Development of a Universal Influenza Vaccine

The 2017–18 influenza season was a stark reminder that outbreaks of influenza virus are associated with significant morbidity and mortality worldwide. The Centers for Disease Control and Prevention reported over 30,000 laboratory-confirmed influenza-related hospitalizations and 171 confirmed pediatric deaths in the United States (1). Severe disease is most commonly seen in adults with underlying medical conditions, including cardiovascular disease, metabolic disorders, and obesity, as examples. A severity assessment classified the 2017–18 season as high overall severity for each age group (children, adolescents, adults, and older adults), something that hasn’t been observed since the 2003–04 season (1). The severity of this influenza season highlights the importance of measures to control and even prevent influenza virus infections.

Arguably, the most effective means to prevent influenza is through vaccination (https://www.cdc.gov/flu/consumer/prevention.htm). Yet the hallmark of influenza viruses is the ability to undergo rapid antigenic variation due to the accumulation of mutations within the Ab-binding sites in the hemagglutinin (HA) and neuraminidase (NA) surface proteins, abrogating the binding of some Abs (2). This antigenic drift requires that the World Health Organization advisory group of experts meet biannually to analyze influenza surveillance data generated by the World Health Organization Global Influenza Surveillance and Response System to determine if the influenza vaccine candidate viruses must be updated (http://www.who.int/influenza/vaccines/virus/recommendations/consultation201809/en/). Continual surveillance of circulating influenza viruses is crucial for the success of this process and timely production of our annual influenza vaccines (3, 4).

Currently, there are three main categories of annual vaccines approved by the Food and Drug Administration, the most common being the detergent-split inactivated influenza vaccine (IIV) (https://www.fda.gov/biologicsbloodvaccines/vaccines/approvedproducts/ucm093833.htm). IIV is composed of three or four candidate vaccine viruses (CVVs), including an influenza A H1N1 and H3N2 virus, as well as influenza B viruses representing either one or both genetically distinct clades (Victoria or Yamagata). Some examples of the IIV include Fluzone, Fluarix, and Flucelvax. CVVs used in the IIV can be grown in embryonated chicken eggs or Madin-Darby canine kidney cells, after which they are inactivated, purified, and detergent split (4). The Ag is primarily composed of the influenza HA protein, although trace amounts of NA protein may also be present. The vaccine is administered i.m. to elicit a protective Ab immune response. The second vaccine category is the recombinant influenza vaccine (RIV), known as Flublok. The RIV is solely composed of the HA protein from the chosen CVVs for that particular year (5). Unlike IIV, RIV is produced and isolated solely in egg-free systems by expressing the HA in insect cells by baculovirus (6). Finally, a live attenuated influenza vaccine (LAIV) is available as FluMist. Like the IIV, the LAIV is composed of three or four CVVs. However, these viruses have been engineered to grow at or below 33°C, limiting replication to the upper respiratory tract (URT) (7, 8). Because they are attenuated live viruses, they elicit a more robust immune response, including both B and T cell responses (9, 10).

Although vaccines are our best line of defense against influenza, they can be improved. Driving Ab responses against the antigenic sites in the HA head is problematic given the constant drift in this region. It requires continually updating the vaccine to keep up with viral evolution. Growth of CVVs in eggs for vaccine production can also lead to mutations in the HA head region, reducing the efficacy of vaccine-generated Ab responses to circulating viruses (11, 12). This has been an issue for the H3N2 component of the vaccine for several recent influenza seasons (13) (https://www.scientificamerican.com/article/ldquo-the-problem-child-of-seasonal-flu-rdquo-beware-this-winter-rsquo-s-virus/). Finally, the IIV drives a strong humoral response to the HA component, which can be impacted by the immunization and infection history of the person (14–16).

In regards to the LAIV, which was initially licensed in the United States in 2003 for use in people ages 2–49 (17), low effectiveness against influenza A(H1N1)pdm09-like viruses circulating in the United States during the 2013–14 and 2015–16 seasons resulted in the Advisory Committee on Immunization Practices recommendation that LAIV not be used in the 2017–18 season (17). On February 21, 2018, the Advisory Committee on Immunization Practices recommended the LAIV as a vaccine option for the 2018–19 season (17). The reason for the low efficacy remains unclear, but a study in the European version of the United States vaccine showed substantial amounts of defective-interfering viral RNAs from both the influenza A and influenza B viruses (18). Given these challenges, the National Institute of Allergy and Infectious Diseases announced a strategic plan to improve current influenza vaccines and eventually lead to the development of a universal influenza vaccine (19). Whereas a universal influenza vaccine would technically target all types of influenza viruses, for the purposes of the National Institute of Allergy and Infectious Diseases strategic plan, “universal” refers to protection against both group 1 and 2 influenza A viruses (independent of influenza B protection). This review will highlight some advances in a few areas of the strategic plan.

The influenza HA and NA surface proteins are the major targets of immune responses elicited by vaccination, especially humoral responses (20). HA is the receptor binding protein in influenza. It consists of a globular head and a stem that is anchored to the viral envelope. The receptor binding portion of the head determines if the virus will bind to an α2,6 (mostly human) or α2,3 (mostly avian) linkage of the sialic acid. Once the virus gets internalized via endocytosis, an HA conformational change allows the virus to be released into the cytoplasm (21). Influenza A viruses are divided into two groups based on phylogenetic analysis of the HA protein (22). Group 1 contains H1 and H2 subtypes, which can sustain circulation in humans, as well as H5, H6, and H9, which can occasionally infect humans (22). The main subtype of group 2 is H3, but H7 and H10 can occasionally infect humans and be associated with severe and even fatal disease (22).

The predominant seasonal influenza vaccine-elicited immune response targets the globular head of HA for neutralization (23, 24). Abs bind to the HA to prevent binding to sialic acid or to prevent the conformational change that leads to fusion (25, 26). However, because of the previously mentioned mutability of the HA head, these responses only protect from identical strains of influenza. Emergence of vaccine escape variants leads to reduced protective immune responses (2, 27). To overcome this challenge, new strategies are being developed to target immune responses against the stem region of HA (28–30). The HA stem is highly conserved compared with the head, making it a strong target for broadly protective immune responses (31). However, it is more difficult to mount immune responses to the stem due to the HA head’s immunodominance and steric hindrance of the stem (23, 24, 32). Thus, chimeric HA, where the head of HA is changed and the stem is maintained, have been developed (29, 33). Vaccination of mice with chimeric proteins of different group 1 virus heads (H1, H9, H6, or H5) with the same H1 stem could protect mice from challenge with a variety of group 1 viruses, suggesting more broadly protective Ab responses. However, mice were not protected from challenge with H3N2 virus, an HA group 2 virus, demonstrating that protection did not extend to intergroup influenza virus strains (29). Similar results were shown in a ferret model (33).

Another strategy used for stalk-directed immune responses is a recombinant headless stem (34). Although this strategy eliminates the immunodominance of the HA head, the resulting protein does not provide complete protection (34). Therefore, an initial dose of a vaccine that mimics a more natural infection will have to be used, along with boosting of the primary immune response to elicit protective stem-directed Ab responses. Indeed, a recent study in ferrets demonstrated that a sequential immunization regimen can redirect the immune response toward conserved epitopes. Briefly, administration of a LAIV chimeric H8/1 HA followed by a heterologous booster vaccination with a chimeric H5/1N1 formalin-inactivated nonadjuvanted whole virus resulted in low or undetectable titers in the URT after the A(H1N1)pdm09 virus challenge, supporting the further development of chimeric HA-based vaccination strategies (35). Boosting HA stem–directed Abs can be done with either chimeric (29) or recombinant stable and correctly folded headless HA stem proteins (36, 37). One advantage of utilizing chimeric HA proteins is that people who have had natural influenza infection will already have some baseline stem-directed Abs, which will benefit the boosting of the stem-directed Ab response (38–40). Of note, HA stem–directed Abs are not necessarily neutralizing and instead can employ different mechanisms, such as Ab-dependent cellular cytotoxicity, to clear the virus after a permissive viral infection (41–43). Identifying approaches that can improve Ab responses to conserved regions of the HA head and stalk are an exciting development toward improved efficacy.

The second main surface protein of influenza virus, the NA protein, is a functional enzyme that cleaves sialic acid, supporting the release of progeny virus during infection (21). Like HA, NA is divided into two groups based on phylogenetic analysis (22). The enzymatic activity of NA is the main target of antiviral drugs. Although targeting this aspect of influenza replication is not necessarily sterilizing, it does limit viral replication and therefore decreases disease symptoms and viral spread (44–46). In fact, antiviral drugs against the enzymatic activity of NA are the only clinically relevant treatment for influenza infection, aside from supportive care. Therefore, targeting immune responses against the NA protein may be an important complement to HA-directed vaccines (47). A recent clinical study challenging young healthy adults with a pandemic H1N1 virus demonstrated better correlates of NA inhibition (NI) Ab titers with fewer, less severe, and less prolonged symptoms, as well as reduced viral shedding (46). In contrast, hemagglutination inhibition titers only correlated with a reduction in virus shedding (46). Similar results were seen with pre-existing anti-HA stalk Abs as with hemagglutination inhibition titers (48). Another important aspect of NA-directed immune responses is that both inactivated and LAIV induce increases in NI Ab titers. Additionally, NI Ab titers correlate with LAIV and IIV effectiveness (49, 50).

Despite these observations, some barriers must be overcome for the successful use of NA-directed immune responses by vaccination. First, like the HA stem, NA immunogenicity can be masked by the immunodominant HA head (51). However, because of the strong emphasis on HA quantification and standardization in influenza vaccines, low NA immunogenicity could be due to a lack of sufficient NA protein present in vaccines. Although we know that vaccines contain some level of NA, given the increase in NI Ab titers after vaccination (52, 53), the content should be standardized. Yet, before this can occur, we need better NA assays. The development of new and simplified techniques to determine NI Ab titers, including enzyme linked lectin assay and ELISA using NA in its native form, will help overcome these hurdles (54, 55). Although these assays have good reproducibility among different laboratories (56), some caveats include steric hindrance or competition of HA-directed Abs with NA-directed Ab activity (51). Therefore, ELISA using recombinant native form NA is probably the best option moving forward (57).

Once Ag content and NI assays have been standardized, the next hurdle to overcome will be the evolution or antigenic drift of NA. Although NA mutability is lower than HA, vaccine escapes and antiviral resistance strains are known to arise (58, 59). However, a positive aspect of influenza viruses is that HA subtypes and NA subtypes are not concordantly paired between group 1 and group 2 members. Driving vaccine responses against the conserved regions of both HA and NA may prove an exciting and important new approach to provide protective immunity and limit antigenic drift in the virus.

The third surface protein of influenza virus is the matrix protein 2 (M2). M2 functions in both viral entry and egress. During entry, M2 acts as a pH-dependent proton-selective ion channel that controls the internal acidity of virus particles in endosomes, allowing for release of the nucleoprotein (NP) components (60). M2 also controls the pH of the Golgi lumen, supporting viral assembly after replication (60–62). However, other functions of M2 are being uncovered (63, 64). In terms of vaccination, M2 serves as an interesting candidate given that it is highly conserved across multiple influenza virus strains (43, 64–66). In fact, avian influenza viruses also cross-react with human sera against the ectodomain of M2 (M2e) (67). Additionally, unlike HA and NA proteins, M2e mutations are nonexistent for up to 11 passages in M2e-vaccinated mice (68), as well as rare and restricted in immunocompromised mice treated with anti-M2e Abs (69). Despite the attractive nature of M2e as a vaccine Ag, its immunogenicity is low after natural infection (70).

To overcome this block, new strategies have been developed to induce M2e-directed Ab responses. The first of these fused M2e to the hepatitis B virus core protein to form virus-like particles with the M2e portion exposed on the surface (71). Those studies demonstrated that both i.p. vaccination with adjuvants and intranasal vaccination without adjuvants protected against both group 1 and group 2 viruses (H1N1 and H3N2). Many other virus-like particle methods have now been employed with M2e (72–74). The protection elicited by Abs to M2e are not sterilizing and instead bind to the surface of virus-infected cells, most likely acting through Ab-dependent cellular cytotoxicity (75–78). The conservation and cross-protective properties of M2e are an exciting aspect of influenza vaccination improvement and will most likely prove indispensable for the development of a universal influenza vaccine. Several recent reviews beautifully cover the state of knowledge of M2 and NP-based vaccines (43, 66, 79).

In addition to HA-, NA-, and M2-directed Ab responses, some vaccine platforms, for example the LAIV, induce T cell responses to conserved Ags in influenza viruses (9, 43, 80). Unlike most Ab vaccine responses, T cell responses are not used to neutralize the virus and prevent infection, but limit viral spread (81). This is achieved by the quick elimination of infected cells by CD8⁺ T cells or by the concerted direction of the immune response by CD4⁺ T cells. Of these, CD8⁺ T cells have been the major focus of influenza-directed T cell responses to date. There are many influenza virus epitopes that are recognized by CD8⁺ T cells (20, 81). Whereas those epitopes include HA and NA, other more highly conserved proteins like NP, matrix protein 1 (M1), and the polymerase proteins are of more interest to the universal influenza vaccination field (82, 83). The T cell epitopes against these proteins are very well conserved among different influenza virus strains (20, 81). It is no surprise, therefore, that there are many reports of CD8⁺ T cell responses correlating with high cross-protection against heterologous strains of influenza viruses in both mice and humans (84–88).

Although CD4⁺ T cell responses to influenza infection have not received as much focus as CD8⁺ T cells, their role is still of importance (81, 89). Memory CD4⁺ T cells help direct a faster Ab response to mutated or immunologically novel viral Ags, as well as the generation of new CD8⁺ T cell responses. In fact, a recent study using a novel platform of influenza vaccination in mice elucidated a major contribution of CD4⁺ T cells to the cross-protective anti-influenza immune response (79). In these experiments, a vaccinia virus encoding five proteins from an H5N1 viral strain was used for immunization and boosting of mice, followed by challenge with an H3N2 virus. This type of vaccination provided complete protection from heterologous virus challenge of mice. Previous studies using this vaccine platform showed the development of Ab responses capable of cross-reacting with different subtypes of viruses (90). However, serum transfer of vaccinated to naive mice was unable to protect from challenge with a heterologous influenza strain (79). In contrast to the lack of protection from adoptive sera transfer, transfer of either CD4⁺ or CD8⁺ T cells protected mice from heterologous challenge. Likewise, depletion of CD4⁺ or CD8⁺ T cells, separately, prior to challenge did not alter the protection conferred by the vaccine (79). Of interest, when CD4⁺ T cells were depleted at time of vaccination, all protection was lost. Therefore, CD4⁺ and CD8⁺ T cells seem to be playing an equivalent protective role during challenge, and CD4⁺ T cells play a necessary role during vaccination.

This previous method of eliciting T cell responses through vector expression is not unique (43, 91), with some studies including mainly T cell Ags (92, 93), whereas other studies combine T cell and Ab Ags (94, 95). One study showed that i.m. vaccination with a Modified Vaccinia virus Ankara (MVA) vector expressing the NP and M1 proteins greatly increased IFN-γ–producing cells after ex vivo restimulation with peptides from the vaccine construct in humans (96). In subsequent studies, participants were vaccinated and challenged with influenza virus (97). When compared with unvaccinated controls, vaccinated subjects had less pronounced symptoms and lower shedding time during infection (97). Furthermore, the MVA-NP+M1 vaccine boosted CD4⁺ and CD8⁺ T cell responses in subjects over 50 y of age (98) and can be used as an adjuvant to increase Ab responses toward IIV components without impacting T cell responses (99, 100). Another advantage of this adjuvant effect of vector vaccines is the possibility to combine with recombinant internal proteins, such as NP, to elicit the often-underappreciated role of nonneutralizing Abs against those virus components (101).

Although it is clear that T cells will be crucial for cross-protective immunity to diverse influenza virus strains, there are drawbacks that make it difficult to implement. First, as with most inflammatory responses, but especially with CD8⁺ T cell responses, the risk of immunopathology is high with such potent effector functions (102, 103). Therefore, it is important to balance a protective CD8⁺ T cell response to influenza with any associated tissue damage (81, 104). One possibility is to direct anti-influenza virus memory CD8⁺ T cell responses to the URT. Studies have shown that URT memory CD8⁺ T cells can prevent influenza virus dissemination to the lower respiratory tract (105). Therefore, a strong CD8⁺ T cell response in the URT could attenuate the immune response necessary in the lungs and maximize nondamaging protection. This is indeed part of the objective when utilizing LAIV, which is given intranasally and is limited to URT replication. However, this leads to another barrier in T cell–directed immune responses to influenza: immune history.

Although immune history is not a problem specific to T cell–directed influenza vaccines, it is particularly apparent in this context. When introduced, the LAIV was of interest because of its live-attenuated nature (106). Therefore, such a vaccine should induce not only the Abs necessary to target HA and NA but also T cell responses to highly conserved Ags to confer cross-protection (9, 80). Although the calculated vaccine efficacy of the LAIV decreased as the years progressed, especially for pandemic H1N1 virus, the reasons behind the inefficacy were not investigated (107). Some studies show that LAIV induces T cell responses in children and does not in adults (14). Thus, on top of defective-interfering RNA, it is possible that the strength of Ab responses to the pandemic H1N1, or other antigenically similar viruses, neutralized the vaccine virus, inhibiting the replication needed to elicit T cell responses (108). Such a limitation could be overcome with platforms such as vector vaccines expressing influenza proteins (79, 92–96). Overall, moving forward, it will be important to elicit strong B and T cell responses in our pursuit of improved influenza virus vaccines.

Although beyond the scope of this review, it is important to note that a universal vaccine must also protect high-risk populations, including the very young, aged adults, pregnant women, people with underlying health conditions, and overweight/obese individuals. We know that the IIV is less effective in high-risk populations (for example, aged adults and overweight/obese individuals) due to underlying immune system complications (109–111). Age-related and potentially weight-related changes in immune function likely contribute to a loss of influenza vaccine efficacy. These changes are likely to limit the applicability of a “universal vaccine” to high-risk populations, outside of the herd immune effects of vaccinating children and adults with a highly efficacious universal vaccine. One strategy for vaccine development in aged is moving toward the notion of “enhanced vaccines” to prevent the serious complications of influenza rather than a universal vaccine that is going to provide sterilizing immunity in this population. These are important consideration for the improvement of current influenza vaccines, as well as in developing universal vaccines.

Influenza virus is a moving target. Rapid evolution allows escape from the protective immune responses generated during natural infection or vaccination to seasonal viruses. Therefore, to provide broad and long-lasting immunity, we need to consider an “all-inclusive” approach to vaccination (Fig. 1). This would include a concerted immune response to conserved regions of different influenza virus proteins driving both B and T cell responses. Many groups are evaluating vaccines to different proteins utilizing distinct platforms for administrations and the impact of immune history or imprinting on vaccine responsiveness, as well as evaluating new methods for preparing vaccine viruses independent of eggs or even cell culture. The results of these studies will lead to improved vaccines against seasonal influenza viruses. They can serve as the template for generation of broadly protective long-lasting immunity that may protect against emerging influenza strains. Yet we must keep in mind that even the best vaccines may have less than ideal efficacy in high-risk populations: a hurdle that might only be overcome by “enhanced vaccines” to prevent serious complications in these populations, or by herd immunity, to ensure population-wide protection against influenza viruses.

The authors have no financial conflicts of interest.