the expression curves, indicating that the assay resolves changes in translational dynamics over time. To determine whether this increase in expression is maintained over a longer period, HEK293 cells were transfected with MEF2C mRNA, and recordings were taken for 72 h (Figure 3G). As with the 18 h experiments, measurements from three biological replicates were taken, and three technical replicates per sample were used in each repeat experiment. Similar to what was observed after 18 h, optimized MEF2C significantly increased expression with a 2.5-fold (P = 0.0061, n = 3, unpaired two-sample t-test) improvement in AUC compared to the unoptimized version. Although there is a substantial increase in expression with optimization, these results demonstrate that optimization does not, in this case, prevent decay over 72 h, with both MEF2C forms exhibiting similar slopes from 24 h onwards. This decay is likely due to linear mRNA degradation and/or cell division. Together, these results demonstrate the potential of mRNA design algorithms to improve mRNA expression; however, they highlight that optimization needs to be undertaken in a POI-dependent manner.

Figure 1: Workflow for mRNA synthesis and split luminescent tag assay in HEK293 cells. To undertake IVT for the split luminescent tag assay, ensure the plasmid contains an RNA polymerase promoter (T7 or SP6), 5’ and 3’ UTRs, and the coding sequences for the protein of interest fused to a short luciferase subunit HB (HiBiT). The plasmid is linearized with a single site-cutting restriction enzyme, providing the template for the Poly T PCR, which is subsequently used as the template for IVT. mRNA is synthesized, complexed with liposomes, and reverse-transfected into HEK293 cells constitutively expressing the larger luciferase subunit (LB), which can be immediately placed into a plate reader where live recordings are measured. After experiment completion, data analysis is performed for normalization and subsequent statistical analysis. Please click here to view a larger version of this figure.

Figure 2: A four-fold reduction in capping reagent concentration produces comparable expression of eGFP with full concentration. (A) Live cellular kinetic assay in HEK293 cells transfected with eGFP mRNA with varying concentrations of capping reagent, MYL3, or Mock, data normalized to t₀; dots represent the mean of at least three biological replicates per timepoint. (B) Fold change in AUC of curves for an 18 h time course of at least three biological replicates was obtained by normalizing to the reference sample MYL3 and compared by one-way ANOVA with Tukey’s multiple comparisons test. Data is shown as mean ± standard error of the Mean (SEM), with each dot representing the mean of a biological replicate. ***P < 0.001. (eGFP – enhanced green fluorescent protein; AUC – Area under the curve; RLU-relative luminescence units) Please click here to view a larger version of this figure.

Figure 3: Sequence optimization improves mRNA translation in a POI-dependent manner. (A) LinearDesign optimization of mRNA sequence resulted in a reduction and an increase of minimum free energy (MFE) and codon adaptability index (CAI) values, respectively, for each transcription factor. (B) RRL reactions were carried out for 120 min with mRNA encoding unoptimized and optimized HAND2, GATA4, MEF2C, TBX5, or no RNA. Resulting proteins were processed for western blotting and detected by chemiluminescence via incubation of the membrane with the enzymatic subunit in the split luciferase system and appropriate substrate. Live cellular kinetic assay in HEK293 cells transfected with unoptimized or optimized HAND2 (C), GATA4 (D), MEF2C (E), or TBX5 (F), along with MYL3, and Mock. Data were normalized to t₀, and dots represent the mean of three biological replicates per time point. Fold change in AUC of curves for an 18 h time course of three biological replicates was obtained by normalizing to the reference sample MYL3 and compared by an unpaired t-test with Welch’s correction. Data is shown as mean ± SEM, with each dot representing the mean of a biological replicate. *P < 0.05. (G) Live cellular kinetic assay in HEK293 cells transfected with unoptimized or optimized MEF2C, MYL3, and Mock data normalized to t₀, dots represent the mean of three biological replicates per timepoint. Fold change in AUC of curves for a 72 h time course of three biological replicates was obtained by normalizing to the reference sample MYL3 and compared by an unpaired t-test with Welch’s correction. Data is shown as mean ± SEM with each dot representing the mean of a biological replicate. **P < 0.01. (MFE – minimum free energy; CAI – codon adaptation index; RRL – rabbit reticulocyte lysate; AUC – area under the curve, RLU – relative luminescence units) Please click here to view a larger version of this figure.

Discussion

The field of mRNA therapeutics has expanded rapidly in recent years, underscoring the importance of sequence design and optimization as a critical stage in the development pipeline. Significant efforts have focused on building computational tools to predict the most effective RNA structures and compositions, with sequence optimization receiving particular attention^8,9. While these tools are powerful, experimental validation of these designs is still an essential part of the pipeline. However, this validation has largely relied on reporter proteins, including GFP⁷, NanoLuc⁸, and the SARS-CoV-2 spike protein⁹ which are already extremely stable, highly translatable proteins. To overcome this limitation, a split luciferase system was incorporated into a pipeline to study the translation dynamics of in vitro transcribed mRNA in HEK293 cells, enabling temporal resolution³, and demonstrated the limitations of using reporter proteins in detecting changes in translation upon mRNA engineering. This method's article demonstrates the steps required to perform this assay, illustrates that the IVT reaction can be miniaturized, and the 5’ cap concentration can be reduced without translational consequences, thereby reducing costs. Building on our previous observations, we highlight that a POI-dependent approach to mRNA sequence optimization is required to achieve increased expression.

As described here, the split luminescent tag assay is a relatively straightforward pipeline enabling medium-throughput measurement of mRNA translation, allowing users to investigate temporal expression of their POI in real time. However, there are multiple critical steps and considerations required to achieve robust results. Firstly, all downstream results are dependent on the quality of the mRNA, which begins with the IVT and the decontamination of the equipment used for this. After synthesis, the storage of the mRNA is key whereby it is kept on ice post-elution and following confirmation of mRNA size, the mRNA is aliquoted into working volumes to prevent freeze-thawing, and the mRNA remains stored at -80°C. Secondly, the 96-well plate used is of utmost importance and must be consistent between experiments. For this protocol, ensure an all-white 96-well plate is used. Clear-bottom, white-walled plates could be used if monitoring of cell morphology is required; these result in a much lower signal, so the two plate types cannot be used interchangeably. It was observed that when measuring the translation of the same mRNAs in a clear-bottom, white-walled plate, the resulting luminescent signal was 5-fold lower than that of a parallel fully white-walled plate. This technical note can be impactful when assessing genes with inefficient translational dynamics. Aside from introducing inconsistency into the data, the reduction in signal may prevent detection of subtle differences induced by mRNA sequence or structural perturbations. Thirdly, the use of L-15 media is key to this assay as it enables cell viability in an environment without CO₂ supplementation²⁵, which provides flexibility with the luminometer used. Additionally, the use of phenol-free L-15 reduces background signal while measuring luminescence. Lastly, culturing and measuring the luminescence at 37°C (or the culture condition of your cell type of interest). Performing this experiment at room temperature will greatly alter the results, as the translation elongation rate is reduced²⁶ meaning normal kinetics will not be observed.

The versatility of in vitro transcribed mRNA design can be transferred to the synthesis process, which allows numerous modifications. As observed, the concentration of 5’ cap can be reduced by 4-fold, yet still achieve comparable expression, enabling researchers to decrease costs. While capping efficiency was not directly measured post-IVT, an increasing proportion of uncapped mRNA within the IVT pool would result in a decreased signal, as demonstrated by uncapped mRNA; this was not observed. Additionally, considering eGFP is a relatively small protein at only 27 kDa, this modulation of cap concentration should readily translate to bigger genes, as there will be fewer transcripts to cap post-IVT. There are multiple cap analogs that possess different methylation modifications, impacting immunogenicity³ and protein expression²⁷. Additionally, the capping process can be performed co-transcriptionally (as demonstrated here) or post-transcriptionally using Vaccinia virus, which results in 100% capping efficiency²⁸ but requires extra purification steps. For lab-scale production of mRNA, purification columns are commonly used as they are quick and efficient, but costly. A cheaper and more versatile approach is lithium chloride precipitation, which, although more time-consuming, allows the user to concentrate the mRNA and does not restrict the yield to column binding capacity; therefore, it is suitable for scale-up. Like the purification columns, this approach does not completely remove double-stranded RNA and truncated RNA fragments, which cause immunogenicity²⁹. Similar to the 5’ cap, the incorporation of the poly(A) tail can be achieved in alternative ways to what was described here. Analogous to the PCR introduction of the tail, plasmid DNA can be engineered to contain a poly(A) tail; however, these long stretches can undergo recombination, resulting in heterogeneous and shortened poly(A) sequences. However, poly(A) segmentation – the insertion of heteronucleotide spacers – has minimized these effects^30,31. Alternatively, the poly(A) tail can be generated using a template-independent poly(A) polymerase, which enables the generation of longer poly(A) tails³² but at the risk of heterogeneity impacting regulatory requirements³³.

While the split luminescent tag assay described is extremely effective in measuring protein expression of exogenously delivered mRNA, there are some limitations that are to be considered depending on the research question. Firstly, there is a need to make a cell line constitutively expressing one half of the split luciferase. Although this can be purchased ready-made, it is currently limited to only HEK293 cells. If the creation of a stable line in the cell type of interest is challenging, a lytic assay exists that we have found illustrates the same outcome between HEK293s, human embryonic stem cells, and differentiated cardiomyocytes³. However, the time point chosen for the lytic assay from this previously published study³ was guided by the live cell HEK293 data. Secondly, the assay provides live-cell temporal resolution but is unable to offer live-cell spatial resolution, which may be of interest to certain researchers. However, there is a primary antibody available that works well for fluorescent immunocytochemistry against the short luciferase peptide tag used in these experiments, enabling spatial interrogation of the translated POI. Additionally, a microscope capable of luminescent imaging could provide single-cell resolution data. An approach offering spatiotemporal information is SunTag, which measures active translation and localization using live cell imaging through interactions between GFP-tagged scFv binding to peptide epitopes^14,34. However, 24 tandem repeats of these epitopes are required, thereby potentially distorting translation, while the workflow is relatively low-throughput. Thirdly, the split luminescent tag enables the measurement of mRNA abundance, but not in live cells. In our previous work³, qRT-PCR was performed on extracted RNA using primers specific to the exogenously delivered transcript, which confirmed mRNA levels for transfection efficiency between constructs and decay between timepoints. Another approach has seen the use of Northern blotting to measure mRNA stability³⁵. However, both approaches are laborious and are unable to provide continuous measurements. Currently, no method exists that enables simultaneous measurements of mRNA translation and abundance in live cells, providing a potential avenue for advancing assay development in this space.

The split luciferase system was initially applied to study protein dynamics, yet it is apparent here that this assay has major implications in other research areas. With mRNA therapeutics at the forefront for new gene therapy approaches, this assay is perfectly suited to allow researchers to characterize the expression of their potential therapeutic and optimize it for enhanced expression. As mentioned earlier, codon optimization has become a powerful new tool in mRNA design with the development of multiple tools^8,9,36. When a uniform sequence optimization approach was applied to four transcription factors, it improved both MFE and CAI for all genes, yet only MEF2C and TBX5 exhibited increased expression. Notably, by enabling the detection of expression differences arising solely from synonymous coding sequence variation, this approach provides a unique platform for interrogating translation efficiency in live cells with high temporal resolution. These findings outline that sequence optimization must consider starting GC content, codon usage, and RNA secondary structure in a POI-dependent manner to achieve enhanced expression. While more constructs may need to be tested, the split luminescent tag assay has the potential for high-throughput analysis to identify the optimal candidate for the POI. Additionally, as shown previously³, this assay can measure the effects of perturbing any sequence or structural component of mRNA further outlining its applications. Aside from investigating alterations to the mRNA, the split luminescent tag assay could also be utilized in delivery and formulation research. Even with improvements in 5’ caps and the use of modified nucleotides, exogenously delivered mRNA requires a packaging system to prevent degradation and cellular uptake³⁷. While lipid nanoparticles³⁸, extracellular vesicles³⁹, and biomimetics⁴⁰ are at the forefront of mRNA delivery, further development is still required. Encapsulation of split luminescent-tagged mRNA in any of these carriers would enable characterization of their delivery efficiency and enable researchers to test modified carriers in a medium-to-high-throughput manner. Outside of translationally focused research, the split luminescent tag assay is readily usable for fundamental biology. Whether this is assessing the effects of sequence and structural element perturbations on mRNA translation, interfering with the function of translational machinery, or probing the proteasome, the split luminescent tagging system provides an attractive option.

Overall, the split luminescent tag system described here offers a flexible, robust assay to monitor synthetic mRNA-driven protein expression temporally as a readout for translational dynamics. Implementing a live cell assay overcomes many of the previous limitations in the mRNA translation field and uncovers crucial information on the dynamics of protein synthesis and POI degradation. As transcripts possess different regulatory and expression profiles, it is imperative that the therapeutic POI is characterized which will allow mRNA therapeutics to continue to evolve and translate to the clinic.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
Acetic Acid	Sigma-Aldrich	A6283	Component of TAE buffer
Agarose	Sigma-Aldrich	A9539	Gel electrophoresis
Agilent 2100 Bioanalyzer	Agilent	G2939B	Bioanalyser used for mRNA quality control
CLARIOstar Plus Luminometer	BMG Labtech	430-501S-F	Plate reader/luminometer used for measuring luminescence
CleanCap AG	Trilink Biotechnologies	N-7113	5' cap used for IVT
DMEM, high glucose, pyruvate	Thermo Fisher Scientific	41966029	Media used for culture of HEK293-LgBiT cells
DNase I	New England Biolabs	M0303	Degrades DNA template post-IVT reaction
Dulbecco's Phosphate Buffered Saline (PBS)	Sigma-Aldrich	D8537	For cell washing and addition to plate to prevent evaporation
EDTA, pH 8.0, RNase-free	Thermo Fisher Scientific	AM9260	Component of TAE buffer
Ethidium Bromide	Sigma-Aldrich	E1510	Stain for gel electrophoresis
Fetal Bovine Serum (FBS)	Sigma-Aldrich	F7524	Supplement added to DMEM
HiScribe T7 High Yield RNA Synthesis Kit	New England Biolabs	E2040	In vitro transcription (IVT) kit
Leibovitz's L-15 Medium, no phenol red	Thermo Fisher Scientific	21083027	Media used for HEK293-LgBiT culture during luminescence measurements
L-glutamine	Thermo Fisher Scientific	25030-024	Supplement added to DMEM
Lipofectamine RNAiMAX	Thermo Fisher Scientific	13778150	Lipid-based transfection reagent
Lithium Chloride Precipitation Solution	Thermo Fisher Scientific	AM9480	Alternative approach for mRNA purification post-IVT
Monarch Spin PCR & DNA Cleanup Kit	New England Biolabs	T1130	Purification kit after plasmid linearization
NanoDrop One Microvolume UV-Vis Spectrophotometer	Thermo Fisher Scientific	ND-ONE-W	Spectrophotometer used for quantifying DNA and mRNA concentration
Nano-Glo Endurazine Live Cell Substrate	Promega	N2570	Substrate required for luminescence added to L-15 media
NucleoSpin RNA Cleanup Kit	Macherey-Nagel	12708612	Purifies mRNA post-IVT reaction
Nunc MicroWell 96-Well, Nunclon Delta-Treated, Flat-Bottom Microplate	Thermo Fisher Scientific	136101	Plate used for HEK293-LgBiT plating and subsequent measurements
Opti-MEM Reduced Serum Medium	Thermo Fisher Scientific	31985062	Media used for transfection reaction
Phusion High-Fidelity PCR Kit	New England Biolabs	E0553	PCR kit
RiboRuler High Range RNA Ladder	Thermo Fisher Scientific	11883993	RNA ladder for gel electrophoresis
RNaseZap	Thermo Fisher Scientific	R2020	RNase decontamination spray
TRIZMA Base	Sigma-Aldrich	T6066	Component of TAE buffer
Trypsin-EDTA Solution	Sigma-Aldrich	T4174	Cell dissociation reagent
UltraPure DNase/RNase free Distilled Water	Thermo Fisher Scientific	10977035	Used for template preparation and for IVT