Correlation of Gut Microbiome Between ASD Children and Mothers and Potential Biomarkers for Risk Assessment

Abstract

Variation of maternal gut microbiota may increase the risk of autism spectrum disorders (ASDs) in offspring. Animal studies have indicated that maternal gut microbiota is related to neurodevelopmental abnormalities in mouse offspring, while it is unclear whether there is a correlation between gut microbiota of ASD children and their mothers. We examined the relationships between gut microbiome profiles of ASD children and those of their mothers, and evaluated the clinical discriminatory power of discovered bacterial biomarkers. Gut microbiome was profiled and evaluated by 16S ribosomal RNA gene sequencing in stool samples of 59 mother-child pairs of ASD children and 30 matched mother-child pairs of healthy children. Significant differences were observed in the gut microbiome composition between ASD and healthy children in our Chinese cohort. Several unique bacterial biomarkers, such as Alcaligenaceae and Acinetobacter, were identified. Mothers of ASD children had more Proteobacteria, Alphaproteobacteria, Moraxellaceae, and Acinetobacter than mothers of healthy children. There was a clear correlation between gut microbiome profiles of children and their mothers; however, children with ASD still had unique bacterial biomarkers, such as Alcaligenaceae, Enterobacteriaceae, and Clostridium. Candidate biomarkers discovered in this study had remarkable discriminatory power. The identified patterns of mother-child gut microbiome profiles may be important for assessing risks during the early stage and planning of personalized treatment and prevention of ASD via microbiota modulation.

Keywords: Autism spectrum disorders; Biomarker; Gut microbiome; Microbiota-gut-immune-brain axis; Mother–child pair.

Figure 1

Comparison of the alpha diversity and relative abundances at the phylum level based on the OTU profile Comparison of the alpha diversity was evaluated using PD_whole_tree based on the OTU profile between the autism groups and the control groups and shown in the top panels for ASD-C vs. H-C (A), ASD-M vs. H-M (B), and ASD-M+C vs. H-M+C (C). P values were calculated using the Wilcoxon rank-sum test. The relative abundances of different taxa at phylum level were shown in the bottom panels for ASD-C vs. H-C (D), ASD-M vs. H-M (E), and ASD-M+C vs. H-M+C (F). OTU, operational taxonomic unit. ASD-C, ASD child; ASD-M, mother of ASD child; H-C, healthy child; H-M, mother of healthy child; ASD-M+C, mother–child pair of ASD child; H-M+C, mother–child pair of healthy child.

Figure 2

Microbiome community and Venn diagram analysis PCoA of bacterial beta diversity based on the unweighted UniFrac distances. ASD-C vs. H-C (A); ASD-M vs. H-M (B); ASD-M+C vs. H-M+C (C). D. Venn diagram displaying the degree of overlap of bacterial OTUs among ASD-C, ASD-M, H-C and H-M. PCoA, principal coordinate analysis.

Figure 3

The relative abundance of the OTUs and ROC curves A. The relative abundance of the top 96 most different OTUs across groups (LDA score >2 and adjusted P < 0.1) according to the Wilcoxon rank sum test. The abundance profiles are transformed into Z scores by subtracting the average abundance and dividing the standard deviation of all samples. Z score is negative (shown in blue) when the raw abundance is lower than the mean. OTUs with adjusted P < 0.01 and P < 0.05 are marked in red and green, respectively. B. ROC curve with adjusted P < 0.01, (Wilcoxon rank sum test) and LDA score >3 (LEfSe analysis). 5 biomarkers were selected to predict the risk of disease in children with autism. These include Betaproteobacteria, Burkholderiales, Pseudomonadales, Moraxellaceae, and Acinetobacter. C. ROC curve with adjusted P < 0.01 (Wilcoxon rank sum test) and LDA score >3 (LEfSe analysis). 6 biomarkers were selected to predict the risk of disease in children’s mothers. These biomarkers are Flavobacteriia, Gammaproteobacteria, Flavobacteriales, Weeksellaceae, Enterobacteriaceae, and Enterobacteriales. The SVM classifier from R package e1071 was adopted to perform the classification analysis for the selected biomarkers. Five-fold cross-validation was used to evaluate the performance of the predictive model. The ROC curves as well as the AUC value was calculated using the ROCR R package. P values were adjusted by FDR. ROC, receiver operating characteristic; LDA, linear discriminant analysis; LEfSe, linear discriminant effect size; AUC, area under the curve; FDR, false discovery rate.

Figure 4

LEfSe analysis between ASD-C and ASD-M groups A. Histogram of the LDA scores computed for differentially abundant taxa between ASD-C and ASD-M. The LDA score indicates the effect size and ranking of each differentially abundant taxon. B. The enriched taxa in ASD-C and ASD-M gut microbiome represented in the cladogram. The central point represents the root of the tree (Bacteria), and each ring represents the next lower taxonomic level (phylum to genus: p, phylum; c, class; o, order; f, family; g, genus). The diameter of each circle represents the relative abundance of the taxon.

Figure 5

Predicted metagenome function based on KEGG pathways analysis Extended error bar plot showed the significantly different KEGG pathways between ASD-C and H-C (A), between ASD-M and H-M (B), between ASD-M+C and H-M+C (C).

Supplementary Figure S1

The diversity and richness of the gut microbiota in the autism groups and the control groups The Shannon index of the gut microbiome in ASD-C and H-C (A), ASD-M and H-M (B), as well as ASD-M+C and H-M+C (C); and the ACE index of the gut microbiome in ASD-C and H-C (D), ASD-M and H-M (E), as well as ASD-M+C and H-M+C (F). ACE, abundance-based coverage estimator.

Supplementary Figure S2

Comparison of relative taxonomic abundance between the autism groups and the control groups at genus level Comparison of relative taxonomic abundance between different groups at the genus level (average of each group). A. ASD-C vs. H-C. B. ASD-M vs. H-M. C. ASD-M+C vs. H-M+C.

Supplementary Figure S3

Comparison of relative taxonomic abundance among ASD-C, H-C, ASD-M, and H-M at the level of phylum (A) and genus (B)

Supplementary Figure S4

LEfSe analysis between the autism groups and the control groups Histogram of the LDA scores computed for OTUs differentially abundant between different groups. A. ASD-C vs. H-C. B. ASD-M vs. H-M. C. ASD-M+C vs. H-M+C. Health-enriched OTUs are indicated with positive LDA scores, and OTUs enriched in autism have negative scores. The LDA score indicates the effect size and ranking of each differentially abundant taxon. Cladograms generated by LEfSe indicate differences in the bacterial taxa between different groups. D. ASD-C vs. H-C. E. ASD-M vs. H-M. F. ASD-M+C vs. H-M+C. Nodes in red indicate taxa that were enriched in autism group compared to those in health controls, while nodes in green indicate taxa that were enriched in health controls compared to those in the autism group.

Supplementary Figure S5

Amino acid level in blood of ASD-C and H-C subjects