Prenatal bisphenol A (BPA) exposure alters the transcriptome of the neonate rat amygdala in a sex-specific manner: a CLARITY-BPA consortium study.

Animal Care

Study animals were housed in an Association for Assessment and Accreditation of Laboratory Animal Care (AALAC) accredited facility. All procedures were approved in advance by the National Center for Toxicological Research Institutional Animal Care and Use Committee (NCTR-IACUC). PND 1 pups were obtained from litters produced for the CLARITY-BPA program [ 35 , 37 ]. Methods for animal husbandry, diet, breeding, dose preparation and administration, and necropsy are described in detail elsewhere [ 34 ]; therefore, only relevant methods are reviewed below. Sprague-Dawley rats from the NCTR colony (NCTR-SD strain code 23) were housed in solid-bottomed polysulfone caging with hardwood chip bedding at 23 ± 3°C with a relative humidity level of 50 ± 20% on a 12:12h light/dark cycle (0600–1800). Food (soy- and alfalfa-free diet verified casein diet 10 IF 5K96; Cat. 1810069; Purina Mills, Richmond, IN) and Millipore-filtered water in glass water bottles with silicone stoppers (#7721 clear, The Plasticoid Co., Elkton, MD) were provided for ad libitum consumption. Extracts of each diet lot were analyzed for BPA and myco/phytoestrogens (genistein, daidzein, zearalenone, and coumestrol) by liquid chromatography and mass spectrometry [ 42 ] and all had levels below the average analytical method blanks [ 34 ]. Drinking water, polysulfone cage leachates, and bedding extracts were also found to have BPA levels below the level of the average analytical method blanks [ 34 ]. Reagents and Dosing The BPA (CAS # 80-05-7, catalog # B0494, TCI America, Portland, OR) and EE 2 (CAS # 57-63-6, catalog #E4876, Sigma-Aldrich, St. Louis, MO) were more than 99% pure and administered in 0.3% aqueous carboxymethyl cellulose (CMC; catalog # C5013, Sigma- Aldrich, St. Louis, MO). The EE 2 groups were included to serve as the “reference estrogen” and to determine if BPA-related effects were consistent with an estrogenic mode of action. Two weeks before mating, dams were randomized to one of eight exposure groups stratified by body weight to produce approximately equal mean body weights in each group. Sires were randomly assigned subject to the constraint that no sibling or first cousin mating was permitted, as previously described [ 42 ]. Mating was confirmed by the presence of a sperm plug or sperm-positive vaginal cytology [defined as GD 0]. To model the exposure route used to establish the NOAEL, dams were gavaged daily with vehicle (0.3% CMC/kg bw/day), BPA (2.5, 25, 250, 2500, or 25000 μg BPA/kg bw/day), or EE 2 (0.05 or 0.5 μg EE 2 /kg bw/day) from GD6 until the day of parturition [postnatal day (PND) 0]. Dams and pups were left undisturbed on PND0. On PND 1, pups (one per sex per litter) were weighed and euthanized by rapid decapitation. Heads were collected, snap frozen, and shipped coded (blinded) to the Patisaul lab where they were stored at −80°C until processing.

Tissue Collection and Preparation

Each whole head was coronally cryosectioned (Leica CM1900, Nussloch, Germany) from the caudal end until the caudal borders of the amygdala were identified. Two sequential bilateral punches, each 1.00 mm in diameter and 1.00 mm in depth, were collected caudally to rostrally; this corresponded with plates 69–75 of the Atlas of the Developing Mouse Brain [ 43 ] ( Supplemental Fig. 1 ). All four punches, which collectively comprised the entire amygdala, were combined and stored in BPA-free Eppendorf tubes at −80°C. These punches were collected at the same time we collected hypothalamic and hippocampal studies for a prior, published study [ 12 ] with the intention of performing the amygdalar assessment as a follow-up (secondary analysis) if any significant observations were found in the other brain regions. The outcomes of that prior study informed the selection of the dose groups and primary genes of interest for the present study.

Quantitative real-time PCR

Analysis was performed on eight exposure groups (n = 5–7 for the predetermined genes, n = 3–7 for the validation genes; sample size based on availability of cDNA): vehicle, BPA 2.5, 25, 250, 2500, and 25000 and EE 2 0.05 and 0.5. Total RNA was extracted with the Qiagen RNEasy Miniprep kit. An Agilent 2100 Bioanalyzer with an RNA 6000 Nano Chip was used to determine RNA purity and concentration and each sample had a RIN of 10. Single-stranded cDNA synthesis was performed with 350 ng of RNA input using the high capacity RNA-to-cDNA kit (Applied Biosystems, Cat. 4387406) and samples were stored at −20°C until use. qRT-PCR was performed as previously published [ 12 ] using a TaqMan probe-based protocol and detected on a StepOnePlus™ Real-Time PCR System (Applied Biosystems, Life Technologies, Grand Island, NY) with the following cycling parameters: 50°C for 2 min, 95°C for 10 min, followed by 40 cycles of 95°C for 15 sec and 60°C for 1 min. Each sample was run in triplicates the sample maximization approach was followed to avoid technical and run-to-run variation [ 44 ]. Cycle threshold (Ct) values for the gene of interest were normalized to the Ct for 18s rRNA and relative data were determined by the Livak ΔΔ cycle threshold (ΔΔ-Ct) method [ 45 ].

RNAseq Data Analysis

The experimental design for transcriptome sequencing was developed in consultation with the NCSU Genomic Sciences Laboratory (GSL). Transcriptome sequencing was performed by the GSL on 24 amygdala samples (n = 3 per sex per group). Four experimental groups were examined: vehicle, 25 BPA, 250 BPA, and 0.5EE 2 . RNA extraction was performed with the Qiagen RNEasy Miniprep kit according to the manufacturer protocol (Qiagen, Cat. 74134). Total RNA samples were submitted to the North Carolina State Genomic Sciences Laboratory for Illumina RNA library construction and sequencing. Prior to library construction, RNA integrity, purity, and concentration were assessed using an Agilent 2100 Bioanalyzer with an RNA 6000 Nano Chip (Agilent Technologies, USA). All samples had an RNA integrity number (RIN) of 10. To optimize library complexity, only samples that had greater than 300 ng of total RNA were used as input material for library preparation. As previously described [ 12 ], messenger RNA (mRNA) was purified using the oligo-dT beads provided in the NEBNext Poly (A) mRNA Magnetic Isolation Module (New England Biolabs, Cat. E7490). Complementary DNA (cDNA) libraries for Illumina sequencing were prepared with the NEBNext Ultra Directional RNA Library Prep Kit and the NEBNext Mulitplex Oligos (New England Biolabs, Cat. E7420 and E7335). Briefly, mRNA was isolated, heat fragmented, and primed with random oligos for first strand cDNA synthesis. Second strand cDNA synthesis was performed with dUTPs to preserve strand orientation information. Next, the double-stranded cDNA fragments were purified using AMPure XP beads (Beckman Coulter Genomics, Cat. A63881), end-repaired, and ligated onto adaptors specifically designed for the Illumina platform. Following ligation, the samples were size-selected to a final library size of 400–550 bp (adapters included) using sequential AMPure XP bead isolation. Protocol-specified PCR amplification was performed to enrich adaptor-ligated cDNA and add specific indexes for each sample. The amplified library fragments were purified and quality and final concentration was assessed using an Agilent 2200 Tapestation. The final quantified cDNA libraries were pooled into equimolar amounts for clustering and sequencing on an Illumina HiSeq 2500 DNA sequencer (4 lanes), utilizing a 125 bp single end sequencing reagent kit (Illumina, USA). Approximately 37.5 million reads were generated per sample. The software package Real Time Analysis (RTA), was used to generate raw bcl, or base call files, which were then de-multiplexed by sample into fastq files for data submission. Data analysis for RNAseq was performed in consultation with the Bioinformatics Core of the NCSU Center for Human Health and Environment. Sequence data was evaluated with FastQC and 12 poor quality bases were trimmed from the 5′-end. The good quality reads were aligned to the Rattus norvegicus (rn6) reference genome (downloaded from UCSC) using the STAR software package [ 46 ]. For each replicate, per-gene counts of uniquely mapped reads were calculated using htseq-count script from the HTSeq python package [ 47 ]. Count data were normalized for sequencing depth and distortion, and dispersion was estimated using the DESeq2 Bioconductor [ 48 , 49 ] package in the R statistical computing environment. We fit a leaner model using treatment levels and differentially expressed genes were identified after applying multiple testing corrections using the Benjamini-Hochberg procedure (padj