Mucosal immunization with an adenoviral vector vaccine confers superior protection against RSV compared to natural immunity

Respiratory syncytial virus (RSV) infections are the leading cause of severe respiratory illness in early infancy. Although the majority of children and adults mount immune responses against RSV, recurrent infections are frequent throughout life. Humoral and cellular responses contribute to an effective immunity but also their localization at respiratory mucosae is increasingly recognized as an important factor. In the present study, we evaluate a mucosal vaccine based on an adenoviral vector encoding for the RSV fusion protein (Ad-F), and we investigate two genetic adjuvant candidates that encode for Interleukin (IL)-1β and IFN-β promoter stimulator I (IPS-1), respectively. While vaccination with Ad-F alone was immunogenic, the inclusion of Ad-IL-1β increased F-specific mucosal immunoglobulin A (IgA) and tissue-resident memory T cells (TRM). Consequently, immunization with Ad-F led to some control of virus replication upon RSV infection, but Ad-F+Ad-IL-1β was the most effective vaccine strategy in limiting viral load and weight loss. Subsequently, we compared the Ad-F+Ad-IL-1β-induced immunity with that provoked by a primary RSV infection. Systemic F-specific antibody responses were higher in immunized than in previously infected mice. However, the primary infection provoked glycoprotein G-specific antibodies as well eventually leading to similar neutralization titers in both groups. In contrast, mucosal antibody levels were low after infection, whereas mucosal immunization raised robust F-specific responses including IgA. Similarly, vaccination generated F-specific TRM more efficiently compared to a primary RSV infection. Although the primary infection resulted in matrix protein 2 (M2)-specific T cells as well, they did not reach levels of F-specific immunity in the vaccinated group. Moreover, the infection-induced T cell response was less biased towards TRM compared to vaccine-induced immunity. Finally, our vaccine candidate provided superior protection against RSV infection compared to a primary infection as indicated by reduced weight loss, virus replication, and tissue damage. In conclusion, our mucosal vaccine candidate Ad-F+Ad-IL-1β elicits stronger mucosal immune responses and a more effective protection against RSV infection than natural immunity generated by a previous infection. Harnessing mucosal immune responses by next-generation vaccines is therefore a promising option to establish effective RSV immunity and thereby tackle a major cause of infant hospitalization.


Introduction
RSV is the leading cause of respiratory disease and hospitalization during early infancy. It is estimated that 33.8 million RSV-associated acute lower respiratory tract infections occur annually in children younger than five years leading to 66,000 to 199,000 deaths (1). By the age of two, almost all children were infected at least once with this virus. While the majority of infections result in mild disease, RSV bronchiolitis in early childhood has been associated with delayed respiratory sequelae such as wheezing cough or allergic sensitization in general (2). Although RSV-specific serum antibodies can be found in almost every child or adult, and RSV is relatively conserved, reinfections occur frequently throughout life (3). Especially in frail, elderly adults, such reinfections bear a risk for severe illness. 78% of all RSV-related deaths occur in elderly >65 years old (4). The reasons for this insufficient immunity after natural infections are incompletely understood. A short half-life of infection-induced humoral responses (5) and a generally weak induction of virus-specific antibodies (6) and T cells (7) by an RSV infection have been reported. This might be explained in part by a suppression of innate immune responses by several RSV proteins that in turn inhibits the establishment of adaptive memory responses (8).
The presence or absence of mucosal immune responses induced by previous RSV infections is increasingly recognized as a decisive factor for disease severity in humans (6,(9)(10)(11) and mice (12, 13) as also reported for other respiratory viruses like influenza A (14-17) or severe acute respiratory syndrome coronavirus 2 (18). Important components of adaptive mucosal immune responses are secretory IgA and T RM . Secretory IgA coats virus particles right after entering the respiratory tract and therefore likely provides an immediate virus neutralization. Moreover, its specific interaction with Fca receptors might enable unique effector functions, whereas its polymeric form might increase antibody avidity (19). Mucosal IgA was reported to correlate with protection against RSV reinfections in humans (6,9), but IgA levels seem to be poorly maintained in convalescents (6). Similar to these humoral immune parameters, T cells can persist in the respiratory tract as well. Those T RM can easily be identified by their expression of CD69 and CD103, which is their prototypic phenotype in murine and human lung tissues (14,15,20,21). A human challenge study with RSV followed pulmonary T RM dynamics upon infection. It found a correlation of preexisting RSV-specific CD8 + T RM and reduced clinical symptoms and viral load upon infection. Therefore, this study implies that T RM provide protection against human RSV infections (11), although they are obviously not maintained at protective levels in all RSV-experienced individuals. Mouse models underline that RSV-induced T RM responses provide protective immunity against secondary infections but wane substantially within five months (12, 13). Importantly, high levels of IgA correlate with protection against PCR-confirmed infection in humans but once an infection occurred do not influence the viral load (6). In contrast, airway T RM might function as a second, complementary layer of immunity that cannot prevent infection but reduces virus replication upon infection (11).
Despite the disease and healthcare burden of RSV, prophylactic measures are still restricted to passive immunization with monoclonal antibodies. For a long time, vaccine research was largely affected by the unexpected outcome of an early vaccine trial in the late 1960s. A formalin-inactivated RSV vaccine given to toddlers and infants did not decrease the rate of community-acquired RSV infections but increased the rate of hospitalization from 5% in the control cohort to 80% in the vaccine group (22). Much effort has been spent on investigating the mechanisms behind this vaccine-enhanced RSV disease. Possible mechanisms include an induction of poorly neutralizing antibodies (nAbs) due to a disruption of important epitopes during the formalin inactivation (23-25) and a suboptimal Toll-like receptor stimulation upon immunization (26). Drivers of the vaccine-enhanced disease include the production of IL-4, IL-5, and IL-13 associated with T H 2 responses, and eosinophilia (27, 28) upon infection. Although there is still no market approved vaccine against RSV available, several platforms including protein-based, nucleic acid-based, and viral vector vaccines are currently at different preclinical or clinical stages (29). One advanced candidate is Janssen's adenoviral vector vaccine named Ad26.RSV.preF. It is based on the rare human adenovirus serotype 26 and encodes the RSV F protein in its stabilized prefusion conformation. In a human challenge study, the vaccine injected intramuscularly prevented about 40% of PCR-confirmed RSV infections compared to the placebo group four weeks after immunization (30) and is now tested in a phase III clinical trial (NCT04908683). Another vaccine candidate is Moderna's mRNA-1345. The messenger RNA vaccine that also encodes for the stabilized prefusion F glycoprotein has recently entered phase III clinical trials (NCT05127434) after been granted a fast track designation by the U.S. Food and Drug Administration (FDA). Since both vaccine candidates are administrated intramuscularly, they predominantly provoke systemic immunity. Although repeated systemic immunizations can elicit low to moderate levels of humoral and cellular responses in the respiratory tract (31), a mucosal vaccination is more effective in this regard (32). Particularly the establishment of respiratory T RM is highly dependent on a mucosal vaccine delivery as shown in different contexts (14,15,18). Overall, mucosal vaccines that generate local immunity in addition to systemic responses appear promising given the protective roles of mucosal immune responses in respiratory virus infections.
Here we describe a mucosal vaccination strategy based on an adenoviral serotype 5 vector expressing RSV F supplemented by vector-encoded IPS-1 (Ad-IPS-1) or IL-1b (Ad-IL-1b) as genetic adjuvants. Without an adjuvant, Ad-F resulted in moderate levels of mucosal immunity and reduced viral burden upon RSV infection, whereas the inclusion of Ad-IL-1b significantly increased the vaccine immunogenicity and protective efficacy. In direct comparison to natural immunity induced by a primary RSV infection, the vaccination with Ad-F+Ad-IL-1b raised stronger F-specific immune responses, particularly in the respiratory tract, and led to less virus replication and decreased weight loss upon an RSV infection. These results are encouraging indicating that an effective immunity against RSV is possible by taking advantage of mucosal vaccine strategies.

Adenoviral vaccines
The vaccine vector encoding RSV-F was constructed by insertion of the codon-optimized full-length sequence of the RSV-A2 F protein (GenBank database entry EF566942) as described before (33). The murine mature IL-1b-encoding vector and the Ad-empty control were constructed as described elsewhere (16). For the generation of Ad-IPS-1, the full-length cDNA sequence (NCBI Ref Seq NM_001206385.1) was initially cloned into the pShuttle vector. Replication-deficient adenoviral vectors were produced according to the AdEasy adenoviral vector system (34). Viral particles were purified and concentrated with the Vivapure Adenopack20 (Sartorius). The concentration of total Ad was measured by optical density at 260 nm (OD 260 ) and infectious particles by Reed-and-Muench's tissue-culture infectious dose 50 assay (35). Ratios of total to infectious particles usually were in the range of 200:1.

Luciferase promoter reporter assays
In a 96-well plate, confluent HEK 293T cells were transfected with 0.2 mg of a luciferase reporter plasmid for promoter (binding) activity of either IFN (Interferon) -b, IRF3 (Interferon regulatory factor 3), NFkB (nuclear factor 'kappa-light-chain-enhancer' of activated B-cells), or ISRE (Interferon-sensitive response element), respectively (36). These plasmids are based on the pGL2-TATA-inr backbone (Stratagene) and contain the respective response elements/promoters that control the expression of firefly luciferase from P. pyralis. 24 hours later, the transfected cells were transduced with the replication-defective adenoviral vectors at a multiplicity of infection (MOI) of 10. After 48 hours incubation at 37˚C and 5% CO 2 , the cells were lysed with 100 ml Glo-lysis buffer, 25 ml Bright-Glo Luciferase substrate (both Promega) was added, and the signal (relative light units per second, RLU/s) was measured on a microplate luminometer (VICTOR X5, Perkin Elmer).

Mice and immunizations
Six to eight-week-old BALB/cJRj mice were purchased from Janvier (Le Genest-Saint-Isle, France) and were kept in individually ventilated cages in accordance with German law and institutional guidelines under specific pathogen-free conditions, with constant temperature (20-24°C) and humidity (45-65%) on a 12 h/12 hlight/dark cycle. Experimental and control animals were co-housed. The research staff was trained in animal care and handling in accordance to the FELASA and GV-SOLAS guidelines. Immunizations were conducted under general anesthesia (100 mg/kg ketamine and 15 mg/kg xylazine). A dose of 2x10 8 particles (as defined by OD 260 ) Ad-RSV-F and 1x10 9 particles encoding for the adjuvants were given intranasally in a total volume of 50µl PBS (phosphate-buffered saline). Blood samples were taken on stated time points from the retroorbital sinus under light anesthesia with inhaled isoflurane. Bronchoalveolar lavages (BAL) were conducted after sacrificing the animals by cannulation of the trachea following two rinses with 1 ml PBS. Lungs and spleens were removed for T cell analyses.

FACS-based antibody analysis
Target cells were either generated by stable lentiviral transduction of human embryonic kidney (HEK) 293A cells (for RSV F) or by transient transfection of HEK 293T cells (for RSV G; 11.5x10 6 cells in a 175cm2 flask, 35 µg pCG-RSV, 17.5 µg blue-fluorescent protein, 78.75 µg polyethylenimine). For the assay, cells were harvested with a plastic scraper and 2x10 5 cells were plated in a 96-well plate and were incubated with sera and BAL samples diluted in FACS-PBS (PBS with 0.5% BSA and 1 mM sodium azide) for 20 min at 4°C. F-specific antibodies were then detected with polyclonal anti-mouse Ig-FITC (1:300, 4°C, 20 min incubation; BD Biosciences) or anti-mouse IgA-FITC (1:300, clone C10-3, BD Biosciences). Date were acquired on an AttuneNxT (ThermoFisher) and the mean fluorescence intensity (MFI) of the respective fluorescence signals were analyzed with FlowJoTM software (Treestar Inc.).

RSV microneutralization assay
As described before (33), twofold dilution series of complement-inactivated (56°C, 30 min) sera and BAL samples were incubated with 50 PFU rgRSV, a recombinant RSV-A2 strain expressing a green fluorescent protein, in a total volume of 100 µl Dulbecco's modified Eagle's medium (DMEM; no FCS, 2 mM L-Glutamine, 100 units/ml penicillin/streptomycin) for 1h at 37°C. Subsequently, 50 µl DMEM (1% FCS, 2 mM L-Glutamine, 100 units/ml penicillin/streptomycin) containing 10 4 Hep2 cells were added per well and incubated for 72h at 37°C. Plaques were counted with an ELISPOT reader and analyzed using CTL Immunospot software (Cellular Technology limited BioSpot). 75% plaque reduction neutralization titers (PRNT75) were defined as the highest reciprocal serum/BAL dilution that inhibited more than 75% of plaques observed in infected control wells without serum/BAL treatment on the same plate (mean of four control wells per plate used).

T cell assays
Where indicated, mice were injected with 3 µg anti-CD45-BV510 (clone 30-F11, Biolegend) intravenously and were euthanized 3 min later with an overdose of inhaled isoflurane to define circulatory and T RM . Lungs were cut in pieces and were digested in 2 ml R10 medium (RPMI 1640 supplemented with 10% FCS, 2 mM L-Glutamine, 10 mM HEPES, 50 mM bmercaptoethanol and 1% penicillin/streptomycin) including 160 units DNase I and 500 units Collagenase D for 45 min at 37°C. Subsequently the digested lungs and spleens were homogenized through 70 µm cell strainers and treated with ammonium-chloride-potassium buffer to lyse erythrocytes.

RSV infections
Experimental RSV infections were conducted under general anesthesia (100 mg/kg ketamine and 15 mg/kg xylazine). The inoculum containing 5x10 5 PFU (primary RSV infection) or 1x10 6 PFU (RSV protection experiments) RSV-A2 (ATCC VR-1540) was administered intranasaly in 50µl PBS. Animals were monitored daily for body weight and clinical score. In the challenge infections, mice were sacrificed five days postinfection. Lungs were harvested and homogenized in M-tubes with a GentleMACS dissociator (Miltenyi Biotec). Viral RNA was detected in cell-free supernatants of the lung homogenates via quantitative reverse transcriptase real-time PCR (qRT-PCR) using the GoTaq ® RT-qPCR 1-Step Kit (Promega) with the following primers for a sequence in the nucleoprotein gene: for agatcaacttctgtcatccagcaa; rev gcacatcataattaggagtatcaat). The lower limit of quantification was at 666 copies/ml. As correlate of tissue destruction, the total amount of protein in cell-free BAL samples was measured by a bicinchoninic acid assay (Pierce).

Statistical analysis
Results are shown as mean ± SEM or as median ± interquartile range except it is described differently. Statistical analyses were performed with Prism 9.0 (GraphPad Software, Inc.). A p < 0.05 was considered statistically significant.

Inclusion of Ad-IL-1b in the vaccine increases systemic and mucosal antibody responses
In the present study, we evaluate the immunogenicity and protective efficacy of an Ad5-based mucosal vaccine encoding for RSV F (Ad-F). Our previous study proved that the inclusion of inflammatory stimuli as genetic adjuvants can substantially increase the vaccine efficacy in mucosal immunizations (16). Therefore, we evaluated the use of vector-encoded IL-1b and IPS-1 (also known as mitochondrial antiviral-signaling protein, MAVS) as adjuvants in this part. Before using them in vivo, the genetic adjuvants were tested in vitro by luciferase reporter assays. Transduction of HEK 293T cells with Ad-IL-1b provoked a strong induction of the NFkB pathway, while Ad-IPS-1 additionally initiated a type I IFN response as seen by activation of the ISRE, IRF3, and IFN-b promotor reporters (Supplementary Figure 1).
Next, BALB/c mice were intranasally immunized with Ad-F plus the adjuvants or an empty vector construct (Ad-empty). Vaccine-induced antibody responses against RSV F were analyzed 35 days and 49 days later in serum and BAL samples, respectively. In a cytometry-based antibody detection method based on HEK 293A cells expressing the F protein in its natural conformation, all immunized groups presented robust F-specific antibody levels in sera ( Figure 1A). The inclusion of Ad-IL-1b but not Ad-IPS-1 further increased antibody levels. Moreover, this adjuvant effect was not restricted to the systemic compartment but was also detectable in BAL samples that showed increased levels of F-specific IgA ( Figure 1B). Correspondingly, all immunization strategies resulted in significant in vitro neutralization activity with the highest titers in the Ad-F+Ad-IL-1b group underlining its adjuvant properties ( Figure 1C).
Inclusion of Ad-IL-1b in the vaccine enhances mucosal T cell immunity Seven weeks after the immunization, T cell responses were investigated in lymphocytes isolated from lung tissue. Using pentamer staining for F 85-93 -specific CD8 + T cells, it became evident that all immunizations established F-specific CD8 + T cells (Figure 2A). To further characterize those antigen-specific CD8 + T cells, we performed a phenotypic analysis to differentiate between different circulating memory T cell subsets (effector memory T cells, T EM : KLRG1 + CD127 + ; central memory T cells, T CM : KLRG1 -CD69 -CD103 -CD127 + ) and T RM (KLRG1 -CD127 +/-CD69 + CD103 + ; gating scheme in Supplementary Figure 2). As expected for a mucosal immunization route, most vaccine-induced, F 85-93 -specific CD8 + T cells in the lung were of a T RM phenotype ( Figure 2B). Substantially less T CM and no T EM or T EFF were passaging through the lung at the time point of analysis. The vaccination with Ad-F+Ad-IL-1b increased lung CD8 + T RM significantly compared to a nonadjuvanted vaccination with Ad-F+Ad-empty. Similarly, also the functional analyses of F-specific CD8 + ( Figure 2C) and CD4 + T cell responses ( Figure 2D) underlined the T cell stimulatory effect of Ad-IL-1b leading to substantial increases of cytokineproducing (IFNg + , IL-2 + , TNFa + ) or degranulating (CD107a + ) T cells upon peptide restimulation (Supplementary Figure 3). Polyfunctional CD8 + T cell populations positive for all assessed effector functions were elevated as well by the coadministration of Ad-IL-1b. In conclusion, a mucosal vaccination with Ad-F results in robust humoral and cellular immunity. The inclusion of Ad-IL-1b but not Ad-IPS-1 further increased systemic and mucosal immune responses.

Ad-IL-1b-adjuvanted mucosal immunization increases protection against RSV infection
To assess the efficacy of the different vaccination strategies, mice were infected seven weeks after the immunization with RSV-A2. The unvaccinated group as well as the groups that received no adjuvant or Ad-IPS-1 showed a peak weight loss on day two post- Vaccine-induced T cell responses in the lung. BALB/cJRj mice were immunized as described above and lung lymphocytes were analyzed 49 days later. Counts of F 85-93 pentamer + CD8 + T cells per lung (A) and counts of specific memory T cell subsets (B) are shown; effector T cells, T EFF , KLRG1 + CD127 -; effector memory T cells, T EM : KLRG1 + CD127 + ; central memory T cells, T CM : KLRG1 -CD69 -CD103 -CD127 + ; T RM : KLRG1 -CD127 +/-CD69 + CD103 + ; gating scheme in Supplementary Figure 2. (C, D) Lung lymphocytes were restimulated with MHC-class I and class II peptides derived from the F protein and the functionality of CD8 + T cells (C) and CD4 + T cells (D) was detected via extracellular staining for CD107a and intracellular cytokine staining (poly; for CD8 + : CD107a + IFNg + IL-2 + TNFa + ; for CD4 + : IFNg + IL-2 + TNFa + ; gating strategy in Supplementary Figure 3). Bars represent group means overlaid with individual data points. n=4-5. Data were analyzed by one-way ANOVA followed by Tukey´s post Test. *p<0.05, ** p<0.005, *** p<0.0005, **** p<0.0001. infection followed by a regain of weight until day five ( Figure 3A). In contrast, the group that received Ad-IL-1b presented only a minor weight loss on day one and then started to recover reaching its initial weight about two days earlier than the other vaccinated groups. Viral load in the lung was investigated five days postinfection by qRT-PCR detection of viral RNA. Naïve animals showed the most pronounced virus replication, whereas all vaccine regimens reduced viral load at least by a factor of 1000 on day 5 ( Figure 3B). In line with the enhanced immune responses provoked by the inclusion of Ad-IL-1b in the vaccination, animals of this group had significantly lower viral loads compared to the ones of the other immunization groups.

Mucosal immunization generates superior mucosal antibody responses compared to natural infection
Next, we compared the immune responses raised by a natural RSV infection with the ones induced by our mucosal vaccine candidate Ad-F+Ad-IL-1b. To this end, mice were immunized as described above or were infected with RSV-A2. Five weeks (serum) and seven weeks (BALs) later humoral immune responses were assessed. In order to capture all RSV-specific antibody moieties, we performed ELISA analyses with coated, heat-inactivated RSV-A2. Humoral responses in sera showed a bias towards IgG1 after mucosal immunization compared to a more IgG2a-prone response after RSV infection ( Figure 4A). Importantly, BAL samples contained about 10-fold higher levels of RSV-specific antibodies of all isotypes ( Figure 4B left) and secretory IgA ( Figure 4B right) after adenoviral immunization compared to the RSV-immune group. Serum neutralization titers were comparable in the RSV-immune and Ad-F+Ad-IL-1b groups ( Figure 4C), while neutralization in BAL samples was overall low. However, there was a trend that all mucosally vaccinated mice mounted a low-level virus neutralization (6/6 mice), whereas in the previously infected group only one mouse showed detectable a neutralization titer (1/6 mice; Figure 4C).
While the adenoviral vaccination encodes only for one RSV protein, the primary infection probably generates humoral immunity against other RSV proteins as well. Therefore, the flow cytometric assay was used to determine the specific antibody levels against the two major surface proteins F and G. It was observed that F-specific serum antibodies were much more abundant in mucosally vaccinated mice compared to RSV-immune mice ( Figure 4D). This was also evident in BAL samples, where the infection resulted in low to undetectable levels of F-specific antibodies, whereas the immunization provoked a response in all animals. However, the primary infection additionally induced Gspecific humoral responses in sera, which were of course absent after Ad-F-Ad-IL-1b immunization ( Figure 4E). Again, the RSVimmune group presented low to undetectable antibody levels in BAL samples. In conclusion, systemic antibody responses are quantitatively similar in both immune groups resulting in a comparable neutralization, but there are clear qualitative differences. Moreover, a mucosal immunization with our candidate vaccine is superior in inducing mucosal antibodies than a primary RSV infection.

Mucosal vaccination but not natural infection establishes robust F-specific T RM responses
As already shown above, our mucosal vaccine candidate elicits robust F 85-93 -specific CD8 + T cell responses in the lung ( Figure 5A; Figure 2). In contrast to that, a primary RSV infection results in a 20-fold weaker F-specific response ( Figure 5A). By using a combination of our phenotyping analysis and intravascular staining to distinguish circulating and resident memory T-cells (37), the vast majority of vaccine-induced, F-specific T cells was identified as T RM as already pointed out earlier ( Figure 5B; Figure 2). A primary infection was not very efficient in inducing T cells of that specificity and, interestingly, those few F-specific T cells show predominantly a T CM phenotype. CD8 + T cell responses against another dominant epitope, M2 82-90 , were about fourfold stronger compared to F 85-93 -specific responses in the RSVimmune group, but do not reach the absolute level of Fspecific T cell immunity observed after immunization ( Figure 5C). Again, there was no clear dominance of T RM over circulatory phenotypes as seen after mucosal Ad-F+Ad-IL-1b vaccination ( Figure 5D) yielding roughly equal amounts of T CM and CD69 + CD103 + T RM sampled in the lung. In order to compare T RM phenotypes in a broader fashion independent of the strict CD69 + CD103 + phenotype, we analyzed M2 82-90specific (RSV-immune) and F 8 5-9 3 -specific (mucosal immunization) CD8 + T cells in the lung that were negative for the intravascular staining. These T cells are per definition tissueresident but also include populations aside from the most stringent CD69 + CD103 + lung T RM phenotype. By this approach, we could again prove that CD69 + CD103 + T RM are indeed the dominant lung-resident phenotype after mucosal vaccination and that most of these cells express CD11a + as well ( Figure 5E). However, such phenotypic dominance was absent after RSV infection. The primary infection mainly resulted in CD11a + T RM subsets with variable expression of CD69 + and/or CD103 + . Of note, the frequencies of all subsets that express CD103 + are higher in the Ad-F+Ad-IL-1b group compared to the RSV-immune animals. To analyze CD8 + T cell functionality, lung lymphocytes were restimulated ex vivo with F-and M2-derived peptides. These analyses confirmed a strong F-specific response after mucosal vaccination, while a primary RSV infection did not raise reliable amounts of F-reactive T cells ( Figure 6A). As expected from the multimer staining, moderate levels of M2-reactive CD8 + T cells were detectable in RSVimmune mice ( Figure 6B). Similar trends were observed in the systemic T cell response as well (Supplementary Figure 4). Thus, a natural infection and our mucosal vaccination strategy led to very different T cell responses regarding amplitude, specificity, and memory phenotype. Substantial numbers of mucosal, Fspecific T cells are not part of a naturally induced immunity in mice but can be established by a mucosal vaccination.
Mucosal immunization establishes more efficient protection than natural immunity The protective effects of natural immunity and the mucosal vaccination were then tested by an experimental infection with RSV-A2. Vaccinated mice showed a transient weight loss of~3% between one and three days post-infection but started to regain weight already on day one reaching the initial weight on day four ( Figure 7A). In contrast, RSV-immune and naïve mice started to lose weight on day two with a constant decrease until the end of the experiment on day five. There was even a trend towards more pronounced weight loss in the RSV-immune group compared to naïve mice, but this trend lacks statistical power (p=0.10 naïve vs. RSV at day three post-infection; one-way ANOVA followed by Tukey´s post test). Nevertheless, virus RNA levels in the lung were significantly reduced in presence of natural immunity compared to naïve animals ( Figure 7B). In line with the lower weight loss, the adenoviral immunization resulted in the lowest viral RNA levels at day 5 after infection. As a surrogate of tissue damage, we measured the amount of total protein in BAL samples. Both immune groups showed a significant reduction of the protein concentration compared to naïve mice, but the RSV group presented a relatively high tissue damage compared to the vaccine group ( Figure 7C). Therefore, although both a primary infection and a mucosal immunization reduce virus replication upon an RSV infection, the mucosal immunization provides a superior protection against weight loss and tissue damage.

Discussion
RSV vaccine research was hampered by the severe complications observed during early clinical trials with a formalin-inactivated vaccine candidate in the 1960s. However, vaccine technologies have diversified and improved significantly in the last decades yielding several promising vaccine candidates that are currently FDA fast-tracked and/or entering phase III clinical trials (38). Importantly, none of these vaccines aims at the induction of mucosal immunity, although preclinical and clinical data suggest an important contribution of local immune responses to the control of RSV infections.
In the present study, we assessed the immunogenicity and efficacy of a mucosally administrated, adenoviral vaccine that encodes the wildtype RSV-A2 F protein. In the first part, we investigated whether the inclusion of genetic adjuvants increases A B FIGURE 6 Mucosal T cell functionality induced by mucosal vaccination and a primary RSV infection. Lung lymphocytes were restimulated with MHC-class I peptides derived from the F (A) and the M2 protein (B) and the functionality of CD8 + T cells was detected via extracellular staining for CD107a and intracellular cytokine staining (poly; CD107a + IFNg + IL-2 + TNFa + ; gating strategy in Supplementary Figure 3). Bars represent group means overlaid with individual data points. n=4-6. Data were analyzed by one-way ANOVA followed by Tukey´s post Test. *p<0.05, ** p<0.005, *** p<0.0005, **** p<0.0001.
vaccine-induced immunity. Two inflammatory proteins were chosen as vector-encoded adjuvants: IL-1b and IPS-1. IL-1b is a hallmark cytokine of tissue inflammation (39, 40) and particularly connected to T cell stimulating properties (41-43). Therefore, this cytokine was already used in several studies as adjuvant for example as mucosal adjuvant in flu vaccinations (16,44) or as T cell stimulus in cancer vaccines (45)(46)(47). IPS-1 was chosen as adjuvant candidate since the retinoic acid inducible gene I (RIG-I)-IPS-1 pathway plays an important role in innate immune sensing of viral infections including RSV (48)(49)(50) and the subsequent activation of adaptive immune responses (51,52) as exploited by the co-application of RIG-I agonists in vaccination settings (53,54). Moreover, IPS-1 activation has been reported as essential for mounting T cell responses upon RSV reinfection (55) and specifically for the induction of mucosal RSV immunity (56). In general, the mucosal administration of Ad-F showed a robust immunogenicity inducing F-specific antibodies and T cells. Mucosal immune responses including BAL IgA and CD8 + T RM were rather low without adjuvantation, but the inclusion of Ad-IL-1b signifi cantly increased those responses. Concomitantly, the protection against an RSV infection was enhanced as well indicated by reduced weight loss and virus replication upon infection. These observations are reminiscent of our previous studies in an influenza A model (16). Ad-IL-1b increased mucosal immunity in these experiments by activation of key steps in the establishment of T RM : mucosal inflammation, attraction of innate immune cells, increased pulmonary infiltration of T RM precursor cells, and finally the upregulation of local factors that support in situ differentiation into CD69 + CD103 + T RM cells. The data generated in the present study underline that the adjuvant activities of IL-1b benefit mucosal vaccinations against different respiratory pathogens. In contrast, Ad-IPS-1 did not enhance vaccine immunogenicity. Although it clearly demonstrated strong activation of NFkB and the type I IFN pathway in vitro, none of the measured immune parameters was increased by the inclusion of Ad-IPS-1 in the mucosal vaccine. This was unexpected, since IPS-1-induced type I IFNs contribute to RSV-specific T RM establishment (56). However, a defective CD8 + T RM response in IPS-1 knockout mice might not directly be translatable into an enhanced response upon in vivo overexpression of IPS-1, and there also exists a contradictory report about the role of IPS-1 in RSV memory T cell formation (49). Moreover, murine type I IFNs consist of IFN-b and 14 IFN-a subtypes that differ in their immunomodulatory effects on CD8 + T cells (57). One cannot exclude that the vector-driven overexpression of IPS-1 results in a type I IFN profile that does not efficiently stimulate T and B cell responses.
After proving that Ad-F+Ad-IL-1b is an effective vaccine candidate that establishes systemic and mucosal immunity, we compared its immunogenicity profile and protective efficacy to natural immunity generated by a primary RSV infection. This is particularly important because natural immunity is obviously not sufficient to promote long-lived and efficient immunity against reinfections. Therefore, vaccine candidates should preferably induce stronger immune responses compared to an RSV infection. Most preclinical vaccine candidates and all vaccine approaches that entered phase III clinical trials aim at the induction of F-specific immunity. One important consideration for this choice is that the F protein is highly conserved and a major target for nAbs. The attachment protein G also raises nAbs, but it shows the highest variability among all RSV proteins suggesting that G-specific immunity is less cross- reactive (58). Our mucosal vaccine candidate Ad-F+Ad-IL-1b and a primary RSV infection both elicit robust RSV-specific antibody responses in serum. However, F-specific responses were substantially stronger after vaccination compared to a primary RSV infection. In contrast, the latter one raised Gspecific antibodies as well probably explaining why both immunizations resulted in comparable virus neutralization by serum samples. Indeed, G-specific virus neutralization is commonly induced by a primary RSV infection in infants (59). Interestingly, the mucosal vaccination provoked robust levels of F-specific antibodies in BAL fluids, while F-and Gspecific humoral responses were rather low or undetectable in BAL samples of previously infected mice. F-specific IgA is an important correlate of protection against RSV in non-human primates (60) and in a human challenge study (6). The latter study reports an impaired induction of IgA + memory B cells after RSV infection, which is in line with our observations. There is emerging evidence that mucosal CD4 + and CD8 + T cells contribute to a rapid and effective clearance of pulmonary pathogens with influenza A being the most investigated, prototypical example (14-16, 61). The Varga lab dissected the role of T cell responses in RSV infections in a series of publications.
In conclusion, they have demonstrated that systemic CD8 + T cells alone mediate severe immunopathology upon infection in mice (62), which is prevented in presence of RSV-specific nAbs (63). The importance and protective capacity of lung T RM against RSV was also underlined by works of this (13) and another group (12). The clinical relevance is highlighted by human challenge studies conducted by the group around Christopher Chiu and Peter Openshaw. In two studies they report on CD4 + and CD8 + T RM dynamics upon human RSV infection and show that CD8 + T RM abundance prior to infection correlates with reduced symptoms and virus replication (10,11). Importantly, our present study proves that a natural RSV infection does not generate substantial T RM responses against the F protein, while inducing moderate levels of M2-specific T RM cells. Additionally, the infection-induced T cell response is not skewed towards mucosal T cell phenotypes but instead established similar levels of T RM and T CM as sampled in the lung. It is tempting to speculate that such a suboptimal generation of mucosal memory T cell responses is one reason for frequent RSV reinfections despite previous encounters with the pathogen. In contrast, our candidate vaccine elicits robust F-specific T RM responses in the lung. Moreover, comparing the phenotype of infection-and immunization-induced T RM , the adenoviral vaccine raised a higher frequency of CD103 + T RM cells. This might be connected to the upregulation of transforming growth factor b (TGF-b) by Ad-IL-1b upon immunization (16) and could promote enhanced cytotoxic T cell functions (64). In accordance with this enhanced immunogenicity, our mucosal vaccination strategy led to reduced weight loss, virus replication in the lung, and tissue damage upon infection. Importantly, although virus replication was also decreased in RSV-immune mice, they experienced a similar or even increased weight loss compared to naïve animals and a relatively high tissue damage. One might speculate that the impaired induction of mucosal immune responses in connection with moderate levels of circulating T cell responses lead to a pronounced immunopathology but this needs further investigation. Nevertheless, these data underline that natural RSV immunity is suboptimal, especially regarding mucosal, F-specific IgA and T RM . At the same time, we could show that a single, mucosal administration of Ad-F+Ad-IL-1b efficiently establishes these promising correlates of protection leading to an improved control of RSV infections.
Although the present study provides valuable insights into infection-versus mucosal immunization-induced immunity, these insights need confirmation in other models and settings as well. The BALB/c mouse model is commonly used as preclinical animal model and enabled important insights in RSV-related disease and immunopathology, but there are limitations (65). One is the semipermissiveness of mice for RSV and therefore a relatively high inoculation dose is needed to achieve virus replication in the respiratory tract (66). The extent of virus replication might influence the resulting immune responses after an RSV infection. Moreover, clinical symptoms are only mild in murine RSV infections despite the high infectious doses. Weight loss upon an RSV infection, a commonly used disease progression marker in murine respiratory infections, is very mild and sometimes varies among different infection studies. Nevertheless, our data hind towards a defective generation of mucosal immunity upon RSV infection. It has to be investigated whether this is a general immune escape feature of RSV or a species-related phenomenon. However, the findings reported here fit well to data from clinical studies investigating RSV-specific immune responses after infection. Since most animal models have limitations, clinical studies with human volunteers remain the most meaningful setting. In particular, smallsized human challenge studies are well suited to investigate correlates of protection and vaccine efficacy, but are extremely difficult to conduct due to ethical and economic reasons.

Data availability statement
The original contributions presented in the study are included in the article/supplementary material. Further inquiries can be directed to the corresponding author/s.

Author contributions
MT and DL designed and conceived the study. CM, JF, PI, MW, JB and DL conducted the experiments. CM, MT and DL analyzed the data and wrote the manuscript. All authors contributed to the article and approved the submitted version.

Funding
The study was supported by funding from the German Research Foundation (DFG, TE1001/4-1, 419013006) and the Doktor-Robert-Pfleger Foundation.