Epigraph hemagglutinin vaccine induces broad cross-reactive immunity against swine H3 influenza viru

Influenza A virus infection in swine impacts the agricultural industry in addition to its zoonotic potential. Here, we utilize epigraph, a computational algorithm, to design a universal swine H3 influenza vaccine. The epigraph hemagglutinin proteins are delivered using an Adenovirus type 5 vector and are compared to a wild type hemagglutinin and the commercial inactivated vaccine, FluSure. In mice, epigraph vaccination leads to significant cross-reactive antibody and T-cell responses against a diverse panel of swH3 isolates. Epigraph vaccination also reduces weight loss and lung viral titers in mice after challenge with three divergent swH3 viruses. Vaccination studies in swine, the target species for this vaccine, show stronger levels of cross-reactive antibodies and T-cell responses after immunization with the epigraph vaccine compared to the wild type and FluSure vaccines. In both murine and swine models, epigraph vaccination shows superior cross-reactive immunity that should be further investigated as a universal swH3 vaccine.

Influenza infection in swine is a highly contagious respiratory virus endemic in pig populations around the world1. Influenza A virus in swine (IAV-S) can cause zoonotic infections in humans, representing a potential threat to human health2,3. When the influenza virus of swine origin infects humans, it is termed a variant infection. Since 2010, there have been >460 reported IAV-S variant infections in humans in the United States of America4. Pigs are susceptible to swine, avian, and human influenza viruses, making them the perfect “mixing vessel” for novel reassorted influenza viruses2,5. These novel reassorted viruses have significant pandemic potential if zoonosis occurs, as seen with 2009 H1N1 “swine flu” pandemic. This highly-reassorted swine-origin influenza virus quickly circulated the globe and infected a staggering 24% of the world’s human population6,7. As the first influenza pandemic of the twenty-first century, this highlights the threat that zoonotic IAV-S poses to human health.

IAV-S not only poses a potential human health threat from zoonosis, but it also represents a significant burden on the pork industry. IAV-S infection of pigs results in high morbidity, with many of the same symptoms as human influenza infections8. IAV-S infection can cause tremendous economic loss to swine producers, with cost estimates as high as $10.31 per market pig9. In the USA, over 95% of swine nursery sites vaccinated weaned pigs against IAV-S infection. However, 50% of those sites also reported IAV-S infections in their herds despite vaccination10. This highlights the ongoing challenge of vaccinating against the highly diverse and evolving influenza virus. Currently, most commercial IAV-S vaccines are traditional whole inactivated virus (WIV) vaccines containing both H1 and H3 subtypes, often with an oil-in-water adjuvant11. However, these commercial vaccines are infrequently updated and do not protect against the large diversity of IAV-S circulating in the swine population. This has led to the use of autogenous, or custom, vaccines that contain herd-specific IAV-S strains and are limited to use within that herd. An estimated 50% of IAV-S vaccines sold are autogenous vaccines10,11,12. However, autogenous vaccines have multiple drawbacks, including labor-intensive laboratory techniques for diagnosis, isolation, virus growth, and purification, which results in a lag period before the vaccine can be administered11. The limited strains that were currently available in commercial swine influenza vaccines paired with the significant drawback to autogenous vaccines highlight the urgent need for a universal swine influenza vaccine. A universal swine influenza vaccine could reduce the economic impact of IAV-S on the pork industry, along with reducing the risk of emergent zoonotic influenza viruses into the human population.

Currently, the IAV-S subtypes H1N1, H1N2, and H3N2 circulating in the swine population worldwide1. We chose to focus on the swine H3 (swH3) subtype for this study because the H3N2 subtype accounted for >90% of the IAV-S variant human infections reported in the US since 20104. The swH3 subtype is highly diverse, with multiple human-to-swine introduction events establishing the contemporary H3N2 strains circulating in different regions of the world. In Europe, the swine H3N2 subtype emerged in the early 1970s from the introduction of a human lineage H3N2 strain8,13. However, in North America, the H3 subtype was not found in the swine population until 1998 when a triple-reassorted H3N2 virus emerged14. The North American strains are divided into clusters I–IV, with cluster IV further divided into A–F, and are divergent from contemporary Eurasian strains8. Additionally, in 2010–2011, a human seasonal H3N2 was transmitted to North American swine and established a lineage of human-like H3 viruses that are antigenically distinct from other North American clusters15,16. The high diversity of the swH3 population represents a significant challenge in the development of a vaccine that induces strong levels of broadly cross-reactive immunity.

This study aims to evaluate a vaccine antigen designer, called the Epigraph vaccine designer tool, for the design of a universal swH3 influenza vaccine17. The epigraph is a graph-based algorithm that creates a cocktail of vaccine antigens designed to maximize the potential epitope coverage of a highly diverse population. This epigraph algorithm has been used to predict therapeutic HIV vaccine candidates18 and has shown promising potential in vivo as a Pan-Filovirus vaccine19. Here, we utilize the Epigraph vaccine designer in the development of a universal swH3 vaccine by computationally designing a cocktail of three swH3 hemagglutinins (HA), a surface glycoprotein of influenza. This is the first report evaluating the epigraph algorithm for the design of a broadly reactive influenza vaccine. The epigraph HA immunogens were expressed in a replication-defective Adenovirus type 5 (HAdV-5) vector and compared to a wild-type HA (TX98) and the commercial inactivated adjuvanted vaccine, FluSure. We evaluated the cross-reactivity of the epigraph vaccine by measuring both antibody and T-cell responses in mice and swine. Additionally, we evaluated cross-protective immunity against three diverse swH3 strains after challenge in mice. These data support the use of epigraph immunogens in the development of a universal swH3 vaccine.

Development and characterization of the swH3 epigraph HA vaccine

We designed the swH3 epigraph HA using the Epigraph vaccine designer tool, a graph-based algorithm that creates a cocktail of immunogens designed to maximize potential epitope coverage in a population17,18. First, the Epigraph vaccine designer determines the frequency of each potential epitope of designated length (k-mer) in the target population. The algorithm then uses a graph-based approach to trace a path across the HA protein that contains the most common epitopes in the population, resulting in a full length computationally designed HA protein (epigraph 1). The first epigraph, by design, tends to be very central in its composition (Fig. 1a). This algorithm then is repeated, to create complementary epigraph sequences that minimize, to the extent possible, potential epitopes contained in the previous epigraph immunogens. In this way, the epigraph 2 and 3 construct generally contain the second and third most common epitopes in the population, respectively. These sequences will appear as outliers in a phylogeny, as their composition reflects different k-mer frequencies from sequences throughout the tree (Fig. 1a). The resulting trivalent set of epigraph sequences provides the optimal coverage of potential linear epitopes in the population for a 3-protein set, minimizes the inclusion of rare epitopes that might result in type-specific immune responses, and although artificial, each epigraph resembles natural HA proteins to enable both the induction of antibody and T-cell responses.

The resulting three epigraph HA sequences were aligned back to the original swH3 sequence population and a phylogenic tree was constructed to visualize their relationship to the swH3 population. The three epigraph swH3 immunogens localize across the phylogenic tree (Fig. 1a). To evaluate the computational design of the epigraph vaccine, we selected a HA gene that localizes near the center of the tree (A/swine/Texas/4199-2/1998 [TX98]) as a wild-type comparator. In addition, we also compared our epigraph vaccine to a commercial IAV-S vaccine, FluSure. FluSure is an inactivated, oil-in-water adjuvanted vaccine that contains two North American swH3 strains (along with two H1 strains), which belong to the North American IV-A and IV-B clusters. The three swH3 epigraph genes and the TX98 wild-type HA comparator were cloned into a replication-defective HAdV-5 vector for gene expression. Gene expression was confirmed via western blot (Fig. 1b) and virus particle (vp) to infectious unit ratios were determined to confirm approximate infectivity between the stocks (Supplementary Table 1).

Vaccination with epigraph lead to the development of a strong cross-reactive antibody response in mice

We first evaluated the immune response after vaccination in mice. BALB/c mice (n = 10) were vaccinated with 1010 vp of the HAdV-5-epigraph vaccine, which consisted of equal ratios of the three HAdV-5-epigraph viruses totaling 1010 vp. Our epigraph vaccine was compared to mice vaccinated with either 1010 vp of the HAdV-5-TX98 wild-type comparator or 50 μL of FluSure (which translates to 10✕ the equivalent dose of a 3-week-old pig). A PBS sham vaccine was used as a negative control. Three weeks later, mice were boosted with the same vaccine. Mice were sacrificed 2 weeks after boosting to examine the humoral and cellular immune response after vaccination (Fig. 2a). The cross-reactivity of the antibody response was examined using a hemagglutination inhibition (HI) assay. We selected a panel of 20 swH3 strains which represent much of the diversity of the swH3 phylogenetic tree. This panel contains representative strains from multiple North American clusters along with Eurasian isolates. In addition, the panel contains human-like strains from both the contemporary 2010 human-like lineage and a historical human-like strain that arose from a human-to-swine transmission event (Colorado/1977). A phylogenetic tree was constructed to examine the relationship of the selected 20 strains to the vaccine strains (Fig. 2b; Supplementary Table 2). Vaccination with the epigraph immunogens resulted in a strong cross-reactive antibody response, with HI titers ≥40 to 14 of the 20 (70%) swH3 strains. Epigraph vaccination showed the greatest cross-reactivity against North American and 2010 human-like strains, with HI titers ≥40 to 11 of the 13 (85%) North American strains and both 2010 human-like strains. For the Eurasian strains, epigraph vaccination induced HI titers ≥40 to 1 of the 4 Eurasian strains tested. Importantly, epigraph vaccination-induced significantly higher antibody titers as compared to the TX98 and FluSure groups for 11 of the 20 of the swH3 strains (Fig. 2c). In contrast, the TX98 wild-type comparator and FluSure vaccinated mice developed strong antibody titers (≥40) to 3 of the 20 (15%) and 4 of the 20 (20%) swH3 strains, respectively. The TX98 group developed a strong antibody response to the matched virus Texas/1998 and limited cross-reactivity with only two other strains (Wyoming/2013 and Minnesota/2012). The FluSure vaccine group developed a strong antibody response to two cluster IV-A viruses and to the Minnesota/2012 cluster IV-B strain (a match for the vaccine strain). However, FluSure vaccination provided only limited cross-reactivity with mismatched viruses.

Epigraph immunized mice have a higher total T-cell response and recognize more epitopes from four divergent swH3 strains

Cross-reactive T cells have been shown to play an important role in viral clearance during influenza virus infection20,21. Therefore, we wanted to evaluate if there was increased cross-reactivity of T-cell responses after vaccination with the epigraph vaccine. To examine the cross-reactivity, we selected four swH3 strains that represent a large portion of swH3 diversity. Peptide arrays for the Ohio/2011 strain (cluster IV-A), Manitoba/2005 (cluster IV), Texas/1998 (cluster I), and Colorado/1977 (human-like) were constructed. The T-cell response to each of the four strains was mapped using an IFNγ ELISPOT with an overlapping peptide array containing 17-mers with 10-amino acid overlap. Peptides were considered positive if the response was greater than 50 spot-forming cells (SFC) per million. Epigraph vaccinated mice recognized a greater number of epitopes across all four swH3 strains as compared to the TX98 vaccinated mice (Fig. 3a). Interestingly, the epigraph vaccine induces a significant and robust T-cell response to the Colorado/1977 virus despite not inducing a detectable antibody response against this strain. Therefore, this strain was selected specifically to examine the potential for cross-reactive T cells in the absence of detectable cross-reactive antibodies. In contrast, FluSure vaccinated mice did not develop significant T-cell responses after vaccination (Fig. 3). The magnitude of the responses to each peptide revealed an immunodominant epitope in the HA1 region (amino acid 120–128 of the HA protein) that was positive in all four strains after vaccination with an epigraph. This epitope was predicted to bind strongly to the MHC-I complex of BALB/c mice22,23 and, therefore, is likely an immunodominant CD8 epitope (Fig. 3b; Supplementary Fig. 1). Interestingly, T cells from epigraph vaccinated mice recognized this immunodominant epitope in the Texas/1998 peptide array, however, T cells from TX98 vaccinated mice did not. One possible explanation may be differences in peptide processing and presentation which are dependent on surrounding sequences. Overall, the total T-cell response was significantly stronger in epigraph vaccinated mice against all four swH3 strains (Fig. 3c).

Vaccination with epigraph reduces weight loss and lung viral titers after swH3 challenge in mice

We next wanted to determine if the increased cross-reactive antibody and T-cell responses translated to increased protection from a panel of diverse swH3 strains. BALB/c mice (n = 10) were vaccinated with a single shot of 1010 vp of HAdV-5-epigraph or HAdV-5-TX98, FluSure, or a PBS sham vaccine. Mice were then challenged 3 weeks later with the mouse-adapted swH3 challenge viruses (Fig. 4a). To examine the antibody response after a single immunization, sera at the time of challenge was examined using an HI assay against each of the three challenge strains (Fig. 4b). A single immunization of epigraph resulted in strong HI titers ≥40 to both Ohio/2011 and Manitoba/2005. In contrast, TX98 vaccination did not result in any detectable antibody responses to these three viruses, while FluSure vaccination resulted in low titers (≤40) to Ohio/2011 and Manitoba/2005. No vaccine groups developed antibody responses to the Colorado/1977 strain, making this an ideal strain to evaluate the potential contribution of cross-reactive T-cell responses. After the challenge, mice were monitored for weight loss over 2 weeks. On day 3 post challenge, five mice were sacrificed to examine lung viral titers. We measured lung viral titers by both TCID50 and qPCR to evaluate infectious virus and viral RNA copies, respectively (Fig. 4).

After challenge with the Ohio/2011 strain (cluster IV-A), only epigraph vaccination completely protected mice from weight loss (Fig. 4c). In contrast, the TX98, FluSure, and PBS vaccinated mice lost 8–12% of their body weight by day 3. The FluSure vaccine contains a similar cluster IV-A strain and, although mice were not protected from initial weight loss, the mice showed faster recovery by day 8 as compared to the PBS vaccinated mice (p < 0.0001). In addition, epigraph vaccinated mice showed significantly reduced day 3 lung viral titers as compared to the TX98, FluSure, and PBS vaccinated mice (Fig. 4c).

Challenge with the Manitoba/2005 strain (cluster IV) resulted in severe weight loss for the FluSure and PBS vaccinated mice, whereas epigraph and TX98 vaccinated mice were protected from weight loss (Fig. 4d). However, epigraph vaccinated mice showed the lowest lung viral titers on day 3 post challenge as compared to the three other vaccine groups. Interestingly, although both epigraph and TX98 vaccination protected from weight loss, there were significantly higher lung viral titers in the TX98 vaccinated group, supporting that weight loss does not always correlate with lung viral titer24,25.

Lastly, we were challenged with the highly divergent Colorado/1977 strain. All vaccination groups lost weight early after the challenge, however, epigraph and TX98 vaccinated mice showed significantly reduced weight loss by day 6 as compared to the FluSure and PBS vaccinated mice (Fig. 4e; p < 0.001). Since epigraph and TX98 vaccination does not induce detectable anti-Colorado/1977 antibody responses, the early weight loss but the increased recovery could be a result of T cell-mediated protection. Again, epigraph vaccinated mice showed significantly lower lung viral titers on day 3 as compared to the TX98, FluSure, and PBS vaccination mice (Fig. 4e).

Epigraph vaccination leads to cross-reactive antibody and T-cell responses against multiple human H3 strains

Reverse zoonosis, the transmission of influenza virus from human-to-swine, is a key factor in driving the diversity of IAV-S in swine1,26,27. Therefore, we wanted to determine if our swH3 epigraph vaccine might induce cross-reactive immune responses to human H3 (huH3) isolates to reduce reverse zoonotic events. We selected a panel of 7 huH3 strains to evaluate cross-reactive antibody responses by HI assay. A phylogenetic tree was constructed to examine the relationship of these seven strains to the vaccine strains (Fig. 5a). Epigraph vaccination led to strong antibody titers ≥40 to 3 of the 7 (43%) huH3 strains (Fig. 5b). TX98 vaccination resulted in antibody titers ≥40 to 2 of the 7 (29%) strains and is closely related to both strains (>95.9% identity; Fig. 5a, Supplementary Table 3). In contrast, FluSure vaccination did not result in cross-reactive antibody responses to any of the huH3 isolates.

We also evaluated cross-reactive T-cell responses to three huH3 strains (Mississippi/1985, Aichi/1968, and Texas/1977) using an IFNγ ELISPOT with overlapping peptide arrays. T-cell mapping was performed as described with the swH3 isolates. Interestingly, epigraph vaccination induced a T-cell response against a single immunodominant epitope conserved in all three huH3 isolates (Fig. 5c, d). This epitope is the same position as the immunodominant epitope induced against the swH3 isolates (amino acid position 120–128). In the huH3 population, the amino acids in this epitope are highly conserved (~94% conserved in the huH3 population; Supplementary Fig. 2). TX98 vaccination did not induce a T-cell response against this immunodominant epitope and FluSure vaccination did not result in a significant T-cell response against any of the huH3 strains. Epigraph vaccination also resulted in significant total T-cell responses against all three huH3 isolates (Fig. 5e).

To determine if the huH3 cross-reactive immune responses resulted in protection, we challenged vaccinated mice with a mouse-adapted huH3 Texas/1977 isolate. BALB/c mice (n= 10) were vaccinated with a single shot of 1010 vp of HAdV-5-epigraph or HAdV-5-TX98, FluSure, or a PBS sham vaccine. Mice were then challenged 3 weeks later with the Texas/1977 challenge strain. Only epigraph vaccination completely protected mice from weight loss and death (Fig. 5f). In contrast, TX98 vaccinated mice lost >16% of their starting body weight before starting to recover. FluSure and PBS vaccinated mice quickly lost weight and were all humanely euthanized by day 7 post infection. Epigraph vaccination also reduced infectious virus in the lungs below the level of detection on day 3 as measured by TCID50 (Fig. 5f).

Epigraph vaccination in swine induced strong cross-reactive antibody and T-cell responses

Lastly, to confirm that the results seen in mice translated to the target animal, we vaccinated 3-week-old pigs intramuscularly with 1011 vp of our HAdV-5-epigraph vaccine and compared the immune responses to swine vaccinated with 1011 vp of the HAdV-5-TX98 wild-type comparator or the commercial vaccine FluSure at the manufacture’s recommended dose. Three weeks later, serum was collected to examine the antibody response after a single immunization (Fig. 6a). A single immunization of the epigraph vaccine led to strong cross-reactive antibody titers ≥40 to 13 out of 20 (65%) swH3 strains, with significantly higher antibody responses to 11 out of 20 of the swH3 strains tested, as compared to the TX98 and FluSure groups. Importantly, the epigraph vaccine resulted in cross-reactive antibodies (≥40) to 11 of the 13 (85%) North American strains and both 2010 human-like strains after only a single immunization. In contrast, TX98 only resulted in strong antibody titers (≥40) to the matched Texas/1998 strain and FluSure vaccination did not result in significant titers to any of the swH3 after a single immunization.

Pigs were boosted with the same vaccine and dose 3 weeks after priming and sacrificed 2 weeks later to examine immune correlates at peak immunity. A second immunization boosted cross-reactive antibody titers in epigraph vaccinated pigs, with titers ≥40 to 15 of the 20 (75%) strains (Fig. 6b). In addition, epigraph vaccination showed significantly higher antibody titers to 15 of the 20 strains as compared to TX98 vaccination and significantly higher antibody titers to 10 of the 20 strains as compared to FluSure vaccination. In contrast, after boosting, the TX98 vaccinated pigs showed strong antibody titers (≥40) to 5 of the 20 (25%) swH3 strains, with the strongest antibody titer against the matched Texas/1998 strain. The strongest antibody responses induced after boosting with FluSure were against similar strains to the vaccine, the cluster IV-A viruses and the matched FluSure virus (Minnesota/2012; cluster IV-B). Interestingly, boosting with FluSure also increased the cross-reactive antibody responses across the swH3 panel, with titers ≥40 to 15 of the 12 (75%) strains. However, the responses to most unmatched viruses were significantly lower than responses after epigraph immunization, with an average of 4-fold lower HI titers. Indeed, a single immunization of HAdV-5-epigraph resulted in comparable cross-reactive antibody levels as two FluSure immunizations.

To confirm that the cross-reactive antibody responses as measured by HI assay were also functionally neutralizing, we performed a microneutralization assay. Neutralization titer patterns matched those seen in the HI assay, confirming the functionality of these cross-reactive antibodies (Supplementary Fig. 3). PMBCs were also collected 2 weeks after boosting to examine the cellular immune response using an IFNγ ELISpot. Epigraph vaccination induced the strongest total T-cell response to all four swH3 strains tested (Fig. 6c). TX98 vaccination resulted in a strong T-cell response against the matched Texas/1998 strain but only modest cross-reactive T-cell levels to the other three swH3 strains. FluSure vaccination did not result in detectable cross-reactive T-cell responses.

The post vaccination swine serum was also examined for the presence of cross-reactive antibodies to the panel of 7 huH3 isolates. After a single immunization, epigraph resulted in strong cross-reactive antibody titers ≥40 to 3 of the 7 (43%) huH3 viruses, the same viruses exhibiting cross-reactivity in the mouse model (Fig. 7a). In contrast, TX98 vaccination resulted in antibody titers ≥40 to 1 of the 7 (14%) huH3 strains and FluSure vaccination did not show any cross-reactive antibodies to the huH3 isolates after a single immunization. After a second immunization, the cross-reactive antibody levels in all three vaccine groups increased (Fig. 7b). Boosting with epigraph resulted in strong antibody titers ≥40 to 6 of the 7 (86%) of the huH3 viruses, while TX98 and FluSure boosting results in antibody titers ≥40 to 3 of the 7 (43%) and 4 of the 7 (57%) huH3 viruses, respectively. However, epigraph showed significantly higher antibody titers to 3 of the 7 isolates as compared to both TX98 and FluSure vaccination.