Research ArticleLYMPHATICS

Fibroblastic reticular cells initiate immune responses in visceral adipose tissues and secure peritoneal immunity

See allHide authors and affiliations

Science Immunology  10 Aug 2018:
Vol. 3, Issue 26, eaar4539
DOI: 10.1126/sciimmunol.aar4539

Setting the stage for attack

In classical secondary lymphoid organs (SLOs) such as lymph nodes, tonsils, and Peyer’s patches, it is well established that fibroblastic reticular cells (FRCs) play an integral role in the generation of immune responses. Nonclassical SLOs, including fat-associated lymphoid clusters (FALCs), also play important roles in systemic immunity. However, the role of FRCs in FALCs has not been previously examined. Here, using Ccl19-driven cre to delete MYD88 in FALC-associated FRCs, Perez-Shibayama et al. report that FRCs in FALCs play both organizational and immunomodulatory roles. These studies add to the growing recognition of the importance of stromal cells in shaping immune organs and immune responses.


Immune protection of the body cavities depends on the swift activation of innate and adaptive immune responses in nonclassical secondary lymphoid organs known as fat-associated lymphoid clusters (FALCs). Compared with classical secondary lymphoid organs such as lymph nodes and Peyer’s patches, FALCs develop along distinct differentiation trajectories and display a reduced structural complexity. Although it is well established that fibroblastic reticular cells (FRCs) are an integral component of the immune-stimulating infrastructure of classical secondary lymphoid organs, the role of FRCs in FALC-dependent peritoneal immunity remains unclear. Using FRC-specific gene targeting, we found that FRCs play an essential role in FALC-driven immune responses. Specifically, we report that initiation of peritoneal immunity was governed through FRC activation in a myeloid differentiation primary response 88 (MYD88)–dependent manner. FRC-specific ablation of MYD88 blocked recruitment of inflammatory monocytes into FALCs and subsequent CD4+ T cell–dependent B-cell activation and IgG class switching. Moreover, containment of Salmonella infection was compromised in mice lacking MYD88 expression in FRCs, indicating that FRCs in FALCs function as an initial checkpoint in the orchestration of protective immune responses in the peritoneal cavity.


Essential internal organs are suspended in the peritoneal, pleural, or pericardial cavities, which are fluid-filled spaces formed by mesothelial membranes. The peritoneum covers the abdominal wall and the visceral organs with a single layer of mesothelial cells that secrete serous fluid. The peritoneal fluid contains immune cells that facilitate rapid binding and disposal of bacteria and bacterial products that have escaped from the intestine (1). For example, peritoneal B-1 B cells produce natural, low-affinity immunoglobulin M (IgM) antibodies (2) that opsonize bacterial antigens and can thereby enhance uptake and elimination of bacteria by myeloid cells (3). Other immune cells in the peritoneal cavity such as B-2 B cells translocate into omental fat-associated lymphoid clusters (omFALCs) (4, 5) and communicate with these nonclassical secondary lymphoid organs in the visceral adipose tissue. Furthermore, CD1d-restricted natural killer T cells in FALCs of the mesentery (mesFALCs) can be activated by lipid-based antigen present in the peritoneal cavity initiating a cascade of innate and adaptive immune responses that mediate expansion and de novo formation of FALCs (6). Although the development of lymph nodes (LNs) or Peyer’s patches requires lymphotoxin (LT)–mediated or receptor activator of nuclear factor κB (RANK) ligand–mediated stimulation of stromal cells (79), formation of FALCs depends, to a large extent, on the activation of stromal cells via the production of inflammatory cytokines, which are induced through the presence of microbiota in the intestine (6). Moreover, fibroblastic stromal cell–derived factors such as interleukin-33 (IL-33) are important for the activation of B-1 B cells in the pleural cavity (10), suggesting that specialized populations of nonhematopoietic cells actively contribute to the initiation and regulation of immune responses in the body cavities.

Stromal cells of secondary lymphoid organs are divided into major populations defined by the expression of podoplanin (PDPN) and the endothelial cell marker CD31 (11). PDPN+CD31+ lymphatic endothelial cells are involved in building lymphatic vessels, whereas PDPNCD31+ blood endothelial cells form the inner lining of arteries, capillaries, and veins. The nonendothelial fraction PDPN+CD31 of lymphoid organ stromal cells is generally referred to as fibroblastic reticular cells (FRCs) (11, 12). FRCs of the T cell zone regulate T cell migration and survival by producing constitutive chemokines such as CCL19 and CCL21, both crucial for the attraction and retention of T cells (1214). Although FRCs of the T cell zone also affect T cell survival and differentiation through the production of IL-7 (13) and IL-15 (15), B cell zone FRCs provide essential factors such as the B cell zone chemokine CXCL13 and the activation factor BAFF to promote recruitment and survival of B cells (16, 17). Likewise, the secretion of CXCL13 in FALCs has been shown to support accumulation of B cells and promote peritoneal antibody responses (18). However, production of CXCL13 in omFALCs appears to be a trait of both fibroblastic stromal cells and CD11b+ myeloid cells (18). Hence, to assess whether cells with a distinct FRC phenotype exist in FALCs and to determine to which extent these cells contribute to peritoneal immunity, we have used the Ccl19-Cre model that permits in vivo targeting of FRCs in classical secondary lymphoid organs (12, 15, 19). We found that Ccl19-expressing FRCs in the visceral adipose tissue were mainly confined to omental and mesenteric FALCs. Cell type–specific ablation of myeloid differentiation primary response 88 (MYD88)–dependent innate immunological sensing abrogated FRC activation and expansion. As a consequence, recruitment of inflammatory monocytes into FALCs and T helper cell–dependent B cell activation was blunted in mice lacking MYD88 expression in FRCs. Generation of bone marrow chimeras and blocking of critical FRC activation pathways with Ig fusion proteins revealed that an FRC-myeloid cell axis in FALCs drives the initiation of peritoneal immune responses. Direct activation of FRCs via MYD88 appears to be the primary event in the immune activation cascade that controls protective immunity in the peritoneal cavity.


FRCs govern the structural foundations of FALCs

FRCs in LNs (11) and Peyer’s patches (15) use Toll-like receptors (TLRs) to sense microbial products and to promote innate and adaptive immune activation pathways. To assess whether innate immunological sensing in FRCs of FALCs affects peritoneal immunity, we have used mice that lack the expression of the adaptor molecule MYD88 in Ccl19-expressing cells (15). Moreover, to locate immune-interacting fibroblasts in FALCs, we crossed Ccl19-Cre mice with R26R–enhanced yellow fluorescent protein (EYFP) mice (Ccl19-EYFP). We found high numbers of Ccl19-expressing cells in omFALCs forming dense clusters that harbored B cells and were pervaded by a glomerular network of blood vessels [Fig. 1A (arrows) and movie S1], whereas fewer Ccl19-expressing cells were located around the main blood vessels leading to the omFALCs (Fig. 1A, arrowheads). The absence of MYD88 expression in Ccl19-Cre+ cells precipitated a malformation of omFALCs with smaller clusters and the lack of an elaborated blood vessel network (Fig. 1A, asterisks). Likewise, the structural integrity of mesFALCs was impaired to a similar degree in Ccl19-Cre Myd88fl/fl mice (fig. S1A). Assessment of the phenotypical properties of the Ccl19-Cre+ fibroblastic stromal cell scaffold of omFALCs (Fig. 1B) and mesFALCs (fig. S1B) revealed that the cells expressed the canonical FRC markers intercellular adhesion molecule–1, vascular cell adhesion molecule–1, platelet-derived growth factor receptor α (CD140a), tumor necrosis factor receptor–1 (TNFR1), and TNFR2 but lacked expression of CD157, indicating that the cells resemble perivascular FRCs present in other lymphoid organs (20). We found a substantial reduction in the frequency of PDPN+EYFP+ cells in both omFALCs (Fig. 1, C and D) and mesFALCs (fig. S1C) when MYD88 signaling was abrogated in a cell type–specific manner. The enumeration of recoverable cells from omFALCs confirmed about 50% reduction of EYFP+ cell content (Fig. 1E) that was associated with significant reduction in FALC numbers and size in the omentum (Fig. 1F). Moreover, the impairment of innate immunological sensing in omFALC FRCs precipitated a reduced accumulation of immune cells (Fig. 1G) with a significant impact on the B-2 B cell compartment (Fig. 1H), suggesting that FRCs not only underpin the structure of FALCs but also actively interact with immune cells.

Fig. 1 Impact of FRC-specific Myd88 ablation on omental FALC organization.

(A) Confocal microscopic analysis of omental FALCs in 8- to 10-week-old Ccl19-EYFP mice (top) and mice lacking MYD88 expression in FRCs (bottom) using the indicated markers. Scale bar, 100 μm. (B) Flow cytometric analysis of canonical FRC marker expression on CD45PDPN+EYFP+ cells from the omentum of the indicated mouse strains; isotype control staining is indicated in gray. ICAM, intercellular adhesion molecule; VCAM, vascular cell adhesion molecule. (C to E) Enumeration of FRCs by flow cytometry. Representative dot plots of PDPN+EYFP+ cells (C) and frequency (D) and absolute numbers (E) of PDPN+EYFP+ cells in the omentum of the indicated mouse strains. (F) Quantification of omental FALC numbers and size in whole mount stains; number of FALCs per square millimeter (left) and FALC-covered area in the omentum (right). Enumeration of hematopoietic cells (G) and B cell subsets (H) in mice lacking MYD88 expression in FRCs or in Cre-negative littermates (Ctrl) by flow cytometric analysis. Data are from one representative sample with five to eight mice (A to C) or pooled data from two independent experiments [n = 5 to 8 mice per group, with bars in (D) to (H) representing mean values ± SEM]. Statistical analysis was performed using Student’s t test with *P < 0.05 and **P < 0.01.

FRC-dependent remodeling of FALCs

Proper activation of Ccl19-expressing FRCs in classical secondary lymphoid organs is important for antiviral immunity (12), LN expansion (21), B cell activation (16), and the cellular constitution of the splenic white pulp marginal zone (19). To assess whether the blocking of innate immunological sensing in FALC FRCs affects peritoneal immunity, we used outer membrane proteins C and F (OmpC/F) of Salmonella Typhi, which induce both T cell–independent and –dependent B cell responses (22). Moreover, MYD88-dependent signaling is crucial for the initiation and sustenance of antibody responses against S. Typhi OmpC/F (23). The initial B cell activation after intraperitoneal OmpC/F immunization was confined to the omentum (fig. S2A) and was accompanied by a substantial remodeling of FALCs in the omentum of Ccl19-EYFP mice (fig. S2B), recruitment of myeloid cells (Fig. 2, C and D), and expansion of FRCs (fig. S2E). Activated omFALCs were surrounded by a corona of CD11b+ myeloid cells (Fig. 2A, top) that was absent in omFALCs of mice lacking MYD88 expression in FRCs (Fig. 2A, bottom). MYD88 deficiency in FRCs abrogated FALC expansion after OmpC/F exposure (Fig. 2, B and C), and enumeration of FRCs by flow cytometry revealed that MYD88-deficient FRCs failed to react and to expand after OmpC/F exposure (Fig. 2, D and E). Moreover, the specific block of innate immunological sensing in Ccl19-expressing FRCs incapacitated recruitment of CD11b+ myeloid cells to (Fig. 2F) and decreased accumulation of B cells in (Fig. 2, G and H) the omentum. Overall, these data indicate that the small subset of Ccl19-Cre+ omental fibroblastic stromal cells is endowed with the ability to drive FALC remodeling and to induce innate and adaptive antibacterial immune responses in these nonclassical secondary lymphoid organs.

Fig. 2 Expansion of omental FRCs and FALC remodeling after intraperitoneal exposure to bacterial antigen.

(A to C) Confocal microscopic analysis of FALCs on day 4 after immunization with S. Typhi OmpC/F. (A) Microscopic structure of omental FALCs from Ccl19-EYFP mice (top) and mice lacking MYD88 expression in FRCs (bottom) using the indicated markers. Representative image from one of four mice per group. Scale bar, 100 μm. Number of FALCs per square millimeter (B) and percentage of FALC-covered area in the omentum (C). Relative frequency (D) and absolute numbers (E) of PDPN+EYFP+ cells in the omentum of the indicated mouse strains on day 4 after immunization with S. Typhi OmpC/F or in untreated mice (naive); bars represent mean values ± SEM. Enumeration of CD11b+ myeloid cells (F), B220+ B cells (G), and B cell populations (H) on day 4 after immunization in mice lacking MYD88 expression in FRCs or in Cre-negative littermates (Ctrl). Data in (B) to (H) represent mean values ± SEM and are from four to nine mice per group from two to three independent experiments; statistical analysis was performed using one-way ANOVA with Dunnett’s multiple comparison test (D and E) or Student’s t test (B, C, and F to H). *P < 0.05, **P < 0.01, and ***P < 0.001.

Myd88-dependent transcriptional reprogramming of FALC FRCs

To elucidate the molecular circuits that control the immunological functions of FALC FRCs, we analyzed the changes in the transcriptome of omental EYFP+ cells from Ccl19-EYFP and Ccl19-EYFP Myd88fl/fl mice. To assure the high quality of the RNA sequencing (RNA-seq) data, we only used sorted cell populations with a yield of >1200 cells and a purity of >95% EYFP+ cells for complementary DNA (cDNA) library construction (fig. S3). Multidimensional scaling (MDS) based on differentially expressed genes showed that OmpC/F-induced immune activation was associated with a shift in the gene expression pattern in both MYD88-proficient and MYD88-deficient FRCs (Fig. 3A). We found that 110 genes were differentially expressed under naive conditions, with 64 genes being down-regulated and 46 genes being up-regulated (Fig. 3B). Gene expression analysis revealed that MYD88-dependent genes in FRCs were involved in innate immunological sensing and inflammation (e.g., Tlr6, Tlr7, Nlrp1b, Nlrp2, and Scarf1), mesenchymal cell activation and growth (e.g., Fgf5, Stam1, and Klf5), and antigen presentation [e.g., the major histocompatibility complex class II (MHCII) molecules H2-Eb2 and H2-DMb1 and the costimulatory molecules Tnfs4 and Tnfsf9; Fig. 3C]. Gene set enrichment analysis indicated four significantly down-regulated pathways in MYD88-deficient omFALC FRCs related to inflammation, response to TNF, tissue remodeling, and stroma and adaptive responses (Fig. 3D). Further analysis showed that MYD88 signaling in FALC FRC is, to a large extent, required to amplify the expression of inflammatory chemokines (Cxcl2, Ccl2, and Cxcl1), homeostatic chemokines (Ccl19 and Cxcl13), and cytokines (Tnf, Il6, and Ifnab) (Fig. 3E) that are involved in steering of innate and adaptive immune responses (24, 25). Moreover, the profound changes in the expression of extracellular matrix proteins (e.g., Col15a1, Fbln1, and Tnc) and metalloproteinases (Adamts1) (Fig. 3E) indicate that FRC activation precipitates a structural adaptation of the FALC microenvironment. In summary, the molecular dissection of the MYD88-dependent transcriptional reprogramming of FALC FRCs suggests that this distinct cell population is involved in the coordination of immune cell trafficking and the provision of dedicated niches for immune activation.

Fig. 3 RNA-seq transcriptional profiling of FRCs in the omentum.

(A) MDS plot of the logCPM values of differentially expressed genes in at least one pairwise comparison between Myd88-proficient and Myd88-deficient Ccl19-EYFP+ FRCs under naive conditions and on day 4 after S. Typhi OmpC/F immunization. (B) Volcano plot indicating 110 differentially expressed genes between denoted FRC populations under naive conditions. Number of up- and down-regulated genes is indicated. Thresholds for differential gene expression, |FC| ≥ 2 and adjusted P ≤ 0.05, are indicated on a log2 and log10 scale, respectively. FC, fold change. (C) Heat map of genes related to innate sensing/inflammation, activation, and antigen presentation for indicated FRC populations under naive conditions. (D) Gene set enrichment analysis of FRC populations denoted with [A] and [B] after OmpC/F immunization. Nominal P values and false discovery rate (FDR)–corrected P values are shown for indicated gene sets with significance level FDR ≤ 0.1. (E) Heat maps of gene expression levels related to enriched pathways in (D). Data are pooled from two to five mice per group. Data represent logCPM values that have been row mean–centered and row-normalized (C and E). Statistical analysis was performed using gene-wise likelihood ratio tests for a negative binomial generalized linear model using edgeR (B) and competitive gene set tests (D), with a Benjamini-Hochberg (FDR) multiple comparison correction.

FALC FRCs control CCR2-dependent inflammatory monocyte recruitment

On the basis of our previous finding that MYD88-deficient FRCs from Peyer’s patches fail to respond to viral TLR ligands (15), we considered it possible that FRCs from FALCs respond directly to TLR-binding bacterial products and thereby contribute to MYD88-dependent innate immune activation in the peritoneal cavity. Mice lacking MYD88 expression in FRCs showed a significantly reduced recruitment of CD11b+Ly6C+ inflammatory monocytes into the omentum (Fig. 4A) and the peritoneal cavity (Fig. 4B). To assess the quality and magnitude of MYD88-dependent FRC activation through bacterial TLR ligands, we first isolated PDPN-expressing fibroblastic stromal cells from the omentum (fig. S4A) and the mesentery (fig. S4B). Compatible with the ex vivo RNA-seq data from omental FALC FRCs (Fig. 3), we found a pronounced MYD88 dependence of the production of TNF and CCL2 after exposure to TLR4 and TLR2 ligands including OmpC/F preparations (Fig. 4C). CCL2 is a chemokine that binds to CCR2, which is highly expressed in inflammatory monocytes. Because entry of monocytes into inflamed tissues is regulated, to a large extent, via the CCL2/CCR2 axis (26, 27), we generated bone marrow chimeric mice to assess whether the swift assembly of inflammatory monocytes is mediated through MYD88-dependent chemokine production in FALC FRCs. We found that reconstitution of mice lacking MYD88 expression in FRCs with either CCR2-competent or CCR2-deficient hematopoietic cells resulted in a profound reduction of CD11b+ myeloid cell accumulation in the omentum after OmpC/F immunization (Fig. 4D). Likewise, recruitment of CCR2-deficient Ly6C+ inflammatory monocytes was impaired to a similar degree when compared with CCR2-proficient monocytes under conditions of MYD88 deficiency in FALC FRCs (Fig. 4E). Because both the mobilization of monocyte progenitors in the bone marrow and the recruitment of monocytes from the bloodstream into inflamed tissues are considered key mechanisms for effective clearance of bacterial infections (28), we assessed Ccl19-Cre transgene activity in bone marrow stromal cells of OmpC/F-exposed Ccl19-EYFP mice. Only a fraction of <0.1% of the fibroblastic stromal cells in the bone marrow expressed EYFP (fig. S4C), rendering it highly unlikely that MYD88 deficiency in this compartment affects myeloid cell mobilization. Moreover, the presence of neither CD11c+Ly6C+ inflammatory monocytes (fig. S4D) nor CCR2-expressing monocytes (Fig. 4F) in the circulation was affected by the FRC-specific ablation of MYD88 under naive conditions. Together, these data indicate that CCL19-expressing FRCs in FALCs are the main initial source of CCL2 that leads to the attraction of inflammatory monocytes into the visceral adipose tissues during peritoneal inflammation.

Fig. 4 Mechanisms of FRC-dependent myeloid cell recruitment.

Flow cytometry–based enumeration of CD11b+F4/80Ly6G-Ly6C+ inflammatory monocytes in the omentum (A) and peritoneal cavity (B) in mice lacking MYD88 expression in FRCs or in Cre-negative littermates (Ctrl) at the indicated days after intraperitoneal immunization with S. Typhi OmpC/F. (C) Production of the inflammatory mediators TNF (left) and the chemokine CCL2 (right) by fibroblastic stromal cells isolated from the omentum of Myd88-proficient or Myd88-deficient mice. Data are representative of one of three independent experiments with at least triplicate measurements. nd, nondetected. (D and E) Recruitment of CD11b+ myeloid cells (D) and CD11b+F4/80Ly6G-Ly6C+ inflammatory monocytes (E) into the omentum of the indicated bone marrow chimeric mice on day 4 after immunization with S. Typhi OmpC/F. (F) Flow cytometric analysis of CD11b+ Ly6C+ CCR2+ monocytes in circulation of the indicated mice under naive conditions with representative dot plots is shown on the left. Data are shown as mean values ± SEM and are pooled from six to nine mice from three independent experiments (A and B) and from six to eight mice per group from two independent experiments (D to F); statistical analysis was performed using Student’s t test (A, B, and F) or one-way ANOVA with Dunnett’s multiple comparison test (C to E) with *P < 0.05, **P < 0.01, and ***P < 0.001.

FRC activation drives IgG class switch in the early germinal center reaction

The antibody repertoire in the peritoneal cavity is generated by T cell–dependent or T cell–independent B cell subsets (2931). The transcriptome analysis of omFALC FRCs (Fig. 3) suggested that the lack of MYD88 in Ccl19-Cre+ FRCs directly affects adaptive immune responses because up-regulation of important chemokines that regulate T and B cell migration in lymphoid organs (i.e., Ccl19 and Cxcl13) was substantially impaired. To assess whether and to which extent MYD88-dependent signals in FALC FRCs affect B cell activation, we followed S. Typhi OmpC/F-specific B cell responses in the omentum and the peritoneal cavity. We found that the numbers of IgM OmpC/F-specific antibody-secreting cells (ASCs) in the omentum (Fig. 5A) and the peritoneal cavity (Fig. 5B) were not affected by FRC-restricted MYD88 ablation. However, production of IgG was already significantly reduced on day 4 after immunization (Fig. 5, A and B). Consistently, OmpC/F-specific IgM levels were not different in the peritoneal fluids from Ccl19-EYFP Myd88fl/fl and control mice, whereas IgG secretion into the peritoneal cavity was significantly reduced when FALC FRCs were MYD88-deficient (fig. S5, A and B). Although FALCs do not develop proper germinal centers (31), B cells rapidly acquire a germinal center–like phenotype that can be assessed by the first apoptosis signal (FAS) and the T and B activation marker (GL7) expression (fig. S2A). The mounting of an expeditious germinal center–like response on day 4 after immunization was completely dependent on MYD88 expression in FRCs (Fig. 5C). The requirement of MYD88 expression by FALC FRCs for the initiation of a germinal center reaction in the omentum was confirmed in bone marrow chimeric mice (fig. S5, C and D) in which the expression of activation-induced deaminase in germinal center B cells can be tracked by green fluorescent protein (GFP) expression (32). Whereas the substantial delay in the establishment of an FRC-dependent germinal center environment did not affect serum titers of OmpC/F-specific IgM antibodies (Fig. 5D), we found that serum IgG titers were significantly reduced on day 4 in Ccl19-EYFP Myd88fl/fl mice (Fig. 5E). To dissect whether the FRC-dependent recruitment of inflammatory monocytes and T cell–dependent IgG switching are linked, we first immunized MHCII–deficient mice with OmpC/F preparations and could confirm the requirement of the CD4 T cell activation pathway for IgG production in the omentum and the peritoneal cavity (fig. S5E). Likewise, the rapid expansion of germinal center–like B cells in the omentum on day 4 was abrogated in the absence of MHCII antigens (fig. S5F). On the contrary, the recruitment of inflammatory monocytes into the omentum (fig. S5G) and the peritoneal cavity (fig. S5H) was independent of the MHCII pathway, suggesting that CCR2-dependent attraction of inflammatory monocytes precedes CD4+ T cell activation. To further assess to which extent recruitment of CCR2-expressing inflammatory monocytes affects the swift IgG switch after OmpC/F administration, we ablated inflammatory monocytes using a highly effective anti-CCR2 antibody (33). This depletion protocol affected mainly Ly6C+ cells in OmpC/F-immunized mice (fig. S5I) and led to significantly reduced proportions of germinal center B cells (Fig. 5F). Again, loss of CCR2-expressing inflammatory monocytes impaired IgG ASC differentiation in both omentum (Fig. 5G) and peritoneal cavity (Fig. 5H). Assessment of B cell differentiation in the omentum of bone marrow chimeric mice revealed that Myd88 expression in FRCs is the major driver for CCR2-dependent IgG production (Fig. 5I). Last, the importance of the connection between MyD88-dependent FRC activation, inflammatory monocyte recruitment, and swift activation of an antibacterial IgG response was validated by challenging Ccl19-Cre Myd88fl/fl mice with S. Typhimurium. The combined action of innate immune cells and B cells is essential for controlling systemic spread of this pathogen (34). The significant exacerbation of S. Typhimurium bacteremia and liver colonization in mice lacking MYD88 expression in FRCs (Fig. 5J) confirms that innate immunological sensing in FALC FRCs is a key driver for the initiation of antibacterial immunity in the peritoneal cavity and thereby permits shielding of the host from excessive bacterial spread.

Fig. 5 Regulation of antibacterial B cell responses in the peritoneal cavity via MyD88 signaling in FRCs.

Enumeration of IgM or IgG ASCs in the omentum (A) and peritoneal cavity (B) in mice lacking MYD88 expression in FRCs or in Cre-negative littermates (Ctrl) using ELISPOT analysis at the indicated days post-intraperitoneal (p.i.) immunization with S. Typhi OmpC/F. (C) Flow cytometry–based assessment of germinal center B cells in the indicated mouse strains after OmpC/F intraperitoneal immunization. Representative dot plot analysis on day 4 after immunization (left) and time course analysis of pooled data (right). (D and E) Serum titers of S. Typhi OmpC/F-specific IgM (D) and IgG (E) antibodies on day 4 after immunization. (F to H) C57BL/6 mice were treated with CCR2-specific depleting antibody (MC-21) or isotype control antibody (isotype), and B cell responses were analyzed at day 4 after immunization. (F) Flow cytometry–based assessment of germinal center B cells and enumeration of IgM or IgG ASC in the omentum (G) and peritoneal cavity (H). (I) ASC in the omentum of the indicated bone marrow chimeric mice on day 4 after immunization with S. Typhi OmpC/F. (J) Bacterial load in blood and liver of mice lacking MYD88 expression in FRCs or in Cre-negative littermates (Ctrl) on day 4 after intraperitoneal infection with 5 × 105 CFU of attenuated S. Typhimurium. Data are shown as mean values ± SEM and are pooled from 4 to 10 mice per group from two (D to H) or three (A to C, and I and J) independent experiments; statistical analysis was performed using Student’s t test (A to H, and J) or one-way ANOVA with Dunnett’s multiple comparison test (I) with *P < 0.05, **P < 0.01, and ***P < 0.001.

TNF-dependent interaction between FRCs and myeloid cells

The development of classical secondary lymphoid organs depends on the interaction of embryonic type 3 innate lymphoid cells or B cells with nonhematopoietic stromal cells through triggering of LTβR, RANK, and/or TNFR1/2 pathways (7, 9, 12, 35). Likewise, both TNF and LT signals can contribute to the development and shaping of FALCs (6, 31). To further dissect the signals that govern the interaction of FRCs and myeloid cells in visceral adipose tissues and to determine the impact of a disrupted communication between the two cell populations after intraperitoneal administration of S. Typhi OmpC/F, we blocked TNFR and/or LTβR signaling through application of the respective Ig fusion proteins. Although blocking of the LTβR alone did not affect the rapid changes in the myeloid cell compartment of FALCs, we found that blockade of TNFR signaling either alone or in conjunction with LTβR-Ig altered the composition of myeloid cells (fig. S6A). Enumeration of Ly6C+ inflammatory monocytes revealed a significantly impaired recruitment of this cell population in terms of both proportion (Fig. 6A) and absolute numbers (Fig. 6B), indicating that the communication between FRCs and myeloid cells is driven mainly via TNF. Neutralization of TNF in Ccl19-EYFP mice reduced the expansion of FRCs (Fig. 6C) and FALCs in the omentum (Fig. 6D) after OmpC/F immunization. Moreover, TNF deprivation deterred class switching in OmpC/F-specific B cells with significantly reduced numbers of IgG-producing B cells in the omentum (Fig. 6E) and the peritoneal cavity (Fig. 6F). Likewise, the expansion of germinal center–like B cells was significantly impaired under TNFR-Ig treatment (Fig. 6G). To determine the directionality of TNF-dependent FRC–myeloid cell interaction, we generated bone marrow chimeric mice that lacked TNFR1/2 signaling either on hematopoietic or on stromal cells. The lack of TNF responsiveness in both compartments resulted in a pronounced impairment of myeloid cell recruitment determined as absolute numbers of CD11b+ cells (Fig. 6H) and CD11b+Ly6C+ inflammatory monocytes (Fig. 6I). Likewise, activation of the germinal center reaction (Fig. 6J) and IgG class switching (Fig. 6K) was significantly impaired when TNFR signaling was abrogated in hematopoietic or in nonhematopoietic cells. Overall, these data indicate that TNF serves as the main means of the communication between FRCs and myeloid cells in FALCs during bacterial antigen–driven immune activation.

Fig. 6 TNF-dependent communication between FALC FRCs and myeloid cells.

C57BL/6 (B6) (A, B, and E to G) or Cc19-EYFP (C and D) mice were treated with the indicated fusion proteins 1 day before immunization with S. Typhi OmpC/F. (A and B) Accumulation of CD11b+F4/80Ly6GLy6C+ inflammatory monocytes in the omentum; relative frequency (A) and absolute numbers (B) of inflammatory monocytes in the omentum. (C) Frequency of PDPN+EYFP+ cells in the omentum after TNFR-Ig treatment as determined by flow cytometric analysis. (D) Number of FALCs per square millimeter (left) and percentage of FALC-covered area in the omentum (right). Enumeration of IgM and IgG ASCs in the omentum (E) and the peritoneal cavity (F) and flow cytometry–based analysis of germinal center–like B cells (G) on day 4 after immunization in TNFR-Ig-treated mice. (H to K) Assessment of CD11b+ myeloid cell (H) and inflammatory monocyte recruitment (I) and accumulation of germinal center B cells (J) and IgG-producing B cells (K) in the omentum of the indicated bone marrow chimeric mice on day 4 after S. Typhi OmpC/F immunization. Data are shown as mean values ± SEM and are pooled from 7 to 8 mice from three independent experiments (B, C, and E to G), from 4 to 6 mice from two independent experiments (C and D), and from 7 to 12 mice from three independent experiments (H to K); statistical analysis was performed using Student’s t test (C to G) or one-way ANOVA with Dunnett’s multiple comparison test (A, B, and H to K) with *P < 0.05, **P < 0.01, and ***P < 0.001.


Our study reveals a critical three-step process for the generation of protective immunity in the peritoneal cavity that is directed by a distinct fibroblastic stromal cell population. We found that recognition of bacterial products by FALC FRCs resulted in MYD88-dependent generation of inflammatory mediators including TNF and CCL2. Recruitment of CCR2-expressing inflammatory monocytes into FALCs and TNF-dependent interaction between FRCs and myeloid cells precipitated the rapid remodeling of the FALC infrastructure that facilitated immunoglobulin class switch and broadening of the antibody repertoire. The orchestration of these processes (i.e., bacterial recognition, immune cell recruitment, and FALC remodeling) through the activation of FRCs facilitated the generation of protective peritoneal immunity.

The mesothelial surface of the peritoneal cavity covers both the internal organs and visceral adipose tissues such as the omentum, which has long been known to react rapidly during immune responses in the peritoneal cavity (36, 37). Here, we found that the basic structure and the cellular composition of FALCs under homeostatic conditions depend on innate immunological signal integration in Ccl19-expressing FRCs. This finding is unexpected because MYD88 deficiency in FRCs affects neither the immune cell content nor the structure of Peyer’s patches when the microbiota is in equilibrium (15). Hence, it is likely that more than one cell type present in Peyer’s patches reacts to the various inflammatory stimuli present in the intestinal content. On the contrary, FRCs in FALCs appear to exert a more unique function because they constantly interrogate their environment for innate immunological signals and thereby maintain the integrity of FALCs in visceral adipose tissues. The microbiome of mice lacking Myd88 expression in FRCs is not different from cohoused Myd88-competent littermates (15). It is thus unlikely that the failure of FRCs to drive full elaboration of FALC structure and cellularity in Ccl19-Cre Myd88fl/fl mice was a consequence of a globally altered intestinal microbiome. Future studies will reveal how the intestinal microbiome (6) and/or microbial metabolites (38) exert remote control on FALC FRC phenotype and function.

The importance of innate immunological sensing by Ccl19-expressing FRCs in FALCs during immune activation could be unveiled through exposure to a mixture of microbial products in the peritoneal cavity, hence mimicking massive breach of bacterial content from the intestine. Although tissue-derived signals such as retinoic acid impinge on localization and functional polarization of resident peritoneal macrophages (39, 40), it is the exposure to microbial products that induces the recruitment of inflammatory monocytes from blood to the peritoneal cavity. The CCR2/CCL2 axis has been shown to direct these efficient antimicrobial effector cells into tissues to reduce pathogen load through the production of TNF and nitric oxide (41, 42). In accordance with previous studies (6, 18, 31), we found that omFALCs reacted instantaneously with recruitment of inflammatory monocytes after exposure to S. Typhi OmpC/F, leading to the subsequent accumulation of the cells in the peritoneal cavity. CCR2-dependent recruitment of inflammatory monocytes was profoundly impaired when MYD88-dependent innate immunological signaling was incapacitated in FRCs. Although it is possible that inflammatory monocytes contribute to the production of CCL2 once they have been recruited to an inflammatory lesion (43), the task of initiating powerful innate immune activation cascades in visceral adipose tissues appears to be restricted to a distinct population of immune-stimulating fibroblastic stromal cells.

Although natural IgM generated by B-1 B cells in the peritoneal cavity provides an important first layer of protection against bacteria that have escaped the gut (44, 45), enhanced production of IgG antibodies that bind their targets with higher specificity and can better engage immune effector pathways is required to halt bacterial invasion (46). Immune-stimulating fibroblastic stromal cells that regulate B cell responses are known as follicular dendritic cells and are marked by the expression of CXCL13 and the complement receptors CD21 and CD35 (47). Despite the lack of follicular dendritic cells, germinal center–like B cell responses can be mounted in FALCs (6, 31, 36), suggesting that other fibroblastic stromal cells substitute the functions of follicular dendritic cells. Here, we observed that S. Typhi OmpC/F induced a fast immunoglobulin class switch with appearance of substantial numbers of IgG-secreting B cells after only 4 days in FALCs and the peritoneal cavity. Moreover, we found not only that the germinal center–like response and the IgG class switch were controlled through MYD88 expression in FRCs but also that FRC-dependent B cell activation was linked to CCR2-dependent myeloid cell recruitment. As inflammatory monocytes promote local protective CD4+ T cell responses during infection (26), it is conceivable that the swift IgG production in omentum was facilitated through potent antigen presentation function of inflammatory monocytes. Overall, synchronized control of these processes through innate immunological signal integration at the locus of the FALC FRC grants swift and simple decision-making. In line with this interpretation is the revelation of an almost exclusive steering of the multicellular interactions in FALCs via the TNF/TNFR axis. TNF appears to act in the context of FALCs as a multifunctional tool that is produced by the main cell populations and that promotes all relevant processes during the initiation of the immune response. It is possible that other members of TNF superfamily members such as the LT-α trimer contribute to fine-tuning of immune responses in FALCs during later stages of antimicrobial immune responses (48). Nevertheless, the initiation of this powerful cascade is provided by the specialized stromal cell population of Ccl19-expressing FRCs.

The omentum harbors several myeloid and lymphoid cell populations that favor preservation of an immune-suppressive environment under steady-state conditions such as IL-10–producing B-1 B cells (49) or regulatory T cells (50, 51), which have been implicated in metabolic activities of the visceral adipose tissues. Thus, further elucidation of FRC functions in FALCs not only will help to identify targets that affect inflammatory responses in the body cavities but also may provide a better understanding of the regulatory pathways underlying metabolic processes in visceral adipose tissues.



C57BL/6 (B6) mice were purchased from Charles River Laboratories (Germany). B6.129P2(SJL)-Myd88tm1.1Defr/J (Myd88−/−), B6.129S4-Ccr2tm1Ifc/J (CCR2-deficient; Ccr2−/−), B6.129S-Tnfrsf1atm1Imx Tnfrsf1btm1Imx/J (TNFR1- and TNFR2-deficient; TNFR1/2 K.O.), and B6.129X1-H2-Ab1tm1Koni/J (MHCII-deficient; Iab−/−) mice were obtained from the Institute for Laboratory Animal Sciences at the University of Zürich. BAC-transgenic C57BL/6N-Tg (Ccl19-Cre)489Biat (Ccl19-Cre) mice (12) and C57BL/6-Tg(Aicda/EGFP)1Rcas/J (AID-GFP) mice (32) have been previously described. To specifically ablate Myd88 expression in FRCs, we crossed Ccl19-Cre mice with Myd88fl/fl mice (the Jackson Laboratory). All mice were on the C57BL/6 genetic background, were maintained in individually ventilated cages, and were used between 8 and 10 weeks of age.

For the generation of bone marrow chimeric mice, recipients were lethally irradiated with 9 gray from a linear accelerator (Clinic of Radio-Oncology, Kantonsspital St.Gallen, St. Gallen, Switzerland) and injected intravenously 1 day later with 2 × 107 of the indicated donor bone marrow cells. Chimeric mice were maintained on antibiotic water containing sulfadoxine and trimethoprim (Borgal; Veterinaria) for the following 4 weeks. Mice were used for experiments 10 weeks after bone marrow reconstitution. Experiments were performed in accordance with federal and cantonal guidelines (Tierschutzgesetz) under permission numbers SG10/16, SG07/16, and SG05/15 after review and approval by the Cantonal Veterinary Office (St. Gallen, Switzerland).

Bacterial infections

Mice were infected intraperitoneally with 5 × 105 colony-forming units (CFU) of the streptomycin-resistant S. Typhimurium ΔAroA. Bacteria were grown in an LB medium supplemented with streptomycin (50 mg/liter). Mice were sacrificed at the indicated time points, and organs were harvested at day 6 after infection, homogenized, and cultured on selective plates. Bacterial load was calculated as previously described (30).

Preparation of stromal cells

Omentum or mesentery was obtained and transferred into a 24-well dish filled with RPMI 1640 medium containing 2% fetal calf serum (FCS), 20 mM Hepes (all from Lonza), collagenase D (1 mg/ml; Sigma-Aldrich), deoxyribonuclease I (25 μg/ml; Applichem), and Dispase (Roche). Dissociated tissues were incubated at 37°C for 30 min. After enzymatic digestion, cell suspensions were washed with phosphate-buffered saline (PBS) containing 0.5% FCS and 10 mM EDTA. For in vitro assays, omental or mesenteric fibroblastic stromal cells were cultured for 7 days in RPMI 1640/10% FCS and 5 × 104 cells were stimulated with lipopolysaccharide (LPS; 1 μg/ml), OmpC/F (10 μg/ml), or Pam3CSK (15 μg/ml) or left untreated (medium). Supernatants were collected after 24 hours, and TNF and CCL2 concentrations were determined using cytometric bead array (BD Biosciences).

Flow cytometry

Single-cell suspensions were incubated for 20 min at 4°C in PBS containing 1% FCS and 10 mM EDTA with the indicated labeled antibodies (table S1). Cells were acquired with a FACS Canto or a BD LSRFortessa (BD Biosciences) and analyzed using FlowJo software (Treestar Inc.).

Histology and image acquisition

Omentum or mesentery was fixed overnight in freshly prepared 4% paraformaldehyde (Merck) at 4°C under agitation. Fixed tissues were blocked in PBS containing 10% FCS, anti-Fcγ receptor (1 mg/ml; BD Biosciences), and 0.1% Triton X-100 (Sigma-Aldrich). Sections were incubated overnight at 4°C with the following antibodies: anti-B220, anti-CD31 (eBioscience), anti-SMA (Sigma-Aldrich), and anti-EYFP (Clontech). Unconjugated antibodies were detected with the following secondary antibodies: Dylight649-conjugated anti-rat IgG, Alexa Fluor 488–conjugated anti-rabbit IgG, Dylight549-conjugated anti–Syrian hamster IgG, and Dylight549-conjugated streptavidin (all purchased from Jackson Immunotools). Microscopic analysis was performed using a confocal microscope (Zeiss LSM 710), and images were processed with ZEN 2010 software (Carl Zeiss Inc.) and Imaris (Bitplane).

Antibody detection

For measurement of anti-OmpC/F antibody titers, high-binding 96-well polystyrene plates (Cornings, New York, NY, USA) were coated with 1 μg of S. Typhi OmpC/F per well. The assay was performed as previously described (22, 23). Serum antibody titers are given as −log2 dilution × 40. Positive titers were defined as 3 SD above the mean values of the negative controls. For the determination of specific IgG and IgM antibody levels in peritoneal fluid, 1:2 dilutions of the fluid were incubated in the plates; values are given as absorbance values (A = 492 nm). ELISPOT assays were performed following the manufacturer’s instructions (Mabtech AB). Plates with 1 μg of S. Typhi OmpC/F per well were incubated for 24 hours at 37°C with 105 cells from peritoneal lavage or omentum from OmpC/F-immunized or naive mice. Plates were counted using an ELISPOT reader and analyzed with the software ELISPOT 3.1SR (AID). Values are expressed as mean number of specific antibody-forming cells (experimental sample, naive control).

Purification of OmpC/F and immunization protocol

OmpC/F were purified from S. Typhi American Type Culture Collection 9993 as previously described (52). LPS content was determined using the limulus amoebocyte lysate (LAL) assay (Endosafe KTA, Charles River Endosafe Laboratories), and all batches were negative for LAL assay (detection limit, 0.2 ng LPS/mg protein). Western blot analysis using anti-LPS polyclonal sera confirmed that LPS was not detectable by these means as previously shown (22, 52). Mice were immunized intraperitoneally on day 0 with 10 μg of the S. Typhi OmpC/F.

Blockade of LTβR and TNFR and depletion of inflammatory monocytes

Mice were administered with 100 μg per mouse of the fusion proteins LTBR-Ig and TNFR-Ig or the control Ig (MOPC-21) before immunization as previously described (53, 54). Mice were analyzed at day 4 after immunization. For inflammatory monocyte depletion, mice were administrated intraperitoneally with 40 μg per mouse of the MC-21 anti-CCR2 depleting antibody or the MC-26 isotype control (33) at days −1 and +1 of OmpC/F immunization.

Quantification of FALCs in the omental tissue

Omental tissues from the genetic ablated and littermate control mice were fixed and stained, as described. B cell clusters containing more than 20 B220+ cells are regarded as a B cell follicle, and the total number of B cell follicles was normalized to an area of 1 mm2 of the omental tissue based on the calculation of DAPI+ (4′,6-diamidino-2-phenylindole–positive) signal acquired using ZEN 2010 software (Carl Zeiss Inc.). Similarly, the area covered by the B cell follicles was calculated as the sum of B220+ signal within B cell follicles and normalized to the area of omental tissue acquired using the ZEN 2010 software (Carl Zeiss Inc.).

RNA-seq analysis of omental FRCs

EYFP+ cells from naive and OmpC/F day 4 immunized Ccl19-EYFP and Ccl19-EYFP Myd88fl/fl omental tissues were separately sorted with a Bio-Rad S3 cell sorter and were collected in 200-μl Eppendorf tubes containing 100 μl of RNAlater reagent (Qiagen) to preserve the RNA after sorting. RNA extraction was performed with Quick-RNA Micro-Prep (Zymo Research). Reverse transcription and cDNA library generation were carried out using the Ovation SoLo RNA-Seq System (NuGEN) as described by the manufacturer. cDNA libraries were quantified by the KAPA Library Quantification kit (KAPA Biosystems) and Agilent Tape station in the Functional Genomics Center Zurich (Zurich, Switzerland). Sequencing of the cDNA libraries was performed on a Illumina HiSeq 2500 by the Functional Genomics Center Zurich.

Single-end reads [126 nucleotides (nt)] were trimmed with Trimmomatic v0.36 (55) and flexbar v2.5 (56) to remove adapters and low-quality bases. The trimmed reads were aligned to the mouse genome (GRCm38) with STAR v2.5.1b (57) and deduplicated with NuDup v2.3. Reads that start at the same genomic coordinate and have the same strand orientation and 8-nt molecular tag sequence were considered duplicates. The deduplicated BAM files were converted back to FASTQ format using bedtools bamtofastq v2.17.0, and duplicate read entries were removed. A transcriptome index was generated from the combined cDNA and noncoding RNA sequence files (Ensembl GRCm38.90), and Salmon v0.9.1 (58) was used to estimate transcript abundances, which were read into R (v3.4.2) and summarized on the gene level with tximport (v1.6.0), together with average transcript length offsets (59). Five samples with low quality or strong evidence of contamination were excluded from further analysis. For differential expression analysis, only genes with estimated CPM above 0.5 in at least two samples were retained, and a generalized linear model with sample type as predictor was fit to each gene using edgeR (v3.20.8) (60). Pairwise tests for differential expression between groups were performed using edgeR’s LRT framework. P values were adjusted for multiple comparisons using the Benjamini-Hochberg method, and genes with adjusted P value below 0.05 were considered significantly differentially expressed. An MDS plot was generated from the logCPM values (calculated with the CPM function of edgeR with a prior count of 2) of the genes that were differentially expressed in at least one comparison, using the plotMDS function of the limma package (v3.34.8) (61). The camera method (62) in the limma package was used to perform gene set analysis for each contrast, considering the C2, C5, and C7 gene set collections from MSigDB v5.2, downloaded from

Statistical analysis

Statistical analyses were performed with GraphPad Prism 7.0 using an unpaired two-tailed Student’s t test. Longitudinal comparison between different groups was performed with one-way analysis of variance (ANOVA) with Tukey’s post test or Dunnett’s post test. Statistical analyses are indicated in figure legends. Statistical significance was defined as P < 0.05.


Fig. S1. Impact of FRC-specific MYD88 ablation on mesenteric FALCs.

Fig. S2. OmpC/F immunization–induced immune responses in the omentum.

Fig. S3. Quality control of FRC sorting for transcriptional profiling.

Fig. S4. Mechanisms of FRC-dependent myeloid cell recruitment.

Fig. S5. Regulation of antibacterial B cell responses in the peritoneal cavity via MYD88 signaling in FRCs.

Fig. S6. Myeloid cell recruitment to omental FALCs under conditions of TNF and LT ablation.

Table S1. Antibodies used in this study.

Table S2. Raw data sets.

Movie S1. Organization and interaction of FRCs in omental FALCs.


Acknowledgments: We thank R. De Guili and A. De Martin for excellent technical support. Funding: This study received financial support from the Swiss National Science Foundation (grants 166500 and 159188 to B.L.); the Consejo Nacional de Ciencia y Tecnología (CONACYT), Mexico (SEP-CONACYT CB-2015-256402 to C.L.-M.); and FIS/IMSS/PROT/G17/1682 to C.L.-M. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Author contributions: B.L. designed the study, discussed data, and wrote the paper; C.P.-S. and C.G.-C. designed the study, performed experiments, and wrote the paper; L.O., H.-W.C., A.P., U.M., and E.S. performed experiments and discussed data; S.J.T., M.B.B., J.G., C.L., C.L.-M., and M.M. discussed data and provided reagents; M.N., C.S., and M.D.R. performed bioinformatics analyses and discussed data. Competing interests: C.L.-M. holds a patent on porins as vaccine (patent: MX346872) and as adjuvant (patent: MX292850). The rest of the authors declare that they have no competing financial interests. Data and materials availability: RNA-seq data have been deposited in the ArrayExpress database at EMBL-EBI ( under accession number E-MTAB-6886. Further information and requests for resources and reagents should be directed to and will be fulfilled by the corresponding author, B.L. (burkhard.ludewig{at}
View Abstract

Navigate This Article