Following a sterile birth, the gastrointestinal (GI) tracts of humans and all mammals coordinately assemble a diverse multitude of microorganisms, collectively known as the microbiota. It has been acknowledged for decades that many of these microorganisms live symbiotically with their hosts, performing beneficial functions such as metabolizing complex carbohydrates and providing essential nutrients [1]. Recent studies have shown that the microbiota critically augments the development and function of the immune system (reviewed in [2] and [3]). Although the microbial diversity in the mammalian gut is vast (with an estimated 500-1000 species of microorganisms present in the human), organisms belonging to the genus Bacteroides represent one of the most abundant microbial taxa in both mice and humans [4]. Bacteroides fragilis is a ubiquitous Gram-negative anaerobic bacterium that inhabits the lower GI tract of most mammals [5]. Recent findings have revealed that this organism possesses the ability to direct the cellular and physical maturation of the host immune system and to protect its host from experimental colitis [6], [7], [8]. Therefore, the contributions of the microbiota to human health appear to be profound.

We wanted to understand how B. fragilis colonizes the mammalian gut. B. fragilis expresses at least eight distinct surface capsular polysaccharides (CPS), and previous studies have shown that CPS expression by the bacterium is required for successful intestinal colonization [9], [10]. How these molecules mediate the initial interactions with the host, and whether they are involved in long-term persistence in the gut are currently unknown. Several mechanisms of gut colonization by symbiotic bacteria have been studied. Some organisms form biofilms, composed of a polymeric matrix secreted by the bacteria, which adhere to the epithelial layer. Others interact with components of the mucosal layer (reviewed in [11]). Mucus is a viscous stratum which separates epithelial cells from the lumen of the gut and acts as a crucial barrier against infection by pathogens. Various membrane-bound or secreted glycoproteins called mucins associate with one another to form the gel-like mucus. Interactions between gut bacteria and mucus have been hypothesized to be important for the assembly and stability of the microbiota [12]. Accordingly, we sought to determine whether or not B. fragilis binds intestinal mucus and purified mucin.

Initially, we visualized the spatial localization in the colon of 2 different commensal bacteria to determine potential differences in association with the mucus layer in vivo. Wild-type Bacteroides fragilis NCTC9343 was grown anaerobically in brain-heart infusion (BHI) supplemented with hemin (5 μg/ml) and vitamin K (0.5 μg/ml), and Escherichia coli BL21 was grown aerobically in BHI at 37°C. Bacteria were grown to OD₆₀₀ of 0.7-0.8 and 1×10⁸ colony forming units (CFUs) were orally inoculated into germ-free Swiss Webster mice by gavage. Following 1 week of colonization, mice were sacrificed and colon tissue was fixed in Carnoy's solution and embedded in paraffin wax for sectioning. Fluorescence in situ hybridization was performed on tissue sections mounted on glass slides using a 6-carboxyfluorescein (6-FAM)-labeled oligonucleotide probe for E. coli (EnterbactB [AAGCCACGCCTCAAGGGCACAA]) and a Cy3-labeled oligonucleotide probe for B. fragilis (Bfra602 [GAGCCGCAAACTTTCACAA]) (Integrated DNA Technologies, Inc.). Briefly, slides were deparaffinized, dried, and hybridized with both probes at 5ng/μl concentration each for 2 hours at 46°C in hybridization buffer (0.9 M NaCl, 15% formamide, 20mM Tris-HCl (pH 7.4), and 0.01% sodium dodecyl sulfate (SDS)). Slides were washed for 15 minutes at 48°C in wash buffer (20mM Tris-HCl (pH 7.4), 318 mM NaCl, and 0.01% SDS). For visualization of the colon epithelial cell nuclei, the slides were counterstained with 4′,6′-diamidino-2-phenylindole (DAPI). The autofluorescence background allowed visualization of the tissue structures. The slides were examined with an Axioplan microscope (Zeiss, Oberkochen, Germany) using a 100× oil immersion objective. Epifluorescence images of a cross section through the colon of gnotobiotic mice that were dual-colonized with both E. coli and B. fragilis reveal that both bacteria are found in the lumen of the gut in high abundance (Fig. 1). Surprisingly however, only B. fragilis is found in the mucus layer that lies between the lumen and the gut epithelium tissue (Fig. 1). The spatial segregation of B. fragilis and E. coli across the colon mucus barrier suggests that B. fragilis may interact with mucus in vivo and this may be important for sustained colonization of commensal B. fragilis. Furthermore, these results reveal that not all bacteria are equally able to penetrate the mucus layer, suggesting dedicated mucus associating functions for B. fragilis.

Open in a separate window

Fig. 1

Colon tissue section from B. fragilis and E. coli dual-colonized Swiss Webster mouse. Epifluorescence image of bacteria visualized by FISH, and the epithelial cells counterstained with DAPI (blue) to visualize DNA. Both E. coli (green) and B. fragilis (red) are found in the lumen but only B. fragilis is found in the mucus layer.

To test the hypothesis that B. fragilis colonization of the distal gut is mediated by mucus binding, a standard mucus binding assay was used to determine if live bacteria are able to bind a crude, intestinal mucus preparation. Crude mucus was isolated from the colon and cecum of conventionally-colonized Swiss Webster mice as described in Cohen et al [13]. Briefly, colonic and cecal mucus was scraped into HEPES-Hanks' Buffer (pH 7.4 with Calcium Chloride and Magnesium Chloride). Next, non-soluble material was removed by centrifuging once at 12,000 × g for 10 minutes at 4°C, and then once at 26,500 × g for 15 minutes at 4°C. The final concentration of the crude mucus solution was determined by the Bradford assay. The mucus was diluted with HEPES-Hanks' Buffer to 1mg/ml. 0.2 ml of mucus was added into the wells of a 24-well tissue culture plate and incubated overnight at 4°C. Controls included wells containing 0.2 ml of a 1mg/ml solution of Bovine Serum Albumin (BSA, which served as a specificity control) or 0.2 ml of HEPES-Hanks' Buffer (which served as a negative control). The wells were washed with HEPES-Hanks' Buffer to remove non-immobilized proteins. The plate was UV-sterilized for 10 minutes and was ready for use in the mucus binding assay. 1×10⁸ CFUs of bacteria were added to immobilized mucus, or BSA control, and incubated at 37°C for 1 hour. Wells were washed with HEPES-Hanks' Buffer, treated with 0.05% trypsin for 10 minutes at room temperature to liberate bacteria. One milliliter of cold BHI was added to quench the trypsin activity. Samples were serially diluted and plated for CFUs. Fig. 2A shows that B. fragilis binds to crudely purified mucus in vitro, as determined by recovered CFUs. The BSA- and buffer-containing wells illustrate low background binding. A mutant strain of B. fragilis (CPM1), which only expresses one of the eight CPS [9], is able to bind mucus as effectively as wild-type B. fragilis, suggesting that CPS expression does not mediate mucus binding. Therefore, B. fragilis specifically binds intestinal mucus via a mechanism that appears not to involve expression of multiple surface polysaccharides.

Open in a separate window

Fig. 2

B. fragilis binds intestinal mucus. (A) Number of B. fragilis (in CFUs) recovered after one hour incubation in wells with an immobilized mucus layer, an immobilized BSA layer, or buffer only. Of the 1×10⁸ CFUs incubated, 1.6×10⁶ (1.6%) bound to immobilized mucus. The CPM1 mutant binds mucus similarly to wild-type bacteria. These data are representative of four independent trials. (B) Number of bacteria recovered from mucus binding assay after a 2 hour pre-incubation with different concentrations of excess mucus. These data are representative of three independent trials.

Next, a mucus binding competition assay was performed to determine if the interaction between B. fragilis and mucus is saturable. We reasoned that as B. fragilis is pre-coated with higher concentrations of excess mucus, fewer putative receptors would be available to bind immobilized mucus in the well. Briefly, 1×10⁸ CFUs of B. fragilis were incubated with excess mucus at 37°C for 2 hours under aerobic conditions with shaking. Bacteria were washed and added to wells of a 24-well tissue culture plate containing immobilized mucus, BSA, or nothing (prepared as above). After 1 hour, samples were treated with trypsin and serially diluted, and plated for CFUs. Unexpectedly, pre-incubation with excess mucus appeared to increase B. fragilis binding to mucus with a bi-phasing profile (Fig. 2B). Binding to immobilized mucus reached a peak when B. fragilis was pre-incubated with 0.2 mg/ml of excess mucus. Pre-incubation of bacteria with excess mucus at concentrations higher than 0.2 mg/ml resulted in a decrease in mucus binding, yet binding was still higher than without pre-incubation with mucus. Pre-incubation of bacteria with 0.4mg/ml and 1mg/ml of BSA did not affect binding, once again showing that the B. fragilis-mucus interaction is specific (data not shown). These results suggest that bacteria pre-incubated with mucus (and not BSA) are increased in their ability to bind immobilized mucus until putative receptors are saturated at the highest mucus concentrations. Further experiments are required to determine if dedicated molecules on the bacterial surface mediate mucus binding.

Intestinal mucus is known to contain host molecules in addition to mucin, such as anti-microbial peptides, immunoglobulin A (IgA) antibodies, and lysozyme [13]. We wished to determine if mucus binding by B. fragilis was specific to mucin. As murine colonic mucin is not commercially available, we purified mucins from Swiss Webster mice based on the protocol by Shekels et al. [14] with a few modifications. Fig. 3 illustrates a schematic of this modified protocol and the analysis of mucin purity. We then tested the purified mucin and BSA for specific binding by B. fragilis. Purified mucin and the BSA control were labeled with Thermo Scientific DyLight Amine-Reactive Fluor 488, and unbound fluorophores were removed from the sample via dialysis against PBS. B. fragilis was pre-incubated with either unlabeled BSA or PBS and was subsequently incubated with labeled mucin or labeled BSA for 30 minutes at room temperature. The bacteria were washed after each incubation to remove non-adherent material. Percentage of mucin-binding bacteria in each sample was determined by flow cytometry (FC). When B. fragilis was incubated with fluorescently labeled BSA, no binding was detected (Fig. 4A). However, when B. fragilis was incubated with labeled mucin, a significant number of B. fragilis was detected by flow cytometry. Pre-incubation with BSA did not diminish the percentage of B. fragilis adherent to mucin (Fig. 4A). Taken together, B. fragilis binds specifically to purified murine colonic mucin and not to BSA.

Open in a separate window

Fig. 3

Schematic of the mucin purification protocol. Briefly, crude mucus was scraped from the colon and cecum of conventionally colonized 8-week-old male Swiss Webster mice into 0.1 M NH₄HCO₃, 0.5 M NaCl, and a cocktail of protease inhibitors on ice. The sample was then homogenized and centrifuged at 45,000 × g for 45 minutes at 4°C. Non-soluble material was removed before centrifuging again at 45,000 × g for another 45 minutes at 4°C. The supernatant was taken and dialyzed against 10 mM Tris pH 8.0 + 150 mM NaCl for about 24 hours. Next, the sample was sonicated at eight 15-second pulses with intermediate 1 minute cooling on ice on a Brason Sonicator at speed 3 to break up large aggregates and then centrifuged once more at 45,000 × g for 45 minute. Next HPLC was employed whereby the supernatant was size fractionated on a XK 26/70 column containing Sephacryl S-400 resin (equilibrated in 10 mM Tris, pH 8.0). The void volume (which contained the large mucin glycoproteins) was collected and dialyzed against water for about 36 hours and then lyophilized. The lyophilized glycoproteins were resuspended in a solution containing RNase A and DNase I and digested for 2 hours at room temperature. After the digestion, the sample was centrifuged at 27,000 × g for 30 minutes at 4 °C and the supernatant was dialyzed against phosphate buffered saline (PBS) for 36 hours. Cesium chloride was added to the dialyzed supernatant to a final concentration of 0.54 g/ml, and then centrifuged at 160,000 × g for 72 hours. One milliliter fractions were collected and analyzed with the Pro Q Emerald Glycoprotein Staining Kit to determine which fractions contained the purified mucins. The mucin-containing fractions were pooled, dialyzed against water for 24 hours, lyophilized, and then stored at -20°C. Positive fractions from gel filtration chromatography were identified by absorbance readings at 280 nm. CsCl fractions and final product were assayed to contain mucin by glycoprotein staining (data not shown).

Open in a separate window

Fig. 4

B. fragilis binds soluble murine and porcine mucin. (A) Flow cytometry plots indicating percentage of B. fragilis bound to fluorescently labeled murine colonic mucin. Cells were either pre-incubated with BSA or not (1°), and secondary incubations were with fluorescently labeled BSA or mucin (denoted by asterisk). Percentages represent bacteria bound to fluorescently-labeled mucin relative to total number of bacteria analyzed per sample. These data are representative of two independent trials. (B) Percentage of B. fragilis bound to fluorescently labeled porcine mucin with no pre-incubation. These data are representative of two independent trials. (C) Percentage of B. fragilis bound to fluorescently labeled porcine gastric mucin following pre-incubation with unlabeled mucin (left) or unlabeled BSA (right). Porcine mucin was purchased from a commercial source and purified as described in Figure 3 from the RNase/DNase digestion step. These data are representative of two independent trials.

B. fragilis colonizes the intestines of most mammalian species studied to date [5]. In order to determine if mucin interactions extend beyond the murine host, we examined the ability of B. fragilis to bind porcine mucin. Starting with partially purified porcine gastric mucin purchased from Sigma Aldrich, we purified mucin to homogeneity using the same protocol as described above. Fig. 4B shows B. fragilis binding to fluorescently labeled purified porcine mucin as significant amount of mucin-binding bacteria were detected by flow cytometry. Both approaches we used in this study to demonstrate mucus binding resulted in only a small portion of bacterial binding (~1.6% for the immobilized plate assay and ~1.5% for the soluble mucin binding assay). This is consistent with the known ability of B. fragilis to be highly phase variable whereby only a portion of the bacterial population express a given surface molecule [15]. Fig. 4C shows that pre-incubation with 1.0 mg/ml of unlabeled mucin was able to compete with the fluorescently-labeled mucin, resulting in a lower percentage of bacteria binding to the fluorescently labeled mucin. Pre-incubation with BSA shows no inhibition (Fig. 4C), serving as a specificity control. Our results show that B. fragilis specifically binds porcine mucin in addition to murine mucin.

B. fragilis has emerged as a model symbiont for the study of host-microbial interactions with the immune system [3]. The mechanism by which B. fragilis maintains long-term colonization of the mammalian intestine remains unknown. Associations with mucus may involve bacterial binding, and/or nutrient utilization of mucin for bacterial growth. If binding to mucus is involved during the colonization process in vivo, we predict that B. fragilis would express defined and dedicated receptor(s) with specific affinity for mucin. Along these lines, the B. fragilis genome and other sequenced Bacteroides species express numerous homologs of the SusC/SusD proteins, which are known to bind starch and other carbohydrates that decorate the mucin glycoproteins [16]. Furthermore, SusC/SusD proteins of B. fragilis were recently shown to be phase variable [17]. This property is similar to the phase variability of capsular polysaccharides, whereby only a small fraction of bacteria express any one of the eight CPS of B. fragilis [9]. If mucus binding is also phase variable, this would explain why only a small percentage of bacteria invade the mucus layer (as shown in Fig. 1), and why only a small fraction of bacteria bind mucus and mucin in vitro (as shown in Figs. 2 and ​and4).4). A non-mutually exclusive function for mucus binding may be the use of host derived sugars as a carbon source. Several studies have shown that B. fragilis can degrade mucin and utilize it as a nutrient source for growth [18], [19]. In fact, B. fragilis can utilize porcine mucin as a sole source for carbon and nitrogen [20], and structural analysis of the SusD homolog of Bacteroides thetaiotaomicron (also found in B. fragilis) suggests it binds sugars liberated from mucin glycoproteins [21]. Therefore, mucus binding may serve as a physical mechanism for sustained colonization, as a means to degrade and import nutrients into the bacterial cell for growth, or both. We have shown here that B. fragilis specifically binds intestinal mucin (although B. fragilis may also bind to other components in the mucus) and associates with the mucus layer in vivo. These findings, along with previous work, suggest that specific interactions between B. fragilis and mucus are relevant for in vivo colonization of animals. The identity of dedicated mucin binding receptor(s), and a molecular mechanism during long-term association of the mammalian gut, await discovery.