Development of a Mouse-Adapted Live Attenuated Influenza Virus That Permits In Vivo Analysis of Enhancements to the Safety of Live Attenuated Influenza Virus Vaccine

Abstract

The live attenuated influenza virus vaccine (LAIV) is preferentially recommended for use in persons 2 through 49 years of age but has not been approved for children under 2 or asthmatics due to safety concerns. Therefore, increasing safety is desirable. Here we describe a murine LAIV with reduced pathogenicity that retains lethality at high doses and further demonstrate that we can enhance safety in vivo through mutations within NS1. This model may permit preliminary safety analysis of improved LAIVs.

INTRODUCTION

Influenza A virus is a respiratory pathogen that infects through the upper airway and leads to pathology via replication in the lower airway (1). The temperature gradient between these two areas in people enabled the development of the cold-adapted, live attenuated influenza virus vaccine (LAIV [FluMist]) that replicates in the cooler upper respiratory tract to trigger a protective immune response but cannot damage the lower respiratory tract due to the elevated temperatures restricting replication (2). This temperature-sensitive (ts) attenuated (att) phenotype is imparted by five mutations within the viral replicative machinery: namely, PB2 N265S, PB1 K391E, D581G, and A661T, and NP D34G (3, 4). Although this vaccine has an overall acceptable safety profile, it is not approved for use in children under 2 years of age due to concerns about elevated hospitalizations due to wheezing (5, 6). For this reason, it is also not approved for use in asthmatics. Therefore, development of vaccines with increased safety over LAIV is desirable. Currently, no mouse model exists for the adequate assessment of the safety of the LAIV. In experimental animal studies with LAIV, elevated doses of LAIV do not elicit pathology, rendering determination of safety impossible (7–10). Here, we describe a model with which we can assess alterations in vaccine safety.

The parental strain for our vaccine is a well-characterized, murine-lethal strain of influenza A virus (A/Puerto Rico/8/34 H1N1, PR8). We introduced four ts att mutations from LAIV into PR8 (NP D34G is natively present) via site-directed mutagenesis (Agilent) and rescued this virus using plasmid-based reverse genetics techniques (11). This ts att virus (referred to henceforth as “PR8 LAIV”) has been previously characterized in cell culture, but its phenotype in mice was not demonstrated (8). PR8 wild-type (WT) virus has a 50% lethal dose (LD₅₀) in C57BL/6 (B6) mice of 10 to 25 PFU (12, 13). Thus, we sought to ascertain the LD₅₀ of PR8 LAIV (Fig. 1). Groups of mice (n = 5) were intranasally inoculated with 10-fold serial dilutions of PR8 LAIV (10⁶ to 10³ focus-forming units [FFU]/mouse), and signs of morbidity (percent loss in body weight) were monitored daily, with animals that lost greater than 25% of their initial weight being sacrificed (Fig. 1A). While PR8 LAIV was indeed lethal at doses of ≥10⁵ FFU, it exhibited no lethality in this experiment at or below 10⁴ FFU (Fig. 1B). Therefore, by introducing the four remaining ts att mutations of LAIV into PR8, the LD₅₀ shifted to 3.16 × 10⁴ FFU (using the method of Reed and Muench, [14]), >1,000-fold greater than that of the WT. Additionally, consistent with FluMist in humans (10, 15), PR8 LAIV replicated in the airways, albeit to lower levels than WT PR8 (Fig. 1C). It is important to note that, unlike humans, mice show a lower body temperature upon influenza virus infection (16). Therefore, the replication of PR8 LAIV in mouse lungs is fully consistent with the ts phenotype of virus as the lung temperature would drop upon infection, and it also suggests that temperature sensitivity is not likely to be the sole mechanism of attenuation of PR8 LAIV, at least in mice.

FIG 1.

PR8 LAIV displayed the att phenotype. (A and B) Morbidity and mortality of PR8 LAIV. Female 6- to 8-week-old B6 mice were inoculated intranasally with the indicated doses of PR8 LAIV. For 2 weeks postinfection, weight loss (A) (plotted data represent means ± standard errors of the means [SEM]) and survival (B) were monitored daily (n = 5). (C) Replication of PR8 LAIV was limited in vivo. Mice were inoculated as described above with 10⁴ FFU of PR8 (shaded columns) or PR8 LAIV (white columns) (n = 3). At 3 and 6 days postinfection, lung virus titers (FFU/ml) were determined from total lung homogenates on MDCK cells using an immunofluorescence assay (30). Columns represent mean virus lung titers ± standard deviations (SD) from individual mice, and the dotted line denotes the limit of detection (20 FFU/ml). Statistical analysis was performed using the Mann-Whitney test. *, P ≤ 0.05; §, no mice surviving at this time point.

To evaluate the protection conferred by PR8 LAIV vaccination, mice were primed with phosphate-buffered saline (PBS) or the highest dose that caused no overt weight loss (10³ FFU) and 14 days later challenged with 10 LD₅₀ of homologous PR8 (n = 9 to 11) (Fig. 2A to C). Whereas all mice mock immunized with PBS rapidly lost weight and succumbed by day 7 postchallenge, PR8 LAIV-primed mice maintained body weight and survived (Fig. 2A and B). The ability of PR8 LAIV-primed mice to overcome homologous challenge was not surprising, as 1 day prior to challenge, the sera contained high titers of PR8 hemagglutination inhibition (HAI) activity (Table 1), indicative of the induction of virus-neutralizing humoral immunity. This is also exemplified by the lack of detectable challenge virus in the lungs of immunized mice at 3 and 6 days postchallenge, while mock-immunized animals showed challenge virus replication of up to 10⁶ FFU/ml lung tissue (Fig. 2C).

FIG 2.

Homologous and heterologous protection induced by PR8 LAIV. (A to C) Protection from lethal homologous challenge conferred by PR8 LAIV. Female 6-to 8-week-old B6 mice were inoculated intranasally with PBS (black circles) or 10³ FFU of PR8 LAIV (white squares). Two weeks postpriming, mice were challenged intranasally with 10 LD₅₀ of PR8 (H1N1) homologous virus. For 2 weeks postchallenge, weight loss (A) (plotted data represent means ± SEM) and survival (B) were monitored daily (n = 3 to 5). (C) Challenge virus lung replication was inhibited by PR8 LAIV priming. Mice primed and challenged as described above were sacrificed at 3 and 6 days postchallenge, and lung virus titers were determined as in Fig. 1 (n = 3). (D to F) Heterologous protection induced by PR8 LAIV. Female 6- to 8-week-old B6 mice were inoculated as described above. Two weeks postpriming, mice were challenged intranasally with 10 LD₅₀ of X31 (H3N2) heterologous virus. For 2 weeks postchallenge, weight loss (D) (plotted data represent means ± SEM) and survival (E) were monitored daily (n = 4). (F) Heterologous challenge virus lung replication was inhibited by PR8 LAIV priming. Mice primed and challenged as described above were sacrificed at 3 and 6 days postchallenge, and lung virus titers were determined as in Fig. 1 (n = 3). Columns represent mean virus lung titers ± SD from individual mice; the dotted line denotes the limit of detection (20 FFU/ml). Statistical analysis was performed using the Mann-Whitney test. *, P ≤ 0.05; §, no mice surviving at this time point; n.d., no virus detected within lung homogenates.

TABLE 1.

HAI titers from mice vaccinated with LAIV viruses

Priming virusa	Dose, FFU	Geometric mean (SD) HAI titer, reciprocal serum dilutionb:
		PR8	X31
PR8 LAIV	103	260 (190)	<16 (0)
	104	390 (140)	<16 (0)
PR8 LAIV-11C	103	300 (130)	<16 (0)
	104	260 (0)	<16 (0)
	105	1,200 (510)	16 (60)

^aVirus was administered intranasally to anesthetized mice (n = 4 to 5), and sera were collected at 13 days postinfection.

^bEight hemagglutinating units (HAU) of the indicated virus was incubated with 2-fold serial dilutions of the indicated sera.

Current influenza virus vaccines are reformulated each year due to the changing antigenicity of the virus, where virus-neutralizing humoral immunity typically only confers protection against matched influenza virus strains (17). Thus, it is desirable to develop vaccines that target conserved viral epitopes. To ascertain whether PR8 LAIV confers protection against heterologous challenge, mice were vaccinated as described above prior to challenge with 10 LD₅₀ of X31, a recombinant influenza virus that contains the hemagglutinin (HA) and neuraminidase (NA) genes derived from A/Hong Kong/1/1968 (H3N2) and the remaining six segments from PR8 (18) (Fig. 2D to F). Antibodies generated from H1N1 viral infections do not typically neutralize H3N2 isolates (19–22), allowing us to evaluate cross-protective immunogenicity. After challenge with X31 virus, mice from both mock-immunized and PR8 LAIV-immunized cohorts rapidly lost weight postchallenge, likely due to an absence of X31 neutralizing antibodies (Table 1). The loss of body weight in mock-immunized animals progressed to fatality, but all of the PR8 LAIV animals regained weight on day 4, completely recovered by day 7 postchallenge, and survived (Fig. 2D and E). Although the onsets of disease were similar for both groups, viral lung burden was significantly reduced (>3.5 logs) in PR8 LAIV- compared to PBS-immunized mice on day 3 postchallenge (Fig. 2F). This heterosubtypic immunity is consistent with the previously reported ability of LAIV to induce flu-specific, lung-tropic CD8 cytotoxic T cells (23–26).

FluMist is now preferentially recommended over inactivated vaccines for children 2 to 8 years of age (27), but contraindications and safety concerns in children under 2 years of age prevent universal licensure. As a result, it is highly desirable to increase the tolerability of FluMist in order to broaden the target population for this effective and needle-free vaccine. Similarly, development of improved LAIV strains requires analyses to ensure comparable safety with FluMist. However, to the best of our knowledge, no reliable animal models have been created thus far to efficiently evaluate the safety of LAIV. As a proof of concept, we sought to further attenuate PR8 LAIV by introducing three previously described mutations into NS1 (11C) that had been shown to affect viral egress and budding in a temperature-sensitive manner (Fig. 3A) (28). We introduced the three ts mutations into the NS1 gene of WT PR8 (11C) or PR8 LAIV (LAIV-11C) via site-directed mutagenesis and rescued these viruses as described above. We then confirmed the ts phenotype of these viruses by plaque assay and growth kinetics in MDCK cells (Fig. 3B and C). As shown in Fig. 3B, the plaque size of LAIV-11C was reduced compared to that of LAIV; this can likely be attributed to the 11C mutations affecting virus egress/budding (28).

FIG 3.

PR8 11C, LAIV, and LAIV-11C maintained the ts phenotype. (A) Schematic representation of PB2, PB1, and NS1 mutations. Residues of PR8 were mutated to match those identified with 11C, LAIV, or the two in combination (LAIV-11C) to confer ts and att phenotypes. Note that PR8 naturally contains NP G34, and this residue was therefore not modified. (B and C) Modified PR8 viruses demonstrated a ts phenotype in vitro. MDCK cells were infected with PR8, 11C, LAIV, or LAIV-11C and incubated at 33, 37, or 39°C for 3 days. (B) Plaque phenotypes were determined by immunostaining using the NP-specific monoclonal antibody HT103 as described previously (12, 30). (C) Multicycle growth curve experiments were performed with MDCK cells as described previously (30). Mean ± SD values for triplicate infections are plotted; the dotted line denotes the limit of detection (100 FFU/ml).

Consistent with previous results, 11C virus was most drastically attenuated at 39°C compared to WT PR8 (28), whereas LAIV showed severe, slight, and absent attenuation at 39, 37, and 33°C, respectively (8). When 11C mutations were introduced in the context of LAIV (PR8 LAIV-11C), a moderate growth defect compared to PR8 LAIV was observed at 72 h postinfection in MDCK cells at 37°C (P = 0.02; two-tailed Student's t test).

We next sought to evaluate if the ts phenotype of 11C could further attenuate PR8 LAIV in vivo. Mice were infected as described above with serial dilutions of 11C or LAIV-11C (10⁷ to 10³ FFU) (Fig. 4), and body weight and survival were then measured over a 14-day period. As expected, PR8 11C demonstrated an attenuated phenotype (LD₅₀, <10³ FFU) (Fig. 4A and B) compared to WT PR8. Strikingly, B6 mice infected with PR8 LAIV-11C tolerated 10-fold-higher doses than LAIV alone (LD₅₀ of 3.16 × 10⁵ FFU for PR8 LAIV-11C [Fig. 4C and D] versus 3.16 × 10⁴ FFU for LAIV [Fig. 1A and B]). LAIV-11C priming also induced similar HAI activity to PR8 LAIV (Table 1). We then sought to determine whether the ts phenotype of 11C impacted protective efficacy. Mice were primed as described above with a low dose of PR8 LAIV, 11C, or LAIV-11C (approximately 0.1 LD₅₀). We delayed delivery of the challenge virus to 3 weeks postpriming to ensure that any residual innate immune response had waned. Mice were then challenged with lethal doses of PR8 and X31. LAIV-11C protected mice from both lethal homologous and heterologous challenges (Fig. 5A and B and D and E). Additionally, vaccination with PR8 LAIV, 11C, and LAIV-11C led to reduced replication of challenge homologous or heterologous viruses, as measured by lung titers at day 3 or 6 postchallenge (Fig. 5C and E). These data show that, despite enhanced attenuation in vivo, priming with PR8 LAIV-11C confers protection (29).

FIG 4.

Tolerability of PR8 11C and LAIV-11C viruses in mice. Female 6- to 8-week-old B6 mice were inoculated intranasally with the indicated doses of PR8 11C (A and B) and PR8 LAIV-11C (C and D) (n = 4). For 2 weeks postinfection, weight loss (A and C) (plotted data represent means ± SEM) and survival (B and D) were monitored daily.

FIG 5.

Protection conferred by PR8 LAIV and NS1 11C mutant viruses. Female 6- to 8-week-old B6 mice were inoculated intranasally with PBS or approximately 0.1 LD₅₀ of virus as follows: PR8 LAIV (300 FFU), PR8 11C (100 FFU), or PR8 LAIV-11C (3,000 FFU). Three weeks postpriming, mice were challenged intranasally with 10 LD₅₀ of PR8 (H1N1) homologous virus (A to C) or 5 LD₅₀ of X31 (H3N2) heterologous virus (D to F). For 2 weeks postchallenge, weight loss (A and D) (plotted data represent means ± SEM) and survival (B and E) were monitored daily (n = 3 to 5). (C and F) Challenge virus lung replication was inhibited by priming. Mice primed and challenged as described above were sacrificed at 3 and 6 days postchallenge, and lung virus titers were determined as in Fig. 1 (n = 3). Columns represent mean virus lung titers ± SD from individual mice. The dotted line denotes the limit of detection (20 FFU/ml). Statistical analysis was performed using the Mann-Whitney test. *, P ≤ 0.05; n.d., no virus detected within lung homogenates.

Collectively, these results demonstrate that mutations that confer a ts phenotype alone further enhanced the attenuation, while maintaining the immunogenicity of murine-lethal PR8 LAIV. Also, dispersing the genetic att loci across more RNA segments can potentially increase the phenotypic stability of the vaccine. Thus, it should be possible to use the PR8 LAIV-mouse model to evaluate whether additional modifications to the LAIV genetic background can further enhance virus safety in vivo. This may have important applications in the development of an improved LAIV suitable for use in young children and infants and other populations for whom the present LAIV is contraindicated.