Inducing broad-based immunity against viruses with pandemic potential

Summary

The brutal toll of another viral pandemic can be blunted by investing now in research that uncovers mechanisms of broad-based immunity so we may have vaccines and therapeutics at the ready. We do not know exactly what pathogen may trigger the next wave or next pandemic. We do know, however, that the human immune system must respond, and must be bolstered with effective vaccines and other therapeutics to preserve lives and livelihoods. These countermeasures must focus on features conserved among families of pathogens in order to be responsive against something yet to emerge. Here, we focus on immunological approaches to mitigate the impact of the next emerging virus pandemic by developing vaccines that elicit both broadly protective antibodies and T cells. Identifying human immune mechanisms of broad protection against virus families with pandemic potential will be our best defense for humanity in the future.

INTRODUCTION

Entering the third year of the COVID pandemic underscores the growing cost of a new infectious agent, whether measured in lives (the staggering number of deaths and long-term complications), or in livelihoods (the toll on education, mental health, businesses and economies). Indeed, the cost of the 2014-2016 Ebola virus epidemic has been estimated at $53 Billion, and that of the SARS-CoV-2 pandemic as $16 Trillion, or 90% of the gross domestic product of the United States. The direct and indirect toll of emerging viruses screams the need for global strategies to anticipate and counteract future pandemics.

Strategies are needed at different levels, from infrastructure and rapid response capabilities to interventions such as broad-based antivirals, monoclonal antibodies, and vaccines. Strategies that induce the broadest possible immunity provide our best hope against as-yet-unknown threats.

In this perspective we focus on broad-based immunity against viruses with pandemic human potential, especially those with potential for zoonotic spillover and emergence of new species of concern. Although global pandemics have this far been caused by influenza viruses (Orthomyxoviridae) and coronaviruses (Coronaviridae), emergence of new species of concern is possible from other viral families, as well including Arenaviridae, Flaviviridae, Bunyavirales, Paramyxoviridae, Togaviridae, Picornaviridae and Filoviridae. Well-known viral species of human health concern from these families include influenza, Lassa, West Nile, Dengue, Nipah/Hendra, Zika, Ebola, SARS, MERS and many others. While we do not know which viral species or variant will next emerge, we do know that the human immune system will need to respond. A focus on human immune mechanisms of broad protection against these families of viruses provides our best defense for humanity in the future.

Here we focus on immunological approaches to blunt the impact of another emerging virus pandemic, namely, development of vaccine concepts aimed at eliciting broadly protective antibodies and T cells alike.

Broadly protective antibodies

Antibodies that achieve protection against multiple viruses must first achieve breadth of recognition. Three points are relevant here. First, although an immune response is polyclonal, mutagenic substitution that reduces recognition of multiple individual antibodies at immunodominant sites will diminish antibody protection: a broadly protective vaccine must elicit antibodies able to recognize an array of mutagenic substitutions. Second, although antiviral antibodies may target a linear sequence or loop, more often, they target a three-dimensional surface conformational in nature, with contributions from amino acids that approach each other in the three-dimensional fold (VanBlargan et al., 2016); Crowe, 2017). In immunogen design, we must consider three-dimensional space, and transient conformational adjustment in that space. Mutations don’t just alter a single binding contact, they can also influence 3D conformation and transient occupancy of different conformers. Third, to target heavily glycosylated viral envelope proteins, antibodies must reach in between carbohydrates or successfully incorporate conserved parts of the carbohydrate into their epitopes (Nguyen et al., 2019); (Hastie et al., 2019; LaBranche et al., 2018); (Dang et al., 2021) (Figure 1A). Some rare antibodies recognize the carbohydrate structure themselves by detecting forms and clusters of carbohydrate unique to a viral surface (Acharya et al., 2020; Saunders et al., 2017; Williams et al., 2021), sometimes even achieving glycan-directed recognition of disparate viral families (Lee et al., 2021). The glycans themselves may not only be essential for recognition by assisting in presentation of the protein structure and conformation. As a result, we must consider glycosylation in immunogen design. Different strategies are needed at different sites. We can block highly variable or undesired antibody sites by adding glycosylation (Chen et al., 2021; Duan et al., 2018; Lin et al., 2021). We can encourage recognition of desired sites by ensuring incorporation of native glycosylation that reflects that of the authentic virus (properly structured antigens). Further, at some difficult to access sites, we can consider deleting glycan in priming steps to facilitate recognition by initial germline antibodies, boosting later with fully and natively glycosylated immunogens (Dubrovskaya et al., 2019; Hastie et al., 2019; LaBranche et al., 2018).

Figure 1: Approaches to the induction of broad-based immunity.

A) To achieve breadth of antibody recognition, vaccine immunogens must faithfully display conformational, heavily glycosylated surfaces and conserved sites, perhaps assisted by multivalent antigen display. Fc-mediated protection should be sought in addition to neutralization. B) Inclusion in the vaccine design of antigens and epitopes from conserved regions of the SARS-CoV-2 genome outside spike could provide additional more broadly cross reactive T cell responses. Image credit: Christina Corbaci.

Most importantly, we must target sites that are conserved in sequence in order to achieve breadth of recognition. Target sites conserved in sequence will frequently be those portions of the envelope protein essential for viral entry, which cannot tolerate mutagenic substitution without a cost to fitness (Figure 1A). For example, stalk, membrane-proximal and stem-helix epitopes contain machinery essential for viral infection and are comparatively conserved in sequence (Guthmiller et al., 2022); (Cagigi et al., 2018; Caillat et al., 2021; Darricarrère et al., 2021; Flyak et al., 2018; King et al., 2019; Li et al., 2022; Pinto et al., 2019; Wang et al., 2021a; Zhou et al., 2022) antibodies against them achieve broad recognition. Other broadly reactive antibodies, for example those that cross react between seasonal coronaviruses and SARS-CoV-2, target sites on the more conserved S2 subunit (Dowell et al., 2022; Ng et al., 2021) rather than the receptor-binding domain. Broad recognition may also occur by contacts to main chain instead of side chain, or to the unchanging carbohydrate cores when the presence of the carbohydrate itself is essential for proper fold of the viral protein.

One key target of broadly reactive antibodies are the hydrophobic fusion loops or fusion peptides indispensable for interaction with and fusion of virus and host membranes (Cheng et al., 2019; Kim et al., 2021; Sauer et al., 2021; Vanderheijden et al., 2021). Not only are these sites more conserved in sequence, protein-protein recognition mediated by hydrophobic surfaces may be more tolerant of side chain substitution than other epitopes that rely on a particular polar hydrogen bond. Antibodies that broadly react with the multiple viruses in the ebolavirus family, for example, recognize the hydrophobic fusion loop largely conserved among the Ebola, Sudan, Bundibugyo, Reston and Tai Forest viruses. These viral glycoproteins are otherwise only ~50% conserved overall (Milligan et al., 2019, 2022; Milligan, J.C., Davis, C.W., Yu, X., Ilinykh, P.A., Huang, K., Halfmann, P., Cross, R.W., Borisevich, V., Agans, K.N., Geisbert,J.B., Chennareddy, C., Goff, A.J., Piper, A.E., Hui, S., Shaffer, K., Buck,T., Heinrich,M.L., Branco, L.M., Crozier, I., Holbrook, M.R., Kuhn, J.H., Kawaoka, Y., Glass,P.J., Bukreyev, A., Geisbert, T.W., Worwa, G., Ahmed,R., and Saphire, E.O., 2022; West et al., 2018). Antibodies against the fusion peptides of previously endemic coronaviruses cross-react with that of SARS-CoV-2 (Vanderheijden et al., 2021).

Another conserved target could be a receptor-binding site shared among multiple viruses in the family, such as those antibodies able to neutralize both SARS-CoV-1 and SARS-CoV-2 by interfering with binding of the shared ACE2 receptor (Jennewein et al., 2021), the broad recognition of influenza viruses by recognition of the shared sialylated glycan receptors (Whittle et al., 2011; Wu and Wilson, 2020); (Lin et al., 2018). Broad recognition of ebolaviruses and even Marburg viruses can occur by recognition of the common NPC1 receptor-binding site (Gong et al., 2016; Hashiguchi et al., 2015). Note however, that the ebolaviruses haven’t changed much in sequence during or between outbreaks: The Ebola virus that broke out in 1976 in the Democratic Republic of the Congo is 99% identical to the Ebola virus that broke out in 2014 in DRC (Jun et al., 2015), and antibodies neutralize both equally well. In contrast, the SARS-CoV-2 pandemic occurred with intense evolution in the virus as it adapted to its new human host. Adaptation through mutations in its receptor binding motif meant a significant drop in binding of many anti-receptor binding site antibodies to the beta and omicron variants (Dejnirattisai et al., 2022; Wang et al., 2021b). Vaccination strategies against rapidly evolving viral sequences, for SARS-CoV-2 or other viruses, will need to involve a variety of sequences (i.e. display of antigen or portions of antigen from multiple members of the virus family), or will need to better display or prime with conserved, perhaps less immunodominant, portions of the envelope proteins conserved among members of the virus family (Cheng et al., 2019; Cottrell et al., 2020).

Breadth of recognition and potency of neutralization is often achieved by recognition of quaternary epitopes, formed when multiple subunits of the viral surface proteins approach each other in three-dimensional space (de Alwis et al., 2012; Dang et al., 2021; Tyagi et al., 2020). Some of the more potently neutralizing antibodies against multiple families of viruses recognize multiple subunits of the viral glycoprotein, whether by bridging the receptor-binding subunit to the fusion subunit, by anchoring neighboring monomers in in a single glycoprotein trimer together, or by bridging multiple copies of envelope on the viral surface (de Alwis et al., 2012; Fedechkin et al., 2020). Vaccine strategies that elicit these types of antibodies must faithfully display the right conformation and the right assembly of these unstable proteins – tactics which require protein engineering of the immunogen to achieve.

Creation of an antibody protein surface (paratope) that achieves precise recognition of a novel pathogen is achieved over time, evolving from initial, low-affinity germline sequences, to higher-affinity binding molecules. Repeated antigen stimulation to evolve these responses is facilitated by longevity/stability of the immunogen, boosting or exposure (Tsai et al., 2013). In design of vaccine strategies, we have the opportunity to prime with one form and boost with another to improve more elusive, but desired types of antibody responses.

Geometry and stoichiometry matter here: in general, recognition of repeated copies of antigen facilities binding by avidity (Jendroszek and Kjaergaard; Oda and Azuma, 2000; Remmel et al., 2021; Yin et al.) (Figure 1A). Recognition by the natural IgG may be monovalent, bivalent, or may occur by bridging separate glycoproteins together. In vaccine design, display of antigen in the correct oligomeric form and in desired immunogen spacing will assist in elicitation of IgG that naturally bridge monomers in the trimer, that bind with avidity (Tsai et al., 2013), and which display Fc for immune effector cells. Multivalent display of antigens on vaccine particles (whether nanoparticles or virus-like particles or the surfaces of attenuated viruses), and display of full oligomeric glycoproteins instead of single subunits, will allow vaccines to elicit IgG that can take advantage of oligomeric state and geometry to achieve both neutralizing and extra-neutralizing functions. In one broad analysis of an array of therapeutic antibodies against SARS-CoV-2, those antibodies that take advantage of avidity achieve more potent neutralization and breadth of recognition than those that engage by one Fab alone (Hastie et al., 2021). Elicitation of these types of antibodies would be facilitated by display of whole glycoprotein oligomers, rather than monomeric receptor-binding domains. Further, display of multiple copies on nanoparticles would afford opportunities to enhance responses against desired, conserved, otherwise weakly immunogenic epitopes (Yang et al., 2020). For viruses like Dengue, one must consider whether the particles are mature: better neutralization results may be achieved when particle preparations are homogeneous and mature (Tsai et al., 2018; Yang et al., 2020), presenting conformations and quaternary sites accessible in the actual infecting virion.

In understanding which are the desirable types of antibodies to elicit with a vaccine, we must note that there are both neutralizing and extra-neutralizing functions of antibodies. Neutralization means the ability of antibody to prevent viral propagation in cell culture, usually of a single cell line. Neutralization in vitro is usually mechanical in nature and involves the biochemistry and biophysics of the protein-protein interactions between antibody and pathogen. Antibody binding that sterically blocks binding of receptor, required protease cleavage of the viral glycoprotein, or conformational changes required for fusion of viral and host membranes will all achieve mechanical neutralization. Neutralization is measured in vitro by measuring viral or pseudoviral infection of cells in culture, often using a GFP- or other reporter gene to measure infection. Because neutralization is easy to measure, and peripheral serum antibody easy to collect, neutralization of serum antibody is a frequently tapped tool to forecast vaccine efficacy. Note also that mechanical neutralization, which involves the biochemical and biophysical events of protein-protein interaction, is more straightforward to recapitulate among tissue culture, rodents, non-human primates and human clinical studies.

However, although neutralization in vitro is a key predictor, it is not the only predictor of antibody efficacy in vivo (Balachandran et al., 1982; Calvert et al., 2022; Corbeil et al., 1996; Kim et al., 2021). Some antibodies have a mechanical activity not recapitulated in the chosen in vitro system, depending, for example, on whether the range of target cells in vivo has different conditions or entry factors than the cell type used in vitro (Neerukonda et al., 2021). Some protective antibodies block egress instead of entry (Williamson et al., 2021); (Dufloo et al., 2022; Jin et al., 2018).

Other antibodies recruit protection in vivo via their Fc domain. In this process, antibody-bound particles bind to and trigger Fc receptors, which result in the generation, secretion or repression of various pro- or anti-inflammatory substances. Fc-mediated immune effector functions are a major contributor of antibody-mediated protection, whether by passive immunotherapy or by vaccines, and are a fundamental source of antibody protection. While the activity of the Fab may inactivate a virus, it is the activity of the Fc that clears the virus and infected cells from the body. These Fc-mediated antibody functions (Forthal, 2014); (Bournazos and Ravetch, 2017) involve lysis of infected cells, immune activation by complement, internalization and destruction of virions and infected cells by phagocytosis, modulation of inflammation, mucus trapping, and recruitment of other factors or cells in the immune system. Fc-mediated internalization of immune complexes results in the engagement of toll-like receptors. Further, Fc-mediated phagocytosis of foreign particles by antigen- presenting cells enhances T cell responses. Hence, achieving extra-neutralizing functions can assist development of broadly reactive and broadly protective T cell responses.

Fc functions occur outside of and independently from in vitro neutralization. They may occur on neutralizing antibodies or on non-neutralizing antibodies. Hence, we prefer to term them “extra-neutralization” functions rather than non-neutralizing functions, because “non-neutralizing” implies that these functions only occur on antibodies that do not neutralize. Instead, Fc-mediated protective functions may occur in the presence or absence of complementary neutralization. Indeed, addition of Fc-mediated functions on top of Fab-mediated neutralization provides additional protective benefit (Hessell et al., 2007) over mechanical neutralization alone.

These extra-neutralization functions are measured less frequently than neutralization as they are more difficult to measure, involving multiple factors and multiple cell types, concerted and sometimes transient or low-affinity interactions. Further, Fc-mediated functions differ among each animal model species and humans. These extra-neutralization functions are, however, important. They have been linked to improved survival in viral infection (Bahnan et al., 2021; Hessell et al., 2007), and may constitute the majority of the antibody response: The neutralizing portion of antibodies, for example, are a fraction of the total antibody elicited in SARS-CoV-2 infection (Yamayoshi et al., 2021), and for Ebola virus, total binding antibody is a stronger correlate of protection than neutralizing antibody (Roozendaal et al., 2020).

Interestingly, in some studies, it is the broadly reactive antibodies that achieve the greatest Fc-mediated effect (Terajima et al., 2011); (Grunst and Uchil, 2022). Hence, antibodies that broadly bind can be protective, even if they do not neutralize (van Erp et al., 2019); (Thulin and Wang, 2018). For coronaviruses, for breadth, one may seek to target the S2 subunit (Dowell et al., 2022); (Pinto et al., 2021), a subunit characterized by much lower potency of neutralization. Vaccines against other families of viruses should seek recognition of more conserved subunits as well (Angeletti and Yewdell, 2018; Ellebedy et al., 2014; Hsieh et al., 2021; Impagliazzo et al., 2015; Mallajosyula et al., 2014). Such antibodies may provide protection in humans even if they do not lead to the most potent mechanical neutralization in cell culture.

The low-hanging fruit, or straightforward way to achieve broad recognition may simply be including antigens from different viral strains or species onto one vaccine particle, for example the receptor-binding domains from multiple coronaviruses (Dowell et al., 2022; Walls et al., 2021) or the envelopes of multiple ebolaviruses (Callendret et al., 2018; Cross et al., 2020; Lehrer et al., 2021; Matassov et al., 2018). The multi-component strategy achieves breadth at the polyclonal level, eliciting some antibodies against each of the components individually and some antibodies that target something shared among them. The high-hanging fruit, identification of immunogens that represent shared features of the multiple strains or viruses in a family, involves engineering of viral antigens to mask immunodominant, but variable sites, and to better stabilize and present conserved and essential, but less immunodominant sites. The vaccines in the SARS-CoV-2 pandemic could be rapidly mobilized because of the prior decades of vaccine research into engineering strategies, display formats, and related viruses. To blunt the impact of the next pandemic, we must continue exploration of vaccine strategies against the array of virus families, seeking to elicit antibodies that are broad in recognition, and broad in types of immune protection achieved. In addition to seeking the most potent mechanical neutralization, we must also think more deeply about engineering immunogens that faithfully present quaternary assembly and glycosylation of the actual virus, correct multimeric presentation to take advantage of beneficial effects of avidity. We must also seek better methods of measuring and encourage the multitude of protective Fc-mediated functions for both clearance and inspiration of complementary T cell activities.

Broadly protective T cells

What is the rationale for the further exploration of a broadly cross-reactive T cell vaccine concept? To illustrate this point we briefly point to some of the lines of evidence supporting the concept in the case of SARS-CoV-2. Early T cell responses correlate with better outcomes (Rydyznski Moderbacher et al., 2020) and lower viral loads (Tan et al., 2021). Agammaglobulinemic and B cell-depleted individuals have only moderately increased risk of hospitalization with COVID-19 (Soresina et al., 2020); COVID-19 in ocrelizumab-treated people with multiple sclerosis (MS) is predominantly mild, suggesting that disease can be controlled and resolved also in absence of antibody responses (Apostolidis et al., 2021; Sabatino et al., 2022). The onset of protection following one-dose of Moderna or Pfizer vaccine correlates with early appearance of T cells responses, with detectable neutralizing antibodies (Kalimuddin et al., 2021). Thus, several lines of evidence point to a potential protective contribution of T cells in the context of control of SARSCoV-2 and modulation of disease severity. Additional lines of evidence are provided from studies linking preexisting cross-reactive T cell immunity with vaccination and infection outcomes, discussed in a separate section below.

SARS-CoV-2 variants have taken center stage in the pandemic developments in late 2021 and early 2022. In general, several variants such as Beta, Delta and Omicron have been associated with increased capacity to evade humoral responses induced by natural infection and vaccination, and have also been associated with higher infectivity and transmissibility. However, it has also been noted that in general the variants have not been associated with more severe disease, and in fact Omicron appears to be associated with relatively less severe disease outcomes. The mechanisms associated with these effects are likely to be diverse and factors such as significant changes in Omicron virulence, memory B cells making anamnestic neutralizing antibody responses and changes in tropism infectivity are all plausible contributing factors.

Several studies have investigated whether the T cell response of vaccinated or naturally infected individuals is impaired by variant-associated mutations. The general consensus is that while the mutations affect some epitopes, the majority of T cell epitopes are fully conserved in the various variants (Tarke et al., 2021a); (Redd et al., 2021; Tarke et al., 2021b). Likewise, at the population level it was found that T cell responses are largely preserved in terms of recognition of variants, including Delta and Omicron (Tarke et al., 2021a); (Tarke et al., 2022); (Geers et al., 2021); (Alter et al., 2021); (Alter et al., 2021; Keeton et al., 2021); (Melo-González et al., 2021); (Madelon et al., 2021); (GeurtsvanKessel et al., 2022); (GeurtsvanKessel et al., 2022; Keeton et al., 2022); (De Marco et al., 2021); (Liu et al., 2022); (Gao et al., 2022). In light of this data, it is plausible that T cells play a role in protection, modulation of disease severity and termination of infection. These data suggest that preservation of T cell responses can be of potential benefit, even in concomitance with suboptimal antibody responses.It was noted early on that memory T cell reactivity is also detected in non-exposed individuals (Grifoni et al., 2020a). It was demonstrated (Mateus et al., 2020) that this was to be ascribed to memory T cells recognizing Common Cold Coronavirus (CCC) epitopes with sequence homology to SARS-CoV-2 sequences, although potential crossreactivity of T cell epitopes derived from other viral species was also reported (Bacher et al., 2020); (Bacher et al., 2020; Le Bert et al., 2021). Pre-existing memory T cells in SARS-CoV-2 unexposed subjects capable of recognizing SARS-CoV-2 sequences was reported by several independent parallel studies (reviewed in (Sette and Crotty, 2020). Thus, this observation was independently reported by several groups in diverse settings and geographical locations. Furthermore, numerous studies reported different epitopes associated with different degrees of CCC-SARS-CoV-2 cross-reactivity (Schulien et al., 2021); (Sekine et al., 2020); (Shomuradova et al., 2020); (Nelde et al., 2021); (Keller et al., 2020; Nelde et al., 2021); (Ferretti et al., 2020); (Ferretti et al., 2020; Prakash et al., 2021).

Several additional studies reported cross-reactivity between SARSCoV-2, SARS-CoV-1 and MERS at the level of T cell responses. In general, the level of cross-reactivity tends to be higher when these viruses are considered as compared to CCC, which is consistent with the closer phylogenetic relation, and the higher degree of structural similarity and sequence homology (Ferretti et al., 2020; Habel et al., 2020; Le Bert et al., 2020; Prakash et al., 2021).

It was hypothesized that this pre-existing immunity could influence the severity of disease associated with subsequent SARS-CoV-2 infection and/or the outcomes of SARS-CoV-2 vaccination (Sette and Crotty, 2020). Our recent studies analyzing responses to low dose Moderna mRNA-1273 COVID-19 vaccine demonstrated that indeed subjects with pre-existing cross-reactive immunity respond faster and better at the level of CD4⁺, T folicular helper (Tfh) and serological responses (Mateus et al., 2021).

Further evidence of cross-reactive T cells roles in prevention of symptomatic COVID-19 was also published by Thiel and colleagues (Loyal et al., 2021). Also consistent with this notion are studies showing a potential protective effect of a conserved HLA-B7 restricted epitope (Peng et al., 2021) and the ability of T cell to cross react with CCC homologous peptides in unexposed individuals (Lineburg et al., 2021). Further studies from Sagar et al showed that recent exposure to common cold coronaviruses correlated with less severe COVID-19 outcomes (Sagar et al., 2021). Finally, we (da Silva Antunes et al., 2021) demonstrated that heavily exposed COVID seronegative donors had high level of CCC-T cell reactivity, and a recent study from the Maini and Bertoletti’s group highlighted that cross-reactive T cells are associated with abortive SARS-CoV-2 infection of health care workers (Swadling et al., 2022). Similarly, a study from Lalvani’s group associated cross-reactive memory T cells with protection against SARS-CoV-2 infection in COVID-19 contacts (Kundu et al., 2022). Thus, cross-reactive T cells can modulate disease severity and dramatically influence infection outcomes (Peng et al., 2021).

The data and studies described above indicate that it might be possible to define antigens, antigen fragments and/or epitopes that are broadly cross-reactive different coronaviruses (Mateus et al., 2020; Prakash et al., 2021)(Fig 1B). Beyond SARS-CoV-1, SARS-CoV-2 and MERS, several other coronaviruses are also of concern, and in particular several zoonotic viruses that could trigger a new pandemic if capable of jumping to human host and causing widespread disease, or even SARS-CoV-2 variants that further evolved in non-human hosts. The identification of cross-reactive epitopes can be accomplished utilizing a combination of immunological, virological and phylogenetic analyses. For example, representative viruses sampling the viral species of concern can be defined, and the sequence conservation along the proteome can be rigorously defined, both within different viral groups and across more distantly related viral groups.

A variety of different immunological techniques can be utilized (Sidney et al., 2020) to define the actual T cell epitopes recognized in individuals infected by CCC or SARS-CoV-2. The conservation and immunogenicity analysis can then be further integrated to map immunogenic and conserved regions in CCC and SARS-CoV-2, and experimentally determining epitopes/ regions that are widely cross-reactive at the level of T. cell responses (Grifoni et al., 2020b; Mateus et al., 2020). The widely cross-reactive antigens, antigen domains and/or epitopes identified could be used as a T cell-inducing vaccine component, either standing alone or in conjunction with antibody-induced components (Dangi et al., 2021).

The definition of conserved/cross-reactive T cell epitopes across different coronaviruses is not without precedent. Cross-reactive and conserved epitopes have been described in a number of different viral species, serotypes and subtypes. Here, as a way of illustration we describe three such instances that we have addressed in past investigations.

In the case of influenza, early studies demonstrated that the human T cell repertoire is broad and multispecific (Gianfrani et al., 2000). A subsequent study described 38 class I and 16 class II nonredundant epitopes which were consistently recognized in multiple subjects, provided high coverage among different ethnicities, and were conserved in the majority of the strains analyzed (Assarsson et al., 2008). Parallel studies focused on HLA DR restricted epitopes conserved in Influenza A virus strains in circulation, associated with past pandemics and of potential zoonotic interest. Immunization of HLA transgenic mice with a DNA plasmid encoding 20 different epitopes was shown to enhance antibody responses and protect from lethal PR8 influenza virus challenge (Alexander et al., 2010a). A similar study studied HLA class I restricted influenza-derived conserved T cell epitopes (Alexander et al., 2010b). Finally, we also reported pre-existing immunity against swine-origin H1N1 influenza viruses in the general human population, presumably due to previous exposure to related influenza strains, and hypothesized that this effect might explain the observed lower disease severity in older subjects, presumably exposed to other H1N1 strains (Greenbaum et al., 2009). Similar approaches were utilized to define sets of epitopes immunogenic in humans and/or restricted by human HLA, and conserved in Old and New World Arenaviruses (Botten et al., 2010; Kotturi et al., 2009, 2010).

A significant amount of data is also available in the case of flaviviruses. In the case of DENV, we have shown that infection with multiple serotypes tend to favor a bias towards recognition of T cell epitopes conserved across serotypes (Weiskopf et al., 2014), and vaccination with a tetravalent attenuated DENV vaccine elicits responses mostly directed against conserved epitopes (Weiskopf et al., 2015). HLA alleles association with disease protection also suggest a potential protective role for T cell responses (Weiskopf et al., 2013, 2016). Several animal model studies have shown that T cell responses can protect from challenge or ameliorate disease (Elong Ngono et al., 2016; Yauch et al., 2009, 2010; Zellweger et al., 2015). Further studies have shown that in DENV infection can influence T cell responses to ZIKV upon ZIKV infection (Grifoni et al., 2017; Wen et al., 2017) presumably because of preexisting ZIKA cross-reactive T cells (Schouest et al., 2021). In, general cross-reactivity amongst different flaviviruses (Dengue, Yellow Fever, Zika) is low (Grifoni et al., 2020b), but conserved epitopes can be identified and shown to mediate protection in animal models of disease (Regla-Nava et al., 2018; Wen et al., 2020).

We conclude that it is possible to entertain a broad concept, aimed at generating T cell inducing vaccine components, for the specific purpose of enhance preparedness against future possible pandemics. Such vaccine components could entail inclusion of additional SARS-CoV-2 antigens such a N or some of the NSP antigens encoded in the Orf1a/b region of the genome, or specific antigen domains of fragments rich in T cell epitope content, or collections of defined T cell epitopes. These T cell vaccine or vaccine components might be designed specifically for the purpose of broadly preventing hospitalizations, severe disease and death. Several groups have reported initial testing of SARS-CoV-2 vaccines including additional components, beyond the spike antigen, which would be expected to broaden the spectrum of T cell reactivities elicited by vaccination. These include NantWorks /Immunitbio, Gritstone, Flowpharma, Walz group, Vaxxinity, several academic groups and potentially others.

We emphasize that this concept should not be seen as an alternative to antibody-inducing strategies, but should rather be expected to be highly synergistic with those strategies. A T cell-inducing component could be produced, tested for safety and immunogenicity in small phase I trials, and even stockpiled, as a first line of defense from a new pandemic, buying time until outbreak-specific vaccine or antibodies become available. Some are in development. This would be done in parallel and as an adjuvant to other antibody inducing vaccine strategies.

Finally, we note that while the approach is being considered for coronaviruses in general, and sarbecoviruses in particular, a similar strategy could be developed for each of several families of viruses of pandemic preparedness concern, namely Arenaviridae, Flaviviridae, Bunyavirales, Paramyxoviridae, Togaviridae, Picornaviridae and Filioviridae. Research now into vaccine strategies that elicit broadly reactive antibodies and T cells alike, against the array of possible threats, will better prepare us, our health systems and our economies, for the next pandemic to come.

HIghlights and eTOC.

Saphire and Sette discuss potential approaches for vaccine development against the broad array of SARS-CoV-2 variants and other viruses of pandemic potential. Approaches that elicit both humoral and cellular responses are considered and are expected to be synergistic with each other, providing humanity with the best chance to defend against the next pandemic.