Bifidobacterium dentium Fortifies the Intestinal Mucus Layer via Autophagy and Calcium Signaling Pathways

Microbe-host interactions in the intestine occur along the mucus-covered epithelium. In the gastrointestinal tract, mucus is composed of glycan-covered proteins, or mucins, which are secreted by goblet cells to form a protective gel-like structure above the epithelium. Low levels of mucin or alterations in mucin glycans are associated with inflammation and colitis in mice and humans. Although current literature links microbes to the modulation of goblet cells and mucins, the molecular pathways involved are not yet fully understood. Using a combination of gnotobiotic mice and mucus-secreting cell lines, we have identified a human-derived microbe, Bifidobacterium dentium, which adheres to intestinal mucus and secretes metabolites that upregulate the major mucin MUC2 and modulate goblet cell function. Unlike other Bifidobacterium species, B. dentium does not extensively degrade mucin glycans and cannot grow on mucin alone. This work points to the potential of using B. dentium and similar mucin-friendly microbes as therapeutic agents for intestinal disorders with disruptions in the mucus barrier.

a similar degree as B. longum subsp. longum, indicating the comparable ability of B. dentium to adhere to mouse MUC2. Additionally, B. dentium colocalized with MUC2 in the mucus-producing human cell line LS174T, as observed by immunofluorescence and scanning electron microscopy (SEM) (Fig. 1C and D). These data indicate that B. dentium, similar to other well-characterized bifidobacteria, is able to adhere to mouse and human MUC2.
Since B. dentium does not encode GH33, which removes terminal sialic acid groups and provides access to underlying carbohydrates in the glycan, we hypothesized that B. dentium would not be able to release carbohydrate residues from MUC2 and use them as a growth source. To address this question, B. dentium was grown in a fully defined medium (lactic acid bacteria defined medium IV [LDMIV]) with the carbon source (100 mM glucose) removed (Fig. 2O). Supplementation of purified germfree cecal MUC2 or MUC2 derived from the human mucin-producing cell lines LS174T or HT29-MTX at 1 mg/ml did not enhance B. dentium growth in LDMIV without glucose. Next, we examined whether B. dentium could grow in minimal medium in the presence of free oligosaccharides that are typically found in mucin (Fig. 2P). The addition of mannose and, to some degree, galactose was able to restore the growth of B. dentium. However, B. dentium was unable to grow on fucose, N-acetylglucosamine, N-acetylgalactosamine, or sialic acid as primary carbon sources. These data further support the idea that B. dentium does not extensively degrade mucin glycans and cannot use them as a fuel source.
B. dentium colonizes the intestinal mucus layer in vivo. Based on our in vitro data demonstrating that B. dentium adheres to purified MUC2 and existing literature demonstrating the mucus-modulatory effect of Bifidobacterium species, we next examined the effect of B. dentium on the intestinal mucus layer in a gnotobiotic model. Adult germfree Swiss Webster mice (male, n ϭ 21; female, n ϭ 23) were orally gavaged with either sterile de Man-Rogosa-Sharpe (MRS) medium, live B. dentium MRS cultures (2 ϫ 10 8 bacteria), or heat-killed B. dentium ( week. To prevent cross-contamination, mice were housed in separate isolators. For comparison, age-matched Swiss Webster specific-pathogen-free (SPF) mice that have a complex murine microbiota were included as controls. Microbe colonization was examined by tissue Gram staining, fluorescence in situ hybridization (FISH), quantitative real-time PCR (qPCR), and conventional plating on bacteriologic medium (Fig. 3). Gram stains of SPF mice indicate a robust microbial community residing in the mucus layer above the epithelium (Fig. 3A). Multiple bacterial morphologies were present in the SPF mice, including cocci and rods (Fig. 3A, inset). Microbial distribution was confirmed by FISH with a probe recognizing diverse bacteria (universal 16S rRNA gene-based probe) (Fig. 3B). In contrast to SPF mice, germfree mice exhibited no bacterial stains by Gram stain or FISH, while B. dentium-monoassociated mice exhibited classic bifid-shaped microbes in the mucus layer adjacent to the intestinal epithelium ( Fig. 3A and B). Negligible bacterial staining was observed by Gram stain or FISH in mice treated with heat-killed B. dentium ( Fig. 3A and B). Colonization of B. dentium was further confirmed using microbiologic plating of stool on MRS plates and qPCR of stool extract genomic DNA (gDNA) with B. dentium-specific primers. Data obtained by qPCR and microbial cultures with CFU resulted in calculations of approximately 7.2 ϫ 10 7 Ϯ 1 ϫ 10 7 bacteria per gram in B. dentium-monoassociated mouse stool, with no observable CFU for germfree or heat-killed B. dentium groups. These findings indicate that live B. dentium can stably colonize the mucus layer of germfree mice.
Gut microbes are known to influence morphological parameters, including weight, cecum size, intestinal parameters, and development of the lamina propria (99)(100)(101). To determine if colonization by B. dentium influences overall mouse weight, mice were weighed on day 17 after monoassociation (Fig. 4A). No significant differences were observed in terms of mouse body mass between germfree, live, or heat-killed B. dentium-monoassociated groups (day 17 germfree, 40.5 Ϯ 5.6 g; live B. dentium, 42.5 Ϯ 4.8 g; heat-killed B. dentium, 39.5 Ϯ 5.3 g). Germfree mice are characterized by an accumulation of mucins, proteins, and carbohydrates which results in abnormally large cecum size (102,103). The morphologic difference between germfree and conventionally colonized mouse cecum is associated with the absence of mucus-degrading are presented in terms of copy number for glycosyl hydrolase (GH) families in the whole genome (A) and predicted to be involved in O-linked (B to L) and N-linked (M and N) mucin degradation. (O) B. dentium was grown in LDMIV with or without glucose or purified MUC2 from germfree cecal contents or human goblet cell line LS174T and HT29-MTX. (P) B. dentium was also grown without glucose in the presence of carbohydrates found in mucins. n ϭ 4 groups/treatment, repeated 4 independent times. *, P Ͻ 0.05, one-way ANOVA.  intestinal bacteria (104,105). Cecal weights revealed that B. dentium monoassociation did not restore cecum size, as both germfree, live, and heat-killed B. dentium-colonized mice exhibited significantly longer ceca than their conventionally colonized counterparts (Fig. 4B). This finding is consistent with the prediction that B. dentium cannot extensively degrade mucins, and several species are likely to be necessary to modulate mucus sufficiently to affect cecum size.
To assess the effects of B. dentium colonization on intestinal morphology, we examined intestinal dimensions and architecture from hematoxylin and eosin (H&E) staining ( Fig. 4C and D). No significant differences were observed for colon diameters or colon lengths between any of the groups (data not shown). H&E staining found no differences in crypt depths between germfree, live, or heat-killed B. dentium-colonized mice. However, H&E staining indicated that B. dentium-colonized mice exhibited an increase in filled goblet cell numbers per crypt relative to germfree mice and heat-killed B. dentium-treated mice ( Fig. 4C and D).
Goblet cells are modulated by B. dentium colonization. Next, we sought to define the effects of B. dentium monoassociation on goblet cell maturation and function. Compared to germfree mice and heat-killed B. dentium controls, B. dentium-colonized mice exhibited increased expression of Krüppel-like family of zinc-finger transcription factor 4 (Klf4), a goblet cell-specific differentiation factor (106) (Fig. 5A). Goblet cells are known to produce a number of important mucosal defense factors, such as resistin-like molecule-beta (Relm-␤) and trefoil factor 3 (Tff3) (107). Both Relm-B and Tff3 expression levels were increased in B. dentium-colonized colons compared with those of germfree or heat-killed B. dentium controls (Fig. 5A). MUC2 is the major gel-forming mucin protein secreted by intestinal goblet cells. Similar to other markers of goblet cells, Muc2 mRNA was also increased in B. dentium-monoassociated mice compared with germfree and heat-killed B. dentium controls (Fig. 5A). Of note, a trend toward increased expression of Relm-␤ and Muc2 was observed in heat-killed B. dentium-treated mice.
Although not significant, this may point to a complementary role of B. dentium surface proteins in modulating goblet cells. Several cytokines are associated with increased mucus production, most notably interleukin-22 (IL-22), IL-33, and IL-13 (108)(109)(110)(111). Interestingly, no changes were observed in any of these mucin-regulating cytokines (Fig. 5B), indicating that B. dentium may be modulating mucin production through secreted factors that affect the epithelium, rather than by stimulating the immune system.  Consistent with increased St6gal1, which appends on sialic acid, B. dentium-colonized mice exhibited increased amounts of ␣-2,6-linked sialic acid in intestinal mucins compared with germfree and heat-killed B. dentium controls ( Fig. 6C and F). These results indicate that B. dentium alone cannot degrade terminal sialic acid and thus cannot efficiently degrade the mucus layer.
B. dentium stimulates autophagy-mediated Ca 2؉ signaling and mucus expulsion. Based on our in vivo data, which demonstrated significant changes in MUC2 production in response to live B. dentium, we speculated that secreted factors play a role in modulating goblet cell function. Similar to other bifidobacteria, the B. dentium genome harbors the machinery required to generate acetate (65, 120) (KEGG Pathway, map 01120), a short-chain fatty acid that has also been shown to stimulate Muc2 expression (121,122). We have also previously shown that B. dentium secretes 3.4 Ϯ 0.4 mg/ml GABA (87), and other groups have found that activation of GABA type A (GABA A ) receptors stimulates the release of stored mucin granules (123). To determine if B. dentium secretes factors that increase MUC2 synthesis, we utilized the T84 intestinal epithelial cell line, which can form goblet-like cells with MUC2-containing secretory granules (124). T84 cells harbor GABA receptors (125,126) and an acetate receptor (GPR43) (127), allowing us to examine the effects of these compounds on MUC2 levels. The addition of B. dentium-conditioned medium resulted in increased MUC2 by immunostaining compared to medium controls (Fig. 7A). Moreover, the addition of acetate, but not GABA or heat-killed B. dentium, increased MUC2 staining. At the mRNA level, the addition of increasing concentrations of B. dentium-conditioned medium, as well as 5 mM acetate, resulted in increased levels of MUC2 (Fig. 7B). This result indicates that B. dentium can secrete products, such as acetate, that are capable of increasing MUC2 synthesis.
In addition to synthesizing MUC2, goblet cells release stored MUC2 granules in response to the proper stimuli. Work by Patel et al. has demonstrated that autophagy is required to elicit Ca 2ϩ signaling within intestinal goblet cells (128), and that defects in autophagy (ATG5 deficiency) result in a loss of Ca 2ϩ mobilization and accumulation of mucin granules within goblet cells. In parallel with these findings, multiple studies  have found that Ca 2ϩ signaling is required for the release of mucin-filled vacuoles (128)(129)(130)(131). We examined a number of known autophagy markers by qPCR and found that B. dentium colonization was associated with increased expression of ATG16L, ATG5, Beclin1, LAMP2, and LC3 (Fig. 8A). To determine if B. dentium was capable of eliciting Ca 2ϩ signaling and MUC2 release, we first created mucin-producing T84 cells that stably expressed the cytoplasmic genetically encoded Ca 2ϩ indicator GCaMP6s using lentivirus transduction (132). We then incubated the T84 GCaMP6s cells with B. dentium FluoroBrite conditioned Dulbecco's modified Eagle medium (DMEM) and measured the relative cytoplasmic Ca 2ϩ concentration using epifluorescence microscopy ( Fig. 8B and C). The addition of B. dentium-conditioned medium significantly increased Ca 2ϩ levels ( Fig. 8C). Mucin release was quantified by the release of CFDA-SE-tagged proteins at 1 h ( Fig. 8D and E) or by alcian blue staining of T84 cell supernatants after 24 h (Fig. 8F). Corresponding to Ca 2ϩ signaling data, B. dentium-conditioned medium elicited mucin release ( Fig. 8D to F). The results obtained with B. dentium conditioned medium (CM) were comparable to results obtained with calcium ionophore A21387. Notably, the intracellular calcium chelator BAPTA-AM [1,2-bis(2-aminophenoxy)ethane-N,N,N=,N=tetraacetic acid tetrakis(acetoxymethyl ester)] prevented the secretion of mucin. Moreover, pretreatment with the autophagy inhibitor 3-methyladenine (3-MA) inhibited B. dentium stimulated-mucus release, indicating that B. dentium requires both autophagy and Ca 2ϩ mobilization to promote mucin secretion ( Fig. 8D and F). We also demonstrated that GABA stimulates mucin expulsion, an effect which can be partially inhibited by 3-MA and completely inhibited by BAPTA-AM ( Fig. 8E and F). Acetate and heat-killed B. dentium had no effect on mucin expulsion ( Fig. 8E and F), indicating that B. dentium-secreted products, including GABA, are capable of stimulating the release of MUC2 from goblet cells. Together, this work provides evidence that B. dentium is unable to degrade mucin glycans or proteins to an appreciable extent to support growth but can stimulate host mucin production and secretion to bolster the mucus barrier. This work highlights the unique nature of B. dentium as a "mucin builder" and potential therapeutic for mucin-related disorders.

DISCUSSION
Herein, we describe the first gnotobiotic mouse model with a mucin-building, human-derived Bifidobacterium species, B. dentium. By studying monoassociated mouse models with human gut microbes adept at modulating intestinal mucus, we have gained new insights into the microbe-mammal interface. We demonstrate that B. dentium adheres to intestinal mucus and colonizes the mucus layer in vivo, but it does not harbor the genetic machinery required to significantly degrade the mucin glycans. We demonstrate that the presence of B. dentium correlates with significant increases in mucin gene expression as well as increased quantities of MUC2 protein. B. dentiumsecreted products, such as acetate, were able to increase MUC2 production in vitro. B. dentium also releases products which regulate intestinal goblet cells, including GABA, which stimulates autophagy-mediated Ca 2ϩ signaling and MUC2 release. These findings are supported by our in vivo findings of increased autophagy gene expression in B. dentium-colonized mice. We postulate that these factors work together in concert to mediate the mucin-building effects of B. dentium. Together, these findings suggest that B. dentium may be a desirable mucin builder Bifidobacterium species for the treatment of mucin-related disorders.
Our work is the first to provide conclusive evidence of Bifidobacterium species directly modulating MUC2 and glycosylation in vivo. Bifidobacteria are known to colonize the intestinal mucus layer, and this location has been speculated to allow for the optimal delivery of microbial metabolites to promote mucin production (43, 44, 59-61, 63, 133-135). Several studies point to the role of bifidobacteria in mucin maintenance in rats and mice with a complex gut microbiota (24)(25)(26)(27)136). The contributions of specific gut microbes is difficult to decipher in SPF animals, making the development of gnotobiotic mouse models attractive for studies of microbiome-mucus layer biology. Our work demonstrates that B. dentium is able to adhere to intestinal MUC2 and expands prior studies by reporting that B. dentium produces compounds that stimulate MUC2 synthesis. Our in vivo gnotobiotic mouse model and our in vitro T84 human cell model support this conclusion. In support of a specific role of secreted metabolites in stimulating mucus production, we found no differences in the cytokines known to stimulate intestinal mucus production (108,109,(137)(138)(139)(140). Additionally, our data using heat-killed B. dentium indicate that B. dentium surface components do not play a large role in modulating MUC2 in vivo or in vitro. Although lipoteichoic acids (LTA) from Gram-positive bacteria and other Toll-like receptor (TLR) agonists have been reported to modulate mucin production (141-147), we did not observe significant changes in MUC2 levels in response to B. dentium surface proteins. Single-cell sequencing indicates that mouse goblet cells exhibit low TLR2 expression compared to other cell types (Single Cell Portal; Broad Institute). We speculate that this observation may account for our lack of significant goblet cell modulation by our heat-killed bacteria. Our in vitro work in human T84 cells also indicated that heat-killed B. dentium has no effect of MUC2 synthesis or expulsion. This finding may be due to our T84 cell line model, and further studies are needed to assess B. dentium-mediated TLR activation in human goblet cells.
Based on our data, we speculate that B. dentium enhances mucus production via the secretion of microbial metabolites such as acetate and GABA. B. dentium does not contain the genes responsible for generating butyrate or propionate, but it is capable of synthesizing acetate (KEGG). As such, we speculate that the MUC2-modulating effects of B. dentium may be due in part to acetate production. Acetate may work through multiple pathways. According to single-cell sequencing of mouse intestinal epithelium, the acetate receptor GPR43 (also known as FFAR2) is highly expressed in goblet cells (Single Cell Portal; Broad Institute), and acetate may directly modulate Engevik et al. ® GPR43 to stimulate MUC2 production. Additionally, acetate is known to lower the intestinal pH (148,149), and the organization of MUC2 in goblet cells is triggered by low pH (150). Moreover, enteroendocrine cells in the intestine can be stimulated by low pH or GPR43 activation to secrete serotonin, which is also known to influence goblet cells (151)(152)(153)(154). In addition to acetate, it is possible that B. dentium secretes other products that regulate MUC2 production. Another lactic acid-producing bacterium, Lactobacillus rhamnosus GG, secretes a 40-kDa protein, termed p40, which has 27 to 78% homology with CHAP domain-containing proteins secreted by various Bifidobacterium strains, including B. dentium (155). These compounds may also be involved in B. dentiummediated mucin modulation.
In addition to promoting MUC2 synthesis, we demonstrate that B. dentium metabolites stimulate autophagy-mediated Ca ϩ2 signaling and mucin release. MUC2 secretion and the renewal of the colonic mucus layer are essential for the protective function of mucins. Goblet cells can release stored mucin granules in response to autophagy and Ca 2ϩ -mobilizing agents, such as acetylcholine or other cholinergic agonists (144,(156)(157)(158)(159)(160)(161). Patel et al. found that autophagy was required for mucin release, and ATG5-deficient mice were unable to release MUC2 (128). Moreover, another autophagydeficient mouse model, ATG7 knockout mice, exhibited a diminished mucus layer and was more susceptible to dextran sulfate sodium (DSS)-induced colitis (161). We are the first to link bacterium-promoted autophagy with goblet cell mucin expulsion. Our work shows that B. dentium can stimulate autophagy-driven Ca 2ϩ signaling. We also demonstrate that GABA is capable of eliciting mucin secretion, an effect which can be minimized by blocking the autophagy pathway by 3-MA. Interestingly, RELM-␤, a compound secreted by goblet cells, has also been shown to act as a mucin secretagogue in the mouse colon and HT29-Cl.16E human gut cell lines (162). RELM-␤ is upregulated in response to B. dentium colonization in our model and may serve to further promote MUC2 expulsion in vivo.
Finally, we demonstrate that B. dentium is unique among several well-characterized human Bifidobacterium strains, as it is unable to extensively degrade mucin glycans. Although the CAZy database does not have GHs annotated for our B. dentium ATCC 27678 strain, a BioCyc Database Basic Local Alignment Search Tool (BLAST) search shows 99 to 100% homology between mucin-degrading GHs in B. dentium ATCC 27678 and B. dentium Bd1 (UniProt D2Q7E7 [GH29], D2Q5B6 [GH125], D2Q767 [GH2], and D2Q935 [GH42]), indicating that the Bd1 genome can be used as a proxy for ATCC 27678 analysis. Based on in silico analysis and in vitro experiments, Bifidobacterium species do not produce mucinases that can cleave the MUC2 mucin protein core (163). However, many bifidobacteria do synthesize mucin carbohydrate-cleaving glycosyl hydrolases (65,164). These human gut species include Bifidobacterium scardovii, B. longum subsp. longum, B. longum subsp. infantis, B. breve, and B. bifidum (165). B. dentium does not contain genes of the GH33 sialidase family or genes of the GH101 or GH129 families that enable cleavage of mucin glycans from the mucin protein. Despite the fact that the B. dentium genome contains more GHs than the average Bifidobacterium species (165), the B. dentium genome only contains a limited assortment of mucin-related GH genes (GH2, GH42, GH29, and GH125). The majority of the B. dentium GH enzymes are predicted to utilize and convert various intracellular or target dietary compounds, such as cellobiose, xylose, ribose, arabinose, mannitol, and sorbitol (136). Interestingly, the glycosyl hydrolases harbored by B. dentium may be advantageous in early life. B. dentium is frequently isolated from healthy breast-fed infants (39,84,85,166). Breast milk glycans commonly contain galactose residues (167), and B. dentium harbors the galactosidases GH2 and GH42. As a result, B. dentium may use milk glycans during postnatal human development to establish a niche.
Consistent with the larger repertoire of mucin-degrading glycosyl hydrolases, previous studies have shown that mucin-maintaining or mucin-degrading species, such as B. bifidum (PRL2010, D119, and L22), B. breve NCIMB8807, and B. longum NCIMB8809, can grow in the presence of mucin alone (72, 75, 168-170). Our data indicate that B. dentium does not harbor the same glycosyl hydrolase genes and is unable to grow in mucin alone. Under pathological conditions with a diminished gastrointestinal mucus layer, a relative preponderance of mucin builders versus mucin degraders may be beneficial. Mucin-building bifidobacteria that enhance mucin production without mucin-degrading abilities would be preferred under specific biological conditions. B. dentium has the distinct ability to upregulate MUC2 synthesis, influence host glycosylation patterns, and promote MUC2 secretion, coupled with the inability to degrade mammalian host mucins. Thus, we speculate that B. dentium may be beneficial in intestinal environments lacking sufficient mucin.
One caveat to our work is that several bifidobacterial species have been identified in dental caries (120,(171)(172)(173)(174)(175)(176)(177)(178)(179)(180)(181). These bifidobacteria include gut-associated strains, such as those of B. breve, B. adolescentis, B. longum, and our microbe of interest, B. dentium. It has been postulated that bifidobacteria exist as stable species in dental caries due to their adhesive properties and their resistance to acidity (173,182,183). However, the results of clinical studies assessing the effects of bifidobacteria on the oral microbiota are controversial. Currently, no studies have dissected the role of bifidobacteria in the pathogenesis of caries. Of note, studies have also shown that bifidobacteria, including B. dentium, inhibit the growth of Porphyromonas gingivalis in in vitro biofilm models (184), which may have a positive effect on subgingival biofilm and thereby may enhance gingival health. One theory, the specific plaque hypothesis, has proposed that only a few specific species, such as Streptococcus mutans and Streptococcus sobrinus, are actively involved in the disease (185). While it is possible that Bifidobacterium species may be indirectly contributing to caries via acid production, the role of bifidobacteria within these carries remains unclear. More studies are needed before it is possible to draw conclusions on bifidobacteria in dental caries.
B. dentium, similar to other Bifidobacterium strains, is a recognized member of the healthy infant and adult human gut microbiota, according to the Human Microbiome Project (86) and sequencing studies (39,84,85,166). Our data indicate that B. dentium is a commensal bacterium of the intestine and harbors beneficial mucin-modulating properties. Collectively, these data point to the potential role of next-generation mucin-building probiotics, such as B. dentium, in rescuing intestinal mucus layer function and treating intestinal disorders in the future.

MATERIALS AND METHODS
Materials. The monosaccharides D-glucose and D-mannose were purchased from Sigma-Aldrich. L-Fucose, N-acetyl-D-glucosamine (GlcNAc), N-acetyl-D-galactosamine (GalNAc), and N-acetylneuraminic acid (Neu5Ac) were purchased from Carbosynth Limited. All other reagents were from Sigma-Aldrich, unless otherwise stated.
Bacteriology. Bifidobacterium dentium ATCC 27678 (adult human fecal isolate), Bifidobacterium longum subsp. infantis ATCC 15697 (infant intestinal isolate), Bifidobacterium longum subsp. longum ATCC 55813 (adult human fecal isolate), and Bifidobacterium breve ATCC 15698 (infant intestinal isolate) (ATCC, American Type Culture Collection) were cultured in an anaerobic workstation (Anaerobe Systems AS-580) in a mixture of 5% CO 2 , 5% H 2 , and 90% N 2. Colonies were grown in de Man-Rogosa-Sharpe (MRS) medium (Difco) anaerobically at 37°C overnight. Single colonies were isolated from cultures plated on MRS agar plates. B. dentium was adjusted to optical density at 600 nm (OD 600 ) of 0.1, subcultured into a fully defined medium termed lactic acid bacteria defined medium IV (LDMIV) (see Table S1 in the supplemental material), and incubated for up to 48 h anaerobically at 37°C. Bacterial growth was measured by the OD 600 and CFU counts on MRS agar. For T84 cell treatment, B. dentium was grown overnight in MRS anaerobically at 37°C, and cells were pelleted by centrifugation at 5,000 ϫ g for 5 min. Bacterial cells were washed three times with sterile anaerobic phosphate-buffered saline (PBS) to wash away residual MRS, and the bacterial pellet was resuspended in anaerobic optically clear FluoroBrite DMEM (catalog no. A1896701; Thermo Fisher) and incubated for 6 h. Following the incubation, cultures were centrifuged to remove bacteria, the pH was adjusted to 7, and the cultures were filtered through a 0.2-m-pore-size polyvinylidene difluoride (PVDF) membrane (Millipore) to sterilize the supernatant, termed "conditioned medium." For animal experiments, B. dentium was grown anaerobically in MRS to log-phase growth (1.6 ϫ 10 9 cells ml Ϫ1 ). Bacterial viability was confirmed for each gavage session by serially plating B. dentium and assessing CFU via conventional plating on MRS agar.
Gnotobiotic mouse model. All experimental procedures and animal care were approved by the Institutional Animal Care and Use Committee (IACUC) at Baylor College of Medicine, Houston, TX. Swiss Webster germfree mice were purchased from Taconic and housed in the Baylor College of Medicine germfree facility. Mice were housed in filter-top cages in sterile isolators, and all food, water, and bedding were irradiated for sterility. Feces from experimental mice and isolator sentinels were collected at various time points and examined routinely for colonization by plating anaerobically and aerobically on Engevik et al. ® blood agar. Male and female adult 7-to 9-month-old mice were housed under gnotobiotic (germfree) conditions. Germfree mice were monoassociated by oral gavage with 3.2 ϫ 10 8 CFU ml Ϫ1 B. dentium ATCC 27678 grown in MRS (total, n ϭ 33; males, n ϭ16; females, n ϭ17). Control germfree mice received sterile MRS gavages (total, n ϭ 31; males, n ϭ 14; females, n ϭ17) or 3.2 ϫ 10 8 CFU ml Ϫ1 heat-killed B. dentium (males, n ϭ 5; females, n ϭ 5). For heat-killed bacteria, B. dentium was heated at 60°C for 30 min. Culture viability was confirmed by plating on MRS agar.
Cultures were maintained in sterile Hungate tubes until administration to the mice to maintain anaerobic conditions. Mice received oral gavage doses once every other day for 1 week and one final gavage a week later. Fecal samples were plated on MRS and blood agar (Hardy Diagnostics) over time to monitor colonization. Plates were incubated anaerobically and aerobically at 37°C. To ensure that the germfree mice harbored no bacteria, gDNA was extracted from stool samples before the start of the experiment and at the time of euthanasia and examined by qPCR using a 16S rRNA gene-based universal bacterial probe for bacterial colonization. Weight was monitored over the course of treatment, and mice were euthanized 72 h after the last gavage.
Intestinal tissue staining. (i) Immunofluorescence. Mouse tissue segments were embedded in paraffin, and 7-m sections were processed for staining. Following dehydration, slides were incubated for 20 min at 100°C in antigen unmasking solution citrate buffer pH 6 (Vector Labs) in a steamer for antigen retrieval. Sections were blocked for 1 h at room temperature in 10% donkey and/or goat serum. Staining was performed with an anti-MUC2 antibody (dilution 1:200, rabbit anti-MUC2, catalog no. sc-15334, Santa Cruz Biotechnology Inc.) and incubated overnight at 4°C. The primary antibody was recognized using donkey-anti-rabbit Alexa Fluor 488 antibody diluted at 1:300 (Life Technologies) and incubated for 1 h at room temperature. Hoechst 33342 (Invitrogen) was incubated at room temperature for 10 min to stain the nuclei. Coverslips were mounted with mounting medium (Life Technologies).
(ii) H&E, PAS-AB, and tissue Gram stain. Paraffin-embedded tissue sections were serially dehydrated in xylene and ethanol. Sections were stained with hematoxylin and eosin (H&E) for intestine architecture or periodic acid-Schiff-alcian blue (PAS-AB) to identify goblet cells and mucus. To identify bacterial colonization, sections were incubated with the following Gram staining reagents: crystal violet, Gram iodine, and safranin, with a series of washes with a decolorizer solvent, followed by picric acid-acetone. After alcohol and xylene dehydration steps, sections were coverslipped with Fluoromount aqueous mounting medium (Sigma-Aldrich catalog no. F4680) and imaged on an upright wide-field Nikon Eclipse 90i microscope. H&E and PAS-AB sections were recorded using a DS-Fi1-U2 camera (Nikon) with a 20ϫ Plan Apo (NA, 0.75) DIC objective.
(iii) Fluorescence in situ hybridization staining. B. dentium localization was examined in mouse colon sections using the Bifidobacterium-specific FISH probe Bif164 (5=-CATCCGGCATTACCACCC-3=), and total bacteria were examined using the universal bacterial FISH probe EUB338 (5=-GCTGCCTCCCGTAGG AGT-3=) (Integrated DNA Technologies [IDT]). Briefly, 7-m sections were dehydrated in ethanol and incubated with the Bifidobacterium probe at 45°C in a dark humidifying chamber. Slides were hybridized for 45 min and counterstained with Hoechst 33342 for 10 min at room temperature (Invitrogen). All slides were imaged on an upright wide-field epifluorescence Nikon Eclipse 90i microscope (Tokyo, Japan) with a 40ϫ Plan Apo (NA, 0.95) DIC objective. FISH images were recorded using a CoolSNAP HQ2 camera (Photometrics) using a Nikon Intensilight C-HGFI mercury lamp.
Germfree mouse mucus isolation. Germfree mouse cecum contents were collected (n ϭ 7) and pooled for mucus extraction. Mucus was prepared similarly to previously described methods (70,91). Briefly, pooled cecum samples were diluted in four volumes of ice-cold phosphate-buffered saline containing EDTA, phenylmethylsulfonyl fluoride, and iodoacetamide. Particulate matter was removed by centrifugation, and mucin protein was isolated by precipitation with ice-cold ethanol. Precipitated mucin was lyophilized and purified by CsCl isopycnic density gradient centrifugation at a starting density of 1.39 g ml Ϫ1 using a 70 Ti rotor (Beckman Coulter, Inc.) (188,189). Dialyzed mucins were lyophilized and resuspended in HEPES-buffered Hanks' salt solution (HH) at a concentration of 1 mg/ml. Mucin concentration was ascertained by a bicinchoninic acid (BCA) assay with purified pig stomach mucin (Sigma-Aldrich) as a standard. For adhesion assays, 100 l of mucus solution (1 mg/ml) was immobilized in 96-well polystyrene microtiter plates (Maxisorp, Nunc; VWR) by overnight incubation at 4°C, as previously described (70). Wells were washed 3ϫ with 200 l of HH prior to the addition of bacteria to remove unbound mucus.
Mucin adhesion assay. Bifidobacterium dentium ATCC 27678, Bifidobacterium infantis ATCC 15697, Bifidobacterium breve ATCC 15698, and Bifidobacterium longum ATCC 55813 were grown anaerobically in MRS. Bacterial suspensions were adjusted to an OD 600 of 2 in PBS and fluorescently tagged by incubating with 10 M carboxyfluorescein diacetate-succinimidyl ester (CFDA-SE; catalog no. C1157; Thermo Fisher) for 30 min at 37°C anaerobically, similar to previous bacterial staining reports (190,191). Fluorescent bacteria were washed 3 times with anaerobic PBS and resuspended in PBS at an OD 600 of 2. To determine bacterial adhesion, 100 l of labeled bacteria was added to the microtiter plates coated with germfree mouse mucus and 2-fold serially diluted in PBS. Following a 1-h incubation at 37°C, the wells were washed 3 times with 200 l of HH to remove unattached bacteria. Bacterial adherence was examined by reading fluorescence (excitation/emission, 492/517 nm) in a microtiter plate reader (Synergy H1; Biotek), and the presence of bacteria was confirmed by microscopy. The results from the adhesion assay are presented as means from four replicates in three independent experiments.
T84 culturing and treatment. (i) T84 cell culturing. Human colon-derived T84 (ATCC CCL-248) cells were obtained from the ATCC. Cells were regularly maintained in a complete growth medium (CGM) of Gibco Dulbecco's modified Eagle medium F-12 (Thermo Fisher) supplemented with 2 mM GlutaMAX (Thermo Fisher) and 10% fetal bovine serum (FBS). Cultures were maintained in a humidified atmosphere at 37°C and 5% CO 2 . Cells were routinely tested for Mycoplasma contamination using the Mycoplasma detection kit (catalog no. LT07-518; Lonza). For an examination of Muc2 mRNA levels, cells were seeded at 5 ϫ 10 4 cells/cm 2 in 12-well tissue culture-treated plates (Corning) until the cells reached confluence. Cells were then treated with various concentrations of B. dentium FluoroBrite conditioned DMEM, heat-killed B. dentium bacteria, ␥-aminobutyric acid (GABA), or 5 mM sodium acetate overnight in DMEM without glucose and without FBS. Following incubation (16 to 18 h), cells were treated with TRIzol reagent for RNA extraction.
(ii) T84 mucin expulsion assays. T84 cells were seeded at 5 ϫ 10 4 cells/cm 2 into a 24-well plate and incubated at 37°C and 5% CO 2 until the cells reached confluence. Cells were then serum starved overnight and treated with various amounts of B. dentium FluoroBrite conditioned DMEM, GABA, or acetate diluted in FluoroBrite DMEM with 1ϫ GlutaMAX (no FBS). To inhibit autophagy, cells were pretreated for 1 h with the PI3K inhibitor 3-methyladenine (3-MA; Sigma-Aldrich catalog no. M9281). As positive and negative controls, cells were treated with 10 M the calcium ionophore A21387, a known stimulator of mucus (124), or 25 M BAPTA, a Ca 2ϩ chelator which is known to inhibit mucus production (192,193). Supernatant (100 l) was collected at 1, 3, and 24 h and incubated with alcian blue, as previously described (194). Briefly, the supernatant was incubated with an equal volume of 1% alcian blue and 3% glacial acetic acid for 2 h at room temperature. Mucins were pelleted by centrifugation at 12,000 ϫ g for 10 min, and pellets were washed three times with PBS. Following washes, mucin pellets were resuspended in PBS containing 10% SDS and sonicated to release the alcian blue color. Absorbance was measured at 620 nm on a spectrophotometer. As an alternative method for analyzing mucus release, confluent T84 monolayers in 96-well plates were thoroughly washed with PBS and incubated with 10 M CFDA-SE in PBS for 30 min at 37°C and 5% CO 2 to fluorescently tag all proteins. Following the incubation, cells were washed three times in PBS and incubated in FluoroBrite DMEM for 3 h at 37°C and 5% CO 2 to allow the CFDA-SE to incorporate into all proteins. Cells were then treated with B. dentium FluoroBrite conditioned DMEM, GABA, acetate, 3-MA, A21387, or BAPTA, as described above. Supernatants (100 l) were collected at 1 h and examined by fluorescent plate reader at an excitation/emission of 492/517 nm.
(iii) Generation of T84 GCaMP6S cells for live-cell calcium imaging. For calcium (Ca 2ϩ ) imaging studies, T84 cells were cultured in 12-well plates to 50 to 60% confluence and transduced using lentivirus pLVX-GCaMP6s with 10 g/ml Polybrene (catalog no. TR-1003-G; Millipore) in CGM and incubated for 48 h. The lentivirus construct and packaging were previously described (195). Two days after transduction, cells were cultured in the presence of puromycin (10 g/ml) to select for stably expressing cells. For Ca 2ϩ measurements, T84 GCaMP6s were grown to confluence on 8-well chamber slides (Matek) and then changed to an optically clear FluoroBrite DMEM (Invitrogen) supplemented with 15 mM HEPES (Invitrogen), 1ϫ sodium pyruvate (Invitrogen), 1ϫ GlutaMAX (Invitrogen), and 1ϫ nonessential amino acids (Invitrogen). B. dentium FluoroBrite conditioned DMEM was added in uninoculated FluoroBrite DMEM (50% final concentration). Cells were placed in an Okolabs stage-top incubation chamber with CO 2 mixing and humidity control and placed on an Nikon TiE inverted wide-field epifluorescence microscope with a motorized X, Y, and Z stage for software-controlled multiposition imaging. Cells were imaged with wide-field epifluorescence using a 20ϫ PlanFluor (NA, 0.45) phase-contrast objective or a 20ϫ Plan Apo (NA, 0.75) differential interference contrast objective, using a Spectra X LED light source (Lumencor). Images were recorded using an ORCA-Flash 4.0 scientific complementary metal-oxide semiconductor (sCMOS) camera (Hamamatsu), and the Nikon Elements v4.5 software was used for data acquisition and image analysis. Following the experiment, cell viability was examined by staining cells with 3 M calcein AM (live-cell dye, catalog no. C3099; Invitrogen) and 2.5 M propidium iodide (dead-cell dye, catalog no. P4170; Sigma-Aldrich).
Scanning electron microscopy. The MUC2-producing human colonic carcinoma cell line LS174T ATCC CL-188 (ATCC) was grown in DMEM supplemented with 10% fetal bovine serum at 37°C and 5% CO 2 . Approximately 1 ϫ 10 5 cells were seeded onto Corning Costar 24-well culture plates containing poly-L-lysine-coated coverslips and grown to confluence. B. dentium was incubated with confluent coverslips at 2 ϫ 10 5 bacteria and incubated for 1 h anaerobically at 37°C. Coverslips were then washed thoroughly with PBS (3 times) and fixed in 2.5% glutaraldehyde in PBS for 1 h at room temperature. Coverslips were dehydrated with ethanol and coated in 20 nm of gold using a desktop sputtering system (Denton Desk II). Slides were viewed in a scanning electron microscope (FEI XL-30FEG) at 12 kV.

Quantitative real-time PCR.
A stool sample was collected to examine bacterial composition, and genomic DNA was isolated using the Zymo gDNA MiniPrep kit (catalog no. 11-317C; Zymo Research) with bead beating. Mouse colon cells were stored in TRIzol reagent, and RNA was extracted according to the manufacturer's details. The SensiFAST cDNA synthesis kit (Bioline USA, Inc.) was used to synthesize cDNA from 1 g RNA. cDNA was examined by quantitative real-time PCR (qPCR) via Fast SYBR green (Thermo Fisher) and primers (Table 1) using a QuantStudio 3 qPCR machine (Applied Biosystems). All reactions were carried out in 96-well plates with melting curves to ensure primer specificity. Relative mRNA levels were examined using the ΔΔCT method with the glyceraldehyde-3-phosphate dehydrogenase (GAPDH) housekeeping gene, generating fold changes for each gene.
Statistical analysis. Graphs were generated using the GraphPad Prism software (version 6; Graph-Pad, Inc.). For statistical analysis, comparisons were made with either one-way analysis of variance (ANOVA) with the Holm-Sidak post hoc test or with a t test using SigmaPlot. The data are presented as mean Ϯ standard deviation, with differences between the groups considered significant at a P value of Ͻ0.05.