rAAV expressing a COBRA-designed influenza hemagglutinin generates a protective and durable adaptive immune response with a single dose

ABSTRACT

Influenza remains a worldwide public health threat. Although seasonal influenza vaccines are currently the best means of preventing severe disease, the standard-of-care vaccines require frequent updating due to antigenic drift and can have low efficacy, particularly in vulnerable populations. Here, we demonstrate that a single administration of a recombinant adenovirus-associated virus (rAAV) vector expressing a computationally optimized broadly reactive antigen (COBRA)-derived influenza H1 hemagglutinin (HA) induces strongly neutralizing and broadly protective antibodies in naïve mice and ferrets with pre-existing influenza immunity. Following a lethal viral challenge, the rAAV-COBRA vaccine allowed for significantly reduced viral loads in the upper and lower respiratory tracts and complete protection from morbidity and mortality that lasted for at least 5 months post-vaccination. We observed no signs of antibody waning during this study. CpG motif enrichment of the antigen can act as an internal adjuvant to further enhance the immune responses to allow for lower vaccine dosages with the induction of unique interferon-producing CD4+ and CD8+ T cells specific to HA head and stem peptide sequences. Our studies highlight the utility of rAAV as an effective platform to improve seasonal influenza vaccines.

IMPORTANCE

Developing an improved seasonal influenza vaccine remains an ambitious goal of researchers and clinicians alike. With influenza routinely causing severe epidemics with the potential to rise to pandemic levels, it is critical to create an effective, broadly protective, and durable vaccine to improve public health worldwide. As a potential solution, we created a rAAV viral vector expressing a COBRA-optimized influenza hemagglutinin antigen with modestly enriched CpG motifs to evoke a robust and long-lasting immune response after a single intramuscular dose without needing boosts or adjuvants. Importantly, the rAAV vaccine boosted antibody breadth to future strains in ferrets with pre-existing influenza immunity. Together, our data support further investigation into the utility of viral vectors as a potential avenue to improve our seasonal influenza vaccines.

INTRODUCTION

Throughout the 2022–2023 influenza season, the Centers for Disease Control (CDC) estimated that the United States encountered over 56 million influenza cases, leading to 650,000 hospitalizations and 58,000 deaths (1). Influenza represents a continual strain on our global healthcare systems, given that lower respiratory infections remain the fourth leading cause of death worldwide; at least half a million deaths are attributed to influenza annually (2, 3). Therefore, innovative ways to defend against influenza infections must be developed.

Vaccination represents the best defense against severe disease, although many obstacles routinely impede vaccine efficacy (4). Despite increased worldwide surveillance and advances in predictive algorithms, viral evolution makes antigen selection difficult (5). The standard-of-care trivalent and quadrivalent vaccine platforms containing inactivated forms of viruses grown in eggs or cells require antigen selection months before the influenza season, leaving ample time for antigenic mismatch to occur. These vaccines can also fail to generate sufficient immune responses following vaccination for several reasons. The overwhelming majority of adults have been exposed to influenza strains during their lifetimes through natural infection or previous vaccinations. This pre-existing immunity leads to original antigenic sin, which correlates with a reduced humoral response following influenza immunization with newer strains (6, 7). Studies have also demonstrated a rapid waning of antibody responses and loss of protection from future infections, especially in high-risk populations (8–12). With seasonal vaccine efficacy routinely below 50%, these challenges support a need to explore new vaccine technologies to overcome the limitations of traditional vaccines (13).

To address these challenges, we explored an easily programmable viral vector platform to deliver expression cassettes encoding influenza virus antigens: adeno-associated virus (AAV) vectors. AAV is a small single-stranded DNA parvovirus not associated with a known disease phenotype. In its recombinant form, AAV can be manipulated such that the capsid and inverted terminal repeats (ITRs) are the only viral elements present (14). Recombinant AAV (rAAV) vectors have shown tremendous promise for effecting therapeutic gene transfer and are the platform of five Food and Drug Administration (FDA)-approved drugs. Although the rAAV genome persists predominantly in episomal forms in transduced cells, it can provide durable transgene expression since it can transduce both dividing and non-dividing cells. In the context of gene transfer for monogenetic disorders, a strong anti-capsid immune response following vector administration was originally described as resulting from unmethylated CpG motifs in the expression cassette (15), which several pre-clinical and clinical studies went on to both confirm and associate with TLR9 activation (16–21). Although unfavorable in this context, we hypothesize that these motifs can be incorporated to augment an improved vaccine response.

The concept of rAAV-based vaccines is not new; other labs have developed vaccines for influenza, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), rabies, hepatitis C virus (HCV), human immunodeficiency virus (HIV), and various cancer models (22–28). The efficacy and durability of this platform as a SARS-COV-2 vaccine have been demonstrated to provide complete protection from viral challenges with long-term humoral responses maintained for at least 88 weeks post-vaccination (18, 23). Although successful, these studies utilize wildtype antigens, resulting in similar susceptibility to viral diversity as current standard-of-care platforms.

In this work, we incorporated the influenza hemagglutinin (HA) antigen designed with the Computationally Optimized Broadly Reactive Antigens (COBRA) methodology developed by the lab of Ted Ross (29). These novel sequences are generated using a layered consensus algorithm to incorporate a diverse range of broadly reactive and conserved epitopes into a single antigen. In adjuvanted dosing regimens, COBRA-derived recombinant proteins have protected against divergent strains, including H1, H2, H3, and H5 strains of prominent healthcare concerns (30–33), with superior breadth compared with traditional antigens (34) in several pre-clinical models with varying immune history profiles (35–37). We hypothesized that the durable COBRA HA presentation achieved following rAAV-mediated gene transfer and inherent immunogenic aspects of the vector would enable us to generate robust antibody and T-cell responses following a single dose without external adjuvants.

We demonstrate the ability of a rAAV9 COBRA HA vector administered intramuscularly at a low dose to elicit an immune response that provided complete protection from morbidity and mortality upon viral challenge in a mouse model. Our highly effective vaccine candidate can be administered at even lower dosages by deliberately increasing the antigen’s CpG motifs as an internal adjuvant. The broadly reactive boosts in humoral immune responses achieved in pre-immune ferrets show that these vectors can afford protection against new influenza strains, suggesting our vaccine’s ability to overcome the challenges traditionally seen with pre-existing influenza immunity in humans. The longevity and breadth of the immune response suggest that it may eliminate the need for annual immunizations. Altogether, our results highlight a promising proof-of-concept study with a product that may bring the field closer to an improved seasonal influenza vaccine that can be manufactured at large scales to increase vaccine access worldwide.

RESULTS

rAAV vector design and validation

We designed and tested a single-stranded AAV (ssAAV) vector (referred to as vector V1) where the expression cassette used a cytomegalovirus (CMV) promoter to drive antigen expression and included regulatory elements to boost mRNA stability and translation efficiency (Fig. 1A). AAV8 was selected as the capsid because of its extensive use in pre-clinical and clinical settings (38–41). As is typical for most influenza vaccines, the antigen was the immunodominant influenza viral surface glycoprotein HA (4–6, 42–46); specifically, the COBRA Y2 sequence (31) was generated from a consensus of H1 strains (referred to as COBRA HA) (31). Plasmids containing rAAV vector genomes encoding expression cassettes were transfected into HEK293T cells to verify transgene expression via western blot analysis (Fig. 1B).

Fig 1.

ssAAV2/8 vectors carrying the influenza hemagglutinin (HA) antigen protect mice against morbidity and mortality during lethal viral challenge. V1, a single-stranded AAV (ssAAV) vector, was designed to carry the COBRA HA antigen using an AAV2/8 capsid (A). The vector’s expression was validated in vitro with a western blot using Hek293T cells alone and recombinant HA protein as controls (B). Four-month-old male BALB/c mice were vaccinated intramuscularly with V1 at doses ranging from low (1.0 × 10¹⁰ vg), medium (1.0 × 10¹¹ vg) to high (1.0 × 10¹² vg), and control mice were vaccinated with either a self-complementary (scAAV) vector expressing green fluorescent protein (GFP) or sterile phosphate-buffered saline (PBS) for the mock-vaccinated group (n = 3/group) (C). Mice were bled at the indicated time points to measure alanine transaminase (ALT) and aspartate aminotransferase (AST) levels as well as antibody responses via hemagglutinin inhibition assays (HAI) (C). The mice were challenged with 1.8 × 10⁴ pfu/mL of pH1N1 strain A/California/04/2009 (CA/09) 6 weeks post-vaccination (C). Morbidity and mortality were monitored via weight loss (D) and survival (E). HAI titers against CA/09 were measured from receptor-destroying enzyme (RDE)-treated sera collected pre- and post-viral challenge (F). The dotted lines indicate HAI titers representative of seroprotection (1:40) and seroconversion (1:80). The “X” symbol represents “no survivors.” Data in panel (E) were analyzed with the Mantel-Cox test. Data in panels (D) (through 8 days post-challenge) and (F) were analyzed with two-way analysis of variance (ANOVA) with Tukey’s multiple-comparison test, with no significant differences detected between the low, medium, and high doses of V1. Data represented are biological replicates (average of two technical replicates) with means (standard error of the mean [SEM]) (*, P < 0.05; ***, P < 0.0005; ****, P < 0.0001). Any statistical comparisons not depicted were not statistically significant (P > 0.05).

As depicted schematically, mice were vaccinated intramuscularly (IM) with low, medium, and high doses (1 × 10¹⁰, 1 × 10¹¹, and 1 × 10¹² vector genomes [vg], respectively) of the ssAAV2/8 COBRA vector V1 (Fig. 1C). Control mice were vaccinated with 1 × 10¹¹ vg rAAV encoding green fluorescent protein (GFP) or sterile phosphate-buffered saline (PBS) (Fig. 1C). Six weeks post-vaccination, mice were challenged with a lethal dose of A/California/04/2009 (CA/09) H1N1 virus and monitored for morbidity via weight loss and mortality via survival. All control mice had substantial weight loss (Fig. 1D) and succumbed to the viral challenge (Fig. 1E). Conversely, all ssAAV2/8 COBRA vector V1-vaccinated mice, regardless of dose, were protected from morbidity (Fig. 1D) and mortality (Fig. 1E) and generated seroprotective hemagglutination inhibition (HAI) titers at all doses pre-challenge; these levels were only boosted in the “low-dose” group post-challenge (Fig. 1F). As expected, the control mice failed to generate HAI titers post-vaccination (Fig. 1F). With these promising initial results, we moved forward with further investigation into improving the ssAAV vector as a vaccine platform.

Capsid screen of rAAV CMV COBRA HA vaccine

AAV capsid pseudotyping, in which a vector genome includes ITRs of one serotype (usually AAV2) that is encapsidated within capsids of an alternate serotype, is an effective way to direct tropism to a desired target (47). To test the robustness and flexibility of our rAAV vaccine platform, we asked if there were capsid-specific differences in vaccine efficacy in our mouse model. We performed a comparative study whereby our AAV CMV COBRA HA vector genome was pseudotyped within capsids of alternative serotypes, some of which have previously been used in rAAV vaccine studies (22–28). Two of the capsids were developed using directed evolution to have high muscle expression and a limited capacity for hepatic transduction (48). We slightly altered the in vivo study design, lowering the dose from the initial study to 1 × 10⁹ vg to enable us to determine the most efficacious capsid for an IM vaccination (Fig. S1A). Regardless of the capsid, all mice given an rAAV vaccine mounted HAI titers and survived the viral influenza challenge with minimal weight loss (Fig. S1B, D through F).

Although the capsid-specific differences were modest, AAV9 yielded (1) consistently high pre-challenge HAI titers, (2) the least weight loss during the challenge, and (3) the highest viral titers produced in small-scale batches. Unlike AAV8, which performed similarly, we would anticipate less hepatic transduction using an IM AAV9 vaccine based on previous publications (49, 50). From this, we selected the AAV9 capsid to move forward for future studies.

ssAAV2/9 CMV COBRA HA is more immunogenic and effective than the split-inactivated quadrivalent vaccine platform

We next sought to evaluate the efficacy of the monovalent ssAAV2/9 COBRA HA vector V2 (Fig. 2A) at a 1 × 10¹⁰ vg dose alongside a quadrivalent vaccine control. The standard-of-care (SOC) Flucelvax is a commercially available quadrivalent inactivated influenza vaccine, administered at 1.5 µg/HA (Fig. 2B). Six weeks post-vaccination, sera samples were collected, and neutralizing antibodies were measured, showing that ssAAV2/9 COBRA V2-vaccinated mice produced stronger virus-specific IgG enzyme-linked immunosorbent assay (ELISA) responses (Fig. 2C) and neutralizing antibodies (Fig. 2D). Upon challenge, rAAV vaccine afforded complete protection against weight loss (Fig. 2E) and lethality (Fig. 2F), more rapid clearance of virus from the upper respiratory tract (nasal passages, Fig. 2G), and complete control of viral replication in the lungs (Fig. 2H). These data suggest a rapid and effective immune response to infection, which is in stark contrast to the mock-vaccinated mice that experienced severe weight loss (Fig. 2E) and less favorable survival outcomes (Fig. 2F). The Flucelvax-vaccinated mice had similar titers in the nasal washes and lung homogenates at days 3 and 6 post-challenge to the mock-vaccinated mice but were better able to control the infection by day 9 post-challenge (Fig. 2G and H). These findings correlate with significantly higher pre- and post-challenge HAI titers observed in the ssAAV2/9 COBRA V2-vaccinated group compared with the control groups; Flucelvax-vaccinated mice had no pre-challenge HAI titers to the CA/09 virus, and they only mounted HAI titers post-challenge (Fig. 2I). Although the COBRA HA antigen was designed from matched strains, it is essential to note that Flucelvax is a quadrivalent vaccine comprised of antigens from the currently circulating H1N1 strains. Collectively, these data lend credibility to rAAV’s promise as a vaccine platform.

Fig 2.

The ssAAV2/9 viral vector vaccine platform induces strong neutralizing antibodies and protection from viral replication, morbidity, and mortality following lethal viral challenge in male mice. V2, a single-stranded AAV (ssAAV) vector, was designed to carry the COBRA HA antigen using an AAV2/9 capsid (A). Two-month-old male BALB/c mice were vaccinated intramuscularly with 1.0 × 10¹⁰ vg of V2 vector, and control mice were vaccinated intramuscularly with Flucelvax (1.5 µg/HA) or sterile PBS (n = 29/group) (B). IgG ELISA (C) and microneutralization assays (D) against CA/09 were run on the RDE-treated sera collected 6 weeks post-vaccination, and the endpoint dilution titers were calculated for the half maximal inhibitory concentration (IC50). The mice were challenged with 1.8 × 10⁴ pfu/mL of pH1N1 strain A/California/04/2009 (CA/09) 6 weeks post-vaccination (B). Morbidity and mortality were monitored via weight loss (E) and survival (F). Viral titers were measured 3, 6, and 9 days post-challenge from nasal washes (G) and lung homogenates (H) from n = 3 mice/time point using a TCID₅₀ assay. HAI titers against CA/09 were measured from RDE-treated sera collected pre- and post-viral challenge (I). The dotted lines indicate the limit of detection (G, H) or HAI titers representative of seroprotection (1:40) and seroconversion (1:80) (I). “X” represents “no survivors.” Data in panels (C, D, E) (through 6 days post-challenge) (G, H, and I) were analyzed with two-way ANOVA with Tukey’s multiple-comparison test. Data in panel (F) were analyzed with the Mantel-Cox test. Data represented are biological replicates (average of two technical replicates) with means (SEM) (ns, not significant; *, P < 0.05; ***, P < 0.0005; ****, P < 0.0001). Any statistical comparisons not depicted were not statistically significant (P > 0.05).

ssAAV2/9 CMV COBRA HA maintains efficacy in female mice

As evidenced by the National Institute of Health’s initiatives over the past decade, it is imperative to investigate sex as a biological variable, especially in the context of preclinical studies for a vaccine designed to be given to all humans. Previous studies have demonstrated sex- and species-specific differences with both liver-directed rAAV-mediated transgene expression being lower and influenza disease pathogenesis being worse in females; these factors raised the concern that the combination may impede influenza vaccine efficacy in female mice (51, 52).

We evaluated ssAAV2/9 COBRA vector V2 administered intramuscularly between 1 × 10¹¹ and 1 × 10⁹ vg alongside Flucelvax and mock vaccine control in female BALB/c mice (Fig. 3A). We observed a dose-dependent effect of pre-challenge HAI titers (Fig. 3B). However, all female mice vaccinated with rAAV had reduced weight loss (Fig. 3C) and enhanced survival (Fig. 3D) compared with the controls. These results closely resemble what was previously observed in the male mice (Fig. 2), suggesting our platform has good efficacy and immunogenicity in both sexes.

Fig 3.

The ssAAV2/9 viral vector vaccine platform also induces strong antibody responses in female mice and protects against morbidity and mortality following lethal viral challenge. Two-month-old female BALB/c mice were vaccinated intramuscularly with the V2 vector at doses ranging from low (1.0 × 10⁹ vg), medium (1.0 × 10¹⁰ vg) to high (1.0 × 10¹¹ vg) (n = 5/group) (A). Control mice were vaccinated with Flucelvax (1 µg/HA) or sterile PBS (n = 2–5/group) (A). The mice were challenged with 9 × 10³ pfu/mL of pH1N1 CA/09 6 weeks post-vaccination (A). HAI titers against CA/09 were measured from RDE-treated sera collected pre- and post-viral challenge (B). The dotted lines indicate HAI titers representative of seroprotection (1:40) and seroconversion (1:80). “X” represents “no survivors.” Morbidity and mortality were monitored via weight loss (C) and survival (D). Data in panels (B and C) (through 10 days post-challenge) were analyzed with two-way ANOVA with Tukey’s multiple-comparison test. Data in panel (D) were analyzed with the Mantel-Cox test. Data represented are biological replicates (average of two technical replicates) with means (SEM) (**, P < 0.005; ***, P < 0.0005; ****, P < 0.0001). Any statistical comparisons not depicted were not statistically significant (P > 0.05).

Longevity of ssAAV2/9 CMV COBRA HA vaccine responses

Previous studies have demonstrated that rAAV vaccines can induce long-lasting immune responses (24). However, this has yet to be shown in the context of influenza. Unchallenged and challenged mice from the experiment described in Fig. 2 were bled monthly to determine the longevity of the immune response (Fig. 4A); these are referred to as the “Vax Only Group” and the “Vax + Challenge Group,” respectively. At 21 weeks post-vaccination, the “Vax Only Group” was challenged with CA/09 virus, whereas the “Vax + Challenge Group” was challenged with A/Hawaii/2019 H1N1—a drift variant of CA/09 virus—to test the breadth of protection; the groups were monitored for weight loss (Fig. 4B and D, respectively) and survival (Fig. 4C and E, respectively).

Fig 4.

The ssAAV2/9 viral vector vaccine platform vectors offer sustained antibody titers and protection against lethal viral challenges in mice for at least 5 months post-vaccination. Mice from the experiment described in Fig. 2B were kept for longevity following either vaccination alone or vaccination and viral challenge (n = 5/group) (A). Sera samples were collected every 6 weeks until the mice were challenged at 21 weeks post-vaccination with 1.8 × 10⁴ pfu/mL of pH1N1 strain A/California/04/2009 (CA/09) for the primary challenge of the “Vax Only Group” or 10⁶ TCID₅₀/mL of Hawaii/19 (HI/19) for the secondary challenge of the “Vax + Challenge Group” (A). Morbidity and mortality were monitored via weight loss (B, D) and survival (C, E) for the CA/09 and HI/19 challenges, respectively. HAI titers were measured against CA/09 from RDE-treated sera collected at the indicated time points for the “Vax Only Group” (F, on the left) and the “Vax + Challenge Group” (F, on the right). The dotted lines indicate HAI titers representative of seroprotection (1:40) and seroconversion (1:80). “X” represents “no survivors.” Microneutralization assays against CA/09 were run on the RDE-treated sera collected pre- and post-challenge from both groups, and the IC50 endpoint dilution titers were calculated for the CA/09 (G, on the left) and HI/19 challenges (G, on the right), respectively. Data in panels (B) (through 6 days post-challenge) (D, F, and G) were analyzed with two-way ANOVA with Tukey’s multiple-comparison test. Data in panels (C and E) were analyzed with the Mantel-Cox test. Data represented are biological replicates (average of two technical replicates) with means (SEM) (**, P < 0.005; ***, P < 0.0005; ****, P < 0.0001). Any statistical comparisons not depicted were not statistically significant (P > 0.05).

For the homologous challenge (Fig. 4B and C), all ssAAV2/9 COBRA vector V2- and Flucelvax-vaccinated mice survived. Sixty percent of mock-vaccinated mice survived, which may be explained by older mice having more balanced immune-pathological responses to influenza infections that contribute to decreased mortality (53). For the heterologous challenge (Fig. 4D and E), all vaccinated mice survived the heterologous challenge with minimal weight loss, indicating that ssAAV2/9 COBRA vector V2 and Flucelvax offer similar long-term protection following an initial viral challenge.

The ssAAV2/9 COBRA vector V2-vaccinated mice from both groups maintained strong HAI titers at 6, 12, and 18 weeks post-vaccination (Fig. 4F). The Flucelvax-vaccinated mice only produced HAI titers in the Vax + Challenge Group (following their initial viral challenge at 6 weeks post-vaccination) after they recovered at 12 weeks post-vaccination (Fig. 4F).

Neutralization titers were measured on pre- and post-challenge sera from both groups, revealing boosted titers following challenge in the “Vax Only Group” and a statistically significant increase in neutralization in ssAAV2/9 COBRA vector V2-vaccinated mice compared with Flucelvax-vaccinated mice following the secondary challenge in the “Vax + Challenge Group” (Fig. 4G). These data suggest that the ssAAV2/9 COBRA vector V2-vaccinated mice with immune history could generate higher neutralizing antibodies to the strain they had previously encountered. In contrast, the Flucelvax-vaccinated antibody levels remained the same.

These findings are consistent throughout our study (Fig. 2D, I, and 3B); other studies in humans and mice have shown low HAI titers and influenza-specific responses following Flucelvax vaccination (43, 54). The ssAAV2/9 COBRA vector V2-vaccinated mice maintain strong neutralizing antibodies, seropositive HAI titers, and complete protection against lethal viral challenges for at least 5 months after vaccination.

CpG enrichment of antigen coding sequence improves vaccine efficacy at low doses

Our initial rAAV vaccine studies demonstrated robust efficacy at all tested doses with symptom-free survival from lethal viral challenge. At this ceiling, it was difficult to determine if modifications to the vector design could improve efficacy. Therefore, we performed a dose de-escalation challenge study from 1 × 10¹⁰ vg of ssAAV2/9 COBRA vector V2 to determine the minimum threshold needed for effectiveness; this range would allow us to identify doses to evaluate our alternative vector designs. We observed minimal weight loss (Fig. 5A) and complete survival (Fig. 5B) at 1 × 10¹⁰ and 1 × 10⁹ vg. Mice vaccinated with 1 × 10⁸ and 1 × 10⁷ vg exhibited significant weight loss and struggled to recover from the viral challenge, whereas mice vaccinated with 1 × 10⁶ vg fared worse and succumbed to the challenge (Fig. 5A and B).

Fig 5.

A dose de-escalation study reveals that the ssAAV2/9 viral vector platform loses protection against morbidity and mortality in mice at 1 × 10⁸ vg doses, but efficacy at lower doses can be rescued by increasing the CpG motifs within the vector’s antigen as an internal adjuvant. To perform the dose de-escalation study, 2-month-old male BALB/c mice were vaccinated intramuscularly with 1.0 × 10¹⁰, 1.0 × 10⁹, 1.0 × 10⁸, 1.0 × 10⁷, and 1.0 × 10⁶ vg of V2 vector (n = 5 mice/group). The mice were challenged with 1.8 × 10⁴ pfu/mL of pH1N1 strain A/California/04/2009 (CA/09) 6 weeks post-vaccination. Morbidity and mortality were monitored via weight loss (A) and survival (B). Two additional ssAAV2/9 vectors were designed to carry the same COBRA HA antigen but were manually reverse codon-optimized to express a medium (V3) or high (V4) number of CpG motifs compared with the original V2 vector (C). The new vectors were validated in vitro and in vivo (Fig. S2). Two-month-old male BALB/c mice were vaccinated intramuscularly with 5.0 × 10⁸ vg of V2, V3, or V4 vector (n = 10 mice/group), whereas control mice were vaccinated with Flucelvax (1.5 µg/HA) or sterile PBS (n = 3–5 mice/group). The mice were challenged with 1.8 × 10⁴ pfu/mL of pH1N1 strain A/California/04/2009 (CA/09) 6 weeks post-vaccination. Virus-specific IgG ELISA titers were measured from RDE-treated sera collected pre-challenge (D). HAI titers against CA/09 were measured from RDE-treated sera collected pre- and post-viral challenge (E). The dotted lines indicate HAI titers representative of seroprotection (1:40) and seroconversion (1:80). “X” represents “no survivors.” Morbidity and mortality were monitored via weight loss (F) and survival (G). Data in panels (A) (through 8 days post-challenge) and (D, E, and F) (through 6 days post-challenge) were analyzed with two-way ANOVA with Tukey’s multiple-comparison test. Data in panels (B and G) were analyzed with the Mantel-Cox test. Data represented are biological replicates (average of two technical replicates) with means (SEM) (*, P < 0.05; **, P < 0.005; ****, P < 0.0001). Any statistical comparisons not depicted were not statistically significant (P > 0.05).

Previous studies have shown that increasing the CpG content of the transgene elicits an increased anti-transgene immune response (15, 19–21, 55). After establishing 1 × 10⁹ vg as the lowest dose that offered ample protection, we asked whether increasing the number of CpG motifs within the antigen would serve as an internal adjuvant to improve the vaccine responses and compensate for the lower doses. Using ssAAV2/9 COBRA vector V2 as our baseline with “low-CpG content,” we codon-optimized the COBRA HA sequence to incorporate 50% (medium-CpG content) and 100% (high-CpG content) of all possible CpG motifs to create ssAAV2/9 COBRA CpG-Medium vector V3 and ssAAV2/9 COBRA CpG-High V4, respectively (Fig. 5C).

Because increased GC content can create secondary structures, we validated vector homogeneity using an alkaline gel (Fig. S2A). Additionally, we observed that the CpG enrichment in the vector antigens did not affect vector production (Fig. S2B). For further initial validation, the ssAAV2/9 COBRA vectors V3 and V4 performed similarly to ssAAV2/9 COBRA vector V2 (Fig. 2) in vivo at a higher dose of 1 × 10¹⁰ vg in terms of HAI titers pre- and post-challenge (Fig. S2C), weight loss (Fig. S2D), mortality (Fig. S2E), and viral titers detected in the nasal washes (Fig. S2F) and lung homogenates (Fig. S2G) during lethal viral challenge.

We then vaccinated mice below the lower protection limit with 5 × 10⁸ vg of ssAAV2/9 COBRA Low-CpG vector V2, ssAAV2/9 COBRA Medium-CpG vector V3, or ssAAV2/9 COBRA High-CpG vector V4. Control mice were vaccinated with 1.5 µg/HA Flucelvax or PBS for mock vaccination. At 6 weeks post-vaccination, mice were bled to assess CA/09 virus-specific IgG ELISA titers (Fig. 5D) and seroconversion via HAI titers (Fig. 5E). At this low dose, we saw a minimal pre-challenge humoral immune response in the ssAAV-vaccinated groups (Fig. 5D and E). No pre-challenge antibodies were detected in the Flucelvax- or mock-vaccinated controls via HAI (Fig. 5E). However, there were low levels of CA/09 virus-specific IgG titers in the Flucelvax-vaccinated mice (Fig. 5D). During the viral challenge, all vaccinated groups showed less severe weight loss (Fig. 5F) and improved survival (Fig. 5G) compared with the mock-vaccinated group. Outside of the mock vaccine control, the ssAAV2/9 COBRA Low-CpG vector V2 and Flucelvax groups experienced the most severe weight loss (Fig. 5F) and poorest survival outcomes (Fig. 5G). Only the CpG-enriched ssAAV2/9 COBRA Medium-CpG vector V3 and ssAAV2/9 High-CpG vector V4 vaccines protected the mice at the lower vaccine dose (Fig. 5G). These results support our hypothesis that CpG-enriched antigens can enable us to maintain efficacy at lower vaccine dosages.

CpG enrichment of vaccine vector impacts CD4+ T cell response

Although we observed improved survival of the mice vaccinated with 5 × 10⁸ vg CpG-enriched ssAAV vectors (Fig. 5G), we did not see significant differences in the humoral response pre- or post-challenge (Fig. 5E). These results suggest that another mechanism mediates the improvement in efficacy. Thus, we explored the impact of CpG enrichment on T-cell responses to COBRA HA peptides. Single-cell suspensions were generated from spleens and lungs harvested 10 weeks post-vaccination. The cells were stimulated with two pools of peptides as listed in Table S1: one specific to the highly conserved HA stem, and the other to the diverse and immunodominant HA head. Intracellular staining following peptide stimulation was also performed to identify cytokine-producing CD4+ and CD8+ T cells.

Using the gating strategy described in Fig. S3, flow cytometry revealed that regardless of the antigen’s CpG enrichment, the rAAV vector can induce CD4+ and CD8+ T cells reactive to the influenza HA head and stalk peptides following administration at 5 × 10⁸ vg (Fig. 6). These responses are observed in the lungs (Fig. 6A through D) and in the spleen (Fig. 6E through H), suggesting a primed local response in the respiratory tract that allows the host to respond swiftly to viral infection. The limitation of this study is that the organ harvest protocol does not distinguish between circulating cells in the vasculature and those in the lung tissue (56), but the T cells are indeed present within the lungs. Although not always statistically significant, the ssAAV2/9 COBRA High-CpG vector V4-vaccinated mice tended to have higher, albeit more variable, frequencies of the cytokine-producing T cells. We observed a significant increase in CD4+IFN-γ+ cells in the lungs (Fig. 6B) and spleens (Fig. 6H) of the ssAAV2/9 COBRA High-CpG vector V4-vaccinated mice reactive to the HA head and stalk, respectively. The Flucelvax-vaccinated control mice could generate CD8+ T cell responses to the more conserved HA stalk but had lower responses to the HA head likely due to strain mismatch (Fig. 6). The rAAV-vaccinated mice mounted stronger responses to the HA head and quite robust CD4+ T cells, as well (Fig. 6). The ssAAV2/9-luciferase and mock-vaccinated mice had negligible responses to the peptide stimulation (Fig. 6). These data support the claim that the CpG enrichment at low doses may generate more antigen-specific T-cell responses, corresponding to increased antigen recognition and protection against morbidity and mortality following viral challenge (Fig. 5).

Fig 6.

The ssAAV2/9 vector vaccine platform induces influenza-specific T-cell responses in both the lungs and spleens of mice following vaccination. To understand if the improved vaccine responses from vectors V3 and V4 with enriched CpG motifs (compared with V2) when administered at a low dose of 5.0 × 10⁸ vg Fig. 5 was attributed to changes in T-cell responses, we performed peptide stimulation with pools of HA-specific epitopes followed by intracellular cytokine staining at 10 weeks post-vaccination (n = 4-5 mice/group). As controls, we compared mice vaccinated with 1.0 × 10¹⁰ vg ssAAV2/9 Luciferase, Flucelvax (1.5μg/HA), or sterile PBS 7 weeks post-vaccination (n = 4 mice/group). We observed influenza-specific responses in the lungs (A–D) and spleens (E–H) generated by CD8+ (A, C, E, G) and CD4+ (B, D, F, H) T cells producing cytokines TNF-α, IFN-γ, or IL-2 specific to the HA head (A, B, E, F) and stalk (C, D, G, H) peptides. Data represented in panels (A–H) are biological replicates (normalized to unstimulated controls) shown as box and whisker plots with the tails representing the minimum and maximum values with mean values marked as lines within the boxes. Data in panels (A–H) were analyzed with two-way ANOVA with Tukey’s multiple-comparison test (*, P < 0.05; **, P < 0.005; ***, P < 0.0005; ****, P < 0.0001). Any statistical comparisons not depicted were not statistically significant (P > 0.05).

Utilization of COBRA antigen generates broadly reactive antibodies

It has been previously demonstrated that using COBRA-derived recombinant proteins improves the breadth of the antibody response (34, 57, 58). To understand if those benefits are maintained in the context of an rAAV vaccine, we generated a ssAAV2/9 vector encoding a wildtype CA/09 H1 HA derived from the sequence of the CA/09 virus used for challenges in mice (referred to as vector V5; Fig. S4A). We validated its immunogenicity via HAI titers (Fig. S4B) and efficacy via weight loss (Fig. S4C), survival ( Fig. S4D), and viral titers from nasal washes (Fig. S4E) and lung homogenates (Fig. S4F) following a lethal viral challenge. Not surprisingly, ssAAV2/9 wildtype vector V5 had equivalent in vivo efficacy to its counterpart, ssAAV2/9 COBRA vector V2.

The COBRA antigen design’s major advantage is the extended breadth of antibody responses compared with wildtype sequences. We compared the breadth of antibody responses generated in vaccinated mice via HAI titers across a panel of contemporary H1 influenza strains using sera collected at 9 weeks post-vaccination (Fig. 7A). Regardless of whether the mice had undergone viral challenge, all rAAV vectors induced an impressive breadth of seroprotective antibody responses across the panel. Without viral challenge, the ssAAV2/9 COBRA vectors V2, V3, and V4 generated a slightly higher magnitude of responses across the panel than the ssAAV2/9 wildtype vector V5. Following viral challenge, this trend continues where the increased CpG content in ssAAV2/9 COBRA Medium-CpG vector V3 and ssAAV2/9 COBRA High-CpG vector V4 offer slightly more substantial breadth and magnitude of antibody titers than the ssAAV2/9 wildtype vector V5 (Fig. 7A). These data suggest that increasing the CpG content of the COBRA antigens further improves the vaccine-induced antibody responses.

Fig 7.

We observe a robust breadth of antibody responses in mice vaccinated with the ssAAV viral vector vaccine platform carrying the COBRA HA regardless of capsid selection or CpG enrichment. Sera samples were collected at 9 weeks post-vaccination to assess differences in antibody breadth due to (i) antigen design by comparing with sera from V2-, Flucelvax-, and mock-vaccinated mice (as described in Fig. 2) and (ii) CpG enrichment of the antigen, by comparing with sera from V3- and V4-vaccinated mice (as described in Fig. S2). HAIs were performed on the RDE-treated sera against a range of drifted H1N1 strains, including California/09, Michigan/15, Idaho/18, and Hawaii/19 from mice that either had (n = 5/group; B, at right) or had not (n = 5/group; B, at left) undergone viral challenge at 6 weeks post-vaccination. The dotted lines indicate HAI titers representative of seroprotection (1:40) and seroconversion (1:80). “X” represents “no survivors.” The breadth of antibody responses was also measured using the same viruses using sera obtained from the experiment described in Fig. S1 (n = 5/group) shows broad responses generated regardless of capsid at 6 weeks post-vaccination, although with differences in the magnitude of those responses (C). Data in panel (B) were analyzed with two-way ANOVA with Tukey’s multiple-comparison test. Data represented are biological replicates (average of two technical replicates) with means (SEM) (*, P < 0.05; **, P < 0.005; ***, P < 0.0005; ****, P < 0.0001). Any statistical comparisons not depicted were not statistically significant (P > 0.05).

To ensure the robustness of the rAAV platform, we also screened the sera obtained from mice 6 weeks after they were vaccinated with the ssAAV encoding the COBRA HA antigen with a diverse range of capsids (from the experiment described in Fig. S2) against the same H1N1 panel. We observed that regardless of capsid choice, there is a decent breadth of antibody responses; however, there are slight differences in the magnitude of responses (Fig. 7B). Although not statistically significant, AAV9 responses tended to be one of the highest out of the screened capsids.

rAAV vaccines are well-tolerated and do not cause liver toxicity in mouse models

A concern for rAAV administration is liver toxicity (23, 49). We measured alanine transaminase (ALT) and aspartate aminotransferase (AST) levels in the circulating sera as a proxy for liver damage. The IM 1 × 10¹⁰ vg ssAAV2/8 COBRA vector V1-vaccinated mice (experiment described in Fig. 1) did not experience heightened ALT or AST levels compared with the control mice (Fig. S5A and B) (49, 59). We also measured ALT and AST levels to test the safety associated with rAAV9 vaccination. Sera samples were obtained from female mice 6 weeks post-vaccination (experiment described in Fig. 3) and male mice 6 and 9 weeks post-vaccination (experiments described in Fig. 2; Fig. S2). ALT and AST enzymes in female mice (Fig. S5C and D, respectively) and male mice (Fig. S5E and F, respectively) revealed no signs of liver damage in the IM 1 × 10⁹, 1 × 10¹⁰, or 1 × 10¹¹ vg ssAAV2/9 COBRA vector V2-vaccinated mice.

In addition to the capsid screen results discussed in Fig. S1, another desirable feature of the AAV9 capsid is that it has less liver tropism than the AAV8 capsid (49, 59). We assessed the biodistribution by intramuscularly vaccinating mice with 1 × 10¹⁰ vg of ssAAV2/9 expressing CMV COBRA HA or luciferase, or sterile 1× PBS for the mock-vaccinated control mice (Fig. S5G). After 12 weeks, we measured the luminescence signal within the liver and muscle (Fig. S5H). The imaging revealed strong transduction within the muscle, and minimal expression within the liver or elsewhere (average values of 90.67%, 0.46%, and 8.87% of total body luminescence, respectively, for luciferase-vaccinated mice) (Fig. S5H). Representative images of the luciferase-vaccinated mice show intense signal in the muscle at the site of injection (Fig. S5I and J) such that it was only possible to measure any other luminescence throughout the body if the muscle was not exposed (Fig. S5K); these images are in stark contrast to the ssAAV2/9 HA and mock-vaccinated controls (representative images shown in Fig. S5L and M).

ssAAV2/9 CMV COBRA HA is effective in ferrets with influenza immune history

We then assessed vaccine immunogenicity and efficacy in our ferret models. As a small pilot study, we compared naïve ferrets with those vaccinated with ssAAV2/9 COBRA vector V2 (Fig. 8A). V2-vaccinated ferrets had broadly neutralizing antibodies to our panel of H1N1 viruses (Fig. 8B) 9 weeks post-vaccination. Following viral challenge with CA/09, the V2-vaccinated ferrets had less weight loss (Fig. 8C) and lower viral titers in their sneezes (Fig. 8D) than the naïve ferrets. Six days post-challenge, the ferrets were humanely euthanized, and viral titers were measured from each lung lobe (Fig. 8E).

Fig 8.

The ssAAV2/9 viral vector vaccine platform overcomes pre-existing influenza immunity in the ferret model to offer an increased breadth of antibody responses. Ferrets were vaccinated with 2.0 × 10¹¹ vg of vector V2 and bled 9 weeks post-vaccination (A). RDE-treated sera were screened for neutralizing ability across a range of drifted H1N1 strains, including California/09, Michigan/15, Idaho/18, and Hawaii/19 (B). Naïve and V2-vaccinated ferrets (n = 2/group) were challenged with 10⁶ TCID₅₀/mL of pH1N1 strain A/California/04/2009 (CA/09) (A). Morbidity was measured via weight loss (C) and viral titers from collected sneezes (D). Following euthanasia at 6 days post-challenge, viral titers were measured from each lung lobe via TCID₅₀ assay (E). The dotted lines indicate the limit of detection (D, E). Ferrets with previous immune history to H3N2 (Memphis/19) and H1N1 (California/09) were vaccinated with 2.0 × 10¹¹ vg of ssAAV2/9 carrying the COBRA HA or GFP (n = 2/group) and then challenged with 10⁶ TCID₅₀/mL of H1N1 strain A/Michigan/45/2015 (MI/15) (F). Sera samples were collected 6 weeks pre-vaccination, 9 weeks post-vaccination, and 21 days post-challenge. RDE-treated sera were screened for HAI titers against the diverse H1N1 panel. The HAI data pre-challenge were graphed as fold change from pre-vaccination titers (G). Ferrets were monitored for morbidity via weight loss throughout the challenge (H). The HAI data post-challenge were graphed as fold change from pre-vaccination titers (I). Data in panels (C, D, E, G, H, and I) were analyzed with two-way ANOVA with Tukey’s multiple-comparison test. Data represented are biological replicates (average of two technical replicates) with means (SEM) (ns, not significant; **, P < 0.05; ***, P < 0.005; ****, P < 0.0001). Any statistical comparisons not depicted were not statistically significant (P > 0.05).

Finally, to look at how vaccine efficacy is influenced by more complex immune history in this model, we assessed the performance of our vector V2 compared with a ssAAV2/9 GFP control vector in pre-immune ferrets with a history of influenza CA/09 H1N1 and A/Memphis/257/2019 (H3N2 MEM/19) viral infections (Fig. 8F). Sera was collected before vaccination for a baseline and again at 9 weeks post-vaccination to measure the increase in antibody breadth following the rAAV administration (Fig. 8F). We observed a significant improvement in HAI titers against all tested H1N1 strains of influenza in our ssAAV2/9 COBRA vector V2-vaccinated ferrets compared with the ssAAV2/9 GFP-vaccinated controls (Fig. 8G). We then challenged the ferrets with MI/15, a drifted H1N1 strain (Fig. 8F). The V2-vaccinated ferrets had slightly less weight loss during the viral challenge (Fig. 8H) and had a significantly improved breadth of HAI responses post-challenge (Fig. 8I). These data suggest the ability of the rAAV platform to not only increase the magnitude of strain-specific antibodies from past exposures but also improve the breadth of those responses to new antibodies as well.

DISCUSSION

Influenza vaccination is arguably the most effective means of reducing disease severity. However, producing effective vaccines is challenging for many reasons, including protecting from viral drift variants and generating long-lasting immune responses. The CDC publishes results from annual studies predicting vaccine efficacy, and although the vaccines offer moderate protection, there is room for improvement.

We developed a series of rAAV vectors that can be administered as a single, unadjuvanted vaccine. These induced strongly neutralizing and seroprotective antibodies, a feat that is difficult for most vaccine platforms to accomplish. The rAAV vectors provided complete protection against morbidity and mortality following lethal viral challenges by reducing viral replication throughout the respiratory tract within the first few days of the viral challenge.

Although the rAAV platform induced a broad, robust adaptive immune response with a wildtype-derived antigen (as used in vector V5), we observed an even greater magnitude of responses when utilizing a COBRA-derived antigen (as used in vectors V1-V4) within our platform in naïve animals (Fig. 7 and 8). Importantly, the COBRA antigen allowed for greater breadth in the face of pre-existing influenza immunity in our ferret model and induced broadly protective antibodies to a panel of H1N1 viruses spanning an entire decade, suggesting good forward protection (Fig. 8). A critically valuable component of our study is that our ferrets had a history of H1N1 and H3N2 infections before rAAV vaccination, better mimicking the pre-existing influenza immunity of people who have previously encountered influenza (either through infection or vaccination); our rAAV vaccine was able to overcome the ferrets’ pre-existing influenza antibodies to generate broader antibodies toward future drifted strains they had not encountered before (Fig. 8). This advantage would reduce the vaccine’s susceptibility to antigenic mismatch during production ahead of the influenza season by generating broadly reactive antibodies against new and unfamiliar strains without being limited by the immune system biases from antigenic sin (6).

This study further validates prior work that rAAV-mediated immunity is long-lasting (23). We observed no signs of waning 5 months post-vaccination in mice, and other groups have tracked larger animal models much longer (23, 25, 60). With such long-lasting, robust responses, our platform may eliminate the need for seasonal vaccinations.

Our vaccine was effective across multiple naturally occurring and lab-engineered capsids with dramatically different tissue tropisms. The lower HAI titers observed in the AAV-MYO vaccine compared with AAV8 or AAV9 suggest that myocyte transduction is less crucial for vaccine efficacy (Fig. S1B). Since capsid tropism has primarily focused on target organs using luciferase, it is unclear what specific cell type drives this efficacy. Direct transduction of antigen-presenting cells (APCs) may be responsible for the robust immune response; however, AAV6/DJ showed no benefit relative to AAV8, which does not transduce APCs (61, 62). Further investigation is needed to characterize the driving force behind rAAV vaccine efficacy.

The concept of an rAAV vaccine is not novel, since vaccine efficacy has been demonstrated in multiple mouse and disease models (22–28). Many studies show strong vaccine efficacy and survival following viral challenges. The published studies utilize various capsids and strong viral promoters for high levels of antigen expression. Most of these studies were done with at least 1 × 10¹⁰ vg/mouse, with some requiring three 1 × 10¹¹ vg/mouse doses to achieve complete efficacy (21–23, 25–28).

In this study, our innovative approach exploits clinical trial findings to augment the immune response. We observed efficacy with doses down to 5 × 10⁸ vg. Enriching CpG motifs in the antigen coding sequence improved vaccine efficacy at a lower dose: a 50% reduction to achieve the same outcomes as our codon-optimized ssAAV8 COBRA vector V1 and ssAAV9 COBRA vector V2. We achieved the same vaccine effectiveness at 600-fold lower dosages than other published work (22). Although dosing in mice does not directly translate to dosing in humans, this reduction suggests an improvement in efficacy.

It is common practice for AAV gene transfer vectors to be CpG-depleted. Although increasing the number of motifs within our vector’s encoded antigen led to improved vaccine responses at low doses, it is also possible that the sequence may lead to increased immunopathology, inflammation, and less predictable influenza-specific responses in terms of T-cell activity. We predict that the rapid silencing of the CMV promoter (63, 64) and CpG-enriched transgenes observed in the BAX-335 trial may prevent prolonged HA expression following vaccination and alleviate some safety concerns (19). It is still unclear how the titration of CpG motifs within the vector’s antigen specifically impacts the cellular and humoral responses following vaccination, but a recent publication has identified a plasmacytoid dendritic cell-like population that is activated by vector administration for gene transfer (65).

Many next-generation influenza vaccines are being designed for intranasal delivery to target mucosal immunity rather than systemic immunity (45). Strong mucosal responses might block viral transmission from occurring or potentially stop the virus from infecting the upper respiratory tract in the first place, which would drastically improve infection control in pandemic responses. However, our data suggest that intranasal administration might not be necessary for an rAAV vaccine to achieve mucosal immunity. In Fig. 2G through H, our vaccine prevents influenza replication in the mouse lung, suggesting the induction of localized mucosal immunity. In Fig. 6, antigen-specific T cells to the HA head and stem in the lung are detected, and CpG enrichment of the vector appears to increase their frequency.

Although we have succeeded in naïve animals and ferrets with pre-existing influenza immunity, whether our vector can overcome pre-existing AAV capsid immunity remains a critical question. It has been estimated that one-third of adults have antibodies against AAV8 and AAV9, which can impact the clinical value of this vaccine platform (66). Furthermore, if vaccination confers anti-capsid responses, there is concern about whether using the same capsid would be effective for future vaccinations or gene therapy. Although the lack of pre-existing humoral immunity to the vector is part of the current criteria for clinical trial eligibility, new treatment regimens are being explored to enable patients with immune responses to the vector to benefit from gene therapy (38, 67, 68).

Pre-existing capsid immunity could, in fact, be helpful by restricting transduction to the administration site and preventing undesired off-target transduction, such as in the liver (69). Current data demonstrate that IM injection of rAAV is still effective in large animal models with pre-existing capsid immunity (70). However, the doses and transgenes assessed are not informative in the context of a vaccine study; hence, this warrants further investigation. Our data suggest that a wide range of divergent and synthetically engineered capsids are similarly effective (Fig. S1), indicating that we may be able to work around individuals’ pre-existing AAV capsid immunity status. To bypass the pre-existing immunity concern, we can simply select a synthetic capsid in which no natural immunity would be present. The Asokan lab has recently published one such strategy to develop capsids in this manner (71). Additionally, extensive structural biology research has been conducted to create novel antibody escape variant vectors, notably from the Agbandje-McKenna lab (72). If vector redesigns are required for seasonal vaccinations, it may be possible to introduce novel vaccine variants each year that are distinct from those typically utilized in gene therapy.

Another commonly referred to limitation of rAAV is its small carrying capacity. Packaging human genes can be challenging, and in some cases, these restrictions force scientists to delete non-essential protein domains or to split the transgene into multiple vectors. However, viral antigens for influenza and other viruses are considerably smaller; it is plausible that multiple antigens can fit into a single vector. The standard-of-care vaccine platforms do not confer robust immune responses against neuraminidase (NA). The rAAV carrying capacity can easily accommodate bivalent vector combinations with HA and NA. Once optimized, they could be administered together as a quadrivalent formula to offer improved protection against circulating strains.

rAAV has been widely explored in gene therapy for rare monogenetic disorders; decades of research have resulted in five FDA-approved drugs. The thousands of patients dosed in these trials demonstrate the safety of this platform, especially when delivered at low doses. The nearly limitless vector design modifications could allow for the creation of multivalent respiratory vaccines that protect a panel of viruses like influenza, SARS-CoV-2, and respiratory syncytial virus (RSV). The findings in this study encourage a serious pursuit into further evaluating the potential for rAAV as an alternative to the traditional inactivated standard-of-care platforms.

MATERIALS AND METHODS

Cell lines and primary culture

Madin-Darby canine kidney cells (MDCK cells; ATCC: CCL-34) were grown at 37°C under 5% CO₂ and maintained in growth media: minimum essential medium (MEM; Lonza) supplemented with 2 mM GlutaMAX (Gibco), 1 mM sodium pyruvate (Gibco), and 10% fetal bovine serum (FBS; Atlanta Biologicals). For the in vitro viral infections, the MDCKs were plated with infection media: serum-free MEM supplemented with 0.075% bovine serum albumin fraction V (BSA; Gibco), 2 mM GlutaMAX, and 1 µg/mL tosylsulfonyl phenylalanyl chloromethyl ketone (TPCK)-treated trypsin. For the microneutralization assays, the MDCKs were plated with microneutralization media: serum-free MEM with 1% GlutaMAX and 1% BSA. Human embryo kidney tissue cells (HEK293T cells; ATCC: CRL-3216) were grown at 37°C under 5% CO₂ and maintained in growth media: Dulbecco’s modified Eagle medium (DMEM) supplemented with 200 mM glutamine and 10% FBS.

Vector assembly

The COBRA Y2 antigen sequence (designed from human H1N1 sequences isolated between May 2014 and September 2016) (31) was provided by the lab of Ted Ross. A ssAAV vector was designed in SnapGene and ordered as fully constructed from Genewiz. Alternative vectors were generated via PCR amplification of COBRA HA using NEB Q5 High-Fidelity 2× Master Mix (New England Biolabs), followed by restriction digest and ligation via T4 ligase into vectors generated for prior publications and transformation into NEB Stable Competent E. coli cells. Plasmid sequences were verified via whole plasmid sequencing by Harvard Biosciences. CpG-enriched COBRA HA sequences were generated via manual codon optimization in Integrated DNA Technologies (IDT)’s online tool to either contain 50% or 100% of possible CpG motifs without altering the amino acid sequence. Sequences were ordered from Genewiz as a plasmid and then cloned into the previously purchased ssAAV CMV COBRA HA vector.

Western blot

On day 1, 735,000 Hek293T cells were plated in a tissue culture (TC)-treated 6-well dish with growth media. On day 0, cells were transfected with 2.5 µg of each COBRA HA transgene plasmid per well using an appropriate amount of polyethylenimine (PEI). On day 2, the media was aspirated, and the cells were washed in 1× DPBS before promptly being lysed in 400 µL of 2× Laemmle buffer. Samples were sonicated, heated at 95°C for 10 minutes, and then, 5 µL was loaded in each well of a Bio-Rad 4%-20% gradient pre-cast SDS PAGE gel alongside the Bio-Rad dual color protein standard. The gel was run at 60 volts for 3 hours and transferred onto a PVDF membrane via the Bio-Rad Turbo Transfer kit. The membrane was washed briefly, blocked in 6% BSA for 3 hours shaking gently at room temperature, and subsequently incubated in 1:5,000 dilution anti-HA Clade 1 antibody (Immune-Tech; Catalog #IT-003–001M14) overnight at 4°C. The primary antibody was removed, and the membrane was washed three times for 5 minutes each using 1× DPBS-T and incubated in 1:10,000 goat anti-mouse IgG antibody (Invitrogen; Catalog #A10680) for 1 hour at room temperature. The secondary antibody was removed, the membrane was washed three times for 5 minutes using 1× DPBST, and the membrane was imaged using a LI-COR after brief incubation in SuperSignal (Thermo Fisher) solution.

Vector production for capsid screen

The vectors for the capsid screen were produced by St. Jude’s Vector Core. Briefly, SJ293TS or VPC cells were transfected with TransIT-VirusGEN transfection reagent (Mirus Bio, Madison, WI, USA) using a three-plasmid system: (i) transfer vector, (ii) rep/capsid helper plasmid, and (iii) HGTI-adenoI. Vector cell pellets were collected 48 hours post-transfection by centrifugation at 932 × g, and the supernatant was discarded. Cells were dispersed and pooled using PBS and centrifuged at 932 × g for 5 minutes. The supernatant was discarded, and the cell pellet was suspended in 1× TD buffer (0.71 mM K2HPO4, 25 mM Tris, 5.0 mM KCl, 140 mM NaCl, and 6.4 mM MgCl2). Cell pellets underwent three freeze/thaw cycles using a −80°C freezer and a 37°C water bath. Two thousand five hundred units of benzonase (Millipore-Sigma, Burlington, MA, USA) were added to cell pellet lysates and incubated at 37°C for 1 hour. Lysates were clarified by centrifugation at 1,900 × g for 20 minutes, overlayed over a 60%/40%/25%/15% iodixanol density gradient, then placed into the VTi50 rotor of the Beckman Coulter Ultracentrifuge and centrifuged at 49,000 rpm at 18°C for 1 hour. The resulting 0.45 mm polyethersulfone (PES)-filtered iodixanol fraction(s) were then diluted in PBS and concentrated using a Vivaspin 20 with a membrane of 100,000 MCO PES (Sartorius, Gottingen, Germany).

Large-scale vector production

Hek239T cells were plated at 10 million cells/15 cm dish. The cells were expanded until they reached a sufficient quantity to seed 4.3 × 10⁸ cells/cell factory. Twenty-four hours after seeding the cell factories, they were transfected using a three-plasmid system with PEI. An adenoviral helper plasmid, AAV2-9 RepCap plasmid, and transgene-containing plasmid were transfected at a 1:1:1 ratio. 72 hours post-transfection, the media was harvested, and the cells were washed in 1× DPBS and lifted via incubation in 1× DPBS + ethylenediaminetetraacetic acid (EDTA) at 37°C. The cell pellet was spun down at 2,500 × g and washed three times in 1× DPBS to remove the EDTA. All solutions were combined, sterile-filtered, and concentrated using the Pellicon (PCC300C01). The cell pellet was freeze-thawed five times to extract virus, spun down at 2,500 × g, and combined with a concentrated medium. The sample was treated with Benzonase with added magnesium for 1 hour at 37˚C, diluted to 300 mL in 1× DPBS, sterile-filtered, and run overnight on either POROS CaptureSelect AAVX Affinity Resin (Catalog #A36740) or POROS CaptureSelect AAV9 Affinity Resin (Catalog number #A27353) depending on the serotype. Fractions containing the virus were concentrated via centrifugation at 1,000 × g in a 30 kd Amicon filter to a total volume of 1 mL. The vector was sterile-filtered, and recombinant mouse albumin was added such that the final concentration was 0.25% by volume.

Vector titers

Vector titers were extrapolated against a plasmid standard as previously described (73). In short, plasmid standards were generated via restriction enzyme linearization followed by phenol-chloroform purification. The DNA was resuspended and aliquoted such that 2.5 × 10⁹ plasmid copies were present in 5 µL volume and promptly frozen for long-term storage. The chromatography-purified vector was mixed sufficiently and serially diluted in nuclease-free ddH₂O with 0.01% F68 detergent. Vector genomes were quantified alongside serially diluted plasmid standards using SYBR Green (Applied Biosystems) and primers targeting the CMV promoter or antigen sequence (FWD ATATGCCAAGTACGCCCCCTATTGAC, REV ACTGCCAAGTAGGAAAGTCCCATAAGGTC). The concentration of vector genomes per mL was then extrapolated from the sample signal in relation to the plasmid standard.

Alkaline gel

The alkaline gel was run as previously described (74). In short, a 0.8% agarose gel was prepared with 50 mM NaOH and 1 mM EDTA. Running buffer was prepared similarly sans agar. 1 × 10¹¹ vg was diluted to 16.5 µL and mixed with 8.5 µL loading buffer (20% glycerol, 1.2% SDS, 2.5 M NaOH, 50 mM EDTA). Samples were run alongside 5 µL DNA HyperLadder (Meridian Bioscience) diluted similarly in loading buffer. The gel was run at 15 v for 18 hours at 4°C, neutralized with 0.1M Tris-HCL, pH 8.0, stained with GelRed (Millipore; Catalog #SCT123) diluted in 0.1M NaCl shaking gently at room temperature shielded from light for 3 hours. The gel was rinsed twice with ddH₂O and imaged on a LI-COR.

Influenza viral titer assays

Influenza viral titers were quantified with a hemagglutination assay (HA) or 50% tissue culture infectious dose assay (TCID₅₀) in MDCK cells as described previously (75). For the HA assay, hemagglutination units (HAU) were determined by incubating serially diluted virus with equal amounts of 0.5% turkey red blood cells (tRBCs) in a V-bottom 96-well plate (Corning) at room temperature for 30 minutes to allow for the erythrocytes to settle at the bottom of the wells. After the incubation, the plate was tilted to score the samples: hemagglutination-negative samples stream in a “tear-drop” fashion, whereas hemagglutination-positive samples do not. The HAU was determined by the endpoint of the virus titration, which is the highest dilution causing complete hemagglutination. For the TCID₅₀ assay, samples were serially diluted in infection media and added to washed MDCKs in flat bottom 96-well TC-treated plates (Corning). The plates were incubated either at 37°C for 3 days. Following this incubation, the samples were scored with the HA. The infectious viral titers were calculated using the Reed-Muench method (76).

Animal vaccinations

Seven- to 8-week-old male or female BALB/cJ (RRID:IMSR_JAX:000651) mice were lightly anesthetized with isoflurane and vaccinated with varying dosages of rAAV vectors or Flucelvax (1.5 µg/HA; 2022–2023 formula, Seqirus) diluted in sterile 1× PBS in a total volume of 50 µL. The mock-vaccinated mice were vaccinated with 50 µL of sterile 1× PBS. All vaccines were administered intramuscularly to the left gastrocnemius muscle.

Eight-month-old male ferrets (Triple F Farms) were vaccinated intramuscularly into the left quadriceps muscle with 2 × 10¹¹ vg rAAV vector expressing COBRA HA or GFP diluted in sterile 1× PBS in a total volume of 200 µL.

Challenge viruses

The mouse challenge viruses included mouse-adapted A/California/04/2009 H1N1 or A/Hawaii/70/2019 influenza virus. The ferret challenge viruses included wild-type A/California/04/2009 H1N1 or A/Michigan/45/2015. As described previously (75), the stock virus was grown in 9- or 10-day-old specific-pathogen-free embryonated chicken eggs at 37°C. After 3 days, allantoic fluid was harvested and cleared of debris by centrifugation (1,800 × g for 5 minutes). The supernatants were used to determine HAU. The virus was aliquoted and stored at −80°C until use. One aliquot of the virus stock was thawed to quantify the infectious virus using a TCID₅₀ assay.

Animal viral challenges

All viral challenges were conducted in animal biosafety level 2. Six to 21 weeks post-vaccination, as indicated, the mice were lightly anesthetized with isoflurane and intranasally challenged with either: 1.8 × 10⁴ pfu/mL of CA/09 virus diluted with sterile 1× PBS in a total volume of 25 µL, corresponding to a 10× mouse median lethal dose (MLD₅₀) viral challenge or 10⁶ TCID₅₀ units of H1/19 virus diluted with sterile 1× PBS in a total volume of 25 µL. At various time points, nasal washes were obtained by flushing 0.5 mL sterile 1× PBS through a catheter inserted into the trachea; lungs were also harvested and homogenized with 0.5 mL sterile 1× PBS through bead-beating for viral titers via TCID₅₀ assay. The mice were monitored for 14 days post-challenge for morbidity via weight loss and clinical scores (0 = smooth, shiny coat; 1 = hunch; 2 = 1 + scruffy fur; 3 = 2 + slow gait; 4 = moribund; 5 = dead) and mortality via survival. Mice were euthanized when they reached their humane endpoints of a maintained score of 3 and >30% body weight loss.

Nine weeks post-vaccination, the ferrets were lightly anesthetized with isoflurane and challenged with 10⁶ TCID₅₀ CA/09 or MI/15 diluted with sterile 1× PBS in a total volume of 500 µL, split between each nostril. The ferrets were monitored for 14 days post-challenge for morbidity via clinical scores, temperatures, and weight loss. At 2, 4, and 6 days post-challenge, the ferrets were injected with 0.25 mL of ketamine (Patterson Veterinary Supply; 100 mg/mL). Once anesthetized, 1.0 mL sterile 1× PBS containing 100 U/mL penicillin and 100 µg/mL streptomycin was dripped onto their nostrils to trigger a sneeze response. The sneezes were captured in specimen collection cups. The cups were placed on ice and spun at 1,000 × g for 5 minutes to collect the samples for viral titers via TCID₅₀ assay. Ferrets were humanely euthanized at the end of the study by a licensed veterinarian via Euthasol injection (Patterson Veterinary Supply). Individual lobes of the lungs were harvested and homogenized to measure viral titers via TCID₅₀ assay.

Blood chemistries

To quantify ALT and AST enzymes, mice were bled via retro-orbital bleeds using non-heparinized micro-hematocrit capillary tubes (Fisher Scientific). Blood was centrifuged at 10,000 × g for 20 minutes, and the sera were collected. The neat sera were added to the ALT (Horiba: A11A01627) and AST (Horiba: A11A01629) cassettes and run on the ABX Pentra 400 analyzer.

Serology

To determine antibody titers after vaccination or challenge, mice were bled retro-orbitally using heparinized micro-hematocrit capillary tubes (Fisher Scientific), and the ferrets were bled via jugular veins following light anesthetization with isoflurane. Blood was centrifuged at 10,000 × g for 20 minutes, and the sera were collected. As described previously (77), one volume of sera was incubated with three volumes of receptor-destroying enzyme (RDE II; Hardy Diagnostics) at 37°C for 20 hours first and then at 56°C for 1 hour. Six volumes of 1× PBS were added to each reaction for a final dilution of 1:10 before freezing at −80°C for 2 hours.

Hemagglutination inhibition assay (HAI)

As described previously (77), the RDE-treated sera were serially diluted and incubated with 4 HAU of virus at room temperature for 15 minutes. An equal volume of 0.5% turkey red blood cells (RBCs) was added to the reaction before incubation at 4°C for 30 minutes. After the incubation, the plate was tilted to score the samples: sera with neutralizing antibodies inactivate the virus, which results in the erythrocytes streaming in a “tear-drop” fashion; however, sera with non-neutralizing antibodies cannot inactivate the virus, which results in cloudy wells without the “tear-drop” streaming of erythrocytes. The HAI titer is determined by the endpoint of the sera titration that causes the viral neutralization.

Microneutralization assays

Hundred microliters of 3 × 10⁴ MDCK cells/well were plated in TC-treated 96-well plates (Corning) and incubated overnight. In fresh plates, serially diluted sera were incubated with 2 × 10³ TCID₅₀/mL reverse genetics CA/04/09 luciferase virus (78) in microneutralization media for 1 hour at 37°C. The MDCK cells were washed twice with 200 µL/well of PBS. Hundred microliters of the RDE-treated sera/virus mixture was added, and the plates were incubated at 37°C for 3 days. The media was then removed, and the plates were frozen at −80°C for 1 hour before thawing at 4°C for 30 minutes. Twenty-five microliters NanoGlo Luciferase substrate (Promega) was added to each well, and the plates were read after 3 minutes on a Xenogen at the 460/40 emission. IC50 values were then calculated for visualization.

Whole-virus IgG ELISAs

Briefly, 5 µg/mL of purified wtCA/09 virus was coated onto 384-well flat bottom plates (ThermoFisher) in 15 µL total volume/well overnight at 4°C. The next day, coated plates were washed three times with 100 µL of PBS-T (1× PBS with 0.1% Tween-20, Sigma), then blocked with 100 µL/well of PBS-T with 3% Goat Serum (Sigma) and 0.5% nonfat milk powder (AmericanBio) for 1 hour at room temperature. The blocking solution was removed and replaced with a fresh blocking solution for the serial sera dilutions. After a 2-hour incubation at room temperature, the plates were washed three times with 100 µL/well of PBS-T. Fifteen microliters/well of horseradish peroxidase (HRP)-labeled goat anti-mouse IgG (Invitrogen, 1:3,000 dilution in PBS-T) was added and incubated for 1 hour at room temperature. Plates were washed three times with 100 µL/well of PBS-T, and 25 µL of SigmaFast o-phenylenediamine dihydrochloride (OPD) substrate (R&D) was added to each well and developed for 10 minutes. Twenty-five microliters of 3M hydrochloric acid was added to each well to stop the reaction. Plates were read at 490 nm using a BioTek plate reader.

Luciferase imaging

At 12 weeks post-vaccination, mice were anesthetized with isoflurane and injected intraperitoneally with 200 µL of luciferin reagent. Briefly, 1 g of In Vivo Imaging System (IVIS)brite D-Luciferin Potassium Salt Bioluminescent Substrate (Perkin Elmer) was dissolved into 5 mL CaCl₂/MgCl-free PBS to generate the stock solution. The stock solution was subsequently diluted to a working solution of 15 mg/mL PBS and sterile-filtered using a 0.2 µm filter. Following the luciferin injection, the mice were placed in a Xenogen IVIS200 instrument. The luminescence was measured for 3 minutes, and the values were extracted from the raw images using the IVIS Living “Image 64-bit” program.

Flow cytometry assays

Mouse spleens and lungs were collected 6–10 weeks post-vaccination and processed to make single-cell suspensions. Briefly, the organs were dissociated in 1× Hank's balanced salt solution (HBSS) using gentleMACS C Tubes with a gentleMACS Tissue Dissociator (Miltenyi Biotec) or mincing with razor blades. The lungs were incubated with digestion media (0.1% collagenase IV [Worthington] and 0.01% DNAseI [Worthington] in HBSS) for 30 minutes at 37°C with 5% CO₂. Cells were filtered through nylon mesh, and then, red blood cells were lysed using ammonium chloride solution (Stemcell). Cells were counted using a Vi-CELL BLU Cell Viability Analyzer (Beckman Coulter) and 1 × 10⁶ cells were plated in flat bottom 96-well plates. For peptide stimulation, the cells were stimmed for 6 hours with 1 µM pooled peptides provided by the lab of Andrea J. Sant at 37°C with 5% CO₂. Control cells received either no stim or phorbol 12-myristate 13-acetate (PMA) and ionomycin (Invitrogen). Following the 4-hour incubation with Golgi Stop/Plug (BD Biosciences), the cells were washed and FC blocked before extracellular staining with an antibody cocktail of CD8 (FITC), CD4 (BV785), TCRb (PerCP-CY5.5), CD44 (APCFire785), CD62L (PE-Cy7), CD103 (PE Dazzle), and CD69 (BV650) at a concentration of 1:200, Ghost Dye Live/Dead (BV510) at a concentration of 1 µL/well. The cells were fixed using the Fixation/Permeabilization solution (BD Biosciences) and washed with Perm/Wash Buffer (BD Biosciences) before intracellular staining with an antibody cocktail of IFN-γ (PE), TNF-α (APC), and IL2 (BV421) at a concentration of 1:100. The cells were washed and transferred to FACS tubes to be run on a DIVA Fortessa instrument. Analyses were performed in FlowJo.

Statistical analyses

Data were organized in Microsoft Excel and analyzed in GraphPad Prism 9.5.1. The particular statistical analyses performed for each experiment are detailed in their corresponding figure legends.

ETHICS APPROVAL

The mice and ferrets for these experiments were provided with food and water ad libitum along with additional cage enrichment as needed. Animals were cared for and handled by trained technicians and routinely monitored by veterinarians. All animal procedures were approved by the St. Jude Children’s Research Hospital Institutional Animal Care and Use Committee and followed the Guide for the Care and Use of Laboratory Animals (79).

DATA AVAILABILITY

The authors confirm that the data supporting the findings of this study are available within the article and the supplemental material. Per NIAID contract requirements, the data sets from these studies have been uploaded to ImmPort under SDY2733.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/jvi.00781-24.

Supplemental material. jvi.00781-24-s0001.docx.

Figures S1 to S5; Table S1.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.