Defective Atg16l1 in intestinal epithelial cells links to altered fecal microbiota and metabolic shifts during pregnancy in mice

Abstract

Throughout gestation, the female body undergoes a series of transformations, including profound alterations in intestinal microbial communities. Changes gradually increase toward the end of pregnancy and comprise reduced α-diversity of microbial communities and an increased propensity for energy harvest. Despite the importance of the intestinal microbiota for the pathophysiology of inflammatory bowel diseases, very little is known about the relationship between these microbiota shifts and pregnancy-associated complications of the disease. Here, we explored the longitudinal dynamics of gut microbiota composition and functional potential during pregnancy and after lactation in Atg16l1^∆IEC mice carrying an intestinal epithelial deletion of the Crohn's disease risk gene Atg16l1. Using 16S rRNA amplicon and shotgun metagenomic sequencing, we demonstrated divergent temporal shifts in microbial composition between Atg16l1 wildtype and Atg16l1^∆IEC pregnant mice in trimester 3, which was validated in an independent experiment. Observed differences included microbial genera implicated in IBD such as Lachnospiraceae, Roseburia, Ruminococcus, and Turicibacter. Changes partially recovered after lactation. Additionally, metagenomic and metabolomic analyses suggest an increased capacity for chitin degradation, resulting in higher levels of free N-acetyl-glucosamine products in feces, alongside reduced glucose and myo-inositol levels in serum around the time of delivery. On the host side, we found that the immunological response of Atg16l1^∆IEC mice is characterized by higher colonic mRNA levels of TNFα and CXCL1 in trimester 3 and a lower weight of offspring at birth. Understanding pregnancy-dependent microbiome changes in the context of IBD may constitute the first step in the identification of fecal microbial biomarkers and microbiota-directed therapies that could help improve precision care for managing pregnancies in IBD patients.

Keywords: Atg16l1; IBD; Pregnancy; microbiota.

Figure 1.

Gut microbiota characterization of Atg16l1^fl/fl and Atg16l1^∆IEC mice during pregnancy and after lactation periods. a. study design. Stool samples of Atg16l1^fl/fl (n = 14 mice) and Atg16l1^∆IEC (n = 15 mice) pregnant mice were collected at baseline (before mating), week 3 (late pregnancy), and week 6 (after lactation period). Samples were submitted to 16S rRNA and shotgun sequencing. b. alpha diversity analysis of gut microbiota during pregnancy and lactation. Shannon diversity was computed at the ASV level. Comparisons between Atg16l1^fl/fl (n = 14 mice, three timepoints) and Atg16l1^∆IEC (n = 15 mice, three timepoints) were performed using the Wilcoxon rank-sum test. c. principal coordinate analysis on Aitchison distance matrix of pregnant mice. Differences between Atg16l1^fl/fl and Atg16l1^∆IEC were tested with PERMANOVA with 10,000 permutations. FDR represents Benjamini-Hochberg corrected p values, and adj.R² represents partial omega squares as effect size in the analysis of variance. d. relative abundances of the top 20 most abundant genera. Unclassified genera and those with low relative abundance are grouped as “others”. Colors represent individual phylum and color gradients represent individual genus within a phylum. e. triangular dot-plot showing significantly changing genera compared to baseline. The triangles show the hues and direction of the effect size (log₂FC). Color intensity and size represent the magnitude of the effect size and FDR significance of each specific genus, respectively. f. top three unstable genera with significantly increased abundances in any timepoint compared to baseline. g. top three unstable genera with significantly decreased abundances in any timepoint compared to baseline. The longitudinal plots are colored to remark differences in genotype.

Figure 2.

Taxonomic profiling differences of the gut microbiota of Atg16l1^fl/fl and Atg16l1^∆IEC during pregnancy and lactation periods. a. variation partition analysis of different taxa-level abundances from 16S rDNA data. Each color represents the contribution of individual covariates as a source of variation. Each stacked-area bar plot compartment (timepoints, BL, w3, and w6) represents the output of different variance partition models. b. top 20 genera mainly contributing to the variation of the maternal genotype at week 3. Cross-sectional comparisons (Atg16l1^fl/fl vs Atg16l1^∆IEC) were performed using the Wilcoxon rank-sum test, and p values were corrected by the Benjamini–Hochberg method. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, empty: not significant. The color of the arrow symbol represents the sense of the increased effect of each significant genus (Atg16l1^fl/fl, blue or Atg16l1^∆IEC, pink). c. heatmap of significant abundance changes in identified species and SGBs found during multivariate comparison of Atg16l1^fl/fl vs Atg16l1^∆IEC. for each cell, colors indicate the row-wise z-score of relative abundances, asterisks denote the FDR < 0.05 significance at each cross-sectional comparison and prevalence represents the percentage of non-zero features. Row-wise clusters represent features that belong to the same phylum. d. Box plots showing the relative abundance of selected species and SGBs, the dots represents fecal samples of Atg16l1^fl/fl (n = 12) and Atg16l1^∆IEC (n = 12) pregnant mice per time point.

Figure 3.

Functional potential predictions of the gut microbiota of Atg16l1^fl/fl and Atg16l1^∆IEC during pregnancy and lactation periods. a. variation partitioning analysis of different functional categories abundances from metagenomics data. Each color represents the contribution of individual covariates as a source of variation. Each stacked-area bar plot compartment (timepoints, BL, w3, and w6) represents the output of different variance partition models. b. top 20 MetaCyc pathways mainly contributing to the variation of the maternal genotype at week 3. c. boxplot of the abundances of polyamine biosynthesis pathway. d. Boxplot of the abundances of chitin degradation pathway. Boxplots are colored to remark differences in genotype. e. top 20 KEGG orthologs that most contribute to the variation of maternal genotype at week 3. The arrow symbol and colors represent the sense of the increased effect of each significant genus (Atg16l1^fl/fl, blue or Atg16l1^∆IEC, pink). f. box plot of the K02377 gene family, GDP-L fucose synthase. Boxplots are colored to remark differences in genotype. Cross-sectional comparisons (Atg16l1^fl/fl , n = 12 mice vs Atg16l1^∆IEC, n = 12 mice) were performed using the Wilcoxon rank-sum test and p values were corrected by the Benjamini–Hochberg method. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, empty: not significant.

Figure 4.

Gut microbiota and cytokine expression characterization of Atg16l1^fl/fl and Atg16l1^∆IEC during pregnancy period (independent validation experiment). (a) Study design. Stool samples of Atg16l1^fl/fl (n = 7) and Atg16l1^∆IEC (n = 7) pregnant mice were collected weekly from BL (before mating) to week 3 (late pregnancy) and submitted to 16S rRNA amplicon sequencing. (b) Alpha diversity analysis of gut microbiota during pregnancy and after lactation. Shannon diversity was computed at the ASV level. (c) Principal coordinate analysis on Aitchison distance matrix of pregnant mice. Differences between Atg16l1^fl/fl and Atg16l1^∆IEC were tested with PERMANOVA with 10,000 permutations. FDR represents Benjamini-Hochberg corrected p values, and adj.R² represents partial omega squares as effect size in the analysis of variance. (d) Relative abundances of the top 20 most abundant genera. Unclassified genera and those with low relative abundance are grouped as “others”. Colors represent individual phylum and color gradients represent individual genus within a phylum. (e, f) Expression levels of the cytokines in nulliparous and pregnant Atg16l1^fl/fl and Atg16l1^∆IEC mice (e) TNFα and (f) CXCL1. (g) Puppy weight. Groups are shown as nulliparous Atg16l1^fl/fl, nulliparous Atg16l1^∆IEC, pregnant Atg16l1^fl/fl and pregnant Atg16l1^∆IEC. Cross-sectional comparisons (Atg16l1^fl/fl vs Atg16l1^∆IEC) were performed using the Wilcoxon rank-sum test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns: not significant.

Figure 5.

Fecal and serum metabolomics of Atg16l1^fl/fl (n = 5) and Atg16l1^∆IEC (n = 4) mice at trimester 3 of pregnancy. a. microbiome-specific and significant different metabolites in serum and feces between Atg16l1^fl/fl and Atg16l1^∆IEC mice. b. N-acetyl glucosamine, c. L-Fucose, d. myo-inositol, e. correlation analysis of metabolites and cytokine expression. f, g. cysteine association with tnf-α, h i. glucose association with CXCL1.