Phenotypic and transcriptional distinction between B cells in Peyer’s patches
The B cells of the PPs in wild-type mice were characterized on cell size using unbiased regular and imaging flow cytometry to account for potential dynamics in the expression of surface markers. We found that the CD3-CD19+B220+ B cells clustered into two subsets according to their size (Fig. 1a, b, Additional file 1: Fig. S1A, B), where 38.7% were large B cells (120–210 μm2) of higher granularity than small B cells (61.3%, 60–120 μm2), while size-separated clusters were not observed for the B cells from spleen or mesenteric lymph nodes (MLNs) in healthy mice. Within the PPs, immunohistochemical staining demonstrated that the two B cell subsets occupied distinct niches (Fig. 1c), where the large B cells were observed at the GL7+/Ki67+/IgA+ region, corresponding to the GC area [10, 15], while the small-sized B cells were predominantly found in the sub-epithelial compartment (Fig. 1c, d). Intravital microcopy confirmed that small B lymphocytes were abundant in the SED pre-GC niche [12, 14], where they were also detected interspersed between the FAE (Additional file 1: Fig. S1C, D, Additional file 2: movie S1, Additional file 3: movie S2), directly facing the luminal antigens.
Transcriptional profiling of the size-sorted PPs CD19+B220+ B cells (Additional file 1: Fig. S2A-G) revealed that both subsets expressed B cell lineage-specific transcripts (Additional file 1: Fig. S2B) and Ig genes in similar patterns but at different levels (Additional file 1: Fig. S2C), indicating a difference in affinity maturation [18]. Clustering analysis showed distinct subgroups of the large and small B cells (Additional file 1: Fig. S2D), where approximately 2290 genes were significantly differentially expressed (FDR q < 0.05). Gene ontology enrichment analysis (Biological Process) demonstrated that large B cells contain significantly more transcripts encoding proteins associated with cell cycle and proliferation, or engaged in metabolic processes including oxidation-reduction, transport, lipid oxidation, and tricarboxylic acid cycle (Fig. 1e, upper panel, Additional file 1: Table S1), especially linked with the glycolysis- and hypoxia-pathways (Additional file 1: Fig. S2F). These transcriptional programs support a high degree of biosynthesis and growth of PPs large B cells, similar to observations of the newly formed GC B cells of spleen and lymph nodes expressing the metabolic checkpoint regulators Gsk-3α (glycogen synthase kinase 3) and Hif-1α [19, 20]. Indeed, the large B cells expressed higher levels of Mki67, Gsk-3α, and Hif-1α compared to the small B cells (Additional file 1: Fig. S2G). Hypoxic signals were identified within the GL7+CD35+ areas using the hypoxia probe pimonidazole (Additional file 1: Fig. S2H). In addition, heat maps of mRNA expression analyzed by qRT-PCR revealed that large B cells expressed significantly higher levels of genes critical for B cell cognition and T cell interaction (BAFFR, CD40, CD274, ICOSL, and TACI) [21], and differentiation (AICDA and TGFβ1), Myd88, and Integrin α4 and β7 compared to small B cells (Fig. 1f).
In contrast, small B cells displayed a gene signature of innate anti-bacterial immune responses (Fig. 1e–g, Additional file 1: Table S1), involving transcripts encoding proteins such as defensins (Defa17, Defa3), lysozymes (Lyz1, Lyz2), and intestinal lactoferrin receptor (Itln1) (Fig. 1e, lower panel). The anti-bacterial potential of small B cells was confirmed by immunohistochemistry, as CD11C-CD19+B220+B cells expressing defensin α3 (DEFa3) were identified in close proximity to the high DEFa3+ FAE region (Additional file 1: Fig. S2I). Notably, genes of pattern recognition receptors TLR2, TLR4, the lipid presenting receptors CD1D, S1PR1, and the chemokine receptor CCR9 were expressed at higher levels in the small compared to the large B cell subset (Fig. 1g). In contrast, the expression of nucleic acid sensing TLR7 and TLR9 as well as CD69, the negative regulator of S1PR1, were expressed at lower levels in small B cells (Fig. 1g). Both large and small B cells expressed similar levels of Bcl6 (Additional file 1: Fig. S2G), the master regulator of GC B cell-fate [22], and genes associated with B cell migration (Fig. 1g), including CXCR5, CCR6, and EBI2 [14, 15, 22]. Further, flow cytometry analysis showed that large B cells more frequently expressed GL7 and at higher levels (Fig. 1h, i), in contrast to the small B cells that were more often IgD+, CD62L+, CD21+ (CR2), and exclusively expressed surface S1PR1 (Fig. 1h, i). Activated B cells at pre-GC stage have been reported to express Bcl6 [23], thus implying a potential pre-GC phenotype of the IgD+/GL7−/S1PR1+/Bcl6, CCR6-expressing small-B subset.
Taken together, the size of B lymphocytes in PPs is a good estimator of the B cell activation state and might therefore allow for separation and studies of specific populations in response to environmental changes. We find that large B cells in the PPs are of GC-like phenotype, expressing GL7+/S1PR1−/Ki67+/Bcl6 and CD69, and display intense metabolic activity. In addition, small B cells represent a heterogeneous group of SED B cells that exhibit bacterial defense properties, mainly comprising of the pre-GC phenotype [22, 23], but also of IgD+/CD62L+ naïve B cells [15].
Limosilactobacillus reuteri expands the different population of Peyer’s patches B cells and enhances their effector functions
Peroral administration of 108 L. reuteri R2LC for seven consecutive days specifically expanded the pre-GC-like and the GC-like B cell subsets, together increasing the CD3-CD19+B220+ B cell population from 60 to 70% of total PPs cells (Fig. 2a, Additional file 1: Fig. S3A, B). As a result, the PPs became substantially enlarged and increased in height (Additional file 1: Fig. S3C), while the B cell population (CD3-CD19+B220+) in MLNs was not altered (Additional file 1: Fig. S3D), demonstrating gut-specific effects. Bacterial strain specificity was tested using two L. reuteri strains isolated from human breast milk (ATCC PTA 4659 and ATCC PTA 6475), whereas R2LC originates from rat intestine. While 4659 expanded only the GC-like B cell population, no difference was observed between 6475 and the control group (Fig. 2a, Additional file 1: Fig. S3B). We therefore mainly focused on R2LC, hereafter referred to as “L. reuteri.”
The expansion of the pre-GC-like B cells in L. reuteri-treated mice was almost exclusively due to increased number of S1PR1+ B cells (Fig. 2b) with decreased surface expression of S1PR1. These observations imply increased availability of the S1P ligand, as previously reported [24]. Indeed, L. reuteri-treatment down-regulated gene expression of S1P lyase (sgpl1) while upregulating expression of ceramide synthase (LASS5) in the PPs (Fig. 2c), suggesting reduced S1P degradation and amplified S1P synthesis. The involvement of the S1P/S1PR1 pathway was further assessed using a S1PR1 agonist FTY720 (Fingolimod) [24]. In line with the observations following L. reuteri-treatment, FTY720-treatment significantly increased the number of pre-GC-like, but not to GC-like B cells (Fig. 2d, Additional file 1: Fig. S3E).
Further, a principal component analysis (PCA) of the transcriptome data for the size-sorted PPs CD19+B220+ B cells from control and L. reuteri-treated mice revealed four separated clusters (Fig. 2e). PC1 variation confirmed the transcriptional distinction between B cells of the SED and the GC, which separated independently of bacterial stimulation. L. reuteri-treatment was found to impact the transcriptional profile of both subsets in two different ways: one was common for both subsets as they were shifted towards the right on PC1, whereas the other was subset-specific, as displacement on PC2 occurred in opposite directions. The differential expression of genes of large and small B cells in response to L reuteri treatment was further analyzed and was found to be consistent with the PCA analysis (Additional file 1: Fig. S2D), indicating distinct responses of the B cell subsets to L. reuteri. Gene ontology enrichment analysis of pre-GC-like B cells showed that L. reuteri-treatment further upregulated expression of gene clusters associated with defense responses to bacteria (Fig. 2f), including Defa3, Defa21, Lyz1, Fbrs, Pira1, Arhgef1, Mptx2, Itln1, Muc2, Agr2, Mmp7, and Angptl1. Although no treatment difference was observed on DEFa3+ protein levels in B cells, the DEFa3+ B cells co-expressed S1PR1 (Fig. 2g, Additional file 1: Fig. S3F). These data demonstrate that L. reuteri promotes the bacterial defense gene program of the S1PR1+ pre-GC-like B cell subset in a S1P/S1PR1-dependent manner.
L. reuteri increases the numbers of B cells in the PPs GC, and modifies their functional gene signature
L. reuteri-treatment expanded the GL7+ GC area of the PPs (Fig. 3a, Additional file 1: Fig. S3G), where more proliferating Ki67+ B cells were found to accumulate (Fig. 3c, Additional file 1: Fig. S3H). Flow cytometry confirmed that L. reuteri (R2LC and 4659) increased the number of GL7+ and Ki67+ B cells (Fig. 3b, d). In addition, active DNA synthesis determined by EdU (5-ethynyl-2′-deoxyuridine) uptake on the last day of the experiment occurred in GL7+ but not GL7- B cells (Fig. 3e, left panel), whereas the number of EdU+GL7+B cells was unaltered (Fig. 3e, right panel). In contrast, transcriptome analysis of the GC-like B cells showed that L. reuteri-treatment downregulated cell division and cell cycle gene clusters (Additional file 1: Table S2), and expression of Mki67, Gsk-3α, and Hif-1α (Fig. 3f). Moreover, the proportion of HIF-1α+ B cells and their HIF-1α expression were reduced by L. reuteri-treatment (Additional file 1: Fig. S3I). These results indicate that L. reuteri stimulate the proliferation and clustering of GC-like B cells at the GC niche. Furthermore, L. reuteri-treatment upregulated the gene signature of GC function in these B cells (Fig. 3g), and expressions of TGFβ1 and TGFβR1. Using flow cytometry, we discovered a significant expansion of TGFβ1+ B cells in PPs of L. reuteri-treated mice (Fig. 3h). The L. reuteri-induction of TGFβ1 was specific to the TGFβ1-TGFβR1-expressing GC-like B cell subset, and the tissue concentration of TGFβ1 did not change (Fig. 3i). In contrast, L. reuteri-treatment reduced cytokine levels in the PPs including TNF-α, CXCL1, IL-6, and IL-10 (Fig. 3i), promoting a non-inflammatory shift of the microenvironment. In addition, L. reuteri-treatment upregulated the IgA germline transcript (αGT) in PPs (Fig. 3j). These data demonstrate that L. reuteri enhances the functional gene expression of GC-like B cells, and promotes their low proliferative/high T cell-interactive post-GC phenotype [21] by autocrine stimulation of TGFβ1 signaling [15].
L. reuteri DSM 17938 has previously been reported to modulate gut microbiome and inhibit autoimmune responses in immune-compromised Scurfy mice via the adenosine receptor A2A [25]. To determine whether L. reuteri R2LC induces PPs B cell responses through adenosine signaling, we generated a mutant strain lacking the ability to convert AMP into adenosine by impaired functional 5′-nucleotidase (L. reuteri R2LC_ΔADO). The effect of the mutant on the PPs B cells subsets did not differ in any respect from the wild-type R2LC strain (Fig. 3b, d, h, k). However, the mutant increased the number of Ki67+ and TGFβ1+ B cells in the PPs compared to control mice (Fig. 3d, h). These results indicate that adenosine signaling alone does not determine the effects by R2LC on PPs B cells.
L. reuteri promotes B cell IgA-responses through the T follicular helper (Tfh)-PD-1 pathway
As L. reuteri was demonstrated to enhance IgA induction in PPs, its effects on IgA production and secretion were investigated. Interestingly, we found that L. reuteri-treatment increased the density of IgA+ plasma cells in the lamina propria of both colon and ileum, as well as increased IgA expression in the ileum (Fig. 4a). Accordingly, higher titers of free IgA were detected in the intestinal lumen, whereas IgA or IgG concentrations in circulation were not altered (Fig. 4b, c). Following secretion, IgA serves as part of innate immune surveillance through binding commensal microbiota [26]. Flow cytometry revealed that L. reuteri-treatment enhanced the proportion of IgA+ bacteria in the ileal flora, as well as the level of IgA bound per bacteria (IgAhigh) (Fig. 4d, Additional file 1: Fig. S4A). Increased number of IgAhigh bacteria was also observed with L. reuteri R2LC_ΔADO compared to control (Additional file 1: Fig. S4B). L. reuteri-treatment did not increase the α-diversity of the microbiota (Fig. 4e, Additional file 1: Fig. S4C), but induced a shift of bacterial community composition (Fig. 4f, Additional file 1: Fig. S4D). Notably, ileal Clostridiaceae, Erysipelotrichaceae and other bacteria belonging to Firmicutes were less abundant in L. reuteri-treated mice compared to in control mice (Additional file 1: Fig. S4E). In contrast, L. reuteri-treatment promoted the expansion of ileal Bifidobacteriaceae, as well as ileal and colonic bacterial phylotype S24-7 (affiliated to Bacteroidetes) (Fig. 4f, Additional file 1: Fig. S4D, E), while the relative abundance of Lactobacillaceae did not increase at either site. Together these results demonstrate that administration of a single species of L. reuteri boosts IgA production, thereby shifts the composition of the intestinal microbiota.
In lymphoid organs, PD-1+CXCR5+ Tfh cells are specialized in promoting T cell-dependent B cell IgA-responses [21]. Immunohistochemical staining revealed that in PPs, the PD-1+ Tfh cells are localized in the GC area (Fig. 4g). L. reuteri-treatment did not affect the CD3+TCRαβ+CD4+ T cell population (Additional file 1: Fig. S4F); however it doubled the number of PD-1+CXCR5high Tfh cells, which expressed higher levels of PD-1 (Fig. 4h) and increased the number of IL-21+CD4+ effector T cells and their IL-21 production (Additional file 1: Fig. S4F). Importantly, the L reuteri-induced expansion of GC-like B cells was completely dependent on the Tfh-PD-1 pathway, as demonstrated by either administration of PD-1-blocking monoclonal antibodies or depletion of CD4+ cells (Fig. 4i). PD-1 inhibition also impaired the ability of L. reuteri-treatment to enhance IgA production, resulting in reduced levels of IgA-binding bacteria (Fig. 4j, k). These data demonstrate that the Tfh-PD-1 pathway is crucial for L. reuteri-induction of B cell IgA-responses.
L. reuteri provides protection against DSS-induced intestinal inflammation and disruption of Peyer’s patches
As previously demonstrated [8], DSS induces colitis and profound inflammation in the colon (Additional file 1: Fig. S5A, B). Here, we found that it also severely damaged the ileal mucosa (Additional file 1: Fig. S5C) and reduced the number of CD3-CD19+B220+ B cells in the PPs by more than 50%. This reduction was due to decreased numbers of both pre-GC-like B cells and GC-like B cells, resulting in reduced PPs size (Fig. 5a, b). Colitis induced significant B cell accumulation in MLNs (Additional file 1: Fig. S5D) and caused ileal mucosal infiltration of IgA+ plasma cells (Additional file 1: Fig. S5E), but did not affect the proportion of IgA+ bacteria in the ileal flora (Additional file 1: Fig. S5F). In contrast, a significant increase of the level of IgA bound per bacteria (IgAlow) was detected in DSS-treated mice when compared to control (Additional file 1: Fig. S5F). Prophylactic treatment of L. reuteri for 14 days prevented DSS-induced ileal disruption (Additional file 1: Fig. S5C), and delayed and attenuated colitis symptoms (Fig. 5c, Additional file 1: Fig. S5A, B). Disease activity index (DAI) [8] revealed that the protective effect was most obvious for R2LC followed by 4659, but not observed with 6475 (Additional file 1: Fig. S5G). L. reuteri R2LC_ΔADO-treatment also lowered the DAI compared to DSS only, demonstrating that the disease-ameliorating effects by R2LC are not mediated through adenosine signaling.
L. reuteri-treatment given to DSS-challenged mice preserved their PPs integrity, the number of both subsets of PPs B cells (Fig. 5a, b), and reduced the B cell accumulation in MLNs (Additional file 1: Fig. S5D). Further, L. reuteri-treatment reduced the ileal infiltration of IgA+ plasma cells (Additional file 1: Fig. S5E), and restored the DSS-induced shift of SIgA-coated bacteria (Additional file 1: Fig. S5F). A strong negative correlation between body weight loss and B cell numbers in PPs following DSS-treatment was detected (Fig. 5d), supporting that the effects by L. reuteri on PPs B cells are important for preventing colitis. Again, part of the probiotic effect can be explained by the local S1P/S1PR1 signaling in the PPs, as co-treatment with FTY720 to DSS-treated mice mimicked the effect of probiotics with maintained pre-GC-like B cells in PPs (Fig. 5b) and delayed disease onset (Fig. 5c, Additional file 1: Fig. S5H).
DSS-treatment induced a major shift of the bacterial community composition into a dysbiotic state (Fig. 5e), including reduced α-diversity in both ileum and colon (Fig. 4e, Additional file 1: Fig. S4C). The bacterial taxa Erysipelotrichaceae in the ileum (Fig. 5f) and Escherichia-Shigella (affiliated with Enterobacteriaceae) in the colon (Additional file 1: Fig. S5I) were identified as the main pathobionts, with their relative abundance increasing from less than 2% in healthy animals to up to 80% in DSS-treated mice, while the relative abundance of S24-7 was decreased by DSS at both sites (Fig. 5g, Additional file 1: Fig. S5J). L. reuteri-treatment preserved the α-diversity of colonic microbiota, the ileal Lactobacillaceae from DSS challenge (Additional file 1: Fig. S4C, Fig. 5h), and inhibited the overgrowth of Erysipelotrichaceae and Escherichia-Shigella (Fig. 5f, Additional file 1: Fig. S5I). It also caused an obvious increase of Bifidobacteriaceae and Coribacteriacease in the ileum (Fig. 5e) and maintained the dominant S24-7 population in the colon (Additional file 1: Fig. S5J).
Thus, we report that L. reuteri-treatment protects against intestinal inflammation by maintaining the functions of PPs, which modifies intestinal IgA production and prevents inflammation-induced microbiota dysbiosis.