M14: COMBINED ANALYSIS OF THE ORAL MICROBIOME AND MICROTRANSCRIPTOME OF AUTISM SPECTRUM DISORDER

Abstract

Background Autism Spectrum Disorder (ASD) affects 1 in 45 children aged 3–17 and is characterized by a wide array of deficits in social interaction, communication, and behavior. Despite significant evidence for genetic contributions to ASD risk, single gene variants as a group do not explain the vast majority of cases. Consequently, considerable interest has turned to the study of epigenetic mechanisms as potential contributing factors. miRNAs are now well-recognized epigenetic regulators of gene expression that influence biological processes in all cell types and are released from the cells in which they are synthesized and circulate throughout the body in extracellular fluids. We previously identified significant alterations in miRNA levels in the saliva of ASD subjects (average age 9) compared with matched typically developing control children. We also observed (but did not report) that the salivary microbiome of ASD children was significantly different than that of the control children. In the present study, we sought to replicate and extend these observations on the oral microtranscriptome and microbiome of ASD children in a larger and younger cohort. Methods Parental consent was obtained for all subjects. More than 350 samples were collected from children aged 2–6, with either a confirmed diagnosis of ASD, Developmental Delay, or healthy controls. Comprehensive medical and demographic information was obtained using detailed questionnaires, the Vineland Adaptive Behavior Scales (2nd Edition), and the Autism Diagnostic Observation Schedule (2nd Edition). Identification and quantification of saliva miRNA and microbiome abundance were performed using NGS on a NextSeq. 500 instrument at the SUNY Molecular Analysis Core (SUNYMAC) at Upstate Medical University. Alignment of NGS reads was performed to the miRbase21 database using the Shrimp2 algorithm in Partek Flow software to identify mature and precursor miRNAs and alignment of metatranscriptome reads was performed to the Human Microbiome Database, using K-SLAM and Kraken software. Data were subjected to quantile normalization, filtered for quality control, and examined for their diagnostic utility using machine learning algorithms. The most promising variables were also examined to identify salient patterns and subjected to functional analysis. Results Saliva miRNA and microbiome taxon variables demonstrated the ability to distinguish children with ASD from typically developing controls, with accuracy approximating 75–80% based on Monte-Carlo Cross Validation. When combined, the best miRNA and taxon classifiers from the learning algorithm performed at an overall accuracy level exceeding 86%, including more than 90% accuracy for ASD children. Notably, the top miRNA classifiers disproportionately targeted mRNAs that were enriched in several key biological pathways of interest for ASD, including Amphetamine Addiction, Axon Guidance, and Oxytocin Signaling. Notably, many miRNAs and taxon classifiers exhibited robust correlations, suggesting possible host-microbiome signaling. Discussion This is the first report of correlated miRNA and microbiome elements in saliva. Our results indicate a high degree of potential diagnostic utility in using these saliva-based data for ASD and Developmental Delay. Ongoing work is establishing the quantitative relationship between these variables and functional measures of ASD symptoms and severity, as well as co-morbid medical conditions including sleep disorders and GI disturbances.