Spatial analysis of gut microbiome reveals a distinct ecological niche associated with the mucus layer

Figures & data

Figure 1. Dense band of bacteria forms overlying colonic mucosa

A. Colonic cross-section stained with fluorescent probes identifying all bacteria (green) and host epithelium (blue) to show dense community structure close to the host (Left: 10x; Right: 60x).B. Mean fluorescence intensity (MFI) measurements from a representative mouse (average N = 3, SEM) to show bacterial concentration of the dense community. Measurements start 10-μm before start of bacterial signal closest to epithelial lining.C. Colonic cross-section stained with Alcian Blue/Pas to show dense mucus layer (Left: 10x; Right: 60x).D. Relative frequency of dense mucus layer thickness measurements, described in Methods section under “Mucus Thickness” (N = 6 mice, 40 measurements each).

Figure 2. Mucus-associated dense community is compositionally distinct from adjoining sparse community

A. Diagram of laser capture microdissection (LCM) strategy for independently obtaining the inner dense community (1^st 100-μm) and outer sparse community (subsequent 50-μm) of the mucosal layer for 16S rRNA sequencing. B. Images of colonic cross-sections showing LCM masks of inner and outer communities followed by the independent extraction of each region separately in order to analyze regions separately. C. Copy number of 16S rRNA per sample from inner and outer communities showing enough material is obtained for sequencing, with more copies in the inner community (N = 6 mice, SD). *p < .05 paired t-test.D. Principal coordinates of analysis (PCoA) plots. Left: Weighted UniFrac shows most significant difference in microbial communities comes from cage effect, with no difference in location. Right: Unweighted UniFrac shows most significant difference in microbial communities comes from cage effect, but also the inner and outer communities are significantly different (N = 11 mice). **p < .01, ***p < .001 PERMANOVA. E. Class-level relative abundance of inner and outer communities (N = 5 mice from Cage 2, SEM).F. Differential abundance of class Gammaproteobacteria calculated using DESeq2, showing significantly more Gammaproteobacteria in the outer community (N = 5 mice from Cage 2, Log2 Fold Change = 4.87, padj = 1.39 x 10⁻⁷). ***p < .001.

Figure 3. Inner dense community has significantly higher species richness due to presence of low abundance unique amplicon sequence variants (ASVs)

A. Alpha rarefaction curves plotting the number of unique amplicon sequence variants (ASVs) from inner (red) and outer (blue) communities, showing more unique ASVs found in the inner community (N = 5 mice from Cage 2).B. Chao1 diversity index showing significantly higher richness in the inner community (N = 5 mice from Cage 2). ***p < .001 paired t-test.C. Shannon diversity index showing no difference in Shannon diversity between inner and outer communities (N = 5 mice from Cage 2).D. Pielou’s Evenness index showing significantly higher evenness in the outer community (N = 5 mice from Cage 2). *p < .05 paired t-test.E. Proportion of ASVs found either unique to either region or shared within each mouse showing the inner community maintains a large proportion of unique ASVs and the outer community mostly consists of ASVs also found in the inner (N = 5 mice from Cage 2, SD). *p < .05, **p < .01 paired t-test.F. Abundance of unique ASVs in the inner community of each mouse condensed by family to show the unique ASVs of the inner community are at low abundance (N = 5 mice from Cage 2, SEM).G. Abundance of shared ASVs between the inner and outer communities of each mouse condensed by family to show the shared ASVs comprise the majority of the community (N = 5 mice from Cage 2, SEM).H. Number of ASVs found either unique to either region or shared within each mouse condensed by class showing the increased richness in the inner community is comprised of unique ASVs from classes Clostridia and Bacteroidia (N = 5 mice from Cage 2, SEM).

Figure 4. Mice deficient in adaptive immunity still form biofilm-like community with some changes to composition

A. Experimental design for comparing dense community formation in WT and Rag1KO mice. Provided germ-free WT and Rag1KO mice with same donor flora and examined community structure after 4 weeks of conventionalization. B. Colonic cross-section stained with fluorescent probes identifying all bacteria (green) and host epithelium (blue) to show dense community structure forms after 4 weeks in WT-conventionalized mice. C. Colonic cross-section stained with fluorescent probes identifying all bacteria (green) and host epithelium (blue) to show dense community structure forms after 4 weeks in Rag1KO conventionalized mice. D. Principal coordinates of analysis (PCoA) plot using weighted UniFrac distances show there is a significant difference between mucosal communities of conventionalized WT and Rag1KO mice (N = 4 mice per group). *p < .05 PERMANOVA. E. Principal coordinates of analysis (PCoA) plot using unweighted UniFrac distances show there is no significant difference between mucosal communities of conventionalized WT and Rag1KO mice (N = 4 mice per group). p = .113 PERMANOVAF. Family level relative abundance of mucosal communities in conventionalized Rag1KO and WT mice (N = 4 mice per group, SD).G. Differential abundance of family Prevotellaceae calculated using DESeq2, showing significantly less Prevotellaceae in the mucosal communities of Rag1KO mice (N = 4 mice per group mice, Log2 Fold Change = 1.0, padj = 6.67 x 10⁻⁵).

Figure 5. Dense community structure is sensitive to Vancomycin

A. Experimental design for comparing dense community structure in Ciprofloxacin and Vancomycin treated mice. Provided SPF WT mice with 500 mg/L Ciprofloxacin or Vancomycin in the drinking water for three days. B. Shannon diversity of untreated, Vancomycin, and Ciprofloxacin treated fecal communities (N = 6 mice per group, SD).C. Class-level relative abundance for untreated, Vancomycin, and Ciprofloxacin treated fecal communities (N = 6 mice per group, SEM).D. FISH images of spatial structure after Ciprofloxacin treatment showing preservation of community consisting of Clostridia and excluding Proteobacteria. i: merged image with host epithelium (DAPI, blue), all bacteria (16s rRNA, green), and Proteobacteria (red). ii. Merged image with host epithelium and Proteobacteria. iii. Merged image with host epithelium, all bacteria, and Clostridia (red). iv. Merged image with host epithelium and ClostridiaE. FISH images of spatial structure after Vancomycin treatment showing decimation of dense band of bacteria as well as the absence of Clostridia and the prevalence of Proteobacteria. i. merged image with host epithelium (DAPI, blue), all bacteria (16S rRNA, green), and Proteobacteria (red). ii: merged image with host epithelium and Proteobacteria. iii. Merged image with host epithelium, all bacteria, and Clostridia (red). iv. Merged image with host epithelium and ClostridiaF. Mean fluorescence intensity (MFI) measurements for representative mice quantifying bacterial density for all bacteria (16s rRNA), Proteobacteria, and Clostridia after Ciprofloxacin (open circles) and Vancomycin (closed circles) treatment. Measurements started 10-μm before start of bacterial signal closest to epithelial lining (N = 3 mice, SEM).

Figure 6. Following antibiotic treatment inner dense community maintains higher species richness and is less prone to dysbiosis than luminal community

A. Experimental design for comparing effects of antibiotics on spatially distinct regions of the gut microbiome. Provided mice water with 100 mg/L of Vancomycin in drinking water. Sacrificed after 10 days of treatment. B. Principal coordinates of analysis (PcoA) plot using weighted UniFrac distances shows both mucosal and luminal communities are significantly altered by antibiotic treatment (N = 7 mice per group).C. Principal coordinates of analysis (PcoA) plot using unweighted UniFrac distances shows both mucosal and luminal communities are significantly altered by antibiotic treatment (N = 7 mice per group).D. Number of unique ASVs found in each sample at rarefied depth of 39,464 reads (N = 7 mice per group) showing a significant depletion in richness in both locations following antibiotic treatment (paired t-test) and the luminal community lost significantly more ASVs than the mucosal community following antibiotic treatment (unpaired t-test, SEM). **p < .01, *** p < .001.E. Phylum-level relative abundance for pre- and post-antibiotic mucosal and luminal communities (N = 7 mice per group, SEM).F. Differential abundance of class Clostridia, showing significantly more Clostridia in the mucosal community after antibiotic treatment (N = 7 mice per group, Log2 Fold Change = 1.33, padj = 0.034). ***p < .001 DESeq2.G. Differential abundance of class β-Proteobacteria, showing significantly more β-Proteobacteria in the luminal community after antibiotic treatment (N = 7 mice per group, Log2 Fold Change = 0.59, padj = 0.034). ***p < .001 DESeq2.H. qPCR quantification of Clostridia relative abundance of mucosal and luminal communities after antibiotics, showing increased Clostridia in the mucosal community following treatment (N = 4 mice per group, SEM). *p < .05 paired t-test.I. FISH images of mucosal communities before and after Vancomycin treatment showing the greater prevalence of Clostridia (red) before treatment and the reduction after antibiotics, and the greater maintenance of Clostridia in the dense community adjacent to the host epithelium (DAPI, blue) following treatment.

Figure 7. Intermittent fasting increases thickness and density of biofilm-like community and significantly alters community composition

A. Experimental design for intermittent fasting experiment. Intermittent fasting (IF) mice were allowed ad libitum feeding during an 8-hour feeding window during the dark cycle and fasted for the remaining 16 hours. This was repeated every day for 30 days. A control group was given continuous ad libitum (AL) feeding. B. The number of grams of food eaten per 24-hour period by each cage showing no significant difference in the amount consumed between AL and IF mice (SD).C. Percent change in body weight over the course of the experiment showing no significant difference in the increase in body weight between AL and IF mice (SD).D. Blood glucose levels taken before, 15 min, 30 min, 45 min, 60 min, and 130 min post glucose injection showing a significant increase in blood glucose levels of AL mice at 30 min post-injection, suggesting reduced insulin sensitivity compared to IF mice (N = 6 mice per group, SEM). *p < .05, **p < .01 2-way ANOVA. E. Colonic cross-section stained with fluorescent probes identifying all bacteria (green) and host epithelium (blue) of dense community structure with Left: Ad libitum feeding and Right: Intermittent fasting, showing increase in thickness and density with IF treatment. F. Overlay of MFI measurements from representative IF and AL mice showing difference in thickness and density (SEM).G. Thickness of dense community structure after AL and IF measured by MFI showing significantly thicker communities after IF (N = 5–6 mice per group, SD). *p < .05 unpaired t-test.H. Peak MFI measurements of dense community structure after AL and IF showing significantly denser communities after IF (N = 5–6 mice per group, SD). *p < .05 unpaired t-test.I. Observed ASVs found in AL and IF LCM-captured communities (150-μm) showing significantly higher richness in IF communities. Samples were rarefied to lowest sampling depth for normalization (N = 4–5 mice per group, SD). *p < .05 unpaired t-test.J. Shannon diversity in AL and IF LCM-captured communities (150-μm) showing significantly higher diversity in IF communities. Samples were rarefied to lowest sampling depth for normalization (N = 4–5 mice per group, SD). **p < .01 unpaired t-test.K. Principal coordinates of analysis (PcoA) plot using weighted UniFrac distances show LCM-captured communities (150-μm) from AL and IF are significantly different (N = 4–5 mice per group). **p < .01L. Principal coordinates of analysis (PcoA) plot using unweighted UniFrac distances show LCM-captured communities (150-μm) from AL and IF are significantly different (N = 4–5 mice per group). **p < .01M. Differential abundance of Clostridia between LCM-captured communities (150-μm) from AL and IF communities, showing significantly more Clostridia in IF communities (N = 4–5 mice per group, Log2 Fold Change = 1.04, padj = 0.0017, SD). ***p < .001 DESeq2.N. Differential abundance of Bacilli between LCM-captured communities (150-μm) from AL and IF communities, showing significantly less Bacilli in IF communities (N = 4–5 mice per group, Log2 Fold Change = 2.59, padj = 0.0004, SD). ****p < .0001 DESeq2.