Cutting Edge: High-Dose Live Attenuated Influenza Vaccines Elicit Pulmonary Tissue-Resident Memory CD8⁺ T Cells in the Face of Pre-Existing Humoral Immunity

The recommended annual influenza virus vaccination schedule, coupled with the widespread prevalence of the virus within the community, has generated a population that possesses high levels of Ab immunity to influenza virus (1). Yet despite achieving a cornerstone principle of vaccination (i.e., high levels of Ab immunity), adults are susceptible to reinfection. This is because influenza virus can frequently mutate and escape the strain-specific Ab-mediated immunity elicited by current vaccines. Developing improved vaccines that provide universal protection against circulating and emerging influenza virus strains remains a health issue of utmost global importance. These next generation influenza virus vaccines will have to evoke broad cross-strain protection and do so in vaccinees with diverse levels and specificities of influenza-reactive pre-existing humoral immunity.

Although multiple reports have illustrated how the pre-existing Ab repertoire of individuals can impact the subsequent influenza-specific Ab response, few studies have explored the effect on T cell responses (2, 3). This is important, because CD8⁺ T cells, specifically tissue-resident memory CD8⁺ T cells (Trms) that permanently reside in the respiratory tract (4), are indispensable for cross-protection against different influenza virus strains (5).

The live attenuated influenza vaccine (LAIV) can elicit T cell responses (in addition to Abs) in humans (6–8). Furthermore, because the LAIV is administered intranasally (i.n.), a key requirement of lung CD8⁺ Trm development (9), this formulation has the potential to generate cross-protective local T cell immunity. Despite this, LAIV fails to provide robust cross-strain protection in humans (10). Whether pre-existing humoral immunity to the backbone of an LAIV impacts the vaccine-induced influenza-specific local CD8⁺ T cell response, and therefore potentially accounts for the suboptimal protection conferred by LAIV, has not been resolved (2, 11). In this study, we sought to understand whether pre-existing Ab immunity directed toward the vaccine backbone impacts the LAIV-induced CD8⁺ Trm response and, thus, vaccine efficacy.

C57BL/6 (CD45.2), uMT^−/−, and OT-I.CD45.1 mice were bred in-house under specific pathogen-free conditions at the Peter Doherty Institute of Infection and Immunity, University of Melbourne, Melbourne, VIC, Australia. All experiments were done in accordance with the Institutional Animal Care and Use Committee guidelines of the University of Melbourne. Viruses used include X31/H3N2, PR8/H1N1, cold-adapted LAIV (PR8/H1N1) (12), PR8-OVA (which encodes the OVA_257–264 epitope within the neuraminidase stalk) (13), S-FLU (PR8) [S-eGFP*/N1(PR8).H1(A/PR/8/1934)], and PR8 (CAM). S-FLU (PR8) is a pseudotyped virus based on A/PR/8/34 and modified to possess a defective hemagglutinin (HA) glycoprotein, restricting it to a single round of replication (14). PR8 (CAM) is a A/PR/8/34 strain with the wild-type HA from the Cambridge strain, matching the HA pseudotyped on S-FLU (PR8) (14). For i.p. infections, mice were infected in a volume of 200 µl with 10⁶ PFUs PR8/X31 or 10⁶ Median tissue culture infectious dose (TCID₅₀) PR8 (CAM). For total respiratory tract infection, mice were anesthetized with inhalation isoflurane anesthetic and infected in a volume of 30 µl with 50–100 or 10⁴ PFUs PR8-OVA, 100 or 2.5 × 10⁵ PFUs LAIV (PR8), or 10³–10⁶ TCID₅₀ S-FLU (PR8). For primary infection with PR8-OVA, a sublethal dose of 50–100 PFUs depending on the batch was used. As previously described (12), the low and high dose of LAIV was defined by extrapolating the clinical dose used in an adult (10⁷ PFUs, 75 kg) to a mouse (10³ PFUs, 20 g).

Naive OT-I CD8⁺ T cells were isolated from OT-I TCR transgenic mice as described previously (15). A total of 1 × 10⁶ purified OT-I.CD45.1 CD8⁺ T cells were labeled with 5 μM CFSE (Sigma-Aldrich) for 10 min/37°C before i.v. injection into mice.

Single-cell suspensions were prepared from spleens, lymph nodes (LNs), and lungs as previously described (16). Virus-specific CD8⁺ T cells were identified using in-house–produced tetrameric complexes of H2D^b and the NP₃₆₆ peptide (ASNENMETM). The conjugated mAbs were obtained from BD Pharmingen, BioLegend, or eBioscience and include mouse anti-CD8 (53-6.7), anti-CD3 (17A2), anti–CD45-1 (A20), anti–CD45-2 (104), anti-Va2 (B20.1), anti-CD44 (1M7), anti-CD103 (2E7), anti-CD69 (H1.2F3), anti-CD62L (MEL-14), and anti-CD25 (PC61). Samples were acquired using a Becton Dickinson Fortessa flow cytometer, and data were analyzed using the FlowJo software package (Tree Star, Ashland, OR).

ViroSpot microneutralization assay was performed as previously described (17).

Heat-inactivated serum (56°C/30 min) was serially diluted in serum-free DMEM (Life Technologies) and incubated with 200 TCID₅₀ of the indicated influenza virus for 1 h at 37°C/10% CO₂. The serum/virus mixture was then cocultured with a confluent layer of PBS-washed MDCKs in a 96-well plate for 2 h at 37°C/10% CO₂. The serum/virus mixture was then removed and replaced with serum-free DMEM additionally supplemented with 2 μg/ml L-1-tosylamido-2-phenylethyl chloromethyl ketone-treated trypsin (Sigma-Aldrich) and incubated for 3 d. The supernatant was combined with 1% (v/v) turkey RBCs (1:1), gently mixed, and the HA titer was read after 30 min to determine titers of anti-influenza neutralizing Abs in serum.

Comparison between two study groups was statistically evaluated by unpaired two-tailed t test or Mann–Whitney U test. Comparisons between more than two groups (single factor) were evaluated using one-way ANOVA with Tukey’s multiple comparison. Two-way ANOVA with Sidak’s multiple comparison was used to evaluate more than two groups at different time points or cohorts. In all tests, statistical significance was quantified as *p < 0.5, **p < 0.01, ***p < 0.001, and ****p < 0.0001. Statistical analysis was performed using GraphPad Prism.

We examined whether pre-existing Ab immunity directed toward the vaccine backbone has an impact on the development of a cellular immune response after immunization with an LAIV. To do this, C57BL/6 (B6) mice were i.p. infected with X31/H3N2 or PR8/H1N1 influenza virus strains and were rested for 20 d to generate influenza-specific neutralizing circulating Abs (Supplemental Fig. 1). Mice were then vaccinated with a cold-adapted LAIV (1e2 PFUs) on a PR8/H1N1 backbone [LAIV (PR8)] i.n., or alternatively were not vaccinated to discern the baseline levels of circulating CD8⁺ T cell responses induced by primary i.p. infections (Fig. 1A). By using distinct serotypes of influenza virus for the primary infections, the magnitude of the vaccine-induced CD8⁺ T cell response could be measured in the presence (PR8 primed) or absence (X31 primed) of neutralizing Abs (Supplemental Fig. 1). At 20 d after vaccination, the endogenous influenza nucleoprotein (NP)-specific CD8⁺ T cell response in the spleen and lung was measured by flow cytometry using H-2D^b tetramer loaded with NP₃₆₆ peptide. D^bNP₃₆₆-specific CD8⁺ T cells were increased by ∼5-fold in hosts that were LAIV vaccinated and X31 primed relative to X31 primary infection controls. In contrast, no increase in D^bNP₃₆₆-specific CD8⁺ T cells was detected in mice previously exposed to PR8 influenza virus postvaccination with LAIV (Fig. 1B–D).

Assessment of the lung influenza-specific Trm response in these cohorts revealed that LAIV immunization of naive animals elicited D^bNP₃₆₆-specific lung CD8⁺ Trms, whereas i.p. X31 or i.p. PR8 infections variably elicited modest Trm populations (Fig. 1E, 1F). While LAIV vaccination of X31-primed mice resulted in 10% of bulk pulmonary D^bNP₃₆₆-specific CD8⁺ T cells developing into CD69⁺CD103⁺ Trms (Fig. 1E, 1F), pulmonary D^bNP₃₆₆-specific CD8⁺ Trms failed to develop in PR8-primed LAIV-vaccinated mice (Fig. 1E, 1F). Because LAIV (PR8) vaccination induced Trm responses in naive and X31-primed mice, but not PR8-primed mice, this would suggest that pre-existing Abs against the vaccine backbone prevented the development of vaccine-induced lung CD8⁺ Trms.

B6 or B cell–deficient uMT^−/− mice were left untreated or infected with PR8 i.p. to further confirm that neutralizing Abs established from prior influenza virus infections affect the development of CD8⁺ T cell immunity evoked by LAIV vaccination. Mice were rested for 20 d to establish circulating Abs and were then seeded with CFSE-labeled transgenic OT-I CD8⁺ T cells specific for the SIINFEKL epitope 1 day before PR8-OVA (50 PFUs) i.n. infection (Fig. 2A). This approach allowed us to track the development of expanded OT-I CD8⁺ T cells induced by PR8-OVA in the presence of PR8-reactive Abs and compare responses with mice that had no previous influenza encounter or no capacity to produce Abs. At day 4 postinfection (p.i.) with PR8-OVA, prior exposure to PR8 influenza virus in B6 mice impaired the division of OT-I CD8⁺ T cells in the mediastinal lymph node (mLN) by 60-fold relative to the expansion observed in the naive B6 cohort (Fig. 2B, 2C). In turn, this resulted in significant reductions in the number of splenic and pulmonary bulk memory OT-I CD8⁺ T cells (Supplemental Fig. 2A, 2B) and CD69⁺CD103⁺ OT-I lung Trms (Supplemental Fig. 2C–E). No impairment was observed in the size of the OT-I CD8⁺ T cell response in PR8-OVA–infected naive or PR8 pre-exposed uMT^−/− mice, although we detected increased numbers of OT-I CD8⁺ T cells in PR8-primed uMT^−/− mice at day 4 p.i. (Fig. 2B, 2C, Supplemental Fig. 2). These results signify that pre-existing Ab immunity to influenza virus impairs the development of a de novo CD8⁺ T cell response against the same influenza strain with a novel T cell epitope.

Computational modeling has shown that boosting the vaccine dose can counteract pre-existing Ab immune responses inhibiting the generation of strong vaccine-induced Ab immunity (18). We wondered whether this strategy could also overcome the impairment of CD8⁺ T cell priming we observed in PR8-OVA–infected animals with pre-existing immunity to the virus. To address this, B6 mice were infected with PR8 i.p. or were untreated, rested for 20 d to establish circulating Abs, and then seeded with CFSE-labeled OT-I CD8⁺ T cells 1 d before either a high-dose (1e4 PFUs) or low-dose (50 PFUs) i.n. PR8-OVA infection (Fig. 2D). As expected, low-dose PR8-OVA infection of PR8-immune mice failed to induce division of OT-I CD8⁺ T cells in the mLN at day 4 p.i., with numbers indistinguishable from background division in PR8-immune mice not infected with PR8-OVA (Fig. 2D). However, high-dose PR8-OVA challenge in PR8-primed hosts resulted in OT-I T cell division on par with PR8-OVA (50 PFUs)-infected mice that had no prior influenza history (Fig. 2D). The capacity of the high-dose PR8-OVA infection to bypass pre-existing Ab responses was dependent on the replication competency of the virus, because OT-I CD8⁺ T cells failed to divide in PR8-primed mice that were infected with a matched high dose of UV-inactivated PR8-OVA (Fig 2D). Interestingly, although high-dose PR8-OVA infection could overcome pre-existing Ab responses to generate de novo CD8⁺ T cell responses, these T cells had an altered phenotype and failed to upregulate CD25 (Supplemental Fig. 3). Because naive OT-I CD8⁺ T cells were adoptively transferred into pre-exposed recipients, this may suggest that the environment caused by prior exposure to influenza rather than an intrinsic T cell property accounts for the differential CD25 phenotype. This was also evident in PR8-immune uMT^−/− mice, which were rechallenged with a low-dose PR8-OVA infection (Supplemental Fig. 3), suggesting that differences in CD25 are independent of dose and influenza-specific Ab responses. Collectively, our data suggest that priming of CD8⁺ T cell responses to a new T cell vaccine epitope is abrogated in immune hosts that possess pre-existing Ab immunity against the vaccine backbone, but this can be overcome by immunizing with a sufficiently high vaccine dose.

Next, we explored whether increasing the dose of LAIV could improve the CD8⁺ T cell memory response in vaccinees with pre-exiting immunity to the vaccine backbone. To do this, we repeated the aforementioned LAIV study with a higher dose of LAIV (2.5e5 PFUs) and compared the T cell response with our existing dataset (refer to (Fig. 1; (Fig. 3A). In brief, B6 mice were i.p. infected with X31 or PR8 to generate influenza-specific circulating Abs (Supplemental Fig. 1) and then 20 d later i.n. immunized with low-dose (1e2 PFUs) or high-dose (2.5e5 PFUs) LAIV (PR8) (Fig. 3A). As expected, 20 d after high-dose LAIV immunization, X31-primed mice exhibited a clear increase in D^bNP₃₆₆-specific CD8⁺ T cells in both the spleen and the lung relative to X31 i.p. alone controls (Fig. 3B, 3C). Similarly, high-dose LAIV immunization in PR8-primed mice resulted in a 19-fold and 34-fold expansion in splenic and pulmonary D^bNP₃₆₆-specific CD8⁺ T cells, respectively, relative to PR8-primed alone mice, which was not apparent in PR8-primed mice vaccinated with a low-dose LAIV (Fig. 3B, 3C). Importantly, the incapacity to evoke D^bNP₃₆₆-specific lung CD8⁺ Trms in low-dose LAIV PR8-immune vaccinees could be overcome in high-dose LAIV recipients, with ∼20% of D^bNP₃₆₆-specific CD8⁺ T cells in the lung coexpressing CD103 and CD69, with frequencies and numbers on par with those generated after LAIV immunization of X31-immune mice (Fig. 3D). Moreover, using an alternative LAIV platform based on single-cycle technology (virus restricted to a single round of replication) called S-FLU (14, 19), we showed that high-dose vaccination once again generated pulmonary CD8⁺ Trm responses in vaccinees with pre-existing immunity to the vaccine backbone (Supplemental Fig. 4). Together, our results demonstrate that high-dose vaccination can overcome the inhibition in lung CD8⁺ Trm development in hosts with pre-existing Abs to the vaccine backbone.

Naive, nonvaccinated, PR8/H1N1-primed mice (established from i.p. PR8 infection) or PR8/H1N1-primed mice that were vaccinated with low- or high-dose LAIV (PR8/H1N1) 28 d earlier were challenged i.n. with a heterologous X31/H3N2 influenza virus to test whether high-dose LAIV vaccination in preimmune hosts could confer protection against heterosubtypic viral challenge (Fig. 4A). At day 3 postchallenge, naive mice or PR8-primed mice that were nonimmunized with LAIV or immunized with low-dose LAIV succumbed to ∼10% weight loss (Fig. 4B). Conversely, high-dose LAIV vaccination in PR8-primed hosts exhibited stable maintenance of weight (Fig. 4B), and this was associated with an ∼16-fold decrease in lung viral titers compared with all cohorts (Fig. 4C). We confirmed that protection was dependent on local CD8⁺ Trms as daily treatment several days before and after challenge in the same experimental challenge model with the sphingosine 1-phosphate receptor-1 agonist FTY720 to inhibit circulating T cell infiltration into the lung showed no difference in viral titers (Fig. 4D). Collectively, we demonstrate that immunizing vaccinees with pre-existing immunity to the vaccine backbone with a high-dose LAIV generates lung CD8⁺ Trm responses that confer heterosubtypic protection.

The effect of prior exposure to influenza on LAIV efficacy has been an ongoing and unresolved topic of discussion (2, 20). A recent mouse modeling study by Roy et al. (11) showed that previous inactivated influenza vaccine (IIV) impaired the capacity of LAIV to protect against heterosubtypic influenza challenge that was otherwise observed in non-pre-exposed mice, speculating that cross-protective CD8⁺ T cell responses were suppressed by neutralizing Abs established by prior IIV vaccination. In this article, we demonstrate that pre-existing Ab immunity to the vaccine backbone can impair the capacity for LAIV to induce circulating CD8⁺ T cell and lung CD8⁺ Trm responses, resulting in defective protection against heterologous challenge. While we demonstrate that an LAIV regimen that uses a sufficiently high dose can overcome the observed inhibition in local CD8⁺ T cell immunity and offers protection against viral rechallenge, it will be essential to determine the safety profile of high-dose LAIV immunization.

Our findings emphasize how the immunological landscape of individuals can impact LAIV or infection-induced T cell responses. In support of this, individuals who underwent the annual IIV vaccination regimen (akin to our mouse i.p. infection model that leaves behind an influenza-reactive humoral immune response) failed to induce strong blood CD8⁺ T cell responses elicited by natural infection compared with nonvaccinated controls (21). Furthermore, circulating HA-specific CD4⁺ T cell responses postvaccination were less expanded in individuals who were vaccinated in the previous year compared with vaccine-naive individuals or individuals who received IIV >1 y ago (22). Although these reports and our work focus on pre-existing Ab immunity, it will be important to also address whether pre-existing lung CD8⁺ Trms induced by natural infection contribute to the inhibition of vaccine-induced T cell responses. Collectively, these studies highlight the need to better understand how prior immunity generated by previous IIV vaccination (or infection) impacts the cellular response evoked by new T cell–based vaccine candidates and the need to design solutions to overcome any negative ramifications.

In this article, we show that low-dose LAIV fails to evoke cross-protective lung CD8⁺ Trms in hosts with neutralizing Abs against the vaccine backbone. This may offer an explanation as to why LAIV immunization in adults who possess a relatively mature immunological profile (from repeat infections and IIV) are less protected against influenza compared with children who receive LAIV (10). Importantly, our work suggests the efficacy of current and next generation LAIVs in the adult population could be improved through high-dose vaccination to evoke cross-protective local resident T cell responses that may otherwise be inhibited in individuals with high levels of pre-existing vaccine backbone–specific Ab immunity.

The authors have no financial conflicts of interest.