Vaccines That Reduce Viral Shedding Do Not Prevent Transmission of H1N1 Pandemic 2009 Swine Influenza A Virus Infection to Unvaccinated Pigs

Immunization, influenza A virus challenge, and contact transmission studies were carried out as detailed in Fig. 1. Six vaccines were evaluated in two separate studies for logistical reasons (Table 1). Groups of five pigs were immunized twice with a 3-week interval. Study 1 included the following groups: G1, mock vaccine allantoic fluid as vehicle control administered intramuscularly (i.m.) (Cont); G2, homologous WIV pH1N1 vaccine i.m. (WIV_hom); G3, heterologous WIV H1_avN1 vaccine i.m. (WIV_het); and G4, H3N2 influenza A virus pseudotype delivered by aerosol (S-FLU). In study 2, pigs were immunized i.m. by group as follows: G5, ChAdOx1 prime and MVA boost containing an unrelated antigen (Cont_Ad/MVA); G6, ChAdOx1 prime and MVA boost expressing homologous HA, NP, and M1 (Ad_hom/MVA_hom); G7, MVA expressing the homologous HA, NP, and M1 (MVA_hom); and G8, MVA expressing homologous NP and M1 and heterologous H1_av-origin HA (MVA_het). Immunizations did not induce any adverse reactions. One pig in the Cont_Ad/MVA group was removed on clinical grounds unrelated to the study.

Ten weeks after the first immunization (7 weeks after the boost), pigs were challenged intranasally (i.n.) with 1 × 10⁷ 50% tissue culture infective doses (TCID₅₀) of pH1N1, monitored daily, and removed for necropsy at 6 or 7 days postinoculation (dpi). Unvaccinated naive “contact” pigs (n = 5) were housed from 2 dpi with each vaccinated and challenged group. Contact was maintained for 4 to 5 days, when the directly pH1N1-challenged pigs were removed. The contact pigs were monitored for a total of 12 days postcontact (dpc). Clinical signs were mild or not apparent for the duration of the study in all inoculated and contact groups, and clinical scores transiently reached no more than 2 out of a possible maximum of 20 for several pigs for the duration of the study (data not shown).

Nasal shedding of viral RNA was monitored daily in the immunized, directly virus-challenged and the contact pigs by quantitative real-time reverse transcription-PCR (RRT-qPCR) (Fig. 2A to H). All directly pH1N1-inoculated pigs shed virus from 1 dpi, except for the WIV_hom- or Ad_hom/MVA_hom-immunized groups. These groups given the homologous vaccine to the challenge virus showed minimal virus shedding at the lower limit of quantification (Fig. 2B and F). Mean viral shedding in the directly infected control-immunized animals (Cont) in study 1 (Fig. 2A) or the vector control Cont_Ad/MVA-immunized group in study 2 peaked between 2 and 5 dpi (Fig. 2E). Significant differences (P < 0.05) were observed between these groups on days 4 and 5 dpi, but as these studies were done on different occasions and there was no unimmunized control, it is not known whether this difference was due to a study effect or possible nonspecific adjuvant effect of the viral vector.

Two days after pigs were challenged, each group was cohoused with naive, unimmunized pigs, and nasal swabs were obtained daily to monitor virus shedding. Transmission occurred in all groups, including the WIV_hom- and Ad_hom/MVA_hom-immunized groups, despite significant reduction of virus shedding after direct pH1N1 challenge. However, in the WIV_hom group, two contact animals started shedding virus at 6 dpc, and the onset of shedding was later in the remaining animals. In the Ad_hom/MVA_hom group, one contact animal started shedding at 5 dpc and the remainder between 8 and 10 dpc. In contrast, in all other groups, the majority of contacts started shedding 2 to 3 dpc. It is possible that in the WIV_hom and Ad_hom/MVA_hom groups, the contact animals showing initial shedding infected the other naive pigs in their group. However, once infected, most contact pigs shed virus to the same level despite the delay, except for two contact animals in the Ad_hom/MVA_hom group.

Statistical analysis of the nasal shedding profile of viral RNA supported these conclusions. There were clear pairwise differences (summarized in Table 2) between the vaccinated and control groups in terms of V_max (the peak viral titer shed), T_max (the time to reach V_max following inoculation or contact), the area under the curve (AUC) as a measure of total amount of virus shed, and T_g (the generation time, or time interval between the onset of virus shedding in directly inoculated pigs in relation to naive contact pigs). The P values are shown for vaccinated pigs (Table 3) and contact pigs (Table 4). The variance between the different groups (Fig. 3) was also assessed. AUC values indicative of total viral RNA shedding (Fig. 3A and Table 3) as well as V_max measuring peak viral RNA shedding (Table 3) were significantly different for the WIV_hom- and Ad_hom/MVA_hom-immunized groups compared to the other groups. Correspondingly, the generation time (T_g), representing the mean interval between inoculation of the immunized pigs and detection of shedding in the naive contact pigs, was significantly longer for the pigs in contact with these two WIV_hom- and Ad_hom/MVA_hom-immunized groups (Fig. 3B and Table 4). T_max, or the time from contact to peak virus shedding, was also significantly longer for these two contact pig groups (Table 4). In addition, there was a significant correlation between the lower AUC value in the vaccinated pigs, indicative of reduced viral shedding, and a longer latent period (interval between virus exposure and first detection of virus) in the contact pigs in the WIV_hom and Ad_hom/MVA_hom groups (Fig. 3C and D). These results show that only the WIV_hom and Ad_hom/MVA_hom vaccines significantly reduced shedding after direct pH1N1 challenge. This did not prevent transmission to unimmunized contact pigs, although the infection of contact animals was delayed.

Prior to immunization, influenza virus-specific antibodies were not detected using both NP enzyme-linked immunosorbent assay (ELISA) and hemagglutination inhibition (HI) assays (Fig. 4). A robust and significant (P < 0.0001) anti-influenza NP immune response, indicative of replicating virus, was detected following boost immunization in the WIV_hom, WIV_het, S-FLU, and Ad_hom/MVA_hom groups (Fig. 4A). The antibody responses in the MVA_hom and MVA_het groups were below the threshold of the assay. As expected, no antibody response to NP was detected in the vehicle control (Cont) and vector control (Cont_Ad/MVA) pigs before infection. Antibody responses to immunization were also measured by HI assay using the homologous (clade 1A.3.3.2) and heterologous (1C.2) HA vaccine antigens (Fig. 4B and C). Pigs immunized with WIV_hom or Ad_hom/MVA_hom mounted significant (P < 0.0001) antibody responses to the homologous HA antigen after boost immunization compared to the vehicle and vector control groups (Cont and Cont_Ad/MVA), with average geometric mean ratio (GMR) titers of 2,941 and 1,470, respectively, peaking at 28 days postvaccination (dpv) (Fig. 4B). MVA_hom immunization elicited a lower HI antibody titer, with a significant peak GMR of 557 at 28 dpv (P < 0.0001). HI antibody titers in pigs immunized with the WIV_het measured against the cognate (1C.2) HA antigen showed a significant (P = 0.008) response a week after prime immunization and peaked at a GMR of just under 300 a week after boost (P = 0.002). The WIV_hom- and Ad_hom/MVA_hom-vaccinated pigs elicited comparable titers of cross-reactive antibodies after boost (P < 0.005) (Fig. 4C). Pigs in the Cont_Ad/MVA and S-FLU groups did not produce influenza A virus-specific antibodies, as expected. Surprisingly, MVA_het-vaccinated animals did not produce antibodies to either antigen detectable by HI before virus infection. Virus neutralization (VN) antibody titers elicited by immunization were considerably lower than the anti-NP ELISA and HI antibody levels. Antibodies elicited by homologous vaccination in the WIV_hom and Ad_hom/MVA_hom groups on 28 to 42 dpv (P < 0.001 or P < 0.0001 for the respective vaccine groups) and to a lesser extent MVA_hom (not significant), neutralized the pH1N1 challenge virus strain (geometric mean titers of 28, 74, and 14, respectively) (Fig. 4D). Neutralizing antibodies against the heterologous HA antigen were elicited by the cognate WIV_het vaccine and reached a peak GMR titer of 16 at 28 dpv (P < 0.0001). Cross-reactive antibodies were also detected in the WIV_hom group (P < 0.0001) (Fig. 4E) but could not be detected in the other groups above the limit of sensitivity of the assay.

Six days after pH1N1 challenge, anti-NP antibodies were detected in all groups, including the vector control group (Cont_Ad/MVA). The mean titer increased in the MVA_hom-immunized group but remained below the threshold of the assay for 4 of the 5 pigs. Only the vehicle control (Cont) group did not have a raised anti-NP response within 6 days postinfection (Fig. 5A). All immunized pigs, except the vehicle control (Cont) and S-FLU groups, showed an increased HI titer to the HA from the pH1N1 challenge strain after infection (Fig. 5B). Evaluation of HI titers to the heterologous H1_avN1 antigen at 6 dpi (Fig. 5C) showed a low response in the groups receiving the cognate WIV_het and MVA_het vaccines. Cross-reactive antibodies were elicited in WIV_hom-vaccinated group (geometric mean titer of 490), as reported previously (9). HI titers were not detected in the S-FLU H3N2 vaccine group, as expected. Interestingly, pigs receiving the control virus-vectored vaccine, Cont_Ad/MVA, produced cross-reactive antibodies to both HA antigens post-virus inoculation, possibly reflecting generalized priming of the immune system by the vaccine vectors. Six days after pH1N1 challenge, increased neutralizing antibody titers were detected in the Ad_hom/MVA_hom and MVA_hom groups, and low levels of neutralizing activity were detected in the heterologous MVA_het-immunized group (Fig. 5D). An increase in the neutralizing titer to the heterologous H1_avN1 virus strain was detected in some animals that received the cognate WIV_het immunization, as well as animals receiving MVA_het, but the antibody did not neutralize the pH1N1 challenge strain (Fig. 5E). Taken together, these results indicate that the heterologous HA antigen from the H1_avN1 strain was less immunogenic than the homologous pH1N1 HA antigen used in this study, irrespective of whether it was incorporated in an inactivated or virus-vectored vaccine. Additionally, as reported previously (9), a close antigenic match between the vaccine antigen and challenge strain was needed in order to reduce nasal shedding of virus following infection.

The cellular immune response, monitored by gamma interferon (IFN-γ) enzyme-linked immunosorbent spot (ELISpot) assay in the virus-vectored vaccine groups, showed that at 1 week postboost (28 dpv), the Ad_hom/MVA_hom immunization elicited the strongest response when peripheral blood mononuclear cells (PBMCs) were stimulated by either the homologous inactivated pH1N1 antigen or the conserved overlapping NP or M1 peptides (means of 131, 300, and 121 spot-forming cells [SFCs], respectively) (Fig. 6A to C). In all four vaccine groups, the number of IFN-γ-producing cells had increased 1 week after the pH1N1 challenge infection (6 dpi) following inactivated-pH1N1 or NP peptide stimulation (Fig. 6A and B). The response to M1 peptides was weaker in all groups (Fig. 6C). Interestingly, MVA_het immunization did not induce IFN-γ-producing cells that were stimulated by NP or M1 peptides 1 week after boost, although MVA_het contained the same homologous NP and M1 genes as MVA_hom. Two weeks postboost (35 dpv), reactivity to the inactivated heterologous H1_avN1 antigen was tested (Fig. 6D). Although responses were low, MVA_het animals showed double the response to the cognate H1_avN1 antigen compared to pH1N1 (26 versus 12 SFCs, respectively), while the MVA_hom animals did not show any response different from the control-vaccinated animals (Cont_Ad/MVA). The Ad_hom/MVA_hom-immunized animals had a higher response to the cognate pH1N1 antigen than the heterologous H1_avN1 antigen (mean of 93 versus 57 SFCs), as expected. Overall, the Ad_hom/MVA_hom combination stimulated the highest cellular immune response, which was most clearly detected using NP peptide stimulation. The other groups displayed lower cellular responses that were more variable, a common finding with commercial outbred animals.

In vivo studies were conducted in accordance with U.K. Home Office regulations under the Animals (Scientific Procedures) Act 1986 (ASPA), with protocols approved by the Animal Welfare and Ethical Review Body (AWERB) of the Animal and Plant Health Agency, and the ARRIVE Guidelines were adopted (26).

A total of 80 6-week-old Landrace cross female pigs were obtained from a commercial high-health-status herd in two batches of 40 each. Pigs were screened for absence of influenza A virus RNA by matrix gene RRT-qPCR (27) and absence of influenza A antibodies by hemagglutination inhibition (28) using four swine influenza A virus antigens representative of endemic strains known to be circulating in U.K. pigs, including one pH1N1 antigen. Throughout the study, all pigs received weaner/grower feed and had access to water ad libitum. Study pigs were observed daily for signs of illness and/or welfare impairment. For 1 week postvaccination and postchallenge or after cohousing, animals were scored using a clinical scoring system to monitor clinical signs, including demeanor, appetite, and respiratory signs such as coughing and sneezing, as well as rectal temperature (29).

Virus strains used to generate the monovalent, whole inactivated virus (WIV) vaccines were a pandemic swine H1N1 isolate, A/swine/England/1353/2009 (pH1N1) from clade 1A.3.3.2 (30) and a Eurasian avian-like swine H1N1 isolate, A/swine/England/453/2006 (H1_avN1) from clade 1C.2 (31). Vaccine antigen was prepared from virus propagated in embryonated eggs and inactivated using β-propiolactone (BPL) at (1:2,000) for 2 h as previously described (32). The WIV antigen payload was assessed by the hemagglutination (HA) test, and each dose was formulated in 1 ml with 1,024 HA units (HAU) per ml in an oil-in-water adjuvant, TS6 (https://patents.google.com/patent/WO2005009462A3/en). Vaccines were administered into the trapezius muscle (i.m.), 25 to 30 mm posterior to the ear, using a 1-in., 19-gauge needle.

Virus-vectored vaccines were constructed by the Viral Vector Core Facility of the Jenner Institute, University of Oxford. The modified vaccinia virus Ankara (MVA) vaccines incorporated the complete NP and M1 from A/swine/England/1353/2009 joined by a 7-amino-acid linker sequence and expressed from the vaccinia virus F11 promoter inserted at the F11 locus of MVA (33). The HA coding sequence was derived from the same A/swine/England/1353/2009 or A/swine/England/453/2006 strains used to generate the WIV vaccines and was expressed from the p7.5 promoter at the B8 locus of MVA. Recombinant MVA-vectored vaccines were produced in chicken embryo fibroblast (CEF) cells. All inserts were confirmed by sequencing and were tested for expression and secretion of the proteins by intracellular staining and Western blotting and by ELISA (data not shown). For production of the ChAdOx1-vectored vaccines, expression cassettes were transferred into an adenovirus shuttle plasmid using Gateway technology (Life Technologies); the resulting constructs were linearized by enzyme digestion and transfected into replication-deficient ChAdOx1 as described previously (34, 35). As a control, MVA and ChAdOx1 vaccines incorporating irrelevant antigens for this study, namely, the Zaire Ebola virus surface glycoprotein, were used (36). Each MVA vaccine was administered at a dose of 1.5 × 10⁸ PFU/ml and the ChAdOx1 vaccines at a dose of 5 × 10⁸ IU/ml i.m. in 1 ml, as were the WIV vaccines. The H3N2 S-FLU vaccine is a broadly protective cell-mediated vaccine candidate. This vaccine was constructed with [eGFP*/N2(×217)].H3/Switzerland/9725293/2013 (encoding N2 from A/Victoria/361/2011 from the vaccine strain ×217 and coated with the 3C.3a H3 HA from A/Switzerland/9725293/2013) at 1.52 × 10⁸ 50% tissue culture infectious doses (TCID₅₀)/ml (95% confidence interval [CI], 1.13 to 2.05 × 10⁸/ml). The internal protein gene segments were from influenza A/Puerto Rico/8/1934 (H1N1) (13). The animals received ∼1.5 × 10⁸ TCID₅₀ in 1 ml in identical prime and boost immunizations with a 21-day interval. The S-FLU vaccine was administered to each animal by aerosol using a vibrating mesh nebulizer (SOLO; Aerogen, Ltd.) attached to a custom-made mask held over the nose and mouth following anesthesia, as described previously (13).

Vaccination challenge experiments were conducted with eight groups (G1 to G8) of animals on two separate occasions (Table 1). The same study structure was used (Fig. 1) with pigs housed together during acclimatization for 7 days, the vaccination phase of prime then boost with a 3-week interval and for a further 7 weeks after the boost. In study 1, the following groups were immunized: G1, mock vaccine allantoic fluid vehicle control (Cont) intramuscularly (i.m.); G2, homologous WIV pH1N1 vaccine formulated with TS6 adjuvant i.m. (WIV_hom); G3, heterologous WIV H1_avN1 vaccine (WIV_het) formulated with TS6 adjuvant i.m.; and G4, influenza pseudotype H3N2 administered by aerosol under anesthesia (S-FLU). In study 2, the following groups were immunized: G5, ChAdOx1 prime- and MVA boost-vectored vaccines incorporating the Zaire Ebola virus surface glycoprotein as an irrelevant antigen (Cont_Ad/MVA) i.m.; G6, ChAdOx1 prime- and MVA boost-vectored vaccines incorporating homologous HA and NP+M1 (Ad_hom/MVA_hom); G7, MVA virus-vectored vaccine for the prime and boost incorporating the homologous HA and NP+M1 (MVA_hom); G8, MVA virus-vectored vaccine for the prime and boost incorporating the heterologous H1_av-homologous NP+M1 vaccine (MVA_het). Pigs were housed separately in their vaccine groups for i.n. challenge with 1 × 10⁷ TCID₅₀ A/swine/England/1353/2009 (pH1N1) virus in 4 ml using MAD300 (Teleflex) delivery of an atomized spray of droplets with diameters ranging from 30 to 100 µm. Two days later, 5 naive “contact” pigs were housed with each vaccinated/challenged group. Directly inoculated pigs were euthanized at 6 or 7 dpi and naive contact pigs at 12 dpc with an overdose of intravenous pentobarbital sodium.

Blood samples (clotted and heparin anticoagulated) were taken prior to vaccination, at 1, 3, 7, 14, 21, 28, 35, 42, 56, and 63 dpv, and at 1, 3, and 6 dpi. Serum was collected and stored at −20°C. PBMCs were isolated and cryopreserved for later determination of the humoral and cellular immune responses triggered by vaccination and/or infection. Nasal swabs (two per nostril) were obtained before vaccination and challenge as well as daily after challenge until the end of the study to determine individual influenza A viral RNA shedding profiles by RRT-qPCR. Swabs were stored dry at −80°C until processing.

The nasal swabs from each nasal sample were placed together into 2 ml of Leibovitz L-15 medium (Thermo Fisher Scientific), containing 1% fetal bovine serum (FBS) and 1% penicillin-streptomycin (5,000 U/ml; Gibco). The tubes containing the swabs were agitated, and the supernatant was aliquoted. Total RNA was extracted from these suspensions using the RNeasy minikit (Qiagen, Crawley, United Kingdom) according to the manufacturer's instructions. Viral RNA was detected by RRT-qPCR directed against the influenza A virus M gene (27). RNA quantity is expressed as relative equivalent units (REU) of RNA using a standard 10-fold dilution series of RNA purified from the same batch of virus, of known TCID₅₀ titer, used for challenge. Although these units measure the amount of viral RNA present and not infectivity, it may be inferred from the linear relationship with the dilution series that they are proportional to the amount of infectious virus present as described previously (9).

Antibody titers were measured using a competitive multispecies ELISA to detect the nucleoprotein of influenza A virus (ID Screen IDvet), as well as hemagglutination inhibition (HI) (28) and virus neutralization (VN) (37) assays using homologous and heterologous antigens or viruses.

PBMCs were isolated as previously described (38), and IFN-γ-producing PBMCs were assessed in the virus-vectored vaccine groups with an ELISpot assay using High Protein Binding Immobilon-P membrane plates (MAIPS4510; Millipore) and the MabTech Porcine IFN-γ ELISpot kit (3130-2A; MabTech). PBMCs were stimulated with M1 and NP peptides (16, 17) and β-propiolactone-inactivated virus as previously described (9). The number of spot-forming cells (SFCs) was counted with an automated ELISpot reader (AID) and corrected for background (tissue culture medium stimulated).

Statistical analyses, including calculation of arithmetic means, geometric means, associated standard deviation (SD) or standard error of the mean (SEM), analysis of variance (ANOVA), and associated post hoc Tukey’s tests were performed using GraphPad Prism 7 (IBM). Titers and REU values were logarithmically transformed. Geometric mean ratios (GMRs) of titers were calculated from the samples for each pig relative to the baseline 0-dpv sample. Values were compared between groups using two-way ANOVA (repeated measurements), and statistically significance differences were identified using Tukey’s multiple-comparison test. For k = 8 groups of n = 5 individuals, with power β = 0.80 and type I risk α = 0.05, an effect size of 0.67 could be detected using ANOVA. Differences were considered significant where P was <0.05. We defined the epidemiological parameters as previously described (23). Briefly, the latent period is the interval between inoculation or contact and first detection of virus, while the duration of shedding is the interval between first and last sample times when virus is detected. Since the times to first and last virus detection are interval censored, the observed latent period and duration of shedding are an upper limit and lower limit, respectively. We also defined V_max as the maximal viral titer shed, T_max as the time to reach V_max, and the generation time, T_g, as the time interval between the onset of virus shedding in a primary case and in its secondary case. The area under the curve (AUC) was evaluated to provide a measure of the total amount of virus shed by an animal. The mean AUC values for each group were collectively compared with the AUC values for all other groups. We assumed that the probability of transmission from a vaccinated pig to a contact pig at time t, also called infectiousness, is proportional (with a constant k₁) to viral shedding at t, V(t) (39, 40), as well as homogeneous mixing independent of infection time course to reflect random contacts between the vaccinated and contact pigs (with a contact rate of k₂). Hence, the probability of observing a transmission event at time t is given by $P\left(E=1\mathrm{|t}\right)=k_1{k}_{2}V(t)$. The total amount of virus shed by a pig (or exposure) can be computed as the area under the curve:Generation time can then be computed as the expectation of $P\left(E=1\mathrm{|t}\right)$. We therefore integrated $P\left(E=1\mathrm{|t}\right)t$ with respect to t and normalized by the total rate of transmissions over time, which is k₁k₂ AUC. The expression can be simplified by dropping the constants k₁ and k₂, leading to

All integrals were computed using the trapezoidal rule as implemented in the caTools package (caTools_1.8.tar.gz; https://cran.r-project.org/web/packages/caTools/caTools.pdf). We presented the estimates as mean ± standard error (SE) for quantitative variables. The effect of vaccine group was tested on the different epidemiological parameters (i.e., latent period, duration of shedding, AUC, and generation time) using first a one-way ANOVA. If the P value was <0.05, we then performed a pairwise permutation test as implemented in the rcompanion package (Table 2). Figure 3 was generated using the ggplot2 package (41).