Extracellular Membrane Vesicles from Lactobacilli Dampen IFN-γ Responses in a Monocyte-Dependent Manner

Secreted factors derived from Lactobacillus are able to dampen pro-inflammatory cytokine responses. Still, the nature of these components and the underlying mechanisms remain elusive. Here, we aimed to identify the components and the mechanism involved in the Lactobacillus-mediated modulation of immune cell activation. PBMC were stimulated in the presence of the cell free supernatants (CFS) of cultured Lactobacillus rhamnosus GG and Lactobacillus reuteri DSM 17938, followed by evaluation of cytokine responses. We show that lactobacilli-CFS effectively dampen induced IFN-γ and IL-17A responses from T- and NK cells in a monocyte dependent manner by a soluble factor. A proteomic array analysis highlighted Lactobacillus-induced IL-1 receptor antagonist (ra) as a potential candidate responsible for the IFN-γ dampening activity. Indeed, addition of recombinant IL-1ra to stimulated PBMC resulted in reduced IFN-γ production. Further characterization of the lactobacilli-CFS revealed the presence of extracellular membrane vesicles with a similar immune regulatory activity to that observed with the lactobacilli-CFS. In conclusion, we have shown that lactobacilli produce extracellular MVs, which are able to dampen pro-inflammatory cytokine responses in a monocyte-dependent manner.

www.nature.com/scientificreports www.nature.com/scientificreports/ Cell enrichment and depletion. Enriched T cells were generated from PBMC through negative selection using EasySep TM Human T cell enrichment kit. To generate monocyte-depleted PBMC or B cell-depleted PBMC we used positive selection with EasySep TM Human CD14 Positive Selection Kit II or EasySep TM Human CD19 Positive Selection Kit II, respectively (all from STEMCELL Technologies Inc.). In brief, frozen PBMC were thawed, washed and resuspended in PBS containing FCS 2% and EDTA 1 mM at 50 × 10 6 cells/ml. PBMC were incubated with an antibody cocktail, followed by magnetic particles and target cells were depleted or recovered using an EasySep magnet according to manufacturer's instructions. Cell purity after depletion or enrichment was confirmed by flow cytometric analyses.
Generation of monocyte-conditioned medium. Isolated monocytes were seeded in a 48-well culture plate at 1 × 10 6 cells/ml and treated for 6 h with 10% LGG-CFS. Next, the monocytes were washed twice with cell culture medium and incubated in fresh cell culture medium for 14 h at 37 °C with 5% CO 2 . Finally, the monocyte-conditioned medium was collected by centrifugation and stored at 4 °C until later use.
Proteome profiler array. The monocyte-conditioned medium was analysed for 105 secreted proteins using the Human XL Cytokine Array Kit (R&D Systems) according to manufacturer's instructions. In brief, the stamped identification number on each membrane was cut off and replaced with an individual mark using a pencil. Each membrane was then blocked, washed and incubated overnight at 4 °C on a rocking platform with 1.5 ml of monocyte-conditioned medium diluted 1:3 using the supplied array buffer. Membranes were washed and incubated with the detection antibody cocktail for 1 h at room temperature (RT) on a rocking platform followed by addition of IRDye 800 CW Streptavidin (LI-COR Biosciences) for 30 min. Finally, the membranes were scanned and images acquired using the LI-COR Odyssey Infrared Imaging System (LI-COR Biosciences).
Lipid removal assay. The commercial kit Cleanascite lipid adsorption and clarification reagent (Biotech Support Group) was used to remove lipids and extracellular membrane vesicles from the Lactobacillus-CFS according to manufacturer's instructions. In brief, one part of Cleanascite reagent was added to four parts of Lactobacillus-CFS and incubated at RT for 40 min under periodical mixing. Finally, the Cleanascite reagent was separated from the CFS by centrifugation at 15000 × g for 1 min after which the supernatant was collected and subjected to fractionation using an Amicon Ultra-4 centrifugal filter unit (Merck Millipore) with a MwCO of 100 kDa.
Proteinase K digestion and heat treatment. The Lactobacillus-CFS was subjected to protein digestion using Proteinase K immobilized to agarose (Sigma-Aldrich) using a modified protocol of that described elsewhere 42 . Proteinase K-agarose was added to Lactobacillus-CFS to a working concentration of 1 mg/ml and incubated with rotation at 37 °C for 1 h. Next, the Proteinase K-agarose was removed by centrifugation at 15000 × g for 1 min. Finally, the proteinase K digested CFS was fractionated using an Amicon Ultra-4 centrifugal filter unit (Merck Millipore) with a MwCO of 100 kDa. Fractionated control-CFS was also subjected to heat treatment at 96 °C for 15 min prior to cell stimulations.

Isolation of extracellular membrane vesicles (MVs). Lactobacillus reuteri DSM 17938 bacterial cells
were grown in MRS broth (Oxoid) for 24 h at 37 °C. The bacterial cells were removed from the culture broth by centrifugation at 5,000 × g for 10 min at 4 °C and followed by another centrifugation at 10,000 × g for 10 min at 4 °C. Then the supernatants were filtrated using 0.45 μm pore filter (Millipore). Cell free supernatants were concentrated using Amicon Ultra filter unit with a MwCO of 100 kDa (which remove proteins and other molecules under 100 kDa). The supernatants were loaded on top of 12% sucrose cushion with 50 mM Tris buffer pH 7,2, with the volume ratio 5:1, and centrifuged by Beckman coulter Optima L -80XP ultracentrifuge (Beckman coulter) at 118,000 × g at 4 °C for 3 h. The supernatants were discarded and the pellet was resuspended in PBS buffer and ultra-centrifuged for the second time (118,000 × g at 4 °C for 3 h). The pellets were then dissolved in PBS, aliquoted and stored at −70 °C. Nanoparticle tracking analysis (NTA). The physicochemical characterization of MVs was also investigated by using the Nanoparticle tracking analysis (NTA). Extracellular vesicles were diluted with PBS, and directly tracked using the NanoSight NS300 system (NanoSightTM technology). The analysis was carried out according to the following instrumental set up: (i) a laser beam of 488 nm (blue); (ii) and a high-sensitivity sCMOS camera. Videos were collected and analysed using the NTA software (version 3.2), capturing a video file of the particles moving under Brownian motion. The software tracks many particles individually and using the Stokes-Einstein equation calculates their hydrodynamic diameters. Multiple videos of 90 s duration were recorded generating replicate histograms that were averaged for each sample.
ELISA. Secreted levels of the cytokines IL-1ra (R&D Systems-BioTechne), IL-1β, IL-6, IL-10, IL-17A and IFN-γ (MabTech AB) were measured in cell culture supernatants using sandwich ELISA kits according to the manufacturer's instructions. Absorbance was measured at a wavelength of 405 nm using a micro-plate reader (Molecular Devices Corp.) and results analysed using SoftMax Pro 5.2 rev C (Molecular Devices Corp.).
Statistics. All statistical tests were done using GraphPad Prism (GraphPad Software). All data was considered non-parametric. For Figs. 1-5, the Wilcoxon test was employed. The Friedman test followed by Dunn's multiple comparisons was employed in Fig. 4a. Differences were considered significant when p < 0.05 and the following significance levels were used *P < 0.05; **P < 0.01; ***P < 0.001.

Lactobacilli dampen ifn-γ responses by induction of a monocyte-derived soluble inhibitor.
We have previously shown that cell free supernatants (CFS) from lactobacilli grown in their regular MRS medium dampen IFN-γ and IL-17A responses from T-and NK cells 17,20 . In the present study we cultivated the lactobacilli in modified cell culture medium, a more physiologically relevant medium for immune cells. PBMC cultured in the presence of CFS from L. rhamnosus GG (LGG) or L. reuteri DSM 17938 secreted IL-6 and IL-10 ( Fig. 1a), showing that lactobacilli promote secretion of innate cytokines when grown in cell culture medium. Furthermore, both LGG-and L. reuteri-CFS significantly dampened S. aureus-induced IFN-γ and IL-17A secretion (Fig. 1b), suggesting that lactobacilli have the capacity to promote innate responses while suppressing adaptive inflammatory responses.
Activation of T-and NK cells is regulated by innate APC such as monocytes 43,44 . We therefore investigated whether the lactobacilli-mediated dampening of IFN-γ requires the presence of monocytes or B cells. We depleted PBMC of either monocytes or B cells or isolated pure T cells, prior to stimulation with a T cells specific activator towards CD3/CD28 alone or in combination with LGG-CFS. Depletion of B cells did not affect the IFN-γ-dampening activity while depletion of monocytes completely abolished the LGG-CFS-mediated IFN-γ suppression. In accordance with this, the LGG-CFS was not able to dampen IFN-γ secretion in pure T cell cultures (Fig. 1c). Next, we sought to investigate if the monocyte-dependent dampening of IFN-γ required physical contact between the monocyte and the IFN-γ-producing cell. We isolated and primed monocytes for 6 h with LGG-CFS, thoroughly washed and incubated them in fresh cell culture medium for an additional 14 h. The conditioned medium from the LGG-primed monocytes was subsequently added to stimulated autologous PBMC cultures. We observed a significant reduction in IFN-γ secretion by the monocyte-conditioned medium in a dose-dependent manner (Fig. 1d), suggesting that LGG dampens IFN-γ responses by inducing a monocyte-derived soluble inhibitor.
Lactobacillus-primed monocytes secrete IL-1ra. In order to identify the IFN-γ suppressive monocyte-derived factor, the monocyte-conditioned medium was analysed on a proteomic array. Only twelve of the measured factors were increased in LGG-treated samples compared with unstimulated conditions (Table 1), of which IL-1ra was the candidate molecule that was most likely to exert immune regulatory effects on T cells (Fig. 2a). IL-1ra is an antagonistic inhibitor of the pro-inflammatory IL-1 signalling pathway and it has been shown to reduce IFN-γ secretion 45 . We observed significantly higher levels of IL-1ra in response to S. aureus alone compared to LGG, however, S. aureus also induced significantly more IL-1β. In fact, we found that LGG alone induced a 3-fold higher ratio of IL-1ra/IL-1β than S. aureus alone (Fig. 2b), indicating that LGG favours an anti-inflammatory IL-1 signalling pathway. Finally, supplementation of recombinant IL-1ra to S. aureus stimulated PBMC resulted in reduced IFN-γ levels in a dose-dependent manner (Fig. 2c). Collectively, these data indicate that lactobacilli promote the anti-inflammatory IL-1ra signalling pathway and that this may contribute to their IFN-γ-dampening effect.

Lactobacilli dampen ifn-γ in T cells and NK cells through multiple factors of different molecular size.
Several components of lactobacilli have been shown to be important for host cell interactions, in particular proteins, nucleic acids, PGN and SCFA 26,46,47 . In order to isolate the factors that mediate the IFN-γ dampening activity, we subjected the lactobacilli-CFS to size exclusion fractionation generating multiple fractions spanning from less than 3 kDa to larger than 100 kDa in molecular weight (Mw) and stimulated PBMC in the presence of each isolated fraction and measured secretion of IFN-γ. Two fractions of the LGG-CFS, the 3 kDa (from hereinafter referred to as "low Mw fraction") and the 100 kDa (from hereinafter referred to as "high Mw fraction"), were able to dampen S. aureus-induced IFN-γ production (Fig. 3a). Interestingly, the dampening activity of the L. reuteri-CFS was also confined within the same two fractions of the CFS as for LGG (data not shown), suggesting that the mechanism of cytokine modulation is conserved between both species. Finally, we analysed IFN-γ expression on a cellular level using flow cytometry and found that both the high and the low Mw fractions significantly reduced the frequency of IFN-γ expressing cells within the CD4 + and CD8 + T cell, γδ T cell and NK cell populations (Fig. 3b).
The lactobacilli-CFS high Mw fractions contain a heat sensitive lipid-based factor. During size fractionation, lactobacilli-derived metabolites and small compounds end up in the low Mw fraction of the CFS. We here focused our attention on the cytokine modulating capacity of the high Mw fraction. In order to increase the resolution of the fractionation, the high Mw fraction was further fractionated using high performance liquid chromatography (HPLC). An additional 30 fractions were generated and tested for IFN-γ dampening activity on S. aureus-stimulated PBMC. We identified two consecutive fractions, named J and K, which consistently dampened IFN-γ secretion (Fig. 4a). Furthermore, comparing the protein absorption chromatogram with the ability www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ to dampen IFN-γ, fractions J and K was found to completely match the first protein peak to be eluted from the HPLC column, which indeed was absent in the growth medium control (See Supplementary Fig. S1). Importantly, the protein absorption chromatogram from LGG and L. reuteri looked near identical, further supporting that the IFN-γ-dampening elicited by both species involved the same factor and mechanism. In order to further characterize the isolated factor, we subjected the lactobacilli-CFS to protein digestion using agarose-immobilized proteinase K followed by size fractionation with a 100 kDa filter and assessed the IFN-γ dampening activity of www.nature.com/scientificreports www.nature.com/scientificreports/ the proteinase K treated high Mw fraction on S. aureus-stimulated PBMC. Interestingly, no difference in the ability of LGG or L. reuteri to dampen IFN-γ was observed after proteinase K treatment. Heat inactivation of the L. reuteri-CFS, but not LGG-CFS, prior to stimulations abolished the mediated IFN-γ dampening (Fig. 4b). The lactobacilli-CFS was treated with a lipid adsorption matrix to remove all lipids, including extracellular membrane www.nature.com/scientificreports www.nature.com/scientificreports/ vesicles (MVs) that are enclosed by a lipid membrane, before being added to stimulated PBMC. Indeed, the IFN-γ dampening was lost for both LGG and L. reuteri after lipid removal (Fig. 4c).

Lactobacilli-derived extracellular MVs dampen ifn-γ and IL-17A secretion.
Lactobacilli are known to produce MVs 48 and these bacterial MVs are also know to contain membrane integrated proteins as well as intravesicular proteins that, due to the membrane enclosure, are protected from enzymatic digestion while still susceptible to heat inactivation 49 . We isolated MVs from both species of lactobacilli and a physiochemical characterization of the L. reuteri-MVs revealed a size distribution ranging from 100 nm to 400 nm in diameter, with an average size of 214 nm (Fig. 5a) and contained proteins, DNA and RNA (Fig. 5b). In order to test whether the MVs are responsible for the cytokine-modulatory effect observed with the lactobacilli-high Mw fraction, isolated MVs were added to PBMC at a MV-to-cell ratio of 500:1, 100:1 and 20:1 and incubated for 48 h. The cell culture supernatants were collected and analysed for induction of cytokines using ELISA. The L. reuteri-MVs clearly induced the production of both IL-6 and IL-10 in a concentration dependent manner, while no IFN-γ or IL-17A was detected (Fig. 5c). Moreover, adding isolated MVs to S. aureus-stimulated PBMC significantly dampened www.nature.com/scientificreports www.nature.com/scientificreports/ IFN-γ and IL-17A secretion to a similar extent as the CFS-high Mw fraction (Fig. 5d). Again, a substantial dampening of IFN-γ and IL-17A was also observed using LGG-derived MVs (data not shown).

Discussion
In the current study, we aimed to explore the underlying mechanism behind the cytokine-modulatory effects of lactobacilli. Here we show that lactobacilli produce soluble factors that induce the production and secretion of chemokines (Table 1) and cytokines such as IL-1β, IL-6 and IL-10 from PBMC cultures. We identify extracellular MVs as being one of the main contributors to both the immune-stimulatory and immune-dampening activity of the lactobacilli-CFS, possibly by increasing the relative abundance of IL-1ra to IL-1β by a monocyte-dependent mechanism.
Lactobacilli and its components have been shown to influence T cell differentiation and proliferation in a T cell intrinsic manner. Lactobacillus-derived RNA suppress CD4 + T cell proliferation in the absence of APC 28 and SCFA increase extrathymic generation of T reg cells by promoting acetylation of the foxp3 gene locus within the naïve T cell population itself 34 . In the current study, we also investigated whether the lactobacilli-CFS-mediated dampening of IFN-γ was dependent solely on T cell intrinsic factors or not. Interestingly, IFN-γ dampening was fully dependent on the presence of monocytes since the Lactobacillus-CFS had no effect on IFN-γ production from isolated T cell or monocyte-depleted PBMC cultures (Fig. 1b). Furthermore, we observed a significant level of dampening by using Lactobacillus-primed monocyte-conditioned medium alone, confirming that the dampening mechanism is cell-to-cell contact independent.
Innate cells are known to regulate lymphocyte activation through soluble mediators 50 . IL-10 is a regulatory cytokine with an important role in preventing excessive inflammation by increasing T reg cell function and reducing inflammatory cytokine responses. Lactobacilli-induced IL-10 is described as an important mechanism underlying the beneficial effects of probiotic bacteria on excessive immune activation 51,52 . However, neutralizing IL-10 does not affect the ability of lactobacilli-CFS to dampen IFN-γ production from stimulated PBMC 17 . Moreover, studies in IL-10 KO mice models show that lactobacilli can still modulate immune responses and ameliorate intestinal inflammation by an IL-10-independent mechanism 53,54 , indicating multiple mechanisms of action involving lactobacilli-mediated immune modulation. IL-1ra is another anti-inflammatory cytokine that suppress immune activation by antagonistic competition with IL-1α/β signalling. IL-1 signalling is important for both T-and NK cells effector responses, in particular for IL-17A and IFN-γ production, respectively 55,56 . IL-1β has also been identified as a driver of inflammation in (certain) intestinal inflammatory diseases 57 and neutralizing IL-1β using IL-1ra inhibits fungal-induced IFN-γ responses 45 . In the current study we confirmed IL-1ra production from PBMC cultured with either Lactobacillus-CFS or S. aureus-CFS. However, Lactobacillus-CFS induced minimal levels of IL-1β (Fig. 2b), while S. aureus induced significantly larger amounts of IL-1β. In fact, Lactobacillus-CFS treatment resulted in a 3-fold higher ratio of IL-1ra/IL-1β compared with S. aureus-CFS. Moreover, adding recombinant IL-1ra to S. aureus-stimulated PBMC resulted in suppressed IFN-γ responses, supporting the involvement of IL-1 signalling in Lactobacillus-mediated dampening of induced IFN-γ.
In a previous study, we confirmed the IFN-γ-dampening effect of lactate, specifically on unconventional mucosal associated invariant T (MAIT) cells, γδ T cells and NK cells but not conventional T cells. Here, we showed that the low fraction of the Lactobacillus-CFS, containing only molecules of less than 3 kDa in molecular weight, was able to dampen IFN-γ production in all T cell subtypes including conventional CD4 + and CD8 + T cells. This indicates that lactobacilli produce low molecular weight molecules, other than lactate, capable of modulating pro-inflammatory responses in T cells. Lactobacilli produce several SCFA such as propionate, acetate and butyrate with known effects on host physiology and immunity 34 .
Lactobacillus is a highly heterogeneous genus with immune stimulatory activity that varies between species and strains 58,59 . Interestingly, the immune dampening activity we observed in our setting was conserved across species and strains (See Supplementary Fig. S2). The ability to produce extracellular MVs is also a generally www.nature.com/scientificreports www.nature.com/scientificreports/ conserved trait as it has been found among multiple bacterial genera, species and strains 60 . Treatment of the lactobacilli-CFS high Mw fraction with a lipid removal agent rendered both LGG and L. reuteri incapable of modulating IFN-γ production, indicating that MVs are involved for both species tested. Proteinase K treatment of the L. reuteri-CFS high Mw fraction had no impact on IFN-γ-dampening while heat inactivation did, suggesting that the MV-dampening involves delivery of intravesicular protein cargo to the target cells. On the other hand, heat inactivation of the LGG-CFS high Mw fraction did not abolish the dampening effect to a significant level, suggesting that there are differences in the MV-mediated modulation of cytokine responses between LGG and L. reuteri.
Recently, bacterial MVs have been implicated in a broad range of functions in relation to bacteria-host interaction. MVs from Lactobacillus have been shown to inhibit hepatic cancer growth 61 , protect against S. aureus-induced atopic dermatitis 62 and upregulate host immune gene transcription associated with defence against pathogenic infections 63 . The exact mechanism behind bacterial MV-associated host modulation is still not clear. MVs isolated from Bifidobacterium longum is internalized by mast cells in a phagocytosis-independent mechanism 64 , while MVs isolated from L. sakei subsp. sakei enhance IgA production by activation of Toll-like receptor 2 signalling 65  www.nature.com/scientificreports www.nature.com/scientificreports/ presence of several proteins (p40, p75) with known probiotic effects 66 . Indeed, the MVs, which we isolated from L. reuteri, did contain proteins and nucleic acids (Fig. 5b). Furthermore, gut bacteria-derived MVs have been isolated from peripheral blood, and the circulating MV-profile has been associated with neurodegenerative disease 67 , strengthening the relevance of our findings.
Collectively, we have shown that lactobacilli secrete multiple immunomodulatory factors with the capacity to counteract lymphocytic pro-inflammatory cytokine production by promoting monocyte-derived IL-1ra. Additionally, we identified Lactobacillus-produced extracellular MVs as a significant component for the cytokine regulatory effects observed, extending the general knowledge of bacteria-host interactions, which could bring new implications for future probiotic design.

Data availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request. www.nature.com/scientificreports www.nature.com/scientificreports/