Transcriptomic profiling of PBDE-exposed HepaRG cells unveils critical lncRNA- PCG pairs involved in intermediary metabolism

Abstract

Polybrominated diphenyl ethers (PBDEs) were formally used as flame-retardants and are chemically stable, lipophlic persistent organic pollutants which are known to bioaccumulate in humans. Although its toxicities are well characterized, little is known about the changes in transcriptional regulation caused by PBDE exposure. Long non-coding RNAs (lncRNAs) are increasingly recognized as key regulators of transcriptional and translational processes. It is hypothesized that lncRNAs can regulate nearby protein-coding genes (PCGs) and changes in the transcription of lncRNAs may act in cis to perturb gene expression of its neighboring PCGs. The goals of this study were to 1) characterize PCGs and lncRNAs that are differentially regulated from exposure to PBDEs; 2) identify PCG-lncRNA pairs through genome annotation and predictive binding tools; and 3) determine enriched canonical pathways caused by differentially expressed lncRNA-PCGs pairs. HepaRG cells, which are human-derived hepatic cells that accurately represent gene expression profiles of human liver tissue, were exposed to BDE-47 and BDE-99 at a dose of 25 μM for 24 hours. Differentially expressed lncRNA-PCG pairs were identified through DESeq2 and HOMER; significant canonical pathways were determined through Ingenuity Pathway Analysis (IPA). LncTar was used to predict the binding of 19 lncRNA-PCG pairs with known roles in drug-processing pathways. Genome annotation revealed that the majority of the differentially expressed lncRNAs map to PCG introns. PBDEs regulated overlapping pathways with PXR and CAR such as protein ubiqutination pathway and peroxisome proliferator-activated receptor alpha-retinoid X receptor alpha (PPARα-RXRα) activation but also regulate distinctive pathways involved in intermediary metabolism. PBDEs uniquely down-regulated GDP-L-fucose biosynthesis, suggesting its role in modifying important pathways involved in intermediary metabolism such as carbohydrate and lipid metabolism. In conclusion, we provide strong evidence that PBDEs regulate both PCGs and lncRNAs in a PXR/CAR ligand-dependent and independent manner.

Fig 1. Workflow depicting preparation and analysis of expression data.

Fig 2. Regulation of protein coding genes (PCGs) and long noncoding RNA (lncRNAs) in HepaRG cells when exposed to rifampin (RIF), constitutive androstane receptor agonist (CITCO), BDE47 and BDE99.

(A) Bar plot depicting the proportion of PCGs that were differentially expressed in at least one treatment group, stably expressed in all treatment groups, and unexpressed in all treatment groups. (B) A 4 way Venn Diagram displaying shared and unique differentially expressed PCGs in all treatment groups. (C) Bar plot depicting the proportion of lncRNAs that were differentially expressed in at least one treatment group, stably expressed in all treatment groups, and unexpressed in all treatment groups. (D) (B) A 4 way Venn Diagram displaying shared and unique differentially expressed lncRNAs in all treatment groups.

Fig 3. Genome annotation of lncRNAs through annotatePeaks in HOMER.

This pie chart shows the proportion of lncRNAs that were mapped proximal to either a 3’UTR, 5’UTR, exon, intron, non-coding region, promoter TSS, or TSS.

Fig 4. Differentially expressed lncRNA-PCGs pairs associated with the Protein Ubiquitination pathway.

(A) 4 way Venn Diagram displaying the common and unique genes differentially expressed in all four pathways.

Fig 5. Differentially expressed lncRNA-PCGs pairs associated with the PPARa-RXRa pathway.

(A-D) Heat maps depicting gene expression patterns of differentially expressed genes associated with this pathway. Blue indicates down-regulation while red indicates up-regulation. (E) 4 way Venn Diagram displaying the common and unique genes differentially expressed in all four pathways.

Fig 6. Differentially expressed lncRNA-PCGs pairs associated with the GDP-L-fucose Biosynthesis I (from GDP-D-mannose) pathway.

(A-B) Barplots depicting gene expression patterns of differentially expressed genes associated with this pathway. Errors bars represent mean ± SE.

Fig 7. Unique canonical pathways induced by BDE99.

(A-D) Heat maps depicting gene expression patterns of differentially expressed genes associated with the Bile Acid Synthesis pathway (A), JAK-Stat Signaling Pathway (B), Sirtuin Signaling Pathway (C), and Autophagy (D). Blue indicates down-regulation while red indicates up-regulation.

Fig 8. Effect of BDE47 on Rho Family of GTPases.

(A-B) Heat maps depicting gene expression patterns of differentially expressed genes associated with the RhoBDI Signaling pathway (A) and Signaling by Rho Family GTPases (B). Blue indicates down-regulation while red indicates up-regulation.

Fig 9

Effect of PBDEs and CITCO on interferon signaling: (A-C) Heat maps depicting gene expression of differentially expressed lncRNA-PCG pairs involved in interferon signaling. (D) Three-way Venn Diagram displaying the common and unique genes differentially expressed by CITCO, BDE-47, and BDE-99 for interferon signaling. Blue indicates down-regulation while red indicates up-regulation.

Fig 10

Effect of PBDEs and CITCO on glutathione biosynthesls: (A) Expression of glutamate—cysteine ligase catalytic subunit (GCLC) by exposure group (B) Expression of glutamate-cysteine ligase modifier subunit (GCLM) by exposure group.

Fig 11

Effect of PBDEs and CITCO on mTOR signaling: (A-C) Heat maps depicting gene expression of differentially expressed lncRNA-PCG pairs involved in mTOR signaling. (D) Three-way Venn diagram showing overlapping and unique genes between the three groups. Blue indicates down-regulation while red indicates up-regulation.