## Introduction

CD8 tissue-resident memory T cells (TRM) in the lungs represent a subset of memory T cells essential for optimal control of respiratory virus infections.1,2 Since TRM were identified a decade ago, they have been found in various tissues including gut, skin, lung, reproductive tract, liver, and brain.3,4,5,6,7 TRM are cells that reside in non-lymphoid tissues and function as a first line of defense against secondary infections. In addition to their unique anatomic location, TRM have transcriptional profiles distinct from both central and effector memory T cells (TCM and TEM, respectively).8,9,10 In peripheral tissues, CD8 TRM cells are primarily identified by their expression of the tissue retention markers CD69 and CD103, combined with the exclusion of cells that stain with an intravascular label.11 However, not all TRM are created equal. In the skin, TRM remain in the tissue long after infection and antigen clearance.12 Conversely, lung TRM are lost a few months after an acute infection, leading to the loss of protection from secondary challenge.1,13 The reason for this rapid decay is still under investigation, but a harsh environment in the lungs and airways is likely to contribute. Recently published evidence shows that the airway environment causes transcriptional and epigenetic changes in the memory T cells resulting from amino acid starvation and leading to increased apoptosis.14 Previous reports have suggested residual antigen is important for continued development of virus-specific CD8 T cells after viral clearance following influenza infection.15 However, the potential role of residual antigen for the differentiation and maintenance of lung TRM has not been investigated.

In the present study, we highlight the connection between persistent antigen expression following AdNP vaccination and maintenance of lung CD8 TRM. We find, in contrast to the rapid loss of lung TRM following influenza infection, mice primed with AdNP maintain a substantially larger population of lung TRM for up to 1 year post vaccination (p.v.). Using Nur77GFP reporter mice, we show lung TRM of AdNP immunized mice continue to interact with cognate antigen for at least 3 months p.v. The expanded lung TRM population in AdNP immunized mice is maintained in the lungs and airways by both in situ proliferation and continual replenishment of the lung TRM pool from the circulation. Finally, we show that i.n. administration of AdNP is critical for the increased longevity of lung and airway TRM. These results highlight the potential benefits of Ad vector vaccination, and underscore the importance of prolonged antigen expression in the lung for extended T-cell mediated protection against respiratory infections.

## Results

### Antigen persists in the lungs and airways after AdNP vaccination

Several studies have investigated the effect of persistent antigen on the CD8 T cell population following vaccination with adenoviral vectors.29,30 It is well established that low-level persistent antigen can stimulate T cells without causing exhaustion. This phenomenon, known as memory inflation has been studied in both murine Cytomegalovirus and Ad infections.29,31,32,33 Moreover, previous work has shown that antigen encounter in the lung is required for establishment of CD8 TRM.34,35,36 Importantly, genomic material from adenoviruses and adeno-vectors persists in the tissue after immunization, in some cases up to a year post injection.30,37 In the present study, we confirmed persistence of the NP antigen by immunofluorescence microscopy in lungs 110 days after AdNP vaccination (Fig. S4). However, the effect of this low-level antigen persistence on lung TRM has never been studied. To investigate this, we utilized a Nur77GFP reporter mouse to visualize antigen-dependent CD8 T cell stimulation in vivo. The Nur77-GFP signal is transient and quickly lost after withdrawal of antigen stimulation, making the readout of GFP in Nur77GFP mice a useful tool to investigate if lung TRM have recently encountered their cognate antigen. Nur77GFP mice were primed with x31 or AdNP and 45 or 90 days later BAL, lungs, and spleens were isolated and Nur77 expression analyzed. As illustrated, there is a significantly higher proportion of Nur77+CD69+ cells among the DbNP366+ CD8 T cells in the lungs and airways of AdNP mice at day 45 (Fig. 2a, b), compared with x31 mice. The frequency of Nur77+CD69+ decreases in both groups at day 90, however, we still find significantly more Nur77+CD69+ T cells in the lung TRM of AdNP vaccinated mice There is no difference in the frequency of Nur77+ cells between x31 and AdNP mice when expression was analyzed based on TRM expression of CD69 and CD103 (data not shown).

Chronic PD-1 expression is believed to reflect ongoing antigen stimulation in TRM populations1,38; therefore, we examined PD-1 expression on CD8 lung TRM as an additional indicator of persistent antigen stimulation (Fig. 2c, d). The majority of the DbNP366+ T cells in the BAL and lung of AdNP primed mice expressed PD-1 at day 60 and the proportion of CD69+PD-1+ cells was significantly higher in lungs and BAL of AdNP vaccinated mice compared with cells from x31 primed mice. PD-1 expression in AdNP mice was maintained until the last time point studied (day 275, data not shown). Since prolonged PD-1 expression can indicate a state of T-cell exhaustion,39 it is important to stress that CD8 TRM in AdNP vaccinated mice still show protection from PR8 challenge up to 255 days p.v. (Fig. S2). To directly address possible concerns regarding the functional relevance of PD-1+ expression on the airway T cells, the cellular response and its protective capacities after AdNP immunization were evaluated in PD-1−/− mice (Fig. S3). We found no significant difference in the numbers of DbNP366+ TRM, nor was there a significant difference in viral titers in the lungs 5 days after PR8 challenge compared with wild-type mice. We also examined TIM-3 expression, as TIM-3 has also been implicated in an exhausted T-cell phenotype.40 DbNP366+ cells in AdNP immunized mice had little to no expression of TIM-3 (data not shown).

Next, we hypothesized that the persistent antigen in AdNP immunized mice results in continual in situ proliferation and thereby maintains the TRM population. To evaluate proliferation, we administered EdU in the drinking water for a period of 7 days in both AdNP and x31 primed mice. Both at day 45 and day 90, mice immunized with AdNP have a significantly higher proportion of EdU+ cells in the BAL and lungs compared with x31 infected mice (Fig. 2e, f). Incredibly, we found EdU incorporation in DbNP366+ T cells of AdNP immunized mice as far out as 270 days p.v., with significantly greater incorporation in the BAL and lungs (Fig. 2e, f). Due to the low number of DbNP366+ T cells in x31 infected mice after day 90, we did not investigate EdU incorporation in these mice at later time points. Moreover, we observed that all lung TRM subsets based on expression of CD69 and CD103 expression had a significantly higher proportion of EdU+ NP-specific cells in AdNP immunized mice than observed in x31 infected mice (Fig. 2g, h). Taken together with the previous results, these data indicate the persistence of local antigen in the lung is driving activation and continued proliferation of lung TRM following AdNP vaccination.

### AdNP intranasal inoculation is indispensable for sustaining the lung TRM pool

In order to further support our hypothesis of persistent local antigen expression as critical for maintenance of the lung TRM population, we compared the NP-specific T-cell response in mice immunized both s.c. and i.n. with AdNP to mice vaccinated s.c. with AdNP and infected i.n. with x31 or mock-infected i.n. with PBS (Fig. 4a). By day 14 (Fig. 4b), mice immunized i.n. with AdNP had significantly more cells in the airway and lungs compared with mice inoculated i.n. with x31. At day 40 post priming, there were still significantly more DbNP366+ T cells in in the airways and lungs of mice immunized with AdNP i.n. Importantly, regardless of s.c. administration of AdNP, the fold reduction of DbNP366+ T cells in the lungs and airways between days 14 and 40 was greater in mice that received i.n. x31 or PBS (Fig. 4c). This emphasizes the necessity of local persistent antigen for the maintenance of a long-lived airway and lung TRM population following AdNP vaccination.

## Discussion

Influenza virus infections represent a global health burden and currently available vaccines are inefficient. First, they are directed at a moving target, the main surface molecule, hemagglutinin, (HA), which is subject to substantial genetic variation as a result of both genetic drift as well as genetic shift. This creates a need for repeated vaccinations to sustain at least some protection. Second, the vaccine-induced immune response to HA consists largely of circulating IgG, whereas local immunity in the airways is limited. One way to remedy both of these deficiencies is to replace the current vaccine approach with one that induces a potent local T-cell response in the respiratory tract. Unlike antibodies, T cells also target the internal viral antigens, which are not subject to the same kind of selection as the surface molecules and therefore much more conserved between strains of influenza. However, a primary challenge in generating effective T cell based vaccines against respiratory pathogens, is the rapid decline of the lung TRM population. However, our recently published results have indicated that combined local and systemic immunization in mice using AdNP induces a long-lasting protective CD8 population.18 Until now the underlying reasons for this sustained response has not been investigated. In this study, we provide evidence indicating that TRM in AdNP vaccinated mice continue to encounter antigen in the lungs up to at least 3 months p.v., undergo in situ proliferation, and are continuously recruited into the lung from the circulating memory T cell pool.

First, our data provided evidence for long-standing proliferation of NP-specific cells in the lungs of Ad immunized mice in contrast to flu infected mice. Second, we observed ongoing recruitment of circulating memory cells as a consequence of prolonged antigen expression. Previous research has suggested that the TRM population of influenza primed mice is maintained through a dynamic process consisting of a high apoptotic rate concurrent with replenishment from circulating memory cells, and that this recruitment is independent of local antigen.41 Based on our parabiosis experiments, we found little to no evidence of recruitment of circulating cells into the lung TRM population in influenza primed mice. Moreover, a circulating memory T cell population generated by s.c. AdNP immunization of i.n. influenza primed mice, did not suffice to maintain the lung TRM population. Together, our data show the importance of prolonged local antigen expression for continuous recruitment of circulating T cells into the lung TRM pool.

These data further emphasize that TRM across tissues cannot be treated as one homogenous population, and that the requirements for establishment and maintenance are as diverse as the anatomic locations in which they reside. For example, antigen is not required for the establishment of skin TRM and persistent antigen has not been shown to be necessary for their long-term maintenance.12 Thus, it is highly plausible that long-term maintenance of a stable TRM population in any given tissue is dictated by how effectively local proliferation and ongoing recruitment from other memory cell subsets balances the local apoptotic rate.

Most importantly, all of these parameters are influenced by the microenvironment in the tissues of relevance. It should be acknowledged that other components of the immune system, such as APCs and even NP-specific antibodies, may also play a role in the maintenance of the CD8 T cell population.42 We have previously shown that CD8 T cells are the primary mechanism of protection after AdNP vaccination, but to what extent NP-specific antibodies are affected by the persiting antigen, and their contribution to protective immunity, remains to be investigated.18 In the lungs, the TRM are subject to a relatively harsh environment, causing a high rate of apoptosis.14 While we cannot significantly change the environment of the lung, we have shown it is possible to expand and maintain the TRM population by prolonged antigen stimulation. As stated previously, this strategy will open new doors for development of vaccines against flu and other viral respiratory diseases. In this context, it should be kept in mind that the dual vaccination approach applied for Ad immunized mice allows for antigen persistence in both the lungs and the periphery and both of these sources are likely contributing factors in the sustained response. It has been clearly documented that repeated antigen exposures gradually improve the quality of the primed cells, local cells become less prone to apoptosis, and circulating precursors develop a higher propensity to home to the lungs and differentiate into TRM.43,44 The prolonged presence of antigen in the periphery is likely to have the same effect, in addition to increasing the number of activated circulating precursors. Consistent with this possibility, we see limited contraction following the initial T-cell burst in s.c. Ad immunized mice, and phenotypic analysis have revealed evidence of ongoing activation e.g. the prolonged-expression of CD43, on remaining antigen-specific CD8 T cells.45,46 The degree to which these factors, local antigen versus prolonged circulation of relevant precursors, contribute to the long-term immunity observed is not clear, but our previous results clearly show that combined s.c. and i.n. vaccination is superior to i.n. vaccination alone.18 Notably, due to the delicate nature of the lung, any vaccination strategy that results in persistent antigen must be carefully assessed for the induction of localized persistent inflammation and immunopathology that could be detrimental to the host. In conclusion, the results presented in this report show us that we can achieve a stable T-cell response in the lung with AdNP vaccination. Importantly, we also show that this long-lived T cell response is due to the presence of persistent local antigen. Not only is antigen required for TRM formation as previously documented,34 but unlike the situation in most other organ sites, persistent antigen is essential for the continual replenishment and long-term maintenance of TRM population in the lungs.

## Methods

### Experimental models

Six- to eight-week-old C57BL/6 mice from Taconic Biosciences were used in this study. All mice were rested upon arrival for at least 1 week. PepBoy/J, Nur77GFP (Nur77-GFPCre B6-820), and PD-1−/− (B6.Cg-Pdcd1tm1.1Shr/J) mice were purchase from Jackson Laboratory. All experimental procedures were approved by the national animal ethics committee (The Animal Experiments Inspectorate or IACUC) of the University of Copenhagen, Emory University, and Kindai University and were conducted in accordance with national guidelines; the mice were housed in an AAALAC accredited facility in accordance with good animal practice as defined by FELASA.

### Virus and vaccines

All mice receiving intranasal (i.n.) inoculation were first anaesthetized by intraperitoneal (i.p.) injection with avertin (2,2,2 tribromoethanol in 2-methyl-2-butanol, 250 mg/kg). Influenza infection with A/Hong Kong/X31 (x31) was used at a dosage of 30,000 EID50 in 30 ul HBSS and administrated i.n. after avertin anesthesia. For Influenza challenge, A/Puerto Rico/8/34 (PR8) was used at a dose of 3LD50 in 30 ul i.n. The production of the replication-deficient adenovirus type 5 expressing influenza PR8 nucleoprotein (AdNP) used in this study has previously been described.18 Mice were immunized with 2 × 107 plaque-forming units (PFU) in 30 μl of PBS i.n. and 30 ul s.c. in the right footpad after anesthesia with avertin.

### Preparation of single-cell suspensions

To isolate resident lymphocytes in lung, mice were intravenously injected with 1.5ug anti-CD3e [145-2C11] fluorophore-conjugated antibody in 200 ul PBS in the tail vein.11,28 Five minutes post injection, mice were anesthetized using avertin and exsanguated. This was followed by harvest of BAL and other tissues.47 After isolation lungs were digested with 5 g/L Collagenase D (Roche) and 2 × 106 Units/L DNAse (Sigma) for 30 min 37 C. Samples were enriched by centrifugation in a 40%/80% Percoll gradient to isolate lymphocytes. Spleen and MLN were mechanically dissociated and passed through a 70 µm nylon filter prior to staining.

### Antibodies for flow cytometry

Cells were first blocked for unspecific binding with αCD16/32 followed by staining with NP366-374/Db tetramer conjugated to allophycocyanin (APC) or Brilliant Violet 421. Tetramer labeled cells was incubated with fluorophore-conjugated antibodies CD8α (clone 53-6.7), CD8β (clone H35 17.2) CD44 (clone IM7), CD45.1 (clone 30-F11), CD45.2 (clone 104) CD69 (clone H1.2F3), CD103 (clone 2E7), PD-1 and live/dead stain Zombie NIR. EdU staining was performed using Click It Plus Alexa Flour 647 Assay kit (Invitrogen) according to manufacturer’s instructions. Samples were analyzed on a Fortessa LSR II (BD Biosciences). Data analysis was conducted using FlowJo v10 software (TreeStar). Gates for CD69, CD103, and PD-1 were set using fluorescence minus one samples. All antibodies were purchased from Biolegend. Relevant tetramers were kindly provided by Søren Buus, Department of Immunology and Microbiology and the NIH tetramer core facility.

### Parabiosis

For parabiosis surgery, mice were anesthetized, and a clipper was used to remove flank hair. This was followed by a longitudinal skin incision from the knee to the elbow on a single lateral side along with a 1-cm lateral peritoneal incision. Suturing each reciprocal peritoneal opening joined the two mice together. To hold the mice in the upright position, two mattress stiches were made on the lateral edges of the skin section. Also, the dorsal and ventral sides of the skin section of each mouse were further joined with wound clips. Equilibrium was confirmed 10 days after surgery with a blood sample.

### MDCK plaque assay

Lungs were homogenized in 1%BSA in PBS 9X the lung weight to obtain a 10% w/v suspension. The mixture was homogenized using sterilized sand, mortar, and pestle. This was followed by centrifugation at 600 × g, 15 min, 4 °C. The supernatant was transferred to a new tube and kept on ice until use. 4.5 × 104 MDCK cells were seeded in 100 ul medium in 96-well plates and the following day, lung supernatant was added in tenfold dilutions in media containing DMEM 1965 medium with 2 mM L-glutamin, 200 IU/ml penicillin, 50 μg/ml streptomycin, 0.2% BSA, 1% sodium pyruvate and 5 units/ml TPCK Trypsin for 2 h. Virus was then removed and samples were incubated for 48 h, 37 °C, 5% CO2, with a 1:1 mixture of medium containing 2× minimum essential medium (MEM) eagle supplemented with 0.4% BSA, 10% NaHCO3, 2% Streptomycin, 2% penicillin and 5 units/ml TPCK trypsin and 1.8% methyl cellulose. After incubation overlay was removed and cells were fixed with 4% formaldehyde in PBS for 30 min at room temperature (RT) and permeabilized using warm 0.5% Triton-X in Hanks balanced salt solution for 10 min at RT. After permeabilization cells were incubated with primary α-influenza nucleocapsid A mAb (Nordic Biosite) diluted 1:1500 in 10% FBS in PBS at 37 °C, 5% CO2. Following the primary antibody, cells were incubated with secondary goat α-mouse HRP conjugated mAb (Dako) diluted 1:500 in 10% FBS in PBS at 37 C. After the secondary antibody, substrate was added containing 3 mg/ml 3-amino-9-ethylcarbazole and 0.07% H2O2 and 5 mM citrate phosphate buffer pH5 and incubated at RT for 30 min. After these PFU were counted and calculated per g lung according to the following formula:

$${\mathrm{Average}}\,\,\# \,\,{\mathrm{of}}\,{\mathrm{plaque}}/{\mathrm{well}} \,\times\, {\mathrm{Dilution}}\,{\mathrm{factor}} \,\times\, 20 = {\mathrm{PFU}}/{\mathrm{g}}\,{\mathrm{lung}}$$

### Immunofluorescence microscopy

For OCT imaging, mice were infected with the specified agent and harvested at indicated days. The mice were euthanized by 2,2,2-tribromoethanol overdose and the ribcage was dissected off. An incision was made in the trachea and the airways were inflated with 0.75 mL OCT via an 18-gauge iv cathedar. The trachea was tied using 5-0 silk suture on a reverse cutting needle and the heart, lungs, and thymus were removed en bloc and flash frozen in OCT. Sections were cut at 7 um and transferred to slides for staining. Slides were fixed in 75% acetone/25% ethanol solution for 10 min and blocked with 10% donkey serum, 10% mouse serum, 10% rat serum, 1 ug/mL anti CD16/32 clone 2.4G2, and 5% FCS in PBS for 30 min on ice. Slides were stained with primary antibodies for 30 min on ice and secondary reagents for 15 min on ice. Coverslips were placed with ProFade Gold mounting media and imaged using a Zeiss AxioObserver microscope using Zen 2 software. Antibodies used include anti-EpCAM-A647 (clone G8.8, Biolegend), anti-Collagen-IV goat pAb (part # AB769, EMD Millipore), Donkey anti-goat-A405 (part # 705-475-147, Jackson Immunoresearch), anti-influenza-nucleoprotein-FITC (clone 431, abcam), anti-FITC-A488 (part # A-11090, Invitrogen), and anti-CD11c-A594 (clone N418, Biolegend).

### Statistical analysis

Statistical analysis was performed using GraphPad Prism 7. ANOVA was used for statistical testing for all experiments and where pair-wise comparison was made Mann–Whitney rank test was used. *p < 0.05, **p < 0.01, ***p < 0.001.