B. subtilis strain DSM 32444 was used for this study. This strain is recommended as safe for human consumption (USA FDA GRAS-notification GRN 000905). Spores were prepared in batch culture (200 mL) and suspensions in dH₂O inactivated by autoclaving (121°C, 20 min., 15 psi) and are referred to henceforth as DSM 32444^K. Validation of spore inactivation was made by serial dilution of heat-treated spore suspensions and plating for viability on agar growth medium and with no resulting bacterial growth meeting the required standard. Aliquots (1 mL) were used for animal studies with each lot containing 5 x 10¹⁰ inactivated spore particles (determined by microscopic counting using a Neubauer counting chamber).

H1N1 (A/PR/8 strain) viral stocks were kindly provided by Dr John W McCauley (and Dr Michael Bennett) from the Francis Crick Institute. The SARS-CoV-2 strains used in this study, Beta variant (Bristol/Liverpool isolate) and Omicron variant BA.1, were grown from the stocks of Prof. Julian Hiscox (University of Liverpool). Virus stocks were propagated using Vero E6 cells and viral titres determined by plaque assays. Work with SARS-CoV-2 was performed at containment level 3, following risk assessments and standard operating procedures approved by the University of Liverpool Biohazards Sub-Committee and by the UK Health and Safety Executive. The respiratory syncytial virus was a kind gift from Prof. Jürgen Schwarze (University of Edinburgh), and the viral stocks were produced and maintained in our lab.

Wildtype C57BL/6J, BALB/c, and K18-hACE2 transgenic mice were purchased from Charles River. All mice used in this study were 7-10 weeks and age-matched animals were randomly assigned to experimental groups. All mice were bred and maintained in individually ventilated cages at 22 ± 1°C and 65% humidity with a 12hr light-dark cycle. Prior to use, mice were acclimatised for one week with free access to food and water. All experimental protocols were approved and performed in accordance with the regulations of the Home Office Scientific Procedures Act (1986), project licence P86DE83DA and the University of Liverpool Ethical and Animal Welfare Committee.

Mice were lightly anaesthetised with O₂/isoflurane and dosed intranasally (i.n.) with 2 or 3 doses of spores (1.5 × 10⁹/30μL per dose, once per week). 7 days (or 27 days for Omicron infection) post-last dose of spores, mice were i.n. challenged with 50μL of virus inoculum. For survival experiment, infected mice were monitored and scored regularly for signs of disease (12). Mice that became lethargic were humanely culled with a rising concentration of CO₂ and counted as death events. Bronchoalveolar lavage (BAL) and lung tissues were harvested at indicated timepoints for the analysis of viral loads and immune responses. BAL fluid was collected as previously described (13), by inserting a catheter into the trachea and instilling 2 × 1mL of cold Hank’s balanced salt solution (HBSS, Sigma Aldrich) with 100μM of EDTA (Fisher Scientific). Cytokine levels in the BAL were measured using LEGENDplex Mouse Anti-Virus Response Panel (13-plex) (BioLegend), following the manufacturer’s instructions.

Lung immune cells were isolated as previously described (14). Briefly, lung tissues were cut into pieces and digested with 300U/ml of type 1 collagenase (Gibco) and 75μg/mL of DNase 1 (Sigma Aldrich). Digested lung pieces were then mechanically dissociated by pushing through a 70μM cell strainer and then subjected to red blood cell lysis to generate single cell suspension for restimulation and flow cytometry staining.

Isolated lung cells from each animal were counted for total live cells based on AO/PI staining. FACS staining was carried out to determine the proportions of immune cell subsets in the whole cell population. Lung cell numbers were calculated based on the total cell count and the proportion of each immune cell subset. BAL cell numbers were determined by acquiring all cell events on the flow cytometry after staining.

For viral quantification, the lung lobes were dissociated by an electronic tissue homogeniser (Ultra Turrax, IKA) for H1N1 and RSV infected lungs. For SARS-CoV-2 infected animals, the left lung lobe was collected in Precellys beads tubes (Bertin Technologies) filled with 1mL of TRIzol reagent and homogenised using a Bead Ruptor 24 (Omni International).

The viral loads in the lung of H1N1 infected animals were determined by 50% tissue culture infectious dose (TCID₅₀) in Madin-Darby canine kidney (MDCK) cells (ATCC, CCL-34) (15).

Immunoplaque assay was used to quantify RSV infected lungs (16). Briefly, serial diluted lung tissue homogenates were inoculated onto HEp-2 cells (ATCC, CCL-23). Viral plaques were stained by monoclonal goat anti-RSV antibody (ABD serotec, UK), followed by a secondary extravidin peroxidase conjugate and an Amino-ethylcarbazole substrate (both obtained from Sigma-Aldrich, UK). The number of viruses were expressed as plaque forming units (PFU) per 1 mL of HEp-2 cell supernatant.

For SARS-CoV-2 infected mice, RNA was extracted from the TRIzol lysed lungs, and viral RNA levels were measured using the Luna® Universal Probe One-Step RT-qPCR Kit (New England Biolabs) following the manufacturer’s instructions. WHO International Standard for SARS-CoV-2 RNA (inactivated England/02/2020 isolate, NIBSC 20/146) were used to produce a standard curve for the quantification of absolute viral copies. qPCR was performed on Bio-Rad CFX96™ using 2019-nCoV N1 primer and probe published by the United States Centres for Disease Control and Prevention (U.S. CDC), the sequences (5’-3’) are as below.

Single cells were incubated with Live/Dead Fixable Aqua Stain (Fisher Scientific) at 1:1000 to determine viable cells. Cell surface staining was performed with fluorochrome-conjugated anti-mouse monoclonal antibodies (mAbs) in the presence of True-Stain Monocyte Blocker (1:20) (BioLegend). mAbs used in this study including CD45-PE-Cy7 (30-F11), CD3-FITC (17A2), CD11b-PE (M1/70), Ly6G-APC-Cy7 (1A8), NKp46-APC (29A1.4), CD8-PE-Cy7 (53-6.7), CD103-PE (QA17A24), CD69-APC (H1.2F3) from BioLegend; SiglecF-BV421 (E50-2440) from BD Biosciences; and CD4-VioBlue (REA604) from Miltenyi Biotec. Stained samples were acquired on BD FACSCanto II and analysed using Flowjo 10.4.

Lungs were dissected and fixed with 10% (v/v) neutral buffered formalin (Sigma Aldrich) and sent to LBIH Biobank, Liverpool for embedding, sectioning and haematoxylin and eosin (H&E) staining. Stained sample slides were scanned and examined using QuPath-0.2.3.

Serum and lung samples taken at day 49 (serum) and 50 (lungs) were taken from animals that had been primed (i.m.) with recombinant SARS-CoV-2 Spike protein (50 μL, 5 μg) of formulated protein (Wuhan variant; SinoBiological (Cat Nos. 40589-V08B1) in each hind quadricep muscle) and then boosted (i.n.) with DSM 32444^K spores (1 X 10⁹ CFU; 10 μL, 5 μL/nare). Antibody (anti-Spike) levels in serum (IgG) and lungs (IgA) were determined by indirect ELISA as described elsewhere (17).

All data were analysed and plotted using GraphPad Prism (version 9.0.1) software. Unpaired two-tailed t-test or Mann-Whitney test was used to compare two datasets. Comparison of multiple datasets was carried out by ordinary two-way ANOVA with Sidak’s multiple comparisons test. Survival curves were analysed by the Log-rank (Mantel-Cox) test. TCID₅₀, PFU and viral RNA copies were log-transformed before analysis. Variance between datasets was tested in all analysis. Data are presented as mean ± SEM for bar graphs and dot plots unless otherwise stated, with sample size (n) specified in figure legends. Absolute p values are indicated in each figure. p < 0.05 is considered as statistically significant.

It has been previously shown that BALB/c mice pre-treated intranasally with two doses of killed spores (on day 1 and 14 respectively) were protected from H5N1 virus infection when challenged 27-days following the last dose of spores (6). Using a different strain of B. subtilis (DSM 32444^K) that had been heat-killed, we first examined whether the same two-dose regimen could protect C57BL/6 mice against lethal infection by another IAV subtype H1N1. Partial protection was observed with 60% of spore-dosed mice surviving the infection versus 0% survival for the control mice pre-treated with Dulbecco’s Phosphate Buffered Saline (DPBS) ( Supplementary Figure 1 ). To optimise the efficacy of spores, we adjusted the dosing regimen by shortening the dosing interval and adding a third dose of spores to the dosing regimen. Male C57BL/6 mice were treated intranasally on day 1 & 7 (two-dose) or day 1, 7 & 14 (three-dose) with 1.5 × 10⁹ spores/dose, or the same volume of DPBS for the infection control group and challenged 7-days following the last spore dose ( Figure 1A ). Encouragingly, both dose groups exhibited 100% survival following viral challenge ( Figure 1B ). Moreover, mice receiving the three-dose regimen showed a more significant improvement in weight loss and disease symptoms ( Figures 1C, D ) together with a significant reduction of viral load in the lungs detected at day 6 post-infection (p.i.) ( Figure 1E ). These results suggest a dose-dependent protection resulting from intranasal dosing of spores against lethal H1N1 infection.

RSV is another common respiratory virus that can cause life-threatening infections in infants and elderly adults (18). It has been reported that intranasal treatment with ‘live’ B. subtilis spores promoted clearance of RSV in the lungs (10). We therefore thought to test the efficacy of the optimised three-dose regimen of killed spores against RSV infection. Since C57BL/6 mice are relatively resistant to RSV, BALB/c mice which are permissive for RSV replication in vivo were used for this pneumonia challenge model. RSV infection alone caused a significant weight loss in BALB/c mice by day 6 and 7 post-infection. In mice pre-treated with spores, no obvious weight loss was observed ( Figure 1F ) accompanied with a significant reduction of viral load in the lungs by day 4 p.i. and a complete clearance of virus by day 6 p.i. ( Figure 1G ), confirming that our spore dosing regimen was also effective against RSV.

Having shown that the optimised spore dosing regimen was highly effective against both H1N1 and RSV infections we next tested its efficacy against SARS-CoV-2 challenge. We used hACE2 transgenic mice expressing the humanised angiotensin-converting enzyme 2 receptor (rendering them susceptible to SARS-CoV-2 infection) (19).

Strikingly, we observed a significant improvement in the survival of mice challenged with the SARS-CoV-2 Beta variant following our three-dose spore dosing regimen, where increased survival was accompanied by a minimum weight loss and ameliorated disease symptoms ( Figures 2A–C ). In accordance with the overall improved performance, the lung viral RNA copies in spore-dosed mice were significantly reduced by day 4 p.i. and a further reduction was detected by day 6 p.i. ( Figure 2D ). This decreased viral load was confirmed by a low viral antigen stain in the pneumocytes of spore-dosed lungs ( Figure 2E ).

To examine whether spore-mediated protection against SARS-CoV-2 could last for more than 7 days post-spore dosing, we challenged the mice 27-days after our three-dose spore dosing regimen with an Omicron variant, which at that time was the main circulating VoC (variant of concern). Mice infected with the Omicron variant did not show obvious signs of disease, such as piloerection and hunched posture, as had been observed with the Beta variant. However, a moderate weight loss was observed following the viral challenge, but not for mice that were pre-treated with spores ( Figure 2F ). In conclusion, these results suggest that spore dosing significantly protects hACE2-transgenic mice from infection caused by both Beta and Omicron variants of SARS-CoV-2. Our next challenge was to determine the immunological factors behind these protective effects.

To investigate the mechanism of spore-mediated protection, we examined leukocyte recruitment to the lungs of spore-dosed mice. Bronchioalveolar lavage (BAL) was collected to determine leukocyte populations that migrated into the airway. By intravascular staining with CD45.2 antibody, lung parenchymal (CD45.2^-) and vasculature (CD45.2⁺) cells were distinguished to accurately define lung tissue infiltrations. Following three intranasal doses of spores, markedly higher numbers of neutrophils and T cells were recovered from the airway, and their frequency and numbers also increased significantly in the lung parenchyma ( Figures 3A, B , Supplementary Figure 2 ). NK cell recruitment was also prominent, but the increase in number was relatively small compared to neutrophils and T cells. The results indicate that spore treatment strongly promotes leukocyte trafficking to the lungs.

Neutrophils and NK cells are important innate infiltrates that promote viral clearance at the early stages of infection and before T cell-mediated adaptive immunity has initiated (20–22). Spore recruited neutrophils and NK cells, however, this did not appear to directly contribute to viral clearance during the early stages of infection, as the lung viral loads in spore-dosed mice were not reduced till day-6 post H1N1 infection ( Figure 1D ).

Apoptosis and necrosis of alveolar epithelial cells induced directly by virus infection and dysregulated inflammation cascades are two major factors contributing to the lethality of highly virulent influenza virus and SARS-CoV-2 (23–26). Alveolar macrophages are critical immune regulators to promote viral clearance and limit tissue damage during respiratory viral infections (27–29). They can be self-renewed in physiological status or replenished by circulating monocytes when their numbers decline during infection. However, tissue-resident alveolar macrophages (TRAMs) can be distinguished from those newly recruited monocyte-derived alveolar macrophages (MoAMs) by expressing higher levels of SiglecF (30). We found that both macrophage populations, SiglecF^high TRAMs and SiglecF^low MoAMs, were significantly increased in spore-dosed lungs ( Figures 3C, D ). Hence, spores not only recruited monocytes, neutrophils, NK cells and T cells from circulation into lungs, but also promoted the expansion of lung-resident alveolar macrophages, which is key to their role in protection against respiratory viral infections. Furthermore, this enhanced activation of macrophages and other immune cells was accompanied by a significantly increased production of anti-viral cytokines, most prominently IFNγ, TNFα, CCL5 and CXCL10, in the airway of spore-treated mice ( Figure 3E ). It is worth noting that spore-treated mice also had a small increase of IFNα in the airway, but they did not produce higher levels of IFNα and IFNβ in response to the viral challenge ( Figure 3E , Supplementary Figure 3 ).

We then examined the effect of spore dosing on leukocyte trafficking in the context of lethal H1N1 infection. Prior to viral challenge and at day 2, 4 & 6 p.i., mouse lungs were harvested to analyse parenchymal leukocyte infiltration. Spore treatment led to a further increase in TRAMs at day 2 p.i. ( Figures 4A, B ), suggesting enhanced self-renewal of spore-activated TRAMs during early exposure to highly pathogenic virus. Whilst the viral infection induced a marked recruitment of MoAMs, neutrophils and NK cells to the lungs of DPBS-treated control mice, mice pre-treated with spores had higher basal levels of these innate immune cells in their lungs but a further recruitment in response to the infection was only seen for NK cells at day 2 p.i. ( Figures 4A–E ). The significant increase in total T cells by day 2 p.i. was also not seen in spore-dosed lungs ( Figure 4F ). In general, pre-treating with spores appeared to dampen the inflammatory responses elicited by highly virulent H1N1 virus, suggesting that spores may prevent the development of severe viral pneumonia by inducing immune tolerance and thereby limiting inflammation associated tissue damage. Furthermore, intranasal boosting with spores significantly increased the numbers of both CD8⁺ and CD4⁺ tissue resident memory T cells (TRMs) following RSV challenge ( Figures 4G–H ). These results show that spores can regulate both innate and adaptive immunity for protection against viral infections.

We demonstrated that intranasal administration of spores protected against multiple viral infections, and this was associated with regulated inflammatory leukocyte trafficking and potentially might also involve enhanced TRM activation. We then further assessed their potential as a nasal booster following parenteral vaccination by first parenterally immunising mice with SARS-CoV-2 recombinant Spike protein, and then boosting with 4 or 9 i.n. doses of spores three weeks after ( Figure 5A ). We detected significantly higher levels of serum IgG and pulmonary IgA against the Spike protein following 9 doses of spores ( Figures 5B, C ), suggesting spores have the potential to ‘pull’ pre-existing immune memory to mucosal sites if used in conjunction with current COVID-19 vaccines.