Repetitive Exposure to Bacteriophage Cocktails against Pseudomonas aeruginosa or Escherichia coli Provokes Marginal Humoral Immunity in Naïve Mice

Phage therapy of ventilator-associated pneumonia (VAP) is of great interest due to the rising incidence of multidrug-resistant bacterial pathogens. However, natural or therapy-induced immunity against therapeutic phages remains a potential concern. In this study, we investigated the innate and adaptive immune responses to two different phage cocktails targeting either Pseudomonas aeruginosa or Escherichia coli—two VAP-associated pathogens—in naïve mice without the confounding effects of a bacterial infection. Active or UV-inactivated phage cocktails or buffers were injected intraperitoneally daily for 7 days in C57BL/6J wild-type mice. Blood cell analysis, flow cytometry analysis, assessment of phage distribution and histopathological analysis of spleens were performed at 6 h, 10 days and 21 days after treatment start. Phages reached the lungs and although the phage cocktails were slightly immunogenic, phage injections were well tolerated without obvious adverse effects. No signs of activation of innate or adaptive immune cells were observed; however, both active phage cocktails elicited a minimal humoral response with secretion of phage-specific antibodies. Our findings show that even repetitive injections lead only to a minimal innate and adaptive immune response in naïve mice and suggest that systemic phage treatment is thus potentially suitable for treating bacterial lung infections.


Introduction
Given the rising incidence of multidrug-resistant (MDR) bacterial pathogens, bacteriophages (phages) are becoming a focus of interest for treating infectious diseases [1,2]. Ventilator-associated pneumonia (VAP), one of the most common nosocomial infections in ventilated patients with a high risk of mortality, is caused mainly by MDR-bacteria [3,4]. These include Pseudomonas aeruginosa (P. aeruginosa) or Escherichia coli (E. coli), which are on the WHO list of "global priority pathogens" continuously developing antibiotic resistance [5,6].
These data potentially open the way for further investigations towards the efficacy of the systemic administration of phages to treat MDR-bacteria-induced VAP.

Animals
All animal studies were approved by the institutional and local governmental authorities at the Charité-Universitätsmedizin Berlin and Landesamt für Gesundheit und Soziales (LaGeSo) Berlin.
Mice were housed under specific pathogen-free conditions with free access to food and water and a 12 h (h) light/dark cycle. Animal housing and experimental procedures complied with the Federation of European Laboratory Animal Science Associations (FELASA) guidelines and recommendations for the care and use of laboratory animals.
The 8-10 week old female C57BL/6J WT mice (Janvier Labs, Le Genest-Saint-Isle, France), a common model organism in preclinical research, were used. Mice were injected intraperitoneally (i. p.) with 100 µL of either active phage cocktail (anti-P. aeruginosa cocktail 5 × 10 7 PFU/injection per phage, anti-E. coli cocktail 1 × 10 8 PFU/injection per phage), UV-inactivated phage cocktail or buffer (saline-magnesium (SM) buffer (0.1 M NaCI, 8 mM MgSO 4 , 50 mM Tris-HCI, pH 7.2-7.5)) once or daily for 7 days (d). The mice were monitored every 24 h for body temperature, body weight and general condition according to pre-defined score sheets. At 6 h (n = 9 per group), 10 d (n = 9 per group) or 21 d (n = 5-6 per group), mice were euthanized with ketamine (200 mg/kg body weight) and xylazine (20 mg/kg body weight). At the 6 h and 10 d time points, intraperitoneal lavage and bronchoalveolar lavage (BAL) were performed, blood samples collected and lungs and secondary lymphoid organs (spleen; draining lymph nodes) removed for further analysis. Blood cell analysis, flow cytometry analysis and determination of plaque-forming units (PFU) were performed. At the 21 d time point, blood samples were collected and spleens were removed for histopathological analysis.

Differential Blood Count and Plasma Preparation
Blood leukocytes and platelet counts were carried out using a Scil Vet abc Hematology Analyzer (Scil animal care company GmbH, Viernheim, Germany). In addition, whole blood was centrifuged at 15,000× g for 10 min at 4 • C, plasma was frozen in liquid nitrogen and stored at −80 • C until further analysis.

Peritoneal Lavage, Bronchoalveolar Lavage and Organ Removal
Mice were subjected to deep anesthesia. Peritoneal lavage was performed by flushing the peritoneal cavity (PC) with 5 mL 1 × phosphate-buffered saline (PBS). After exsanguination via the vena cava, the tracheas were cannulated and lungs lavaged twice with 0.8 mL 1 × PBS protease inhibitor (PI) solution (cOmplete TM , Mini Protease Inhibitor Cocktail; Roche, Basel, Switzerland). BAL and peritoneal lavage samples were stored on ice and analyzed for PFUs.
Lavaged lungs were perfused via the right heart chamber with 10 mL 1 × PBS and homogenized in 1 mL 1 × PBS PI-solution with gentleMACS TM M tubes (Miltenyi Biotec, Bergisch Gladbach, Germany). The homogenates were stored on ice for PFU analysis.
Spleens were removed and stored in 1 × PBS on ice. PFU determination was performed with the supernatant of the single-cell suspensions described below. Draining lymph nodes (inguinal, mesenteric) were collected, pooled in 1 mL PBS and stored on ice for further use.
All stained cells were analyzed using the BD FACS Canto II with the BD FACSDiva software. CountBright Absolute Counting Beads (Thermo Fisher Scientific, Waltham, USA) were used for calculation of total cell numbers.

Histopathology Analysis
The spleens of mice at 21 d were carefully removed and fixed in 4% buffered formaldehyde solution for 24-48 h, embedded in paraffin and cut into 2 µm sections. After routine dewaxing and dehydration, sections were stained with GL7 (GL-7; eBioscience; Frankfurt, Germany) mAb and hematoxylin. Histopathological analysis of the germinal center B cells was performed by the Institute of Veterinary Pathology, Faculty of Veterinary Medicine, Freie Universität Berlin, Germany. The forming of germinal centers was scored into nonexistent (0), minimal (1), low-grade (2), moderate (3) and intense (4). Section analysis was performed in a blinded manner.

Anti-Phage Antibody Measurement via ELISA
Specific anti-phage antibodies (IgG, IgM, IgA) in plasma were measured by ELISA. The anti-P. aeruginosa phage cocktail used for these experiments were prepared from plates with confluent lysis for each individual phage. After resuspension in phage buffer, centrifugation at 1254× g for 15 min w/o break, phage suspensions were filtered consecutively with 0.45 µm and 0.22 µm and stored at 4 • C until further use. Flat-bottom 96-well plates (Sarstedt) were coated with the anti-E. coli phage cocktail (1 × 10 8 PFU/phage) in 100 µL/well or with the anti-P. aeruginosa cocktail (5 × 10 7 PFU/phage in 100 µL) overnight at 4 • C. Wells were washed 5 times with PBS with 0.05% Tween 20 (Sigma-Aldrich, St. Louis, USA) and blocked with 1% bovine serum albumin (BSA; Sigma-Aldrich, St. Louis, USA) in PBS (100 µL/well) at room temperature (RT) for 45 min. Plates were washed 5 times with PBS/0.05% Tween 20. Diluted plasma samples in duplicates were added to the wells (100 µL/well) and incubated at 37 • C for 2 h. Samples were diluted as follows: 1:10 for plasma IgM, IgA and 1:500 for plasma IgG testing for the active phage-cocktail treated mice, while 1:10 each for the UV-inactive cocktail groups and the buffer controls. After washing the plates 5 times with PBS/0.05% Tween 20, 100 µL/well of the corresponding detection antibodies were added and the plates were incubated for 1 h at RT in the dark. The following secondary antibodies were used: biotinylated goat anti-mouse IgG (ThermoFisher scientific, Waltham, USA; SA58-10239), biotinylated rabbit anti-mouse IgM (ThermoFisher scientific, Waltham, USA; SA5-10242), biotinylated rabbit anti-mouse IgA (ThermoFisher scientific, Waltham, USA; SA5-10236). The antibody solution was removed and the plates BioLegend, San Diego, CA, USA) and the corresponding software provided by BioLegend. Samples were analyzed once, according to the manufacturer's recommendations.

Plaque Assays
Blood and organ samples were serially diluted (1:10) in phage buffer shortly before performing the plaque assays, either by double agar overlay assays (anti-E. coli phages) or spot tests (anti-P. aeruginosa phages).
Bacteria cryostocks were streaked on blood agar plates (PA74) or lysogeny broth (LB) agar plates (AN33) one day before. Bacterial cultures (tryptic soy broth (TSB), PA74; LB, AN33) were prepared (20 mL) with single colonies (OD 600 = 0.05-0.08) and cultured with shaking (220 rpm, at 37 • C) until reaching early logarithmic phase (OD 600 = 0.2-0.3 PA74; Viruses 2023, 15, 387 6 of 21 OD 600 = 0.1-0.2 AN33). Low melting agar (soft agar) (4 mL/glass tube) was melted in a heating block (110 • C for 10 min) and cooled down to 48 • C. For the double agar overlay assay, 100 µL of the sample dilution and 100 µL of bacterial suspension were added to the liquid soft agar (top), gently mixed and poured on agar plates (bottom). For the spot test, only 100 µL bacterial suspension was added to the liquid soft agar. Subsequently, sample dilutions were spotted (4 µL/spot) in triplicate directly on the plates. Agar plates were incubated overnight (at 37 • C, 5% CO 2 ) before calculating PFU/mL. The detection limit of the overlay assay for the E. coli phages against AN33 was 10 PFU/mL and the detection limit of the triplicates from the spot test for the P. aeruginosa phages against PA74 was 83 PFU/mL.

Transmission Electron Microscopy Analysis of Phage Suspensions
The two phage cocktails (active and UV-inactivated; 100 µL each) were analyzed via transmission electron microscopy by the Core Facility for Electron Microscopy of the Charité Berlin. For the negative staining, carbon-coated mesh grids were hydrophilized with Alcian blue solution (1% in 1% acetic acid) followed by washing steps in dH 2 O. The grids were placed on a drop of 20 µL particle solution for 10 min. Any residue of the solution on the grids was removed with a filter paper, followed by washing in dH 2 O and finally placing the grids on a drop of freshly prepared solution of 1% aqueous uranyl acetate for 20 s. The droplets on the grids were removed with filter paper and the grids dried until use. Imaging was performed on a Zeiss Leo 906 electron microscope at 80 kV acceleration voltage equipped with a slow scan 2K CCD camera (TRS, Moorenweis, Germany).

Data Analysis
Data are expressed as mean ± SD. For grouped analyses, two-way analysis of variance (ANOVA) with Tukey's multiple comparisons test was performed. Results were considered significant if P was less than 0.05. Significance levels are indicated in the figures. Statistical analysis was performed using GraphPad Prism 9 (San Diego, CA, USA). Sample sizes of individual groups are indicated in the figure legends.

Intraperitoneally Injected Phages Reach the Lungs and Are Harmless to Naïve Mice
In this study, we evaluated the immunogenicity of two intraperitoneally (i. p.) injected phage cocktails in naïve mice. The first cocktail includes two phages targeting P. aeruginosa (5 × 10 7 PFU/phage in 100 µL) and the second includes five phages targeting E. coli (1 × 10 8 PFU/phage in 100 µL).
To determine whether the phage cocktails are recognized by the immune system of naïve mice, we first examined their distribution throughout the body at different time points. C57BL/6J wild-type mice were i. p. injected with 100 µL of either active or UV-inactivated phage-cocktail or buffer every 24 h for 7 days. Blood, spleen, lungs, bronchoalveolar lavage fluid (BAL) and i. p. lavage fluid (peritoneal cavity, (PC)) were collected at either 6 h after the first injection or 72 h after the last injection (10 d) ( Figure 1a) and plaque assays were performed. Six hours following phage injection, we detected active anti-P. aeruginosa and anti-E. coli phages in all analyzed organs and compartments (PC, blood, spleen, lungs), except for the alveolar spaces (BAL). Phages in the BAL were only detected in one of nine (anti-P. aeruginosa phages) or five of nine (anti-E. coli phages) mice at this time point (Figure 1b). On day 10, we still detected anti-P. aeruginosa phages in the PC of all animals and one mouse carried respective phages in the spleen. Anti-E. coli phages were found in few individuals in the spleen (n = 1), BAL (n = 1) and lungs (n = 3) 10 days following treatment initiation (Figure 1b).
Body temperature was similar in all groups and remained in a physiological range. The mice from all groups displayed an in line increase in body weight over time as expected for healthy maturing animals.
Taken together, these data indicate that i. p. injection allows for phage dissemination via the bloodstream as early as 6 h after application. Most importantly, phage treatment was well-tolerated without obvious adverse effects. DL-P. aeruginosa, detection limit of anti-P. aeruginosa phages against PA74 (83 PFU/mL); DL-E. coli, detection limit of anti-E. coli phages against AN33 (10 PFU/mL). Results are shown as mean ± SD; n = 9 mice per group. (c) Graphs displaying body temperature and body weight change over time (dpi, days post-injection). Results are shown as mean ± SEM; n = 9 mice per group (6 h, 10 d) or n = 5-6 (21 d). PMNs, polymorphonuclear neutrophils.
To check whether mice showed clinical signs of illness related to the phage treatment, we monitored them for alteration in body temperature (Figure 1c, left) and body weight (Figure 1c, right), as well as general behavior for up to 21 days after the start of treatment. Body temperature was similar in all groups and remained in a physiological range. The mice from all groups displayed an in line increase in body weight over time as expected for healthy maturing animals.
Taken together, these data indicate that i. p. injection allows for phage dissemination via the bloodstream as early as 6 h after application. Most importantly, phage treatment was well-tolerated without obvious adverse effects.

Blood Neutrophils and Monocytes Remain Unaffected by Phage Treatment
To evaluate the systemic activation and sustainment of innate immunity triggered by the phage cocktails, we carried out complete blood counts at 6 h, 10 d and 21 d after start of treatment (Figures 2 and S3). 2a). However, we did observe a significant increase in the fraction of monocytes at 6 h after the first anti-P. aeruginosa phage cocktail treatment, the biological impact of this 3% increase is likely to be minor, particularly since the number of monocytes per mL blood did not change (Figure 2b). At the later time points of 10 d and 21 d after the start of treatment, no changes in innate immune cell numbers and proportions in blood were detected (Figure 2a,b).  Generally, the parameters, including blood leukocytes and subtypes, platelets, hematocrit (HCT) and hemoglobin (HGB), were inconspicuous and did not differ between treatment groups and controls ( Figure S3). Specifically, the mobilization of polymorphonuclear neutrophils (PMNs) into blood from bone marrow, which has been shown to occur within 6 h after exposure to a substance recognized as foreign [49], was not detected upon anti-P. aeruginosa and anti-E. coli phage cocktail treatments at this time point (Figure 2a). However, we did observe a significant increase in the fraction of monocytes at 6 h after the first anti-P. aeruginosa phage cocktail treatment, the biological impact of this 3% increase is likely to be minor, particularly since the number of monocytes per mL blood did not change (Figure 2b). At the later time points of 10 d and 21 d after the start of treatment, no changes in innate immune cell numbers and proportions in blood were detected (Figure 2a,b).

Lymphatic Innate Immune Cells in the Spleen and Draining Lymph Nodes Do Not Display Signs of Activation after Phage Treatment
To examine the adaptive immunity in lymphoid organs, which is initiated by activated antigen-presenting cells (APCs) loaded with antigen, we first probed spleens and draining lymph nodes for the presence of activated APCs in response to active phage cocktail encounter. Using flow cytometry for expression of costimulatory molecules and MHCII, we did not observe a significant increase in either PMNs, macrophages or dendritic cells (DCs) in the spleen at 6 h and 10 d (Figure 3a-c). PMN frequencies and total cell numbers were increased only at the 6 h time point (Figure 3a). vated antigen-presenting cells (APCs) loaded with antigen, we first probed spleens and draining lymph nodes for the presence of activated APCs in response to active phage cocktail encounter. Using flow cytometry for expression of costimulatory molecules and MHCII, we did not observe a significant increase in either PMNs, macrophages or dendritic cells (DCs) in the spleen at 6 h and 10 d (Figure 3a-c). PMN frequencies and total cell numbers were increased only at the 6 h time point (Figure 3a). . Results are shown as mean ± SD, as determined by 2way ANOVA with Tukey's multiple comparisons test: ** p < 0.01; n = 7-9 mice per group. For full gating strategy and cell type identifying markers see Figure S1. PMNs, polymorphonuclear neutrophils; DCs, dendritic cells.
As DCs are the most competent APCs [50,51], we tested whether they upregulate MHCII on their surface as a marker of activation-induced enhanced antigen processing and presentation [52,53]. The proportion of MHCII-expressing DCs in the spleen did not increase at any of the time points analyzed, but remained in the same range as observed . Results are shown as mean ± SD, as determined by 2-way ANOVA with Tukey's multiple comparisons test: ** p < 0.01; n = 7-9 mice per group. For full gating strategy and cell type identifying markers see Figure S1. PMNs, polymorphonuclear neutrophils; DCs, dendritic cells.
As DCs are the most competent APCs [50,51], we tested whether they upregulate MHCII on their surface as a marker of activation-induced enhanced antigen processing and presentation [52,53]. The proportion of MHCII-expressing DCs in the spleen did not increase at any of the time points analyzed, but remained in the same range as observed for the buffer and UV-inactivated phage cocktail controls (50-90%). Surface mean fluorescence intensity (MFI) of MHCII on DCs also failed to increase. MFI can be used as an equivalent to the number of MHCII molecules present on their surface, as the MFI directly correlates with the amount of stained antigen (Figure 4a). In line with this, the proportions of DCs expressing the costimulatory molecules CD80 or CD86 on their surface did not increase, nor did the MFI of CD80 or CD86 (Figure 4b,c). equivalent to the number of MHCII molecules present on their surface, as the MFI directly correlates with the amount of stained antigen (Figure 4a). In line with this, the proportions of DCs expressing the costimulatory molecules CD80 or CD86 on their surface did not increase, nor did the MFI of CD80 or CD86 (Figure 4b,c).
Matching measurements of draining lymph nodes confirmed the absence of innate immune activation of DCs in secondary lymphoid organs at all time points analyzed (Figure S4).

Phage Treatment Does Not Trigger Adaptive T-Cell Responses in Draining Lymph Nodes or Spleen
Repetitive phage therapy carries a risk of inducing memory cells, which could be detrimental for any potential subsequent phage treatment. As a prerequisite for T-cell memory, we investigated whether repeated phage treatments induced T-cell responses in Matching measurements of draining lymph nodes confirmed the absence of innate immune activation of DCs in secondary lymphoid organs at all time points analyzed ( Figure S4).

Phage Treatment Does Not Trigger Adaptive T-Cell Responses in Draining Lymph Nodes or Spleen
Repetitive phage therapy carries a risk of inducing memory cells, which could be detrimental for any potential subsequent phage treatment. As a prerequisite for T-cell memory, we investigated whether repeated phage treatments induced T-cell responses in the spleen and draining lymph nodes. In lymph nodes, CD4, CD8 and γδ T cell frequencies remained similar in all groups at 6 h and 10 d and represented around 60% (a, CD4+), 40% (b, CD8+) and 1.3% (c, γδTCR+) of all T cells (CD3+) (Figure 5a-c). At 6 h post-treatment with the anti-P. aeruginosa phage cocktail, the proportions of CD4 T cells were moderately yet significantly increased at the expense of CD8 T cells (Figure 5a,b). We did not observe any signs of phage-induced T-cell proliferation. Repetitive treatment with phage cocktails against P. aeruginosa or E. coli did not result in an increase in CD4, CD8 or γδ T cell numbers in the draining lymph nodes at 6 h and 10 d after treatment start compared to the buffer or UV-inactivated phage cocktail controls. ment with the anti-P. aeruginosa phage cocktail, the proportions of CD4 T cells were moderately yet significantly increased at the expense of CD8 T cells (Figure 5a,b). We did not observe any signs of phage-induced T-cell proliferation. Repetitive treatment with phage cocktails against P. aeruginosa or E. coli did not result in an increase in CD4, CD8 or γδ T cell numbers in the draining lymph nodes at 6 h and 10 d after treatment start compared to the buffer or UV-inactivated phage cocktail controls.
Our analysis of the same T-cell populations in the spleen also failed to reveal evidence of T-cell proliferation ( Figure S5). They displayed similar numbers and frequencies in all groups at both the time points analyzed. Figure 5. T cell populations in the draining lymph nodes show no marked changes after phage treatment. Analysis of (a) CD4 + , (b) CD8 + and (c) γδTCR + T cells in the lymph nodes. Shown are representative dot plots (left) and bar graphs depicting the percentage of remaining cells (middle) or total cells (right). Results are shown as mean ± SD, as determined by 2-way ANOVA with Tukey's multiple comparisons test: * p < 0.05; n = 7-9 mice per group. For full gating strategy and cell type identifying markers see Figure S2.
Despite the lack of changes in CD4, CD8 and γδ T cell numbers, we probed if CD4 effector subtypes were induced in the spleen and draining lymph nodes at 6 h and 10 d. T-helper cells type 1 (Th1) were identified by the surface expression of CD4 (CD4 + ), high Our analysis of the same T-cell populations in the spleen also failed to reveal evidence of T-cell proliferation ( Figure S5). They displayed similar numbers and frequencies in all groups at both the time points analyzed.
Despite the lack of changes in CD4, CD8 and γδ T cell numbers, we probed if CD4 effector subtypes were induced in the spleen and draining lymph nodes at 6 h and 10 d. T-helper cells type 1 (Th1) were identified by the surface expression of CD4 (CD4 + ), high expression levels of the activation marker CD44 (CD44 hi ) and intranuclear staining for the Th1-transcription factor T-bet. Only 1% of all CD4 T cells were of Th1 type and no increase in frequencies was observed in response to phage treatment at either time point (Figure 6a). Th17 cells were identified by CD4 + , CD44 hi and intranuclear expression of the Th17-transcription factor RORγt. The proportion of Th17 cells among CD4 T cells was around 0.5%. As for Th1 cells, we did not observe changes in response to phage treatment at either time point (Figure 6b). Regulatory T cells (Tregs) are characterized by CD4 and transcription factor FoxP3 expression. We did not discriminate between natural and peripheral Tregs [54]. Their frequencies were around 12% of CD4 T cells and remained unaffected by phage cocktail treatments at 6 h and 10 d (Figure 6c). As observed for CD4, CD8 and γδ T cells, the findings were similar in the spleen ( Figure S5).
Th1-transcription factor T-bet. Only 1% of all CD4 T cells were of Th1 type and no increase in frequencies was observed in response to phage treatment at either time point ( Figure  6a). Th17 cells were identified by CD4 + , CD44 hi and intranuclear expression of the Th17transcription factor RORγt. The proportion of Th17 cells among CD4 T cells was around 0.5%. As for Th1 cells, we did not observe changes in response to phage treatment at either time point (Figure 6b). Regulatory T cells (Tregs) are characterized by CD4 and transcription factor FoxP3 expression. We did not discriminate between natural and peripheral Tregs [54]. Their frequencies were around 12% of CD4 T cells and remained unaffected by phage cocktail treatments at 6 h and 10 d (Figure 6c). As observed for CD4, CD8 and γδ T cells, the findings were similar in the spleen ( Figure S5).

Minimal Pro-Inflammatory Cytokine and Chemokine Secretion after Phage Treatment
The activation of innate and adaptive immune cells and the strength of initiated immune responses rely on cytokine and chemokine milieus [55,56]. Using the Mouse Anti-Virus Response Panel (13-plex, Legendplex TM ; BioLegend, San Diego, CA, USA), soluble cytokines and chemokines were analyzed in plasma samples at 6 h, 10 d and 21 d after the

Minimal Pro-Inflammatory Cytokine and Chemokine Secretion after Phage Treatment
The activation of innate and adaptive immune cells and the strength of initiated immune responses rely on cytokine and chemokine milieus [55,56]. Using the Mouse Anti-Virus Response Panel (13-plex, Legendplex TM ; BioLegend, San Diego, CA, USA), soluble cytokines and chemokines were analyzed in plasma samples at 6 h, 10 d and 21 d after the first injection ( Figure S6). Six hours after treatment with the active anti-P. aeruginosa phage cocktail, the pro-inflammatory factors CXCL10, CCL2, CCL5 and IL-1β, were significantly increased, while treatment with the active anti-E. coli phage cocktail led to a significant increase in IL-6 and IL-10. At 10 d, IFN-γ levels were increased in the UV-inactivated anti-P. aeruginosa group compared to the buffer control or the active phage cocktail group. However, these changes may be reflective of individual outliers and were generally low.

Phage Treatment Induces Minimal to Low-Grade Germinal Center Formation
Humoral immunity, particularly neutralizing antibodies, could impair phage therapy. We therefore investigated if signs of germinal center reaction were visible at 21 d after treatment start, which would indicate the initiation of humoral B cell responses. Immunohistochemistry identified differences between the individual treatment groups. Spleens from mice that received buffer (I), UV-inactivated anti-P. aeruginosa phage cocktail (II) or UV-inactivated anti-E. coli phage cocktail (III) did not show any specific positive signals. In contrast, both the active phage cocktail groups (IV, V) showed minimal to minor positive signals for germinal center B cells stained by GL7 (Figure 7). These two groups also showed morphological evidence of an incipient formation of follicular centers. cocktail, the pro-inflammatory factors CXCL10, CCL2, CCL5 and IL-1β, were significantly increased, while treatment with the active anti-E. coli phage cocktail led to a significant increase in IL-6 and IL-10. At 10 d, IFN-γ levels were increased in the UV-inactivated anti-P. aeruginosa group compared to the buffer control or the active phage cocktail group. However, these changes may be reflective of individual outliers and were generally low.

Phage Treatment Induces Minimal to Low-Grade Germinal Center Formation
Humoral immunity, particularly neutralizing antibodies, could impair phage therapy. We therefore investigated if signs of germinal center reaction were visible at 21 d after treatment start, which would indicate the initiation of humoral B cell responses. Immunohistochemistry identified differences between the individual treatment groups. Spleens from mice that received buffer (I), UV-inactivated anti-P. aeruginosa phage cocktail (II) or UV-inactivated anti-E. coli phage cocktail (III) did not show any specific positive signals. In contrast, both the active phage cocktail groups (IV, V) showed minimal to minor positive signals for germinal center B cells stained by GL7 (Figure 7). These two groups also showed morphological evidence of an incipient formation of follicular centers.

Repetitive Phage Treatment Evokes Phage-Specific Antibody Response
The observed formation of minimal splenic germinal centers prompted us to further investigate the humoral immune response against the two phage cocktails. To this end, we measured phage-specific antibodies (IgG, IgM, IgA) in plasma by ELISA (Figure 8). treatment with the active anti-P. aeruginosa phage cocktail only induced a marked increase in IgG and only at the 21 d time point, but this did not reach significance. A small increase in IgG was also induced by the UV-inactivated phage cocktails but may be reflective of individual outliers. These findings suggest that the repetitive injection of each of the two phage cocktails initiated a minimal humoral B-cell response with secretion of phage-specific antibodies.

Destruction of Phage Structures by UV-Inactivation
For evaluating the immunogenicity of the two phage cocktails, we assumed that the immune response against the UV-inactivated and the active phages would be comparable, as we expected the UV treatment to merely cross-link phage DNA [57]. To understand the low to absent humoral response against the UV-inactivated phage cocktails (Figures 7 and  8), we performed negative-staining transmission electron microscopy of all phage preparations ( Figure S7). In the active phage cocktails, virus-like structures were identified (Figure S7a,b). In contrast, following UV-inactivation, only artifacts that may be fragments of phages could be found ( Figure S7c,d).

Discussion
Phage treatment of bacterial lung infections may be developing into an effective resource in the fight against the increasing antibiotic resistance of clinically relevant pathogens. As natural or therapy-induced immunity against therapeutic phages remains a safety and efficacy concern, in this study we assessed the immunogenicity of two phage In congruence with the presence of germinal centers, the active anti-E. coli phage cocktail induced an increase in IgG, IgM as well as IgA by 21 days compared to the buffer or UV-inactivated control, which for IgG was already evident by 10 d. In contrast, the treatment with the active anti-P. aeruginosa phage cocktail only induced a marked increase in IgG and only at the 21 d time point, but this did not reach significance. A small increase in IgG was also induced by the UV-inactivated phage cocktails but may be reflective of individual outliers. These findings suggest that the repetitive injection of each of the two phage cocktails initiated a minimal humoral B-cell response with secretion of phage-specific antibodies.

Destruction of Phage Structures by UV-Inactivation
For evaluating the immunogenicity of the two phage cocktails, we assumed that the immune response against the UV-inactivated and the active phages would be comparable, as we expected the UV treatment to merely cross-link phage DNA [57]. To understand the low to absent humoral response against the UV-inactivated phage cocktails (Figures 7 and 8), we performed negative-staining transmission electron microscopy of all phage preparations ( Figure S7). In the active phage cocktails, virus-like structures were identified ( Figure S7a,b). In contrast, following UV-inactivation, only artifacts that may be fragments of phages could be found ( Figure S7c,d).

Discussion
Phage treatment of bacterial lung infections may be developing into an effective resource in the fight against the increasing antibiotic resistance of clinically relevant pathogens. As natural or therapy-induced immunity against therapeutic phages remains a safety and efficacy concern, in this study we assessed the immunogenicity of two phage cocktails targeting either P. aeruginosa or E. coli in naïve mice without the confounding effects of a bacterial infection. Our results show that phages reached the lungs after systemic injection. Even repetitive exposure to the phage cocktails led to only a minimal innate and adaptive immune response, while all mice remained healthy without evidence of any adverse effects.
Assessing the possible immune responses towards the two different phage cocktails, we did not observe a significant activation of innate immune cells in blood or lymphoid organs at 6 h after treatment, except for splenic PMNs in response to the active anti-P. aeruginosa phage cocktail. There was no upregulation of MHCII or co-stimulatory molecules on DCs that would indicate maturation into APCs and only a minimal proinflammatory cytokine profile at this time point. Despite the lack of evidence for an early innate response, we did observe a minimal adaptive immune response. Although we failed to observe the activation of main T cells or sub-populations at 10 days post-treatment, the histopathology analysis at 21 days nonetheless revealed a minimal to low-grade germinal center formation with increasing levels of phage-specific antibodies in plasma. The active anti-E. coli cocktail induced IgG and lower levels of IgA already at 10 days and IgM at 21 days, while the anti-P. aeruginosa phages induced only IgG and only at the late point of 21 days.
Phages can induce pro-and anti-inflammatory responses, depending on phage type, the route and duration of administration and the overall amount of phage virions present in the organism [58,59]. Consistent with our results, minimal cytokine release induced by phage treatment has also been reported in several other studies, indicating promising signs of immunogenic tolerance. Low cytokine levels after prophylactic phage treatment by intraperitoneal application were also reported in a murine A. baumannii infection study [60] or after the oral application of bacteriophage T7 for 10 days [61]. Differences due to the application routes of phages were demonstrated in a murine efficacy study of two A. baumannii phages, where intraperitoneal application resulted in the most pronounced, yet still low, immune response compared to oral or intranasal application [62].
Our results of increasing IgG titers and decreasing IgM titers over time are consistent with an antibody class switch. Similar findings, as well as weaker systemic IgA titers at later time points, have been shown by other studies [29,33]. IgA is the predominant antibody in mucosal immunity [63,64]. It is induced upon oral or intranasal immunization with bacteria [65,66] or even oral application of phages: Majewska et al. reported an IgA rise after the oral application of T4 phages in a long-term study [37] and after therapeutic administration of two staphylococcal phages mixed with the drinking water [33]. Interestingly, secreted phage-specific IgA appeared to be the major factor limiting phage activity in the gut, but the authors also noted that once the phage was removed from the diet, secretory IgA decreased over time [33]. We did not measure phage-specific IgE, which would indicate an allergic reaction. However, Cha et al. could show only a low increase in IgE production in serum upon daily intraperitoneal injection of an anti-A. baumannii phage cocktail [62].
Although we did find some evidence of a weak antibody response to our phage cocktails, antibodies do not necessarily have neutralizing capacity-the main concern regarding the immune response to therapeutic phages-since phage-specific neutralizing antibodies could diminish therapeutic success. However, although the presence of phage-specific antibodies with neutralizing capacity, in particular IgG2a and IgG2b, was detected after repeated intraperitoneal applications of an A. baumannii-specific phage cocktail in naïve mice, this did in fact not negatively affect the efficiency of the phages in a wound infection model [60]. Antibody formation and neutralizing capacity depend on the phage type, especially the capsid and tail protein composition; therefore, phages vary in immunogenicity and can induce different antibody responses [58,59,67]. In a murine study, Chechushkov et al. [67] found that phages with podovirus morphotype were nonimmunogenic. In contrast, they observed an increase in neutralizing antibodies after triple immunization with phages of myovirus and siphovirus morphotypes [67]. Additionally, Majewska et al. identified multicopy proteins of phages [33], the T4 phage proteins Hoc, Soc and gp12 [37,68], as well as the conserved structural proteins gp22 and gp29 of Pseudomonas phages [69] as highly immunogenic. In our study, the two P. aeruginosa phages belonged to the myovirus morphology, as well as four E. coli phages, the fifth belonging to the podovirus morphotype. In addition, our findings, as well as those of other groups [33,37] align with the significant rise in neutralizing antibodies observed after 3 weeks in a rabbit model [32]. Thus, there appears to be a "therapeutic window" of phage treatment prior to the production of phage-specific antibodies [29]. Whether the antibodies observed in our model have a neutralizing capacity has to be investigated further, ideally in a murine VAP model, which will also allow analysis of any effects on therapeutic efficacy. So far, however, most animal [32][33][34] and human [36][37][38]70] studies that have investigated antibody responses to phages found that they do not interfere with phage therapy.
Natural antibodies result from continuous contact with endogenous phages, which make up a significant part of the mammalian virome [26,71]. If such pre-existing natural antibodies against phages in our cocktails were present in the mice, we would have expected to observe them even in the groups that received the UV-inactivated phage cocktails or buffer only. Hodyra-Stefaniak et al. reported natural antibodies against PB1related Pseudomonas phages [69] as well as Staphylococcus phages [15] in the serum of healthy human volunteers. However, they had only a low neutralizing capacity. A rapid increase in IgG levels after initial immunization may indicate the previous contact of experimental mice with the applied phages [67]. Conversely, naturally present phages in the gut may lead to immunological tolerance [59]. Priming the immune system with phages from the microbiome could explain the stronger reaction to the E. coli phage cocktail we observed in the present study.
However, the different responses to the different phage cocktails may not root in the immunogenicity of the phages themselves but rather in technical aspects: the E. coli cocktail consisted of five different phages each at 1 × 10 8 PFU/phage in 100 µL, while the P. aeruginosa phage cocktail contained only two each at 5 × 10 7 PFU/phage in 100 µL, resulting in a higher total phage sum applied for the E. coli cocktail. Endotoxins and other bacterial components that remain in the solution during phage production may elicit immune responses by themselves [72]. Notably, both phage cocktails were highly purified, however the finally applied mix of active E. coli phages had a lower endotoxin level compared to the anti-P. aeruginosa phage cocktail, indicating that in our study endotoxin concentration did not correlate with immunogenicity. Despite the two different protocols, the immune responses to each phage cocktail remain globally weak showing that different processes can lead to similar outcomes. The absence of humoral immunity observed with UV-inactivated phages is elucidated by the EM analysis. It revealed that the UV-inactivation protocol led to major damage to viral particles. Thus, intact phages seem to elicit minimal humoral-immune responses, while phage fragments alone seem non-immunogenic.
Our study has some limitations. Since we chose three different time points to study the possible early innate and late adaptive immune responses, we might have missed some responses that occurred in between. We also did not extend our analysis beyond 3 weeks, which may have provided insight into the later humoral response. In addition, we did not investigate B-and T-cell memory, which are the main parameters of rapid induction of phage-specific antibodies following phage treatment after a second interval. Moreover, our study is limited to the two phage cocktails investigated here with their corresponding phage concentration. Preparations with lower or higher phage content might induce differential immune responses. In our study, we used naïve mice under SPF conditions without a bacterial infection, to avoid any confounding immune effects by the infection itself. However, the presence of bacteria may indirectly affect the immunogenicity of phages as well. Phage-induced bacterial lysis can further boost inflammation via bacterial particles and PAMPs, as well as cell-free phage DNA and higher numbers of phage virions released locally [58], potentially acting adjuvant-like and triggering the immune system to perceive phages as foreign [73]. However, during phage treatment in a murine pneumonia model caused by pathogenic E. coli strains, the rapid lysis of bacteria by phages did not increase the innate inflammatory response compared to that observed after antibiotic treatment [39]. Nevertheless, the interaction of phages with immune cells during an active infection has to be further investigated since a possible synergistic mechanism of phages with innate immune cells has been reported [44].

Conclusions
In the context of lung infections with MDR-bacteria, phage therapy is of great interest for treating VAP patients. While recent studies used intratracheal applications [74] or nebulized phage suspensions [7,75,76], here we show that systemic applications of two different phage cocktails are suitable for the dissemination of phages to the lung. The phage cocktails are well-tolerated and do not induce a marked immune response in naïve mice warranting their future evaluation as antibacterial therapeutics.
Supplementary Materials: The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/v15020387/s1. Figure S1. Exemplary flow cytometricgating strategy for the analysis of innate immune cells in secondary lymphoid organs; Figure S2. Exemplary flow cytometric gating strategy for analysis of adaptive immune cells (T cells) in secondary lymphoid organs; Figure S3. Blood cells remain unaffected by phage treatment; Figure S4. No changes of innate immune cell frequencies and numbers in draining lymph nodes in response to phage cocktail treatment observed; Figure S5. T-cell populations in the spleen show no marked changes after phage treatment; Figure S6. Minimal cytokine release in response to phage treatment; Figure S7. Negative staining transmission electron microscopic images of phage cocktails reveal that UV-inactivation may lead to destroyed phage structures.