Role of bacteria and microbial metabolites in immune modulation during early life

Host–microbiome interplay from birth is essential for immune imprinting and tuning. Live gut microbes and microbial‐derived metabolites regulate the development and modulation of the immune system, but whether microbial metabolites solely are sufficient to induce immune maturation remains unclear. Sterile faecal filtrates (FFT) were generated from murine gut contents. Newborn germ‐free (GF) mice were treated twice daily with FFT (GF‐FFT) or saline (GF‐NaCl) from post‐natal day 5 until 4 weeks of age. A third group of GF neonates were conventionalized by the transfer of caecal microbiota with live gut microbes. Host immune compartments were comprehensively immunophenotyped and systemically analysed in all available immune‐related organs using flow cytometry. Oral FFT was associated with reduced survival among neonates (n = 7/19; 36.8% mortality), while saline treatment was well tolerated (n = 1/17, 5.9% mortality). Four‐week‐old FFT‐treated pups were comparable in body weight to GF‐NaCl, and the major B‐cell, conventional T‐cell and unconventional T‐cell subsets were unchanged from saline‐treated mice. Live bacteria administered during early life induced clear changes in proportions of B cells, T cells and T‐cell subsets in all mucosal tissues and secondary lymphoid organs compared to GF‐FFT, including restoration of intestinal natural killer T (NKT) cells with characteristics similar to conventional pups. Our findings show that oral administration of a FFT made of microbial metabolites, antigens and bacteriophages alone is insufficient to induce normal immune development elicited by the presence of live bacteria. Reduced survival during neonatal FFT treatment suggests a potential bioactive attribute of sterile faecal filtrates.


| INTRODUCTION
The development, education and tuning of the immune system is a governed interplay between the host and its commensal microorganisms, starting at birth and continuing throughout life. 1 Evident correlations between microbial perturbations and disease in numerous human diseases 2 advocate a role of gut microbiome manipulation and microbiome-based therapeutics in future clinical practice.Whether dysbiosis represents causation and correlation or is a consequence of disease largely remains elusive.However, use of faecal microbiota transplantation (FMT) in patients with recurrent Clostridioides difficile infection (rCDI) is highly effective 3 suggesting a primary pathogenic role in some diseases.
Despite its effectiveness, established FMT treatment entails the risk of adverse off-target effects 4 that may be prevented using more targeted approaches.To improve safety of microbiota-restoring therapies, there is a need for an improved mechanistic understanding of the hostmicrobiota interplay.A sterile faecal filtrate transfer (FFT) processed from faecal microbiota slurry but devoid of live bacteria has demonstrated clinical effects in patients with Clostridioides difficile infection (CDI) 5 and improved resilience against necrotizing enterocolitis in preterm piglets, presumably by transfer of bacterial components, metabolites and engraftment of donor bacteriophage populations. 6Also, transfer of bacteriophages during FMT in patients with CDI has been reported to correlate with treatment outcome. 7e used germ-free (GF) mice as recipients due to their 'naïve' immunity, not yet encountered by microbes, and chose the time point of early life when the immune system is especially receptive to microbiota-dependent immune modulation.In GF mice, structural and functional immune compartments are affected in absence of microbes as evidenced by diminished lymphoid structures in close proximity to mucosal surfaces 8 and altered immune cell frequencies of myeloid progenitors, innate lymphoid cells and regulatory T cells. 1 Pathologic imprinting triggered by a disrupted early-life microbiota alters colonic transcriptional signatures of upregulated proinflammatory genes 9 and elevates proinflammatory cytokines 10 with the risk of increasing susceptibility to proinflammatory immune responses, allergies and autoimmunity later in life. 11,12n the current study, we aimed to investigate the immunomodulatory potential of FFT consisting of microbial antigens, metabolites and bacteriophages, within the gut and beyond.Previous reported effects of microbial-derived molecules or metabolites have demonstrated immunomodulatory effects of microbial antigens when introduced either purified or jointly with a specified bacterial monoculture to GF mice, 13 and studies have indicated modulatory effects of bioactive metabolites in recipients with or without an already evolved microbiome causing profound changes in immune cell populations 14 and reprogramming of host immune responses. 15,16Whether bacteriophages residing in the gut can interact directly with the host to induce immune responses is uncertain, but bacteriophages can translocate over the intestinal barrier and induce innate immune responses. 17Herein, we explored to what degree feeding with a mixed lysate of microbial-derived metabolites, antigens, bacteriophages and debris could restore the juvenile mucosal immunophenotype evident in mice introduced to microbes during early life.We believe the experimental approach presented may serve as a framework for future studies of microbiome-based interventions and its effects on gut ecology, metabolomics and host mucosal immunity.

| Mice
GF C57BL/6J mice were obtained from the University of Bern Clean Mouse Facility.Both GF and conventionalized offspring from these GF animals, housed in a specific pathogen-free (SPF) environment at the animal facility at Oslo University Hospital Rikshospitalet, Oslo, Norway, for more than two generations were used in the experiments.All animals were housed in a humidity and temperature-controlled environment on a 12/12-hour day/night cycle and had ad libitum access to water and autoclaved rodent chow diet (Labdiet 5021, IPS Products

| Mouse stool collection
Caecal and colon content was collected from adult SPF mice.The genetic background of donor mice can be reviewed in Table S1.All mice were sacrificed by cervical dislocation, and intestinal content from caecum and proximal colon was harvested in an anaerobic nitrogen saturated local environment.The intestinal contents were mixed with yeast peptone dextrose (YPD) broth to nourish heterotrophic microorganisms and snap-frozen in liquid nitrogen at −80°C for preservation.

| Preparation of sterile faecal filtrate
The filtration method was modified from a filtration process previously described for human material. 5Briefly, 175 g of snap-frozen caecal contents harvested from mice of various genetic backgrounds in our SPF facility, was thawed and mixed immediately before processing.One part of murine caecal contents was homogenized and ground using a bead beater (Glen Mills Inc, Clifton, NJ, USA) with three parts of sterile and cold isotonic saline and two parts 0.1-mm refrigerated glass beads while maintained on ice.Next, the faecal slurry was distributed into 500-mL centrifuge tubes and centrifugated at 14 000 g for 10 minutes at 4°C to separate faecal supernatant from the solid content.Faecal supernatant was collected and transferred to an autoclaved pressure chamber and prepared at room temperature using three filtration steps connected with sterile tubes.At 2 bar pressure, the supernatant was first filtered through 6-15 μm (Supracap 50, Seitz K700P, Pall Europe, Portsmouth, UK) and 0.4-to 0.8-μm filters (Seitz KS50, Pall Europe) and lastly through a 0.2-μm filter for bacterial depletion (Mini Kleenpak, Pall Europe).The sterile-filtered faecal filtrate (FFT) was aseptically aliquoted into autoclaved 1.5-mL Eppendorf tubes.Bacterial sterility was confirmed by microbial cultivation as described below.Aliquots of FFT were stored at −80°C and thawed immediately before oral administration to mice.

| Caecal microbiota transfer (CMT)
Caecal contents collected as described above were thawed and homogenized in saline, vortexed and passed through a 100-μm filter.Germ-free animals (mothers and pups) were given caecal microbiota transfer orally on the postnatal day (PND) 4, 5 and 9.In addition, caecal microbiota was applied on the mothers' fur.The conventionalized animals were handled by aseptic technique in autoclaved IVC cages and provided with autoclaved diet and water throughout the experiment to prevent colonization of other microbiological flora than those originating from the CMT slurry.

| Study design
Four GF litters were divided and allocated into two treatment groups for either treatment with FFT (GF-FFT) or saline (GF-NaCl).Based on date of birth, the inclusion of the litters was staggered by 2 days.First, two litters born within 24 hours were allocated to the two treatment groups.Two days later, two additional litters born within 24 hours were likewise allocated to treatment with either FFT or saline.The GF pups were fed orally by gavage two times daily with FFT or saline from PND 4-5 until PND 30 (±3 days) with a starting dose of 20 mL/kg at PND 4-14, increasing to 40 mL/kg at PND 15.After administration, all mice were monitored for clinical signs, including regular breathing, skin colouring and normal activity and behaviour for 30 minutes.GF mice were conventionalized by caecal microbiota transfer (GF/CMT) of a faecal slurry of live microbes received orally at PND 4, 5 and 9. Ageand gender-matched conventional (CONV) and GF mice reared without any treatment were included as controls.Experiments were terminated by PND 30 ± 3 days, and all experimental mice were harvested and processed in four batches during an 8-day period before pooling results of each treatment group.GF microbiological status of GF-FFT, GF-NaCl and GF mice was confirmed by bacterial cultivation of faecal pellets as described below.

| Bacterial cultivation
Processed FFT samples and faecal pellets underwent bacterial cultivation for the detection of anaerobes and aerobes.Filtrate samples and faecal pellets were seeded on HMO (human milk oligosaccharides) agar plates and incubated for 64-72 hours at 35-37°C in an anaerobic atmosphere.Aerobic cultivation was conducted by seeding samples in HIB (brain heart infusion) broth at 35-37°C for 5 days in an aerobe atmosphere containing 4%-6% carbon dioxide followed by incubation on chocolate agar for 18-24 hours.

| Tissue collection and lymphocyte isolation
All mice were euthanized using CO 2 gas at 4 weeks of age.Thymus, spleen, liver, mesenteric lymph nodes, small intestine and colon were harvested.In brief, spleen, thymic tissue and mesenteric lymph nodes were homogenized and filtered through a 40-μm cell strainer and centrifugated at 300-400 g.After centrifugation, the red blood cells in spleen samples were lysed (RBC lysis buffer 10x; BioLegend, San Diego, CA, USA) before a final PBS wash.The livers were minced in petri dishes with cold PBS.The suspensions were then filtered through a 70-μm cell strainer, washed with PBS and pelleted.Liver lymphocytes were isolated using Percoll 40%/70% gradient centrifugation at 700 g for 30 minutes and followed by a final wash with PBS.Peyer's patches collected from the anti-mesenteric side of the small-intestinal serosa were homogenized through a 70-μm cell strainer, flushed with a washing buffer consisting of PBS supplemented with 5% foetal bovine serum (FBS) (Sigma-Aldrich, Saint Louis, MO, USA), 15 mM Hepes (Sigma-Aldrich) and 200 U penicillin/mL, 200 μg/mL streptomycin and 0.5 μg/mL amphotericin (Gibco Antibiotic-Antimycotic; ThermoFisher Scientific, Waltham, MA, USA) and centrifugated.
Small and large intestines were cut in pieces and flushed with the washing buffer described above followed by EDTA wash in the same washing buffer supplemented with 5 mM EDTA (Sigma) in a shaker at 37°C for 30 minutes.The supernatants containing epithelial cells and intraepithelial lymphocytes (IEL) were filtered through a 70-μm cell strainer.IELs were isolated using 40%/70% Percoll density gradient centrifugation at 1600 g for 20 minutes.Remaining intestinal pieces were washed in PBS and minced upon digestion with Gibco Dulbecco's modified Eagle's medium (DMEM) (ThermoFisher Scientific) supplemented with 10% FBS, 0.25 mg/mL Collagenase type 1 (Sigma-Aldrich) and 15 mM Hepes (Sigma-Aldrich).Following digestion, the remaining tissues were homogenized and filtered through a 70 μm cell strainer.Lamina propria lymphocytes (LPL) were isolated using 40%/70% Percoll density gradient centrifugation at 1600 g for 20 minutes.Finally, both IEL and LPL were washed in PBS.All cell pellets were freshly stained for flow cytometry.

| Flow cytometry staining
All cell suspensions were resuspended in PBS containing 2% FBS (Sigma-Aldrich) and 2% Gibco Antibiotic-Antimycotic (ThermoFisher Scientific) and incubated with anti-CD16/32 (Fc block, 101 302; BioLegend, San Diego, CA, USA) in 1:100 dilution for 30 minutes.Cells were stained with individually titrated fluorescent labelled antibodies for 60 minutes at 4°C and fixed by incubation in the FoxP3/Transcription Factor Fixation/ Permeabilization solution (eBioscience, San Diego, CA, USA) for 60 minutes.A full list of titrated antibodies can be found in Table S2.5-OP-RU MR1 tetramer and CD1d-PBS-57 tetramer were a kind gift from the NIH Tetramer Facility at Emory University, Atlanta, GA, USA.For intracellular staining, cells were incubated overnight with Fc block 1:100 diluted in FoxP3/Transcription Factor Permeabilization buffer (eBioscience) followed by washing the next morning.Cells were then incubated for 45 minutes with intracellular staining antibodies diluted in permeabilization buffer.All cells were washed and resuspended in flow buffer (PBS supplemented with 2% FBS and 2% Antibiotic-Antimycotic) for data acquisition.

| Flow cytometry data analysis
Data acquisition was performed on a BD FACSymphony A5 (BD Biosciences, Franklin Lakes, NJ, USA) and analysed using FlowJo software version 10.8.1 (FlowJo LLC Biosciences, Ashland, OR, USA).An automated Cytometer Setup & Tracking (BD Biosciences) procedure was applied to assess instrument stability prior to every run.Single-stained compensation beads (AbC Total Antibody Compensation Bead Kit, Invitrogen / ArC Amine Reactive Compensation Bead Kit, Invitrogen, Waltham, MA, USA) were used for compensation.Tetramers were replaced in the compensation controls by APC-CD4 or PE-CD4.The concentrations and clones of the antibodies were kept constant throughout the experiment to ensure staining consistency.FlowJo's AutoSpill/AutoSpread algorithm was applied, and additional manual compensation adjustments were performed.All data underwent a manual clean-up and quality control before further analysis.Live CD45 + cells were gated and characterized for deeper immunophenotyping.An analytical strategy of investigated populations is shown in Tables S3, S4.Immune cell populations not detected due to staining issues were excluded from further analyses.The mean percentage of every population were calculated, and all values were log2 transformed and normalized to the mean percentage of the conventional group.Heatmaps were generated in GraphPad Prism statistical software version 9.3.1 (GraphPad Software, San Diego, CA, USA).The complete dataset of mean cell type frequencies for each tissue per experimental group is summarized in Table S5.

| Flow virometry
Flow virometry was performed as previously described. 18n short, FFT samples and Myoviridae bacteriophage controls 10 11 PFU/mL (Custus®YRS, ACD Pharma, Ås, Norway), were thawed, diluted in TE-buffer (10 mM Tris, 1 mM EDTA) and stained with SYBR Green-I Nucleic Acid Gel Stain (Invitrogen) diluted 1:200 in distilled water with a final concentration when added to samples or controls of 1:20.000.FFT samples and bacteriophage controls were diluted 1:50 and 1:100.000,respectively, in TE-buffer before staining. Untained FFT samples and TE-buffer served as negative controls.All samples were heated up to 80°C in a water bath and incubated for 10 minutes.After 5 minutes cooling in room temperature, samples were immediately analysed using a standard flow cytometer (MACSQuant Analyzer 10, Miltenyi Biotec, Bergisch Gladbach, Germany).The flow cytometer was cleaned thoroughly by performing the instrument clean programme before running samples to reduce noise and background signals.The flow cytometer was considered clean when detecting <75 events/s when running distilled water at low rate.Fluorescence-labelled viral DNA was singled out according to side scatter signals and obtained green fluorescent signal after staining.Data were analysed in FlowJo software version 10.8.1 (FlowJo LLC Biosciences).

| Statistics
Statistical data analyses were performed using GraphPad Prism statistical software version 9.3.1 (GraphPad Software).Data distributions were assessed visually using Q-Q plots.Two groups were compared using Student's t test or Mann-Whitney test for data not following a normal distribution as indicated.For multiple comparisons, one-way ANOVA with Bonferroni post hoc test was used for normally distributed data with equal variances and Kruskal-Wallis one-way ANOVA with Dunn's multiple comparisons test was used for data not following a normal distribution.Welch ANOVA with Dunnett's T3 multiple comparisons test was used for data with unequal variances.F test was used to assess the equality of variances between the experimental groups.Survival was visualized by Kaplan-Meier survival curves and distributions compared using the log-rank test.Corrected P-values <.05 were considered statistically significant.

| Filtrate-feeding of germ-free pups does not impair normal growth
Germ-free newborn mice were orally fed either sterile-filtered faecal filtrate (FFT) or saline or conventionalized by repeated oral caecal microbiota transfer (CMT) (Figure 1A,  B).During gavage, seven filtrate-treated GF mice (GF-FFT) and one saline-treated mouse (GF-NaCl) died from acute respiratory distress (Figure 1C).All casualties occurred shortly after gavage, either at the first day of treatment or coincided with a step-up in dose to match weight estimations at post-natal day 15.Significantly reduced survival was observed among GF-FFT (n = 7/19; 36.8% mortality) compared to GF-NaCl (n = 1/17, 5.9% mortality) during the treatment period inferring an active component in the filtrate (P = .043).After finishing 4 weeks of treatment, the developmental characteristics as body growth and behaviour were unchanged among GF-FFT and GF-NaCl and similar as the conventional (CONV) untreated group (Figure 1D).Spleen weight as per cent of total body weight was similar between all groups (Figure 1E).GF mice undergoing caecal microbiota transfer (GF/CMT) were successfully conventionalized and adapted to the microbial environment demonstrated by reduction of caecum size corresponding to CONV age-matched mice (Figure 1F).Although the FFT was devoid of live bacteria and sterility was verified (data not shown), positive nuclear stain was detected by flow virometry most likely representing viral DNA and bacteriophages in the FFT (Figure S1).

| Microbial effects on T-cell phenotype at inductive sites for mucosal immunity
Immune compartments from immune-related organs were processed, broadly immunophenotyped and systemically analysed in 4-week-old mice representing the juvenile state.The following analyses generated comprehensive data of 137 individual immune cell populations locally in the gut, in gut-associated lymphoid tissues and secondary lymphoid organs.Gating strategies are shown in Figure S2.Immunologic effects of both early colonization and filtrate treatment in gut-associated lymphoid tissues specialized for mucosal immune induction and regulation were evaluated (Figures 2 and 3).In Peyer's patches, the small-intestinal lymphoid aggregates forming the primary sites for mucosal immune responses, the presence of commensal gut microbes had no impact on the overall proportions of B cells and T cells (Figure 2B, C), but T cells expressing the γδ T-cell receptor were modified the microbial microenvironment and proportionally decreased in both CONV and conventionalized GF/CMT (Figure 2D).However, filtrate-feeding of microbial byproducts did not induce a clear immunologic response in γδ T cells as the subset was proportional similar with saline-treated mice.Subsets of conventional T cells in Peyer's patches were altered by gut microbial exposure as mice with an established microbiome displayed a CD4 + T cell-skewed immunophenotype contrasting the CD8 + T cell-skewed GF phenotype (Figure 3B, C).Additionally, the presence of commensal microbes proportionally reduced FoxP3 + CD4 + T cells (Figure 3D).Filtrate-feeding did not change the proportions of CD4 + T cells, CD8 + T cells or FoxP3 + CD4 + T cells from saline-treated GF mice.Mesenteric lymph nodes (MLNs) draining lymphatics from the intestinal wall of the small and large bowel demonstrated that T cells, B cells and conventional NK cells (cNK) in these lymph nodes seem to be influenced by an intact microbiota when comparing conventionalized and saline-treated GF mice (Figure 2A), despite that the GF untreated group had similar levels of cNK cells as CONV (Figure 2E-G).Filtrate-feeding failed to induce a similar immunologic shift in small-intestinal (SI) MLNs as evidenced by early-life colonization (GF/CMT) inducing reduction in B cells and increased T cells and cNK cells (Figure 2E-G).A notable finding was a significant reduction of cNK cells among GF-FFT and GF-NaCl mice compared to both untreated GF mice and colonized mice (Figure 2G).The microbial presence had limited immunomodulating impact on immune cell proportions in SI MLNs among investigated CD4 + and CD8 + T-cell subsets (Figure 3E-J).Absolute cell counts of CD45 + lymphocytes in germ-free mice were significantly lower compared to conventional mice in Peyer's patches, while lymphocyte counts were not significantly reduced in small-intestinal MLNs (Figure S3).For the lymphocyte subsets, GF mice exhibited lower absolute cell counts of B cells and T cells in the Peyer's patches and lower counts of T cells and NK cells in small-intestinal MLNs compared to conventional mice (Figure S4).

| Microbial exposure with live bacteria impacts the immune phenotype at mucosal effector sites while a germ-free immunophenotype is preserved after filtrate-feeding
Next, we assessed microbial and potential immunomodulatory effects of colonization with live bacteria or FFT feeding on the selected immune cell types at mucosal effector sites represented by intestinal epithelial lymphocytes (IEL) and lamina propria lymphocytes (LPL) (Figure 4).The B-cell proportions in the small-intestinal (SI) LPL compartment among CONV and GF/CMT mice were not significantly different from the three GF groups (Figure 4B), while the T-cell proportions among CONV and GF/CMT mice were decreased when compared to the GF-NaCl and GF-FFT groups (Figure 4C).Exposure to live microbes at birth or within the first post-natal days increased SI LPL CD4 + T cells when compared to mice raised GF, while CD8 + T cells were reduced when compared to the GF-NaCl and GF-FFT groups (Figure 4D, E).Filtrate-feeding was unable to reverse SI LPL accumulation of NKT cells in GF mice (Figure 4F) with a known dependency on microbial exposure during early life. 19Similarly, the levels of CD8αα T cells in SI LPL remained low after FFT treatment (Figure 4G), although our flow cytometry results showed an increase in proportional CD8αα T cells among untreated GF controls.Exposure of antigenic contents in the distal segment of the small intestines comprising a blend of dietary antigens and live commensal microbes in a conventional environment 20 thus shape a defined CONV mucosal immunophenotype differing from the GF phenotype.Early conventionalized mice (GF/CMT) adapted to this overall conventional immunophenotype deviating from a GF phenotype within 4 weeks post-conventionalization, while GF mice undergoing filtrate treatment (GF-FFT) retained a preserved GF phenotype like GF-NaCl and GF mice.Absolute cell counts of lamina propria lymphocyte subsets in germ-free mice were not significantly

| Impact of microbial metabolites and by-products beyond the mucosal immune system
Lastly, we investigated immunomodulatory effects of bacterial colonization and exposure of microbial metabolites beyond the mucosal induction and effector sites.The liver, as a gatekeeper for sensing and responding to gut-derived metabolites and microbial fermentation products transported through the portal vein, and the spleen representing a secondary lymphoid organ and systemic immunity, displayed an unambiguous although less distinct conventional and germ-free immunophenotype (Figure 5) in line with the immunologic features observed at mucosal inductive and effector sites.In the liver, the faecal filtrate did not induce immunologic shifts of immune cell subsets associated with immune tolerance and responses to commensal microbes as demonstrated by CD8αα T cells, ILC1 or FoxP3 + CD4 + T cells (Figure 5C-E).Similarly, conventional NK cells among GF-FFT and GF-NaCl were significantly reduced from CONV levels, but this was not reflected in the untreated GF controls (Figure 5F).Spleen subsets like FoxP3 + CD4 + T cells (Figure 5G) and cNK cells (Figure 5H) displayed corresponding trends as observed in liver.In the spleen, the percentages of cNK cells among GF-FFT and GF-NaCl mice were reduced compared to GF untreated and colonized mice (Figure 5H).Absolute cell counts of lymphocytes in spleen and liver were similar between conventionalized and GF mice (Figure S3D, E).

| DISCUSSION
Microbes, microbial-derived bioactive compounds and metabolites are pivotal in mucosal immune homeostasis and regulation. 1 Our primary objective was to investigate the immunomodulatory potential of a bacteria-free faecal lysate of bacteria-derived by-products and potentially bioactive molecules like lipids or peptides serving as a postbiotic mixture.
Our study demonstrate that a bacteria-free faecal filtrate does not alter immune cell composition contrasting the evident immunomodulatory effects of live bacteria, despite the previously reported clinical effects of a similar sterile-filtered faecal filtrate in rCDI in human 5 and necrotizing enterocolitis in mice. 6Exposure to live microbes is needed to restore 'conventional' mucosal immunity in GF mice.
Daily filtrate administration demonstrated decreased host survival among GF-FFT pups appearing immediately after administration and manifested as aspiration and sudden respiratory distress with fatal outcome.Saline-feeding (GF-NaCl) performed in parallel with filtrate treatment was well tolerated, suggesting that filtrate-related mortality may represent a biologic or toxic effect rather than a technical issue.All treated pups were handled similarly and dosed in a randomized alternating dosing order.A possible explanation of the significant difference in survival could be an acute inflammatory response deteriorated by repeated oral lipopolysaccharide challenge 21 although simple suffocation as a cause of death for some of the pups cannot be excluded.
The increased mortality observed in the mice receiving FFT did not translate into changes in the immune compartments as a GF immunophenotype was preserved following neonatal filtrate-feeding.We did not observe immune effects indicative of either local or systemic biologic effects of the sterile faecal filtrate after daily administration for 25 days.
Explanations of clinical effects and mechanisms of action of a sterile faecal filtrate in previous studies remain elusive. 5,6Postulated underlying mechanisms suggest that postbiotics can modulate resident microbiota by interaction with bioactive compounds as outer-membrane vesicles or bacteria and bacteriophages already present in the receiving gut, 22 restoring beneficial microbes and depleting unfavourable strains by antimicrobial actions. 23econd, postbiotics potentially enhance gut epithelial barrier function by promoting tight junctions mediated by postbiotics through signalling pathways not completely understood. 23Host-microbe interactions facilitated by transfer of microbial-derived molecules, bacterial fermentation products and metabolites in a natural habitat with live microbes like segmented filamentous bacteria adhering to the mucosal epithelial layer 24 do not necessarily translate into similar effects under GF conditions if adhesins as fimbria and bacterial lectins facilitating hostmicrobe interaction are disintegrated during filtrate-processing.Both the faecal slurry and sterile faecal filtrate used in our experiments were of murine origin due to previous reported species-specific microbial gut signatures. 25hird, in vitro cultivation of murine enterocyte organoids together with a sterile stool-supernatant has demonstrated enhanced enterocyte proliferation, 26 and a sterile faecal lysate disrupted colonic epithelial barrier integrity in the IL-10 −/− colitis model 27 implying potential effects on gut barrier function beyond immune-priming.This does not necessarily translate into in vivo approaches as maternal factors in the neonatal gut as milk-derived substances and maternal antibodies orchestrate the host-microbiota interplay 28 and shape the microbiota during early life, 29 and may have influenced the bioactive potential of our filtrate.Additionally, although the faecal filtrate was 'sterile-filtered', a viral component of bacteriophages might be a constituent of our bacteria-free FFT, as flow virometry indicated presence most likely of viral DNA in the faecal filtrate.A shortcoming of our study is the limited profiling of the virome in the faecal filtrate.However, under GF conditions, the potentially dynamic microbiota-modulatory effects prompted by bacteriophages targeting specific bacterial strains are lost. 30Bacteriophages can modulate intestinal permeability and increase serum levels of inflammatory cytokines, 31 but most reported immune effects are through phage-bacterial interactions, which are then lost in the GF state. 32ur findings show that a FFT in a GF environment did not have the potency to induce overt immune responses, albeit we did not characterize the immune profile at the transcriptome level.This opens an opportunity for future studies as transcriptome profiling of intestinal epithelial cells has previously demonstrated temporal and dynamic modulation by the microbiota of genes related to basic functions, immunity and metabolism. 33he established FMT therapy in rCDI entails considerable safety concerns despite a high treatment recovery rate, 4 thus promoting surveys of other opportunities as bacteriophage therapy 5,30 and other microbiome-based therapeutics. 2FFT seems to have an acute bioactive effect although we were unable to detect long-term changes of investigated immune cell proportions, yet we did not examine immune cell activation status or immune cell functions.However, we present an experimental approach that may serve as a framework for future studies of microbiota-targeted therapeutics and its implications on gut microbiome composition, metabolic signalling pathways and host immunity.Implementation of alternative and tailored microbiota-targeting therapies in the treatment of both recurrent C. difficile and non-communicable diseases associated with alterations in the microbiome holds a potential therapeutic opportunity but warrants further mechanistic investigations of the host-microbial interplay.
restoration of intestinal natural killer T (NKT) cells with characteristics similar to conventional pups.Our findings show that oral administration of a FFT made of microbial metabolites, antigens and bacteriophages alone is insufficient to induce normal immune development elicited by the presence of live bacteria.Reduced survival during neonatal FFT treatment suggests a potential bioactive attribute of sterile faecal filtrates.Supplies, Alfreton, Great Britain).GF mice were housed in open cages (Eurostandard type II, 11bbB, Tecniplast, Buguggiate, Italy) maintained in sterile flexible-film isolators.Conventional and conventionalized mice were housed in individually ventilated cages (IVC, GM500, Tecniplast).All animal experiments were approved by the Norwegian National Animal Research Authority (project licence no.11245/25770) and carried out in full accordance with the EU Directive on Protection of animals used for scientific purposes (2010/63/EU) and the Norwegian Animal research legislation.

F I G U R E 1
Experimental design.(A) Orogastric administered feeding of germ-free (GF) mice with either saline (GF-NaCl) or sterilefiltered faecal filtrate (GF-FFT), and GF mice conventionalized with caecal microbiota (GF/CMT) at PND (post-natal day) 4-5.Organ sampling at PND 30 for flow cytometry immunophenotyping.(B) Detailed timeline showing orogastric treatments, the different conditions at birth (GF, CONV) and the experimental groups (GF, GF-NaCl, GF-FFT, GF/CMT and CONV).(C) Survival plot of neonatal mice treated with sterile-filtered faecal filtrate or saline, starting at post-natal day (PND) 5 and continuing to PND 30 ± 3. The number at risk is shown for the days elapsed after birth when casualties occurred.(D) Body weight at termination.(E) Spleen weight as percentage of total body weight, and (F) caecum weight as percentage of total body weight.CONV n = 10, GF/CMT n = 10, GF-FFT n = 12, GF-NaCl n = 16 and GF n = 6.Survival was visualized by Kaplan-Meier survival curves and distributions compared using the log-rank test (C).Statistical significance was evaluated using one-way ANOVA with Bonferroni correction for multiple comparisons (D-F).***P < .001,****P < .0001.ns, not significant.Illustrations generated with BioRender.

F
I G U R E 2 B cells, T cells and NK cells at mucosal inductive sites.(A) Heatmap displaying log2 fold change of relative proportions of immune cell subtypes in Peyer's patches, small-intestinal mesenteric lymph nodes and colon mesenteric lymph nodes compared to the mean of the CONV group.Each square represents an individual mouse (CONV: n = 10, GF/CMT: n = 10, GF-FFT: n = 10, GF-NaCl: n = 13, GF: n = 6).Log2 fold change values are displayed on a scale from −2.5 (blue) to 2.5 (red).Excluded values are marked with 'X'.Significant comparisons of GF/CMT and GF-FFT are marked with ' † ' and asterisk denoting level of significance.Relative proportions of (B) B cells, (C) T cells and (D) TCRγδ T cells in Peyer's patches and (E) B cells, (F) T cells and (G) conventional NK cells in small-intestinal MLNs (SI MLNs).Statistical significance was evaluated using Student's t test (A), regular one-way ANOVA with Bonferroni correction for multiple comparisons or Welch ANOVA with Dunnett's T3 multiple comparisons test (B-G).*P < .05,**P < .01,***P < .001,****P < .0001.C MLN, colon mesenteric lymph nodes; ns, not significant; SI MLN, small-intestinal mesenteric lymph nodes.compared to conventional and colonized mice except for the CD4 + T cells (Figure S4G-K).