Amber codon is genetically unstable in generation of premature termination codon (PTC)-harbouring Foot-and-mouth disease virus (FMDV) via genetic code expansion

ABSTRACT

The foot-and-mouth disease virus (FMDV) is the causative agent of FMD, a highly infectious and devastating viral disease of domestic and wild cloven-hoofed animals. FMD affects livestock and animal products’ national and international trade, causing severe economic losses and social consequences. Currently, inactivated vaccines play a vital role in FMD control, but they have several limitations. The genetic code expansion technology provides powerful strategies for generating premature termination codon (PTC)-harbouring virus as a live but replication-incompetent viral vaccine. However, this technology has not been explored for the design and development of new FMD vaccines. In this study, we first expanded the genetic code of the FMDV genome via a transgenic cell line containing an orthogonal translation machinery. We demonstrated that the transgenic cells stably integrated the orthogonal pyltRNA/pylRS pair into the genome and enabled efficient, homogeneous incorporation of unnatural amino acids into target proteins in mammalian cells. Next, we constructed 129 single-PTC FMDV mutants and four dual-PTC FMDV mutants after considering the tolerance, location, and potential functions of those mutated sites. Amber stop codons individually substituted the selected amino acid codons in four viral proteins (3D, L, VP1, and VP4) of FMDV. We successfully rescued PTC-FMDV mutants, but the amber codon unexpectedly showed a highly degree of mutation rate during PTC-FMDV packaging and replication. Our findings highlight that the genetic code expansion technology for the generation of PTC-FMD vaccines needs to be further improved and that the genetic stability of amber codons during the packaging and replication of FMDV is a concern.

Introduction

Foot-and-mouth disease (FMD) is one of the most feared highly contagious animal viral diseases in countries with a developed livestock production industry [1]. This disease remains enzootic in many regions of the world. It leads to substantial economic losses in the livestock industry due to its rapid transmission, a severe reduction in animal productivity, and high mortality in newborn animals caused by myocarditis [2,3]. FMD is listed by the World Organization of Animal Health (Office International des Épizooties, OIE) as a notifiable disease and upon disclosure of an outbreak, severe limitation of movement, and trading are imposed [4]. Its aetiological agent is the FMD virus (FMDV) in the genus Aphthovirus of the family Picornaviridae. The FMDV infects important domesticated production animals including pigs, cattle, sheep, goats and deer about 70 species of other cloven-hoofed wildlife animals. There are seven distinct serotypes: O, A, C, Asia-1, and Southern African Territories (SAT) 1, 2, and 3. The FMDV genome consists of a positive-single-strand RNA (8400 nt). The viral RNA contains a large open reading frame (ORF) that encodes a large polyprotein, that is then cleaved by viral proteases to form four different structural proteins and seven other non-structural proteins plus various precursors. The P1 encoding four structural proteins, VP1, VP2, VP3, and VP4, forms an icosahedral capsid without an envelope containing the viral RNA genome. The L, P2, and P3 encoding non-structural proteins are involved in protein processing and genome replication [5–8]. Since FMDV shows a high degree of variation in genetic and antigenic, and vaccination does not offer cross-protection in different serotypes. Therefore, it is necessary to design recombinant viruses using new technologies for the rapid development of effective novel vaccine candidates.

In recent years, the genetic code has been expanded to encode unnatural amino acids (UAAs) that are applicable to site-specific incorporation into proteins in various cells and organisms [9–12]. Genetic code expansion with the engineered orthogonal aminoacyl-tRNA synthetase (aaRS)/tRNA pair suppresses a repurposed nonsense codon to directly incorporate UAAs at the desired site in genes of interest [11,13]. At present, more than 200 different UAAs have been genetically encoded, providing exciting new approaches to studying outstanding problems in biology using genetic code expansion technology [14–19]. An orthogonal aaRS/tRNA pair directs the incorporation of UAAs, most commonly in response to the amber stop codon (UAG). The pyrrolysyl-tRNA synthetase (PylRS, encoded by PylS)/tRNA^Pyl_CUA (encoded by PylT) pair from Methanosarcina species (commonly M. mazei or M. barkeri) is arguably the most useful pair. Thus, this pair was developed into a highly versatile system for orthogonal translation. The pylRS/tRNA_CUA pair is orthogonal in a range of hosts, including Escherichia coli, yeast, mammalian cells, Caenorhabditis elegans, and Drosophila melanogaster [10,20]. This pair has evolved to direct the incorporation of multiple structurally and functionally distinct UAAs, and it has also been developed for UAA incorporation into different viral proteins. The genetic code expansion technology has been developed to incorporate UAAs into virus-associated proteins using engineered nonsense-suppressing aaRS/tRNA pairs. Lin et al. first used hepatitis D virus (HDV) as a model system, combining the genetic code expansion strategy with an engineered virus assembly process to site-specifically introduce several different UAAs into the capsid of a mammalian virus; this demonstrated the feasibility of incorporating UAAs into proteins from an intact and infectious virus [21]. This technology has greatly improved the diversity of protein chemical functions once UAAs were precisely incorporated, and the technology has been used to generate premature termination codon (PTC)-harbouring viruses [21–33].

The genetic code expansion technology was successfully applied to several additional mammalian viruses, which has provided a novel strategy to create viral vaccines based on the PTC-viruses. The PTC-viruses as live but replication-incompetent viruses have been constructed by introducing nonsense codons into the viral genomes’ essential genes [21,26,27,29,34]; consequently, the PTC-viruses replicate only in cells containing orthogonal aaRS/tRNA_CUA pairs and their substrate UAAs [27]. Attention has been given to PTC-virus strains. They present a novel general approach for viral vaccine development by incorporating UAAs into essential viral proteins, thereby controlling virus replication through the genetic code expansion [11,26,27]. In a proof-of-concept study, Guo et al. reported the first manipulation of the essential HIV-1 protein biosynthesis through UAA-mediated suppression of genome-encoded blank codons to construct live but replication-incompetent virus vaccines [27]; they then further addressed the construction of all-in-one HIV-1 mutants that carried genomic copies of the aaRS/tRNA_CUA pair via a synthetic biology approach, demonstrating that HIV-1 multicycle replication in vitro could be controlled through a UAA switch [27,29]. Using a strategy similar to the one used in HIV studies mentioned above, the PTC-influenza virus vaccines were generated with transgenic cells, HEK293T-tRNA/pylRS/GFP^39TAG, by introducing one or more amber codons into the influenza A genome [26]. Moreover, vaccination of animal models with the PTC-influenza A virus elicited robust humoral, mucosal, and cell-mediated immunity and protected against subsequent infection with influenza A and heterologous viruses [26]. Therefore, these PTC-virus vaccines have received considerable attention because they represent an attractive middle zone between the inactivated and live-virus vaccines [13]. However, the generation of PTC-virus for FMDV through genetic code expansion has not been empirically investigated so far.

In this study, we first evaluated the feasibility and effectiveness of generating PTC-virus for positive-single-strand RNA viruses with genetic code expansion technology by using FMDV as a virus model.

Results

Establishment and characterization of a BHK-21 cell line harbouring the orthogonal translation machinery

Previous research has indicated that one of the most critical challenges to producing PTC-virus is the requirement of a special packaging cell line that harbours the orthogonal translation machinery [26,27]. We first established the viral packaging cells that would carry the orthogonal translation system according to the reported approach [35,36]. Two vectors, PB2 (pPB-4×PylT/PylS-IRES-puro) and PB4 (pPB-4 × PylT/mCherry-TAG-eGFP-IRES-neo) were constructed for PiggyBac transposase-mediated integration of amber suppression machinery into the baby hamster kidney cells (BHK-21) (Figure 1(a)). We first tested the compatibility of the orthogonal translation system, the Methanosarcina mazei PylRS/tRNA_CUA pair (MmpylRS/pylT) and three orthogonal unnatural amino acids (UAAs) N^ε-tert-Butoxycarbonyl-L-lysine (BocK), N^ε-Benzyloxycarbonyl-L-lysine (Lys (Z)) and N^ε-2-azidoethyloxycarbonyl-L-lysine (NAEK), with the viral packaging cells. The functioning of the orthogonal translation mechanism was confirmed through transient transfection experiments with the PiggyBac constructs. The BHK-21 cells were co-transfected with plasmid PB2 and the encoding eGFP-Y39-TAG vector in the absence and the presence 1 mM of the UAAs (Bock, Lys (Z) or NAEK), known PylRS substrates [30,37], in response to the amber stop codon in eGFP^39TAG. The eGFP expression was positive in BHK-21 cells in the presence of 1 mM of the UAA (Figure S1). Next, the BHK-21 cells were co-transfected with two targeting vectors (PB2 and PB4) and a plasmid that carried the PiggyBac transposase gene integrate a pylT and pylS expression cassette with an amber stop codon between mCherry and eGFP (mCherry-TAG-eGFP) into the cell genome (Figure 1(b)). After selecting with Puromycin and G418, we isolated more than 300 positive cell clones with strong mCherry fluorescence via the limited dilution method (Figure 1(c)). The clonal cell population responded uniformly to the addition of 1 mM NAEK.

Figure 1.

Schematic representation and characterization of the transgenic cells, BHK21-tRNA/pylRS/mCherry-TAG-eGFP, and their genetic stability by constitutive eGFP expression in the presence of UAA

(a) The pPB2 vectors (pPB-4 × PylT/PylS-IRES-puro) and pPB4 vector (pPB-4 × PylT/mCherry-TAG-eGFP-IRES-neo) used in transient transfection and the generation of stable cell lines. INS, insulators; IRES, internal ribosome entry site; prom, promoter; pA, polyadenylation signal; NeoR and PuroR represent resistance markers; grey arrows denote inverted terminal repeats. The amber codon between mCherry and eGFP is indicated as ‘TAG’. (b) Schematic representation of the stable transfection of BHK21 cells for generation of transgenic BHK21-tRNA/pylRS/mCherry-TAG-eGFP cells compatible with the orthogonal translation machinery. UAA: unnatural amino acid. (c) Representative images of an engineered cell clone grown in the presence of 1 mM NAEK for 5 days. Scale bar = 100 µm. (d) Quantifying the RNA expression levels of pyltRNA and pylRS in engineered cell clones. RNA expression ratios are relative to BHK-21 B2M RNA (N = 3). Expression of pylRS and eGFP was detected in the transgenic cells by western blotting against HA-tag and Flag-tag; An N-terminal Flag tag was added to PylRS. The mCherry-TAG-eGFP double fluorescent reporter carried a C-terminal HA tag for detecting full-length products. (e) Genetic stability of the transgenic cells according to the UAA-dependent eGFP phenotype and verified as tested in their 50th generation. Scale bar = 400 µm.

The Relative expression levels of pylRS and pylT were analysed from 39 different clonal cell lines. The data showed that the relative expression ratio of pylT varied substantially (988.12–1.70 × 10⁵); the relative expression ratio of pylRS ranged was 1.45 to 47.74 (Figure 1(d)). The expression of pylRS and eGFP was detected by western blotting with or without 1 mM NAEK. Full-length eGFP was detected only in cells supplemented with 1 mM NAEK, while no eGFP was observed in the absence of NAEK (Figure 1(d)). The final transgenic cells, BHK21-tRNA/pylRS/mCherry-TAG-eGFP, were selected according to the UAA-dependent eGFP phenotype and were identified by constitutive expression of the orthogonal tRNA/pylRS pair as tested in their 50th generation (Figure 1(e)). In addition, no significant differences were observed for amber suppression-dependent eGFP expression in the presence of different NAEK concentrations (1, 2, 5, 10 mM), indicating that there was no effect on the UAA incorporation efficiency (Figure S2). These results indicated that the stable amber suppression cell line, BHK21-tRNA/pylRS/mCherry-TAG-eGFP, was successfully constructed.

Effect of the orthogonal tRNA/pylRS pair on FMDV packaging and propagation

The amber suppression-dependent FMDV can only complete the infection cycle in cells with an exogenously supplemented orthogonal tRNA/pylRS pair and UAA (Figure 2(a)). To define the differences in FMDV packaging efficiency among transgenic tRNA/pylRS/mCherry-TAG-eGFP cells and parental BHK-21 cells, the transgenic cells’ capability to assemble viruses was tested by transfecting the cells with packaging plasmids of wild-type FMDV (Figure 2). The time of apparent CPE was reasonably consistent among transgenic cells and parental BHK-21 cells (Figure 2(b)). Furthermore, in the presence or absence of UAAs, the TCID₅₀ of the FMDV were detected in both cells, and thus they displayed identical virus replication efficiency from each infection (Figure 2(c)). The packaging of FMDV and viral titration assays suggests no detrimental effects of the orthogonal tRNA/pylRS pair integrated into the cells’ genome on FMDV packaging and propagation.

Figure 2.

Establishment and characterization of a virion packaging system that is compatible with the orthogonal translation machinery

(a) Schematic representation of the generation of amber suppression-dependent FMDV. The FMDV mutant can only complete a single infection cycle in the cells in the absence of an exogenously provided orthogonal pyltRNA/pylRS pair. (b) Evaluation of the packaging and production of wild-type FMDV after transfecting the transgenic and parental cells with the FMDV rescue plasmid by the CPE assay (N = 3). Scale bar = 100 µm. (c) Functional evaluation of the orthogonal tRNA/pylRS pair’s effect on the propagation of the wild-type FMDV by comparing the parental BHK-21 and transgenic tRNA/pylRS/mCherry-TAG-eGFP cells in the presence or absence of UAA (N = 3).

Construction and examination of PTC-FMDV mutants

The 3D polymerase (3D^pol) of FMDV is the virus-encoded RNA-dependent RNA polymerase (RdRp) that carries out replicating the viral genome in the cytoplasm of host cells [38,39]. 3D^pol sequences are highly conserved among the different serotypes and subtypes of FMDV [40]. To recover infectious PTC-FDMV from the full-length cDNA clone, FMDV 3D^pol was first chosen to construct PTC-virus through substituting 25 selected amino acid codons with an amber stop codon based on considering the tolerance, location, and potential functions of those mutated sites (3D-1) (Figure 3). The PTC-viruses were recovered via transfecting transgenic cells with mutant FMDV rescue plasmids with or without 1 mM NAEK (Figure 2(a)). The production and propagation of putative PTC-viruses were verified by the cytopathic effect (CPE) assay. The CPE phenotype was observed in the transgenic cells with 1 mM UAA (Figure S4). One of the most critical concerns is the potential reversion of the amber nonsense codon to a sense codon during PTC-FMDV replication and propagation. Therefore, the stability of the amber stop codon was verified with next-generation sequencing (Figure S5). Seventeen CPEs were observed in 25 PTC-FDMV mutants, and the relative packaging efficiencies of these viruses were not completely consistent (Figure 4). PTC-virus production and packaging were significantly slower by more than 50% compared to wild-type viruses, except for two mutations (H14 and M16) (Figure 4, 3D-1). However, gene sequencing showed that 9/17 amber stop codons reversed to sense codons at the second passage, while 8/17 were mutated during the first passage (Figure S5). This result indicated that the PTC-FMDV could be rescued successfully in transgenic cells. However, the artificial introduction of amber stop codons into the viral genome will affect the translation of viral proteins and virus packaging efficiency. Moreover, these amber stop codons in 3D^pol of FMDV were generally not very stable.

Figure 3.

The FMDV genome organization and selection of the amino acid residues for mutation

(a) Genome organization of FMDV. The positive-sense single-stranded RNA genome is approximately 8,400 bases long and contains a single ORF encoding structural and non-structural proteins. The black triangles indicate the position in the FMDV genome of mutant proteins for which an amino acid codon was substituted using an amber codon. The figure is based on Dill and Eschbaumer [61] with modifications. (b) The amino acid residues selected for replacement by UAAs were labelled within each viral protein’s tertiary structure. 3D, PDB: 6S2I; L, PDB: 2JQF; VP1 and VP4 (yellow), PDB: 1FOD. The gap of VP4 is due to the unresolved 3D structure of residues 40 − 64 [62].

Figure 4.

The packaging efficiency of PTC-FMDV with an amber codon introduced at different sites located in the variable, average or conserved domains based on ConSurf analysis

The relative efficiency represented a normalization of the days required to show the typical CPE at each test site compared to the wild-type FMDV (N = 3).

Analysis of the genetic stability for the amber stop codons in FMDV packaging and replication

To generate the PTC-FMDV, we further completed the evolutionary conservation analysis of the amino acid residues of viral proteins to ConSurf calculations (Figure S3) [41]. Based on the analysis, mutant sites on four different FMDV proteins were selected, replacing the corresponding codons with amber stop codons to rescue the PTC-virus. A total of 104 amino acid codons were replaced individually, including two non-structural proteins (3D and L) and two structural proteins (VP1 and VP4) of FMDV, most of which were located in conserved domains or sites. Using the same strategy, the following codons for the generation of PTC-FMDV were systematically explored: 23 codons in 3D (3D-2), 26 codons in L, 16 codons in VP4, and 39 codons in VP1 (Figure 3(b)). Similar to the result of 3D-1, the packaging process of PTC-FMDV was significantly longer than that of WT-FMDV. The replacement of some amino acid sites by the amber stop codon could not rescue the corresponding PTC-virus, which may be related to amino acids’ function at different sites in the virus genome. Different degrees of packaging and propagation efficiency in transgenic cells were observed: 10 UAG codon mutations in 3D-2, 10 in L, 11 in VP4 and 10 in VP1 that caused evident CPE (Figure 4, S4).

The reversion and lost UAA dependency of the amber stop codons were verified by next-generation sequencing. Seven out of 10 amber codons were reversed in 3D-2, 2/11 in VP4, 7/11 in L, and none in VP1 at the first passage (Figure S5). Other PTC mutants still retained the amber stop codon in the viral genome of the first passage (UAG codon in 3D-D43, H60, K177; L-P66, K84, E93; VP1-Q25, Q28, K41, T43, N46, R67, T102, Y107, H108, A116; VP4-Q29, D66, L71, S73, F76, L79, F80, A82, A85), but UAG reversion occurred during the second passage (Figure S5). In addition, the bimodal signal of sequencing was observed in VP1-N46 mutants (Figure S5). There may have been a mixture of revertant and PTC-viruses, and therefore, 23 viral plaques were selected from the first passage of the VP1-N46 mutant after a plaque purification assay. However, the reversion or mutation also appeared in the verified seven plaque viruses at the mutant locus by sequencing at the third passage (Figure S5). Meanwhile, the CPE-negative mutants were also amplified by RT-PCR, and no signal was detected in PCR product by sequencing analysis, indicating that there may no virus and/or virus replication in those CPE-negative cells. Furthermore, the CPE-negative cells were passaged blindly from the 1^st to 5^th passage; CPE-positive samples after the passaging showed that the amber codon was mutated. In contrast, no signal was detected in CPE-negative cells after passaging based on sequencing analysis (each sample was first amplified by PCR and then the PCR products were sent to sequencing, data not shown).

Construction and examination of dual-PTC-FMDV mutants

The dual-PTC mutants containing double amber stop codons were constructed to prevent the reversion of amber stop codons in the PTC-FMDV genome. The five sites with the highest packaging efficiency were selected based on four proteins (L, VP1, VP4 and 3D) of FMDV. Among these, only VP1-N46 mutant was still observed with TAG signal in the 3^rd passage sequencing analysis. Although the TAG occurred mutation during passaging, the process of mutation was relatively slow compared with other sites. Thus, the VP1-N46 mutant was selected and combined with other mutated sites of three proteins (L, VP4 and 3D) to construct dual-PTC mutants (L-Y42/VP1-N46, 3D-H14/VP1-N46, 3D-M16/VP1-N46, and VP4-F76/VP1-N46) to reduce the risk of PTC mutation further. The CPE was not observed at the first passage for these dual-PTC mutants only in the presence of UAA, but four dual-PTC mutants appeared CPE at the second passage with or without UAA (Figure S4). Next-generation sequencing showed that the amber codons at these mutation sites (L-Y42, 3D-H14, 3D-M16, VP4-F76, and VP1-N46) also mutated to sense codons (Figure S5). This result suggested that the infectious dual-PTC-FMDV could be recovered after transfecting with a dual-PTC FDMV packaging plasmid, that targeted mutation sites of different viral proteins. The packaging and propagation efficiency of PTC-FMDV generally decreased when introducing more amber stop codons into the FMDV genome. Finally, there was the possibility that all of the amber codons revert to sense codons during PTC-FMDV replication and propagation.

Discussion

Currently, commercial inactivated FMD vaccines are widely applied to prevent and control FMD [42]. However, as FMDV shows a high degree of genetic and antigenic variation, vaccination with vaccine containing one serotype FMDV does not offer cross-protection to other serotypes [43,44]. Therefore, to protect against each serotype of FMDV, vaccine strains of FMDV for each type must be rapidly developed and implemented. Although a vaccination policy has been implemented to prevent FMD in practice, inactivated FMD vaccines have various drawbacks [45,46]. The shortcomings include the inability to properly stimulate the innate immune response and induce the cellular immune response, the need for regular and repeated vaccinations, the long time required to induce a protective level of vaccine-mediated antibodies, and the inability to induce cross-protection within and between serotypes [47]. Thus, to overcome the limitations of currently available FMD vaccines and improve FMD vaccine efficacy, innovative vaccines are required.

The reported PTC-virus vaccines may be an ideal vaccine type and will likely surmount the drawbacks of the live-attenuated vaccines and the inactivated vaccines [26,27]. The genetic code expansion technology has experienced remarkable progress over the last decade, facilitating its application to various domains of life. In recent years, there have been exciting developments at the interface of genetic code expansion technology and virology [11,13]. Several previous studies have shown that the methods of incorporating genetically encoded UAAs from this discipline are particularly well-suited to viral protein modification, and researchers have engineered live replication-incompetent virus as vaccines [21,24,26,27,29–31]. The PTC-hepatitis D virus, HIV-1, Zika virus and influenza A were created by introducing amber stop codons into viral genomes. These engineered viruses replicated only in cells that could decode the amber codon [21,26,27,34]. In this study, we speculated that if an early translation termination codon at a suitable position is introduced within the FMDV genome, it will block proper translation and assembly of FMDV particles in normal cells. As demonstrated in PTC-influenza A, live but replication-incompetent virus vaccines can elicit robust immunity against both parental and antigenically distinct strains [26].

Our proof-of-concept experiment envisaged introducing amber stop codons into the FMDV genome to generate efficient live-attenuated PTC-viruses as vaccine candidates. We successfully established a stable transgenic cell line, BHK21-tRNA/pylRS/mCherry-TAG-eGFP, containing the orthogonal translation machinery. The stop codon read-through and unnatural amino acid incorporation were verified according to the UAA-dependent eGFP phenotype. Consistent with the report by Elsässer et al. [35], the pylS and pylT expression cassettes were integrated rapidly and efficiently into the genome of BHK-21 cells by PiggyBac transgenesis. FMDV can be packaged in transgenic cells with a high efficiency using an infectious cDNA clone after transfection. Then FMDV titre was almost identical with or without UAA in the transgenic cells and parental cells, and thus demonstrated that these transgenic cells were stable and no effect in replication of FMDV. Based on the above observations, we concluded that there are no detectable detrimental effects of the orthogonal tRNA/pylRS pair on the packaging and production of the FMDV in transgenic BHK-21-tRNA/pylRS/mCherry-TAG-eGFP cells compared with the parental cells.

The FMDV 3D^pol was firstly selected to investigate the feasibility of constructing infectious PTC-FMDV using genetic code expansion technology. We constructed 25 PTC-FMDV mutants (3D-1) by introducing a single amber stop codon into the 3D^pol gene. These 25 PTC-viruses encoding plasmids were transfected into transgenic cells individually. Similar to the report for influenza A [26], the production and packaging of the majority of PTC-FMDV were slower than with wild-type virus. There were 17 CPEs observed in the presence of UAA, indicating that these amino acid sites substituted with amber stop codons did not have any negative effects on 3D^pol function. However, other eight mutants were could not produce active viruses, likely suggesting that these mutated sites are critical for virus replication. The amino acid codon substitutions disrupted the function of 3D^pol. However, all of these amber stop codons mutated to sense codons at the first or second passage, which suggested that the substitution sites of these amber codons in the 3D^pol gene was genetically unstable.

Due to the RNA virus genome is susceptible to mutations, that may result in the genetic instability of amber stop codons introduced [48]. We suppose that may be associated with the conservation of different virus genome regions. The evolutionary conservation of amino acid residues in FMDV proteins was analysed according to ConSurf [41]. We generated 104 PTC-FMDV mutants by substituting 104 codons in four different proteins of FMDV (L, VP1, VP4 and 3D), most of which were located at conserved amino acid sites. The majority of PTC-FMDV were observed to have similar levels of packaging efficiency relative to those 3D-1 mutants. These results are virtually identical to those of the influenza A study [35]. The conservation level of amino acids in each viral protein had no obvious correlation with the PTC virus packaging. The CPE rate of PTC-FMDV was different in each FMDV protein mutant. However, unexpectedly, these amber stop codons showed very high escape frequencies, losing UAA dependency in the first or second passage, which could be ascribed to mutation of PTC codons as verified by gene next-generation sequencing (Figure S5). To further confirm and clarify mutant sites’ stability, VP1-N46 mutants were selected by virus plaque purification. The mutation of the verified seven plaque viruses also appeared at the mutant locus at the third passage. This mutation also exists in the influenza A study. Six amber codons in the influenza genome showed relatively high escape frequencies. Finally, they lost UAA dependency, but the PTC-virus escape frequency was decreased by introducing multiple amber stop codons at different sites [26]. The possibility for the virus to recover all amber mutations is low; however, increasing the number of amber codons may affect the PTC-virus packaging efficiency. The rescue result for the dual-PTC-FMDV indicated that the packaging efficiency of the PTC-FMDV was significantly decreased.

It has been demonstrated in studies of influenza A and HIV-1 that the UAAs could be used to control the multicycle replication of PTC-viruses via genetic code expansion. However, the genetic stability of amber codons was not reported during the in vitro proliferation of PTC-HIV vaccine candidate, a systematic evaluation of genetic stability was not conducted [26,27,29]. The problem of escape mutation of amber codons was mentioned in the study of influenza A and Zika virus [26,34]; therefore, the amber codons in recombinant PTC-viruses had a probability to be substituted by other sense codons during the successive passaging with the presence of UAAs. In our study, a total of 129 single-PTC FMDV mutants and four dual-PTC FMDV mutants were constructed by substituting one amino acid codon in the viral genome with an amber stop codon, followed by the detection of the packaging efficiency of these PTC-viruses and the genetic stability of the amber stop codon. After transfecting the transgenic cell lines with individual FMDV mutant packaging plasmids, PTC-FMDVs were successfully rescued. However, we found that the amber stop codons introduced into the FMDV genome lacked genetic stability, and as almost all of them could mutate to the sense codons during packaging and replication of FMDV. The amber codons were inserted in different sites in FMDV genome (including variable, average and conserved domains). The mutations randomly occurred during virus replication and were not related with conservation level of certain amino acid or protein. Our results also showed similar problems of the genetic instability of PTC-viruses compared with previous studies [26,34]. One possible explanation is that the FMDV of the Picornavirus family has a relatively high genetic variability under survival pressure. The amber codons installed into the viral genome are more likely to mutate to a sense codon to regain viral functional replication. Another possibility is that the function of RdRp is critical not only for the virus life cycle but also for its adaptive potential. The combination of low fidelity to RNA virus replication, the absence of proofreading and excision activities within the RdRPs resulting in a dynamic genome population that is constantly selecting for beneficial mutations during infection, and a high mutation frequency allow RNA viruses to adapt to changing environments rapidly [38,48,49].

The reversion and mutation will cause safety concerns if PTC-viruses are used as vaccines [48–50]. Here, we also observed such phenomena in many in vitro experiments, and this might become more likely in a longer time span when the PTC-virus replicates in vivo. While our results suggest that this approach can construct PTC-viruses and has potential safety concerns regarding the design for other picornaviruses, there is a need for further optimization. From this perspective, our study and the previous reports on potential “general’ methods for PTC-virus vaccines suggest that extrapolation from one species to another could be risky. This surprising result could be attributed to the FMDV, which is different from other viruses, such as influenza virus and HIV, in the genome structure or virus replication cycle. Moreover, as the read-through of stop codons by endogenous processes occurs in some viral genes and certain cell types, our understanding of this phenomenon is still incomplete for mammalian genes [51]. The use of UAAs to control an organism’s viability provides a strategy for the development of conditional ‘kill switches’ for live vaccines. This strategy has been applied to the development of live attenuated vaccines, and it offers an attractive feature for controlling the viability of the virus [11,13]. However, mutations at permissive sites typically suffer from high escape frequencies [52,53]. Therefore, this promising approach should be used with caution [11]. In future studies, the feasible mutation site combinations containing multiple UAG codons will be screened and evaluated. Quadruplet decoding methods could also be used to further explore and improve the genetic stability of this system. Another possible approach would be making the functions of essential proteins explicitly dependent on the identity of an encoded UAA rather than on amber suppression per se. This would improve the safety of strategies for creating attenuated pathogens [11,54,55].

In summary, this study is the first to apply the genetic code expansion technology to the assembly process of FMDV, and establishes proof-of-concept for a reliable production platform to generate UAA-controlled FMDV for use in a potentially large variety of pathogenic study and vaccine design applications. Our results indicated that the UAA-mediated amber suppression strategy could be used to produce PTC-FMDV and the resulting infectious virus. However, both single- and dual-PTC lack sufficient genetic stability during FMDV packaging and replication, suggesting that the amber stop codon introduced into the FMDV genome is susceptible to mutation and thereby loses its dependence on UAAs. Therefore, it should be cautious to apply this approach to developing new FMD vaccines. Our study may also provide a valuable reference for future research in other Picornaviridae viruses, and highlighting the possibility and challenge for the future use of this emerging technique.

Materials and methods

Biosafety statement and facility

All experiments with rescue of recombinant PTC-FMDV were carried out in a biosafety laboroty-3 level (BSL-3) in the Lanzhou Veterinary Research Institute (LVRI), Chinese Academy of Agricultural Sciences (CAAS) accredited by China National Accreditation Service for Conformity Assessment (CNAS) and approved by the Ministry of Agriculture and Rural Affairs. In the laboratory, to reduce any potential risk, the protocols must be strictly followed, and all activities are monitored by the professional staff at LVRI and randomly inspected by local and central governmental authorities without notice in advance.

Cell culture

The baby hamster kidney cells (BHK-21; Chinese Type Culture Collection) were cultured at 37 °C with 5% CO₂ in Dulbecco’s Modified Eagle Medium with 4500 mg/L glucose, L-glutamine, sodium pyruvate, and 25 mM HEPES (DMEM; Gibco), supplemented with 10% foetal bovine serum (FBS), penicillin 100 IU/ml and streptomycin 100 mg/ml.

Plasmid construction

The PiggyBac plasmid pPB-EF1a-eGFP-puro and PiggyBac Transposase Plasmid pPBase were kindly provided by Professor Sen Wu from the State Key Laboratory of Agrobiotechnology, College of Biological Sciences, China Agricultural University, Beijing, China.

The Methanosarcina mazei pyrrolysyl-tRNA synthetase (PylRS)/tRNA_CUA pair (MmpylRS/tRNA_CUA) [35] was used for site-specific incorporation of the unnatural amino acids (UAAs) N^ε-tert-Butoxycarbonyl-L-lysine (BocK) (CAS:2418–95-3) or N^ε-Benzyloxycarbonyl-L-lysine (Lys(Z)) (CAS:1155–64-2) and N^ε-2-azidoethyloxycarbonyl-L-lysine (NAEK) (CAS:1,167,421–25-1).

The construction of all of the plasmids in the current study was carried out according to standard procedures, and recombinant plasmids were transformed and maintained in Escherichia coli (E. coli) DH5α. Two plasmids PB2 (pPB-4 × PylT/PylS-IRES-puro) and PB4 (pPB-4 × PylT/mCherry-TAG-eGFP-IRES-neo), were constructed according to the method described by Elsässer et al. [35,36]. First, pPB-4 × PylT/EF1a-eGFP-puro plasmid was obtained by replacing the EF1a promoter in a pPB-EF1a-eGFP-puro vector with the synthetic DNA fragment 4U6-PylT/EF1a (BGI, Beijing); then, the PB2 plasmid was constructed by inserting a synthetic fragment containing PylRS-IRES (BGI, Beijing) into a pPB-4× PylT/EF1a-eGFP-puro vector to replace the eGFP gene and SV40 promoter. Next, an amber stop codon between mCherry and eGFP was indicated as ‘TAG’. The mCherry-TAG-eGFP-IRES-neo gene fragment was synthesized (BGI, Beijing) and cloned into the pPB-4× PylT/EF1a-eGFP-puro plasmid to substitute the eGFP-SV40-puro fragment. The mCherry-TAG-eGFP was expressed under an EF1a promoter to obtain the PB4 plasmid that confers neomycin resistance. Plasmids PB2 and PB4 with complementary selection markers were used to integrate both the pylT and PylS gene and the gene of interest (here the double-fluorescent reporter mCherry-TAG-eGFP-HA for the generation of cell lines). An N-terminal FLAG tag was added to PylRS. The double fluorescent reporter carried a C-terminal HA tag for detecting the full-length product. The vector backbone contained a tandem cassette of four PylT genes driven by a U6 Pol III promoter, a strong EF1-α Pol II promoter, and IRES driving a Puro or Neo resistance gene. All of the plasmids were confirmed by next-generation sequencing (AuGCT, Beijing).

The 1-plasmid FMDV rescue system (pFMDV) was preserved in our lab [56–58]. FMDV mutant plasmids (pFMDV-L-TAG, pFMDV-VP1-TAG, pFMDV-VP4-TAG, and pFMDV-3D-TAG) containing amber stop codons within the open reading frame were obtained from the wild-type plasmids (pFMDV) via site-directed mutagenesis (NEB) and confirmed by next-generation sequencing (AuGCT, Beijing).

According to the manufacturer’s instructions, all of the plasmids used for transfection were amplified using jetPRIME (Polyplus transfection).

Establishment of the transgenic cell line BHK-21-tRNA/pylRS/mCherry-TAG-eGFP

BHK-21 cells were cultured in DMEM medium (Gibco) supplemented with 10% FBS (Gibco). Sub-confluent BHK-21 cells in six-well plates were co-transfected with 1 µg of pPBase plasmid, 2 µg of PB2, and 2 µg of PB4 plasmid using the transfection reagent jetPRIME (Polyplus transfection). Then, 6 h later, the transfection medium was replaced by DMEM medium supplemented with 10% FBS, and 1 mM NAEK. After 24 h post-transfection, using the mCherry-TAG-eGFP reporter, all of the transfected cells expressed mCherry (mCherry+/eGFP-), but only cells capable of amber suppression expressed the mCherry-eGFP fusion (mCherry+/eGFP+). Next, 48 h after transfection, the cells were split 1:6 into six wells of the original size, and the selection was begun under the pressure of 5 μg/mL puromycin (Amresco) and 1000 µg/mL G418 (Thermo Fisher). The cells were maintained under selection for at least 7 days to ensure stable integration, and the medium was replaced every 2–4 days. Then, surviving cells were collected from wells with the highest drug concentration and expanded. Single clones were derived from a polyclonal pool by single-cell sorting or serial dilution. In the presence of UAA, the stable integration cells, BHK-21-tRNA/pylRS/mCherry-TAG-eGFP, were selected according to the eGFP phenotype and verified by their dependence on UAA for eGFP expression.

Generation of wild-type FMDV and FMDV mutants harbouring amber stop codon(s) in their genome

For the generation of wild-type FMDV, 2 × 10⁵ cells per well (six-well plate) from the BHK-21-tRNA/pylRS/mCherry-TAG-eGFP cell line were seeded in DMEM supplemented with 10% FBS 24 h before transfection. The FMDV rescue system of 2 µg plasmids was transfected into the cells using the transfection reagent jetPRIME (Polyplus transfection) according to the manufacturer’s instructions. Six hours later, the medium containing the mixture of plasmids and jetPRIME reagent was replaced with DMEM supplemented with 1% FBS. The cells were further incubated at 37 °C in 5% CO₂ until a > 90% cytopathic effect (CPE) was observed, and the supernatant containing the generated virus was harvested after freezing and thawing three times, then centrifuged at 6000 × g for 10 min at 4 °C to remove contaminating cells [56].

To generate FMDV mutants harbouring amber stop codon(s) in their genome, an almost identical procedure was carried out, with the following changes: The plasmid expressing wild-type viral RNA was replaced by the corresponding mutant plasmid, and the medium was further supplemented with 1 mM UAA, e.g. N^ε-2-azidoethyloxycarbonyl-L-lysine (NAEK). To identify the UAA-dependent viral strains, a parallel packaging experiment was conducted in which the medium was not supplemented with UAA.

Evaluation of relative packaging efficiency

To explore the effect of the amber codon introduction on PTC-FMDV proliferation at different test sites located in variable, average or conserved domains in different viral protein, the relative packaging efficiency represented as normalization of the days required for formation the typical CPE at each test site was compared to the wild-type FMDV and evaluated according to the described method [26]. Relative packaging efficiency (%) = days of typical CPE in wild-type FMDV/days of typical CPE in PTC-FMDV.

Virus TCID₅₀ assay

BHK21 cells or transgenic BHK21-tRNA/pylRS/mCherry-TAG-eGFP cells were cultured in a 96-well plate to produce a confluent monolayer, and inoculated in eight replicates with 10-fold serial dilutions (100 μL) of virus. Cells were incubated at 37 °C for 48–72 h. Virus titres were calculated by determining the dilution that resulted in 50% of wells containing cells that displayed a cytopathic effect. Virus titres were calculated by the Reed-Muench method [59] and were expressed as median tissue culture infective doses (TCID₅₀) per 0.1 ml.

Amplification and sequencing for FMDV mutant

BHK-21-tRNA/pylRS/mCherry-TAG-eGFP cells at 5 × 10⁶ cells per well in 60-mm dishes were infected with mutant FMDV strains in DMEM supplemented with 1% FBS and 1 mM NAEK. When >90% CPE was observed, the supernatants were collected and used for infection in the next round of investigation. A parallel experiment in which the medium was not supplemented with NAEK was conducted to detect the viral UAA-dependency. After each passage, viral RNA was isolated from 200 µL cell supernatants using E.Z.N.A.® Viral RNA Kit (Omega Bio-Tek, Guangzhou, China), according to the manufacturer’s specifications. Next, RNA was used for amplifying different specific FMDV gene segments by RT-PCR using PrimeScript ® One Step RT-PCR Kit Ver.2 according to the manufacturer’s specifications (Takara, Dalian, China). The PCR conditions were 1 cycle at 50 °C for 30 min and at 94 °C for 2 min, followed by 30 cycles at 94 °C for 30 sec, 68 °C for 30 sec, 72 °C (1 min/kb), and finally 1 cycle at 72 °C for 5 min. The resulting PCR products were sequenced to investigate whether any mutation occurred during viral passages.

Quantitative reverse transcription-polymerase chain reaction (qRT-PCR)

To detect the pylRS and pyltRNA relative expression levels in different cell clones of BHK-21-tRNA/pylRS/mCherry-TAG-eGFP, total RNA was isolated using TRIzol (Invitrogen, Carlsbad, CA, USA) from different clonal cell lines derived from a host capable of amber suppression. RNA samples were analysed by one-step qRT-PCR using a One Step PrimeScript RT-PCR Kit (Perfect Real Time) according to the manufacturer’s specifications (Takara, Dalian, China). The primers and probes of pylRS and pyltRNA used in this study were based on reported sequences [60]. The specific primer and probe for BHK-21 Beta-2 microglobulin (BHK-21 B2M) were designed by entering a single BHK-21 B2M exon from GenBank XM_005068531.3 into Genscript’s Real-time PCR (TaqMan) Primer and Probes Design Online Tool.

All of the assays were performed on the CFX96 Touch Real-Time PCR instrument (Bio-Rad, USA) using a ‘quick 96-well plate’ with each reaction containing 5 ng RNA in a 20 μl reaction volume. Thermocycler conditions were 1 cycle of 50°C for 15 min for Reverse Transcription then 1 cycle of 94°C for 2 min followed by 40 cycles of 94°C for 15 sec, 60°C for 15 sec and 72°C for 30 sec. All of the samples were analysed in triplicate. The result of RNA expression levels is referred to as expression ratios calculated using the following formula: RNA expression ratio = 2^−ΔCT where ΔCT = (Target Mean Ct)-(B2M Mean Ct) [60].

Western blotting

Cells were grown in six-well plates in media containing DMEM with 10% FBS. Cultured cells were washed once with PBS buffer and lysed in RIPA lysis buffer (Solarbio, China) containing complete protease inhibitor for 15 min. The protein concentrations were determined by the BCA method (Thermo Scientific). The pylRS blots were probed with mouse monoclonal antibodies against Flag (Sigma, St. Louis, MO, USA) at a 1:5000 dilution; eGFP blots were probed with anti-HA antibody (Covance, Emeryville, CA, USA) at a 1:4000 dilution, and β-actin blots were probed with anti-β-actin antibody (Santa Cruz, CA, USA) at a 1:2000 dilution. Then, the above blots were probed with the corresponding secondary HRP-linked antibodies and detected with Immobilon™ western HRP substrate (Millipore).

Funding Statement

This work was supported by the the Chinese Academy of Agricultural Science and Technology Innovation Project [CAAS-ASTIP-2020-LVRI]; the Frontier Exploration Project of Chinese Academy of Agricultural Sciences [1610312017003]; the National Key R&D Program of China [2018YFD0500103]; National Natural Sciences Foundation of China [31672585].

Disclosure statement

The authors declare that they have no conflict of interest.

Author contributions

Rongzeng Hao, Yi Ru, Dan Li and Kun Ma designed experiments. Rongzeng Hao, Kun Ma, Gao yuan Song, Bingzhou Lu, Huanan Liu, Yajun Li, Jiaoyan Zhang, Chunping Wu, and Guicai Zhang performed experiments and analyzed the data. Rongzeng Hao wrote the original draft manuscript. Haixue Zheng, Haitao Hu and Jianxun Luo conceived and supervised the study. All authors read and approved the final version of the manuscript.

Supplemental material

Supplemental data for this article can be accessed here.