Guidelines for minimal reporting requirements, design and interpretation of experiments involving the use of eukaryotic dual gene expression reporters (MINDR)

Abstract

Dual reporters encoding two distinct proteins within the same mRNA have played a crucial role in identifying and characterizing new instances of unconventional eukaryotic translation mechanisms. These mechanisms include initiation by internal ribosomal entry sites (IRESs), ribosomal frameshifting, stop codon readthrough, and reinitiation. This design allows one reporter’s expression to be influenced by the specific mechanism under investigation, while the other reporter serves as an internal control. However, challenges arise when intervening test sequences are placed between these two reporters. Such sequences can inadvertently impact the expression or function of either reporter, independently of translation-related changes. These effects may occur because of cryptic elements inducing or affecting transcription initiation, splicing, polyadenylation, and antisense transcription, as well as unpredictable effects of the translated test sequences on the stability and activity of the reporters. Unfortunately, these unintended effects may lead to incorrect conclusions being published in the scientific literature. To address this issue and assist the scientific community in accurately interpreting dual reporter experiments, we have developed comprehensive guidelines. These guidelines cover experiment design, interpretation, and the minimal requirements for reporting results. They are designed to aid researchers conducting these experiments, as well as reviewers, editors, and other investigators who seek to evaluate published data.

Background

Fusions of genes expressing reporters have been used for characterizing mRNA translation mechanisms since the 1990s^1–5. The popularity of this approach increased with the invention of the dual luciferase reporter in 1998⁶. Since then, dual reporters have been used to study ribosomal frameshifting, stop codon readthrough, reinitiation, and internal initiation^6–12. Dual reporter systems have proven instrumental, in combination with site-directed or random mutagenesis, in characterizing mRNA features involved in specific translation mechanisms^11,13–23 and for the identification of underlying cellular factors impacting these mechanisms^24–32. Dual reporters have also found application in the identification and study of drugs targeting specific mechanisms of translation^25,33–39 as well as the mechanisms of disease associated polymorphisms⁴⁰. Increasingly, dual reporters are being used for discovery of novel translational mechanisms and processes^41,42.

The principle is based on encoding two distinguishable reporters within the same mRNA so that expression of one reporter is dependent on the studied mechanism while the other reporter is used for normalization, i.e. as an internal control (Figure 1). As reporter proteins are synthesized from the same mRNA, it is assumed that any differences in the expression of the reporter used to characterize the mechanism under study is independent of confounding variables, e.g. transfection efficiency or RNA stability. The assay can be performed in a single tube or a well with the readout being the ratio of upsteam to downstream reporter activity. However, differences in reporter expression can occur for several reasons other than differential translation, potentially leading to false interpretation of reporter readout (Figure 1). Furthermore, since the readout of dual reporter assays is usually the ratio of downstream to upstream reporter, then changes in the upstream internal control reporter without changes to the downstream reporter will result in misinterpretations without careful consideration of absolute reporter values.

Figure 1.

Principles of a dual reporter strategy and the most common artefacts (shaded background) arising from unexamined assumptions. It is expected that a single RNA species is transcribed from the DNA sequence shown at the top of the illustration. Even if a single RNA species is transcribed, it is assumed that the fusion of the insert sequences to either reporter does not alter reporter protein activity or stability. However, several shorter possible RNA species are shown that could lead to artefactual reporter expression.

There are various possible sources of dual reporter artefacts. Firstly, DNA-encoded dual reporters may contain cryptic promoters, cryptic splice sites, or cryptic polyadenylation signals that could generate unexpected mRNA transcripts encoding only one of the two reporters^11,43–54. Plasmid DNA can also produce unexpected antisense transcripts that may affect reporter expression⁵⁵. Secondly, the protein extensions encoded by the test sequence may alter the stability or activity of one or both reporters, if it is synthesized as a part of the same polypeptide chain^12,56. Thirdly, in certain applications, the downstream reporter is placed under a known (control) or a putative (test) IRES. However, IRES activity may be influenced by surrounding sequences^20,21,57,58; it has also been shown that the presence of an IRES may influence mRNA stability⁵⁹ and translation^60,61 of the upstream reporters in a sequence-dependent manner.

Although various artefacts generated by experiments involving dual reporters have been discussed extensively^{11,12,46–50,52,57,62–70}, the misinterpretation of these assays continues to result in inaccurate conclusions^53,54,56.

Therefore, we developed guidelines for the design, interpretation, and reporting of experiments involving single mRNA-encoded dual reporters. In the first part, Principles and Recommendations (PR), we cover the design and interpretation of dual reporter experiments. We provide guidance on the selection of appropriate vectors to reduce risks of false positive findings and design of critical controls and additional experiments that may help to identify and avoid erroneous conclusions for a wide range of mechanisms. The second part, Minimal INformation on Dual expression Reporters (MINDR), outlines proposed reporting requirements, which establish the essential data that should accompany any publication describing dual reporter experiments. MINDR is designed to facilitate reproducibility and equip editors, reviewers, and readers with the information necessary for evaluating the reliability of the employed dual reporter strategies and assessment of the likelihood of false positive findings.

PR: Principles and Recommendations

In the following subsections we consider reporter design, possible controls, and data interpretation. These are intended to equip researchers with the means to investigate and avoid potential artefacts. We first describe general aspects regarding most dual reporter applications followed by considerations relevant to specific types of assay or phenomena which are also summarsied in Table 1.

Table 1.

Recommended controls and orthogonal evidence

Phenomenon	Negative control	Positive control	Orthogonal evidence
Frameshifting	Mutate putative frameshift site; Introduce stop codon downstream of frameshift site in frame with downstream reporter.	Introduce an indel to place reporters in same reading frame and make synonymous changes to the slip site.	Immunoblotting if reporters are fused; ribosomal profiling: when frameshifting produces longer proteoform, ribosome footprint density is expected dowsntream of zero frame stop codon with triplet periodicity phase matching new frame. When frameshifting is efficient and is predicted to produce a shorter proteoform, drop of ribosome footprint density is expected downstream of stop codon in the new frame.
Stop codon readthrough	Introduce double stop codon (e.g. two TAA codons instead of a readthrough stop codon candidate.	Stop to sense codon substitution.	Immunoblotting if reporters are fused. Ribosome profiling: ribosome footprint density is expected downstream of the stop signal with triplet periodicity matching the zero frame.
Reinitiation	Overlap between upstream and downstream reporter ORFs; introduce a stop codon or indel in the intercistronic region closer to the 2nd cistron.	For reinitiation after short uORFs, delete AUG of an upstream uORF; for long uORFs, reporters could be fused in-frame and separated with StopGo	Immunoblotting to show that reporters are NOT synthetized as a single fusion protein. Use several different reporters as a 2nd cistron.
Internal initiation	In DNA construct, delete the promoter directing the bicistronic transcript. Compare with bicistronic constructs harboring long non-specific intergenic sequences.	Compare with known effective viral IRES (e.g. EMCV) and with the corresponding monocistronic constructs.	mRNA transfection instead of DNA reporters; compare bicistronic construct with the corresponding monocistronic ones (m7G-capped, A-capped); insert 5’-terminal stem-loop and/or long overlapping uORF; siRNA targeting upstream reporter; polysome analysis; using appropriate cell-free systems..

General sources of artefacts

Unexpected mRNA species

Assumptions and potential problems –

It is often assumed that all measured reporter activity is derived from translation of a single RNA species generated in strict accordance with the researcher’s design. However, unexpected mRNA species may be produced through either cryptic splicing, cryptic transcription, or internal polyadenylation (Figure 1). These aberrant monocistronic transcripts could be translated much more efficiently than their bicistronic counterparts and, even if they represent only a tiny fraction, could substantially distort the overall interpretation of the results.

Recommendations -

The most straightforward way to control for these possibilities is to introduce in vitro transcribed RNA encoding the reporter proteins instead of using DNA-based constructs. Any discrepancies between DNA-encoded and RNA-encoded reporter protein ratios may indicate the existence of unexpected cryptic transcripts, although there may be other factors explaining these differences as discussed below. There may also be situations when RNA transfections are impractical as described in the RNA vs. DNA Reporters section below.

For DNA transfections using reporters that are expected to express fusion proteins (e.g. by readthrough or frameshifting), western blotting is an effective and sensitive approach to ensure that downstream reporter protein expression is not derived from aberrant mRNA species producing shorter variants of fusion proteins.

Where products of dual reporters are not expected to be fused (e.g. in StopGo vectors and reinitiation or IRES studies), it may be desirable to identify possible aberrant mRNA species directly. Because even a minor mRNA species could produce a major signal if translated very efficiently, highly sensitive techniques for detecting low abundance mRNA species are required. To this end, emerging long read sequencing technologies (such as nanopore or PacBio) or RT-qPCR may be sufficiently sensitive approaches to detect low abundance RNA species whereas northern blotting is unlikely to be sensitive enough. RT-qPCR can be used for assessing the levels of different RNA segments⁴⁶ with specific primers as shown in Figure 2. Careful design and validation of RT-qPCR amplicons with in vitro transcript standards and melting curves is necessary to determine whether the amounts of each amplicon are equal⁷¹. As the same PCR amplicons can be derived from alternative transcripts and because RT-qPCR optimization is not always straightforward, it is advisable that this method is not the sole approach used to analyze the occurrence of transcript isoforms⁶⁸. An in vitro transcribed full-length control RNA can be used to assess the signal expected from a single RNA species containing both reporters (Figure 2).

Figure 2.

Assessment of dual reporter transcriptional integrity. Red arrows show the design of forward and reverse primers for the dual reporter construct and double red lines depict the resulting cDNA products. Using a synthetic RNA construct (bottom) it is possible to estimate the ratio between cDNA products for the transcript containing all three regions. Horizontal blue lines represent siRNAs for knockdown experiments expected to downregulate expression of both reporters if they are produced from the full-length RNA template as indicated with vertical yellow and green arrows.

Furthermore, an elegant approach employing RNA interference can be utilized to regulate the authenticity of bicistronic RNA⁴⁶. An siRNA probe designed to knock down expression of the upstream reporter directs specific degradation of the bicistronic reporter mRNA, thereby decreasing the activity of both the upstream and downstream reporters to a similar extent, unless they are expressed from different mRNA species (Figure 2). Transcript isoforms could also be explored by methods like Cap Analysis of Gene Expression (5’-CAGE)⁷² or with direct RNA sequencing using nanopore technologies, bearing in mind that low abundance alternative products can produce most of the reporter activity.

Potential existence of aberrant transcripts may also be explored with study-specific controls, the design of which is specific to the studied phenomena.

Altered activity or stability of fused reporter proteins

Assumptions and potential problems –

When testing relative activities from fused reporter proteins (e.g. for readthrough or frameshifting), it is often assumed that the reporter protein activities accurately reflect reporter mRNA translation. Frameshifting and stop codon readthrough efficiencies are generally estimated by determing the ratio of upstream to downstream reporter protein of a test construct with the reporter protein ratio from a positive control (100%), where both reporters are in the same, uninterrupted reading frame. However, the part of the protein fusion encoded by the test sequence may influence the stability or activity of one or both reporter proteins. This can be especially problematic if fusions to the upstream reporter only influence the in-frame control reporter protein and not the test reporter, or vice versa (Figure 1)^12,56.

Recommendations –

Place StopGo/2A sequences on both sides of the test sequence¹² (Figure 3). StopGo is an atypical translation mechanism where the elongating ribosome fails to form a peptide bond and then efficiently continues translation – the outcome of this phenomenon is the production of two peptide chains (from the same ORF) separated by the StopGo motif ⁷³.

Figure 3.

Fused versus unfused dual reporters. Both systems have their merits and drawbacks. The advantage of fused reporters is the ability to estimate recoding efficiencies by western blotting or in vitro translation reactions while controlling for unintended D-Reporter translation by internal initiation or cryptic splicing. One potential downside of fused reporters is erroneous estimations of recoding efficiencies based solely on reporter protein activities because of possible changes in D-Reporter activity resulting from its fusion to the test and U-Reporter proteins. This can be mitigated with unfused reporter systems where the test peptide is released from the reporter proteins by flanking StopGo (SG) motifs. While western blotting or in vitro translations can still be used to estimate recoding efficiencies with unfused reporter systems, due to the similar sized D-Reporter protein produced, spurious D-Reporter protein expression by internal initiation or cryptic splicing may not be as obvious.

A peptide motif enabling StopGo was initially discovered during translation of the 2A region of foot-and-mouth disease virus⁷⁴, and since then a number of 2A peptides with varying efficiencies have been described. While in theory StopGo use should produce reporter proteins with identical amino acid sequences irrespective of the inserted test sequence, this may not always be the case. StopGo is not 100% efficient (generally 80–90%⁷⁵), therefore a certain amount of fusion between either of the reporters and the test sequence-encoded fragment will still be produced. Nonetheless, for example, a comparison of absolute Renilla luciferase activities (when used as an upstream reporter) from a series of fused and unfused test constructs indicated much more variability with fused reporters compared to StopGo-containing plasmid ¹². Additionally, the potential for ribosome drop-off during StopGo that may influence the ratio between two reporters can be easily tested with an appropriate in-frame control construct. When using a new StopGo-containing plasmid, it is important to monitor the efficiency of StopGo by western blotting in the intended biological system. Although variability in StopGo activity has been reported⁷⁶, it can be mitigated by using longer StopGo motifs (>30 aa), which are less sensitive to the effects of proximal sequences, including test inserts, while having higher activities and negligible ribosome drop-off⁷⁵. It should be noted that, although StopGo is active in all eukaryotes tested so far, different 2A peptide variants have different activities in various organisms, and none have been found to function in bacteria⁷⁷. Furthermore, the kinetics of StopGo is still poorly understood and it is likely to cause some ribosome pausing, which may interfere with some of the studied phenomena.

Alterations in reporter ratios due to artefactual changes in reporters’ absolute readouts

Assumptions and potential problems –

It seems reasonable to assume that when transfecting the same reporter construct into two different cell lines or under different conditions, the ratiometric readout should be similar. However, this is not always the case. Not surprisingly, absolute reporter values may be quite different between cell lines/conditions, sometimes spanning orders of magnitude. There may be several reasons for this, including differences in the reporter delivery efficiencies between the cell lines. Reporter RNAs may trigger the innate immune response resulting in global suppression of translation, e.g. through activation of PKR, cGAS, TLRs, NLRs, or RIG-I-like receptors^78,79. Reporter expression in particular cell lines can be compromised at either the transcriptional (e.g. weak promoter activity in DNA reporters) or translational (e.g. low global translation rates) level. When dual reporters are used to compare the efficiency of the studied mechanism across cell lines or under specific conditions (including stress versus control, overexpression or depletion of specific factors, treatment with small molecule inhibitors, and NMD activation) it is important to understand that differences in global translation may affect the ratio of the measured reporter activities without affecting their real relative activities. This is because any measured activity is a combination of genuine reporter protein activity with background levels due to biological noise and technical limitations. We can represent the ratio of measured activities as $(\mathit{X}+\mathit{n})/(\mathit{Y}+\mathit{n})$, where $\mathit{X}$, $\mathit{Y}$ are bona fide activities of the two reporter proteins and $\mathit{n}$ is the constant background signal. It is clear then to see that if $\mathit{X}$ and $\mathit{Y}$ are changed proportionally (e.g. reduced twice by general translation inhibiton) and $\mathbf{X}/\mathbf{Y}$ ratio stays the same, the measured $(\mathit{X}+\mathit{n})/(\mathit{Y}+\mathit{n})$ ratio will change.

Perhaps more problematically, there may be cell specific differential reporter activity due to the presence of a cell-specific repressor/activator that acts on one of the reporters. Another factor to consider is the reporter overexpression driven by a strong promoter that could further exacerbate cell-specific effects by e.g. titrating essential translation components.

Recommendations –

Here it is critically important to report absolute readouts of reporter and background activities to reveal potential misinterpretations of observed changes in relative activities of the reporters, as e.g. in Khan et al⁵³.

Assay-specific principles and recommendations

RNA vs DNA reporters

The use of mRNA transfection is a powerful strategy to avoid the generation of unexpected RNA species by cryptic transcription, polyadenylation, and/or splicing events. However, its advantages are not limited to this. For several applications, RNA transfection may be preferred irrespective of transcriptional artefacts. Analysis of the immediate stress response is particularly difficult when using plasmid DNA reporters. There is little point in applying stress stimuli immediately after DNA transfection as substantial time is required for the accumulation of sufficient mRNA. In contrast, for RNA transfection, the stress stimuli can be applied immediately or shortly before/after the transfection (1–2 hours), here, newly synthesized protein products are responsible for most of the reporter activity. It is also difficult to synchronize the expression of DNA reporters in the entire cell population without synchronizing their cell cycle, as the plasmid only enters the nucleus during mitosis. Furthermore, transfection of non-dividing cells (e.g. matured neurons or cardiac myocytes) with plasmid DNA is highly inefficient.

Nonetheless RNA transfection is not a panacea and should be used appropriately. Firstly, activity values are generally much lower from mRNA transfections as compared to DNA transfections. For this reason, RNA transfections should use reporters with a high signal-to-background ratio such as luciferase reporters. Secondly, RNA is relatively unstable, so it is better to analyze the activity shortly after transfection, ideally in 1–4 hours post transfection. At later time-points, e.g. 8–10 hours post transfection, accumulated reporter products start to exceed newly synthesized ones – even more so than in the case of DNA transfections. It is also important to use a transfection protocol that minimally stresses the cells and to avoid stress stimuli like cell re-plating, electroporation, or plate/dish cooling immediately before or during transfection⁸⁵. Thirdly and most importantly, one should keep in mind that transfected artificial mRNA has not passed through the nucleus and may lack specific and potentially critical features including epitranscriptomic marks or associated mRNA-binding proteins of nuclear origin. Furthermore, in primary cells, mRNA transfection often triggers innate immune responses, so additional efforts are needed to reduce its activation, like the use of transcripts with 5’-cap and modified nucleotides (e.g. m¹Ψ in place of U)^78,79. It is also important to note that lipocationic transfection impedes the analysis of reporter mRNA stability, as only a tiny fraction of transfected mRNA is released from endosomes into the cytosol⁸⁶. Accordingly, total RNA extraction from the transfected cells yields mRNA that is not predominantly derived from the cytosolic fraction but from endosomal and endolysosomal compartments^86,87. Finally, transfected RNA may still be processed by cytoplasmic RNA cleaving enzymes, such as IRE1^88,89 or RNase L⁹⁰, or during alternative polyadenylation that generates translated RNA fragments downstream of it⁹¹. Therefore, one cannot exclude the possibility that even in the case of mRNA transfection, aberrant RNA species may also be present. A more detailed comparison of DNA and RNA transfections can be found in a dedicated review⁹².

The use of dual reporters for massively parallel assays

In recent years, dual reporter vectors have become popular in Massively Parallel Reporter Assays (MPRA) that simultaneously evaluate thousands of test sequences. This is a powerful approach that allows screening of a diverse pool of sequences for specific regulatory properties, e.g. driving internal initiation^41,42, ribosomal frameshifting³⁴ and other translation mechanisms. It can also be applied to screen a pool of all possible variants of a particular sequence to comprehensively characterize cis-acting regulatory elements.

While many of the general principles that should guide the use of dual reporters outlined above also apply here, the high-throughput nature entails several specific considerations associated with measuring reporter protein activity in a pooled manner. Certain guidelines such as reporting absolute measured expression levels of the reporter genes is typically harder to achieve than in the case of single reporter measurements. On the other hand, dual reporter-MPRAs also allow for the inclusion of a much larger number of controls that can be measured in the same experiment.

Particularly suitable for MPRA are dual fluorescent reporter constructs, as they allow fluorescence-activated cell sorting based on relative reporter expression levels. This is then followed by DNA sequencing-based identification and quantification of the underlying sequences.

An important consideration for MPRA is to ensure equal vector copy number and similar expression levels among the cells. Utilization of systems for genome integration⁹³ enables the integration of a single copy of the test vector in the same genetic locus for all cells. However, this may not be strictly required. When this issue was carefully examined recently⁹⁴, a high degree of correlation was found between the expression of two cistrons across a polyclonal population of cells, regardless of integration site and number of integrated copies of the bicistronic construct.

A significant challenge in screening large pools of diverse sequences is discerning artifacts amidst the positive hits. Reliance on validating only a subset of hits can lead to misinterpretation, as each individual sequence may possess unique properties, and even a single nucleotide change can alter cryptic promoter activity. Given that MPRA-based screens typically yield numerous hits, it is impractical to validate each one individually, making it difficult to estimate the rate of false positives^41,42. Therefore, general conclusions should not be drawn from limited validated cases. Instead, conducting a meta-analysis of positive hit sequences may help identify shared features and specific types of artifacts. While rare artifacts are unlikely to significantly impact conclusions drawn from MPRAs aimed at dissecting the regulatory code of a gene regulatory mechanism, findings related to specific “positive” events should be limited to candidates that have been appropriately validated to minimize the potential for misinterpretation.

Cell-free translation systems and other in vitro assays

Another potential source of false positives from dual reporter assays is inappropriate use of in vitro translation systems. For example, the commercially available and widely used nuclease-treated Rabbit Reticulocyte Lysate (ntRRL), which is prepared from specialized cells with a limited range of RNA-binding proteins, has been repeatedly shown to inaccurately reproduce conditions found in normal cells^62,69. Firstly, exogenously added mRNAs translated in ntRRL exhibit a relatively weak reliance on the 5’ cap^51,95, although optimizing the buffer conditions can substantially increase cap-dependency and start site recognition⁹⁶. Moreover, as some eIF4G molecules are sequestered by the capped 5’-terminal mRNA fragments remaining in the ntRRL after limited hydrolysis of endogenous reticulocyte transcripts, the addition of cap-dependent initiation inhibitors (such as m⁷GTP, 4E-BP, or proteases that cleave eIF4G) may release this factor and artificially stimulate translation of uncapped mRNAs. This system also does not recapitulate the cap/poly(A) synergy⁹⁷. Taken together, all this is often misinterpreted as an indication of cap-independent translation of a particular studied mRNA. Secondly, ntRRL is prone to aberrant internal initiation at AUG codons located within extended unstructured regions⁹⁸, causing artificial expression of the second cistron in bicistronic reporters even in the absence of bona fide IRESs. Finally, translation in RRL is highly sensitive to even moderately stable RNA secondary structures in the 5’ leader, which is not the case in living cells⁹⁵.

It is likely that similar artefacts are present in cell-free systems derived from budding yeast and wheat germ, at least under some conditions^67,99. In contrast, such effects are not typically observed in cytosolic extracts of cultured mammalian cells^51,95. However, in any in vitro system, results can greatly depend on the specific preparation conditions and component concentrations. For example, varying polyamine concentrations in the yeast extract can modulate stop codon readthrough efficiency as much as 4-fold¹⁰⁰. Therefore, caution is necessary when comparing findings from a specific cell-free system with cultured cells or in vivo observations.

Less commonly, bicistronic constructs are used in in vitro systems reconstituted from purified components^101,102. Although such analysis is very informative, it also should be done with caution. Similar to RRL, in these systems, which are usually devoid of mRNA-binding proteins, the ribosomes are able to bind to internal AUG codons located within long unstructured regions^103,104. This risks confusion of an authentic mechanism with an artificial one that should be excluded by validation in complete in vitro or in vivo systems.

Another potential source of artefacts specific for in vitro assays is related to partial hydrolysis of an in vitro synthesized reporter mRNA that can produce truncated versions of bicistronic constructs. This leads to inappropriate ribosome loading to the second cistron on transcripts that do not contain the first cistron. Thus, the extensive analysis of mRNA integrity should be carried out. In some cases (i.e. IRES studies), analysis of polysome-associated mRNA fractions can be helpful. In particular, total RNA isolated separately from monosome and polysome fractions of each sample tested would be subject to RT-qPCR using primers covering either each cistron alone or both cistrons together to determine the integrity of the full-length reporter mRNA.

Phenomenon-specific principles and recommendations

Specific considerations in the assessment of ribosomal frameshifting and stop codon readthrough

The assessment of ribosomal frameshifting and stop codon readthrough using dual reporters involves a calculation of the relative reporter activities (downstream to upstream ratio). The relative reporter activity of the test construct is then compared to that of a control construct. It is not advisable to calculate reporter activities relative to a negative control rather than a positive control, as this can result in the unintended consequence of an insignificant number having an outsized effect. An appropriate positive control has both reporters encoded within the same ORF (in-frame control). Often a single in-frame control is compared to several test sequences. We caution against this practice; when the sequence of the test constructs is considerably different from the sequence of a single in-frame control as some of the test sequences may contain cryptic splice sites or promoters. The ideal in-frame control should have no amino acid differences from the expected frameshift or readthrough product and minimal nucleotide differences.

For frameshifting, an in-frame control can be obtained with either insertion or deletion of a single nucleotide in the putative frameshifting site depending on the expected direction of frameshifting (−1 or +1). For stop codon readthrough, the stop codon should be replaced with a sense codon, ideally one encoding the same amino acid that is expected to be inserted in place of the stop codon. However, the identity may not be known in advance and there may be one of several possible amino acids inserted^105,106. Reporter ratios in such positive control constructs is considered to correspond to 100% efficient frameshifting or stop codon readthrough. For studying frameshifting, it is advisable to introduce synonymous changes to disrupt the putative slippery sequence within the in-frame control to create a more accurate readout of 100% frameshifting. If the frameshifting site is not disrupted in an in-frame control and frameshifting occurs (say at 10% efficiency), only 90% of the ribosomes would synthesize the downstream reporter. Thus, ideally, each tested sequence should have its own positive in-frame control. Including a +1 or −1 frame termination codon (depending on the reading frame of the downstream reporter) 5’ of a putative frameshifting site is also worth considering to ensure that only frameshifting within the test sequence is reported.

When using fused reporters, orthogonal validation of frameshifting by western blotting is highly desirable when reporting novel instances. Most reporter proteins can be detected by commercially available antibodies that can detect frameshifting or stop codon readthrough efficiencies as low as 1%. As mentioned above, when using fused reporters, western blotting can also control for cryptic splicing and cryptic promoter activity.

Another important way to validate ribosomal frameshifting and stop codon readthrough is the use of a negative control. Ribosomal frameshifting normally occurs at specific frameshifting sites accompanied with stimulatory elements such as specific mRNA structures. While single point mutations in stimulatory signals rarely abolish frameshifting completely, disruptions of the frameshifting site are expected to eliminate frameshifting as tRNA re-pairing in the new frame is precluded. A useful negative control for stop codon readthrough is the insertion of tandem in-frame TAA stop codons that in most organisms represent the most efficient terminators.

Although not always possible, it is desirable to avoid AUG codons within the test sequence that are in the same reading frame as the downstream reporter ORF, as they may serve as initiation codons on cryptic transcripts missing the upstream reporter, entirely or partly. However, a deliberate insertion of an AUG codon in-frame with the downstream reporter may be used as a control for the existence of cryptic transcripts. In the absence of such transcripts, introduction of an AUG codon in a good Kozak context should not substantially alter the activity of the downstream reporter unless there is reinitiation. Although reinitiation is extremely rare after translation of long ORFs, there are certain signals that could enable reinitiation even after translation of long ORFs^107–110. Another consideration is potential initiation at near-cognate starts which is generally inefficient, but is known to be highly productive and even comparable to that of AUG in certain contexts, therefore it is advisable to examine the sequences for the presence of such near-cognate start codons in a good Kozak context. Additional evidence may be derived from orthogonal methodological approaches, such as ribosome profiling by accessing publically available data in dedicated browsers^111,112. While a detailed discussion of this technique is beyond the scope of this paper, we recommend that interested readers consult specialized reviews^113,114.

Specific considerations in the assessment of reinitiation

Reinitiation is a process in which a ribosome initiates translation downstream of the stop codon at which it terminates (reviewed in^115,116). Usually, this process is not very efficient in eukaryotes, unless the translated upstream ORF is short or mRNA-specific mechanisms are used^{107–110,115,117}. However, reinitiation can be greatly facilitated under some physiological stress conditions or when ribosome recycling factors are artificially depleted^29,118.

The rate of translation reinitiation after long ORFs can be assessed with the dual reporter assay; other methods are more suitable to examine reinitiation after short uORFs (see^115,116 and references therein). However, due to the inefficiency of this process under normal conditions, absolute values of reporter activity and appropriate background correction should be thoroughly considered. As in other cases, cryptic promoters in intercistronic spacers must be excluded and only appropriate cell-free systems should be used for in vitro studies to exclude false (or true) internal initiation. Moreover, stop codon readthrough or frameshifting can be erroneously attributed to reinitiation, so these possibilities should also be excluded. The most appropriate control is to extend the coding region of the upstream ORF into the downstream ORF by mutating the stop codon of the former in a way to avoid matching reading frames of both ORFs – such a construct should allow zero reinitiation unless the AUG of the first ORF is “leaky” for initiation, which would still generate some background activity from the downstream reporter. Additionally, readthrough can be specifically excluded by inserting an additional stop codon (or a tandem of stops) between the translated upstream ORF and the putative reinitiation site, which would reduce readthrough, but not reinitiation. Similarly, the possibility of ribosomal frameshifting can be reduced by the insertion or deletion of a nucleotide within the intercistronic region to disrupt the reading frame and eliminate the possibility of frameshifting without affecting reinitiation (unless the nucleotide indel disrupts a cis-acting signal responsible for reinitiation). Alternatively, western blotting could be used to rule out the possibility of either readthrough or frameshifting.

Specific considerations in the assessment of internal initiation

When testing for IRES-dependent translation, the mRNA sequence suspected to promote internal initiation is inserted between two reporters to enable the translation of the downstream reporter. Unlike scenarios involving frameshifting or stop codon readthrough, no fusion product is expected. Therefore, the potential for the translation of the first cistron to interfere with the proposed IRES must be minimized. This can be achieved by incorporating additional stop codons to prevent any readthrough and by positioning the IRES at a sufficient distance from the first cistron stop codon to ensure that terminating ribosomes do not compromise the structural integrity of the IRES. However, it is important to consider the possibility that integrating an RNA fragment into an unnatural context may affect any potential IRES activity.

Although distant elements affecting stop codon readthrough have been reported^119,120 and may also exist for frameshifting, it is often the case that only short motifs or structures are examined in those instances. Conversely, since internal initiation relies on more elaborate structures, testing longer sequences is essential. Accordingly, all the aforementioned concerns regarding cryptic promoters and splice sites are particularly relevant in the context of IRES testing. It is important to consider that nearly any arbitrarily selected long fragment of a mammalian 5’ UTR could exhibit cryptic promoter activity, attributable to the abundance of transcription factor binding sites within these regions. Furthermore, a significant proportion of human genes contain alternative transcription start sites (TSS)¹²¹, and TSS switching can occur during acute stress^122–124. Additionally, long insertions increase the risk of false positives when using in vitro systems like ntRRL, which are not strongly cap-dependent.

To reveal cryptic splicing (but not cryptic promoter) mediated events, the independence of first and second cistron expression can be verified by designing constructs containing a sufficiently long uORF in the 5’ UTR (preferentially overlapping with the main AUG of the first cistron and with the uAUG codon in strong Kozak context) or by inserting a stable hairpin at the very 5’ end of the transcript to diminish the translation efficiency of the first cistron⁶². However, this approach requires analysis of the transcript level, as the downstream reporter may be affected if such modifications alter the stability of the whole mRNA.

In IRES research, the use of mRNA reporters is distinctly preferable to DNA reporters⁶². Using in vitro transcribed reporters eliminates artefacts arising from cryptic promoter activity, unintended splicing, or premature transcription termination. When employing DNA transfection to evaluate putative IRES activity, promoterless and siRNA-mediated controls (see above) are essential.

Most importantly, however, the bicistronic assay used to examine IRES activity possesses a notable intrinsic limitation that was not initially apparent. Comparing the expression ratios of upstream and downstream cistrons between bicistronic reporters containing different IRESs is often uninformative. This is because any two long non-specific arbitrary sequences placed between reporters are highly unlikely to yield equal readouts, while different bona fide IRESs can also exhibit significantly different activities (Figure 4A). As a result, it becomes challenging to confidently design negative or positive controls for such assays.

Figure 4.

Peculiarities of dual reporter assays in the study of IRES activity. A. Simulated results from typical bicistronic assays and their potential interpretation. Unambiguous conclusions cannot be drawn solely from comparisons of the activities of different bicistronic mRNAs with each other, as control values (both negative and positive ones) may exhibit significant variability. B. A mechanism of ribosome recruitment employed by a specific mRNA fragment (for example, a 5’ UTR) can be elucidated through three related dual reporter assays using the mRNA constructs shown in the upper subpanel. Three distinct translation initiation mechanisms can be distinguished: canonical cap-dependent scanning, CITE-directed, or IRES-directed, as indicated below. To prepare m⁷G-capped transcripts, it is advisable to employ CleanCap, ARCA, or post-transcriptional enzymatic capping instead of m⁷GpppG, as these methods yield more natural, non-immunogenic, and efficiently translated mRNAs. Additionally, ApppG or ApppA can be utilized to produce transcripts that are stabilized by non-functional cap analogs at the 5’ ends. The bottom panel represents typical simulated results. Bkg, the background level of the reporter activity in mock-transfected cells.

Many putative IRES sequences are derived from natural 5’ UTRs, which, in the case of cellular and many viral mRNAs, are naturally capped and therefore can be bound and scanned by the canonical initiation machinery. Therefore, even if the sequence under investigation suggests internal initiation in a bicistronic context, it is not straightforward to evaluate the contribution of internal initiation to overall translation in its natural context. These complications can be addressed by comparing expression from bicistronic and monocistronic reporters, allowing for an assessment of the relative contributions of different initiation mechanisms (Figure 4B).

Finally, it is important to note that the term “IRES” specifically refers to internal ribosome entry rather than cap-independence. While the latter arises as a consequence of the former, some mRNAs that definitely lack an IRES and strictly require a free 5’ end can nevertheless be efficiently translated in an uncapped (or capped with the artificial non-functional A-cap analog) form. In such cases, specific elements known as cap-independent translation enhancers (CITE) facilitate cap-independent translation⁶². In contrast to IRESs, CITEs cannot direct translation of the second cistron in a bicistronic mRNA (Figure 4B). Furthermore, views regarding the widespread presence of IRESs in the 5’ UTRs of cellular mRNAs are largely based on the assumption of low processivity of the translation initiation complex, which is, however, significantly underestimated^51,125,126. This has contributed to the perception that any long 5’ UTR must use a non-canonical translation initiation pathway, which does not seem to be the case. Given the considerable proportion of purported “cellular IRESs” that have been challenged to date^{11,44,46,48–52,54,62,65,67–70,116}, one should bear in mind that the detected cap-independent activity of a sequence under investigation may easily be attributed to CITEs or even some of the experimental artefacts described above.

MINDR: Minimal INformation on Dual expression Reporters.

Based on the caveats described above, we propose the following three minimal reporting requirements that should mitigate many of these issues and should accompany any study with data obtained with dual reporters:

1. A list of positive and negative controls

As discussed above, the use of appropriate positive and negative controls is critical. However, it may be too impractical to design all the controls described in the previous sections. Nevertheless, the level of confidence in the reported results depends on the specific controls used in an experiment. Therefore, the authors should explicitly describe which positive and negative controls have been used for detecting potential artefacts, to help reviewers and readers assess the reliability of the study.

2. Full sequences of all vectors and inserts

Sequences responsible for potential artefacts, such as transcription enhancers or donor/acceptor splice sites, may be distant from the sequence encoding reporters^52,55. Therefore, for the reproducibility and future interrogation of reported results, the exact sequence of all plasmids used should be provided.

3. Absolute readout values for each reporter and for the background

The use of dual reporters in general requires the analysis of their ratios rather than absolute values, which are subject to high variability due to technical reasons, such as transfection efficiencies.

Nonetheless, as discussed earlier, consistent differences in absolute values are often indicative of artefacts. Therefore, it is important that in addition to providing the ratios between the reporter activities, the absolute raw readouts should be made available for each replicate of each construct including background values from control experiments without reporter constructs.