The B. subtilis strain employed in this study was the ABP1 strain, previously isolated from European seabass (Dicentrarchus labrax) gut and defined as a probiotic (50). The ABP1 strain was grown in 500 ml of Luria-Bertani (LB) broth (Difco) at 25°C in an orbital shaker (100 rpm) for approximately 18 h, until the bacterial culture reached an exponential growth phase as determined by the optical density at 600 nm. EVs were then isolated by a physical separation protocol reported before (51), with minor modifications. Briefly, bacterial cells grown overnight were pelleted by centrifugation (10,000 x g for 20 min at 4°C). The supernatants were filtered through a 0.22 µm-pore-size filter (Millipore) and then concentrated by centrifugation on a Centricon Plus-70 filter Device (Millipore), followed by an additional filtration step. At this point, EVs were collected by centrifugation at 100,000 x g for 1 h at 4°C, being this the only step that was different to the previously described protocol (in which this centrifugation was performed at 150,000 x g). The pellets were resuspended in 75 μl of sterile and filtered phosphate-buffered saline (PBS) and stored at -80°C until use. In parallel, as a negative control, an LB medium without bacteria was subjected to the same process, to discard the possibility of lipids in the culture media forming vesicles. The sterility of all samples was assessed on LB plates.

In addition, the immunomodulatory properties of B. subtilis EVs were analyzed and compared with another strain of Bacillus species, namely B. megaterium (52). This species was grown in the same conditions and EVs were isolated in parallel, following the methodology just described.

In all cases, isolated EVs were characterized using different techniques. First, the protein concentration in the obtained preparations was determined using the Micro BCA™ Protein Assay Kit (Thermo Scientific). Thereafter, 18 μg of protein were mixed with 2X loading buffer [0.125 M Tris-HCl, 10% glycerol, 10% 2-mercaptoethanol, 100 mM dithiothreitol (DTT), 4% sodium dodecyl sulfate (SDS) and 0.05% bromophenol blue], boiled for 5 min and loaded on a 15% acrylamide gel containing 6 M of urea. After electrophoresis, the gel was stained with Coomassie blue. The size distribution and concentration of the EVs were analyzed by Dynamic Light Scattering (DLS) using the Zetasizer Ultra (Malvern analytical) device. DLS analysis allows quantitative and high-throughput characterization of vesicles in a low sample volume (53). The Zetasizer Ultra is equipped with a 633 nm He–Ne laser, operates at an angle of 173° and follows a mixed-mode measurement phase analysis light scattering (M3-PALS). For this, EVs were diluted 1,000-fold in PBS and loaded in a polystyrene cuvette (SARSTEDT). For each sample, three measurements were performed with standard settings (Liposomes; fixed position with an automatic attenuator; temperature of 25°C) and eventually averaged. The Zetasizer software reported intensity, size and concentration (particles/ml).

Finally, in the case of B. subtilis, the morphology of the isolated EVs was observed by transmission electron microscopy. For this, EVs were negatively stained with 2% uranyl acetate before visualization in a JEOL JEM 1400 Flash Transmission Electron Microscope (TEM) (Japan) operating at 100 kV. Images were recorded with an OneView CMOS digital camera (Gatan, USA).

The intestinal epithelial cell line RTgutGC (Rainbow Trout gut Guelph Canada) has been isolated from the distal portion of the intestine of a rainbow trout (Oncorhynchus mykiss) (54). The cell line was cultured in Leibovitz’s medium (L-15, Gibco) supplemented with 100 I.U./ml penicillin and 100 μg/ml streptomycin (1% P/S, Life Technologies) and 10% fetal calf serum (FCS, Gibco) in T75 flasks at 19°C. The day before the stimulation, confluent cells were washed with medium, detached using trypsin (Gibco) and cell viability was determined using trypan blue (Sigma). Cells were then adjusted to 4 x 10⁶ cell/ml in complete L-15 medium (supplemented with antibiotics and 10% FCS). Subsequently, cells were seeded into 24-well plates (1 ml per well) and incubated for 24 h at 19°C. At this point, cells were stimulated with B. subtilis EVs to reach final concentrations of 3 x 10⁵, 1.5 x 10⁶ and 3 x 10⁶ particles/ml in the wells. As negative controls, cells were exposed to the same volume of the control preparation (LB subjected to the same process of EV isolation). Six replicate wells were included in all cases. After stimulation, cells were incubated at 19°C for 24, 48 and 72 h. At these points, RNA was extracted from the cells to determine the transcriptional response.

In some experiments, the effects of B. megaterium EVs on RT-gutGC cells were compared to those of B. subtilis EVs. For this, RTgutGC cells disposed in 24-well plates as described above were stimulated with 1.5 x 10⁶ particles/ml. Negative control cells were exposed to the same volume of the control preparation. In all cases, five replicate wells were included. In this case, cells were incubated at 19°C for 48 h before RNA extraction.

Rainbow trout (Oncorhynchus mykiss) of approximately 400 g were obtained from Piscifactoria Cifuentes (Cifuentes, Guadalajara, Spain). Fish were maintained at the animal facilities of the Animal Health Research Centre (CISA-INIA-CSIC) in a re-circulating water system at 14°C with a photoperiod of 12:12 h light/dark and were fed twice a day with a commercial diet (Skretting Spain S.A). Animals were acclimatized for two weeks before they were used for experimentation. During this time, they were regularly checked for clinical signs, which were never observed.

Rainbow trout were sacrificed by benzocaine (Sigma Aldrich) overdose (50 mg/l) and peripheral blood was extracted from the caudal vein. The spleen was then collected and used to isolate total leukocyte populations. Hence, cell suspensions were obtained by passing the spleen through a 100 µm nylon mesh (BD Biosciences) using L-15 containing 1% P/S, 10 U/ml heparin (Sigma- Aldrich) and 2% FCS. Cell suspensions were then placed onto 30/51% Percoll (GE Healthcare) density gradients and centrifuged at 500 x g for 30 min at 4°C, without brake. The interface cells were collected, washed with L-15 supplemented with antibiotics and 2% FCS and resuspended in L-15 medium containing P/S and 5% FCS. The viable cell concentration was then determined by Trypan blue (Sigma-Aldrich) exclusion and cells were adjusted to a concentration of 2 x 10⁶ cells/ml. Depending on the experiment, splenic leukocytes were seeded in 24-well plates (Nunc) at a concentration of 2 x 10⁶ cells per well (for transcriptional studies) or in 96-well plates at 5 x 10⁴ cells per well (to study effects on IgM⁺ B cells). In all cases, splenocytes were exposed to EVs to make final concentrations of 3 x 10⁵, 1.5 x 10⁶, and 3 x 10⁶ particles/ml in the wells. Negative controls in which cells were exposed to the same volume of the control preparation were always included (LB subjected to the same process of EV isolation). Splenocytes were then incubated at 19°C for 24 h (for transcriptional studies) or 72 h (to study effects on IgM⁺ B cells).

RTgutGC cells and splenic leukocytes exposed or not to EVs were collected in 1 ml of TRI Reagent solution (Invitrogen), after discarding the culture supernatants. Total RNA isolation was performed following the manufacturer´s specifications. RNA quantification was performed with a NanoDrop 1000 Spectrophotometer (Thermo Fisher Scientific). One μg of RNA was then treated with DNase I (Invitrogen) and used to synthesize cDNA using the RevertAid Reverse Transcriptase (Thermo Fisher Scientific) with oligo (dT)₂₃VN, following the manufacturer’s instructions.

Real-time PCR analyses were performed with the LightCycler96 System instrument (Roche) using FastStart Essential DNA Green Master reagents (Roche) and specific primers previously described ( Supplementary Table S1 ). Each sample was incubated for 10 min at 95°C, followed by 40 amplification cycles (10 s at 95°C, 10 s at 60°C and 10 s at 72°C). A dissociation curve was obtained by reading fluorescence every degree between 60°C and 95°C to ensure only a single product had been amplified. The relative expression levels of the genes were normalized to the expression of b-actin, as a reference control gene. This reference gene was selected among two candidate genes after verifying that no statistical differences were detected among b-actin Ct values from different samples, following the MIQE guidelines (55). Expression levels were calculated using the 2^-ΔCt method, where ΔCt is determined by subtracting the b-actin value from the target cycle threshold. Negative controls with no template and minus reverse transcriptase controls were included in all cases.

We first evaluated how B. subtilis EVs affected the presence of different B cells in splenocyte cultures by flow cytometry. For this, splenic leukocytes incubated for 72 h with different doses of EVs or left unstimulated in the same conditions were harvested, washed with staining buffer (phenol red-free L-15 medium supplemented with 2% FCS) and then incubated for 1 h at 4°C with specific monoclonal antibodies (mAbs), anti-trout IgM (1.14) [mAb mouse IgG1 coupled to R-phycoerythrin (R-PE), 1 μg/ml] (56) and anti-trout IgD [mAb mouse IgG1 coupled to allophycocyanin (APC), 5 μg/ml] (57) diluted in staining buffer. Negative controls in which cells were exposed to the same volume of the control preparation were also included (LB subjected to the same process of EV isolation). After the incubation, cells were washed twice with staining buffer and counterstained with 0.2 μg/ml 4′, 6-diamidino-2-phenylindole (DAPI, Sigma) to determine cell viability. Only live cells were included in the analysis, which was performed on a FACS Celesta^™ flow cytometer (BD Biosciences) equipped with FACSDiva software. In all cases, cells incubated with R-PE or APC-conjugated mouse IgG1 isotypes (clone MOPC-21, Biolegend) were used as controls, to confirm the specificity of the mAbs used. Doublets and dead cells were excluded from the flow cytometry analysis following the gating strategy described in Supplementary Figure S1 . Data analysis was performed with FlowJo^® v.10 (Tree Star).

B cells are professional antigen-presenting cells that express major histocompatibility complex (MHC) II, and possess all the necessary machinery for antigen uptake, processing and presentation (58). Therefore, we also determined how EVs affected the levels of expression of surface MHC II on IgM⁺IgD⁺ B cells. For this, in some experiments, anti-MHC II β-chain ([mAb mouse IgG1 coupled to fluorescein isothiocyanate (FITC), 2 μg/ml] (59) was added in combination with the anti-IgM and anti-IgD mAbs. All the staining process was then performed as described above.

The antigen-processing capacity of IgM⁺ B cells was determined using the EnzChek protease assay kit (Invitrogen). For that, splenic leukocytes incubated with different doses of EVs or left unstimulated for 72 h as described above, were incubated with green fluorescent BODIPY DQ-CASEIN at 5 μg/ml for 1 h. DQ-CASEIN is a self-quenched form of fluorescently labelled casein (60), commonly used to study protease-mediated antigen processing because it exhibits bright green fluorescence upon proteolytic processing due to the released dye molecules (61). Negative controls in which cells were exposed to the same volume of the control preparation were also included (LB subjected to the same process of EV isolation). Afterwards, the cells were washed with staining buffer three times and labelled with the anti-trout IgM (clone 1.14) mAb coupled to APC (1 μg/ml) for 30 min at 4°C, rewashed and analyzed by flow cytometry as described above.

The Click-IT^® EdU Alexa Fluor^® 488 flow cytometry assay kit (Life Technologies) was used to determine whether B. subtilis EVs could induce the proliferation of IgM⁺ B cells, following the manufacturer’s instructions. For this, splenic leukocytes were incubated with different doses of EVs or left unstimulated in the same conditions in 96-well plates for 72 h at 19°C. 24 h prior to analysis, 1 μM EdU (5-ethynyl-2′ -deoxyuridine) was added to the cultures and incubated until next day. Negative controls in which cells were exposed to the same volume of the control preparation were also included (LB subjected to the same process of EV isolation). The day after, the cells were collected, and their viability determined before fixation and permeabilization using the LIVE/DEAD^™ Fixable Near-IR Dead Cell Stain kit (Invitrogen) for 30 min, following the manufacturer´s specifications. Cells were then washed and stained with the anti-trout IgM (1.14) mAb coupled to R-PE (1 μg/ml) for 30 min at 4°C. Cells were then fixed, permeabilized, and incubated with specific reagents to detect the incorporation of EdU into the DNA of proliferating cells following the manufacturer’s instructions. Samples were then analyzed on the flow cytometer as described above.

ELISpot was used to evaluate the effect of the B. subtilis EVs on the total number of IgM-secreting B cells in splenic leukocyte cultures. For this, ELISpot plates containing Inmobilon-P membranes (Millipore) were activated with 70% ethanol for 30 s, coated with an anti-trout IgM mAb at 2 μg/ml in phosphate buffer saline (PBS) and incubated overnight at 4°C in agitation. To block non-specific binding to the membrane, plates were then incubated with 2% bovine serum albumin (BSA) in PBS for 2 h at room temperature (RT). Splenic leukocyte suspensions of individual fish that had been stimulated with different doses of B. subtilis EVs for 72 h at 19°C, with 1.5 x 10⁶ particles/ml of EVs from either B. megaterium or B. subtilis or left unstimulated in the same conditions, were then added to the wells in duplicate at a concentration of 1 x 10⁵ cells per well. Negative controls in which cells were exposed to the same volume of the control preparation were also included (LB subjected to the same process of EV isolation). After 24 h of incubation at 19°C, cells were washed away five times with PBS and plates were blocked again with 2% BSA in PBS for 1 h at RT. After blocking, biotinylated anti-IgM mAb at 1 μg/ml was added to the each well and incubated for 1 h at RT in agitation. Following additional washing steps (five times in PBS), the plates were developed using streptavidin-HRP (Thermo Scientific) at RT for 1 h, washed again with PBS and incubated with 3-amino-9-ethylcarbazole (Sigma Aldrich) for 30 min at RT in the dark. The substrate reaction was stopped by washing the plates with water. Once the membranes had dried, they were digitally scanned and the number of spots in each well determined using an AID iSpot Reader System (Autoimmun Diagnostika GMBH).

Data handling, statistical analyses and graphic representation were carried out using GraphPad Prism version 8 (GraphPad Software). Statistical analyses were performed using a two-tailed paired Student’s t-test. The differences between the mean values were considered significant on different degrees, where ∗ means p ≤ 0.05, ∗∗ means p ≤ 0.01, and ∗∗∗ means p ≤ 0.001. Furthermore, significant differences between mean values were denoted by letters in some experiments, with lowercase letters indicating differences among groups (p < 0.05).

We used different complementary approaches to confirm the purification of B. subtilis EVs and determine their integrity, size, shape, and concentration. When we observed the protein content by SDS-PAGE gel, a banding pattern was obtained that was not present in the control media subjected to the same purification process ( Figure 1A ). The size distribution of vesicles was confirmed by DLS measurements ( Figure 1B ), revealing a single peak size ranging from 90 to 100 nm in diameter, which provided an indication of the high purity of the EV preparation obtained. Moreover, EVs displayed the typical “round-shaped” morphology when visualized by TEM, with a diameter of up to 100 nm ( Figure 1C ). These structures were not visualized under the TEM or detected by DLS measurements when the culture media was subjected to the same purification protocol, confirming that they were produced by the bacteria. The typical yield obtained in the EV purification process was between 7x10¹¹ to 1x10¹² particles for each 500 ml of starting bacterial culture.

We first studied how EVs affected the levels of transcription of a range of immune genes on the RTgutGC intestinal epithelial cell line. These included several genes coding for pro-inflammatory cytokines such as tumor necrosis factor α (tnfa), interleukin 1β (il1b) and il8 (62). We also studied the transcriptional response of genes coding for antimicrobial peptides (AMPs), such as hepcidin and cathelicidin 2. Finally, we evaluated genes involved in intestinal barrier integrity and homeostasis, including genes involved in the intercellular tight junctions [e-cadherin (cdh1), claudin 3 (claud3), zonula occludens (zo1)] and a gene responsible for intestinal mucin production (imuc) (54, 55, 63, 64).

The levels of transcription of il1b were significantly up-regulated in RTgutGC cells after 24, 48 and 72 h of stimulation with the different doses of EVs ( Figure 2 ). Likewise, il8 transcription augmented with the different doses of EVs, but only after 24 and 48 h of stimulation. On the other hand, the expression levels of tnfa remained stable throughout the experiment. Concerning the antimicrobial peptides, B. subtilis EVs up-regulated the levels of transcription of both hepcidin and cathelicidin 2 ( Figure 2 ). In the case of cathelicidin 2, this up-regulation was observed in response to the three EV doses after 24 h and 48 h, whereas the up-regulation was only significantly achieved in response to specific EV doses for hepcidin. Lastly, the expression levels of claud3 and zo1 were up-regulated with some EV doses, mostly after 72 h of stimulation, whereas after 48 h of incubation, EVs provoked the down-regulation of zo1 transcription ( Figure 2 ). Similarly, the transcription of cdh1 significantly decreased in response to the lower EV doses at the later time point ( Figure 2 ). Finally, imuc was significantly up-regulated at all sampling times in response to some EV doses ( Figure 2 ).

We then determined whether the B. subtilis EVs affected the transcriptional response of total splenic leukocyte cultures stimulated for 24 h with different concentrations of EVs. In this case, we studied the transcription of tnfa, il1b and il8, as well as il10, a gene mostly catalogued as anti-inflammatory (65) and the two AMPs tested before. Furthermore, in this case, we included CD4, CD8, FoxP3, GATA3 and Tbet as markers of different T cell populations, all forms of Ig genes and several genes known to be up-regulated during B cell differentiation, including irf4 and the four isoforms of the prdm1 genes coding for Blimp1 that are present in rainbow trout (66).

We observed that mRNA levels of tnfa, il1b and il8 were significantly up-regulated in splenic leukocytes exposed to any of the EV doses tested ( Figure 3 ). Additionally, both hepcidin and cathelicidin 2 were also transcriptionally up-regulated in response to B. subtilis EVs at some doses ( Figure 3 ). On the other hand, many genes related to B cell differentiation were transcriptionally increased in response to the highest EV dose, including irf4, prdm1a-1, prdm1a-2, prdm1b-1, and prdm1b-2 ( Figure 3 ). Genes related to T cell function were not significantly regulated in the presence of the EVs ( Figure 3 ).

Having seen an important transcriptional effect of B. subtilis EVs on genes related to B cell function in splenocyte cultures and given that B. subtilis had been shown to modulate the functionality of B cells in rainbow trout (49), we decided to investigate if B. subtilis EVs were capable of exerting similar effects. We first determined how the presence of EVs in the culture affected the percentages of IgM⁺IgD⁺, IgM⁺IgD^-, and IgD⁺IgM^- B cell subsets. As previously reported, in rainbow trout spleen, IgM⁺IgD⁺ B cells make up for the vast majority of the IgM/D B cell population (52) and it was precisely this population, the one that significantly increased in the cultures in response to EV exposure ( Figure 4 ). A small percentage of B cells in these cultures corresponded to IgM⁺IgD^- and IgD⁺IgM^- B cells, yet the survival or viability of these populations did not seem to be affected by B. subtilis EVs ( Figure 4 ).

We also investigated the effects of exposure to B. subtilis EVs on the antigen presenting capacities of B cells. For this, we first studied the levels of MHC II on the surface of rainbow trout splenic cells. Our results clearly show that EVs significantly increased the levels of surface MHC II on IgM⁺IgD⁺ B cells from the spleen at all tested concentrations ( Figure 5A ). Additionally, the antigen-processing capacity of splenic IgM⁺ B cells was also significantly higher in cultures exposed to the highest concentration of EVs ( Figure 5B ).

The fact that B. subtilis EVs up-regulated many genes related to B cell differentiation in splenic cultures suggested an effect on the number of IgM-secreting cells. To test this hypothesis, we evaluated the effects of EVs on the number of IgM-secreting in splenic leukocyte cultures by ELISpot. The number of IgM-secreting cells significantly augmented in splenic leukocyte cultures incubated with 1.5x10⁶ and 3x10⁶ EVs per ml when compared to the number of IgM-secreting cells found in non-stimulated cultures ( Figure 6 ). This increase was not significant in splenic leukocyte cultures stimulated with the lowest dose of EVs.

To establish the lymphoproliferative potential of EVs derived from B. subtilis, we next studied the effects of EVs on the proliferation of IgM⁺ B cells in splenocyte cultures. Our results demonstrated that EVs were capable of provoking a slight but significant proliferation of IgM⁺ B cells in a dose-dependent fashion ( Figure 7 ). Interestingly, the highest EV dose also provoked a significant increase in the number of proliferating IgM^- cells in these cultures ( Figure 7 ).

To establish if the immunomodulatory effects provoked by B. subtilis EVs are specific to EVs from this probiotic species, we performed a series of experiments in which their effects were compared to those of EVs isolated from B. megaterium, another probiotic bacterium (67). We first evaluated their transcriptional effects on RTgutGC cells after 48 h, using the intermediate EV dose (1.5 x 10⁶ particles/ml) and studying the same genes studied before. The response to the two EVs was different for many genes. As seen before, B. subtilis EVs significantly increased the transcription of il1b and cathelicidin 2, while those of B. megaterium EVs significantly decreased them (il1b) or did not significantly affect them (cathelicidin 2) ( Figure 8A ). On the other hand, the transcription of zo1 gene was negatively regulated by B. subtilis EVs, as described above, but not by B. megaterium EVs ( Figure 8A ). Moreover, claud3 mRNA levels were significantly increased in response to B. megaterium EVs but not in response to B. subtilis EVs ( Figure 8A ). On the other hand, both types of EVs were able to stimulate significantly the transcription of cytokine il8 and imuc ( Figure 8A ).

Finally, we compared the effects that a single dose of both types of EVs had on the number of IgM-secreting cells in splenic leukocyte cultures. As seen in the previous experiments ( Figure 6 ), B. subtilis EVs provoked a significant up-regulation of the number of IgM-secreting cells when compared to non-stimulated cells ( Figure 8B ). However, this effect was not seen in response to B. megaterium EVs which even provoked a significant decrease in the number of IgM-secreting cells in the cultures ( Figure 8B ).