Indigenous opportunistic bacteria inhabit mammalian gut-associated lymphoid tissues and share a mucosal antibody-mediated symbiosis

Presence of Indigenous Opportunistic Bacteria on the Interior of PPs.

To determine the bacterial composition at the surface and on the interior of PPs in naïve, specific-pathogen-free (SPF) mice, we first used the 16S rRNA gene clone library method. Consistent with a previous report (10), segmented filamentous bacteria were the predominant species detected on the surface of the follicle-associated epithelium covering PPs (Fig. 1A). In contrast, several species of indigenous microbiota, including Alcaligenes spp., Ochrobactrum spp., Serratia spp., and Burkholderia spp., were detected on the interior of PPs. Of these, Alcaligenes, which are opportunistic bacteria (11), were dominant (72%; Fig. 1A).

To confirm the presence and localization of Alcaligenes on the interior of PPs, we next performed a whole-mount FISH analysis to identify the bacterial distribution in this tissue (12). The microbial cells were visualized by three distinct probes used in several previous studies (12–14) (Table S1). EUB338 is routinely used for detecting bacterial species in an indiscriminate manner (12). ALBO34a is a specific probe for Alcaligenes and Bordetella (13), and BPA is for Alcaligenes, Burkholderia, and Comamonas (14). Thus, Alcaligenes are identified as ALBO34a and BPA double-positive cells.

Consistent with the 16S rRNA analysis (Fig. 1A), EUB338-positive bacteria morphologically similar to segmented filamentous bacteria were observed over the entire surface area of PPs covered by wheat germ agglutinin positive (WGA⁺) epithelial cells (Fig. 1B). ALBO34a and BPA double-positive Alcaligenes were detected on the interior of PPs, where WGA⁺ epithelial cells were not observed (Fig. 1B). Sequential analysis through the z axis convincingly showed that Alcaligenes were present on the interior of PPs (Movie S1). We also confirmed the presence of Alcaligenes by the PCR method in a separate study using the 16S rRNA-gene-targeted group-specific PCR primers for Alcaligenes.

In contrast to the preferential localization of Alcaligenes in PPs, this species was essentially absent in the diffuse lamina propria (LP) region of the small intestine (Fig. 1B), whereas EUB338-positive bacteria were scattered throughout the surface layer of the LP (Fig. S1A). Thus, although some antigen-sampling cells [e.g., villous M cells (15) and epithelial DCs (16)] are located in the epithelium covering the more diffuse LP region, it seems that antigen-sampling M cells and DCs in the follicle-associated epithelium of PPs are responsible for the entry of Alcaligenes. Furthermore, the presence of Alcaligenes inside PPs was demonstrated to be a common feature by the characterization of different species of mice housed in various SPF-maintained experimental animal facilities (Fig. S1B). These findings suggest a possibility that commensal bacteria live within the tissues of the organized lymphoid structures associated with the GI tract.

Alcaligenes-Ingested PP-DCs Migrate into Mesenteric Lymph Nodes but not Spleen.

We next investigated the fate of Alcaligenes inhabiting PPs, and particularly their interactions with mucosal immunocompetent cells. When the microbial populations within DCs purified from different tissues were characterized by the 16S rRNA analysis, Alcaligenes were detected within PP-DCs and mesenteric lymph node (MLN) DCs (Fig. 2A) but not splenic DCs (Fig. S2). Our findings support the presence of a restricted PP-MLN axis for migration of DCs that have taken up indigenous microbiota and suggest that MLNs act as reinforcement to help prevent intrusions by indigenous microbiota into the systemic compartment (17). By using FISH analysis, we also found substantial numbers of Alcaligenes in the MLNs of SPF mice (Fig. 2B).

To investigate whether PP-DCs are the main source of MLN-DCs harboring Alcaligenes, PP-null mice were generated by in utero treatment with an anti-IL-7 receptor α chain mAb (18). In PP-null mice, negligible numbers of Alcaligenes were detected in their MLNs (Fig. 2B); these bacteria presumably originated from isolated lymphoid follicles (ILFs) (Fig. S1C and Movie S2), which resemble PPs and still develop in PP-null mice (19). This result was identical to previous reports showing that PPs are the major sites for uptake of orally inoculated bacteria and the subsequent induction of host immune responses (e.g., Salmonella typhimurium and Helicobacter pylori) (20, 21).

Preferential Induction of Alcaligenes-Specific Mucosal Ab Responses for the Establishment of Symbiosis.

To elucidate whether the intratissue presence of Alcaligenes and their uptake by PP-DCs affect intestinal mucosal Ab responses, we next examined IgA Ab responses to Alcaligenes because IgA is the major isotype of mucosal Abs (4). We used Alcaligenes faecalis subsp. faecalis NBRC (National Institute of Technology and Evaluation Biological Resource Center) 13111^T, which was the predominant species in the PPs (Fig. S3A), for the analysis of antigen-specific immune responses. Substantial amounts of Alcaligenes-specific IgA Abs were detected in the feces of SPF mice, whereas GF mice failed to produce this isotype of antigen-specific Abs (Fig. 3A, Left). No serum IgG Abs specific for Alcaligenes were seen in either SPF or GF mice (Fig. 3A, Right). This result reflected the localization of Alcaligenes in PPs, a major mucosal Ab-inductive lymphoid tissue, and not spleen, where systemic IgG Ab responses predominate (Fig. 1 and Fig. S2).

In agreement with this finding, an enzyme-linked immunospot (ELISPOT) assay showed that naïve, SPF mice possessed Alcaligenes-specific IgA Ab-forming cells (AFCs) in their intestinal compartments, including PPs and the LP region, but not in the spleen (Table 1). Additionally, no Alcaligenes-specific IgG-AFCs were seen in MLNs or spleen (Table 1). Alcaligenes-specific IgA-AFCs were more commonly observed in the PPs than in the LP region: more than 2% of IgA-AFCs in the PPs were reactive to Alcaligenes, whereas only approximately 0.5% of IgA-AFCs in the LP were specific for Alcaligenes (Table 1). This tissue-specific pattern of Alcaligenes-specific IgA-AFCs was further confirmed by FACS analysis using GFP-Alcaligenes (Fig. S3B): 5.3% of IgA-positive B cells (including 2.3% of IgA plasmablasts) were specific for Alcaligenes in the PPs, whereas only 1.1% of IgA-positive B cells in the LP were specific for this bacterium (Fig. S3B). In addition, when we examined LP-homing properties of local IgA class-switched (or IgA committed) B cells in PPs, Alcaligenes-specific IgA⁺ B cells expressed fewer gut-homing receptors (α4β7, CCR9, and CCR10) than the rest of the PP-IgA⁺ B cells (Fig. S3C). Therefore, Alcaligenes-specific IgA-committed B cells most likely remained in PPs, which accounted for the presence of elevated Alcaligenes-specific IgA-AFCs in PPs compared with LP.

Some intestinal IgA Abs are derived from B1 B cells and recognize T cell-independent antigens commonly expressed by commensal bacteria. Thus, it is possible that Alcaligenes-specific IgA Abs show some cross-reactivity with other commensal bacteria. We tested this possibility by FACS analysis and found that Alcaligenes-specific Abs did not cross-react with other bacteria (e.g., Escherichia coli; Fig. S4A). This view was further supported by the analysis of Alcaligenes-specific IgA mAb (#3E-12A-6D-3G) developed by fusion of B cells from the PPs of SPF mice. This mAb did not cross-react with E. coli. In addition, impaired intestinal IgA Ab responses to Alcaligenes were noted in TCRβ^−/− δ^−/− mice (Fig. S4B). These data suggest that Alcaligenes-specific IgA Abs are mostly derived from B2 B cells producing T cell-dependent, antigen-specific Abs. This agrees with the evidence that PPs are major sites for the induction of intestinal mucosal Ab responses to T cell-dependent microbial antigens regardless of whether the microbes are commensal or pathogenic (4).

Although PPs are thought to play a major role in the induction of IgA-committed B cells and plasmablasts, but not plasma cells (4), these data suggest that a large part of Alcaligenes-specific fecal IgA Abs are derived from PP IgA-producing cells in a T cell-dependent manner. In fact, markedly decreased levels of anti-Alcaligenes fecal IgA Abs were seen in PP-null mice (Fig. S4C). These findings are in agreement with previous reports demonstrating that PP-DCs are involved not only in the class-switching of IgM⁺ B cells to IgA⁺ ones and the determination of gut-tropism via retinoic acid synthesis (22, 23), but also in regulating IgA secretion in the PPs through the stimulation signal provided by the Ab-enhancing cytokine IL-6 (24). We examined IL-6 production by PP cells from GF mice after treatment with Alcaligenes and found that Alcaligenes induced mainly PP-DCs to produce substantial levels of IL-6 (Fig. S5A). When PP-DCs were isolated from WT mice and cocultured with Alcaligenes, the synthesis of the IgA isotype-switching cytokines TGF-β and B-cell-activating factor belonging to the TNF family (BAFF) were also elevated in addition to IgA-enhancing cytokine IL-6 (Fig. S5B).

Taken together, these findings suggest that mucosal Abs, including locally produced, antigen-specific IgA Abs, may play a critical role in the intratissue cohabitation of Alcaligenes in PPs. Supporting this view, Alcaligenes numbers were much lower in the PPs of CBA/N xid mice, which exhibit a B cell defect, than in WT mice (Fig. 3B and Fig. S6A). Further, Alcaligenes levels tended to be lower also in PPs of IgA-deficient mice, although no statistically significant differences were observed (Fig. S6B). Because the IgA-deficient condition did not lead to the complete removal of PP intratissue Alcaligenesis, it is also possible that Alcaligenes-specific IgA Abs may not be fully involved in the presence of Alcaligenes in PPs. Alternatively, this lack of significant differences may offer another explanation due to the compensation of IgA function by IgM Abs in deficient mice because the numbers of anti-Alcaligenes IgM-AFCs was much increased in IgA-deficient mice when compared with WT mice (Fig. S6C).

Ability of Alcaligenes to Colonize the Interior of PPs.

Intratissue cohabitation of Alcaligenes in PPs should be addressed formally and directly by the establishment of a gnotobiotic mouse model monoassociated with Alcaligenes. The current technology, however, does not permit the isolation and culture of Alcaligenes from PPs. Previous studies have shown that Alcaligenes have the distinctive feature of being resistant to multiple antibiotics (25, 26), suggesting to us a unique strategy to directly assess the presence of intratissue Alcaligenes in PPs. By isolating PPs from antibiotic-treated mice under sterile conditions for the preparation of homogenized tissue and its subsequent oral administration to GF mice, we were able to establish PP-derived, Alcaligenes-associated mice. When we examined the antibiotic-treated mice, no bacteria were seen at the intestinal epithelial surface (including the follicle-associated epithelium), whereas Alcaligenes were present inside PPs (Fig. S7A). Three weeks after oral inoculation, Alcaligenes were again noted on the interior of PPs of ex-GF mice (Fig. 3C). The colonization of Alcaligenes in the PPs of ex-GF mice was further supported by the presence of antigen-specific fecal SIgA but not serum IgG Abs (Fig. 3D). A significant increase in antigen-specific IgA- but not IgG-AFCs was also observed in these mice (Table 1). Furthermore, the levels of total IgA were partially increased in the Alcaligenes-associated mice (Fig. S7B). When we examined PPs of GF mice, the numbers of total IgA-AFCs were 143 ± 45 per 10⁵ lymphocytes. On the other hand, the numbers of total IgA-AFCs in PPs isolated from both SPF and the monoassociated mice were 1,304 ± 364 and 625 ± 307, respectively (Table 1). A similar tendency was also seen when total IgA levels were examined in fecal samples taken from monoassociated, GF, and SPF mice (Fig. S7B). These findings further suggest that the intratissue habitation of Alcaligenes in the PPs may contribute to not only the induction of Alcaligenes-specific IgA but also the development of at least a portion of mucosal IgA-associated humoral immunity.

Alcaligenes Were Present on the Interior of Monkey and Human PPs.

On the basis of the findings demonstrated by a variety of mouse experiments as described above, we next examined the presence of Alcaligenes inside PPs of higher mammals, namely nonhuman primates and humans. This bacterium was observed on the interior of monkey PPs by FISH analysis (Fig. 4A, Left), and anti-Alcaligenes IgA Abs were also detected in the feces of these monkeys (Fig. 4A, Right). To further demonstrate the intratissue habitation of Alcaligenes in monkey PPs, an Alcaligenes-specific mAb (#11E-8C-7A, IgM isotype) was developed. Immunohistochemical analysis with Alcaligenes-specific mAb #11E-8C-7A showed the presence of this bacterium on the interior of primate PPs (Fig. 4C, Left). When human PPs were obtained from noninflamed sites of healthy patients who underwent endoscopic biopsy, the intratissue habitation of Alcaligenes was demonstrated inside human PPs by FISH analysis (Fig. 4B, Left). In addition, anti-Alcaligenes fecal IgA Abs were also detected in human fecal samples (Fig. 4B, Right), consistent with the murine and nonhuman primate studies (Fig. 3A, Left and Fig. 4A, Right), The intratissue habitation of Alcaligenes in human PPs was further confirmed by the use of Alcaligenes-specifc mAb #11E-8C-7A (Fig. 4C, Right).

Animals and Human Samples.

BALB/c and C57BL/6 mice were obtained from CLEA Japan. CBA/N xid and control DBA/2 mice were purchased from Japan SLC. TCRβ^−/− δ^−/− mice were obtained from the Jackson Laboratory. IgA^−/− mice were originally generated by Dr. Gregory Harriman and were kindly provided by the Baylor College of Medicine. Mice were maintained under SPF conditions at the Institute of Medical Science, University of Tokyo and the Immunobiology Vaccine Center, University of Alabama at Birmingham (UAB). GF mouse experiments were performed at the Yakult Central Institute for Microbiological Research. All experiments were conducted in accordance with the guidelines for the Animal Care and Use Committees of the University of Tokyo and UAB.

Nonhuman primate PPs were obtained from cynomolgus macaques housed in the Tsukuba Primate Research Center (TPRC), National Institute of Biomedical Innovation (Tsukuba, Japan). All procedures were conducted in accordance with the guidelines for the Animal Care and Use Committees of the TPRC.

Human PPs were kindly provided by healthy patients without irritable bowel disease who underwent endoscopic biopsy at Osaka University Hospital. All of the subjects provided written informed consent, and the study protocol was approved by the Ethics Committee of Osaka University Graduate School of Medicine (approval no. 08243) and Institute of Medical Science, University of Tokyo (IMSUT) (approval no. 20-67-0331).

16S rRNA Analysis.

The 16S rRNA gene was amplified by PCR with two universal primers (27F: 5′-AGAGTTTGATCCTGGCTCAG-3′; 1492R: 5′-GGTTACCTTGTTACGACTT-3′) ligated into plasmid vector pCR2.1 and transformed into INVαF' competent cells by using a TA Cloning Kit (Invitrogen). Plasmid DNA of randomly selected transformants was prepared by using a TempliPhi DNA Amplification Kit (GE Healthcare) and sequenced by using the primers 27F and 520R (5′-ACCGCGGCTGCTGGC-3′). All sequences were examined by BLAST search to identify the closest relatives. Representative nucleotide sequences obtained in this 16S rRNA gene clone library analysis have been deposited in the International Nucleotide Sequence Database (accession nos. AB453241–AB453250).

Whole-Mount FISH Analysis.

To detect the domain Bacteria or Alcaligenes, oligonucleotide probes were purchased from Invitrogen-Molecular Probes (Table S1). Isolated tissue segments were fixed in 4% paraformaldehyde at 4 °C overnight and washed with PBS. Tissues were hybridized in hybridization buffer [0.9 M NaCl, 20 mM Tris-HCl, 45% (ALBO34a, BPA) or 0% (EUB338) formamide, 0.1% SDS, and 10 μg/mL DNA probe] at 60 °C (ALBO34a, BPA) or 42 °C (EUB338) overnight. After washing twice in washing buffer [0.45 M NaCl, 20 mM Tris-HCl, 45% (ALBO34a, BPA) or 0% (EUB338) formamide, and 0.01% SDS] at 60 °C (ALBO34a, BPA) or 42 °C (EUB338) for 10 min, tissue segments were flushed with PBS. Lectin-labeling experiments were performed Alexa Fluor 633-labeled WGA (Invitrogen-Molecular Probes) and biotinylated UEA1 (Vector Laboratories) followed by Alexa 633-conjugated streptavidin (Molecular Probes) at a concentration of 10 μg/mL for 1 h. After being washed with PBS, the tissue samples were mounted and examined by DM IRE2/TCS SP2 confocal microscopy (Leica Microsystems).

Statistical Analysis.

Data were expressed as the mean ± SD or SEM and evaluated by an unpaired Student's t test. Significance was defined as P < 0.01.