Cytomegalovirus vaccine vector-induced effector memory CD4 + T cells protect cynomolgus macaques from lethal aerosolized heterologous avian influenza challenge

Abstract

An influenza vaccine approach that overcomes the problem of viral sequence diversity and provides long-lived heterosubtypic protection is urgently needed to protect against pandemic influenza viruses. Here, to determine if lung-resident effector memory T cells induced by cytomegalovirus (CMV)-vectored vaccines expressing conserved internal influenza antigens could protect against lethal influenza challenge, we immunize Mauritian cynomolgus macaques (MCM) with cynomolgus CMV (CyCMV) vaccines expressing H1N1 1918 influenza M1, NP, and PB1 antigens (CyCMV/Flu), and challenge with heterologous, aerosolized avian H5N1 influenza. All six unvaccinated MCM died by seven days post infection with acute respiratory distress, while 54.5% (6/11) CyCMV/Flu-vaccinated MCM survived. Survival correlates with the magnitude of lung-resident influenza-specific CD4 + T cells prior to challenge. These data demonstrate that CD4 + T cells targeting conserved internal influenza proteins can protect against highly pathogenic heterologous influenza challenge and support further exploration of effector memory T cell-based vaccines for universal influenza vaccine development.

Introduction

The world remains at risk of another influenza pandemic. The four influenza pandemics of the past 100 years killed tens of millions of people, yet a universal influenza vaccine capable of protecting against future pandemic influenza viruses still does not exist. Current antibody-mediated influenza vaccines are strain-specific due to targeting of the highly variable hemagglutinin (HA) and neuraminidase (NA) glycoproteins. Indeed, given the continual sequence evolution of HA and NA through antigenic drift and ability of the segmented virus to recombine two or more different strains through antigenic shift, seasonal influenza vaccine effectiveness ranges from 30% to 60% depending on matching of the vaccine sequence to influenza viruses subsequently circulating that year^1,2. Furthermore, these vaccines provide little, if any, protection against pandemic influenza viruses. Of particular concern are highly pathogenic avian influenza (HPAI) viruses circulating among wild and captive birds, such as H5N1, which has a documented fatality rate of 52% in humans³. Given the increasing number and geographical spread of HPAI infections in birds⁴, and the minimal amino acid changes necessary for avian influenza viruses to become transmissible via aerosol droplets in mammals^5,6,7, the world is at severe risk of an HPAI pandemic. Indeed, the first case of mammal to human transmission of H5N1 was recently reported in a dairy farm worker⁸, highlighting the potential for HPAI to transmit to humans. Thus, new vaccine approaches capable of protecting against all influenza strains, and particularly against HPAI with pandemic potential, are urgently needed.

While current antibody-based vaccine approaches can provide sterilizing immunity, they narrowly focus on subtype-specific HA and NA sequences and are thus susceptible to antigenic mismatch with new strains that arise annually via antigenic drift or emerge suddenly with pandemic potential via genetic reassortment. In contrast to the high sequence diversity of influenza HA and NA glycoproteins, the internal structural proteins such as matrix (M) and nucleocapsid protein (NP), and viral polymerases like PB1, are highly conserved across all strains⁹. Indeed, pre-existing T cells targeting internal influenza proteins can recognize disparate influenza variants and provide heterosubtypic protection from disease against novel influenza strains in humans^10,11,12. Harnessing T cell immunity against internal influenza proteins therefore represents a potential pathway towards universal influenza vaccine development. However, there is a dearth of vaccine vectors capable of priming and maintaining the high frequencies of pulmonary influenza-specific effector memory T cells (T_EM) likely needed for protection. Indeed, almost all currently utilized clinical vaccine platforms, including the current whole inactivated influenza virus and messenger RNA delivery approaches, induce pathogen-specific T cells with a predominantly central memory (T_CM) phenotype that require a period of anamnestic expansion prior to exerting antiviral activity following infection^13,14. In contrast, the β-herpesvirus cytomegalovirus (CMV) elicits high-frequency T_EM that home to peripheral organs, particularly the lung, where they are pre-positioned to intercept pathogens soon after infection¹⁵, suggesting that CMV may be an ideal vector for development of an influenza-specific T cell-based vaccine.

Despite differences in their etiology, both influenza and HIV possess viral glycoproteins with high sequence diversity that stymie broadly-neutralizing antibody-based vaccine development, and, therefore, vaccine advancements that circumvent glycoprotein diversity in one virus may be transferrable to the other¹⁶. A pre-clinical HIV vaccine approach based on strain 68-1 rhesus CMV expressing internal simian immunodeficiency virus (SIV) antigens (RhCMV/SIV) elicits high-frequency SIV-specific T_EM that control SIV replication in 59% of vaccinated rhesus macaques (RM) across multiple studies in the absence of vaccine-elicited antibodies¹⁷. Strain 68-1 RhCMV, which lacks the pentameric receptor complex components Rh157.5/Rh157.4 and the viral CXC chemokine-like Rh158-Rh161 gene products (orthologs of human CMV [HCMV] UL128/UL130 and UL146/UL147 genes, respectively), induces MHC-E- and MHC-II-restricted CD8 + T cells^17,18. However, while strain 68-1 RhCMV elicits unconventionally MHC-E- or MHC-II- restricted CD8 + T cells, RhCMV vectors can be genetically modified via repair of the Rh157.5/Rh157.4 and Rh158-Rh161 genes to generate full length (FL) RhCMV vectors that elicit conventionally MHC-Ia-restricted CD8 + T cells¹⁹. Therefore, CMV vectors can be generated to elicit the type of MHC-restricted CD8 + T cell required for protection against a particular pathogen. Indeed, while SIV-specific MHC-E-restricted CD8 + T cells are required for RhCMV/SIV-mediated protection against SIV replication^17,20,21, they are not required for RhCMV/Mycobacterium tuberculosis (MTB)-mediated protection against MTB in RM²². Regardless of MHC-restriction, RhCMV-induced CD4+ and CD8 + T cells persist longitudinally for years after the initial vaccination and accumulate to high levels in lung^23,24,25. Finally, given the unique protection against SIV replication observed in RhCMV/SIV-vaccinated RM, clinical trials are currently underway to test the safety and immunogenicity of a HCMV vaccine vector for HIV²⁶, providing a potential pathway for clinical CMV-based vaccines against other pathogens. Based on these attributes, we hypothesized that a CMV-based vaccine expressing conserved internal influenza proteins would generate high-frequency, pulmonary-resident, influenza-specific T_EM with the ability to protect against a HPAI isolate with pandemic potential such as H5N1.

Results

Generation and immunogenicity of CyCMV vaccine vectors for influenza

To determine whether a CMV-based vaccine could protect against HPAI, we elected to utilize a stringent model of aerosolized H5N1 challenge of Mauritian cynomolgus macaques (MCM) where infection is uniformly lethal²⁷. Given the strict species-specificity of CMV, strain 68-1 RhCMV does not infect MCM and cannot elicit T cell responses²⁸, thereby precluding its use as a vaccine vector in the MCM model of aerosolized influenza. To facilitate use of MCM for CMV-based experiments, we recently isolated and characterized a full length cynomolgus macaque CMV (FL CyCMV) isolate, and subsequently generated a strain “68-1 like” double deleted (dd CyCMV) vaccine vector, whereby the CyCMV orthologues of the RhCMV pentameric receptor complex components Rh157.5/Rh157.4 and the viral CXC chemokine-like Rh158-Rh161 gene products are deleted to reflect genetic deletions present in strain 68-1 RhCMV²⁹. Vaccination of MCM with FL CyCMV expressing SIV Gag elicited Gag-specific MHC-Ia-restricted CD8 + T cells, while vaccination of MCM with dd CyCMV expressing SIV Gag generated Gag-specific CD8 + T cells that were either MHC-II- or MHC-E-restricted, mirroring the MHC restriction patterns elicited by FL or strain 68-1 RhCMV/Gag in RM, respectively²⁹. Furthermore, half of dd CyCMV/SIV-vaccinated MCM controlled SIV replication post infection, and manifested a vaccine-induced IL-15 transcriptomic signature that is associated with efficacy in RhCMV/SIV-vaccinated RM^29,30. Therefore, central features of the RhCMV vaccine vector in RM are conserved with CyCMV in MCM, facilitating pre-clinical studies in MCM-based models of disease.

In addition to MHC-Ia, the major cellular targets of influenza virus, epithelial cells and type I and II pneumocytes, express MHC-E and MHC-II^31,32,33,34. Therefore, we generated two sets of CyCMV vaccine vectors expressing influenza antigens, one set based on FL CyCMV to induce MHC-Ia-restricted CD8 + T cells, and another based on dd CyCMV to induce MHC-II- and MHC-E-restricted CD8 + T cells (Supplementary Fig. 1A–C). We selected 1918 H1N1 influenza M, NP, and PB1 as vaccine antigens for the following two reasons: 1) in contrast to the more variable HA and NA glycoproteins, the M, NP, and PB1 proteins are highly conserved among the human and avian influenza viruses recorded across the previous decades³⁵, making them ideal T cell targets (Supplementary Fig. 1D, and 2) using 1918 influenza antigen sequences would yield nearly a century of natural global influenza evolution between the 1918 influenza vaccine antigen sequences and the heterologous influenza A/Vietnam/1203/2004 (H5N1) challenge virus, facilitating a stringent test of the protective capabilities of a T cell-based vaccine for influenza. We generated three separate vectors by inserting the 1918 influenza M1, NP, or PB1 sequence into the Cy110 open reading frame (ORF) of either FL or dd CyCMV, and confirmed protein expression in each vector in infected fibroblasts in vitro, to generate a set of vaccine vectors collectively named FL CyCMV/Flu or dd CyCMV/Flu, respectively (Supplementary Fig. 1A–C). The Cy110 gene, the orthologue of the HCMV UL82 gene encoding the pp71 protein necessary for lytic replication, was selected for the site of antigenic insertion as this configuration retains genome and transgene stability while rendering a spread-deficient CMV vector with an increased safety profile that maintains immunogenicity³⁰.

We vaccinated six influenza seronegative MCM subcutaneously with 1 × 10⁷ PFU with each of three individual vectors comprising either FL CyCMV/Flu or dd CyCMV/Flu and boosted with the same dose at 15 weeks post prime Fig. 1A; Supplementary Table 1). One FL CyCMV/Flu vaccinated MCM died from study-unrelated causes at 129 days post immunization, leaving five MCM in that group for all timepoints onward. We monitored the influenza transgene-specific CD4+ and CD8 + T cell response in peripheral blood and found that these responses persisted throughout the vaccine induction phase with no significant differences between the magnitude of the response engendered by FL or dd CyCMV (Fig. 1B). All three influenza transgenes were recognized by both CD4+ and CD8 + T cell responses, with no significant differences observed between those elicited by FL versus dd CyCMV in peripheral blood or lung, as measured by bronchoalveolar lavages (BAL) (Fig. 1C). In line with previous observations of CMV vaccine vectors in nonhuman primates^22,24,29, high frequencies of transgene-specific CD4+ and CD8 + T cells were observed in the BAL at the final timepoint measured prior to influenza challenge in both FL and dd CyCMV/Flu-vaccinated MCM (Fig. 1C, bottom row). As expected with CMV-based vectors, CyCMV/Flu-elicited CD4+ and CD8 + T cells recognizing influenza antigens exhibited a predominantly T_EM phenotype in peripheral blood as measured by expression of CD28 and CCR7 (Fig. 1D). Next, we defined the MHC-restriction of influenza-specific CD8 + T cells engendered by FL CyCMV or dd CyCMV vectors. To this end, we first performed an intracellular cytokine staining (ICS) assay with the first fifteen 15mer-peptides overlapping by 11 amino acids that span the 1918 NP open reading frame to identify CD8 + T cell responses (Fig. 1E). To define the MHC restriction of peptides eliciting a CD8 + T cell response, we repeated the above ICS assay in the presence of each of the following reagents: the pan MHC-I-blocking antibody W6/32, the leader sequence-derived MHC-E-blocking VL9 peptide, the MHC-II-blocking G46.6 antibody, or isotype control reagents. NP-specific CD8 + T cells in dd CyCMV/Flu-vaccinated MCM recognized peptides in the context of MHC-E or MHC-II, while NP-specific CD8 + T cells in FL CyCMV/Flu-vaccinated MCM recognized peptides in the context of MHC-Ia (Fig. 1E). Finally, we measured whether CyCMV/Flu-elicited influenza-specific T cells could recognize diverse influenza isolates in vitro via incubation of PBMC from CyCMV/Flu-vaccinated MCM with inactivated influenza isolates. Both CD4+ and CD8 + T cells from FL and dd CyCMV/Flu-vaccinated MCM responded to all isolates tested, with the majority of isolates recognized similarly by T cells, regardless of the vaccine vector set utilized (Fig. 1F). Thus, both FL and dd CyCMV/Flu-vaccinated MCM generated robust influenza-specific T_EM capable of recognizing disparate influenza isolates in vitro, regardless of MHC-restriction, suggesting that these T cell responses might offer protection against challenge with heterologous HPAI.

Challenge with aerosolized avian H5N1 influenza

To measure the protective capabilities of CyCMV/Flu-induced influenza-specific T_EM cells against a heterologous challenge virus, we challenged all MCM with small-particle aerosols containing a target dose of 5.5 log₁₀ PFU of the HPAI isolate A/Vietnam/1203/2004(H5N1). To ensure experimental rigor and avoid introduction of any potential bias in clinical scoring, all study staff involved in the aerosolized influenza challenges were blinded to the vaccine status of the MCM until after completion of the challenge phase of the study (Fig. 1A). The inhaled dose was calculated by collecting an aerosol sample during each exposure to measure viral concentration in the aerosol and then multiplying aerosol concentration by the volume of total air inhaled by the MCM during the exposure, as previously described in ref. ²⁷. No statistical difference in the mean inhaled dose of aerosol virus was measured between the groups (Fig. 2A).

Following challenge with aerosolized influenza, all MCM were monitored for severity of infection via influenza titers in lung via BAL, changes in body temperature via telemetry, pulmonary infiltration via chest radiographs, and development of acute respiratory distress syndrome (ARDS) requiring euthanasia via a pre-defined clinical scoring sheet. As expected, based on experience with CMV-based vectors for SIV and MTB^22,24,29, CyCMV/Flu vaccination did not prevent infection. Infectious influenza virus was found in BAL fluid in every MCM following challenge (Fig. 2B). There was a trend towards lower viral titers in BAL fluid from dd CyCMV/Flu-vaccinated MCM at two and four days post infection, but this difference was not statistically significant. Infectious influenza began to clear from BAL fluid from survivors at day seven post infection, with no virus found in samples from any MCM at 15 days post infection. Longitudinal chest radiographs were performed and scored by radiologists blinded to the treatment of the animals, which revealed pulmonary infiltrates in the lungs of all MCM, with no statistically significant differences observed between the vaccine and control groups (Fig. 2C). All MCM developed fever following infection, which eventually resolved in CyCMV/Flu-vaccinated survivors but not in the unvaccinated controls (Fig. 2D). In line with previous results from this aerosol challenge model²⁷, all unvaccinated MCM developed ARDS and met humane endpoint criteria within seven days of exposure (Fig. 2E). In contrast, four of six dd CyCMV/Flu-vaccinated MCM survived through the 14-day post-challenge monitoring period, resulting in statistically significant protection from HPAI-induced death. Two out of five FL CyCMV/Flu-vaccinated MCM survived through day 14 post exposure, but protection in this group did not reach statistical significance. When both vaccine groups were combined, six of 11 CyCMV/Flu-vaccinated MCM survived, yielding overall vaccine-mediated statistically significant protection against lethal disease, regardless of the specific CyCMV/Flu vaccine vector utilized (Fig. 2E). Thus, while CyCMV-based vaccination did not prevent infection or significantly alter influenza-induced fever and pulmonary infiltration, it did significantly protect against HPAI-induced death.

Correlates of protection in CyCMV/Flu-vaccinated MCM

We recently demonstrated that a vaccine-phase, IL-15-based, whole blood transcriptomic signature of protection against SIV replication in strain 68-1 RhCMV/SIV-vaccinated RM was also present in dd CyCMV/SIV-vaccinated MCM that subsequently resisted SIV replication^29,36. To determine if a similar transcriptomic signature might exist in dd or FL CyCMV/Flu-vaccinated MCM that survived HPAI infection, we performed RNA sequencing of whole blood from all dd and FL CyCMV/Flu-vaccinated MCM immediately prior to vaccination and 1, 3, 7, and 14 days post prime and boost vaccination (Fig. 3A). We then assessed if the previously described RhCMV/SIV protection signature IL-15 response pathway genes enriched in dd CyCMV/SIV-vaccinated MCM could distinguish outcome following HPAI infection. The 122 genes previously identified in CyCMVSIV-mediated protection against SIV replication included genes involved in death receptor signaling, immune cell signaling programs, pattern recognition receptor signaling, and NK cell response (Supplementary Data 1). We plotted the fold change in these genes across the vaccination time series and assessed for conservation of this IL-15-based signature in CyCMV/Flu-vaccinated MCM that survived challenge with otherwise lethal aerosolized HPAI as previously described^29,36. This analysis revealed that the CyCMV/SIV-associated protection signature, although present in some vaccinated MCM, did not correlate with CyCMV/Flu-mediated protection against HPAI in either FL or dd CyCMV/Flu-vaccinated MCM, indicating a mechanistically distinct means of protection mediated by CyCMV/Flu against lethal HPAI compared to CyCMV/SIV-mediated protection against SIV replication (Fig. 3B).

The inability of the IL-15-based transcriptomic protection signature to distinguish protection outcomes following HPAI challenge indicated that unlike both RhCMV/SIV- and dd CyCMV/SIV-mediated protection against SIV replication^29,36, CyCMV/Flu-mediated protection against lethal HPAI infection did not depend upon MHC-E-restricted CD8 + T cells and the IL-15 signaling pathway. Therefore, we next assessed if the magnitude of the vaccine-induced T cell response might correlate with outcome. To this end, we examined the magnitude of influenza transgene-specific T cells immediately prior to entry into the ABSL3 for influenza challenge, which revealed that MCM surviving HPAI challenge mounted significantly higher influenza-specific CD4 + T cell responses in peripheral blood compared to MCM that succumbed (Fig. 3C). In particular, CD4 + T cells targeting the M protein in blood were significantly associated with survival. In contrast, there was no association observed with the magnitude of influenza-specific CD8 + T cells in blood prior to challenge. Based on these observations, we performed the same analysis using T cell frequencies measured in BAL, which again revealed that influenza-specific CD4 + T cells, but not CD8 + T cells, correlated with survival following challenge with HPAI (Fig. 3D). Cumulatively, these data indicate that unlike CMV vector-mediated protection against SIV replication, protection against lethal HPAI infection does not depend on CD8 + T cells, regardless of their MHC restriction, but rather on the magnitude of influenza-specific CD4 + T cell prior to infection.

Discussion

The development of a universal influenza vaccine remains a top global health priority, but cannot be achieved via current approaches that generate strain-specific humoral immune responses. Indeed, as recently demonstrated by SARS-CoV-2, despite the ability to rapidly produce lipid nanoparticle mRNA-based vaccines, pathogen variability inevitably still yields viral variants that escape vaccine-induced neutralizing antibodies and subsequently circulate in the human population³⁷. As such, novel vaccine approaches are needed that target conserved viral regions and provide heterosubtypic protection. To this end, we explored the potential of a CyCMV-vectored vaccine expressing conserved internal influenza antigens to protect against heterologous HPAI in a stringent macaque model of aerosolized influenza challenge. While murine CMV (MCMV) vectors expressing either a single influenza CD8 + T cell epitope or the entire HA protein have previously been shown to confer protection against influenza in mice^38,39, the data presented here represents, to our knowledge, the first test of CMV vectors against influenza in primates. We found that CyCMV/Flu vectors induced influenza-specific CD8+ and CD4 + T_EM, with the influenza-specific CD8 + T cells being restricted by MHC-Ia in FL CyCMV/Flu-vaccinated MCM and MHC-II or MHC-E in dd CyCMV/Flu-vaccinated MCM. Regardless of MHC-restriction, both CD8+ and CD4 + T cells from CyCMV/Flu-vaccinated MCM recognized a panel of diverse influenza isolates in vitro. As expected, based on results from CyCMV/SIV-vaccinated MCM challenged with SIV²⁹, CyCMV/Flu vaccination did not prevent acquisition of influenza in MCM. However, CyCMV/Flu-vaccinated MCM exhibited statistically significant protection from an otherwise lethal HPAI aerosolized challenge. This protection illustrates the potential of T_EM responses for development of a universal influenza vaccine given that the vaccine encoded 1918 influenza M1, NP, and PB1 protein sequences and the H5N1 HPAI isolate circulated in 2004, yielding 86 years of global influenza evolution between the vaccine and heterologous challenge virus. Therefore, CMV-induced T_EM responses targeting conserved viral antigens should be considered for inclusion in preventative approaches against pathogens such as influenza where sequence diversity in targets of antibody-mediated neutralization have stymied production of universally protective vaccines.

There exist many advantages for the CMV vaccine platform, which are not shared by other virus-based vaccine vectors. First, CMV is able to efficiently superinfect previously CMV-infected individuals despite pre-existing anti-CMV immunity^22,23,24,40. Second, CMV is highly immunogenic and elicits robust memory CD4+ and CD8 + T cell immunity. Indeed, in natural HCMV infection in humans, approximately 10% of circulating memory CD4+ and CD8+ memory T cells are HCMV-specific⁴¹. Third, CMV induced T cell responses are highly effector differentiated, thereby equipping these T cells with the ability to mediate an immediate effector response without the need to first anamnestically expand and differentiate^22,23,24. Finally, in keeping with an effector memory phenotype, CMV-induced CD4+ and CD8 + T cells are widely distributed at high frequency in various anatomical tissues, including the lungs. Although the CyCMV vaccine vectors here are spread-attenuated³⁰, they may still establish persistence in pulmonary myeloid lineage cells, thereby recruiting CyCMV/Flu-specific T cells to the lung. Based on these intriguing properties, and the unique protection against SIV replication observed in RhCMV/SIV-vaccinated RM, clinical trials are currently underway to test the safety and immunogenicity of a HCMV vaccine vector for HIV²⁶, Thus, the HCMV vaccine platform could be utilized to clinically test the approach presented here for a universal influenza vaccine.

Correlates analysis revealed that the previously identified IL-15 transcriptomic signature did not predict protection from lethal HPAI infection and that CD8 + T cells, regardless of MHC-restriction, did not associate with protection. Rather, the magnitude of the CyCMV-induced, influenza-specific CD4 + T cell response correlated to protection from HPAI-induced death. Although initially surprising given that both CyCMV/SIV- and RhCMV/SIV-mediated protection against SIV replication depends on the presence of SIV-specific, MHC-E-restricted CD8 + T cells and an IL-15-based signaling pathway^17,29,36, this observation is in line with previous reports of the importance of CD4 + T cells in influenza infection. Multiple studies have demonstrated that memory CD4 + T cells mediate heterosubtypic protection in murine models of influenza^42,43,44,45. While influenza-specific CD4 + T cells can orchestrate an effective immune response via interactions with both B cells and CD8 + T cells, they also demonstrate direct antiviral activity mediated via production of perforin and IFN-γ ^42,46. Furthermore, the frequency of pre-existing NP- and M-specific CD4 + T cells correlated with less severe disease in human clinical studies following infection with previously unencountered influenza strains¹⁰. Thus, our correlate of protection based on the magnitude of pre-infection frequencies of vaccine-induced IFN-γ + CD4 + T cells is in line with previous studies, and indicates that CMV vector-mediated protection against influenza could be improved by refining the vaccine to optimize the priming of influenza-specific CD4 + T cells. However, the mechanism of protection mediated by influenza-specific CD4 + T cells following CMV vector vaccination remains undefined. We previously demonstrated that CyCMV vaccine vectors elicit little to no antibody responses against antigenic inserts present in the vaccine vector²⁹. Therefore, it is unlikely that the CyCMV/Flu-elicited CD4 + T cells participated in priming a pre-existing antibody response targeting the internal viral proteins utilized as vaccine targets. Given that influenza-specific CD4 + T cells can exhibit direct antiviral activity^42,46, we hypothesize that CyCMV/Flu-elicited CD4 + T cells mediated protection by directly inhibiting viral replication. IL-15 has previously been identified to support the generation of superior, lung-resident antiviral CD4 + T cells with the ability to protect against lethal influenza infection in mice⁴⁷.

Since IL-15 plays a critical role in the development of effector differentiated CD4 + T cells, including cytotoxic CD4 + T cells⁴⁸, it is interesting that the IL-15-based transcriptomic signature that correlated with SIV protection mediated by MHC-E-restricted CD8 + T cells did not predict the protection against lethal HPAI observed here in the current study. It is therefore possible that IL-15 signaling was not a limiting factor for the function of the influenza-specific CD4 + T cells here as for the MHC-E-restricted CD8 + T cells in SIV protection^29,36. Subsequent statistically powered CyCMV/Flu vaccine-based HPAI challenge studies in MCM are required to determine the mechanism of protection mediated by influenza-specific CD4 + T cells.

Although MHC-E-restricted CD8 + T cells were not required for protection in the MCM model of influenza, the data presented here furthers our understanding of the epitope binding capacity of MHC-E by demonstrating that this monomorphic MHC molecule can bind a variety of influenza peptides for presentation to CD8 + T cells. Indeed, this observation is in line with two recent reports describing the ability of Qa-1, the murine orthologue of HLA-E in humans, and HLA-E to bind influenza-derived epitopes^49,50. Of note, the HLA-E-binding influenza epitopes identified were derived from NP, which we also found can generate peptide epitopes capable of binding Mafa-E, the HLA-E orthologue in MCM, and activating MHC-E-restricted CD8 + T cells. Interestingly, Jost et al. further demonstrate that the NP-derived, HLA-E-binding peptide epitopes are targeted by antigen-specific NK cells, suggesting that such antigen-specific NK cell responses could contribute to the antiviral response against influenza⁴⁹. Whether such responses could be elicited by CyCMV/Flu vaccine vectors remains unknown, and subsequent studies are required to fully investigate for the presence of memory NK cells following vaccination with CMV vectors. However, given the myriad of interactions documented to occur between CMV and NK cells^21,51,52, it is possible that CMV vectors may be able to prime memory NK cell responses.

Overall, we demonstrate that CyCMV vaccine vector-induced, influenza-specific CD4 + T_EM protect MCM from an otherwise lethal HPAI infection. While it remains likely that a universal influenza vaccine will have to engender both humoral and cellular immune responses, the CMV vaccine vector platform is unique given its ability to elicit and maintain long-lived T_EM, which could pair with a separate platform to stimulate both arms of the immune system. Furthermore, given that CMV vectors are now in clinical trials for HIV²⁶, there exists a direct pathway forward for testing influenza-specific HCMV vectors in humans. Further such studies are required to determine if CMV-induced T_EM can contribute to the development of a universal influenza vaccine.

Methods

Production of CyCMV/influenza vaccines

The CyCMV vector constructs used in this study were based on the published FL-CyCMV (strain 31908) bacterial artificial chromosome (BAC) and were generated by en passant homologous recombination^28,29. The codon optimized DNA sequences of the Influenza A virus (IAV) (A/Brevig Mission/1/1918(H1N1) derived NP (AY744935), M1 (AY130766) and PB1 (DQ208310) genes were synthesized by GENEWIZ (Azenta Life Sciences). No M2 sequences were included. Within each DNA sequence, a 50 bp stretch was duplicated and the two homologous sequences were separated by an I-SceI restriction site and an aminoglycoside 3’-phosphotransferase (Kanamycin resistance, KanR) selection cassette. The vaccine inserts were amplified with recombination primers carrying a 50 bp overhang homologous to the upstream and downstream region of the Cy110 (UL82, pp71) ORF. Homologous recombinations were performed in E.coli strain GS1783 which can be used to express the λ phage derived Red recombination genes after heat shock induction and recombination resulted in the substitution of the Cy110 ORF with the vaccine inserts using endogenous viral regulatory elements to drive transgene expression^25,29,30. Successful recombinants were analyzed by XmaI restriction digest and Sanger sequencing across the altered genomic locus to ensure genome integrity. The KanR cassette was removed from the vaccine insert by inducing a DNA double strand break in the BAC at the I-SceI- recognition site through the arabinose induced expression of I-Sce I in E.coli strain GS1783. Simultaneous expression of the Red recombination genes through heat shock induction resulted in homologous recombination of the introduced 50 bp homologous sequences in the DNA insert leading to seamless removal of the selection maker from the BAC. Final clones were once more analyzed by XmaI restriction digest and Sanger sequencing across the altered genomic locus and next generation sequencing of the full BAC was performed to exclude off-target mutations….