Live-attenuated pandemic H1N1 influenza vaccines expressing Computationally Optimized Broadly Reactive Antigens (COBRAs) are immunogenic and protective in mice and ferrets

Abstract

Computationally optimized broadly reactive antigens (COBRAs) induce broad and protective immune responses across multiple viral vaccine platforms. However, their suitability for incorporation into live attenuated influenza vaccines (LAIVs) remains uncertain, as antigen modifications could potentially impact LAIV generation, replication, stability, or immunogenicity. In this study, COBRA hemagglutinin (HA) and neuraminidase (NA) antigens designated Y2 and N1-I, respectively, of the influenza H1N1 subtype were reverse-engineered into A/Puerto Rico/8/1934 (PR8)-based LAIVs. The impact of HA stability and NA composition on LAIV properties was evaluated in cell culture, mice, and ferrets. COBRA LAIV yields were higher in MDCK cells compared to Vero cells, and a higher HA activation pH was associated with increased LAIV growth in cell culture. The COBRA LAIVs elicited broad antibody responses against pandemic H1N1 viruses and provided robust protection in both mice and ferrets. The standard COBRA LAIV, containing unmodified HA Y2 and NA N1-I, had virus inactivation pH and HA activation pH values of 5.4 and 5.6, respectively. In contrast, a modified COBRA LAIV, containing an HA2-K153E mutation and NA from the vaccine strain A/Hawaii/70/2019 (HI19), had a virus inactivation pH of 5.3 and an elevated HA activation pH of 6.0. This modified LAIV had improved growth in cell culture and greater protection from challenge virus lung titers in elderly ferrets. These studies demonstrate the successful integration of COBRA antigen engineering into a LAIV platform. Furthermore, fine-tuning HA stability and NA composition appears to be a promising strategy to enhance LAIVs containing modifications to computationally optimized antigens.

1. Introduction

Seasonal influenza vaccine effectiveness (VE) ranges from approximately 10–60% due to factors such as antigenic mismatch, egg-adapted mutations, egg-derived glycosylation, pre-existing immunity, and limited immunogenicity in high-risk groups [1]. To address these challenges, cell-based production platforms and alternative vaccine technologies have been developed, including nucleic acid, peptide-based vaccines, virus-like particles (VLPs), and viral vectors [2]. Adjuvants such as aluminum salts, MF59, and TLR ligands have been used to enhance immunogenicity, particularly in vulnerable populations [3]. Despite these advancements, predicting antigenic drift remains difficult, contributing to fluctuating VE. Modified hemagglutinin (HA) antigens, including stalk-only HA [4], glycosylation-masked HA [5], and chimeric HA proteins [6], aim to broaden immunity by targeting conserved stalk regions. Additional strategies, such as mosaic [7] and computationally optimized broadly reactive antigens (COBRAs) [8], enhance responses by incorporating epitopes from diverse strains or multi-step consensus sequences to minimize sampling bias, respectively. This study evaluates introducing COBRA methodology within a live attenuated influenza vaccine (LAIV) platform to determine its impact on viral replication and immunogenicity.

Consensus-based approaches, including COBRA, have improved vaccine immunogenicity for viruses like chikungunya, hepatitis C, and HIV [9]. COBRA antigens have been studied in influenza contexts across various platforms, including recombinant HA/NA proteins, VLPs, and mRNA vaccines, eliciting robust antibody responses and broad protection compared to wild-type antigens [10–13]. However, their application in LAIVs remains unexplored. LAIVs mimic natural infection, inducing mucosal and systemic immunity [14], and provide better cross-protection against mismatched antigens than inactivated vaccines [15, 16]. Here, we sought to incorporate COBRA-HA-Y2 and NA-N1-I antigens into a candidate H1N1 LAIV using a modified A/PR/8/34 (PR8) master donor virus [17, 18]. This modified PR8 virus contains the mutations PB2-N265S, PB1-K391E, PB1-E581G, PB1-A661T, resulting in a cold adapted (ca) and attenuated (att) phenotype. COBRA-Y2-HA was computed using amino-acid sequences from pandemic H1N1 (pH1N1) isolates from 2014–2016, and COBRA-NA-N1-I was derived from N1-subtype NA sequences from avian, swine, and human isolates from 1990–2015 [19, 20].

The HA and NA glycoproteins are key immunodominant antigens in influenza, with hemagglutination inhibition (HAI) titers often correlating with VE [21]. HA binds sialic acid, facilitating viral entry, while NA cleaves sialic acid to enable viral movement [22, 23]. HA stability (the pH at which HA undergoes irreversible conformational changes) impacts infectivity, host range, and pandemic potential [24]. Although COBRA computation optimizes antigen sequences, it does not account for HA stability or HA-NA functional balance, properties critical for LAIV replication and antigen expression. To address these uncertainties, we developed a cold-adapted PR8 (ca-PR8) LAIV encoding COBRA-HA-Y2 and COBRA-NA-N1-I, evaluated HA stability mutations, and assessed HA-NA balance effects using a seasonal NA strain.

2. Materials and methods

2.1. Cells and viruses

Madin-Darby canine kidney (MDCK; ATCC-CCL-34) and Vero (ATCC-CCL-81) cells were cultured as described [25]. A/California/04/2009 (CA09), A/Michigan/45/2015 (MI15), A/Brisbane/02/2018 (BR18), and A/Guangdong-Maonan/SWL1536/2019 (GD19) were grown in MDCK cells at 33°C and, after 3-day incubation, were harvested, aliquoted, sequence-confirmed, and stored at −80°C. For pHW2000 COBRA-HA-Y2, the 12 necessary amino-acid substitutions were introduced into pHW2000 A/Tennessee/1–560/2009 (TN09) HA (Supplementary Figure S1). HA2-K153E and HA1-Y17H mutations were generated by site-directed mutagenesis. pHW2000 COBRA N1-I was designed similarly to Y2, using A/Brevig Mission/1/1918 (H1N1) NA as a reference gene (Supplementary Figure S2). The six internal-gene pHW2000 ca-PR8 plasmids have been described [17]. LAIVs were recovered by reverse genetics using Vero cells then amplifying in MDCK cells as described [26].

2.2. HA stability assays

Virus inactivation pH was measured as described [25]. Briefly, LAIV aliquots were treated with pH-adjusted buffer (1:100) at 37°C for 1h and neutralized before titration by TCID₅₀ in MDCK cells at 33°C. Inactivation pH was calculated as the inflection pH using GraphPad Prism 10. HA activation pH was measured by syncytia assay [25]. Briefly, Vero cells were infected at a multiplicity of infection (MOI) of 3 TCID₅₀/cell and incubated at 33°C for 16–18h. Cells were treated with M199 + 5 μg/ml TPCK-trypsin at 37°C for 15 minutes, neutralized with M199 + 10% FBS medium, rinsed with dPBS^+/+, and treated again with pH-adjusted buffer at 37°C for 15 minutes. The cells were overlaid with M199 + 10% FBS medium at 37°C for 3h, fixed, and stained with Hema 3 Stat Pack (Fisher Scientific, Kalamazoo, MI). Syncytia were photographed via brightfield microscopy. HA activation pH was measured as the highest pH value of syncytia formation.

2.3. LAIV growth

Multiple-step replication was measured in MDCK and Vero cells at 33°C or 37°C using an MOI of 0.001 PFU/cell in the presence of 1 μg/ml TPCK-treated trypsin. These cell types were selected as opposed to primary human respiratory epithelial cells (which are well suited to studying LAIV attenuation and replication) because MDCK and Vero cells are used for production of LAIVs. Infection was initiated by covering cells with diluted viruses and incubating for 1h. The inoculum was replaced with infection medium (containing 1 μg/ml TPCK-treated trypsin), and the cells were incubated for 72h. Samples were collected every 24h and stored at −80°C.

2.4. Mice

Groups of five 6-week-old female DBA/2 mice were intranasally inoculated with 5×10⁴ PFU. At 3d post-immunization (DPI), nasal turbinates and lungs were collected and homogenized using Qiagen Tissue Lyser II homogenizer (30/s frequency for 45 s, operated twice). After ultracentrifugation, the supernatant was collected for virus titration. To measure immunogenicity, groups of ten 6-week-old female DBA/2 mice were intranasally immunized with 5×10⁴ PFU of LAIV, and weight loss was monitored for 14 days. Blood was collected at 28-DPI and 60-DPI. Mice were then challenged with 2×10⁵ PFU of BR18 at 61-DPI. Half of the mice were euthanized 3-days post-challenge (DPC) to measure lung titers. Remaining mice were monitored for weight change.

2.5. Ferrets

Groups of six adult (5 months old) and elderly (>6 years old) ferrets were immunized (1×10⁶ TCID₅₀/ferret), and nasal washes were collected at 1- and 3-DPI. Blood was collected at 7-, 9-, and 28-DPI for HAI and neutralization assays, which were performed according to WHO standard procedures [27]. At 31- or 33-DPI, ferrets were challenged with BR18 (1×10⁶ TCID₅₀/ferret). Nasal washes were collected two days after challenge. Three days after challenge, the ferrets were sacrificed, and blood and tissues were collected.

2.6. ELISAs

NUNC^™ MaxiSorp^™ 96 Well Plates were coated with purified inactivated virus particles (4 μg/ml; 50 μl/well). 3% BSA in PBS was used as a blocking buffer. Nasal washes were 2-fold serial diluted in blocking buffer, added to the plates (50 μl/well), and incubated at room temperature for 2h. Anti-ferret IgA-ALP (Goat Anti-Ferret IgA (α-chain specific)-Alkaline Phosphatase, SAB3700788, Sigma) was used for IgA detection. The cut-off was the average absorption of blank wells using a 3-fold standard deviation. The endpoint titer was defined as the highest dilution of sera exhibiting absorption signal higher than the cut-off value.

2.7. LAIV stability

To measure stability at 4°C, LAIV aliquots were transferred from −80°C to a 4°C water bath sequentially over the course of 12 weeks so that they could be titrated for remaining infectivity by TCID₅₀ at the same time.

2.8. NA activity

NA activity was measured as the amount of fluorescent product, 4-MUSS, which was cleaved from MUNANA (Sigma) by viral neuraminidase after 30 minutes of incubation at 37°C [28]. Briefly, virus dilutions were prepared with 1X enzyme buffer (32.5 mM MES (Sigma) and 4 mM CaCl₂ (Fisher), pH 5.0–6.5) and then mixed with MUNANA working solution (200 μM of MUNANA with 1X enzyme buffer). The reaction was incubated at 37°C for 30 minutes in the dark and terminated by stop solution (0.1M glycine (Sigma) in 25% ethanol, pH 10.7). The production of 4-MUSS was detected using 360/460 nm (excitation/emission) wavelengths and calculated using a titration of 4-MUSS from 2.5 μM to 20 μM as a standard curve. The amount of virus was measured by HA assay, which was shown to correlate with NA content in pilot studies. The NA activity was shown as the production of 4-MUSS per HAU after 1h reaction time (μM × HAU⁻¹ × hr⁻¹).

2.9. Statistical analysis

Statistical significance was determined by one-way ANOVA comparing multiple groups with GraphPad Prism 10.

3. Results

3.1. COBRA proteins can be incorporated into LAIVs

The sequences of COBRA-HA-Y2 and COBRA-NA-N1-I were derived by computationally optimizing amino-acid sequences [19, 20]. Their cDNA sequences were codon-optimized for expression in mammalian cells, ignoring gene-packaging sequences required for virus growth. Here, to maintain viral sequence fidelity, we reverse-translated the COBRA protein sequences into full-length cDNA gene segments using related influenza strains and codon preferences. The resulting wild-type (WT) COBRA-HA-Y2 and COBRA-NA-N1-I genes were synthesized and inserted into pHW2000 virus rescue plasmids, which were used to generate LAIVs containing the six internal genes of ca-PR8. A second LAIV was generated containing COBRA-HA-Y2 alongside the NA gene from A/Hawaii/70/2019 (HI19), a WHO-recommended vaccine strain for the 2020–21 North American influenza season. Four additional LAIVs were rescued, using COBRA-HA-Y2 having an HA-stabilizing (HA2-K153E) or HA-destabilizing (HA1-Y17H) mutation [29] paired with either COBRA-NA-N1-I or vaccine-strain HI19-NA. Overall, six LAIVs were generated (Table 1).

Table 1.

Properties of live-attenuated influenza vaccine viruses carrying COBRA antigens (COBRA LAIVs) ¹.

NA	HA (COBRA Y2) 2	Virus inactivation pH 3	HA activation pH 4	Shelf life at 4°C (Days) 5
N1-I 6	Y17H 7	5.65	6.1	69.7
N1-I	WT	5.35	5.6	105.1
N1-I	K153E 8	5.28	5.3	85.8
HI19 9	Y17H	5.67	6.0	42.3
HI19	WT	5.30	5.7	58.7
HI19	K153E	5.26	6.0	46.6

¹COBRA LAIVs contain six internal genes from cold-adapted A/PR/8/34 [17].

²COBRA Y2 was derived from the HA amino-acid sequences of pandemic H1N1 viruses isolated from 2014 to 2016. [20]

³Virus inactivation pH was determined using MDCK cells at 33°C.

⁴HA activation pH was determined by syncytia assay using Vero cells.

⁵To measure the shelf life, aliquots that had been incubated at 4°C were titrated in MDCK cells at 33°C.

⁶COBRA N1-I was derived from the NA amino-acid sequences of N1 subtype avian-, swine-, and human-influenza viruses isolated between 1900 and 2015. [19]

⁷HA1-Y17H.

⁸HA2-K153E.

⁹A/Hawaii/70/2019 (pH1N1) virus was the WHO-recommended cell-based vaccine candidate virus for the North American 20202021 influenza season.

3.2. NA and HA properties of COBRA LAIVs

The NA activities of LAIVs containing COBRA-NA-N1-I were higher than those of the corresponding LAIVs containing HI19-NA (Figure 1A). The HA mutations had minimal impact on NA activity.

Figure 1.

NA and HA properties of COBRA LAIVs. (A) NA activities of COBRA LAIVs at various pH values. Data was analyzed by 2-way ANOVA. *p<0.05, ***p<0.001, ****p<0.0001. (B and C) Virus inactivation pH titrations of COBRA LAIVs. The NA genes were either COBRA-NA-N1-I (B) or HI19-NA (C). In panels A-C, viruses containing COBRA-NA-N1-I have square symbols and those containing HI19-NA have circle symbols. Wild-type (WT) COBRA-HA-Y2 is black or gray, Y17H is burnt orange or orange, and K153E is blue or light blue. (D and E) HA activation pH measured by syncytia assay using Vero cells. HA activation pH values were taken as the highest pH of syncytia formation (center panel for each). Assays were conducted either in the absence of oseltamivir (D) or in the presence of 5 μM oseltamivir (E).

The two LAIVs containing COBRA-HA-Y2 had virus inactivation pH values of 5.30–5.35 (Figure 1B-C; Table 1) and HA activation pH values of 5.6–5.7 for viruses containing COBRA-NA-N1-I or NA-HI19 (Figure 2D; Table 1). Therefore, for LAIVs containing COBRA-HA-Y2, HA activation pH was higher than virus inactivation pH. HA1-Y17H increased the virus inactivation pH to 5.65–5.67, while HA2-K153E slightly lowered inactivation pH to 5.26–5.28. For LAIVs containing HA1-Y17H, HA activation pH values were about 0.33–0.45 pH units higher than their corresponding virus inactivation pH. When paired with COBRA-NA-N1-I, HA2-K153E had HA activation and virus inactivation pH values of 5.3 and 5.28, respectively (Figure 1B, 1D; Table 1). The LAIV containing HA2-K153E alongside HI19-NA had a higher-than-expected HA activation pH of 6.0 given its virus inactivation pH of 5.26 (Figure 1C-D; Table 1). For this LAIV, HA was highly destabilized for cell-to-cell fusion, yet virions were resistant to mild acid inactivation.

Figure 2.

Growth properties of COBRA LAIVs in vitro. MDCK and Vero cells were infected with COBRA LAIVs at an MOI of 0.001 PFU/cell and cultured at 33°C or 37°C. Virus titers were measured in MDCK cells at 33°C. (A) Growth of LAIVs containing COBRA-NA-N1-I. (B) Growth of LAIVs containing HI19-NA. (C-D) Comparisons of peak titer and HA activation pH in MDCK (C) and Vero (D) cells. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 (2-way ANOVA using the LAIV containing COBRA-HA-Y2 in comparisons).

To assess the impact of NA activity on HA activation pH, syncytia assays were performed in the presence of the NA inhibitor oseltamivir (Figure 1E). For LAIVs containing HI19-NA, which had lower NA activity than COBRA-NA-N1-I, oseltamivir did not significantly alter HA activation pH (Figure 1D-E). In contrast, for LAIVs containing COBRA-NA-N1-I, which has greater NA activity, oseltamivir led to an increase in HA activation pH. Specifically, the HA activation pH values of LAIVs containing COBRA-HA-Y2 and HA2-K153E increased from 5.6 to 5.9 and from 5.3 to 5.8, respectively, after oseltamivir treatment.

3.3. Growth properties of COBRA LAIVs

Growth curves were conducted at 33°C and 37°C using MDCK and Vero cells, both GMP-approved cell lines used for influenza vaccine production. The LAIVs containing COBRA-NA-N1-I replicated to low (<2 log₁₀ TCID₅₀/ml) or undetectable levels at 37°C (Figure 2A), confirming the cold-adapted phenotype of the ca-PR8 backbone. Growth in Vero cells at the permissive temperature (33°C) was both delayed and reduced compared to MDCK cells. In both cell lines at 33°C, virus growth ranked as Y17H > WT > K153E. LAIVs containing HI19-NA were also attenuated at 37°C (Figure 2B). When paired with HI19-NA, HA2-K153E grew better at 33°C (Figure 2B, left panel) than it did when paired with COBRA-NA-N1-I (Figure 2A, left panel) in both MDCK cells (day 1 p<0.0001; day 2 p=0.028; and day 3 p=0.004) and Vero cells (day 2 p=0.019; day 3 p<0.001). This suggested that HA activation pH may be a more significant determinant of COBRA LAIV growth in cell culture than virus inactivation pH. Indeed, HA activation pH and peak virus titer were highly correlated (R²=0.9275) in MDCK cells at 33°C (Figure 2C-D). A similar trend was observed in Vero cells, and even though the correlation was lower (R² = 0.6609) the effect was stronger (i.e., the increase in peak titer was higher for a given increase in HA activation pH). Overall, COBRA LAIV growth was (a) attenuated at 37°C, (b) higher in MDCK than in Vero cells, and (c) positively correlated with HA activation pH.

3.4. Shelf-life at 4°C

COBRA LAIV aliquots were incubated at 4°C for up to 13 weeks and titrated by TCID₅₀ in MDCK cells (Figure 3). The 90% reduction time was defined as the duration required for a 10-fold decrease in virus titer. Viruses containing COBRA-NA-N1-I had longer 90% reduction times compared to those containing HI19-NA; all exceeded 42 days (Table 1; Figure 3).

Figure 3.

Stability of COBRA LAIVs at 4°C. (A) LAIVs containing COBRA-NA-N1-I. (B) LAIVs containing HI19-NA. Aliquots were frozen at −80°C, individually placed in a 4°C water bath at one-week intervals, and at the end of the 13-week study were titrated altogether by TCID50 assay.

3.5. COBRA LAIV vaccination of mice

To assess LAIV replication in mice, groups of five DBA/2J mice were intranasally inoculated with the six COBRA LAIVs or with WT-PR8, a control virus containing the six WT internal genes of PR8, COBRA-HA-Y2, and HI19-NA. Three days post-inoculation (DPI), lungs and nasal turbinates were collected for virus titration by TCID₅₀ (Figure 4A-B). In the WT-PR8 control group, average virus titers in nasal turbinates and lungs were 1.8×10⁴ and 3.4×10⁵ TCID₅₀/ml, respectively. By comparison, average titers in the nasal turbinates and lungs for all COBRA LAIVs were less than 1×10³ TCID₅₀/ml. As expected, the COBRA LAIVs were attenuated in mice. The COBRA LAIVs containing HA1-Y17H did not yield measurable infection in mice, and as pH1N1 viruses with HA1-Y17H are highly attenuated in mice and ferrets [30], the two COBRA LAIVs containing HA1-Y17H were excluded from further animal studies to reduce animal use.

Figure 4.

Immunogenicity and protection of COBRA LAIVs in mice. Groups of mice were immunized with LAIVs containing COBRA-HA-Y2 (WT, Y17H, or K153E) and NA (COBRA-NA-N1-I or HI19-NA), and a control virus with COBRA-HA-Y2, HI19-NA, and six internal genes from A/PR/8/34 (PR8), labeled as WT-PR8. (A) Viral loads in the nasal turbinates 3 DPI after immunization. (B) Viral loads in the lungs 3 DPI after immunization. For pathogenicity and immunogenicity tests, groups of mice were immunized with LAIVs containing COBRA-HA-Y2 (WT or K153E) and NA (COBRA-NA-N1-I or HI19-NA) and monitored for weight change and survival for 14 days. Sera were collected at 28 and 60 DPI for HAI assays. (C) Weight change after immunization. (D) Survival after LAIV immunization. The legends for panels C and D are the same. (E and F) HAI titers against pH1N1 viruses using sera collected at 28 and 60 DPI, respectively. Data was analyzed by 2-way ANOVA. At 61 DPI, immunized mice were challenged with BR18 and monitored for weight change and survival for 14 days or were sacrificed at 64 DPI (3 DPC, day post-challenge) to measure challenge virus replication in lungs. (G) Weight change after BR18 challenge. (H) Survival after BR18 challenge. (I) Viral loads in the lungs after BR18 challenge. CA09: A/California/04/2009; MI15: A/Michigan/45/2015; BR18: A/Brisbane/02/2018; GD19: A/Guangdong-Maonan/SWL1536/2019.

To evaluate immunogenicity and protection against challenge, groups of ten DBA/2J mice were intranasally inoculated with (a) one of four COBRA LAIVs containing either COBRA-HA-Y2 or HA2-K153E, (b) the WT-PR8 control virus, or (c) PBS. Mice were monitored for 14 days for clinical symptoms, peripheral blood sera were collected at 28 and 60 DPI, and mice were challenged with 2×10⁵ TCID₅₀ of BR18 at 61 DPI. Following vaccination, all mice in the WT-PR8 group lost over 20% body weight and succumbed within six days (Figure 4C-D). In contrast, mice vaccinated with COBRA LAIVs had average weight loss of 13.4% or less (Y2+N1-I: 9.3%; K153E+N1-I: 1.9%; Y2+HI19-NA: 13.4%; and K153E+HI19-NA: 8.5%). These groups began regaining weight by day 8 and survived, except for one mouse in the COBRA-HA-Y2/HI19-NA group. The COBRA LAIVs induced protective antibody titers (HAI > 40) against BR18 by 28 DPI (Figure 4E). By 60 DPI, most mice had protective antibody titers against four pandemic H1N1 viruses isolated between 2009 and 2019, indicating a broad serological response to pH1N1 viruses (Figure 4F).

At 61 DPI, vaccinated mice were challenged with 2×10⁵ TCID₅₀ of BR18. Five mice per group were monitored for clinical symptoms over 14 DPC, while the remaining five mice per group were euthanized at 3 DPC to recover, homogenize, and assay lung tissues for viral titers. In the PBS (negative control) group, the average lung titer reached 7.4×10⁵ TCID₅₀/ml, with all animals losing more than 25% of their body weight and succumbing to infection within 6 days (Figure 4G-I). In contrast, all groups vaccinated with a COBRA LAIV had no detectable lung titers, an average weight loss of 8% or less (Y2+N1-I: 2.8%; K153E+N1-I: 5.8%; Y2+HI19-NA: 1%; and K153E+HI19-NA: 8%), and 100% survival. Overall, the four COBRA LAIVs elicited strong serological responses to a broad panel of pH1N1 viruses and provided full protection against challenge in mice.

3.6. COBRA LAIV vaccination of ferrets

Groups of six adult (5 months old) or elderly (6–7 years old) ferrets were inoculated intranasally with 1×10⁶ TCID₅₀ of each of the four COBRA LAIVs containing COBRA-HA-Y2, HA2-K153E, COBRA-NA-N1-I, and HI19-NA. Across all groups, average peak nasal titers post-vaccination exceeded 1×10⁴ TCID₅₀/ml (Figure 5A-D). Replication dynamics were generally similar between adult and elderly ferrets, with the exception that the LAIV with the HA2-K153E / COBRA-NA-N1-I combination cleared approximately two days earlier in adults. COBRA LAIV vaccination did not induce clinical symptoms in any group. Adult ferrets maintained stable body weight throughout the 14-day monitoring period, while elderly ferrets gained 5–10% body weight during this time (Figure 5E-F).

Figure 5.

Primary immunization of COBRA LAIVs in ferrets. Each group contained six adult ferrets (5 months old) or six elderly ferrets (6 to 7 years old). Ferrets were immunized with COBRA LAIVs and monitored for LAIV replication in the nasal cavity (A, B, C, D) and body weight (E, F) for 14 days.

Peripheral blood sera and nasal washes were collected 28 days post-vaccination to evaluate serological and mucosal immune responses. All adult and elderly ferrets developed protective antibody titers (HAI > 40) against BR18, with nearly all showing similar protection against pH1N1 viruses isolated in 2009, 2015, 2018, and 2019 (Figure 6A-B). However, none of the COBRA LAIVs elicited serological responses against seasonal (pre-2009) H1N1 or H3N2 viruses (Figure 6C). Among adult ferrets, the two COBRA LAIVs containing HI19 NA produced higher HAI titers compared to COBRA-HA-Y2 / COBRA-NA-N1-I (Figure 6A). HAI titers were comparable between adult and elderly ferrets (p > 0.05).

Figure 6.

Immunogenicity of COBRA LAIVs in ferrets. (A-B) HAI titers against pH1N1 viruses from 2009 to 2019 using serum collected at 28 DPI from (A) adult and (B) elderly ferrets. (C) HAI titers against seasonal H1N1 and H3N2 viruses using serum collected at 28 DPI from adult ferrets. (D-F) Comparison of serum IgA titers against CA09 using samples collected at 7, 9, and 28 DPI. (G-H) Comparison of mucosal IgA titers against CA09 using samples collected at 7 and 9 DPI. (I-J) Comparisons of serum and mucus IgA titers using samples collected at 7 and 9 DPI from (I) adult and (J) elderly ferrets. (K-J) Comparisons of (K) serum and (L) mucosal IgA titers after BR18 challenge. Data was analyzed by 2-way ANOVA. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Mucosal and serological antibody responses were further assessed using an IgA ELISA assay. Adult ferrets exhibited significantly higher serum IgA titers than elderly ferrets at 7- and 9-days post-vaccination with COBRA LAIVs (Figure 6D-E). By 28 days post-vaccination, serum IgA titers were comparable between adult and elderly ferrets across all groups; however, ferrets vaccinated with HA2-K153E / HI19-NA had higher IgA titers than those vaccinated with COBRA-HA-Y2 / COBRA-NA-N1-I (Figure 6F). These findings indicate that adult ferrets mounted stronger immune responses than elderly ferrets during the early stages of vaccination.

Mucosal IgA titers also followed a similar pattern, with adult ferrets showing higher levels than elderly ferrets at 7- and 9-days post-vaccination (Figure 6G-H). A comparison of serum and mucosal IgA titers revealed a stronger correlation in adult ferrets (R² = 0.55) than in elderly ferrets (R² = 0.23) (Figure 6I-J).

After BR18 challenge (Figure 6K), adult ferrets maintained robust serum IgA responses compared to 28-days post-vaccination (Figure 6F) with average titers of 5,120 or higher. In contrast, serum IgA titers in elderly ferrets decreased to below 2,560 after challenge.

Mucosal IgA titers were generally low in both adult and elderly ferrets, possibly due to antibody-mediated viral neutralization in the nasal cavity. Adult ferrets had mucosal IgA titers ranging between 8 and 32, whereas most elderly ferrets had titers below 4, except for those vaccinated with HA2-K153E / HI19-NA (Figure 6L).

Following challenge with BR18, no adult ferret vaccinated with a COBRA LAIV had detectable challenge virus in the nasal wash, nasal turbinates, or lungs (Figure 7). In contrast, approximately half of the elderly ferrets across all four vaccine groups showed detectable levels of challenge virus in nasal wash samples collected at 2 DPC. For tissue samples collected from elderly ferrets at 3 DPC, those in the COBRA-HA-Y2 / COBRA-NA-N1-I group had no detectable virus in the nasal turbinates (Figure 7D). However, all six ferrets in this group exhibited low levels of virus in the lungs, with titers below 1 × 10³ TCID₅₀/ml (Figure 7F). Notably, elderly ferrets vaccinated with HA2-K153E / HI19-NA demonstrated greater protection in the lungs, with 5 out of 6 animals showing no detectable virus (Figure 7F).

Figure 7.

Protection of ferrets against challenge. At 31 and 33 DPI, immunized ferrets were challenged with BR18 virus. (A and B) Viral loads of nasal wash samples at 2 days post-challenge (DPC). (C and D) Viral loads of nasal turbinates collected at 3 DPC. (E and F) Viral load of lungs recovered at 3 DPC. Samples were titrated in MDCK cells at 33°C. (A, C, E) Samples collected from adult ferrets. (B, D, F) Samples collected from elderly ferrets. * p<0.05 (compared to the LAIV containing COBRA-HA-Y2 and COBRA-NA-N1-I using one-way ANOVA).

4. Discussion

This study examined how HA stability and NA composition influenced the production, stability, and immunogenicity of pH1N1 influenza virus LAIVs incorporating COBRA antigens. COBRA LAIV growth was greater in MDCK cells compared to Vero cells. Decreased HA stability (reflected by higher HA activation pH) correlated with higher LAIV growth in both cell types. As anticipated, COBRA LAIV growth was temperature-sensitive in cell culture and attenuated in mice. Immunization with COBRA LAIVs elicited broadly reactive antibody responses against pH1N1 viruses in mice and ferrets, providing robust protection against pH1N1 challenge.

One COBRA LAIV, containing the HA2-K153E mutation and the NA from the vaccine strain HI19, had an unusual phenotype of a high HA activation pH (6.0) and a low virus inactivation pH (5.3). This LAIV replicated efficiently in MDCK cells and better protection from challenge virus in the lungs of elderly ferrets compared to other COBRA LAIVs. Overall, COBRA LAIVs were found to be highly immunogenic in animal models, and the results suggested they may be further enhanced by HA modifications and NA substitutions that result in a high HA activation pH but low virus inactivation pH.

4.1. COBRA HA and NA sequences can be incorporated into LAIVs that retain growth and stability

The cDNA sequences of COBRA antigens, including COBRA-HA-Y2 and COBRA-NA-N1-I, were originally designed based on consensus amino-acid sequences and optimized for codon usage and expression to facilitate the production of recombinant proteins or VLPs [19, 20]. However, these sequences were not suitable for virus transcription, translation, or vRNA packaging. To conform COBRA HA and NA amino-acid sequences for incorporation into the LAIV platform, we developed a four-step process. First, we identified a closely related HA or NA amino acid sequence to serve as a reference for the COBRA antigen. Second, we introduced in silico mutations into the cDNA sequence of the reference strain, guided by sequence abundance in the GISAID database. Third, untranslated regions (UTRs) were added to replicate the natural gene segment architecture. Finally, we constructed cDNA reverse-genetics plasmids and successfully rescued COBRA LAIV viruses. This approach enabled the generation of COBRA LAIVs that replicated similar to WT pH1N1 influenza viruses in MDCK cells and remained stable for up to six weeks at 4°C. These findings demonstrate the feasibility of applying COBRA antigen sequence optimization to the development of LAIVs.

4.2. COBRA antigens expressed in various vaccine platforms are immunogenic and protective in animal models.

COBRA antigens have been developed through multiple rounds of consensus alignments using varying sequence selection parameters. These antigens have been shown across various vaccine platforms (i.e., recombinant HA, IIVs, and mRNA vaccines) to elicit broader HAI and neutralizing antibody responses compared to seasonal influenza control vaccines [10, 11, 13]. For instance, a combination vaccine containing COBRA-HA-X6 and -P1 antigens, derived from comprehensive sequence alignments of H1N1 viruses across different time periods, induced antibody responses against viruses from 1934, 1957, 1976, and 1991–2013 [31]. However, the limited protection provided by COBRA-HA-X6 and -P1 against pH1N1 viruses isolated after 2012 prompted the development of COBRA-HA-Y2, which was designed using sequences from pH1N1 isolates collected between 2014 and 2016 [20].

In prior studies, COBRA-HA-Y2 was incorporated into recombinant HA protein and VLP vaccine platforms [20]. Vaccination in mice elicited strong antibody responses against pH1N1 viruses from 2009–2019 and provided protection against challenge with CA09 and BR18 viruses. When formulated in an mRNA-LNP vaccine platform or an IIV platform, COBRA-HA-Y2 similarly induced robust immune responses and conferred protection in mice [11, 32]. COBRA-NA-N1-I, derived from multiple N1 subtypes (including avian N1.1, human N1.2, and swine N1.3 clade), has been studied as a recombinant protein vaccine. It elicited cross-reactive antibodies targeting these clades and provided protection against viral challenge in both mice and ferrets [19, 33]. However, NA-based vaccines did not prevent virus transmission between ferrets. When testing different vaccination routes, intranasal inoculation with COBRA recombinant viruses (P1, X3, or X6) showed better IgA and IgG antibody responses than intraperitoneal or intramuscular injection [34]. It is noteworthy that intranasal prime immunization of mice with inactivated COBRA P1, X3, or X6 recombinant viruses elicited a higher peripheral IgA and IgG antibody response compared to prime immunization delivered through the intraperitoneal or intramuscular route [34]. This observation further supports intranasal antigen delivery as a favorable route of administration for eliciting strong immune responses.

In the present study, COBRA-HA-Y2 and COBRA-NA-N1-I were introduced into a LAIV platform, which was then used to immunize mice and ferrets. Consistent with previous findings using other vaccine platforms, COBRA antigens in the LAIV platform induced broadly reactive antibody responses against pH1N1 viruses from 2009–2019 and conferred protection against BR18 challenge. Collectively, these studies demonstrate that COBRA antigens are compatible with diverse vaccine platforms, including recombinant protein, IIV, mRNA, and LAIV vaccines. Future studies could explore the effects of priming and boosting COBRA antigens using different vaccine platforms to potentially enhance vaccine breadth, durability, and protection.

4.3. HA and virus stability modulate LAIV growth, immunogenicity, and protection.

This study showed that the growth of COBRA LAIVs in MDCK and Vero cells increased as a function of HA activation pH. The LAIV containing HA2-K153E and HI19-NA, which had a high HA activation pH and a low virus inactivation pH, grew to higher titers in cultured cells than HA2-K153E paired with COBRA-NA-N1-I. This suggests that, for the viruses studied, LAIV growth correlates more strongly with HA activation pH than with virus inactivation pH.

Previous research supports the observation that HA destabilization, reflected by an elevated activation pH, can enhance influenza virus growth in cell culture within certain limits. For instance, an HA2-N116D mutation acquired during the adaptation of A/PR/8/1934 (H1N1) to Vero cells was found to increase both HA activation pH and the growth of seasonal H1N1, pandemic H1N1, and H3N2 viruses [35]. Similarly, a cell-adapted H3N2 virus with a destabilized HA (inactivation pH 5.83) exhibited greater growth in MDCK cells compared to its parental strain with a more stable HA (inactivation pH 5.57) [36]. Additionally, TN09 containing the destabilizing mutation HA1-Y17H, has been shown to achieve higher replication levels in MDCK, A549, and RAW 264.7 cells compared to the TN09-HA-WT [30].

While destabilizing HA mutations can enhance LAIV growth in cell culture, they may also introduce undesirable traits for vaccines. Because HA conformational changes are irreversible, influenza viruses, including LAIVs, with destabilized HA proteins become more susceptible to inactivation by mildly acidic environments outside the cell [30]. For instance, TN09-HA-Y17H, which had higher growth in cell culture, was highly attenuated in murine lungs and ferret nasal cavities compared to TN09-HA-WT [30, 37]. For A/California/7/2009 (HA2–47E) and A/Brisbane/10/2010 (HA2–47K) (pH1N1), viruses harboring the destabilizing HA2–47E variation had reduced infectivity in ferrets compared to those with the more acid-stable HA2–47K variant [38]. For both seasonal and pandemic H1N1 viruses, an HA activation pH of approximately 5.4–5.5 is generally associated with optimal growth in the lungs of mice, while viruses with either higher or lower HA activation pH are attenuated [39].

Suboptimal HA stability that reduces LAIV infectivity may also lead to reduced immunogenicity. For example, in LAIVs containing H5N1 HA and NA, and lacking the NS1 gene, a virus with WT HA (inactivation pH 5.8) had lower infectivity and elicited weaker serological and mucosal antibody responses in mice compared to an analogous LAIV carrying a stabilizing HA2-K58I mutation (inactivation pH 5.4) [40]. HA destabilization in the CA09 LAIV, combined with its low affinity for α2,6-sialic acid receptors, likely contributed to its reduced vaccine efficacy (VE) in humans [41].

Here, the LAIV containing HA2-K153E and HI19-NA had an unusual phenotype with a relatively high HA activation pH (6.0) and a relatively low virus inactivation pH (5.3). Interestingly, this LAIV provided better protection from challenge virus in the lungs of elderly ferrets. The underlying mechanism for this enhancement remains unclear. Its low virus inactivation pH may help the LAIV resist extracellular inactivation in the mildly acidic upper respiratory tract of ferrets [24, 37, 42, 43]. Alternatively, the higher HA activation pH might facilitate viral entry in early endosomes [30, 44].

4.4. Impact of NA composition and NA activity

The ca-PR8-based LAIVs containing the COBRA-NA-N1-I protein had higher NA enzymatic activity compared to those containing the HI19-NA protein. The shelf lives of the LAIVs at 4°C were longer when COBRA-HA-Y2 was paired with COBRA-NA-N1-I (~105 days) compared to HI19-NA (~59 days). Alternatively, pairing COBRA-HA-Y2 with HI19-NA, as opposed to COBRA-NA-N1-I, resulted in modest increases in serum HAI and IgA levels in ferrets, though mucosal IgA levels were unaffected. When paired with COBRA-HA-Y2 or Y17H, the NA composition (COBRA-NA-N1-I vs. HI19-NA) did not significantly alter HA activation pH or virus inactivation pH. Thus, the level of NA enzymatic activity did not appear to intrinsically alter HA stability in the context of PR8-based LAIVs containing COBRA-HA-Y2 (pH1N1). When paired with HA2-K153E, a lower enzymatic activity by HI19 NA compared to NA N1-I was associated with an increase in HA activation pH and concomitant increases in LAIV growth in MDCK cells and immunogenicity in ferrets as described above in section 4.3.

4.5. Immune responses in adult versus elderly ferrets

Immunosenescence reduces VE and increases the susceptibility of elderly individuals to infections, cancer, and autoimmune diseases [45, 46]. Numerous human studies have shown a decline in humoral responses with age [47, 48], a trend we also observed in ferrets in the present study. On days 7 and 9 after vaccination, serological and mucosal IgA antibody titers in adult ferrets were significantly higher than those in elderly ferrets. However, by day 28, IgA titers in elderly ferrets were comparable to those in adult ferrets, and HAI titers were also similar between the two groups. These findings suggest that elderly ferrets require more time to develop a robust immune response compared to adults.

Although HAI and IgA titers were similar between adult and elderly ferrets prior to BR18 challenge, the level of protection conferred by COBRA LAIVs differed between the two groups. In adult ferrets, COBRA LAIVs provided sterilizing immunity, whereas in elderly ferrets, challenge viruses were detected in nasal wash, nasal turbinate, and lung samples. Upon analyzing mucosal and serological IgA antibody responses post-challenge, we found that IgA titers were generally higher in adult ferrets than in elderly ferrets, except in those immunized with HA2-K153E / HI19-NA, which had IgA levels comparable to their adult counterparts. The importance of local IgA protection has been previously reported [49, 50] and may contribute to the stronger protection observed in adult ferrets in the present study. Elderly ferrets vaccinated with HA2-K153E / HI19-NA had higher IgA titers and better lung protection compared to those immunized with COBRA-HA-Y2/COBRA-NA-N1-I. Additionally, the effects of immunosenescence on cellular immune responses may also contribute enhanced protection in adult versus elderly ferrets [45, 46].

5. Conclusions

In this study, we incorporated COBRA antigens COBRA-HA-Y2 and COBRA-NA-N1-I into PR8-based LAIVs and showed that COBRA LAIVs elicited robust antibody responses and protected mice and ferrets from pH1N1 infection. The pairing of the stabilizing mutation HA2-K153E and HI19-NA resulted in a high HA activation pH and low virus inactivation pH, which enhanced virus growth in cultured cells. The COBRA LAIV with HA2-K153E and HI19-NA also provided better protection from challenge virus in the lungs of elderly ferrets. This study highlights the potential of COBRA antigens for use in LAIVs and underscores the importance of HA stability and NA composition in vaccine design.

Highlights.

• Computationally optimized COBRA antigens can be successfully applied to LAIV vaccines

• COBRA LAIVs induce robust antibody responses and protect against pH1N1 virus

• COBRA HA Y2 and NA N1-I induce broad immune responses across vaccine platforms

• LAIV growth in cells can be enhanced by mutations that increase HA activation pH

• HA- and NA-modified COBRA LAIV is more immunogenic and protective in elderly ferrets