https://orcid.org/0000-0001-9163-927X
Abstract
Recent evidence suggests that complex diseases can result from early life exposure to environmental toxicants. Polybrominated diphenyl ethers (PBDEs), and polychlorinated biphenyls (PCBs) are persistent organic pollutants (POPs) and remain a continuing risk to human health despite being banned from production. Developmental BPA exposure mediated-adult onset of liver cancer via epigenetic reprogramming mechanisms has been identified. Here, we investigated whether the gut microbiome and liver can be persistently reprogrammed following neonatal exposure to POPs, and the associations between microbial biomarkers and disease-prone changes in the hepatic transcriptome in adulthood, compared with BPA. C57BL/6 male and female mouse pups were orally administered vehicle, BPA, BDE-99 (a breast milk-enriched PBDE congener), or the Fox River PCB mixture (PCBs), once daily for three consecutive days (postnatal days [PND] 2–4). Tissues were collected at PND5 and PND60. Among the three chemicals investigated, early life exposure to BDE-99 produced the most prominent developmental reprogramming of the gut-liver axis, including hepatic inflammatory and cancer-prone signatures. In adulthood, neonatal BDE-99 exposure resulted in a persistent increase in Akkermansia muciniphila throughout the intestine, accompanied by increased hepatic levels of acetate and succinate, the known products of A. muciniphila. In males, this was positively associated with permissive epigenetic marks H3K4me1 and H3K27, which were enriched in loci near liver cancer-related genes that were dysregulated following neonatal exposure to BDE-99. Our findings provide novel insights that early life exposure to POPs can have a life-long impact on disease risk, which may partly be regulated by the gut microbiome.
Recent data suggest that there is a sensitive developmental time window for toxicant exposures that may produce long-lasting effects later in life, resulting in increased susceptibility to complex diseases (Almeida et al., 2019; Garry et al., 2015; Goyal et al., 2019; Hoffman et al., 2017; Wassenaar and Legler, 2017; Weinhouse et al., 2014). For example, the use of diethylstilbestrol (DES), a synthetic estrogen first manufactured in 1938 and prescribed by US physicians to pregnant women to prevent miscarriages led to a 40-fold increase in the vaginal and cervical cancer risks in daughters who reached puberty (Kinch, 1982; Troisi et al., 2016). Bisphenol A (BPA), a more recent example, has been used to make polycarbonate plastics in consumer products and is ubiquitously detected in humans. BPA has been proposed as a possible human carcinogen (Konieczna et al., 2015; Seachrist et al., 2016). Perinatal exposure to BPA increased liver cancer incidence when the indirectly exposed mouse pups reached adulthood (Weinhouse et al., 2014, 2015). Furthermore, early-life exposure to BPA leads to perturbations in the epigenetic memory in the liver, and such developmental reprogramming combined with a western-like diet resulted in altered transcription rates of downstream targets that persistently altered liver metabolism in rats (Trevino et al., 2020). Together, these studies on early life toxicant exposure-mediated delayed onset of human diseases (such as cancer) form the basis of the Developmental Origins of Health and Diseases (DOHaD) hypothesis (Heindel and Vandenberg, 2015; Mandy and Nyirenda, 2018).
Previous investigations of liver developmental reprogramming from early life exposures have used BPA as a representative compound. Relatively less is known regarding how early-life exposure to other human health-relevant environmental chemicals, such as persistent organic pollutants (POPs), modulate acute and persistent health outcomes in the liver and other biocompartments. POPs often resist natural degradation and decomposition and bio-accumulate in the body of higher vertebrates including humans. One class of POPs is polybrominated diphenyl ethers (PBDEs), which were previously used as flame retardants. PBDEs have been linked to thyroid hormone disorders, neurotoxicity, oxidative stress in the liver, and liver cancer in both animal models and humans (Albert et al., 2018; Blanco et al., 2014; Bruchajzer et al., 2014; Cowell et al., 2019; de Water et al., 2019; Dorman et al., 2018; Dunnick et al., 2018; Gibson et al., 2018; Yuan et al., 2017). PBDEs are known endocrine disruptors (EDCs) and have been implicated in the disruption of estrogenic activity, and estrogen levels regulate thyroid hormones (Allen et al., 2016). Polychlorinated biphenyls (PCBs) are another class of POPs and endocrine disruptors (Ma and Sassoon, 2006) that were manufactured for use as coolants and insulators. PCBs are implicated in diseases, such as type II diabetes (Dzierlenga et al., 2019; Meek et al., 2019; Rahman et al., 2019; Tornevi et al., 2019) and cancer (Georgiadis et al., 2019; Ghosh et al., 2018; He et al., 2017; Leng et al., 2016). Due to the environmental persistence and lipophilicity of PBDEs and PCBs, neonates are at risk for higher exposure levels from sources, such as human breast milk, as well as lipid-rich diet, such as meat and dairy products (Quinn et al., 2011; Schecter et al., 2006). Although the acute toxicities of PBDEs and PCBs have been well-documented, there is a critical knowledge gap regarding the extent of early life exposures to PBDEs and PCBs in producing persistent toxicological outcomes. Thus, the persistent effect from early-life exposure to PBDEs and PCBs remains an important area of research.
The intestine is the first major drug-metabolizing organ to encounter xenobiotics following oral exposure to xenobiotics. The gut microbiome is increasingly recognized as an important regulator for xenobiotic biotransformation within the intestinal environment, mainly through reduction and hydrolysis reactions (Spanogiannopoulos et al., 2016; Zimmermann et al., 2019). In addition, the gut microbiome converts certain food constituents and host intermediary metabolites into microbial metabolites that may act as signaling molecules to interact with host receptors in other xenobiotic-metabolizing organs, such as the liver. Among the byproducts produced and regulated by the gut microbiome, short-chain fatty acid (SCFA) metabolites, such as acetate and butyrate, can modulate host cells through epigenetic modification and gene expression (Davie, 2003; Sabari et al., 2017). For example, acetate and butyrate have been shown to act as histone deacetylase (HDAC) inhibitors (Bolduc et al., 2017; Davie, 2003; Soliman and Rosenberger, 2011), thereby increasing histone acetylation levels, which lead to a more transcriptionally active chromatin state (Bolduc et al., 2017). Furthermore, succinate, a precursor for propionate, can promote histone 3 lysine 4 monomethylation (H3K4me1) (Xiao et al., 2012); histone succinylation is linked to active gene transcription (Smestad et al., 2018).
Importantly, early-life disruption of the gut microbiome is linked to the onset of human diseases later in life, including immune-related disorders, asthma (Martinez and Guerra, 2018; Stein et al., 2016; von Mutius and Vercelli, 2010), and metabolic diseases, such as obesity and Type II diabetes (Bassols et al., 2016; Kalliomaki et al., 2008; Scott et al., 2016). This indicates that the gut microbiome is a critical target for DOHaD. Altered gut microbiome assemblages may serve as novel biomarkers and key regulators of pathogenesis following toxic environmental chemical exposure. In adult mice, it has been previously demonstrated that acute oral exposure to BPA, PBDEs, and PCBs resulted in changes in gut microbial composition corresponding to changes in microbial metabolites and host metabolic functions (Cheng et al., 2018; Lai et al., 2016; Li et al., 2017, 2018; Lim et al., 2020; Petriello et al., 2018). However, during development, most studies on DOHaD and environmental chemicals have only focused on the host factors, such as developmental reprogramming, and very little is known about how the gut microbiome contributes to persistent effects in the host. Investigations of perinatal BPA exposure have found a positive association between BPA-mediated changes in microbial SCFAs and chronic inflammation in rabbits (Reddivari et al., 2017), as well as metabolic syndrome and immune response in mice (Malaise et al., 2017). In addition, previous investigations have shown associations between early-life exposure to PBDEs or PCBs and gut dysbiosis in humans and animal models (Gomez et al., 2021; Iszatt et al., 2019; Laue et al., 2019; Rude et al., 2019).
The liver is a major organ for xenobiotic biotransformation and nutrient homeostasis. In the US and worldwide, the incidence of liver diseases, such as nonalcoholic fatty liver disease and liver cancer are increasing (Bosch et al., 2004; Younossi et al., 2016, 2018), some of which can be attributable to environmental exposures. There are extensive communications between the liver and the gut microbiome, forming the gut-liver axis. The liver regulates the gut microbiome through biliary excreted molecules including xenobiotics, nutrients, and host-derived intermediary products, such as the primary bile acids (BAs), whereas the gut microbes communicate with the liver via enterohepatic circulation of metabolized xenobiotics, secondary BAs, and other microbial metabolites (Konturek et al., 2018; Tripathi et al., 2018). Thinning of the gut lining, which can cause inflammation in host cells and changes in microbial composition, may increase gut permeability and increase the systemic circulation of bacterial constituents, such as lipopolysaccharides, which can lead to inflammation in the liver and other organs. Within the gut-liver axis, through comparing adult conventional mice (with normal gut microbiome) and germ-free mice (without microbiome), we have previously shown that the presence of the gut microbiome is necessary for modulating PBDE- and PCB-mediated regulation of genes encoding drug-metabolizing enzymes, such as cytochrome P450s, glutathione-S-transferases, as well as xenobiotic transporters (Li et al., 2017; Lim et al., 2020). Therefore, it is important to understand the role and regulation of the gut-liver axis during environmental chemical-induced toxicity.
Most recent investigations have focused on host mechanisms underlying DOHaD, and very little is known regarding the contribution of the gut microbiome to DOHaD within the context of environmental toxicology. The objective of the present study was to fill two critical knowledge gaps: first, we aimed to examine the effects of early-life exposure to POPs, i.e., PBDE and PCBs, and compare their effects to the well-studied environmental chemical BPA. The second goal of the study was to test our working hypothesis that early life exposure to human health-relevant environmental pollutants, i.e., BPA, PBDE, and PCBs, persistently alters the gut microbiome, which in turn leads to dysregulation of microbial metabolites that are known as epigenetic modifiers of the host genome. These epigenetic modifications may contribute to the adult onset of complex diseases, such as liver cancer.
MATERIALS AND METHODS
Chemicals
2,2′,4,4′,5-pentabromodiphenyl-ether (BDE-99) (CAS No. 60348-60-9) and BPA were purchased from AccuStandard, Inc (New Haven, Connecticut). The Fox River PCB mixture was prepared from technical PCB mixtures as we described previously (Cheng et al., 2018; Kostyniak et al., 2005; Lim et al., 2020). Specifically, the Fox River PCB mixture is composed of 4 Aroclors, namely Aroclor 1242 (Lot No. KB 05-415), Aroclor 1248 (Lot No.: 106-248), Aroclor 1254 (Lot No. KB 05-612), and Aroclor 1260 (Lot No. 021-020-1A), which were all purchased from Accustandard. Solutions of each Aroclor (50 mg/ml in acetone) were combined in a ratio of 35:25:15:15% by weight for Aroclor 1242, Aroclor 1248, Aroclor 1254, and Aroclor 1260, respectively (Kostyniak et al., 2005). The Fox River mixture was authenticated by the Synthesis Core of the Iowa Superfund Research Program, supported by NIH Grant P42 ES013661. Corn oil vehicle was purchased from Sigma-Aldrich (St. Louis, Missouri). All other chemicals, unless otherwise noted, were also purchased from Sigma-Aldrich. Cholic acid (CA), chenodeoxycholic acid (CDCA), deoxycholic acid (DCA), lithocholic acid (LCA), α-muricholic acid (αMCA), β-muricholic acid (βMCA), ω-muricholic acid (ωMCA), ursodeoxycholic acid (UDCA), hyodeoxycholic acid (HDCA), murideoxycholic acid (MDCA) and their taurine conjugated forms (T-CA, T-CDCA, T-DCA, T-LCA, T-αMCA, T-βMCA, T-UDCA, and T-HDCA) were purchased from Steraloids (Newport, Rhode Island). T-ωMCA was a kind gift from Dr Daniel Raftery’s laboratory at the University of Washington Northwest Metabolomics Research Center. Five deuterated internal standards (IS) were used in total for LC/MS including d4-CA, d4-CDCA, d4-DCA, d4-LCA, and d4-GCDCA. d4-CA was purchased from Toronto Research Chemicals, Inc (Toronto, ON, Canada); 4-DCA and d4-CDCA were purchased from CDN Isotope, Inc (Pointe-Claire, QC, Canada); d4-LCA was purchased from Steraloids; and d4-glycine conjugated CDCA (d4-G-CDCA) was purchased from IsoSciences (Ambler, Pennsylvania).
Animals
All mice were housed according to the Association for Assessment and Accreditation of Laboratory Animal Care International guidelines (https://aaalac.org/resources/theguide.cfm), and studies were approved by the Institutional Animal Care and Use Committee at the University of Washington. Eight-week-old specific pathogen-free C57BL/6J mice were purchased from the Jackson Laboratory (Bar Harbor, Maine) and were acclimated to the animal facility at the University of Washington for at least two breeding generations before experiments. Mice were housed in standard air-filtered cages using autoclaved bedding (autoclaved Enrich-N’Pure) (Andersons, Maumee, Ohio), and were exposed ad libitum to nonacidified autoclaved water, as well as to LabDiet No. 5021 for breeding pairs, or to LabDiet No. 5010 for postweaning pups (LabDiet, St. Louis, Missouri). All chemical solutions were sterilized using the Sterflip Vacuum-Driven Filtration System with a 0.22 μm Millipore Express Plus Membrane (EMD Millipore, Temecula, California). As shown in the study design diagram (Figure 1A), litters were randomly assigned to chemical or vehicle exposures. Beginning on postnatal day 2 (PND2), pups were supralingually exposed to corn oil (vehicle, 5 ml/kg), BPA (250 μg/kg), BDE-99 (57 mg/kg), or the PCB Fox River Mixture (30 mg/kg), once daily for three consecutive days (i.e., PND2, PND3, and PND4; n = 5/sex/exposure). Mice were euthanized using either CO2 and decapitation (PND5) or CO2 followed by exsanguination via cardiac puncture (PND60). Various bio-compartments were collected at PND5 (24 h after the last exposure) and PND60 (56 days after the last exposure), including small and large intestine sections (duodenum, jejunum, ileum, and colon), feces, liver, and serum. Samples were frozen in liquid nitrogen and stored in a −80°C freezer until further analysis, as described in Figure 1A and detailed below.
Persistent dysregulation of the liver transcriptome following early-life exposure to environmental toxicants. A, Corn oil (vehicle control), BPA, BDE-99, and the Fox River PCB mixture were used to orally expose neonatal mice on PND2, PND3, and PND4. Twenty-four hours after the final dose, RNA-seq in the liver and 16s rDNA-seq in the small and large intestines were performed to characterize the acute effects from the exposure. When the exposed mice reached young adulthood at PND60, RNA-seq, metabolomics, ChIP-seq, and 16s rDNA-seq in the small and large intestines were conducted to investigate the persistent effects from the early life exposure. B, Number of dysregulated protein-coding genes (PCGs) at PND5 (black) and PND60 (red) for males (top) and females (bottom). C, Venn diagram showing intersections of dysregulated genes in neonates. D, Venn diagram showing intersections of dysregulated genes in young adulthood. E, Summary of gene ontology enrichment results by age, sex, and exposure groups. Up-regulated (left panel, red) and down-regulated (right panel, blue) biological terms related to liver function are included. The color gradient describes the average significance (p < .01) in each category per group. The top enriched proportion (size of circles) indicates the proportion of GO terms in each category relative to the top 50 terms (p < .01), i.e., larger circles indicate a higher proportion relative to the top 50 GO terms. Please refer to the article in the online version for colored figures.
16s rDNA Sequencing and Data Analysis
Microbial DNA was isolated from un-flushed duodenum, jejunum, ileum, colon, and feces at PND5 and PND60, as well as liver and serum at PND60, using an OMEGA E.Z.N.A. Stool DNA Kit (OMEGA Biotech Inc, Norcross, Georgia). To note, the microbial DNA in the liver and serum serves as an indirect biomarker for gut permeability. The concentration of DNA was determined using a Qubit fluorometer (Thermo Fisher Scientific, Waltham, Massachusetts). The integrity and quantity of all DNA samples were confirmed using an Agilent 2100 Bioanalyzer (Agilent Technologies Inc, Santa Clara, California). Bacterial 16S rDNA hypervariable region 4 (V4) amplicon sequencing was performed using a HiSeq 2500 second-generation sequencer (Illumina, San Diego, California) with 250 bp paired-end reads (n = 5 per group) using a similar method as we previously described (Cheng et al., 2018; Gomez et al., 2021; Lim et al., 2020; Scoville et al., 2019).
Microbial 16S rDNA paired-end sequence reads were merged, de-multiplexed, quality-checked, and chimera-filtered using QIIME 2 version 2020.2 (Bolyen et al., 2019). The Silva 99 version 132 reference was used (Quast et al., 2013). The operational taxonomic unit (OTU) table annotated with classified bacteria information was read into R for further analysis (R Development Core Team, 2019). Differential abundance testing was performed using ANCOM (Mandal et al., 2015), a plugin in QIIME2. Differentially abundant bacteria (relative abundance of OTUs) and differentially regulated hepatic genes were correlated (Pearson’s correlation), and significant correlation was defined as |r| > 0.8 and false discovery rate (FDR)-adjusted p value < .1 using base R. Correlation results were plotted using the R package ComplexHeatmap (Gu et al., 2016). Functional predictions of the microbial composition were conducted using PICRUSt2 (Douglas et al., 2020).
Metabolite Quantification Using GC-MS and Data Analysis
SCFAs and their intermediate precursors were quantified as we described previously (Gomez et al., 2021; Gu et al., 2021; Jasbi et al., 2019). Briefly, approximately 50 mg of each tissue sample was homogenized with 20 μl hexanoic acid-6,6,6-d3 (IS; 200 µM in H2O), 20 μl sodium hydroxide solution (NaOH, 0.5 M in water), and 480 μl methanol (MeOH). Afterward, 400 μl MeOH was added, and the pH of the mixture was adjusted to approximately 10. Upon storage at −20°C for 20 min and centrifugation at 21 694 × g for 10 min, 800 μl of supernatant were collected. Samples were then evaporated to dryness, reconstituted in 40 μl of methoxyamine hydrochloride in pyridine (20 mg/ml), and stored at 60°C for 90 min. Subsequently, 60 μl of N-Methyl-N-tert-butyldimethylsilyltrifluoroacetamide was added and samples were stored at 60°C for 30 min. Each sample was then vortexed for 30 s and centrifuged at 21 694 × g for 10 min. Finally, 70 μl of supernatant were collected from each sample for gas chromatography-mass spectrometry (GC-MS) analysis.
GC-MS experiments were performed on an Agilent 7820A GC-5977B MSD system (Santa Clara, California) by injecting 1 µl of prepared samples. Helium was used as the carrier gas with a constant flow rate of 1.2 ml/min. The separation of metabolites was achieved using an Agilent HP-5ms capillary column (30 m × 250 × 0.25 µm). The column temperature was maintained at 60°C for 1 min, and then increased at a rate of 10°C/min to 325°C and held at this temperature for 10 min. The injector temperature was 250°C, and the operating temperatures for the transfer line, source, and quadruple were 290°C, 230°C, and 150°C, respectively. Mass spectral signals were recorded after a 4.9 min solvent delay (n = 4–6 per group).
One-way analysis of variance (ANOVA) followed by Tukey’s post hoc test was performed for each metabolite in R for the analysis of liver metabolites (adjusted p < .1).
BA Quantification Using LC-MS
One mg/ml stock solutions of individual BAs (for standard curve) and IS were prepared in 50% methanol-water solution (1:1). The 19 individual BA stock solutions were further diluted in nanopure HPLC grade water to obtain 10 working standard solutions (0.05–10 000 ng/ml). The 5 ISs (d4-CA, d4-CDCA, d4-DCA, d4-LCA, and d4-GCDCA) were mixed to obtain a working IS solution in which the concentration of each IS was 10 μmol/ml.
Liver BA Extraction
All organic solvents (analytical grade) used in the procedure, unless noted otherwise, were purchased from Sigma-Aldrich. BAs were extracted from frozen liver samples (n = 4–6/group) using a similar method as we described previously (Dempsey et al., 2019; Gomez et al., 2021; Kim et al., 2021; Li et al., 2018, 2021). Briefly, approximately 50 mg of liver tissue were weighed for each sample and homogenized using nanopure HPLC grade water. For each homogenate, 10 μl of IS solution were added and mixed with 1.5 ml of acetonitrile with 5% ammonium hydroxide solution. The suspension mixtures were vortexed and shaken for 1 h at room temperature. The mixtures were then centrifuged at 12 000 × g for 15 min at 4°C, and the supernatants were collected. The second step of BA extraction was performed using 750 μl HPLC grade methanol, vortexed thoroughly, sonicated for 5 min, and centrifuged at 15 000 × g for 20 min. The supernatants were collected and were pooled with the first extracts. Supernatants were dried under vacuum and the dried residue was reconstituted in 100 μl 50% MeOH. The reconstituted sample was transferred onto the 0.22-μm Costar Spin-X centrifuge tube filter (MilliporeSigma, Burlington, Massachusetts), and centrifuged at 20 000 × g for 10 min before injection. Standard and different quality control samples were extracted using the similar sample preparation procedure described above.
UPLC-MS/MS Analysis
BAs were quantified using UPLC-MS/MS as we described before (Cheng et al., 2018; Dempsey et al., 2019; Gomez et al., 2021; Kim et al., 2021; Li et al., 2021). Briefly, an Agilent 1290 UPLC ultra-high pressure liquid chromatography (UPLC) system combined with an Agilent 6460 triple quadrupole mass spectrometer (MS) via an electrospray ionization interface was used. Chromatographic separation described was performed using a ZORBAX Eclipse Plus C18 analytical column (2.1X100 mm; id: 1.8 µm). Samples were eluted using mobile phase A (20% acetonitrile and 10 mmol/l ammonium acetate in aqueous solutions) and mobile phase B (80% acetonitrile and 10 mmol/l ammonium acetate in aqueous solutions) at a flow rate of 0.400 ml/min. The gradient profile of the mobile phase A and mobile phase B for the LC pump under the final chromatography conditions were as follows: 0 min, 95:5; 5 min, 95:5; 14 min, 86:14; 14.5 min, 75:25; 17.50 min, 75:25; 18 min, 50:50; 22 min, 50:50; 22.50 min, 20:80; 24.50 min, 20:80; 25–28 min, 95:5 (mobile phase A: mobile phase B, v/v). The injection volume of the samples was 5 µl. The column temperature was set at 45°C, and the sample tray temperature was maintained at 9°C. MS/MS spectra were produced using the negative ionization mode.
Serum Alanine Aminotransferase Activity Quantification
Serum alanine aminotransferase (ALT) levels of PND60 males and females neonatally exposed to vehicle (corn oil [CO]) or BDE-99 were measured using the ALT Activity Assay Kit following manufacturer’s procedures (Cat. MAK052, Sigma-Aldrich, Missouri). Data were corrected by blank wells and standard readings, and the linear change in values from the initial and final measurements were used to calculate ALT activity.
RNA Isolation
Total RNA was isolated from frozen livers (PND5 and PND60) and large intestinal tissue (PND60) using the RNA-Bee reagent (Tel-Test Inc, Friendswood, Texas) according to the manufacturer’s protocol. RNA concentrations were quantified using a NanoDrop 1000 Spectrophotometer (Thermo Scientific, Waltham, Massachusetts) at 260 nm. The integrity of total RNA samples was evaluated by formaldehyde-agarose gel electrophoresis with visualization of 18S and 28S rRNA bands under UV light and confirmed by an Agilent 2100 Bioanalyzer (Agilent Technologies Inc, Santa Clara, California). Liver RNA samples with RNA integrity (RIN) values above 8.0 were used for RNA-Seq.
RT-qPCR Quantification of Large Intestinal Tissue at PND60
Total RNA isolated from large intestinal tissues (n = 3–5) at PND60 was reverse transcribed into cDNA using a High Capacity cDNA Reverse Transcription Kit (Life Technologies, California). The resulting cDNA products were amplified by qPCR, using the Sso Advanced Universal SYBR Green Supermix in a Bio-Rad CFX384 Real-Time PCR Detection System (Bio-Rad, Hercules, California). Primer sequences are shown in Supplementary Table 1. Data were normalized to the housekeeping gene Gapdh using the ΔΔCq method and were expressed as % of Gapdh.
Small RNA Sequencing of Liver Samples and Data Processing
In triplicates, small RNA sequencing was performed subject to adult male and female livers. The cDNA library was constructed using an 18–40 bp insert method, followed by second-generation sequencing (HiSeq SE50, 20M raw reads/sample, Illumina, San Diego, California). Quality assurance and control of the FASTQ files was performed using FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Reads were mapped to the mouse reference genome and counts were generated using STAR (Dobin et al., 2013) following the pipeline from ENCODE (https://www.encodeproject.org/microrna/microrna-seq/). Transcript abundance was normalized to counts per million (CPM). Differential expression analysis was performed using DESeq2 (Love et al., 2014). Plots were generated using the R package ComplexHeatmap (Gu et al., 2016).
Whole Transcriptome Sequencing of Liver Samples and Data Processing
In triplicates, the cDNA library was constructed using a ribosomal depletion method, and reads were sequenced (75 bp paired-end reads) per the manufacturer’s protocol (Illumina, San Diego, California). FASTQ files were de-multiplexed and concatenated for each sample. Quality control of the FASTQ files was performed using FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Reads were then mapped to the mouse reference genome (National Center for Biotechnology Information [NCBI GRCm38/mm10]) using HISAT2 version 2.1 (Kim et al., 2015). The sequencing alignment/map (SAM) files were converted to binary alignment/map (BAM) format using SAMtools version 1.8 and were analyzed by Cufflinks version 2.2.1 to estimate transcript abundance using Gencode mouse version 19 (vM19) gene transfer format (GTF). Long noncoding RNA (lncRNA) GTF from Gencode (vM19) was used to estimate the lncRNA transcript abundance. The abundance was expressed as fragments per kilobase of transcript per million mapped reads (FPKM) which was then converted to transcripts per million (TPM). Differential expression analysis was performed using Cuffdiff (Trapnell et al., 2010). The differentially expressed genes were defined as false discovery rate Benjamini-Hochberg adjusted p value (FDR-BH) < .05 in the chemical-exposed groups compared with the vehicle-exposed control group.
RNA-Seq Data Analysis
All analyses were done in R unless stated otherwise. For RNA-Seq, samples were categorized by age, sex, and exposure. Genes were considered expressed if the average TPM was > 1 for at least one group. Protein coding genes (PCGs) were separately categorized based on the Ensembl BioMart gene category. Differentially expressed PCGs were also categorized and matched with genes in categories of interest (xenobiotic biotransformation, epigenetic modifiers, nuclear receptors, oxidative stress, and cellular death). These specific gene categories were based on literature searches and extracting gene sets from the Gene Ontology consortium and the Kyoto Encyclopedia of Genes and Genomes (KEGG). Venn diagrams were made for differentially regulated PCGs and lncRNAs for each sex and age group (PND5 and PND60) using the R package VennDiagram (Chen and Boutros, 2011). Hierarchical clustering was performed using the R package ComplexHeatmap. Up- and down-regulation of transcripts were defined as the absolute value of the fold change greater than 1.5. Lists of up- and down-regulated transcripts were used as input for gene ontology enrichment using the R package topGO (Alexa and Rahnenfuhrer, 2020) for all groups. The list of genes in the unfiltered expression table was used as the background for gene ontology. Upstream regulators for all differentially regulated genes were determined using the Ingenuity Pathway Analysis (IPA) software (Qiagen, Hilden, Germany). Gene expression values were shown as bar plots for specific genes of interest using the arithmetic mean TPM in SigmaPlot (SPSS Inc, Chicago, Illinois).
Chromatin Immunoprecipitation Coupled with Second-Generation Sequencing (ChIP-Seq) and Data Analysis
Frozen pooled livers from PND60 males and females neonatally exposed to BDE-99 were subjected to ChIP with validated antibodies (Active Motif, Carlsbad, California) for the following three histone marks associated with active gene transcription: histone 3 lysine 4 monomethylation (H3K4me1; Cat No. 39297) and trimethylation (H3K4me3; Cat No. 39159), and H3K27 acetylation (H3K27ac; Cat No. 39133). Following DNA extraction, 75-nt single-end (SE75) sequence reads were generated by Illumina sequencing (using a NextSeq 500) and mapped to the mm10 mouse reference genome using the BWA algorithm (Li and Durbin, 2009). Reads that passed Illumina’s purity filter, aligned with no more than 2 mismatches, and mapped uniquely to the genome were used for the subsequent analysis. Peak calling, standard normalization, spike-in adjustment, merged region analysis, and genomic annotations were performed by Active Motif. Genes that had positive epigenetic mark enrichment within ±10 kb of the gene loci were considered target genes of that epigenetic mark. The union of overlapping intervals (merged region) was used to make Venn diagrams to illustrate common and unique epigenetic marks in control and BDE-99 exposed groups. A 30% change in the average peak value (i.e., BDE-99/vehicle > 1.3 or < 0.7) was considered as a significant alteration in the epigenetic mark associated with the gene. ChIP-Seq and RNA-Seq datasets were overlaid to examine the association between the alterations in epigenetic marks and changes in target gene expression, and data were visualized as pie charts (Huang et al., 2013) and analyzed using topGO (Alexa and Rahnenfuhrer, 2020) and STRING (https://string-db.org/).
RESULTS
Transcriptomic Changes in Liver following Early-Life Exposure to Environmental Contaminants BPA, BDE-99, and PCB
We used RNA-Seq to investigate the impact of early life environmental chemical exposure on the liver transcriptome at PND5 (i.e., 24 h postdose) for acute response to chemical exposure in newborns, and PND60 (i.e., 2 months after the chemical exposure stopped) for persistent responses to early-life chemical exposure in young adult males and females. As shown in Figure 1B top panel, at the given doses, early life exposure to BDE-99 amplified the transcriptional effects on PCGs along the developmental trajectory, evidenced by more PCGs that were differentially regulated in PND60 than in PND5. Fewer transcriptomic changes were associated with BPA exposure, and exposure to PCB mixture had the least effect at both ages. In addition, males were more susceptible than females to the persistent effect. Specifically, in livers of PND5 male pups, acute exposure to BDE-99 altered the expression of the greatest number of genes (271), followed by BPA (113) and the PCB mixture (79). Interestingly, in livers of PND60 males, the persistent effect of BDE-99 was greater than BPA, as evidenced by 792 persistently regulated genes by neonatal BDE-99 exposure, and 555 persistently regulated genes by neonatal BPA exposure. The effect of neonatal exposure to the PCB mixture was minimal, with 68 persistently regulated genes by the PCB mixture in livers of PND60 male pups. As shown in Figure 1B bottom panel, in general, livers from female pups were more resistant to both the acute and the persistent hepatic transcriptomic effects from all three chemicals at both PND5 and PND60 ages as compared with livers from the age- and exposure-matched male pups. Among the three chemicals, BPA had the most persistent effect (179 genes persistently regulated genes by BPA in PND60) in adult livers of female pups.
Venn diagrams (Figs. 1C and 1D) display the common and uniquely regulated genes by the three chemicals in both neonatal and adult ages. At both PND5 and PND60, BDE-99 had the greatest number of uniquely regulated genes, followed by BPA and the PCB mixture in livers of male pups. Conversely, BPA had the most numbers of uniquely regulated genes in female pups of both ages. The commonly regulated genes by all three chemicals were minimal at both ages. Specifically, in PND5 males, as shown in Figure 1C and Supplementary Tables 2A–C, among the three chemicals, acute and persistent responses to BDE-99 had the most prominent effect, having 199 uniquely regulated PCGs. PCBs and BDE-99 shared the highest number of commonly dysregulated PCGs. In contrast, PND5 females had fewer differentially regulated PCGs in response to all three chemicals, with BPA having the most prominent effect (Figure 1C and Supplementary Tables 2D–F). For adult males, BDE-99 also had the greatest number of dysregulated genes, of which over 50% were shared with BPA (Figure 1D and Supplementary Tables 2G–I). In adult females, BPA had the most prominent transcriptomic impact, followed by BDE-99 and PCBs (Figure 1D and Supplementary Tables 2J–L).
We predicted the upstream transcriptional regulators by examining differentially expressed genes of mouse livers in PND5 (Supplementary Table 3) and PND60 (Supplementary Table 4) using the Ingenuity knowledge base (QIAGEN, Inc, Redwood City, California). As shown in Supplementary Table 3, at PND5, the top upstream regulators involved in the acute response to BPA included the major xenobiotic-sensing nuclear receptors PXR (Nr1i2) and CAR (Nr1i3), and both were predicted to be inhibited in livers of male pups. The fatty acid beta-oxidation enzyme Acyl-CoA oxidase 1 (Acox) was also predicted to be inhibited in livers of BPA-exposed male pups. For BDE-99, the top upstream regulators in livers of PND5 male pups are mostly involved in immune response, including a predicted activation of the nuclear receptor mediator Tripartite Motif Containing 24 (Trim24) and Il10ra, both of which are involved in the inhibition of the synthesis of pro-inflammatory cytokines (Iyer and Cheng, 2012; Perez-Lloret et al., 2016). There was also a predicted inhibition of interferon beta 1 (Ifnb1), Stat1, interferon-gamma (Ifng), as well as interferon regulatory factors (Irf) Irf3 and Irf7 in livers of PND5 male pups. There were no significant predicted alterations in any upstream regulators in BPA- and BDE-99 exposed female pups. PCBs had a minimal predicted effect on the upstream regulators, except for a predicted inhibition of Bruton tyrosine kinase (Btk), which is involved in B cell maturation, in livers of female PND5 pups.
As shown in Supplementary Table 4, regarding the persistent effect of BPA, at PND60, the predicted upstream regulators include a persistent inhibition of the tumor suppressor Trp53 (P53) and its target gene cyclin-dependent kinase inhibitor 1A (Cdkn1a) involved in cell cycle G1 phase arrest, as well as a persistent activation of the cancer-promoting E2F transcription factor 1(E2f1) and the pro-inflammatory cytokine colony-stimulating factor 2 (Csf2), in both sexes. In addition, in livers of PND60 males, there was a predicted activation of additional genes that are frequently overexpressed in cancers, including Erb-B2 receptor tyrosine kinase 2 (Erbb2), cyclin D1 (Ccnd1), E2F transcription factor 3 (E2f3), as well as the nuclear receptor peroxisome proliferator-activated receptor alpha (Ppara), all of which are involved in cell proliferation. In livers of PND60 females, there was a predicted inhibition in transcription factor 3 (Tcf3), which is involved in lymphogenesis, as well as a predicted activation of the inflammation-related prostaglandin E receptor 2 (Ptger2), the oncogene RAB member RAS oncogene family-like 6 (Rabl6), and the tumorigenesis-related gene hepatocyte growth factor (Hgf). Together these predicted gene expression signatures indicate cancer- and inflammation-prone outcome by early-life BPA exposure.
Regarding the persistent effect of BDE-99, at PND60, similar to BPA, the predicted upstream regulators also include the inhibition of the tumor suppressor Trp53 (P53) (Supplementary Table 4). However, this effect was only observed in males but not in females. In livers of PND60 males, upstream regulators involved in basal liver functions, such as lipid homeostasis (SREBF chaperone [Scap]), drug metabolism (CAR [Nr1i3]), and copper elimination (Atp7b) were also predicted to be inhibited. Conversely, cancer markers, such as Erbb2 and Ccnd1 and the pro-inflammatory cytokine interleukin 6 (Il6) were predicted to be up-regulated. Fibroblast growth factor 15 (Fgf15), which is involved in the down-regulation of BA biosynthesis, as well as cytochrome P450 Oxidoreductase (Por), were also predicted to be up-regulated in livers of PND60 males. In livers of PND60 females, Insulin 1 (Ins1) was the only significantly activated upstream regulator, and there were no other regulators predicted to be altered by early life BDE-99 exposure in females.
Neonatal PCB mixture exposure had a minimal persistent effect on the predicted upstream regulators (Supplementary Table 4). The two factors predicted to be altered were an inhibition in Myogenic Differentiation 1 (Myod1) and the lipid metabolism-related protein Scap, both occurring in livers of PND60 males but not in females.
To predict the functional consequences of the acute and persistent transcriptomic changes following neonatal exposure to the three chemicals, we performed gene ontology (GO) enrichment on differentially expressed genes. We classified GO terms into 35 liver-relevant categories, including chromatin epigenetic modification, cell cycle, and DNA synthesis, drug metabolism, and lipid metabolism (Supplementary Table 5). To characterize and compare functional enrichment, the top 50 enrichment results were summarized in biological GO terms that had at least 5 significant terms (p < .01) in the same category (Figure 1E). Overall, following acute exposure, dysregulated pathways include blood and immune response, chromatin epigenetic modification, cell cycle, and DNA synthesis, drug and lipid metabolism, and transporters. Interestingly, we observed persistent up-regulation in the pathways involved in cell cycle, DNA synthesis, and chromatin epigenetic modification in livers of adult pups that were neonatally exposed to BPA (males and females) and BDE-99 (males). In addition, neonatal exposure to BPA and BDE-99 resulted in persistent down-regulation of drug and lipid metabolism pathways in livers of adult male pups.
In summary, among the three toxicants, BDE-99 produced the most prominent persistent liver transcriptome alterations, followed by BPA and PCBs. Early life BPA and BDE-99 led to cancer- and inflammation-prone gene expression signatures and suppressed basal liver functions, such as lipid and drug metabolism, with males being more susceptible than females.
Differential Regulation of Hepatic Genes Involved in Xenobiotic Biotransformation and Their Regulators in Response to Early-Life Exposure
One of the major functions of the liver is xenobiotic biotransformation, including phase-I and -II drug metabolism as well as transporters. Furthermore, the ontogeny of various PCGs encoding drug-metabolizing enzymes and transporters is under the tight control of transcriptional and epigenetic regulation. As a result, we examined the acute and persistent regulation of genes involved in xenobiotic biotransformation, epigenetic modification, nuclear receptors, and oxidative stress and inflammation markers in livers of PND5 (Supplementary Figures 1 and 2) and PND60 mice (Figure 2 and Supplementary Figure 3A) after chemical exposures.
Specific categories of persistently dysregulated genes following early-life exposure to environmental toxicants in males. Two-way hierarchical clustering of differentially regulated genes involved in xenobiotic biotransformation (A) and oxidative stress and inflammation (B) in livers of PND60 adult mice following neonatal chemical exposure. C, Disease enrichment of persistently dysregulated genes following early-life exposure to BDE-99. The color gradient in the legend represents the adjusted p value in −log10 scale. D, Gene expression values of PCG and long noncoding RNAs (lncRNAs) in the Dlk1-Dio3 imprinted cluster. Expression values are shown in the same y-axis scale for comparison. E, Two-way hierarchical clustering of differentially regulated microRNAs (miRNAs) in the Dlk1-Dio3 imprinted cluster. F, An example of differentially regulated lncRNAs associated with liver tumors. All individual bar graphs with gene symbols show the average transcripts per million (TPM) ± standard error (SE) of the genes in the cluster. Asterisks represent statistically significant differences compared with vehicle-exposed groups. Asterisks (“*” p value < .05) were placed for lncRNAs that were differentially expressed but are not visible due to low expression values within the same y-axis scale. Colored bars in heatmaps represent genes that are differentially regulated in a particular exposure group, i.e., BPA—red, BDE-99—blue, PCB mixuture—green. Please refer to the article in the online version for colored figures.
Specifically, as shown in Supplementary Figure 1, in livers of male PND5 pups, in general, BPA had the most prominent effect on the transcriptional regulation of phase-I and -II drug-metabolizing enzymes. This was followed by BDE-99, which had less dramatic changes. Furthermore, most of the differentially regulated genes were up-regulated by BPA (and to a lesser extent by BDE-99). For transporters, whereas BPA mainly up-regulated the targeted transporter genes (Slco1a1, Scl2b1, Slc22a30, Slc22a26, Abcc3, Slco1a4, Atp6v0d2, and Slc17a4), BDE-99 mainly down-regulated the targeted transporter genes (Slc17a4, Abca7, Slc38a5, Slc14a1, Slc43a3, Slc22a4, Slc7a1, Atp1b2, Slc25a38, Slc4a9, Slc6a9, and Slc16a10). In general, there was minimal overlap between BPA and BDE-99-targeted transporters. Females were more resistant to the acute chemical effects, and most differentially regulated genes in livers of female pups were P450s that were up-regulated by BDE-99 (Cyp1a2, Cyp2b10, Cyp2c37, Cyp2c50, and Cyp2c54) except for Cyp2f2 which was down-regulated by BPA. In both sexes, PCBs had minimal effect than the other two chemicals, including a down-regulation of Cyp4a10, Slc7a1, Atp1b2, Slc4a9, Scl25a47, and an up-regulation of Cyp1a2, Cyp2c50, Cyp2c54, Cyp2a4, Ugt3a2, Slc17a3, and Slc17a4 in males, with no observed effects in females. Supplementary Figure 2 shows the acute chemical effect on the expression of genes involved in oxidative stress, inflammation, epigenetics, and nuclear receptors in livers of PND5 male and female pups. In general, genes encoding epigenetic modifiers and nuclear receptors were minimally influenced by these chemicals. RAR-related orphan receptor gamma (Rorc) was the only nuclear receptor that was differentially regulated in livers of PND5 females (up-regulated by BDE-99). Acute BDE-99 exposure up-regulated the expression of the epigenetic modifier Smarca2, which can alter chromatin structure around targets, and down-regulated H2afx (histone subunit). Genes involved in oxidative stress and inflammation were regulated by BDE-99, and to a lesser extent by BPA, in livers of male pups. The PCB mixture had minimal effects. Notably, there was an apparent down-regulation pattern by BPA and BDE-99 in the expression of genes involved in oxidative stress. In addition, BDE-99 had the most prominent effect on genes involved in inflammation (Supplementary Figure 2, left panel), including a down-regulation of genes involved in the regulation of chemokine production and interleukin-1; cytokine signaling, such as early growth response 1 (Egr1), Il1rl1, metallothionein 2 (Mt2), erythropoietin receptor (Epor), and sphingosine kinase 1 (Sphk1); and up-regulation of genes involved in interferon signaling and myeloid cell differentiation, i.e., interferon-induced protein with tetratricopeptide repeats (Ifit3), interferon-activable protein 204 (Ifi204), interferon regulatory factor 1 (Irf1), signal transducer and activator of transcription 1 (Stat1), and colony-stimulating factor 1 (Csf1). Female pups were more resistant to the acute effects of all three chemicals on the regulation of genes in these categories (Supplementary Figure 2, right panel).
In PND60 male pups, whereas neonatal exposure to the PCB mixture had a minimal persistent effect, we observed a robust BDE-99 and BPA-mediated persistent down-regulation of the mRNAs of numerous phase-I and -II drug-metabolizing enzymes in the liver. This may indicate that the trajectory of the regulatory pattern is reversed compared with PND5 and that detoxification capabilities are decreased following early-life exposure to BPA and BDE-99 (Figure 2A). Most of the differentially regulated transporter genes were persistently up-regulated by BPA and BDE-99; many of these are involved in nutrient homeostasis. Among all three chemicals, the most prominent persistent effects were from BDE-99 exposure (Figure 2A). Specifically, early life BDE-99 exposure persistently down-regulated genes involved in phase-I oxidation (Cyp1a2, Cyp2d10, Cyp2c50, Cyp4f14, Cyp7a1, Cyp8b1, Cyp27a1, and Adh1), and phase-II glutathione conjugation (Gsta’s, multiple Gstm isoforms, Gstp1, and Gstt1). BDE-99 persistently up-regulated metal transporters (Slc39a4, Atp7a, Slc40a1) and xenobiotic uptake transporter Slc22a7 (Oat2) but decreased the xenobiotic uptake transporter Slco2b1 (Oatp2b1) and the xenobiotic efflux transporter Abcc3 (Mrp3) in livers of PND60 males (Figure 2A). In contrast, PND60 females were less responsive to neonatal exposure to the three chemicals, as evidenced by fewer numbers of persistently regulated genes in all categories (Supplementary Figure 3A, left panel).
Early-life exposure to BDE-99 and BPA persistently up-regulated most of the differentially regulated genes involved in oxidative stress and inflammation, whereas the PCB mixture had minimal effect in PND60 males (Figure 2B). Early-life exposure to BPA or BDE-99 up-regulated phospholipase C gamma 2 (Plcg2), spleen-associated tyrosine kinase (Syk), annexin A1 (Anxa1), lipocalin 2 (Lcn2), heme oxygenase 1 (Hmox1), Mt1, Mt2, and leptin receptor (Lepr) in adult males. These genes are involved in B cell receptor signaling, metal ion homeostasis, Jun kinase regulation, or T cell differentiation. Neonatal BDE-99 exposure also up-regulated other genes involved in pattern recognition and intercellular adhesion, e.g., C-type lectin domain family 7 member A (Clec7a), toll-like receptor 7 (Tlr7), integrin subunit beta 2 (Itgb2), and intercellular adhesion molecule 1 (Icam1). In adult females, neonatal exposure to BPA persistently up-regulated pro-inflammatory genes, such as Lcn2, which was also up-regulated by the PCB mixture exposure, and S100 Calcium Binding Protein A9 (S100a9). Egr1 was down-regulated by the PCB mixture (Supplementary Figure 3A, right panel). Neonatal exposure to BDE-99 down-regulated metal binders (i.e., Mt1 and Mt2) and Cdh1, which are involved in cell adhesion, and up-regulated BCL6 transcription repressor (Bcl6), which is a regulator of naive helper T cell differentiation (Supplementary Figure 3A, right panel). The persistent up-regulation of inflammation and oxidative stress in males may predispose them to liver injuries and cancer later in life.
Because among all the three toxicants investigated, BDE-99 produced the most persistent hepatic transcriptomic signatures that are predicted to be involved in liver cirrhosis and cancer in adult male pups (Figure 2C), we quantified serum ALT, which is a prototypical hepatocyte injury biomarker in control and BDE-99 exposed adult pups. As shown in Supplementary Figure 3B, early life BDE-99 exposure did not alter the serum ALT levels in either male or female adult pups. Although there was a trend of increase in serum ALT in adult male pups, it was not statistically significant, and the absolute values were below levels normally associated with the liver injury. In addition, we further examined cell death-related gene expression profiles in all exposure groups at both developmental ages (Supplementary Figs. 3C and 3D). In PND5, the numbers of differentially regulated death-related genes were minimal in male pups across all three chemical-exposed groups (2–4 genes); whereas in female pups, the numbers were slightly higher than males for all three chemical exposed groups, with the highest number observed in BDE-99 exposed group (33 genes), followed by the PCB mixture (9 genes) and BPA (6 genes). Interestingly, in postnatal day 60, whereas early life PCB exposure did not affect many cell-death related genes in either sex (13 genes in males and 3 genes in females), early life exposure to the other two chemicals persistently regulated many cell-death related genes in livers of adult male pups, and BDE-99 had the most prominent effect (113 genes in males) followed by BPA (75 genes in males). In livers of adult female pups, relatively less cell-death-related genes were affected (29 genes by BPA and 8 genes by BDE-99). In summary, these observations indicate that the early life chemical exposure-mediated transcriptomic changes are early biomarkers that may precede the onset of hepatocyte injury, and the differential expression is likely mediated by specific transcriptional mechanisms rather than injury-related secondary effects.
Because early life BDE-99 generally produced the most persistent transcriptomic response than the other two chemicals, and adult male pups were more sensitive than females, we conducted additional downstream analyses focusing on male adult pups following early life BDE-99 exposure to explore disease-related biomarkers. First, using DisGeNET (Pinero et al., 2020), we investigated the gene-disease associations in adult male pups following early life BDE-99 exposure (Figure 2C). Liver cirrhosis was among the top matches, followed by liver carcinoma, carcinogenesis, liver neoplasms, and adenocarcinoma (Figure 2C). Second, the imprinted DLK1-DIO3 locus, containing numerous maternally expressed ncRNAs including the lncRNA maternally expressed gene 3 (Meg3) and a cluster of over 50 miRNAs, is a modulator of stemness in embryonic stem cells and cancer progression, potentially through the tumor suppressor role of Meg3, whose altered expression is associated with cancer initiation, progression, metastasis and chemo-resistance (Budkova et al., 2020; He et al., 2017; Miguel et al., 2020). Additionally, dysregulation of the miRNA hotspot within the Dlk1-Dio3 cluster can lead to cancer formation, metastasis, and epithelial to mesenchymal transition (Benetatos et al., 2014; Haga and Phinney, 2012), and up-regulation is associated with poor clinical outcome of hepatocellular carcinoma (Luk et al., 2011). Interestingly, both the lncRNAs and the miRNAs in the Dlk1-Dio3 gene cluster were persistently up-regulated (Figs. 2D and 2E). Specifically, Meg3 and RNA imprinted and accumulated in nucleus (Rian), both of which are lncRNAs closely linked to cell cycle regulation and tumorigenesis (Luisier et al., 2014; Wang et al., 2012), were persistently upregulated by BPA in both sexes in PND60, and by BDE-99 in PND60 male pups. We observed a persistent up-regulation of all miRNAs in the Dlk1-Dio3 locus following early-life exposure to BDE-99 in males, which was consistent with the disease prediction observed using the dysregulated PCGs (Figs. 2C and 2E). Furthermore, lncRNAs involved in cell cycle regulation, i.e., H19, small nucleolar RNA host genes (Snhg) Snhg1 and Snhg4 were persistently up-regulated from early life exposure to BDE-99 (Figure 2F). Together these results suggest that early-life exposure to BDE-99 reprogrammed the liver transcriptome and persistently up-regulated PCGs and ncRNAs involved in cell cycle regulation toward a consistently proliferative state.
Persistent Gut Dysbiosis Throughout the Intestinal Tract and Feces Following Early Exposure to Environmental Contaminants
In parallel with the hepatic transcriptomic alterations, we tested our hypothesis that the gut microbiome can also be developmentally reprogrammed both acutely and persistently following early life chemical exposures. A diagram of various sections of the intestinal tract and the fecal compartment is shown in Figure 3A. As shown in Supplementary Figure 4, left panel, at PND5, alpha diversity tended to increase by acute BDE-99 and PCB mixture exposure in the duodenum of male pups, but it decreased in female pups. In other bio-compartments, BPA and BDE-99 generally increased the alpha diversity in both sexes (Supplementary Figure 4, left panel). We did not observe apparent separations in beta diversity at PND5 among the chemical exposure groups (Supplementary Figure 4, right panel, principal coordinate analysis [PCoA] plots).
Persistent dysbiosis of the gut microbiome and upregulation of key metabolites involved in liver reprogramming and inflammation in the liver from early life exposure to environmental toxicants. A, Diagram of the duodenum (duo), jejunum (jej), ileum (ile), colon (col), and feces (fec) in physical order from left to right. Note: the intestinal contents were kept in each of the tissue sections. B, Top 10 most abundant taxa in each compartment at the family level in adult age. Asterisks represent statistically significant differences compared with the vehicle control. C, Persistently increased abundance of Akkermansia municiphila in all bio-compartments from neonatal exposure to BDE-99 in males. D, Box plots of persistently increased metabolites from early life exposure to BDE-99 in males. E, Persistently dysregulated colon microbiota significantly correlated with the persistently upregulated lactate and succinate in livers of males neonatally exposed to BDE-99. F, Significant linkage (r > 0.8, FDR < 0.1) between dysregulated colon microbiota and upregulated liver metabolites involved in epigenetic reprogramming and inflammatory signaling. Please refer to the article in the online version for colored figures.
Regarding the persistent effect at PND60, early life exposure to all three chemicals lowered the alpha diversity in the duodenum of both sexes (Supplementary Figure 5). However, in jejunum and ileum, BDE-99 increased the alpha diversity in males, whereas both BDE-99 and the PCB mixture increased the alpha diversity in females in PND60. In colon and feces, all three chemicals tended to lower the alpha diversity in both sexes in PND60 (Supplementary Figure 5). Similar to the acute changes in the gut, no qualitative changes were present for beta diversity in the small intestine (Supplementary Figure 5, right panel, PCoA plots). However, in colon and feces, male and female adults neonatally exposed to BDE-99 tended to form a separate cluster compared with other groups, indicating that neonatal exposure to BDE-99 results in a persistent alteration in beta diversity in the large intestine (Supplementary Figure, 5 right panel).
At PND5, the top 10 most abundant microbiota at the family level were visualized to determine the gross change in bacterial composition (Supplementary Figure 6). The abundance of Lactobacillaceae dominated over other taxa, likely due to consumption of breast milk (Murphy et al., 2017). At PND5, the abundance of Lactobacillaceae tended to decrease, whereas the abundance of other taxa (e.g., Muribaculaceae and Staphylococcaceae) increased. In PND5 females, BPA increased the abundance of Lactobacillaceae and Rikenellaceae in the duodenum, and decreased the abundance of Muribaculaceae, whereas PCB increased the abundance of Enterococcaceae in the jejunum (Supplementary Figure 6). As shown in Supplementary Figure 7, the effect of acute exposure to BPA, BDE-99, and the PCB mixture is intestinal-section-specific in both sexes, likely due to section-specific constitutive levels of the bacteria and the section-specific expression of host receptors.
Interestingly, early life exposure to all three chemicals produced distinct persistent gut dysbiosis (Figure 3B). For example, BPA exposure persistently increased the abundance of Muribaculaceae, whereas early life BDE-99 exposure persistently increased Akkermansiaceae (Figure 3B). Variations in the constitutive abundance of bacteria from various intestinal sections under basal conditions were also observed. For example, the abundance of Prevotellaceae and Rikenellaceae was higher in the colon and feces in both males and females compared with other compartments (Figure 3B).
Interestingly, the abundance of Akkermansia muciniphila, which is known to produce acetate and succinate in monoculture (Chia et al., 2018), tended to be increased in all intestinal compartments and fecal compartments in adult male pups (significant in jejunum and colon) and adult females (significant in colon and feces) following early life BDE-99 exposure (Figure 3C and Supplementary Figure 8). To note, A. muciniphila was below the detection limit in PND 5 pups in all exposure groups. In addition, other bacteria with the capability to produce acetate, succinate, or lactate, such as Lachnospiraceae, and Ruminococcaceae, were persistently up-regulated in colon and feces in adult males following early life BDE-99 exposure (Supplementary Figure 8A). Consistent with the early life BDE-99 mediated persistent increase in A. muciniphila in intestinal sections and fecal compartments in young adults, we observed a trend of persistent increase in acetate, lactate, and succinate uniquely by BDE-99 early life exposure in males (Figure 3D) and females (Supplementary Figure 9A). In PND60 males, the BDE-99 mediated persistent up-regulation of hepatic lactate and succinate were both positively associated with enrichment of A. muciniphila and Ruminoccocaceae UCG-010, whereas hepatic succinate was negatively associated with the enrichment of Ruminococcaceae UCG-005 (Figure 3F). There were no significant correlations between these metabolites and microbial taxa in adult females neonatally exposed to BDE-99.
Microbial-derived secondary BAs were minimally regulated by early life chemical exposure; however, the primary BAs CDCA and its taurine-conjugated form T-CDCA were persistently down-regulated in livers of PND60 males by BDE-99 (Supplementary Figures 9B and 10A). Taurine-conjugated ursodeoxycholic acid (T-UDCA) was persistently up-regulated, and lithocholic acid (LCA) was persistently down-regulated for females neonatally exposed to BPA and the PCB mixture, respectively (Supplementary Figure 10B).
Recently, microbial sequencing strategies in bio-compartments with low microbial abundance have been performed in various disease states (Dong et al., 2019; Mishra et al., 2020; Song et al., 2014; Sookoian et al., 2020). To investigate the potential changes in gut permeability following early life chemical exposure, we performed 16S rDNA sequencing in the serum and liver of pups at PND60 (Supplementary Figs. 8C and 8D). In general, serum 16S rDNA signal was minimally regulated by early life chemical exposure in adult age. However, in livers of PND60 pups, BDE-99 showed the greatest trend in increasing the total assigned reads in males, followed by BPA. However, the PCB mixture showed the greatest trend in increasing the total assigned reads in livers of PND60 females (followed by BPA), suggesting a leaky gut following early life chemical exposure and compensatory uptake in the liver to avoid systemic exposure to microbial constituents.
The large intestine contains one of the largest microbial ecosystems in the human body (Hillman et al., 2017). Therefore, we used the colon as an example to estimate the relationship between the gut and liver. Based on bacterial abundance, we predicted several metabolic potential alterations of the colon microbiome. In PND60 males neonatally exposed to BDE-99, functions related to nucleotide and menaquinol metabolism were predicted to be increased, whereas carbohydrate and SCFA metabolism were predicted to be decreased. In particular, the utilization of succinate and acetate to produce compounds, such as butanoate and methane, were down-regulated, likely due to negative feedback regulations (Supplementary Figure 9C, Supplementary Table 6). Interestingly, persistently up-regulated GO enrichment results in the livers of males neonatally exposed to BDE-99 contained terms related to the microbial influx, such as defense to gram-positive bacterium, response to protozoan, and adhesion of symbiont to host and its process (Supplementary Figure 9D). Females neonatally exposed to BDE-99 followed similar persistent changes but contained a smaller number of predicted colon microbial metabolic potential compared with males and did not show evidence for microbial infiltration signatures in the liver (Supplementary Figure 9E).
Cancer-Prone Transcriptomic and Epigenomic Signature in Adult Male Mouse Livers following Early-Life BDE-99 Exposure
Acetate and succinate are known epigenetic modifiers that were persistently upregulated in adult livers following neonatal exposure to BDE-99 and this was associated with the most profound transcriptomic alterations in males. Furthermore, neonatal chemical exposures can lead to developmental reprogramming of the epigenome (Trevino et al., 2020). Therefore, we hypothesized that BDE-99 exposure in early life persistently reprograms the liver epigenome in young adults. To test this hypothesis, we performed ChIP-seq on three histone modification marks linked to active gene transcription, namely H3K27Ac, H3K4me1, and H3K4me3, in the livers of PND60 male pups that were neonatally exposed to BDE-99. The choice of these epigenetic marks was based on the related functions of the upregulated SFCAs and/or their intermediate products: specifically, acetate can promote histone acetylation via inhibiting histone deacetylation (Qiu et al., 2019), whereas succinate can promote H3K4me1 and H3K4me3 (TeSlaa et al., 2016; Xiao et al., 2012).
On the global scale, no marked differences were observed in distributions of the genomic peak locations in the early life vehicle- and BDE-99-exposed groups (Supplementary Figures 11 and 12), and the number of per chromosome and total peak intervals remained similar for all histone marks and exposure groups (Supplementary Figure 13). The distribution of H3K4me1 was dispersed either upstream or downstream from the transcription start sites (TSS), whereas H3K4me3 and H3K27ac marks were primarily enriched on the TSS (Supplementary Figure 14A). In addition, H3K27ac was also enriched downstream of TSS in the third k-means clustered (C3) region. This observation is consistent with previous findings that the H3K4me1 is enriched in enhancers, H3K4me3 marks the promoters, and H3K27ac is enriched in promoters and enhancers (Heintzman et al., 2007, 2009).
To link the epigenetic marks with transcriptomic changes in adult age following early life BDE-99 exposure, we matched the change in histone modifications of BDE-99 relative to the vehicle control with differentially regulated genes (Supplementary Figure 14B). Overall, most of the differentially regulated genes have no change or no enrichment in any of the three epigenetic marks, suggesting that the differential expression of these genes may be downstream of the initial transcriptional changes of the direct epigenetic mark targeted genes. However, among the direct target genes (defined as the genes with positive associations between differential gene expression and enrichment of the permissive epigenetic marks), H3K4me1 is the most attributable mark that explains 10% of up-regulated genes (with increased H3K4me1) and 17% of down-regulated genes (with decreased H3K4me1). H3K27ac is attributable to 6% of up-regulated genes (with increased H3K27ac) and 12% of down-regulated genes (with decreased H3K27ac), whereas H3K4me3 is attributable to 6% of up-regulated genes (with increased H3K4me3) and 1% of down-regulated genes (with decreased H3K4me3) (Supplementary Figure 14B).
Interestingly, for the liver genes persistently dysregulated by early life BDE-99 exposure that are also direct target genes of at least one of the three epigenetic marks, GO enrichment showed that most of these genes were involved in cancer-related pathways. The GO terms included negative regulation of cell cycle phase transition and positive regulation of cell cycle processes; and regulation of transcription involved in G1/S transition. Most of these biological terms indicate less efficient DNA repair and higher rates of cell proliferation, which are hallmarks of tumorigenesis (Figure 4A, Supplementary Figure 15, left panel, and Supplementary Figure 16A). Examples of BDE-99 mediated persistently up-regulated genes with a persistent increase in at least one of the epigenetic marks include enhancer of zeste homolog 2 (Ezh2) (H3K4me1), cyclin-dependent kinase 1 (Cdk1) (H3K4me3), E2f1 (H3K27ac), and Stathmin (Stmn1) (H3K27ac), all of which are involved in the regulation of neoplasms in the liver (Bisteau et al., 2014; Gan et al., 2010; Huang et al., 2019; Zhang et al., 2020). Genes that were decreased by at least one epigenetic mark by BDE-99 were enriched in basal liver functions, such as lipid and xenobiotic biotransformation including the metabolism of sterols, fatty acids, small molecules, alcohols, and toxins (Figure 4B, Supplementary Figure 15, right panel, and Supplementary Figure 16B). Examples include the transcription factor peroxisome proliferator-activated receptor delta (Ppard) and the nuclear hormone transcription factor Rorc. We also observed a moderate down-regulation of genes involved in detoxification and metal response linked to the reprogramming of liver H3K27ac in females neonatally exposed to BDE-99 (Supplementary Figure 17). Therefore, our evidence suggests that following early-life exposure to BDE-99, the liver transcriptome is reprogrammed toward a highly proliferative state that contains key signatures of neoplasms and has a compromised capacity to metabolize endogenous and exogenous compounds. Together with persistently increased metabolites, i.e., succinate, acetate, lactate, the early life exposure to BDE-99 may lead to an increased risk of liver cancer in adult age.
Developmentally reprogrammed liver epigenome following early-life exposure to BDE-99. Top 20 gene ontology enrichment terms from overlay results (left panel), examples of peak alterations (middle panel), and average peak and gene expression values comparing vehicle control and BDE-99 for up (A) and downregulated (B) genes. Colored bars show histone marks with altered peaks (> 30% in A and < 30% in B) compared with the control for genes that comprise particular GO terms, i.e., purple—H3K4me1, orange—H3K4me3, blue—H3K27ac. Pound signs represent absolute peak fold change > 30%, and asterisks show differential regulation. Please refer to the article in the online version for colored figures.
DISCUSSION
Taken together, as summarized in Figure 5, we have demonstrated that similar to the well-known epigenetic modifier BPA, which is known to produce the adult-onset of liver cancer, early-life exposure to BDE-99 also produced persistent changes in the liver transcriptome toward a cancer-prone signature, associated with persistent gut dysbiosis including up-regulated A. muciniphila in the intestine and its known product cancer-promoting succinate in the liver. In contrast, the PCB mixture had a minimal persistent effect on the gut microbiome and hepatic transcriptome. The early life BPA- and BDE-99-mediated transcriptomic changes were amplified along the developmental trajectory with aggravated transcriptomic dysregulation toward oxidative stress, inflammation, and cancer, with compromised xenobiotic detoxification and intermediary metabolic functions. At the given doses, males were more susceptible than females. These results are among the first to demonstrate that the gut microbiome can also be developmentally reprogrammed and is associated with altered cancer-prone microbial metabolites as well as persistent changes in liver epigenome and transcriptome.
Summary figure of results. Neonatal oral exposure to human health-relevant environmental toxicants BPA, BDE-99, or PCBs (Fox River Mix) was sufficient to acutely and persistently alter the liver transcriptome and changing the ontogenetic trajectory of many key signaling pathways, including cell cycle, epigenetic remodeling, drug metabolism, lipid metabolism, and immune response. In general, the effect of neonatal chemical exposure is markedly amplified in adult age: at the given dose, BDE-99 had the most prominent hepatic response, followed by BPA, then the PCB mixture, for which males were more susceptible than females. For BDE-99, the persistent transcriptomic changes revealed downregulated signatures of endogenous liver functions, i.e., xenobiotic and lipid metabolism, and upregulated cellular proliferation signatures. These changes are possible due to developmental reprogramming through upregulated liver metabolites involved in the epigenetic modification and inflammatory signaling, i.e., acetate, lactate, succinate. Interestingly, neonatal exposure to BDE-99 persistently upregulated gut microbiota capable of producing these metabolites, suggesting a persistently dysregulated cross-talk between the gut microbiome. The persistent effect due to neonatal exposure may be a point source for altered hepatic biotransformation leading to increased risk for chronic disease. Please refer to the article in the online version for colored figures.
Akkermansia muciniphila is known to be an anti-inflammatory and anti-obesity bacterium (Dao et al., 2016; Schneeberger et al., 2015). However, recent studies have shown a positive association between A. muciniphila and hepatocellular carcinoma with cirrhosis in patients (Lapidot et al., 2020). A. muciniphila is also known to induce pro-inflammatory responses in vitro and in vivo (Cekanaviciute et al., 2017; Jangi et al., 2016). Therefore, the effect of A. muciniphila appears to be context specific. It is well-established that A. muciniphila monoculture in vitro produces succinate, acetate, 1,2-propanediol as the major metabolites from pure mucin degradation (Chia et al., 2018). In the present study, both acetate and succinate have a persistently increasing trend associated with up-regulated A. muciniphila following early life BDE-99 exposure. Notably, we did not include 1,2-propanediol as an analyte in our GC-MS panel as it is not a SCFA. Therefore, although at this stage we cannot rule out the possibility that succinate may come from other sources, the observation that both metabolites that are known to be produced by this microbe are persistently increased strongly suggests that A. muciniphila may be one of the contributing factors. Future studies using isotopic tracing and germ free or antibiotic-treated mice are needed to further elucidate the source of production of succinate.
In the present study, we have demonstrated that neonatal exposure to BPA persistently dysregulates the gut microbiome. Our results align with previous findings, which showed that perinatal exposure to BPA resulted in persistent dysbiosis in rabbits and mice (Malaise et al., 2017; Reddivari et al., 2017). We have also recently demonstrated that in utero and lactational exposure to other environmental toxicants, including another diet and breast milk enriched PBDE congener BDE-47, the currently used PBDE-alternative tetrabromobisphenol A (TBBPA), and the BPA-alternative bisphenol S (BPS), also lead to persistent gut dysbiosis in adult male mice, corresponding to increased fecal BA output (Gomez et al., 2021). The present study showed that acetate, succinate, and lactate were the key persistently increased metabolites, whereas BAs were in general minimally impacted. Such discrepancy between Gomez et al. and the present study may be due to different types of toxicants, different exposure windows/routes, and different strains of mice. Even though the specific microbial metabolites or taxa were not regulated in the same manner between the present study and those described in the literature, our data show that the gut microbiome can be reprogrammed following early life toxic exposures in both studies. Given the importance of the gut microbiome in drug metabolism and nutrient homeostasis, these studies suggest that the gut microbiome as a novel target of toxicant exposures needs to be incorporated into the mechanistic investigations of DOHaD.
In the gut microbiome, notably, there was a persistent up-regulation of A. muciniphila, Lachnospiraceae UCG-004, and Ruminococcaceae UCG-010. Lachnospiraceae. Importantly, certain Ruminococcaceae are involved in lactate formation (Shkoporov et al., 2016; Tao et al., 2017; Wang et al., 2012) and A. muciniphila is considered as a succinate producer (Chia et al., 2018). The introduction of A. muciniphila has been previously shown to modulate the epigenetic regulation of host cells in vitro (Lukovac et al., 2014) and alter the gene expression of host cells in germ-free mice (Derrien et al., 2011). Ruminococcaceae UCG-005, which was persistently downregulated, is a marker of a healthy gut (Mancabelli et al., 2017) and was inversely correlated with the succinate in males neonatally exposed to BDE-99. This suggests that neonatal exposure to BDE-99 can decrease normal commensal bacteria, which provides an altered gut environment for lactate, succinate, and acetate producers to bloom. Furthermore, previous findings showed that acetate can reprogram host cells and tissues (Astakhova et al., 2016; Bachem et al., 2019; Schulthess et al., 2019). Succinate, which is a widely accepted inflammatory signal, is involved in histone modification and is closely linked to tumorigenesis (Xiao et al., 2012; Zhao et al., 2017). Combining these findings, our results establish for the first time a novel link between epigenetic modifying microbes and microbial metabolites and the liver epigenome and transcriptome toward a proinflammatory and pro-cancerous stage.
The differential effect of BPA, BDE-99, and the PCB mixture may be due to expression changes in drug metabolism-related genes. This in turn involves differential activation of the corresponding upstream transcription factors (Figure 2 and Supplementary Tables 3 and 4). Overall, our results are in line with previous studies that investigated the developmental origins of health and diseases (DOHaD), in particular, the reprogramming of the liver epigenome following early-life exposure to environmental toxicants. Trevino et al. (2020) found that early-life exposure to BPA reprograms the liver epigenome to transform the liver to nonalcoholic fatty liver disease (NAFLD)-like state in rats (Trevino et al., 2020), preferentially in males. Khalil et al. (2018) showed that neonatal BDE-47 exposure in mice had a persistent effect in adulthood by altering the blood-liver balance of lipids (Khalil et al., 2018). The rat liver epigenome was reprogrammed and associated with metabolic dysregulation signatures following perinatal exposure to BDE-47 (Suvorov et al., 2020), and a higher incidence of liver cancer was found in adults from perinatal exposure to BPA (Weinhouse et al., 2014, 2015). Our results showed that early life exposure to BDE-99 in males leads to a reprogrammed epigenetic landscape of H3K27ac, H3K4me1, and H3K4me3. The enrichment of one of these permissive marks is positively associated with persistently up-regulated genes involved in liver tumors. In contrast, the loss of one of these permissive marks is negatively associated with persistently down-regulated genes involved in lipid metabolism and xenobiotic-sensing transcription factors. However, the persistent up-regulation of immune response features was not related to the reprogramming of the liver epigenome, suggesting that other factors may be involved, such as other microbial constituents that directly trigger pro-inflammatory response without the need to alter gene transcription.
We observed that the mRNA levels of Tjp1 and Tjp2 were not statistically significantly different from the control, and the mRNA levels of Tjp1 tended to be increased in males neonatally exposed to BPA or BDE-99 (Supplementary Figs. 8E and 8F). However, we also observed that neonatal exposure to BDE-99 contained up-regulated signatures related to the increased influx of microbial components (Supplementary Figs. 9C–F), and the corresponding assigned microbial taxa tended to be increased in adult livers that were neonatally exposed to BPA or BDE-99. Therefore, the tendency of increase of Tjp1 and Tjp2 mRNA levels may be indicative of a compensatory mechanism of the large intestine to re-arrive at a homeostatic state from the persistent damage from early life exposures to the environmental toxicants. To further validate our findings, future investigations should involve direct histological measurements of the large intestine for gut permeability and methods including percent sucralose excretion. To establish the role of the gut microbiome in the developmental reprogramming of the liver, germ-free and antibiotic-mediated microbiome depletion methods can be used. Furthermore, the role of specific bacteria in modifying the host epigenome can be identified by tracing the long-term behavior of radioactively labeled carbohydrates.
Although the external chemical exposure stopped in the neonatal period, it is possible the internal exposure to chemicals still occurred due to the persistent nature of these chemicals. Even though we did not quantify the remnant chemicals in PND60, it is known that the estimated half-life of BDE-99 is 6 days in rodents (Haak et al., 2002). Thus, the remaining amount 2-months later, which is over 9 half-lives postdose, is expected to be minimal. In addition, due to the lipophilic nature of BDE-99 and PCBs, most of the remnant chemicals are expected to be stored in adipose tissue. Thus, the direct chemical effect on the gut microbiome and hepatic transcriptome is also expected to be minimal at PND60. Therefore, the persistent responses are likely due to the initial toxicity at early life-induced persistent gut dysbiosis and epigenetic reprogramming of the host liver.
Sex differences occur in a wide array of toxicological responses including behavior, anatomy, physiology, biochemistry, and genetics, accounting for the male to female differences in response to environmental chemical exposures, diet, and pharmaceuticals (Kim et al., 2018; Vyas et al., 2019). From early life exposure to BPA or BDE-99, we show sex differences of developmental reprogramming of the liver and the gut microbiome. Sex differences in disease and developmental reprogramming outcomes have been noted in past reports (Deane et al., 2001; Kundakovic et al., 2013; Trevino et al., 2020). The sex differences may be due to dissimilarities in the endocrine system, such as the normal average production of androgen and estrogen, and the activities of their receptors.
The present study has certain limitations, including doses that are higher than estimated human exposures and small sample sizes (n = 3 per exposure group) for liver transcriptomics. The dose of BPA at 50 mg/kg body weight in laboratory mice resulted in a serum concentration that is close to humans unknowingly exposed to BPA (Sieli et al., 2011). The chronic oral reference dose, 50 ug/kg, defined to not result in adverse health effects from chronic exposure to BPA, was used to investigate the developmental reprogramming of the liver in rats (Trevino et al., 2020). Although the dose used in this study is higher than the chronic oral reference dose, it is within the range of unintentional exposure to BPA in humans, as well as other investigations that have used BPA as a chemical that can exert developmental reprogramming (Dolinoy et al., 2007; Kaur et al., 2020; Malloy et al., 2019; Weinhouse et al., 2014). The concentration of BDE-99 in serum in US adults is between 1.3 and 290 pg/g (Makey et al., 2016). However, the amount of BDE-99 exposed to human newborns is higher than that of adults, as newborns are vulnerable to being exposed to BDE-99 through breast milk (approximately 5.7 ng/g liquid weight) (Frederiksen et al., 2009; Lorber, 2008). We then extrapolated this dose by a factor of 10 for the comparison across different species (Li et al., 2017, 2018; Scoville et al., 2019). The dose used for BDE-99 is similar to doses that have been used to investigate hepatic metabolism in mice (Li et al., 2017, 2018; Pacyniak et al., 2007; Scoville et al., 2019). The PCB mixture dose used in the present study is less than the dose of another study that used a PCB mixture found in food, i.e., PCB-138, PCB-153, and PCB-180 (150 μmol/kg or 43 mg/kg), which results in a PCB plasma level of 5 μM (Choi et al., 2010) and is also comparable with PCB plasma levels in an acutely exposed human population (Jensen, 1989; Wassermann et al., 1979). Because the molecular targets downstream of the three chemicals are different, we did not use the same dose for all three chemicals. Importantly, the doses of the chemicals used in this study were used in other published studies in the published literature (Cheng et al., 2018; Lim et al., 2020; Scoville et al., 2019; Trevino et al., 2020; Weinhouse et al., 2014), which allows for comparisons of toxicological outcomes between the present study and those studies. Regarding sample size, even in triplicates, we show that early-life exposure to toxicants can alter the developmental trajectory and have persistent responses, supporting the hypothesis of our study.
Despite the limitations, our study has provided the first evidence of how early-life exposure to the environmental toxicants BPA, BDE-99, and PCBs can alter the developmental trajectory of the liver and established a novel association between gut dysbiosis and host liver transcriptome and epigenome, which subsequently modulate a cancer-prone hepatic transcriptomic signature in adulthood. As many of the dysregulated gene functions and machineries are evolutionarily preserved among species, our findings provide novel insights that early life exposure to toxicants can have a life-long impact on disease risk, which may at least partly be regulated through the gut microbiome.
SUPPLEMENTARY DATA
Supplementary data are available at Toxicological Sciences online.
DECLARATION OF CONFLICTING INTERESTS
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
DATA AVAILABILITY
Raw data for RNA-seq, 16s rDNA-seq, and ChIP-seq are uploaded to Dryad (https://doi.org/10.5061/dryad.gf1vhhmpz).
Dryad Digital Repository DOI: https://doi.org/10.5061/dryad.gf1vhhmpz
ACKNOWLEDGMENTS
The authors would like to thank Drs Cristian Coarfa, Sandra Grimm, and Matthew Robertson for training Dr Julia Yue Cui on bioinformatics as part of the NIEHS Center Assistant Professor Externship (3P30ES007033-24S1).
FUNDING
This work was supported by the National Institutes of Health (NIH) (grants R01ES025708, R01ES030197, R01GM111381, and R01ES031098), the University of Washington Center for Exposures, Diseases, Genomics, and Environment (P30ES007033), Environmental Pathology/Toxicology Training Program (T32ES007032), Baylor College of Medicine Gulf Coast Center for Precision Environmental Health (P30ES030285), and the University of Washington Sheldon Murphy Endowment.
REFERENCES
R Development Core Team. (
Comments