Generation of influenza A viruses as live but replication-incompetent virus vaccines

See allHide authors and affiliations

Science  02 Dec 2016:
Vol. 354, Issue 6316, pp. 1170-1173
DOI: 10.1126/science.aah5869

Protecting by changing the code

Live attenuated vaccines can be very potent, but their potential to revert to their pathogenic form limits their use. In an attempt to get around this, Si et al. expanded the genetic code of influenza A viruses. They propagated viruses that were mutated to encode premature termination codons (PTCs) in a cell line engineered to be able to express these flu proteins. Despite not being able to replicate in conventional cells, PTC-containing viruses were highly immunogenic and protected mice, guinea pigs, and ferrets against influenza challenge.

Science, this issue p. 1170


The conversion of life-threatening viruses into live but avirulent vaccines represents a revolution in vaccinology. In a proof-of-principle study, we expanded the genetic code of the genome of influenza A virus via a transgenic cell line containing orthogonal translation machinery. This generated premature termination codon (PTC)–harboring viruses that exerted full infectivity but were replication-incompetent in conventional cells. Genome-wide optimization of the sites for incorporation of multiple PTCs resulted in highly reproductive and genetically stable progeny viruses in transgenic cells. In mouse, ferret, and guinea pig models, vaccination with PTC viruses elicited robust humoral, mucosal, and T cell–mediated immunity against antigenically distinct influenza viruses and even neutralized existing infecting strains. The methods presented here may become a general approach for generating live virus vaccines that can be adapted to almost any virus.

The greatest challenge for converting viruses— such as those responsible for influenza, Ebola, and AIDS pandemics—into live whole-virion vaccines is to render them as avirulent as possible while maintaining their full infectivity to elicit sufficient immunity (14). However, inactivated and attenuated virus vaccines against influenza often exhibit a decrease or even loss of productivity and efficacy (5). In addition, immune escape due to antigenic drift and shift (5, 6) introduces a further challenge for the efficacy of conventional influenza vaccines. The modern technology of genetic code expansion (7, 8) and its application (913) to the viral genome may provide the potential to generate live but replication-incompetent virus vaccines eliciting strong and broad immunity (Fig. 1A). A key challenge in designing such a vaccine is to maintain the reproductive potential of the progeny viruses during vaccine production, which requires a special packaging cell line.

Fig. 1 Establishment of a virion packaging system that is compatible with the orthogonal translation machinery.

(A) Schematic representation of the generation of premature termination codon (PTC) influenza viruses that are characterized by replication incompetence in conventional cells but high reproduction in transgenic cells that contain integrated cassettes for the expression of orthogonal tRNA (tRNACUA), tRNA synthase (pylRS), and a gene encoding an amber codon–containing GFP (GFP39TAG). NP, nucleoprotein; PB1 and PB2, polymerase basic proteins 1 and 2; PA, polymerase acidic protein; M, matrix protein; NS, nonstructural protein; HA, hemagglutinin; NA, neuraminidase.(B) Characterization of the transgenic cells and their genetic stability by constitutive expression of GFP in the presence of UAA (top) and expression of pylRS (middle) and tRNA (bottom) (N = 3). (C) Functional evaluation of the effect of the orthogonal pylRS/tRNA pair on the propagation of the wild-type WSN viruses by comparing the parental HEK293T and transgenic tRNA/pylRS/GFP39TAG cells in the presence or absence of UAA (N = 3). (D) Characterization of the infectivity of the progeny viruses by the CPE assay and the replication incompetence in the presence of UAA by comparing the transgenic and parental cells (N = 3). (E) Reproduction and stepwise accumulation of the PTC viruses in transgenic cells; 10 ± 3%, 50 ± 5%, and 99 ± 1% CPE were observed at days 3, 4, and 5 after inoculation, respectively, conditional on the presence of UAA (N = 3). Error bars denote SEM.

We first tested the compatibility of the orthogonal translation system—the Methanosarcina barkeri MS pyrrolysyl tRNA synthetase/tRNACUA pair (MbpylRS/tRNACUA) and the orthogonal unnatural amino acid (UAA) Nε-2-azidoethyloxycarbonyl-l-lysine—with the viral packaging cells. Human embryonic kidney (HEK) 293T cells were transduced with lentivector pSD31s (14) for integration of the genes encoding MbpylRS and an amber codon–containing green fluorescent protein (GFP39TAG) into the host genome (fig. S1). The resultant transduced cells were stably transfected with the bjmu-12t-zeo vector (fig. S1), which harbored 12 tandem tRNA-expression cassettes (15, 16). The final transgenic cells, HEK293T-tRNA/pylRS/GFP39TAG, were selected according to the UAA-dependent GFP phenotype and verified by the constitutive expression of the orthogonal tRNA/pylRS pair as tested in their 200th generation (Fig. 1B). The capability of the transgenic cells to propagate viruses was tested in parallel with parental cells by transfection with the packaging plasmids of influenza A/WSN/33 (H1N1; WSN) virus (Fig. 1A) (17). The plaque formation assay (18, 19) indicated almost identical viral titers and thus identical packaging efficiency from each transfection (Fig. 1C).

We then randomly selected one codon in the gene encoding the viral NP protein, Asp101, for amber codon replacement via site-directed mutagenesis. The mutant viral genome was reciprocally packaged by the transgenic and parental cells to generate the PTC viruses (Fig. 1A). The production and infectivity of the putative PTC viruses were verified by the cytopathic effect (CPE) assay (18, 19). The CPE phenotype was observed in the transgenic cells only in the presence of 1 mM UAA (Fig. 1D). Clearly, the generation and stepwise reproduction of PTC viruses, as shown in the time-course of CPE accumulation and viral growth curve, occurred only in the transgenic cells and not in the conventional cells (Fig. 1E and fig. S2). This result suggests that PTC viruses had a higher level of safety (i.e., the viruses were replication-incompetent in conventional cells) than did the clinically used cold-adapted live attenuated influenza vaccine (CAIV) (fig. S2) and the attenuated viruses by codon deoptimization, which were still able to replicate in conventional cells and kill the mice at a high dose (2022).

One concern over such a design is the potential reversion of the amber codon to a sense codon during PTC virus replication and propagation. To generate genetically stable PTC viruses, we individually replaced 21 extra codons across the NP gene, most of which were located at conserved sites (23), with an amber codon (Fig. 2A and figs. S3 to S5). Seven of them—NP-Asp101, NP-Gly102, NP-Gly126, NP-Asp128, NP-Arg150, NP-Met163, and NP-Gly169—caused UAA-dependent CPE, and such dependence was stable with a low escape frequency (7 × 10−10 to 5.9 × 10−7) even over 20 passages (Fig. 2A, fig. S4, and table S1). In addition, the amber codon substitutions at Asp101, Gly102, and Met163 had much less effect on virus packaging efficiency and replication kinetics relative to the wild-type virus (Fig. 2A and fig. S6).

Fig. 2 Genome-wide investigation and characterization of the PTC influenza viruses.

(A) Systematic exploration of the effect of the introduction of the amber codon at different test sites located in variable, average, or conserved domains, based on the ConSurf analysis (23), on PTC virus production. Relative efficiency represents a normalization of the days required to attain ~100% CPE at each test site compared to the wild-type WSN virus. (B) Multicycle growth kinetic curves of different PTC viruses in transgenic and conventional cells in the presence of UAA. (C) Escape frequencies of different PTC viruses at the first and 20th passage. (D) Comparisons of the morphologies of the PTC viruses and the wild-type WSN by transmission electron microscopy Scale bars, 100 nm. (E) Direct observations of the UAA-dependent plaque phenotypes of PTC viruses; N = 3. Abbreviations for amino acid residues: A, Ala; D, Asp; G, Gly; H, His; K, Lys; L, Leu; M, Met; N, Asn; R, Arg; S, Ser; Y, Tyr. Error bars denote SEM.

Using the same strategy, we systematically explored the following codons for the generation of replication-incompetent viruses: 22 codons in PB1, HA, NA, and NS; 8 codons in PA, PB2, and M2; and 14 codons in M1 (fig. S3). Different degrees of packaging and propagation efficiency in transgenic cells were observed (figs. S4 to S6 and table S1): 11 UAG codon mutations in PB1, 10 in NA, 5 in HA and PB2, 18 in NS, 6 in PA and M2, and 0 in M1 caused clearly UAA-dependent CPE (figs. S4 and S5). As expected, most of them replicated only in the transgenic cells in the presence of UAA with various replication kinetics and low escape frequencies (<10−6) (fig. S6 and table S1). However, six of them (HA-Lys57, NS-Phe103, PB2-Gln13, PB2-Thr35, M2-Lys49, and M2-Lys60) showed relatively high escape frequencies (3.5 × 10−4 to 8.9 × 10−3) and finally lost UAA dependency, which could be ascribed to mutation of PTC codons as verified by sequencing (fig. S5 and tables S1 and S2). Given the odds of reversion during viral propagation, multiple UAG codons, from two to eight, were introduced into viral genome with each at one RNA segment. The packaging and propagation efficiency of the resultant PTC viruses generally decreased after the introduction of more PTCs into the viral genome (Fig. 2B and table S1). Their escape frequencies also decreased: 1.0 × 10−8 for mono-PTC virus (PTC-1), 1.2 × 10−10 for dual-PTC virus (PTC-2), and undetectable (<10−11) for strains containing three or more stop codons (Fig. 2C and table S1).

We next explored the in vivo safety of PTC viruses by intranasally infecting BALB/c mice, ferrets, and guinea pigs. PTC-4A was chosen as a representative (Fig. 2, D and E) because it maintained the intact surface antigens by harboring four stop codons in PA, PB2, PB1, and NP, rather than touching envelope HA and NA genes (figs. S7 and S8 and table S1). In the mouse model study, the median lethal dose (LD50) of wild-type WSN virus was 8 × 103 plaque-forming units (PFU), and 10 × LD50 of wild-type viruses killed all mice, with a significant loss of body weight preceding death (fig. S9A). In contrast, no mice were killed by PTC-4A even at a dosage of 109 PFU, a factor of 105 higher than the LD50 of the wild-type viruses; no body weight loss or other health issues were observed (Fig. 3A). Detection of viruses in the turbinate, trachea, or lung of mice 3 days after inoculation of 105 PFU of viruses indicated that the mean viral titers were 101.6, 101.5, and 101.3 PFU/g (turbinate, trachea, and lung, respectively) for the PTC group, 103.5, 102.9, and 101.8 PFU/g for the CAIV group, and 105.5, 105.7, and 106.5 PFU/g for the wild-type group (Fig. 3B). Unlike the rescued wild-type viruses, no plaque was observed for the rescued PTC viruses from mice in the absence of UAA (Fig. 3C), which suggests that PTC reversion did not occur in vivo. In addition, respiratory droplet transmission experiments indicated that PTC viruses were not detected in noninoculated guinea pigs when caged together with inoculated guinea pigs, in contrast to the findings for wild-type viruses and CAIV (fig. S9B). All data indicated that influenza viruses have been rendered avirulent via the introduction of amber codons, as confirmed by ferret and guinea pig model studies (fig. S9, C and D).

Fig. 3 Characterization of the in vivo safety of the PTC viruses.

(A) Effect of intranasal virus infection with the indicated viruses or vaccines on the survival rates and body weights of BALB/c mice (n = 10). Error bars denote SEM. (B) Detection of the virus titers in mouse tissues (n = 5) on day 3 after infection with 105 PFU of the wild-type, CAIV, or PTC-4A viruses. Data are plotted for individual mice (n = 5) and overlaid with means ± SD. *P < 0.05, **P < 0.01, ***P < 0.001 [one-way analysis of variance (ANOVA) with Newman-Keuls multiple-comparisons test]; n.s., not significant. (C) Plaque phenotypes of the rescued wild-type and PTC-4A viruses in mice.

Next, we tested the immunogenicity of PTC-4A by comparison with the commercially available inactivated influenza vaccine (IIV) and CAIV. Three weeks after immunization of 106 PFU for mice and 107 PFU for ferrets and guinea pigs, PTC-4A and CAIV, both as whole virions, induced robust antibodies in sera according to hemagglutination inhibition (HAI) and neutralization (NT) antibody assay, but IIV did not do so, even after the second vaccination (Fig. 4A and figs. S10 to S12). Impressively, the second vaccination by PTC-4A caused the HI and NT antibody titers to increase further by a factor of approximately 6 to 8, comparable to the effect of CAIV. Such a PTC virus–mediated immune response was also observed against internal NP antigen; this effect was not observed for IIV even after a second vaccination (Fig. 4B and fig. S11A). Furthermore, a high level of secretory immunoglobulin A (IgA), representing the mucosal immune response, was elicited in the lungs by both PTC-4A and CAIV but not by IIV vaccine (Fig. 4A and fig. S11A). Moreover, approximately 10 times as many virus-specific cytotoxic T lymphocytes (CTLs) were elicited by PTC-4A and CAIV than by IIV, which only elicited a basic level of CTLs in the lungs (Fig. 4A). Our results indicate that the PTC virus vaccines elicit robust humoral, mucosal, and cell-mediated immunity, which may be due to their full infectious form carrying the native conformation of all surface antigens and internal viral components.

Fig. 4 Characterization of the immunogenicity and protective efficacy of the PTC-4A virus in BALB/c mice that were intranasally inoculated with one or two doses.

(A) Antibody responses (including serum IgG and mucosal IgA) and virus-specific CD8+ T cell responses, induced by PTC-4A (n = 5). Error bars denote SD. (B) Mouse serum IgG antibody responses to the influenza internal protein NP (n = 5). Error bars denote SD. (C) Plaque titration of the WSN viruses in the lungs (n = 5) to compare the protective efficacy of the PTC-4A virus versus IIV or CAIV at one or two doses. Error bars denote SD. (D) Characterization of the protective efficacy of the PTC-4A viruses in terms of survival rates and body weights of the vaccinated mice (n = 10). Error bars denote SEM. *P < 0.05, **P < 0.01, ***P < 0.001 (one-way ANOVA with Newman-Keuls multiple-comparisons test).

The protective efficacy of PTC-4A virus was tested and compared with that of CAIV and IIV. Three weeks after vaccination, 15 mice in each group were challenged intranasally with 50 × LD50 of wild-type viruses. Three days after challenge, five mice from each group were killed for plaque titration of wild-type viruses in the lungs (Fig. 4C). We found a significant decrease of viral titers in both PTC-4A and CAIV groups relative to the IIV group, in particular for dual vaccinations, which led to a further decrease of viral titers by a factor of 103 for both PTC-4A and CAIV. In terms of survival rate and body weight for the remaining mice, all mice in the vehicle group succumbed by 9 days after challenge, whereas all mice survived after immunization even with just one dose of PTC-4A or CAIV (Fig. 4D). The mice inoculated with one dose of PTC-4A regained their body weight 5 days after challenge, whereas no weight loss was observed in mice inoculated with two doses of PTC-4A or CAIV, which we also confirmed in ferret and guinea pig model studies (figs. S11 and S12).

We also tested whether the PTC viruses exhibit heterologous protection against antigenically distant influenza strains. Fifteen mice, vaccinated with a single dose or two doses of PTC-4A, were challenged with 106 PFU of influenza A/reassortant/NYMC X-179A (pH1N1) or A/Aichi/2/68 (H3N2) viruses. Significant heterologous protection was observed, according to plaque titration and observations of the survival rates and body weight changes (fig. S13). Such heterologous protection was ascribed to the conserved regions of viral proteins, including surface and internal antigens among distant influenza viruses, and to virus-specific CTLs that collectively contribute to the broad protective effect (5, 2426).

We then tested whether the existing viral virulence of PTC viruses could be enhanced by co-propagating wild-type and PTC viruses in conventional cells. Rather than more plaques, we observed remarkably fewer plaques or even undetectable plague formation, depending on the ratio of wild-type versus PTC viruses and the number of PTC codons in the PTC viruses (fig. S14, A and B). This contrasted with CAIV, which exerted no inhibitory effect on the propagation of wild-type viruses. Furthermore, the capability of attenuating wild-type viruses by PTC viruses was verified by in vivo experiments (fig. S14C). This was due to the genetic reassortment between the existing and PTC viruses, as verified by sequencing (fig. S14D). Progeny reassortants harbored at least one replication-incompetent gene segment, providing an effective approach for “neutralization” of infectious viruses.

Our results show that we succeeded in generating live but replication-incompetent virus vaccines by applying genetic code expansion to the influenza virus genome. Such live vaccines elicited robust immunity against both parental and antigenically distinct strains. Generation of such PTC virus vaccines can be potentially adapted to almost any virus (27) so long as their genome could be manipulated and packaged in a cell line. Furthermore, the multiple PTC-harboring viruses are not only prophylactic but also therapeutic vaccines in the neutralization of the replicating viruses.

Supplementary Materials

Materials and Methods

Figs. S1 to S15

Tables S1 and S2

References (2841)

References and Notes

Acknowledgments: We thank Sinovac Biotech Ltd. (Beijing) for technical assistance with the animal work. Supported by National Natural Science Foundation of China grants 81530090 and 21572015 and by National Key Research and Development Program of China grant 2016YFA0501500. All authors declare no competing financial interests. D.Z. and L.S. are inventors on a patent application (PCT/CN2016/092778) held by Peking University that covers the preparation of PTC viruses through genetic codon expansion. Sharing of materials will be subject to standard material transfer agreements. The nucleotide sequences used in the study have been deposited in GenBank under accession numbers CY034139.1, CY034138.1, X17336.1, HE802059.1, CY034135.1, CY034134.1, D10598.1, and M12597.1; additional data are presented in the supplementary materials.
View Abstract

Stay Connected to Science

Navigate This Article