Integrated RNA Sequencing Reveals Epigenetic Impacts of Diesel Particulate Matter Exposure in Human Cerebral Organoids

Abstract

Autism spectrum disorder (ASD) manifests early in childhood. While genetic variants increase risk for ASD, a growing body of literature has established that in utero chemical exposures also contribute to ASD risk. These chemicals include air-based pollutants like diesel particulate matter (DPM). A combination of single-cell and direct transcriptomics of DPM-exposed human-induced pluripotent stem cell-derived cerebral organoids revealed toxicogenomic effects of DPM exposure during fetal brain development. Direct transcriptomics, sequencing RNA bases via Nanopore, revealed that cerebral organoids contain extensive RNA modifications, with DPM-altering cytosine methylation in oxidative mitochondrial transcripts expressed in outer radial glia cells. Single-cell transcriptomics further confirmed an oxidative phosphorylation change in cell groups such as outer radial glia upon DPM exposure. This approach highlights how DPM exposure perturbs normal mitochondrial function and cellular respiration during early brain development, which may contribute to developmental disorders like ASD by altering neurodevelopment.

Keywords: Autism; Cerebral organoids; Diesel particulate matter; Epigenetics; Transcriptomics.

Fig. 1

Overview of environmental exposure experiment with cerebral organoids. Cerebral organoids are matured to day 30 and then exposed to DPM for 7 days. These can then be analyzed using different sequencing technologies to probe the effects of DPM.

Fig. 2

Nanopore PCR-amplified vs. Direct RNAseq. A) Principle component analysis of cerebral organoid samples for Nanopore PCR (cyan) or Direct (red) RNAseq. B) Volcano plot for genes significantly differentially expressed in PCR vs Direct RNAseq with RNA binding associated genes highlighted. C-E) Details for most significant modification of control Direct RNAseq using comparison to PCR amplified data (C), 5mC model prediction (D), and de novo predictions (E). Shown on the left of each is the voltage overlap of control reads (red) relative to PCR or expected positions (gray) for the most significant site detected. On the right is the meme motif from the top 100 significant with statistics shown, and below that a String network with enriched ontology terms. The panel C does not have any significant enriched protein networks so shown is the second predicted motif.

Fig. 3

Control vs. DPM treatment using Nanopore Direct RNAseq. A) Voltage overlay for the top predicted modification found in the TMEM14B gene. DPM treatment reads are in red and control in black. B) Density plot for the voltage signal in panel A with DPM in red and control in black. C) Meme motif prediction from the top 100 predicted modified sites between control and DPM treatment. D) String network of genes found in the top 100 modified sites between control and DPM. Colors corresponded to the labeled enriched terms.

Fig. 4

Single Cell (sc)RNAseq of control and DPM treated cerebral organoids. A). UMAP of cell expression (each dot) pooled from both control and DPM organoids identifying 19 cell clusters, labeled A-S with manually curated prediction of identifies. B) Comparison of DPM exposed vs control (Ctrl) cerebral organoids plotted as UMAP. Note the Outer radial glia cell clusters (group F) has a reduced number of cells in DPM exposed conditions (red circle). C) Significant genes of the DPM treatment shown as the highest -log10 adjusted Pvalue (y-axis) and the number of cell clusters that have an adjusted Pvalue <0.0001. Top ten genes are labeled in red. ASD SFARI genes are colored cyan. D) The number of significant genes in each cell cluster (x-axis) and the number of ASD genes significantly changed by DPM treatment. E) Top enriched KEGG pathways for genes found in ¾ of the group J, A, L, I.