The fecal mycobiome in non-alcoholic fatty liver disease

Abstract

Background & aims: Studies investigating the gut-liver axis have largely focused on bacteria, whereas little is known about commensal fungi. We characterized fecal fungi in patients with non-alcoholic fatty liver disease (NAFLD) and investigated their role in a fecal microbiome-humanized mouse model of Western diet-induced steatohepatitis.

Methods: We performed fungal internal transcribed spacer 2 sequencing using fecal samples from 78 patients with NAFLD, 16 controls and 73 patients with alcohol use disorder. Anti-Candida albicans (C. albicans) IgG was measured in blood samples from 17 controls and 79 patients with NAFLD. Songbird, a novel multinominal regression tool, was used to investigate mycobiome changes. Germ-free mice were colonized with feces from patients with non-alcoholic steatohepatitis (NASH), fed a Western diet for 20 weeks and treated with the antifungal amphotericin B.

Results: The presence of non-obese NASH or F2-F4 fibrosis was associated with a distinct fecal mycobiome signature. Changes were characterized by an increased log-ratio for Mucor sp./Saccharomyces cerevisiae (S. cerevisiae) in patients with NASH and F2-F4 fibrosis. The C. albicans/S. cerevisiae log-ratio was significantly higher in non-obese patients with NASH when compared with non-obese patients with NAFL or controls. We observed a different fecal mycobiome composition in patients with NAFLD and advanced fibrosis compared to those with alcohol use disorder and advanced fibrosis. Plasma anti-C. albicans IgG was increased in patients with NAFLD and advanced fibrosis. Gnotobiotic mice, colonized with human NASH feces and treated with amphotericin B were protected from Western diet-induced steatohepatitis.

Conclusions: Non-obese patients with NAFLD and more advanced disease have a different fecal mycobiome composition to those with mild disease. Antifungal treatment ameliorates diet-induced steatohepatitis in mice. Intestinal fungi could be an attractive target to attenuate NASH.

Lay summary: Non-alcoholic fatty liver disease is one of the most common chronic liver diseases and is associated with changes in the fecal bacterial microbiome. We show that patients with non-alcoholic fatty liver disease and more severe disease stages have a specific composition of fecal fungi and an increased systemic immune response to Candida albicans. In a fecal microbiome-humanized mouse model of Western diet-induced steatohepatitis, we show that treatment with antifungals reduces liver damage.

Keywords: Fungi; NAFLD; NASH; gut pathogens; metagenomics; microbiome; microbiota.

Figure 1.. Liver disease severity is associated with a specific fecal mycobiome composition in non-obese patients with NAFLD.

(A-B) Principal coordinate analyses (PCoA) based on the Jaccard dissimilarity matrices for NASH and F2-F4 fibrosis. All of the detected 120 fungal taxa at were included. P values were determined by permutational multivariate analysis of variance (PERMANOVA) and adjusted for age, gender, BMI, type 2 diabetes, proton pump inhibitor use and alcohol consumption. Bold font indicates statistical significance (P<0.05). (C-D) No significant differences in the alpha diversity and richness among groups. In (A-D) 78 patients with NAFLD were included in the analysis of whom 24 had NAFL, 54 had NASH; 40 had minimal fibrosis (F0-F1) and 38 had at least moderate fibrosis (F2-F4). In addition, 16 subjects without any known disease were included. For the box and whisker plots (C-D), the box extends from the 25th to 75th percentile, with the center line indicating the median; the bottom whiskers indicate the minimum values and the top whiskers indicate the 75th percentile plus 1.5-fold the inter-quartile distance (the distance between the 25th and 75th percentiles). Values greater than this are plotted as individual dots. P values determined by Kruskal-Wallis test with Dunn’s post-hoc test for skewed distributions followed by false discovery rate (FDR) or Whitney-Wilcoxon rank-sum test. NAFL, non-alcoholic fatty liver; NAFLD, non-alcoholic fatty liver disease; NASH, non-alcoholic steatohepatitis.

Figure 2.. Compositional changes in the fecal mycobiome of patients with NASH.

(A) Differential abundance of fungal taxa among groups were calculated using Songbird under consideration of clinical metadata (age, gender, BMI, type 2 diabetes, proton pump inhibitor use and alcohol consumption). Fungal taxa with at least 90% confidence level alignment in the UNITE database were colored. (B) Log-ratios of specific fungal taxa that were significantly different among groups, either when including all patients, or (B) when investigating subgroups based on obesity. We determined the denominator for the calculation of the log-ratios by choosing a fungal taxon that is present in all samples. This was the case for Saccharomyces cerevisiae (S. cerevisiae). In order to handle problems occurring when calculating log-ratios in zero-inflated microbiome data, we added a constant small value (0.1) to each count before calculating log-ratios. In (A-C), 78 patients with NAFLD were included in the analysis of whom 24 had NAFL, 54 had NASH. Of the patients with NAFL, 17 (70.8%) were non-obese and 7 (29.2%) were obese. Of the patients with NASH, 21 (38.8%) were non-obese and 33 (61.2%) were obese. In (B-C), 16 subjects without any known disease were additionally included. P values determined by Kruskal-Wallis test with Dunn’s post-hoc test for skewed distributions followed by false discovery rate (FDR) or Whitney-Wilcoxon rank-sum test. *P<0.05, >0.01; **P <0.01, P>0.001; ***P <0.001. For the box and whisker plots (b-c), the box extends from the 25th to 75th percentile, with the center line indicating the median; the bottom whiskers indicate the minimum values and the top whiskers indicate the 75th percentile plus 1.5-fold the inter-quartile distance (the distance between the 25th and 75th percentiles). Values greater than this are plotted as individual dots. NAFL, non-alcoholic fatty liver; NASH, non-alcoholic steatohepatitis.

Figure 3.. Compositional changes in the fecal mycobiome of patients with F2-F4 fibrosis.

(A) Differential abundance of fungal taxa among groups were calculated using Songbird under consideration of clinical metadata (age, gender, BMI, type 2 diabetes, proton pump inhibitor use and alcohol consumption). Fungal taxa with at least 90% confidence level alignment in the UNITE database were colored. (B) Log-ratios of specific fungal taxa that were significant among groups, either when including all patients, or (C) when investigating subgroups based on obesity. In (A-C), 78 patients with NAFLD were included in the analysis of whom 40 had minimal fibrosis (F0-F1) and 38 had at least moderate fibrosis (F2-F4). Of the patients with F0-F1 fibrosis, 23 (57.5%) were non-obese and 17 (42.5%) were obese. Of the patients with F2-F4 fibrosis, 15 (39.5%) were non-obese and 23 (60.5%) were obese. In (B-C), 16 subjects without any known disease were additionally included. P values determined by Kruskal-Wallis test with Dunn’s post-hoc test for skewed distributions followed by false discovery rate (FDR) or Whitney-Wilcoxon rank-sum test. *P<0.05, >0.01; **P <0.01, P>0.001; ***P <0.001. For the box and whisker plots (B-C), the box extends from the 25th to 75th percentile, with the center line indicating the median; the bottom whiskers indicate the minimum values and the top whiskers indicate the 75th percentile plus 1.5-fold the inter-quartile distance (the distance between the 25th and 75th percentiles). Values greater than this are plotted as individual dots. NAFLD, non-alcoholic fatty liver disease.

Figure 4.. Associations between compositional fecal mycobiota changes and liver histology features.

Univariate and multivariate ordinal regression models using liver histology features as outcome parameter, were used to associate specific log-ratios with liver biopsy features. Predicted probabilities for (A) individual stages of fibrosis, (B) grades of inflammation and (C) the NAFLD activity score. Univariate and multivariate analyses, adjusted for age, gender, BMI, type 2 diabetes, PPI use and alcohol consumption as well as multivariate analyses introducing the BMI as moderating factor, were performed. Only significant (P<0.05) associations are shown. In (A-C), 69 biopsy-proven patients with NAFLD were included. Bold font indicates statistical significance (P<0.05). BMI, body mass index; NAFLD, non-alcoholic fatty liver disease.

Figure 5.. Plasma antibodies against Candida albicans are increased in patients with NAFLD and advanced fibrosis.

Antibodies specific to Candida albicans (Anti-C. albicans IgG) were measured in plasma samples from 96 subjects (17 controls and 79 patients with NAFLD) of whom 74 were overlapping with the subjects for the fecal mycobiome analysis. (A) Spearman correlation between anti-C. albicans IgG and C. albicans/ Saccharomyces cerevisiae log ratio. (B) Anti-C. albicans IgG in n=17 controls, n=58 subjects with NAFLD and F0-F2 fibrosis and n=21 subjects with NAFLD and F3-F4 fibrosis. For the box and whisker plot (B), the box extends from the 25th to 75th percentile, with the center line indicating the median; the bottom whiskers indicate the minimum values and the top whiskers indicate the 75th percentile plus 1.5-fold the inter-quartile distance (the distance between the 25th and 75th percentiles). Values greater than this are plotted as individual dots. P values determined by Kruskal-Wallis test with Dunn’s post-hoc test for skewed distributions followed by false discovery rate (FDR). **P <0.01, P>0.001. C. albicans, Candida albicans; Ig, Immunoglobulin; NAFLD, non-alcoholic fatty liver disease

Figure 6.. Amphotericin B decreases Western diet-induced steatohepatitis and fibrosis in fecal microbiome-humanized mice.

Germ-free C57BL/6 mice were colonized with feces from two patients with NASH and subjected to 20 weeks of Western diet-feeding with or without amphotericin B (n=13-16 per group), or chow diet with or without amphotericin B (n=3-5 per group). (A) Plasma ALT levels. (B) Hepatic triglyceride content. (C) Representative liver sections after hematoxylin and eosin staining (bar size = 100 μm). (D) Hepatic cholesterol content. (E-H) Hepatic mRNA expression of (E) Tnf-α, (F) F4/80, (G) Col1a1, and (H) Tgfb1. (I) Hepatic hydroxyproline content. (J) Representative liver sections after Sirius Red staining (bar size = 100 μm). (K-L) Fecal content of (K) Candida albicans and (L) Mucor/Rhizopus spp. in Western diet-fed groups. Results are expressed as mean ± s.e.m. P values are determined by One-way ANOVA with Holm’s post-hoc test (A-I) or unpaired student t-test (K-L). *P<0.05. ALT, alanine aminotransferase; Col1a1, collagen type I alpha 1; Mucor/Rhizopus spp, Mucor/Rhizomucor/Rhizopus species; NASH, non-alcoholic steatohepatitis; Tgfb1, transforming growth factor-beta 1; Tnf-α, tumor necrosis factor-alpha.