1mΨ influences the performance of various positive-stranded RNA virus-based replicons

Abstract

Self-amplifying RNAs (saRNAs) are versatile vaccine platforms that take advantage of a viral RNA-dependent RNA polymerase (RdRp) to amplify the messenger RNA (mRNA) of an antigen of interest encoded within the backbone of the viral genome once inside the target cell. In recent years, more saRNA vaccines have been clinically tested with the hope of reducing the vaccination dose compared to the conventional mRNA approach. The use of N1-methyl-pseudouridine (1mΨ), which enhances RNA stability and reduces the innate immune response triggered by RNAs, is among the improvements included in the current mRNA vaccines. In the present study, we evaluated the effects of this modified nucleoside on various saRNA platforms based on different viruses. The results showed that different stages of the replication process were affected depending on the backbone virus. For TNCL, an insect virus of the Alphanodavirus genus, replication was impaired by poor recognition of viral RNA by RdRp. In contrast, the translation step was severely abrogated in coxsackievirus B3 (CVB3), a member of the Picornaviridae family. Finally, the effects of 1mΨ on Semliki forest virus (SFV), were not detrimental in in vitro studies, but no advantages were observed when immunogenicity was tested in vivo.

Introduction

With the implementation of global immunization protocols against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), including messenger RNA (mRNA) vaccines, during the COVID-19 pandemic, renewed attention has been focused on RNA-based vaccine modalities. Self-amplifying RNAs (saRNAs) are the next generation of RNA vaccines¹ that are usually constructed based on replicons of positive-sense ((+)-RNA) viruses, where the coding sequence of the viral structural proteins (SP) is replaced with that of a gene of interest (GOI), while retaining the coding sequences of non-structural proteins (NsPs), including the viral RNA-dependent RNA polymerase (RdRp)^2–4. Thus, their advantage over conventional mRNA vaccine platforms relies on the viral replication machinery, which amplifies the mRNA of the encoded GOI within target cells⁵. Several positive-sense and negative-sense RNA viruses have been used as backbone for the development of replicons, including those of the Togaviridae, Flaviviridae, Paramixoviridae, and Rabdoviridae families, among others^6,7. These have served as important tools for in vitro studies, and in vivo studies have been mainly performed using mice^8,9. Only Venezuelan equine encephalitis virus (VEEV) and Semliki forest virus (SFV) of the Togaviridae family, have been tested in clinical settings^8,10–12. However, it is unlikely that a single viral backbone is sufficiently versatile to allow the insertion of a wide range of GOIs. Therefore, it is crucial to further understand the practical uses of this innovative and promising vaccine modality by considering other viruses¹³.

The Nodaviridae family comprises a group of small non-enveloped bipartite (+)-RNA viruses classified into two genera, Alphanodavirus and Betanodavirus, which infect insects and fish, respectively¹⁴. The viral genomic RNAs (RNA1 and RNA2) that are capped on their 5′ end, but not polyadenylated on their 3′ ends¹⁵, are packaged in the virions. RNA1 encodes the viral RdRp protein A (Supplementary Fig. S1a). A subgenomic RNA3 comprising the end of RNA1, not co-packaged in the virions, carries two open reading frames (ORF). B1 protein of unknown function corresponds to the C-terminus of protein A. B2 protein, in the + 1 reading frame relative to protein A, is a known suppressor of RNA silencing. The Nodamura virus (NoV), the first nodavirus to be discovered, is known to be pathogenic to its natural insect hosts and is lethal when infecting suckling mice¹⁶, although not adults. Another member of the Alphanodavirus genus within this family is Flock House virus (FHV). This virus has served as a model to study the replication process of (+)-RNA viruses owing to its simplicity, and is thus the most extensively studied species of the family. FHV RNA has been reported to replicate in insect, yeast, plant, and mammalian cell lines^17–19. These attributes make the members of this family attractive candidates for biomedical applications.

Knowledge pertaining to the structural or chemical modifications that play important roles in RNA stability and antigen expression, in addition to formulation compositions for delivery in vivo, has been gained as a consequence of COVID-19^20–22. These advancements have been based on previous crucial research. For example, the use of modified nucleosides such as pseudouridine (Ψ), an isomer of uridine (U), for the synthesis of in vitro-transcribed mRNAs was shown to increase the expression of the encoded protein and reduce the activation of the innate immune response^23,24 that would otherwise lead to the degradation of the mRNA. The inclusion of N1-methyl-pseudouridine (1mΨ), was significantly pivotal for the COVID-19 mRNA vaccines^25,26. This modified nucleoside was reported to also enhance protein expression from mRNAs containing it, and to decrease toll-like receptor 3 (TLR3) activation²⁷. Alterations in the RNA secondary structure that might influence protein-RNA interactions, and increase in the size and abundance of polysomes, are among the mechanisms proposed for these effects^28,29. Viruses have been reported to contain RNA modifications that affect various steps of their life cycle in different ways, thus shaping their interactions with their host^30,31. For example, N(6)-methyladenosine (m⁶A), the most abundant modified nucleoside, has been reported to regulate gene expression and reverse transcription of the pre-genomic RNA of the hepatitis B virus³². The influence of modified nucleosides on saRNA vaccines has not been extensively explored. We hypothesized that saRNAs based on (+)-RNA viruses might benefit from the advantages provided by the modified nucleosides used for conventional mRNAs vaccines, at least early after delivery into the target cells, when the saRNAs serve as mRNA of the viral proteins involved in the replication process. However, delivered saRNAs must also act as template during the first rounds of replication. Thus, in the current study, we evaluated the effects of 1mΨ on viral replicons, first using a nodavirus as a model. We found that it interfered with viral replication, and that translation of viral RNAs was not abrogated. This was in contrast to what we observed for a replicon based on coxsackievirus B3 (CVB3), where expression of a reporter protein was not detected when synthesized with 1mΨ. Furthermore, we found that SFV-based replicons containing 1mΨ were functional in in vitro tests, but did not show benefits when used as saRNA vaccines in vivo. Our results reaffirm the importance of testing the effect of specific modified nucleosides on each particular backbone virus, considering the functional and structural complexities of their genomes.

Results

1mΨ negatively affects TNCL RNA1-based replicons

Since 1mΨ has been shown to improve the performance of mRNA vaccines^27,33, we sought to evaluate the effect of this modified nucleoside on replicons and assess whether similar benefits can be expected for saRNAs constructed using various viral RNAs. In the current study, we used replicons based on the TNCL nodavirus, which is closely related to FHV³⁴. We first confirmed the expression of the green fluorescent protein-HiBiT (GFP-HiBiT) reporter in BHK-21 cells transfected with conventional mRNAs containing U or 1mΨ, incorporated during in vitro transcription (IVT) (Fig. 1a). Next, we tested a TNCL RNA1-based reporter replicon where the GFP-HiBiT reporter was expressed as a cleavable fusion of protein B2 due to the presence of the T2A peptide. This replicon produced a luminescent signal when synthesized with U alone at one day post-transfection (dpt) (Fig. 1b). This signal increased significantly one day later, suggesting proper translation and accumulation of the reporter protein. However, when the replicon contained 1mΨ, the expression was significantly reduced, particularly when U was completely substituted with 1mΨ. These results correlated with the levels of GFP mRNA, which increased for the replicon containing U, but not 1mΨ (Fig. 1c). The expression of the reporter from this replicon results from the translation of RNA3, which is separately transcribed from a subgenomic promoter present in RNA1³⁵ (Supplementary Fig. S1a). The GFP sequence is present not only in RNA3 but also in genomic RNA1, which is amplified in transfected cells. When we amplified a region of the viral RdRp protein A ORF present only in RNA1 (Fig. 1d), the results correlated, as did those for the detection of the intermediate (−)-strand RNA by qPCR where a region within GFP was targeted for amplification (Fig. 1e): there was a significant increase of RNA measured for the replicon prepared with U, while no increase was observed for that prepared with 1mΨ. Furthermore, we assessed the presence of double-stranded RNA (dsRNA) 24 h post-transfection (hpt) by immunofluorescence and detected positive staining only in cells transfected with U-containing replicons (Fig. 1f). For replicons synthesized with a mixture of U and 1mΨ, a significant increase in protein and RNA levels was observed, albeit smaller than for replicons prepared exclusively with U (Fig. 1b–e).

Figure 1.

1mΨ negatively affects the replication of TNCL gRNA1-based replicons. BHK-21 cells were transfected with the GFP-HiBiT mRNA (a) or a TNCL RNA1-based reporter replicon (LC831250), coding for a cleavable GFP-HiBiT fusion protein of B2 (T2A-GFP-HiBiT) (b–f), that were transcribed in vitro using U, 1mΨ or a 50% mixture of both nucleotides. (a) Reporter expression from both types of mRNAs. Unpaired t-test was used to calculate the p-value. (b) Luminescent signal resulting from the reporter’s expression from TNCL RNA1-based replicons prepared with 1mΨ significantly decreased compared to that of U- or U/1mΨ-containing replicons. Two-way analysis of variance (ANOVA) with Tukey’s multiple comparison test was performed for the statistical analysis. GFP RNA (c), RNA1 (d) (−)-RNA1 (e) levels in samples taken at the indicated time-points from transfected cells, were measured by RT-qPCR. The converging arrows in the schematic representations of the replicons indicate the location of the amplified targets. Two-way ANOVA with Tukey’s multiple comparison test was performed for the statistical analysis of RT-qPCR results. The red arrow in (e) represents the primer used for the RT step in the detection of the (−)-strand RNA. (f) dsRNA was detected by immunofluorescence in cells transfected with U- or 1mΨ-containing TNCL replicons 24 hpt. DAPI staining was performed for nuclear counterstaining.

Translation of TNCL RNA1 or RNA2 is not abolished by 1mΨ

TNCL gRNA1 serves as the mRNA for the viral RdRp protein A³⁶. Thus, upon transfection with TNCL RNA1-based replicons, translation of protein A is a critical initial step that drives downstream events in the nodavirus replication process. The results shown in Fig. 1 suggest that replication is the step affected most, but the translation of protein A may also be affected by 1mΨ. To evaluate the effect of 1mΨ on the translation of TNCL RNAs, we constructed another replicon with a reporter gene cloned in-frame with the protein A ORF, where the expression of the reporter reflected that of protein A itself. The signal 3 hpt was not significantly different between the two types of replicons, and the expression of the reporter from the replicon containing 1mΨ was not completely abrogated, since an increase in the signal was detected between 3 and 24 hpt, indicating cumulative translation of the encoded protein (Fig. 2a). These results indicate that protein A was expressed at levels that would allow the replication process and amplification of the replicon.

Figure 2.

1mΨ does not abrogate the translation of TNCL-based replicons, but hinders the replication step. A TNCL gRNA1-based replicon with the cleavable reporter (T2A-GFP-HiBiT) encoded in-frame with the protein A ORF (LC831256) (a) or TNCL gRNA2-based RNA (LC831258) (b) was transcribed in vitro using either U or 1mΨ and transfected into BHK-21 cells. Expression of the GFP-HiBiT reporter was measured using HiBiT luminescent signals at 3 and 24 hpt. Two-way ANOVA with Tukey’s multiple comparison test was used for statistical analysis. (c) Schematic representation of the TNCL trans-replication system composed of protein A RdRp mRNA (top, LC831257) and the defective template TNCL reporter replicon (bottom, LC831255). (d) BHK-21 cells were transfected with U- or 1mΨ-containing target TNCL reporter replicons, in the presence or absence of PA mRNA. GFP expression was detected by fluorescence microscopy. (e) HiBiT luminescence was measured in the cell lysates of transfected cells. Statistical analysis was performed using one-way ANOVA with Tukey’s multiple comparison test. (f) Relative GFP RNA levels measured by RT-qPCR are shown. Two-way ANOVA with Sidak’s multiple comparison test was performed. (g) RNA immunoprecipitation was performed with HA-tagged protein A and template TNCL defective replicons containing either U or 1mΨ. The region amplified by qPCR is indicated by the converging arrows. Two-way ANOVA with an uncorrected Fisher’s least significant difference (LSD) was performed for statistical analysis.

Furthermore, we sought to test the effect of 1mΨ on the translation of TNCL RNA2 that encodes the viral capsid precursor protein. To verify, we constructed a reporter RNA2 containing the GFP-HiBiT-coding sequence between the TNCL RNA2 untranslated regions (UTRs), replacing the coding sequence of the capsid. Twenty-four hpt we observed that the luminescent signal from the RNA2 reporter prepared with 1mΨ was significantly lower than that prepared with U, but was still detectable and higher compared to 3 hpt (Fig. 2b), even in the absence of protein A. Overall, these results suggest that 1mΨ might restrain the translation of TNCL RNA2 but does not cancel it, and that it limits the early stages in the replication of TNCL RNA1.

1mΨ impairs the binding of the RdRp to TNCL RNA1 replicon

To provide more evidence on the replication impairment exerted by 1mΨ on TCNL RNA1-based replicons, we analyzed the expression of a reporter gene (GFP-HiBiT) using a trans-replication system (Fig. 2c). A defective template replicon containing a premature stop codon within the protein A coding sequence was constructed based on the replicon used in Fig. 1, and RNA was synthesized by IVT using either U or 1mΨ. An additional source of functional viral RdRp was provided by co-transfecting the mRNA of protein A, containing a C-terminus hemagglutinin (HA) tag, and the encephalomyocarditis virus (EMCV) internal ribosome entry site (IRES) sequence to facilitate its translation. The lack of reporter expression from the replicon prepared with U in the absence of protein A (Fig. 2d,e) confirmed that the addition of a separate source of RdRp was required. When viral protein A was provided by co-transfecting IVT protein A mRNA, reporter expression was only detected from the template replicon containing U. The levels of GFP RNA (Fig. 2f) were consistent with these results, as an increase of RNA was only detected when the template replicon was synthesized with U in the presence of protein A, suggesting that replication, and thus amplification of the replicon and GOI mRNA, were negatively affected by the presence of 1mΨ. This modified nucleoside has been reported to provide increased stability and/or rigidity to RNA molecules^29,37 and to affect RNA–protein interactions²⁸, which might affect the function of regulatory sequences such as viral UTRs. We performed structure prediction analyses of the wild type (WT) 3′ end, comprising part of the B2 coding sequence and the 3′ UTR of TNCL RNA1 (Supplementary Fig. S2a). We hypothesized that the bond between G at position 7 and C at position 31 of the analyzed sequences played an important role in establishing the predicted structure (Supplementary Fig. S2a). Indeed, point mutations aimed at disrupting this bond led to changes in the predicted secondary structures (M1 and M2, Supplementary Fig. S2b) that were rescued by swapping the positions of the interacting G and C nucleotides (M3, Supplementary Fig. S2b). Experimentally, we observed that clones with the disrupted 3′ end (M1 and M2) showed a decreased level of reporter expression and RNA replication (Supplementary Fig. S2c,d, respectively), suggesting that the conformation of the TNCL RNA1 3′ end is important for these processes. Similar results on the importance of the 3′ end structure have been reported for replicons based on Nodamura virus³⁸. Thus, we hypothesized that the presence of 1mΨ in regions of the 3′ end, by affecting U-A interactions within the stems of the 3′ UTR, might lead to a disrupted binding or recognition of the viral RNA by the RdRp. We performed an RNA-immunoprecipitation assay (RIP), using the trans-replication system described in Fig. 2c. As shown in Fig. 2g TNCL RNA1 containing U was more significantly enriched in the precipitate than in the control IgG, suggesting that the interaction of viral RdRp might be hindered by the presence of 1mΨ in the target RNA. Nevertheless, we observed that the pull-down of 1mΨ-containing RNA was still significantly enriched compared to its control IgG.

1mΨ abrogates the translation of CVB3-based replicons

To determine whether our findings on the effect of 1mΨ on TNCL nodavirus can be reproduced in other RNA viruses, we also tested a CVB3-based replicon (Fig. 3a). When the replicon was synthesized to contain U, the expression of the reporter was detected as luminescence as early as 3 hpt. The signal increased significantly after 6 h. However, when the replicon containing 1mΨ was transfected, no expression was detected, suggesting that either translation or replication was impaired (Fig. 3a). To determine which step was mainly affected, we examined dsRNA in replicon-transfected cells (Fig. 3b) and found that only replicons containing U showed positive staining, suggesting the abrogation of the replication step. CVB3 RNA is translated immediately after entering the target cell to produce a polyprotein, which is cleaved by viral proteases into various functional proteins³⁹. The translation is dependent on the viral IRES sequence located at the 5′ end of the viral genome⁴⁰. Ψ has been reported to exert deleterious effects on EMCV IRES sequences⁴¹ and 1mΨ prevented translation driven by CVB3 IRES in circular RNAs⁴². To confirm the effects of 1mΨ on replication in the context of linear CVB3-based replicons, we constructed a reporter RNA, containing the CVB3 IRES sequence that includes a part of the viral 5′ UTR, upstream of the GFP-HiBiT reporter, followed by the polyA tail, also present in the original replicon (Fig. 3c)⁴³. The results for this reporter were concordant with those obtained for the full replicon (Fig. 3a), indicating that translation was crippled by 1mΨ. Overall, our findings support the hypothesis that 1mΨ interferes with the normal IRES-mediated translation of CVB3 RNA, and hence, the downstream replication process of CVB3-based replicons.

Figure 3.

1mΨ impairs the translation of CVB3 replicons. (a) Schematic representation of CVB3 reporter replicon. VPg virion protein genome-linked; viral peptide corresponding to 3B in the early translated polyprotein. HEK 293T cells were transfected with in vitro-transcribed reporter replicons containing either U or 1mΨ, and HiBiT luminescence was measured in the lysates of transfected cells harvested at the indicated hpt. (b) dsRNA was detected by immunofluorescence in the transfected cells at 24 hpt. (c) A shorter reporter RNA, containing CVB3 IRES and part of the viral 5′ UTR upstream of the GFP-HiBiT coding sequence was constructed and in vitro transcribed, using either U or 1mΨ. RNA was transfected into HEK 293T cells, and HiBiT luminescence was measured at the indicated time points. Two-way ANOVA with Tukey’s multiple comparison test was performed for statistical analysis of (a,c).

The effect of 1mΨ on Semliki forest virus (SFV) replicons varies in vitro and in vivo

Alphaviruses are among the most commonly used viruses for developing saRNA vaccines. We tested the use of 1mΨ on a SFV replicon containing the coding sequence for GFP-HiBiT (Fig. 4a), driven by the sub-genomic promoter⁴⁴. Cells transfected with in vitro-transcribed replicons containing either U or 1mΨ, were evaluated for expression of the encoded reporter and amplification of RNA. As shown in Fig. 4b, expression of GFP was confirmed at 1 dpt. HiBiT quantitative analysis showed an increase in the luminescent signal originating from replicons containing either U or 1mΨ between 8 and 24 hpt (Fig. 4c), demonstrating translation and accumulation of the protein. The signal increased even more 48 hpt for U-containing replicons, but remained at similar levels for 1mΨ-containing replicons. Although at this time point there was a significant difference in the levels of luminescence between the replicons synthesized with different nucleotides, these results suggest that 1mΨ-containing SFV replicons are still functional. Consistent with these protein expression results, increasing levels of SFV RNA were detected in both types of replicons (Fig. 4d), indicating that replication was not abolished. There was, however, a delay in the increase of RNA when the replicon was synthesized with 1mΨ, indicating that it might generate a barrier during the replication process. These results were in line with the dsRNA staining of transfected cells that showed lower levels for 1mΨ-containing replicons 24 hpt (Fig. 4e). Next, we evaluated the efficacy of the modified SFV replicon to induce an immune response against an antigen of interest in vivo. For this purpose, we used the SARS-CoV-2 receptor-binding and transmembrane domains (RBD-TM) of the spike protein as a model antigen and cloned it into the SFV replicon, replacing the GFP-HiBiT sequence of the previous reporter (Fig. 4f). In vitro, this replicon led to the expression of the antigen (Supplementary Fig. S3a,b) and replication of the RNA (Fig. 4g, Supplementary Fig. S3c) with a tendency that correlated with that observed with the previous replicon containing GFP-HiBiT (Fig. 4d, Supplementary Fig. S3d). For the in vivo experiments, lipid nanoparticles (LNPs) encapsulating the in vitro-transcribed replicons were injected intramuscularly into BALB/c mice following the schedule detailed in Fig. 4h. Mice immunized with the U-containing replicon showed a clear increase in antibodies targeting the RBD antigen as determined by enzyme-linked immune-sorbent assay ELISA (Fig. 4i). In contrast, mice injected with the modified replicon showed very low and variable levels of antibodies, with titers that differed by approximately seven orders of magnitude. The physical characterization of the generated LNPs did not show significant differences that could explain the in vivo results (Supplementary Fig. S3e–g). These results suggest that, at least at the dose tested, 1mΨ did not show beneficial effects on the performance of SFV replicons in vivo.

Figure 4.

SFV replicons containing 1mΨ do not effectively induce an immune response in vivo. (a) Schematic representation of the SFV reporter replicon containing the GFP-HiBiT-coding sequence replacing the viral structural proteins. SFV reporter replicons were transcribed in vitro using U or 1mΨ, and transfected into HEK 293T cells. (b) GFP expression was confirmed by fluorescence microscopy 24 hpt. Brightness of the counted spots (n indicated in brackets) in each frame were quantified and shown in the plot. Unpaired t test was performed for the statistical analysis. (c) HiBiT luminescence of the lysate of transfected cells was measured at the indicated time-point (d) The presence of SFV genomic RNA was analyzed by RT-qPCR, at the indicated time-points post-transfection, targeting the region indicated by the converging arrows in (a). (e) Replication of SFV replicons in BHK-21 cells was confirmed by immunofluorescence microscopy after staining of dsRNA. DAPI staining was performed for nuclear counterstaining. (f) Schematic representation of the SFV replicon containing the coding sequence of an antigen of interest (SARS-CoV2 spike protein RBD and trans-membrane domain) for in vivo use. (g) The presence of the SFV genomic RNA from the replicon was first analyzed in vitro by RT-qPCR, at the indicated time-points. (h) Immunization protocol for mice injected with the SFV-based saRNA. (i) ELISA results using the sera of the immunized mice. Two-way ANOVA with Tukey’s multiple comparison test was performed for statistical analysis of (c,d,g) data. All data were collected from at least two independent experiments.

Discussion

During the COVID-19 pandemic, mRNA vaccines marked a pivotal milestone in the therapeutic usage of RNA molecules and opened doors to related next-generation vaccine platforms, including saRNAs, which are constructed based on viral RNAs. The incorporation of modified nucleosides such as 1mΨ was among the strategic modifications included in the current mRNA vaccines that was demonstrated to reduce the innate immune response triggered by foreign RNAs. However, the effects of modified nucleosides on saRNA vaccines have not yet been investigated in detail. In the current study, we found that the translation of nodavirus TNCL RNA 2 was not abrogated when they contained 1mΨ (Fig. 2a), and that the replication of TNCL RNA1 was significantly hindered (Figs. 1, 2).

RNA viruses rely on regulatory sequences within their genomes for replication in infected cells. The translation of uncapped viral mRNAs is sometimes facilitated by the presence of an IRES. These regulatory sequences, which were first discovered in picornaviruses, usually adopt structural conformations that vary in complexity and convey different degrees of reliance on endogenous factors for translation. In addition to increasing ribosome recruitment, Ψ and 1mΨ increase the stability of RNA molecules by improving base-pair interactions and base stacking^33,37,45, which may affect the ability of the IRES to recruit the host translation machinery to the viral RNA⁴¹. The results of the CVB3 replicons analyzed in the current study (Fig. 3) are in line with previous reports, where CVB3 IRES was abrogated by 1mΨ⁴². The same mechanism may underlie the interference exerted by 1mΨ on the replication of TNCL nodavirus. In the in vitro characterization of the constructed replicons, we observed that single point mutations in the 3′ end of RNA1 led to a decreased or delayed expression of the encoded GOI. Changes in the predicted secondary structure were observed in the fragments containing these point-mutations. These prediction analyses were limited to a portion of the 3′ end of TNCL RNA1, but suggested that the structural conformation of the 3′ end is important for viral replication. Considering the characteristics attributed to Ψ and 1mΨ on RNA molecules, it can be inferred that U-A interactions in regions of the 3′ end (Supplementary Fig. S2a) might be affected when 1mΨ is used instead (∆, Supplementary Fig. S2a) influencing the overall replicative capacity of the replicon. Within the context of the entire replicon sequence, the interpretation of these structural changes induced by 1mΨ become more difficult due to the complexity of the viral genomes. However, the RIP results showed a hindrance to the recognition or proper functioning of viral RdRp (Fig. 2g). RNAs synthesized with different modified nucleotides were reported to differ in their affinity to bind RIG-I and in their ability to enable or prevent its activation²⁸, suggesting that protein-RNA interactions are influenced by modified nucleosides. Therefore, although 1mΨ provides advantageous effects on mRNAs, those on replicons may vary depending on the viral RNA backbone used, as observed in our study. Our findings using SFV-based replicons synthesized with 1mΨ led to the expression of the GOI in vitro (Fig. 4a–d). Nevertheless, when tested in vivo as an saRNA vaccine, the modified version did not perform as well as the unmodified version (Fig. 4i). These results suggest that replication inefficiencies and SFV replicon degradation might occur in vivo. In vitro, we observed a delay in the increase of RNA levels for 1mΨ-containing SFV replicons (Fig. 4d,h), which could be indicating that by hindering replication, the difficulty of overcoming the innate immune response is larger. However, after the first rounds of replication, the newly synthesized U-containing replicons will continue to function as templates without the restraints of 1mΨ. A faster replicating U-containing replicon, on the other hand, would be better fit to confront the immune response. Another possibility for the poor in vivo performance could be the generation of proteins different from the GOI, as suggested in a recent report, where the presence of 1mΨ in COVID-19 mRNA vaccines was shown to induce frame shifts, resulting in the expression of aberrant proteins⁴⁶. To improve this outcome, it could be more beneficial to use a mixture of saRNAs, some containing U and some containing 1mΨ; or containing both U and 1mΨ (as shown in Fig. 1) may provide a balance between RNA stability, RdRp and GOI expression and, saRNA replication. In this case, we could not ascertain where each nucleotide was incorporated within the replicon. However, based on our results, we can infer that once the barrier of the first round of replication is overcome, 1mΨ no longer constitutes a hindrance, and the expression of the GOI may increase (Fig. 1b–e). Novel approaches to the synthesis of therapeutic RNAs might enable the use of specific modified nucleosides for specific regions of RNAs which, in the case of replicons, might help avoid the crippling of important regulatory sequences. Furthermore, other studies have provided alternative strategies to optimize the performance of saRNA vaccines, such as the inclusion of coding sequences for modulators of innate immunity⁴⁷ or the use of trans-amplifying systems⁴⁸ overall increasing the success of this vaccine modality. Modern technologies, such as reverse genetics, have enabled the construction of more customizable designs for recombinant viruses, widening the possibilities for their medical applications^49,50.

The number of RNA-based vaccines is rapidly increasing, reflecting the expectations placed on this vaccine platform, which includes a broad range of RNA molecules, in addition to the conventional mRNAs currently in use or the saRNAs analyzed here. Our study demonstrated that the inclusion of 1mΨ, although expected to improve the performance of saRNAs, resulted in different outcomes based on the backbone virus used, acting on different stages of the replication cycle. Other modified nucleosides such as 5 methylcytidine, with similar immunological benefits as 1mΨ, have shown better results in platforms based on other viruses^51,52. Thus, increased knowledge on basic virology and the strategies used to optimize RNA-based vaccine development will pave the way for the application of this versatile vaccine modality, using a variety of viruses, in the near future.

Methods

Cell lines

Baby hamster kidney (BHK-21) cells were purchased from the Japanese collection of research bioresources (JCRB) cell bank. They were maintained in Minimum Essential Medium (MEM) (Sigma-Aldrich, St. Louis, MO, USA) supplemented with 10% fetal bovine serum (FBS) (Sigma-Aldrich), penicillin (100 units/mL) (Nacalai Tesque, Kyoto, Japan), and streptomycin (100 µg/mL) (Nacalai Tesque), at 37 °C with 5% CO₂. When transfected with TNCL replicons, cells were incubated at 30 °C. Human embryonic kidney (HEK 293T) cells were purchased from ATCC (Manassas, VA, USA) and were maintained in Dulbecco’s Modified Eagle’s Medium (DMEM) (Sigma-Aldrich) supplemented with 10% FBS, penicillin, and streptomycin. High Five insect cells were purchased from Thermo Fisher Scientific (Waltham, MA, USA), and cultured in SF-900 II SFM medium (Gibco-Thermo Fisher Scientific) in the absence of FBS or additional antibiotics, at 27 °C. Vero cells were purchased from ATCC and maintained in DMEM, supplemented with 10% FBS, penicillin, and streptomycin.

Mice

All experiments and protocols were approved by the Animal Care and Use Committee of Osaka University and BIKEN. Seven-week-old BALB/c female mice were used to test the immunogenicity of SFV replicons in vivo. Mice were immunized with two doses of SFV replicon prepared with either U or 1mΨ (1 µg of RNA/30 μL/dose) administered intra-muscularly, three weeks apart, and sera was analyzed by ELISA three weeks after the second dose. Two independent experiments using 3 or 5 mice per group were performed. The mice were kept in a temperature- and light-controlled room at the animal facility of Osaka University with free access to food and water. All methods were carried out in accordance of relevant guidelines or regulations, and reported in accordance with the ARRIVE guidelines.

Cloning of TNCL and CVB3 genomes

High Five cells (Thermo Fisher Scientific) were infected with a baculovirus to induce the reactivation of the TNCL nodavirus that is latently present in this cell line³⁴. Four days later, the supernatant was collected and subjected to ultra-centrifugation (100,000×g, 4 h) over a 10% sucrose cushion at 4 °C. The pellet containing the virus was resuspended in PBS and stored at − 80 °C until use. Viral RNA was extracted from an aliquot of the stored virus using TriReagent (Molecular Research Center, Cincinnati, OH, USA) following the manufacturer’s protocol. The cDNA of TNCL genomic RNA1 and RNA2 were synthesized using SuperScript III reverse transcriptase (Thermo Fisher Scientific) according to the procedure described by Li et al.³⁴, and cloned into Invitrogen’s pCR4 Blunt TOPO plasmid (Thermo Fisher Scientific). To clone the CVB3 genome, Vero cells were infected with the Nancy strain of CVB3 (ATCC VR-30). Three days later, the supernatant was collected and viral RNA was extracted using TriReagent (Molecular Research Center), as described above for TNCL. The cDNA of CVB3 was synthesized with SuperScriptIII, and used as a template to amplify two viral fragments containing the required additional flanking sequences: the hammerhead ribozyme⁵³ was inserted upstream of the 5′ UTR, and a poly(A)₂₅ tail was added downstream of the region encompassing P2 to the 3′ UTR, in addition to a Mlu I site, used to linearize the plasmid for IVT^54,55. These sequences were then added to the primers used for PCR. The EMCV IRES sequence was separately amplified from the pAAV-IRES-Puro expression vector (Cell Biolabs, San Diego, CA, USA) with an inserted Hpa I restriction site in the 5′ end. These three fragments were assembled into pcDNA3.1 plasmid (Invitrogen) using the NEBuilder HiFi DNA Assembly Master Mix (New England Biolabs).

Generation of various viral replicons

TNCL RNA1-based replicons. We used the WT TNCL RNA1-containing TOPO vector as the backbone to be modified for different purposes. We constructed template plasmids for IVT by adding the T7 promoter and terminator sequences, and other regulatory sequences such as the hepatitis virus D ribozyme, based on a previous report⁵⁶. A schematic representation of the constructs is shown in Supplementary Fig. S1b. SFV-based replicons. A LacZ SFV-based replicon (pSFV3-LacZ) was purchased from Addgene (Watertown, MA, USA) (plasmid #92074)⁵⁷ and used as the backbone for the SFV replicons tested in the current study, replacing the LacZ sequence with that of GFP-HiBiT (Fig. 4a) or the SARS-CoV-2 spike protein RBD-TM (Fig. 4f). For the CVB3-based reporter experiments, the sequence encoding GFP-HiBiT was inserted into the Hpa I site using the wild-type replicon described above.

In vitro transcription (IVT)

TNCL template plasmids were linearized with Not I restriction enzyme (New England BioLabs, Ipswich, MA, USA) and purified using a Qiagen gel extraction kit (Qiagen, Hilden, Germany). One microgram of the digested plasmid was used for each reaction using the MEGAscript T7 transcription kit (Invitrogen), where 1mΨ triphosphate (TriLink, San Diego, CA, USA) was used instead of the U triphosphate contained in the kit (or a 50% mixture of U and 1mΨ), where necessary. Capping was performed using 50 μg of RNA with ScriptCap Cap1 system reagent (CellScript, Madison, WI, USA). This procedure was the same for mRNAs but required polyA tailing, which was performed using a poly(A) tailing kit (Invitrogen). In vitro transcription of CVB3 replicons was performed following the same protocol, but omitting the capping reaction. CVB3 replicons were constructed on a pcDNA3.1 vector background, and linearized with Xha I (New England BioLabs) restriction enzyme prior to IVT. Plasmids encoding SFV replicons were linearized with Spe I (New England BioLabs) restriction enzyme for IVT using the MEGAscript SP6 transcription kit (Invitrogen), following the procedure described above, as well as the capping step. The polyA tailing reaction was not necessary in this case because the polyA tail was encoded within the template plasmid and was thus generated during the IVT reaction. The SFV replicon used for in vivo experiments was prepared with an additional purification step using a cellulose column to eliminate dsRNA impurities, between the IVT and the capping steps.

Transfection of cells was performed with Lipofectamine2000 reagent (Invitrogen) at a ratio of 150 ng of RNA/µL of reagent.

Assessment of GOI expression from the replicons

The GOI cloned in the reporter replicons consisted of the 11-amino acid peptide of the NanoLuc luciferase (HiBiT) (Promega, Madison, WI, USA) fused to the C-terminus of GFP. The expression of the GFP-HiBiT reporter at the protein level was confirmed by fluorescence microscopy (Keyence BZ-X710, Osaka, Japan) and luminescence measurements of lysates with the same total protein content. Luminescence is generated upon the addition of the substrate together with the larger NanoLuc luciferase fragment LgBiT that binds to the HiBiT tag. To confirm replication, quantitative real-time PCR was performed to detect viral and GOI RNAs. The primer sequences are listed in Table 1. Total RNA was extracted from the transfected cells using the RNeasy Plus kit (Qiagen) following the manufacturer’s instructions. Five hundred to 1000 ng of RNA was used for reverse transcription using random oligomers and the PrimeStar RT kit (Takara, Shiga, Japan). For the detection of TNCL (−)-RNA1, cDNA was prepared using a virus-specific primer (RT_RNA1 primer in Table 1) and SuperScript III first-strand synthesis system for the reverse transcription step (Invitrogen). A tag sequence was added upstream of the virus-specific sequence of the primer to increase specificity in the amplification by PCR⁵⁸, when using a tag-specific and a GFP-targeting primers (Table 1). Real-time PCR was performed using Applied Biosystems PowerUp SYBR Green Master Mix (Thermo Fisher Scientific) in a QuantStudio1 thermal cycler (Applied Biosystems). Relative expression to the initial analyzed time point was calculated following the 2^ΔΔCt method, using hamster gapdh or human GAPDH genes as internal controls for normalization. The details of the primers used are listed in Table 1. For Western blot analysis of the RBD-TM-encoding SFV replicon, an anti-RBD Ab (40592-T62 SinoBiological, Wayne, PA, USA) was used and was detected with HRP-labeled anti-rabbit IgG Ab (Sigma-Aldrich AP307P). Human GAPDH was used as a protein-loading control and was detected using mouse anti-human GAPDH (Abcam ab8245, Cambridge, England) and anti-mouse IgG-HRP (Sigma-Aldrich).

Table 1.

Primers used in the current study.

Primer name	Sequence	Target	Assay
RNA1_F	AAAGTGAGCGGCTTTGATGC	TNCL gRNA1	qPCR
RNA1_R	TTGTAAAAACCATTCCTTCC
hGapdh_F	TCTTCCAGGAGCGAGATCCC	Hamster gapdh
hGapdh_R	ACTTGTCATGGTTCACACCC
GAPDH_F	CACATCGCTCAGACACCATG	Human GAPDH
GAPDH_R	TGACGGTGCCATGGAATTTG
GFP_F	AAGTTCATCTGCACCACCGG	GFP
GFP_R	AGAAGATGGTGCGCTCCTGG
SFV_F	GAGCTGAAAGAACTGACGCCG	Semliki forest virus gRNA
SFV_R	CGGCCTGATCTTCAGCCC
RBD_F	CGCCGACTACAATACAAGC	SARS-CoV-2 spike mRNA
RBD_R	GCTCGAAGGGCTTCAGATTG
RT_RNA1	CGGTCATGGTGGCGAATAAAACCAACAATCGAAGAACGC	TNCL (−) RNA1	RT
Tag	CGGTCATGGTGGCGAATAA		qPCR
GFP_R2	AACTTCAGGGTCAGCTTGCCGTAGGTGGC

Sequence of the primers used for detection of the indicated target mRNAs or viral RNAs (gRNA) by real-time PCR.

qPCR quantitative real-time PCR, RT reverse transcription.

Immuno-fluorescent staining

TNCL or SFV replicons were transfected into BHK-21 cells, which had been pre-seeded onto glass slides and placed inside the wells of a 12-well plate. At 24 hpt, the glass slides were removed, washed with PBS, and fixed with a 4% paraformaldehyde solution for 15 min at room temperature. After washing again with PBS, the cells were permeabilized with 0.5% solution of Triton X-100 in PBS. Blocking with 0.5% BSA in PBS (for 30 min at room temperature) was performed before staining with anti-dsRNA antibody (Ab) (Millipore, Burlington, MA, USA). Detection was performed using an Alexa568-labeled anti-mouse IgM Ab (Abcam) and observed under a Keyence BZ-X710 fluorescence microscope. The same procedure was performed on HeLa cells transfected with CVB3 replicons.

RNA immunoprecipitation (RIP)

BHK-21 cells were transfected with HA-tagged IRES-protein A mRNA and defective TNCL replicons in vitro-transcribed with either U or 1mΨ. Lysates were collected at 48 hpt and RIP was performed using the Magna RIP RNA-binding protein immunoprecipitation kit (Millipore) with an anti-HA tag Ab (Thermo Fisher Scientific, cat# 26183) or mouse isotype control IgG (Invitrogen cat# 02-6100), following the manufacturer’s instructions. Ten percent of the raw lysate was processed in parallel with the precipitated samples for RNA extraction using TriReagent (Molecular Research Center), and the RNA pellet was dissolved in 20 μL of water. The precipitated RNA (2 μL) was used for cDNA synthesis using a SuperScript III reverse transcription kit (Invitrogen) with random hexamers. A standard curve was constructed using the RNA of the raw lysates (input), and the amount of precipitated RNA was expressed as “percentage of input.” TNCL viral RNA1 in the precipitated RNA was detected by RT-qPCR using the primers RNA1_F and RNA1_R listed in Table 1.

Formulation of SFV replicons for in vivo experiments

In vitro-transcribed SFV replicons containing the SARS-CoV-2 spike RBD-TM coding sequence were encapsulated in LNPs, prepared with: ssPalmE-P4C2 (NOF Co, Tokyo, Japan), 1,2-dimyristoyl-rac-glycero-3-methoxypolyethylen glycol 2000 (NOF Co.), 1,2-dioleolyl-sn-glycero-3-phosphoetyanoamine (DOPE) (Avanti Polar lipids Inc., Alabama, USA), and cholesterol (Sigma-Aldrich), as previously described⁵⁹ and using a microfluidics device, kindly provided by Manabu Tokeshi and Masatoshi Maeki of Hokkaido University. Physical characterization of the prepared LNPs was performed in a Zetasizer Nano ZS90 nanoparticle analyzer (Marlvern Paralytical, Worcestershire, UK), where size and polydispersity index were determined. The encapsulation efficiency was calculated based on RNA contents measured with Quant-iT RiboGreen RNA assay kit (Invitrogen).

Enzyme-linked immune-sorbent assay (ELISA)

Plates were coated with recombinant RBD peptide dissolved in PBS (1 µg/mL, 100 µL/well, 96-well plates), and were incubated at 4 °C overnight. Wells were washed three times using 0.05% Tween20 in PBS, and blocking was performed with a solution of 1% BSA (Sigma-Aldrich) in PBS at room temperature for 2 h. After another washing step, aliquots of twofold serially diluted serum from immunized mice were added to the coated plates and incubated at room temperature for 2 h. The dilution buffer consisted of PBS with 1% BSA and 0.05% Tween20. Goat horseradish peroxidase-labeled anti-mouse Ab (Sigma-Aldrich) was used to detect RBD-specific antibodies together with the BioFix TMB One component HRP microwell substrate (Surmodics, MN, USA). A solution of 0.5 N HCl was used to stop the reaction, and the optical density was measured at 450 nm using an SH-9000 microplate reader (Corona Electric, Ibaraki, Japan).

Statistical analysis

Statistical analysis was performed using GraphPad Prism version 9.4.1 (Graphpad Software LLC). Figures were generated using the same software. The details of each analysis are provided in the corresponding figure legends.

Ethics statement

All methods were carried out in accordance of relevant guidelines or regulations.

Author contributions

P.M. designed and performed the experiments, analyzed the data, and wrote the manuscript. T.N., F.O., T.S., Y.F., and R.S. performed the experiments and analyzed the data. H.E. conceived and designed the study, supervised the project and reviewed the manuscript.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request. The sequences of the viral constructs are available in the DNA Data Bank of Japan (DDBJ). The accession numbers of the isolated TNCL genomic RNA1 and RNA2 are LC831249 and LC831259, respectively. The accession numbers for the replicons used in the different figures, are as follows. Figure 1b–f: TNCL RNA1 reporter replicon: LC831250. Figure 2a: The TNCL RNA1 reporter replicon: LC831256. Figure 2b: TNCL RNA2 reporter replicon: LC831258. Figure 2c–g: TNCL protein A mRNA and TNCL RNA1 defective reporter replicon: LC831257 and LC831255, respectively. Supplementary Figure S2: TNCL RNA1 wild-type (WT_C10) and mutant (M1, M2 and M3) reporter replicons: LC831251, LC831252, LC831253 and LC831254, respectively.

Competing interests

P.M., T.N., F.O., T.S., Y.F., R.S., and H.E. are employed by BIKEN. H.E. holds a managerial position at BIKEN.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-024-68617-y.