Does consumption of a high-fructose diet during pregnancy and lactation exacerbate the effects of maternal exposure to cadmium on development and metabolic function of mouse offspring?

Abstract

Exposures to pollutants rarely occur in isolation, and often exist concurrently with other potentially harmful circumstances such as dietary stressors (e.g., over- and/or under-nutrition). The impact of such combined exposures may be particularly insidious if it occurs during early life-stages. This study evaluated the effects of maternal exposure to cadmium (Cd) and consumption of a high-fructose diet (HFrD) on development of the offspring. Female CD-1 mice were given either 0.5 or 5 ppm Cd in drinking water with or without a ~60% fructose diet for 3 weeks before mating; controls received deionized water and calorie-matched diet. Dams were maintained on the same exposures until postnatal day (PND) 16. Cd concentrations in maternal, fetal, and neonatal liver increased in a dose-dependent manner; HFrD did not alter the hepatic bioaccumulation of this metal. Approximately 55% of Cd was cleared from livers of offspring by 6 months of age. Five endpoints known to be associated with Cd or HFrD adverse effects were evaluated longitudinally in the offspring from birth to young adulthood, which included growth trajectory, pubertal development, body composition, glycemic tolerance, and hepatic lipid accumulation. Maternal exposure to either Cd or HFrD alone significantly advanced onset of puberty, led to hypoglycemia and reduced adiposity in adulthood. The dietary stressor rarely exacerbated the metal effects in most of the endpoints examined, with incidental age-specific exceptions in adiposity. In fact, the responses elicited by HFrD appeared to alter those produced by Cd in some cases (such as pubertal development in females). Because of the long-lasting effects of Cd and HFrD on metabolic function (e.g., glucose tolerance), transcriptomics and gene methylation analyses were performed on livers from neonatal and adult offspring, and results were largely consistent with the phenotypic findings. In summary, maternal exposure to Cd or HFrD alone drove perturbations in growth and development, producing long-lasting changes in metabolic function in adult offspring. High fructose diet did not appear to significantly exaggerate the adverse outcomes from chemical exposure in the endpoints examined herein.

Introduction

Exposures to environmental pollutants, particularly ones that are ingested, do not occur in isolation. Consumption of such environmental toxicants, whether it be from contamination of the water supply or through food packaging, often coexist with nutrient stressors that spanning from an excess of specific macronutrients (e.g., added sugars, saturated and/or trans fats, total calories) to common micronutrient deficiencies (Vitamin D, Zinc, Iron) (Miller and Rayalam, 2017). Subsequently, such dietary or specific nutrient stressors may serve as factors that can modify the deleterious effects of pollutants. In general, studies addressing the interactions between chemical toxicants and dietary manipulations have rarely been characterized, especially for exposures during development (Callahan and Sexton, 2007; Linder and Sexton, 2011; Moretto et al., 2017; US EPA 2003; Vesterinen et al., 2017). Notably, maternal high fat-diet has been shown to exacerbate cadmium induced fatty liver (Young et al., 2022) and to enhance susceptibility to metabolic alterations induced by ozone exposure (Snow et al., 2020; Rouschop et al., 2021). Dietary manipulations have been shown to act as non-chemical stressors that modify toxicant responses when exposed concurrently, such as: maternal lead exposure and prenatal stress (Varma et al., 2017; Sobolewski et al., 2018; 2020); manganese exposure and maternal stress (Oshiro et al., 2022); and exposure to phenols, parabens or phthalates and prenatal stress (Aker et al., 2020; Ferguson et al., 2019).

Here, we examined outcomes of mouse offspring after developmental (maternal) exposure to cadmium (Cd) in drinking water and a high-fructose diet (HFrD). Cadmium is a toxic heavy metal widely found in the environment, with human exposures commonly occurring via cigarette smoke or industrial emissions, or through consumption of contaminated food and water (Satarug and Moore, 2004); its adverse health effects in adults have been well characterized (Genchi et al., 2020; Schaefer et al., 2022; Tinkov et al., 2017; 2023). In addition, several recent reviews have summarized the current understanding of cadmium developmental toxicity in laboratory animals and humans (Young and Cai, 2020; Saedi et al., 2023; Lawless et al., 2023; Chandravanshi et al., 2021). Fructose is a natural sugar found in fruits, but commercial production of crystalline fructose and high-fructose corn syrup have promoted its use in processed food and beverages. This has greatly increased its dietary consumption in the general population. The rise in fructose intake has raised considerable public health concern (Rizkalla 2010; Dornas et al., 2015) because it appears to parallel the obesity epidemic, and the increased incidence of metabolic syndrome and non-alcoholic fatty liver disease (Taskinen et al., 2019; ter Horst and Serlie, 2019). The adverse impacts of high fructose diet on maternal health and development of the offspring have also been described (Thompson and DeBosch, 2021).

We chose to examine the developmental exposure to cadmium and HFrD because they share several common features of adverse health effects. These include growth impairment (Wang et al., 2016; Kozlosky et al., 2023; Asghar et al., 2016; Liu et al., 2021; Bo et al., 2021) and developmental alterations (such as onset of puberty, Parodi et al., 2017; Hernandez-Rodriguez et al., 2021; Li et al., 2022; Mueller et al., 2015), induction of metabolic syndrome that involves obesity (Ba et al., 2017; Jackson et al., 2020; Saad et al., 2016; Jurgens et al., 2005), diabetes (Saedi et al., 2023; Jackson et al., 2020; Yi et al., 2021; Jung et al., 2022), and hepatosteatosis (Pillai et al., 2009; Riegl et al., 2023; Jensen et al., 2018; Koo et al., 2021). Accordingly, we examined the effects of Cd and HFrD individually or in combination on these five endpoints. We focused on low doses of cadmium relevant to human exposure (Jackson et al., 2020; Young et al., 2022), with said exposures limited to the prenatal and lactational window. We thus examine the "Developmental Origins of Health and Disease" concept, that postulates early-life insults (nutrition, stress, chemicals) can predispose an organism to adverse health outcomes and diseases later in life, possibly involving alterations of developmental programming through epigenetic mechanisms (Heindel et al., 2017). Hence, we evaluated DNA methylation profiles of mouse offspring livers to compare changes between early-life and young adult stages; we also ascertained whether changes in the epigenome were connected to the expression of specific genes identified by transcriptomic evaluation.

Methods

Materials.

Cadmium Chloride (CdCl₂) was purchased from Sigma Chemical Co. (St. Louis, MO). 50% Dextrose solution stock was purchased from Covetrus North America (North Dublin, OH). Calorically matched AIN-93G and 59% Fructose diets were purchased from Research Diets (New Brunswick, NJ). Diet information can be found in Suppl Table 1.

Animal treatment.

All procedures involving the use of laboratory animals were conducted in accordance with the guidelines set forth by the U.S. EPA Office of Research and Development, Center for Public Health and Environmental Assessment Institutional Animal Care and Use Committee. Animal care procedures and facilities (AAALAC accredited) were consistent with the recommendations provided by the 2011 National Research Council’s “Guide for the Care and Use of Laboratory Animals”, the Animal Welfare Act, and the Public Health Service Policy on the Humane Care and Use of Laboratory Animals. Male and female CD-1 mice (8-10 weeks of age, 20-35 g) were purchased from Charles River Laboratory (Raleigh, NC). Upon arrival, animals were housed individually in polycarbonate, solid bottom cages. Lab Diet 5001 pellet chow (PMI Nutrition International, Brentwood, MO) and tap water were provided ad libitum. Environmental controls for the animal facility were set to maintain a mean temperature of 20-24°C, relative humidity of 40-60%, and a 12-h light-dark cycle (lights off 7PM – 7AM).

All animals were allowed 4 days to acclimate to their environment prior to the beginning of the study. At the end of acclimation, female mice were switched to either a control (AIN-93G) or a 59% high-fructose diet (HFrD) and given deionized pico-pure water (Milli-Q system, Millipore Corp, Burlington, MA) only (control), or pure water containing either 0.5 ppm or 5 ppm of Cd for 3 weeks prior to mating. Hence, there were a total of six treatment groups: (i) Control Diet and pure drinking water (ConD); (ii) Control diet and 0.5 ppm Cd in drinking water (ConD+0.5Cd); (iii) Control diet and 5 ppm Cd in drinking water (ConD+5Cd); (iv) High-Fructose Diet and pure drinking water (HFrD); (v) High-fructose Diet and 0.5 ppm Cd in drinking water (HFrD+0.5Cd); and (vi) High-fructose Diet and 5 ppm Cd in drinking water (HFrD+5Cd). Males were housed separately and maintained on the Lab 5001 diet and tap water. One female and one male mouse were coinhabited in each cage overnight; females were checked for the presence of a vaginal plug in the following morning, returned to her home cage and maintained on her exposure during the gestation and lactation periods until postnatal day (PND) 16 to limit direct ingestion of Cd or fructose in the offspring. After PND 16, both dams and pups of all treatment groups received pico-pure water and the AIN-93G diet. Pups were weaned on PND21 into sex specific housing conditions. Dams were euthanized using CO₂ and decapitation10 days after weaning; blood, livers and kidneys were collected for determination of metal concentrations; an additional subset of livers were collected to measure triglyceride levels and for transcriptomic and gene methylation analyses.

Evaluation of dams, and growth and development of the offspring.

Pregnant mice were weighed after mating and at term (GD18) to determine maternal weight gain. Newborn mice were weighed on PND1. The sex of each pup was determined on PND3, and each litter was standardized to 4 males and 4 females to maintain an appropriate nutritional status. Neonatal body weight was monitored until weaning. After weaning, the body weight of male and female pups was measured weekly until PND75, and again, at PND150.

Determination of developmental landmarks.

A single investigator was assigned to monitor eye-opening and onset of puberty of mouse offspring in all treatment groups to ensure consistency of animal handling and observation. Beginning on PND10, pups in each litter were examined daily in the morning to determine if both eyes were fully opened. The day of eye-opening and body weight of the pup within each litter were averaged to derive a mean value, as litter was used as statistical unit. The examination continued until the eyes of all pups were opened. Beginning on PND24, female and male pups were identified by ear marks and evaluated daily to determine the onset of puberty by vaginal opening or preputial separation. The age of vaginal opening or preputial separation, and weight of each pup were tabulated accordingly.

Intraperitoneal glucose tolerance test.

Mice were fasted for 6 h (Andrikopoulos et al., 2008) beginning at 7AM, weighed and fasting blood glucose was determined with a handheld glucometer (Contour^® next EZ Blood Glucose Monitoring System, Ascensia Diabetes Care, Parsippany, NJ). Blood collected from the tail vein (~2 μL) was applied directly to a glucose strip. Each mouse was then given 20% dextrose (10 μL/g body weight) by intraperitoneal injection and blood glucose was measured at 15, 30, 60, 90, and 120 min after injection on PND 47, 97 and 137.

Determination of body composition of adult mouse offspring.

The body composition of male and female mouse offspring during adolescence (PND 38 and 90) and adulthood (PND 114 and 150) were measured noninvasively by the Bruker LF90 II “minispec” body composition analyzer equipped with a nuclear magnetic resonance system (Bruker Optics, Inc, Billerica, MA), as described previously (Gordon et al., 2016). In brief, on each day of measurement, a quality control (QC) sample of safflower seeds of a set weight was tested to ensure that the instrument provided consistent readings between runs. Each mouse was weighed and placed in a clear, plastic ventilated cylinder (with a diameter of 5 cm), which closed at each end. The cylinder was then inserted into the bore of the instrument for approximately 1 min to complete the scan. Measurements of %fat, %lean, and %fluid were normalized to body weight and recorded for each animal.

Determination of hepatic triglycerides.

Liver triglycerides concentrations were measured with an assay kit purchased from Zenbio (Research Triangle Park, NC, cat# TG-1-NC). Frozen liver samples were thawed, weighed, and homogenized (polytron) in 1:10 volume of distilled, pico-pure deionized water. Each sample was then further diluted from 1:50 to 1:400 with wash buffer to ensure that the triglyceride values would fall within the glycerol standard curve performed concurrently with each experiment. Triglyceride levels were determined according to the manufacturer’s instructions and optical density at 540 nm was measured by a microtiter plate reader (BMG CLARIOstar, BMG Labtech, Cary, NC). Triglyceride values represent the mean of two technical replicates for each sample.

Liver sectioning, homogenization, and nucleic acid isolation.

Approximately 15-50 mg of frozen mouse livers were sectioned on dry ice and immediately placed into a 2 ml vial with ceramic homogenization beads (MP Biomedicals, Lysing Matrix D) containing 360 μl of Qiagen buffer ATL (for DNA isolation; Qiagen GmbH, Hilden, Germany) or 600 μL RNAzolRT (for RNA isolation; Qiagen). Samples were stored at 4 °C until tissue homogenization and nucleic acid isolation. The tissue was allowed to equilibrate to room temperature, 20 μl of 40 mg/ml Proteinase K was added and homogenized with a Precellys 24 homogenizer (Bertin Technologies, Villeurbanne, France) at 5500 rpm for 20 s. For DNA isolations, the homogenized sample tubes were placed in a 56°C thermomixer at gentle agitation for 3 h. After proteinase treatment, the liver homogenates were treated with RNAse A, and DNA was isolated using Qiagen’s DNeasy^® kit following the manufacturer’s instruction. Yield was determined by Qubit broad range dsDNA assay kit and protocol (Life Technologies, Carlsbad, CA), which averaged ~190 ng/μl after isolation. The isolated DNA was checked for initial purity (A260/280 of ≥ 1.8, A260/230 ≥ 1.0) using a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, Delaware). DNA was stored at −20°C until assayed by Infinium Mouse Methylation BeadChip (MM285; Illumina, Inc., San Diego, CA).

For RNA isolations, homogenates were centrifuged at 2000xg for 4 min. The supernatants were then transferred to a fresh 1.5mL tube and 100 μL of Ambion Nuclease-free water (Cat #AM9938) was added to every 500 μL of RNAzolRT supernatant transferred. Samples were then vortexed briefly and incubated at room temperature for 10 min. Following the room temperature incubation, samples were centrifuged at 12,000xg for 15 min. The supernatant was then transferred to a fresh 1.5 mL tube and 6.5 μL of 4-bromoanisole (BAN reagent; MRC, Cincinnati, OH) was added to the supernatant. Samples were vortexed vigorously and incubated for 5 min at room temperature. Samples were then centrifuged at 12,000xg for 10 min. The supernatant was transferred to a fresh 1.5 mL tube. The equivalent volume of 100% isopropanol was then added to the supernatant, vortexed, and incubated at 4°C for 15 min. Samples were then centrifuged at 16,000xg for 10 min. The supernatant was then discarded, leaving the RNA pellet. The RNA pellet was then washed by adding 500 μL of 75% ethanol and then centrifuged at 16,000xg for 2 min. This was repeated for two total wash steps. The remaining ethanol was removed from the RNA pellet. 100 μL of nuclease-free water was added to the pellet and mixed to resuspend the pellet. The RNA pellets were then incubated at room temperature for 5 min for complete dissolution. 350 μl Buffer RLT (Qiagen) was added to each resuspended pellet. RNA was then purified using Qiagen’s RNeasy Mini Kit, following manufacturer’s instructions. RNA yield and purity was assessed using the NanoDrop spectrophotometer. The isolated RNA was checked for initial purity (A260/280 of ≥ 1.9, A260/230 ≥ 1.0) and quantified using a NanoDrop ND-1000 spectrophotometer. RNA samples were stored at −80 °C until further use.

Targeted RNA sequencing and analysis.

Using targeted Templated Oligo-Sequencing (TempO-Seq; BioSpyder, Inc., Carlsbad, CA), whole transcriptome measurements on liver RNA were derived from male and female offspring at PND3, 38, 88, and 150. Library preparation and sequencing were performed at BioSpyder using previously described procedures (Harrill et al. 2021). Library preparation was performed using the TempO-Seq Mouse Whole Transcriptome Assay (v1.1 rev B; 30146 probes, 21398 genes). After passing quality control checks for total mapped reads in positive control samples (>1.5M mapped reads, we observed 8.2M reads per sample), signal-to-noise ratio of the number of mapped reads in positive controls versus negative controls (>20:1, we observed 51:1), and percentage of mapped reads in positive controls (>70%, we observed 95.3%), sequence data were aligned and matched to the probed gene for the assay. Gene count data were provided to EPA for further analysis. Raw count data are available on http://data.gov.

Count data for each sample were further examined using quality control metrics that examined the number of probes capturing the top 80% of the signal (flagged samples < 1000 probes meeting criteria), the number of probes with at least five reads (flagged samples < 5000 probes meeting criteria), and the proportion of reads captured by the top 10 probes (flagged samples with 20% or more reads meeting criteria). Two samples contained less than 5000 probes that had more than 5 counts (a control female PND 150 sample and a high fructose diet male PND 150 sample) and were removed from further analyses (Harrill et al. 2024). All other QC thresholds were met. Samples were then separated by PND and by sex, before further analysis. In a Partek Flow software environment (Build 10.0.23.0823), probes of a mean count in each sample group of < 5 were filtered out. Probe counts were then normalized by counts per million (CPM, added a count of 1.0, then log₂-transformed). Samples with normalized counts were then visualized by principal components analysis (PCA; first 3 principal components observed) to identify any outliers. Two samples were removed by this method: a high fructose diet PND38 female exposed to 5ppm Cd sample and a control diet PND3 female with no exposure to Cd sample. Using these normalized counts, we determined differentially expressed genes from age-match and diet-matched controls using DESeq2 (significance based on FDR ≤0.1) (Love et al. 2014). A fold-change cut-off was not applied to generate DEG lists for each age/sex grouping.

Further gene pathway analysis sets were carried out using Ingenuity Pathways Analysis (IPA) (Qiagen Inc., https://www.qiagenbioinformatics.com/products/ingenuitypathway-analysis). Some visualizations were generated using R package VennDetail for Venn and UpSet plot generation (v 4.3 https://github.com/guokai8/VennDetail).

cDNA synthesis.

Fresh RNA dilutions were used for each cDNA synthesis reaction. cDNA was made using the iScript cDNA synthesis kit (BioRad, Carlsbad, CA), following manufacturer’s instructions. Each cDNA synthesis reaction contained 4 uL of 5x iScript Advanced Reaction Mix, 1 uL of iScript Advanced Reverse Transcriptase, 5 uL of the diluted RNA template, and 10 uL of nuclease free water. Samples were then incubated using a BioRad T100 Thermal Cycler using the following steps: Priming for 5 minutes at 25 °C, reverse transcription for 20 minutes at 46 °C, reverse transcription inactivation for 1 minute at 95 °C. Samples were then moved to 4 °C for storage.

Quantitative reverse-transcriptase polymerase chain reaction (qRT-PCR) and analysis.

A mastermix of SsoAdvanced Universal Supermix, PrimePCR Assay (BioRad), nuclease-free water, forward and reverse primers (PrimePCR assays, BioRad, see below) was vortexed briefly and plated with cDNA samples to a 384-well plate, using 10 μL technical triplicate. The 384-well plate was then sealed, briefly vortexed, and spun down to dispel bubbles. Samples were then cycled in a BioRad CFX384 RealTime System with the following steps: Activation for 2 minutes at 95 °C (1 cycle), Denaturation for 5 seconds at 95 °C then annealing/extension for 30 seconds at 60 °C (40 cycles), and finally a melt curve at 5 seconds a step from 65-95 °C in 0.5 °C increments (1 cycle). PrimePCR (BioRad) Assay IDs for gene expression targets were: qMmuCED0041303 (Ppia), qMmuCED0040751 (Rplp0), qMmuCID0011348 (Scd1), qMmuCED0045676 (Fasn), qMmuCED0044791 (Acyl), qMmuCID0021174 (Pklr), qMmuCID0024354 (Elovl6), qMmuCID0008222 (Rgs16), and qMmuCID0010304 (Acss2). Relative gene expression was assessed between experimental groups using baseline (initial target amount or N₀) estimation based on PCR kinetics and log-linear extrapolation from the early plateau phase of PCR (Peirson et al., 2003; Ruijter et al., 2009). Target gene expression N_o values were normalized to a geomean of reference genes Ppia and Rplp0 N_o values for matched samples. These reference genes were determined from a panel of putative reference genes (Pgk1, Hmbs, Tbp, Alas2, G6pdx, Ywhaz, Ipo8, Sdha, Tfrc, Actb, Nono, Hsp90ab1, Hprt, Ldha, Rplp0, Gapdh, Psmc4, Rpl13a, Rpl30, Ppia, Rps17, and B2m) using targeted RNA-seq data. The top two stable genes in all PND88 and PND150 sample groups were Ppia and Rplp0, as determined by NormFinder (https://www.moma.dk/software/normfinder).

DNA methylation array and data analysis.

DNA was diluted to 50 ng/μl in Qiagen Buffer AE and arranged in a 96-well plate for dry-ice shipment to Diagenode Laboratory (Hologic, Inc., Denville, NJ) for DNA methylation array assessment of ~285K CpG mouse loci using the Infinium Mouse Methylation BeadChip (MM285) array. Briefly, DNA was bisulfite converted using the EZ-96 DNA Methylation Kit (Zymo Research, Inc., Irvine, CA), following manufacturer’s instructions. C->T conversion was conducted for 16 cycles of 95°C for 30 s, 50°C for 50 min, holding at 4°C after conversion. After washing, preparation for the array followed Illumina’s kit instructions. As part of this protocol, DNA is enzymatically fragmented, ethanol precipitated, resuspended in RA1 buffer, and denatured at 95°C for 20 m. 15 μl of each sample is then loaded onto the room temperature equilibrated BeadChip lanes, then prepped DNA is hybridized at 48°C for 16-24 hrs in a hybridization chamber. BeadChips are then washed, stained, and dried. Prepped BeadChips are then loaded onto an Illumina iScan Reader and generated IDAT files which captured methylated (“M” probe) or unmethylated (“U” probe) statuses of measured loci, labeled with green or red fluorophores, respectively.

IDAT files of mouse methylation arrays were read into R with the minfi package (Aryee et al. 2014). Then, data were preprocessed with background and dye bias correction using the preprocess Noob method (Fortin et al., 2017). DNA methylation data were filtered based on these criteria: any samples having more than 5% probes failed array QC standards, all CpG probes on the X, Y and MT chromosomes, probes that did not meet design objective (sequence mismatches and suboptimal hybridization performance) were excluded. There were approximately 275-278K CpG probes remaining after exclusions, depending on PND/sex group assessed. For each sample, the sex was predicted based on DNA methylation profile and compared with recorded sex. If predicted sex was different from estimated sex, predicted sex was used for further analysis (only one sample did not match estimation in a PND3 sample). Principal component analysis (PCA) was performed to identify any potential batch effect, where one array slide (12 samples) clustered separately from other array slides. ComBat function in sva package (Leek et al. 2012) was used to obtain batch corrected methylation profiles.

To investigate associations between treatment and control groups, normalized and batch-corrected beta-values were transformed to log ratio, defined as log₂[β/(1 − β)], and then fitted using robust linear regression (Fox and Weisberg 2011). Significance was assessed using a Benjamini–Hochberg false discovery rate (FDR) of less than 0.05 (Benjamini and Hochberg 1995). IDAT and processed DNA methylation files are available on http://data.gov.

Tissue extraction and metal analysis by Inductively Coupled Plasma-Mass Spectrometry (ICP-MS).

Trunk blood was collected after euthanasia by decapitation, liver and kidneys excised, weighed, and stored frozen at −20°C until being processed. All tissues were dissociated with 0.5 mL of 25% tetramethylammonium hydroxide (TMAH, Thermo Scientific, Electronic Grade) at 65°C for 2 h. The resulting colloidal suspension was digested with 1 mL of 69% (v/v) concentrated nitric acid (Optima Grade, Fisher Scientific) at 95°C for 2 h. The digestate was diluted gravimetrically, centrifuged, and an aliquot was transferred for elemental analysis after further dilution to achieve a final acid concentration of 2%.

Elemental analyses were performed using a Nu AttoM Single Collector-ICP-MS (Nu Instruments Ltd., Wrexham, UK). The Nu AttoM is a double focusing magnetic sector mass spectrometer, with forward (Nier–Johnson) geometry. It was equipped with a concentric glass nebulizer connected to a Peltier cooled spray chamber, and platinum skimmer and interface cones. The radio frequency power was set at 1300 W. Samples were introduced using a SC-FAST flow injection analysis system (ESI, USA) and an ESI’s MC2 peristaltic pump at 0.20 mL/min with on-line internal standard Indium at 2 ppb level. Analytical calibration was performed using multielement standards from SCP Science, USA. NIST SRM 1643a and 1640a were analyzed in each analytical sequence as a quality control of the instrument. NIST SRM 1577c Bovine liver was used to validate the overall method for Cd and other elemental analyses. Limits of detection limits (LOD) for Cd, Cu, Fe, Mn, and Zn were 0.001 ng/mL, 0.21 ng/mL, 1.27 ng/mL, 0.065 ng/mL, and 1.02 ng/mL, respectively. For metal concentrations below LOD, a value equal to LOD divided by the square root of 2 was estimated (Hornung and Reed, 1990).

Statistical analysis.

Data are presented as means and standard errors (S.E.). Statistical comparisons were conducted by multivariate ANOVA (data log-transformed when required by heterogeneous variance), incorporating all the factors in a single test to avoid an increased probability of type 1 errors that might otherwise result from multiple tests of the same data set. The factors were Cd dose, diet, sex, and age (repeated measure for weights, where the same animal was weighed at multiple time points; not repeated measure when different animals were euthanized for determinations at various ages). Where we identified interactions of Cd dose and/or diet with each other or with age, data were then subdivided for lower-order ANOVAs. Where pairwise comparisons were enabled by the subdivisions, we used Tukey-Kramer post hoc tests to establish individual differences among groups. In the absence of interactions, we compiled only the main treatment effects. Significance was assumed at the level of p < 0.05, two-tailed.

Results

Cadmium and other essential metals in mouse dams and offspring.

Exposure to Cd in drinking water to mouse dams for 8 weeks (3 weeks premating, 2.6 weeks of gestation, and 2.4 weeks of lactation) led to dose-dependent accumulation of the metal in maternal liver; co-treatment with a high-fructose diet (HFrD) did not significantly alter the body burden of Cd (Table 1A). Only a small fraction of Cd (<0.1% of maternal level) was detected in the fetal liver at term, regardless of Cd dose or sex. In the Cd-exposed offspring at weaning, only traces of Cd (~0.06-0.16 ng/g) were detectable in blood, although significant amounts of the metal were accumulated in liver (~0.8-7.2 ng/g), and even greater in the kidney (~2-28 ng/g), in a dose-dependent manner (Table 1B); there were no significant effects of HFrD or sex. Consistent with its known long half-life, about 45% of Cd was still detected in the liver of 27-week old adult offspring exposed to the 5 ppm dose, after reaching a peak level at weaning age and cessation of exposure at PND16 (Table 1C). Concurrent with measurements of Cd in mouse liver, levels of Cu, Fe, Mn, and Zn in this tissue were also determined in the dams and offspring at fetal, neonatal, and adult ages, but there were no consistent patterns of change for these essential metals in response to maternal Cd exposure (Tables S1-S4).

Table 1A. Liver Cd concentrations (ng/g).

Treatment	ConD+0Cd	ConD+0.5Cd	ConD+5Cd	HFrD+0Cd	HFrD+0.5Cd	HFrD+5Cd
Maternal (postweaning)	4 ± 0 (n=13)	157 ± 17 (n=7)	8831 ± 1440 (n=13)	5 ± 0.2 (n=11)	539 ± 178 (n=7)	7804 ± 1346 (n=7)
Female fetus (GD18)	0.59 ± 0.3 (n=3)	0.22 ± 0.01 (n=4)	2.29 ± 0.71 (n=4)	0.18 ± 0.04 (n=4)	0.35 ± 0.08 (n=4)	1.92 ± 0.51 (n=4)
Male fetus (GD18)	0.33 ± 0.04 (n=6)	0.27 ± 0.06 (n=6)	1.72 ± 0.25 (n=5)	0.39 ± 0.09 (n=5)	0.21 ± 0.02 (n=6)	1.78 ± 0.41 (n=6)

Data represent means ± S.E. of numbers of litters examined as indicated in parenthesis. 2-way ANOVA (factors = Cd and diet) indicates significant main effects in tissue Cd concentrations (p<0.0001), with the high dose group distinguishable from the control and low dose. There is no interaction with diet.

Table 1B. Neonatal tissue Cd concentrations (ng/g) at PND22.

Tissue	Sex	ConD+0Cd	ConD+0.5Cd	ConD+5Cd	HFrD+0Cd	HFrD+0.5Cd	HFrD+5Cd
Blood	Female	0.07 ± 0.01 (n=27)	0.06 ± 0.01 (n=11)	0.16 ± 0.02 (n=10)	0.06 ± 0.01 (n=21)	0.11 ± 0.06 (n=6)	0.18 ± 0.03 (n=11)
	Male	0.07 ± 0.01 (n=25)	0.02 ± 0 (n=10)	0.19 ± 0.03 (n=10)	0.05 ± 0.01 (n=21)	0.05 ± 0.02 (n=7)	0.12 ± 0.01 (n=10)
Liver	Female	0.51 ± 0.03 (n=26)	0.79 ± 0.09 (n=11)	6.62 ± 0.44 (n=9)	0.49 ± 0.03 (n=21)	0.94 ± 0.10 (n=6)	7.33 ± 1.21 (n=12)
	Male	0.40 ± 0.02 (n=25)	0.80 ± 0.11 (n=11)	8.89 ± 2.18 (n=10)	0.42 ± 0.02 (n=21)	0.73 ± 0.05 (n=7)	6.12 ± 0.98 (n=10)
Kidney	Female	0.51 ± 0.06 (n=26)	1.68 ± 0.23 (n=11)	27.6 ± 2.0 (n=10)	0.52 ± 0.07 (n=20)	2.60 ± 0.57 (n=6)	28.5 ± 3.9 (n=12)
	Male	0.42 ± 0.03 (n=25)	2.07 ± 0.31 (n=11)	31.2 ± 4.3 (n=10)	0.84 ± 0.41 (n=21)	1.63 ± 0.17 (n=7)	25.5 ± 3.7 (n=9)

Data represent means ± S.E. of numbers of litters examined as indicated in parenthesis and were log-transformed because of heterogeneous variance across tissue type. 4-way ANOVA (factors = tissue, Cd, diet, and sex) indicates a main effect of tissue types; hence, each tissue is further analyzed. For all three tissues, 3-way ANOVA (factors = Cd, diet, and sex) indicates a significant main effect of Cd that is dose-dependent (p<0.0001) but not for diet or sex. There is an interaction between Cd and diet: HFrD increases Cd levels at the low dose but decreases Cd levels at the high dose.

Table 1C. Adult liver Cd concentrations (ng/g) at PND188.

Sex	ConD+0Cd	ConD+0.5Cd	ConD+5Cd	HFrD+0Cd	HFrD+0.5Cd	HFrD+5Cd
Female	0.82 ± 0.05 (n=4)	—	2.88 ± 0.56 (n=3)	1.19 ± 0.26 (n=4)	1.16 ±0.22 (n=4)	2.93 ± 0.82 (n=4)
Male	0.94 ± 0.05 (n=4)	—	1.92 ± 0.43 (n=4)	1.13 ± 0.20 (n=4)	0.94 ± 0.04 (n=4)	1.19 ± 0.03 (n=3)

Note. Data represent means ± S.E. of numbers of litters examined as indicated in parenthesis. 3-way ANOVA (factors = Cd, diet, and sex) indicates significant main effects in tissue Cd concentrations (p<0.0004). There is an interaction between Cd and sex, only females showed a main effect of Cd.

Maternal effects.

Exposure to 0.5 ppm Cd before and during pregnancy significantly reduced maternal weight gain during gestation, but the effect was not seen at the higher dose (5 ppm) (Table 2). HFrD by itself reduced maternal weight gain. All maternal weight effects were no longer apparent after weaning, nor were there any effects on liver weight and triglyceride content. Neither Cd nor HFrD altered the percentage of live pups at birth.

Table 2. Maternal weight gain and pregnancy outcomes.

Treatment	ConD+0Cd	ConD+0.5Cd	ConD+5Cd	HFrD+0Cd	HFrD+0.5Cd	HFrD+5Cd
Maternal body weight gain (g)	27.1 ± 1.2 (n=36)	22.7 ± 1.7 (n=16)	28.6 ± 1.2 (n=19)	20.4 ± 1.2 (n=31)	20.6 ± 1.9 (n=11)	22.2 ± 1.0 (n=15)
% Live pups per litter	89.8 ± 1.6 (n=40)	88.3 ± 2.0 (n=18)	88.8 ± 2.3 (n=22)	86.5 ± 2.5 (n=34)	88.1 ± 2.2 (n=14)	89.9 ± 2.2 (n=19)
Pup weight at birth (g)	1.68 ± 0.02 (n=39)	1.66 ± 0.02 (n=17)	1.67 ± 0.03 (n=22)	1.55 ± 0.03 (n=33)	1.60 ± 0.04 (n=12)	1.54 ± 0.03 (n=19)
Weaned dam weight (g) †	37.2 ± 0.8 (n=14)	37.0 ± 1.1 (n=7)	34.3 ± 1.1 (n=13)	35.4 ± 0.9 (n=11)	34.4 ± 0.5 (n=7)	35.3 ± 0.9 (n=7)
Weaned dam Relative liver weight (%)	5.29 ± 0.18 (n=14)	5.16 ± 0.09 (n=7)	5.21 ± 0.22 (n=13)	5.48 ± 0.14 (n=11)	5.75 ± 0.26 (n=7)	5.75 ± 0.19 (n=7)
Weaned dam liver triglycerides (μmol/g)	70.3 ± 8.2 (n=14)	66.7 ± 8.5 (n=7)	66.3 ± 12.1 (n=13)	60.1 ± 7.3 (n=11)	41.2 ± 7.4 (n=7)	39.0 ± 3.2 (n=7)

Data represent means ± S.E. of numbers of litters examined as indicated in parenthesis. 2-way ANOVA (factors = Cd and diet) indicates a significant diet effect in maternal weight gain (p<0.0001) and pup weight at birth (p< 0.0005). ^†A subset of dams was weighed 10 days after weaning and evaluated for hepatic triglycerides.

Growth trajectory of offspring.

Preweaning body weights of offspring under control diet (ConD) were not affected by Cd but were slightly (5-8%) and significantly lower in the HFrD group, independent of Cd exposure (Fig. 1). After weaning, we saw a different profile of growth effects (Fig. 2). The high dose of Cd by itself reduced growth selectively in female offspring. In females for dams exposed to HFrD alone or in conjunction with Cd, an overall growth deficit was seen, but a significant interaction between chemical and non-chemical exposure was not detected. By adulthood, the sex differences disappeared, and only the residual effects of HFrD but not those of Cd remained observable.

Fig. 1.

Effects of maternal exposure to cadmium (Cd) and high-fructose diet (HFrD) during pregnancy and lactation on neonatal growth of mouse offspring. Average body weights (males and females combined) of pups from each litter were determined at intervals of several days from PND1 until weaning. Data represent means ± S.E. of 12-40 litters for each treatment group. 3-way ANOVA with repeated measures (factors = age, Cd, and diet) indicated a significant main effect of HFrD (p<0.001) but not of Cd, and there was no significant interaction between HFrD and Cd treatments. A significant interaction between HFrD and age was noted (p<0.001), but the HFrD effect was significant at each age point examined.

Fig. 2.

Effects of maternal exposure to cadmium (Cd) and high-fructose diet (HFrD) on growth of female (upper panels) and male (lower panels) offspring from weanling to adolescence. Average body weight of littermates (male or female) was determined, and litter means were tabulated. Data represent means ± S.E. of 14-40 litters for each treatment group. 4-way ANOVA with repeated measures (factors = sex, age, Cd, and diet) indicated significant main effects of sex (p<0.001), Cd (p<0.01), and HFrD (p<0.001); and a significant interaction between age and Cd (p<0.05), but not between sex and HFrD, nor between Cd and HFrD. The HFrD effect was significant at every age point determined; the Cd effect was also significant at PND28, 42, 56 and 63 but not at PND35, 49 and 70. Treatment effects in adult (PND150) offspring were presented in a Table (bottom panel). 3-way ANOVA (factors = sex, Cd, and diet) indicated significant main effects of sex (p<0.001) and HFrD (p<0.001) but no interaction between sex and HFrD. There was no significant effect of Cd nor interaction between Cd and HFrD.

Developmental landmarks of offspring.

Maternal exposure to Cd and/or HFrD did not alter the age of eye-opening in the offspring (Table 3A); however, their pubertal development was significantly advanced. The onset of preputial separation in males exposed to Cd alone at both concentrations occurred at an earlier postnatal age compared to controls (Table 3C); those males exposed to HFrD alone or in combination with Cd showed similar advancement, although the dietary manipulation did not exacerbate the Cd effect. The body weights of Cd or HFrD groups were significantly lower than those of untreated controls when puberty was reached. A similar profile of accelerated pubertal onset was seen in the Cd-exposed female offspring, as vaginal opening was observed at an earlier age compared to untreated controls (Table 3B); however, unlike the observation with males, HFrD alone or in combination with Cd did not significantly alter the timing of vaginal opening.

Table 3. Developmental landmarks of mouse offspring.

A: Eye opening (EO)
Treatment	ConD+0Cd (n=40)	ConD+0.5Cd (n=18)	ConD+5Cd (n=22)	HFrD+0Cd (n=33)	HFrD+0.5Cd (n=14)	HFrD+5Cd (n=19)
Age (PND)	14.7 ± 0.1	14.6 ± 0.2	14.4 ± 0.1	14.6 ± 0.1	14.4 ± 0.1	14.6 ± 0.1
Data represent mean age ± S.E. of numbers of litters examined as indicated in parenthesis. 2-way ANOVA (factors = Cd and diet) did not indicate a significant main effect of Cd or diet on timing of eye opening.

B: Vaginal opening (VO)
Treatment	ConD+0Cd (n=33)	ConD+0.5Cd (n=15)	ConD+5Cd (n=17)	HFrD+0Cd (n=30)	HFrD+0.5Cd (n=13)	HFrD+5Cd (n=16)
Age (PND)	30.7 ± 0.4	28.1 ± 0.5	28.2 ± 0.5	29.3 ± 0.4	29.5 ± 0.6	28.6 ± 0.3
BW(g)	24.2 ± 0.4	22.7 ± 0.4	21.7 ± 0.3	21.9 ± 0.4	23.1 ± 0.5	21.0 ± 0.5
Data represent mean age ± S.E. of numbers of litters examined as indicated in parenthesis. 2 way ANOVA (factors = Cd and diet) indicates a significant main effect with Cd (p<0.001) but not with diet on vaginal opening (VO), and a significant interaction between Cd and diet (p=0.01). The responses to Cd were further analyzed by individual diet. With ConD, VO observed in 0.5 or 5 ppm Cd dose group was significantly earlier than 0Cd (p<0.001). With HFrD, the timing of VO in 0, 0.5 or 5 ppm Cd was not significantly different from 0Cd. The body weight of pups exposed to 5 ppm Cd at puberty was significantly lower than those of 0Cd (p<0.05).

C: Preputial separation (PS)
Treatment	ConD+0Cd (n=33)	ConD+0.5Cd (n=15)	ConD+5Cd (n=17)	HFrD+0Cd (n=30)	HFrD+0.5Cd (n=13)	HFrD+5Cd (n=16)
Age (PND)	30.5 ± 0.2	27.3 ± 0.4	26.7 ± 0.3	27.6 ± 0.3	27.4 ± 0.4	27.2 ± 0.3
BW (g)	31.0 ± 0.4	27.5 ± 0.7	24.7 ± 0.5	25.4 ± 0.6	26.2 ± 0.9	23.7 ± 0.6
Data represent mean age ± S.E. of numbers of litters examined as indicated in parenthesis. 2 way ANOVA (factors = Cd and diet) indicates significant main effects of Cd (p<0.001) and diet (p<0.005), and a significant interaction between Cd and HFrD (p<0.001). With ConD, PS observed in 0.5 or 5 ppm Cd was significantly earlier than 0Cd (p<0.001) although a significant difference was not seen between these 2 dose groups. With HFrD, the timing of PS was advanced in all (0, 0.5 and 5 ppm) Cd dose groups to the same extent. Body weight of pups at puberty was significantly lower in the Cd or HFrD exposed groups (p<0.001) compared to the unexposed controls (ConD+0Cd).

Body composition of adult offspring.

The percent of body fat increased significantly with age in female offspring, but less so in the males (Fig. 3A). Exposure to Cd did not affect the body composition in either sex. HFrD alone significantly lowered the percentage of body fat, and the effect was compounded by co-exposure with 5 ppm Cd. It should be noted that significant changes in body composition were inconsistent across ages. For instance, significant interactions between HFrD and Cd were only detected at PND38 and PND90, but not at other ages. A similar profile of changes was seen when data were expressed as ratios of body fat/lean mass (Fig. 3B).

Fig. 3.

Effects of maternal exposure to cadmium (Cd) and high-fructose diet (HFrD) on body fat in female and male offspring from adolescent (PND38) to adult (PND150) ages (top panels) and their ratio of body fat/lean (bottom panels). Data represent means ± S.E. of 7-10 litters for each treatment group. For %body fat, 4-way ANOVA (factors = sex, age, Cd and diet) indicated significant main effects of sex, age, and HFrD (p<0.005), but not of Cd; no significant interactions were detected between HFrD and sex nor between HFrD and age. However, a significant interaction was found between HFrD, Cd and age (p<0.05). For ratio of body fat/lean, the statistical profile was identical to that of % body fat.

Glucose tolerance of adult offspring.

Fasting glucose was significantly lower in adult male offspring exposed to high dose Cd alone, as well as in those exposed to HFrD alone; co-exposure to HFrD did not alter the Cd effect. Low dose Cd exposure did not lead to any significant changes with either diet (Fig. 4A). A similar pattern of changes was observed for glucose tolerance (Fig. 4B). Male offspring exposed to 5 ppm Cd alone or to HFrD alone, displayed reductions of area under the curve (AUC), but the Cd effect was not impacted by co-exposure to HFrD.

Fig. 4.

Effects of maternal exposure to cadmium (Cd) and high-fructose diet (HFrD) on fasting glucose levels (top panels) and glucose tolerance (AUC, bottom panels) in female and male offspring from adolescent (PND47) to adult (PND137) ages. Data represent means ± S.E. 7-16 litters for each treatment group, except those for PND 47 where n=3-16. For fasting glucose levels, 4-way ANOVA (factors = sex, age, Cd, and diet) indicated significant main effects of sex (p<0.001), Cd (p<0.05) and HFrD (p<0.03), but no significant interaction between Cd and HFrD. For glucose tolerance (AUC), 4-way ANOVA (factors = sex, age, Cd, and diet) indicated significant main effects of sex (p<0.001), Cd (p<0.03) and HFrD (p<0.001), and a significant interaction between sex and Cd (p<0.05).

Hepatic lipid accumulation in adult offspring.

As shown in Fig. 5, neither maternal Cd nor HFrD significantly affected liver weight or hepatic triglycerides in the adult offspring, nor were there significant interactions between the two factors.

Fig. 5.

Effects of maternal exposure to cadmium (Cd) and high-fructose diet (HFrD) on liver weight (top panels) and hepatic triglycerides (bottom panels) in young adult female and male offspring. For liver weight, 4-way ANOVA (factors = sex, age, Cd, and diet) indicated significant main effects of sex (p<0.001) and age (p<0.001) but not of Cd or diet, and a significant interaction between age, Cd, and diet (p<0.03). Data were therefore further analyzed at separate ages. On PND38, a significant main effect of diet (p<0.03) was found, and there was a significant interaction between Cd and diet on PND88 (p<0.01). For hepatic triglycerides there were significant main effects of sex (p<0.005) and age (p<0.001) but not of Cd or diet, and a significant interaction between Cd and diet (p<0.02). Hence, data were further analyzed at different Cd doses. A significant main effect of diet was not noted in the 0 Cd group, but significant main effects of diet (p<0.02) and sex (p<0.03) were found for both 0.5 ppm Cd and 5 ppm Cd groups.

Hepatic gene expression profiles of offspring.

Given the impacts on fasting glucose levels and on glucose tolerance, we more closely examined liver gene expression involved in insulin signaling, specifically on the downstream genes of the insulin-responsive transcription factors 3 sterol regulatory element binding protein 1 (Srebp1; Seo et al. 2009), carbohydrate responsive element binding protein (ChREBP; Benichou et al. 2024), and forkhead box protein O1 (FoxO1, Sullivan et al. 2015). Only Acyl and Fasn demonstrated any significant increase in gene expression with maternal Cd treatment in PND88 female offspring when compared to controls that were not exposed to Cd or HHrD maternally (p≤0.05, pairwise two-tailed t-test; Suppl Fig 1A). For offspring who were exposed to HFrD maternally, Acyl expression also increased with low dose Cd in PND150 female offspring (Supp. Fig. 1A), whereas Fasn expression was elevated in PN150 males in the high Cd group (Supp. Fig. 2A). We also examined whole-transcriptome assessment using targeted RNA-sequencing of offspring liver sampled at PND3, 38, 88 and 150. We observed few alterations caused by maternal exposure to either dose of Cd , with or without HFrD (FDR q-value≤0.01; Table 4, Supp. Data 1), with the most robust alterations occurring with 0.5 Cd maternal exposure in the male offspring at PND38 (13 DEGs). Targeted RNA-seq did not indicate significant changes in Srebp1, ChREBP, or FoxO1target genes with maternal Cd, HFrD or both, in either sex (Supp. Figs. 1B and 2B).

Table 4. Differentially expressed genes (DEG) summary (female/male).

	PND 3		PND 38		PND 88		PND 150
Cadmium	ConD	HfD	ConD	HfD	ConD	HfD	ConD	HfD
+0Cd	-	0/0	-	1/0	-	0/0	-	1/0
+0.5Cd	0/0	2/1	0/13	2/0	6/0	0/3	0/0	1/4
+5Cd	1/2	0/0	0/0	0/0	0/0	0/0	0/0	1/0

Note. Number of DEGs were determined by comparison to matched sex-age ConD+0Cd samples. The first number are DEGs observed in female offspring mouse liver, the second number were DEGs observed in male offspring mouse liver. Significant DEGs were determined by FDR step up threshold of 0.1 or less. Greyed-out boxes represent same sample-group comparisons. n=3-8, with the exception of female PND150 HfD samples (n=2; bolded and italicized).

Hepatic DNA methylation profiles in offspring.

We further examined alterations to CpG methylation in PND3 and PND88 male and female offspring liver using an Illumina Infinium Mouse Methylation BeadChip (~285K methylation sites). We focused on differentially methylated CpGs (dmCpGs) compared to controls (normal diet, 0 Cd) due to maternal HFrD alone, 0.5 or 5 ppm Cd alone, and combined effects (Table 5, Supp. Data 2 & 3). dmCpGs were determined by pair-wise robust linear regression between the control and comparison groups, where significance was noted by Benjamini–Hochberg false discovery rate (FDR) of less than 0.05. Overall, the number of identified dmCpGs were a small fraction compared to the overall CpGs assayed (no greater than 0.15% in any of the comparisons examined). Among all examined groups in both sexes, there were more dmCpGs at PND3 on average (220.5) compared to PND88 (132.5); however, females exhibited more dmCpGs than males at PND3 (284.8 vs. 156.2) and males more than females at PND88 (177.2 vs. 87.8). In addition, when compared to controls, most PND3 female experimental groups (regardless of maternal diet or Cd exposure) exhibited greater hypomethylation (average 63%) than the PND3 males, whereas at PND88, females were hypermethylated (average 62%) (Supp. Figs. 3A, E). PND3 and PND88 male groups exhibited more mixed methylation states (averaged 51% and 56% hypermethylation, respectively), with strongest hypermethylation seen in HFrD+0.5Cd (compared to ConD+0Cd) at both time points (Supp. Figs. 3C, G). Very few shared dmCpGs were observed in both female and male experimental groups; only 25 and 11 dmCpGs were observed throughout the PND3 and PND88 experimental groups, respectively, indicating overall unique dmCpG responses between the two sexes. In addition, within the same sex and PND, most dmCpGs were largely independent among experimental groups, with some commonality observed due to HFrD and/or same ppm Cd treatment (Supp. Figs. 3B, D, F, H). There was some evidence of “persistent” dmCpGs that occurred in matched experimental groups at both PND3 and PND88 (Table 6).

Table 5. Differentially methylated CpGs (dmCpGs) in PND3 and PND88 offspring liver.

						dmCpGs		in cis dmCpGs		# nearest gene TSSs (total unique)
Sex	PND	Cd (ppm)	Diet	# of samples	Total probes	Hyper	Hypo	Hyper	Hypo
Female	PND3	0.5	Con	9	275813	19	29	12	17	28
		5	Con	8	275813	122	313	46	179	221
		0	HFr	8	275813	83	222	27	84	98
		0.5	HFr	8	275813	117	98	37	51	80
		5	HFr	7	275813	152	269	58	158	208
	PND88	0.5	Con	9	270535	70	54	22	22	44
		5	Con	8	270535	39	15	15	5	20
		0	HFr	9	270535	48	32	17	12	29
		0.5	HFr	10	270535	83	39	26	15	38
		5	HFr	7	270535	31	28	12	13	25
Male	PND3	0.5	Con	7	278122	73	83	26	30	55
		5	Con	8	278122	43	69	15	32	48
		0	HFr	8	278122	97	84	36	42	75
		0.5	HFr	7	278122	128	46	60	20	79
		5	HFr	9	278122	66	92	29	37	65
	PND88	0.5	Con	8	272914	197	89	77	28	99
		5	Con	7	272914	30	30	12	12	24
		0	HFr	8	272914	72	58	39	28	63
		0.5	HFr	6	272914	213	113	70	44	112
		5	HFr	9	272914	36	48	10	18	28

Note: Differentially methylated CpGs (dmCpGs) were determined compared to age- and sex-matched ConD+0Cd group using robust linear regression and significance determined by adjusted p-value (FDR) <0.05. “In cis dmCpGs” were further filtered by those dmCpGs that exhibited > 1% methylation change compared to reference group and were within 5000bp of a gene transcription start site (TSS).

Table 6. Common dmCpGs between PND3 and PND88 offspring based on treatment group.

Treatment group	Sex	Assay ID	Location	Closest gene TSS	PND3 dM	PND88 dM
ConD+0.5Cd	Female	-	-	-	-	-
ConD+5Cd	Female	cg29535263	chr11:52338517	ENSMUSG00000139961	0.074	0.190
HFrD+0Cd	Female	cg47125677	chr9:97007142	ENSMUSG00000127397	0.183	0.024
HFrD+0Cd	Female	cg44216183	chr7:7206884	Zfp772	0.050	0.044
HFrD+0.5Cd	Female	cg28753519	chr10:82720172	Eid3	−0.250	−0.328
HFrD+5Cd	Female	-	-	-	-	-
ConD+0.5Cd	Male	cg43024296	chr5:146690622	Usp12	−0.241	−0.264
ConD+0.5Cd	Male	cg36156841	chr19:11570376	Ms4a6d	−0.141	−0.210
ConD+5Cd	Male	-	-	-	-	-
HFrD+0Cd	Male	cg45122815	chr7:126358895	Mapk3	0.216	−0.071
HFrD+0Cd	Male	cg43528611	chr6:68078695	Igkv1-117	0.176	0.097
HFrD+0.5Cd	Male	-	-	-	-	-
HFrD+5Cd	Male	-	-	-	-	-

Note. Assay ID based on Illumina-assigned ID from the Infinium Mouse Methylation Array (MM285). dmCpGs based on adjusted p-value (FDR) <0.05. dM is calculated as the treatment group methylation ratio subtracted from the reference group methylation level.

Overlaps between the handful of DEGs observed and genes that were closest to dmCpGs based on their transcription start site (TSS) were also examined to determine if there was any connection between methylation alterations and differentially expressed genes. In female offspring, no intersections were observed, regardless of age or treatment group. In males, three DEGs, Cirbp (in PND38 ConD+0.5Cd), Eci1 and Atf5 (DEGs in PND150 HFrD+0.5Cd) were linked to a dmCpGs (chr10:79995865 [Cirbp] in PND88 ConD+0.5Cd; chr17:24639928 [Eci1] in PND88 HFrD+0.5Cd; chr7:44465252 [Atf5] in PND3 HFrD+0.5Cd).

Further analyses were performed on lists of genes that were “in cis” with dmCpGs (defined as within 5000 bp of a gene TSS and exhibited >1% methylation change compared to reference group). Closest TSS-linked genes for in cis dmCpGs were analyzed for enrichment of gene pathways using IPA (Supp. Figs. 4A-E, Supp. Data 4). Largely, enriched pathways were unique to the offspring of the individual treatment groups. Commonalities in pathways between female treatment groups at the same age were mediated primarily by HfrD at PND3 (Supp. Fig. 4A) including pathways of amino acid biosynthesis and transcriptional regulation. In males, we additionally saw common HFrD-mediated pathways at PND3 (involved in injury response, inflammation, stellate cell activity, and insulin response pathway), but also a 12-pathway intersection between PND3 male ConD+5Cd and HFrD+0.5Cd offspring, indicative of liver injury response, metabolism, inflammation, and the insulin pathway (Supp. Fig. 4B). PND88 in cis dmCpGs-linked pathways were largely independent based on maternal treatment groups.

When common pathways between sexes in matched treatment groups at PND3 (Supp. Fig. 4C) and PND88 (Supp. Fig. 4D) were examined, PND3 ConD+5Cd male and females intersected the most (20 pathways). These pathways included ERK/MAPK Signaling, CDK5 Signaling, ID1 Signaling Pathway, PTEN Signaling, and Neuregulin Signaling, but also additional insulin responsive pathways such as Type II Diabetes Mellitus Signaling and injury/regenerative signals such as RAC Signaling and Hepatic Fibrosis Signaling Pathway.

Finally, we examined “persistent” pathways linked to in cis dmCpGs between PND3 and PND88 of same sex treatment groups (Supp. Fig. 4E). While there was little evidence of carryover for most groups, there was a 9 pathway overlap for male HFrD+0Cd, indicating maternal diet impacted male offspring until PND88. These pathways were indicative of xenobiotic metabolism (Aryl Hydrocarbon Receptor Signaling) and injury response and regeneration (multiple pathways involved in Activin signaling).

Discussion

This study evaluated the adverse effects of maternal exposure to cadmium in drinking water or a high-fructose diet on health consequences of the mouse offspring from early life to young adulthood, and to investigate the potential interactions between these two chemical and non-chemical stressors. Five developmental and metabolic endpoints previously shown to be associated with Cd or HFrD toxicity in experimental models were evaluated. Each individual endpoint illustrated a unique profile of changes induced by Cd exposure alone or HFrD treatment alone, with or without chemical and non-chemical interactions.

The diabetogenic potential of Cd has been well investigated and described in the literature (Tinkov et al., 2017), although findings from both human and animal studies were quite inconsistent. Several recent studies have evaluated the risks of developing diabetes in the adult offspring after maternal exposure to Cd (Saedi et al. 2024), including glucose intolerance with low dose (0.5 ppm) exposure in CD-1 mice and (Jackson et al., 2020) and high dose (150 ppm) exposure in ICR mice (Yi et al., 2021), while similar effects were not detected by Young et al. (2022) in C57BL/6J mice and Jacquet et al (2019) in Wistar rats. Hence, the glycemic responses in rodent offspring after maternal exposure to Cd may depend on species, rodent strain and sex, as well as study designs. Similarly, although fructose has been implicated for induction of diabetes in humans and laboratory rodents (Jung et al., 2022), this issue remains to be a subject of debate (Kearns et al., 2024). Saad et al. (2016) and Rodriquez et al. (2015) showed that HFrD during pregnancy led to insulin resistance in adult mouse and rat offspring, respectively; however, Alzamendi et al. (2010) failed to detect such an effect. Results in the current study provided differing findings from those reported previously. Maternal exposure to either 5 ppm Cd alone or 60% fructose diet alone during pregnancy and lactation led to fasting hypoglycemia in male CD-1 mouse offspring only, which persisted from adolescence to young adulthood. However, HFrD did not significantly affect the Cd effects, indicating a lack of interactions between these chemical and dietary stressors. Consistent with the fasting hypoglycemia, these offspring mice also displayed enhanced glucose tolerance (assessed by AUC) after maternal exposure to 5 ppm Cd alone or HFrD alone, rather than intolerance, as suggested by some of the previous studies. Nevertheless, significant interactions between Cd and HFrD were not detected.

The mechanism(s) underlying our observations are not clear, but we do not hypothesize that there was a direct involvement of persisting Cd on pancreatic β-cells in the offspring. While we did not measure metals in the pancreas, Cd is known to impair β-cell function (Hong et al., 2021), which could lead to hyperglycemia and glucose intolerance. Rather, we propose that exposure to either Cd or HFrD might have induced hyperglycemia in the dams during pregnancy (Ghafghazi and Mennear, 1973; Blakely et al., 1981), which in turn altered the early life programming of glucose metabolism in the offspring for maintenance of homeostasis. Accordingly, we examined the hepatic epigenomic and transcriptomic profiles at various life stages of the offspring liver to examine any possible impact on the methylome that have been described in previous studies. Indeed, human studies have indicated Cd-induced alterations in DNA methylation profiles in the blood of offspring due to maternal exposures (Park et al., 2022; Sanders et al., 2014), some of which persisted into childhood (Gliga et al., 2022). In a mouse model of maternal Cd exposure, observed phenotypes such as fetal growth restriction was linked to hypermethylation of the glucose transporter Glut3 in the placenta (Xu et al. 2016). In the rat liver, maternal Cd exposure resulted in sex-specific methylation changes in the fetal offspring of the glucocorticoid receptor promoter (Nr3c1) (Castillo et al., 2012), a gene that was also associated with hypermethylation in cord blood of offspring prenatally exposed to heavy metals including Cd (Appleton et al., 2017). Maternal Cd exposure was also linked to alterations in the imprinted gene network (IGN) in mouse offspring (Xu et al., 2017; Simmers et al., 2023); although there is no current evidence to support mediation by DNA methylation (Riegl et al., 2023). In our study, maternal Cd or HFrD exposure alone led to altered DNA methylation in both sexes; however, these changes were sparse and mostly independent. Similar to findings reported by Castillo et al. (2012), we also observed significant 5.8% hypermethylation near Nr3c1, but only in female offspring exposed maternally to 0.5 ppm Cd + HFrD and not with Cd exposure alone. When comparing to mouse IGN (https://www.geneimprint.com/site/genes-by-species.Mus+musculus.imprinted-All), only a handful of dmCpG in cis to imprinted genes and most were observed at PND3 and not PND88. Tle3 (TLE family member 3, transcriptional corepressor, involved in adipogenesis and lipid accumulation in the liver (Villanueva et al., 2011)) and Tpx2 (TPX2, microtubule-associated, a mitotic marker and linked to liver metabolism (Wang et al., 2023)) were the only genes consistently linked to hypomethylated CpGs in female offspring with 5 ppm Cd maternal exposure, regardless of diet. Indeed, most dmCpGs occurred at PND3 for the female offspring, associated with higher Cd exposure regardless of diet and were primarily hypomethylated. Conversely, most dmCpGs in males occurred at PND88, with lower, 0.5 ppm Cd exposure (regardless of diet) and were primarily hypermethylated. Therefore, maternal Cd exposure alone seemed to drive most of the dmCpGs and liver methylome responses were sex and age dependent. With very few exceptions (see Table 6), dmCpGs that were altered at PND3 did not persist to PND88 in the same treatment group. Because we observed such little transcriptional response, it is difficult to attribute functional consequences to these DNA methylation alterations, including the few links we have to genes that may contribute to the glucose metabolism regulation in the liver.

The effects of early life Cd exposure on adiposity of the offspring at later ages have been described previously, although the findings were also inconsistent. Ba et al. (2017) showed that adult male C57BL/6J mouse offspring from dams given ~0.2 ppm Cd in drinking water during gestation and lactation had significantly elevated fat mass at 20 weeks of age; while females were not affected. In contrast, maternal exposure to 0.5 ppm Cd in CD-1 mice led to > 2-fold increases of perigonadal fat weight in adult female but not male offspring at PND90 (Jackson et al., 2020). Hence, there are sex and strain differences in responses to Cd. Yet, maternal exposure to similar low doses of Cd (0.5 or 5 ppm) alone in the present study did not significantly alter the percent of body fat in either male or female offspring at young adult ages. The reasons behind these disparate responses to Cd between studies are not clear. However, a different profile of responses was observed with maternal exposure to HFrD in the current study, where small reductions of body fat were consistently detected in both sexes. Interestingly, Saad et al. (2016) have shown that maternal consumption of 10% fructose fluid led to an increase of visceral adipose tissue in female C57BL/J offspring mice but not males at one year of age, but Toop et al. (2017) did not detect significant changes of fat mass in male or female Wistar rat offspring from dams given high fructose corn syrup during pregnancy and lactation, likely reflecting the differences between rodent species and strains as well as the forms of HFrD (fluid vs. solid). Our findings of reduced adiposity in adult mouse offspring after maternal consumption of HFrD alone, though subtle, are in stark contrast to the two-fold increase of body fat in adult mice after consuming 15% fructose solution for 10 weeks (Jurgens et al., 2005), which may suggest involvement of early life reprogramming of lipid homeostasis that lasts into adulthood. In addition, a statistically significant interaction between Cd, HFrD, and age was noted in this study, suggesting that at adolescent ages (PND 38 and 90) exposure to high dose (5 ppm) of Cd might counteract the HFrD effects; however, these effects were transient and no longer significant at older ages. The variable changes in responses to developmental Cd+HFrD across the lifespan thus make it difficult to discern a clearcut and consistent interaction between the chemical and non-chemical stressors regarding adiposity of the adult offspring after early life insults.

The adverse effects of cadmium and fructose on liver function and their association with non-alcoholic fatty liver disease (NAFLD) are well documented (Tinkov et al., 2023; Jensen et al., 2018), although the health consequences in offspring after only maternal exposure to these two chemical and dietary stressors remain sparsely examined. Pillai et al. (2009) reported suppression of hepatic metabolizing enzymes in adult female rats exposed to Cd in early life. Jackson and colleagues (2020) described morphological changes in the liver of young adult CD-1 mouse female offspring indicative of hepatosteatosis and fibrosis after maternal exposure to 0.5 ppm Cd during gestation. Indeed, Riegl et al. (2023) showed upregulation of the target genes that are markers for steatosis in the liver of juvenile C57BL/6J mouse offspring after maternal exposure to 50 ppm Cd and suggested the involvement of imprinted gene Zac1. Young et al. (2022) also investigated the hepatic effects of whole-life Cd exposure adult male C57L/6J mouse offspring but did not detect significant changes in morphologic pathology scores or hepatic triglyceride concentrations. Thus, the Cd influences on developmental induction of fatty liver in mice may depend on strain, sex, exposure, and study design. Very few studies evaluated the effects of maternal fructose intake on hepatic function of the offspring. Smith et al. (2022) demonstrated developmental reprogramming of hepatic mitochondrial metabolism in adult guinea pig after maternal consumption of 10% fructose solution during pregnancy, which led to significant increases of triglycerides in circulation. In the current study, maternal exposure to Cd alone or HFrD alone did not significantly alter the liver weight or hepatic triglycerides of adult offspring; HFrD also failed to affect the lack of Cd responses. The absence of overt hepatosteatosis in the mouse offspring after maternal exposure to Cd and/or consumption of HFrD compared to that seen after direct exposure in adults may be related to the low Cd doses used in this study, and the choice of a less sensitive CD-1 mouse strain (vs. C57BL/6J, Fengler et al., 2016), or detection methods (biochemical measurement of triglycerides vs morphological staining). These factors may have hampered the evaluation of hepatic endpoint for interactions between these two chemical and non-chemical stressors.

Maternal exposure to Cd or HFrD has been reported to alter pubertal development, although findings have been inconsistent. For instance, Hernandez-Rodriguez et al. (2021) reported respectively, Cd-induced delays in vaginal opening and preputial separation in rats, and Li et al. (2022) in mice, although the exposure doses were as high as 200 ppm; in contrast, Parodi et al. (2017) and Li et al. (2018) described precocious puberty caused by Cd exposure in female rats. In our study, a sex-dependent effect of Cd and HFrD was observed with relatively low doses of maternal Cd exposure compared to previous studies, advanced onsets of puberty were detected in both male and female mouse offspring. In comparison, maternal HFrD exposure by itself produced a sex-dependent acceleration of pubertal development only in the male offspring. The dietary stressor did not affect the females appreciably, except in combination with 5 ppm Cd. The Cd or HFrD-induced precocious onset of puberty observed in the presence of body weight reduction is intriguing. Typically, growth deficits are accompanied by delays in pubertal development. The advanced onset of puberty observed in this study may therefore reflect specific changes of hormonal and neural signals that orchestrate this developmental event (Livadas and Chrousos, 2019), as general developmental landmarks such as eye-opening was not affected by the chemical and non-chemical stressors. Indeed, accelerated pubertal onset in mice by pyrethroid insecticide was shown to involve disruption of the hypothalamic-pituitary-gonadal axis (Ye et al., 2017), and hyperinsulinemia could induce early puberty in females (Saleh et al., 2022). The mechanism(s) underlying the Cd and HFrD effects on pubertal development will require additional exploration. However, this accelerated developmental event should not be viewed as beneficial or even innocuous. Several studies have previously shown that precocious onset of puberty by environmental pollutants such as phthalates and bisphenols could lead to deleterious consequences in reproductive health later in life (Rattan et al., 2018; Ma et al., 2021; Nah et al., 2011). Thus, the early onset of puberty induced by maternal exposure to Cd and/or HFrD observed in the current study clearly warrants additional evaluations of reproductive functions of the mouse offspring in adulthood.

Previous studies have indicated that early life exposure to Cd or HFrD alone produced growth deficits and developmental alterations that might be linked to adverse health consequences later in life. Prenatal exposure to Cd was associated with low birthweight and adverse effects on organ systems in infants and children (Chandravanshi et al., 2021); similar findings were reported with rodent models indicating fetal growth restriction (typically at high doses (e.g., 250 ppm), Wang et al., 2016; Kozlosky et al., 2023), metabolic syndrome and non-alcoholic fatty liver disease later in life (Jackson et al., 2020; Saedi et al., 2023; Riegl et al., 2023). Interestingly, maternal high-fructose intake during pregnancy and lactation also led to fetal growth restriction (Asghar et al., 2016; Liu et al., 2021) and induced metabolic syndrome in adult offspring (Saad et al., 2016; Koo et al., 2021). In the current study, Cd exposure did not alter maternal weight gain or PND1 body weight of the offspring consistently, perhaps due to the low doses of 0.5 and 5 ppm; however, both measures were significantly reduced by consumption of HFrD, likely reflecting placental insufficiency and fetal growth restriction as shown previously. These differing profiles of changes between Cd and HFrD were extended to postnatal growth and development. Similar persistent growth deficits were seen in the offspring whose dams were fed a HFrD, although the effects were more profound than those produced by Cd fed a control diet. In fact, the dietary effects masked those caused by 5 ppm Cd, such that the responses between HFrD+5Cd and HFrD+0Cd are indistinguishable, in either sex. Essential metals such as zinc, copper, iron, and manganese are known to play critical roles in early life growth and development. Previous studies have shown that exposure to Cd even at low doses disrupted concentrations of essential metals in various tissues (Young et al., 2019, Jackson et al., 2022). However, corresponding changes of these metals were not detected in the liver despite persistence of Cd in the present study, suggesting that disruption of essential metal homeostasis was unlikely a critical mechanism for the growth deficits associated with maternal exposure to Cd and HFrD. It is also interesting to note that catch-up growth typically observed growth retardation induced by malnutrition (Gat-Yablonski and Phillip, 2015) was not detected with either chemical or non-chemical stressors in this study.

Our exposure design produced a dose-dependent increase of Cd in maternal liver, in a range consistent with previous reports (Young et al., 2019; Jackson et al., 2022). Only a small fraction (~0.02%) of the maternal Cd level was detected in the fetal liver, suggesting limited placental transfer. Traces of circulating Cd were detected in the offspring blood at weaning. However, hepatic Cd increased by 10-fold compared to the fetal liver tissue, likely derived from lactational intake. Hepatic Cd declined approximately by only 55% from weaning age to young adulthood, within range of a previous half-life estimate (Taguchi and Suzuki 1981) and highlighted the extraordinary persistence of this metal in the offspring well into adulthood, despite only a small window of exposure from placental and lactational transfer. Maternal consumption of a HFrD did not alter Cd tissue levels in the offspring, suggesting that this combination of environmental stressors does not simply involve direct effects of metal absorption or bioaccumulation.

In summary, maternal exposure to Cd and HDFr alone altered development and had lasting effects on glucose and fat metabolism in the adult offspring. Although they interacted with each other for several specific endpoints, there were no cumulative effects that would reflect a global change indicative of synergism between the toxicant and dietary manipulation. An explanation for such observation is not readily available largely because the underlying mechanisms responsible for the developmental (at early ages) and latent (at adulthood) effects of cadmium and fructose remain largely undefined. It is tempting to speculate an involvement of epigenetic control or cellular reprogramming of functional responses for individual organ systems at early life stages in accordance with the "Developmental Origins of Health and Disease" concept, but further investigation to support and clarify such hypothesis will be required. Findings from our study are largely in line with others that addressed other combinations of environmental stressors, such as Snow et al. (2020) and Oshiro et al. (2022), in that the dietary and behavioral stresses tended to mask the chemical effects. They highlight one of the challenges with such investigation. Typically, in chemical mixture studies, the optimal doses chosen for each chemical would approximate ED₅₀, to allow a dynamic range of responses to discern interactions statistically. However, in chemical and non-chemical mixture studies, the non-chemical stressors such as dietary perturbations (high-fat, high-fructose, low-protein, etc.) and stresses (physical, behavioral, noise, etc.) invariably lack dose-response characterization, which may lead to an over- or underwhelming effect on a specific endpoint, compared to the chemical stressors. In addition, the mechanisms underlying the non-chemical stressors are largely non-specific or ill-defined, compared to those for the chemicals, which will hamper the detection and interpretation of any interactions between these two stressors. Future investigation should take these challenges into consideration.

Supplementary Material

Supplemental tables figures

Supplemental Fig. 1. Gene expression alterations of insulin pathways in female offspring liver. (A). Heatmap PND 3, 38, and 88 gene expression of targets for transcription factors Srebp1, ChREBP, and FoxO1 dervied from targeted RNA-sequencing data (* = downstream target of transcription factor). “id” denotes gene and probe ID. Color scale represent ratio of gene expression compared to the heatmap row median. There were no significant changes (adjust p-value (FDR) <0.1). n=3-8. (B). Quantitative RT-PCR analysis of select insulin pathway genes in PND 38, 88, and 150 female offspring liver samples. Bars indicated mean normalized gene expression values (normalized to the geomean of reference genes, Rplpo and Ppia). Error bars indicated standard error of the mean. * denoted significant difference from ConD+0Cd samples (two-tailed t-test, p-value<0.05). ± denoted significant different from HFrD+0Cd samples (two-tailed t-test, p-value<0.05). n=2-4.

Supplemental Fig. 2. Gene expression alterations of insulin pathways in male offspring liver. (A). Heatmap PND 3, 38, and 88 gene expression of targets for transcription factors Srebp1, ChREBP, and FoxO1 dervied from targeted RNA-sequencing data (* = downstream target of transcription factor). “id” denotes gene and probe ID. Color scale represent ratio of gene expression compared to the heatmap row median. There were no significant changes (adjust p-value (FDR) <0.1). n=3-8. (B). Quantitative RT-PCR analysis of select insulin pathway genes in PND 38, 88, and 150 male offspring liver samples. Bars indicated mean normalized gene expression values (normalized to the geomean of reference genes, Rplpo and Ppia). Error bars indicate standard error of the mean. * denotes significant difference from ConD+0Cd samples (two-tailed t-test, p-value<0.05). ± denoted significant different from HFrD+0Cd samples (two-tailed t-test, p-value<0.05). n=2-4.

Supplemental Fig. 3. Hypo- and hypermethylation summary and shared dmCpG for treatment groups. Specific CpGs are listed in Supp. Data 2 and 3. (A) Graph displaying number of significant hypo- (black bar) and hypermethylated (white bar) dmCpGs (p-adjusted (FDR) value <0.05) for female PND3 treatment group samples. (B) UpSet plot and select Venn diagrams (HFrD groups and Cd treatment groups) to show numbers of intersecting dmCpGs within the female PND3 treatment group samples. (C) Graph displaying number of significant hypo- (black bar) and hypermethylated (white bar) dmCpGs (p-adjusted (FDR) value <0.05) for male PND3 treatment group samples. (D) UpSet plot and select Venn diagrams (HFrD groups and Cd treatment groups) to show numbers of intersecting dmCpGs within the male PND3 treatment group samples. (E) Graph displaying number of significant hypo- (black bar) and hypermethylated (white bar) dmCpGs (p-adjusted (FDR) value <0.05) for female PND88 treatment group samples. (F) UpSet plot and select Venn diagrams (HFrD groups and Cd treatment groups) to show numbers of intersecting dmCpGs within the female PND88 treatment group samples. (G) Graph displaying number of significant hypo- (black bar) and hypermethylated (white bar) dmCpGs (p-adjusted (FDR) value <0.05) for male PND88 treatment group samples. (G) UpSet plot and select Venn diagrams (HFrD groups and Cd treatment groups) to show numbers of intersecting dmCpGs within the male PND88 treatment group samples.

Supplemental Fig. 4. Intersecting significant canonical pathways (−log(p-value>1.3) linked to dmCpGs in cis with genes closest to the TSS (<5000 bp) and displaying > 1% dM. (A) UpSet plots and select Venn diagrams for female PND3 and PND 88 treatment groups. (B) UpSet plots and select Venn diagrams for male PND3 and PND 88 treatment groups. (C) UpSet plots and select Venn diagrams demonstrating intersecting canonical pathways between male and female groups at age PND3. (D) UpSet plots and select Venn diagrams demonstrating intersecting canonical pathways between male and female groups at age PND88. (E) Intersecting canonical pathways of same sex treatment groups between PND3 and PND88.

Table S1. Maternal (postweaning) hepatic metal concentrations.

Table S2. Fetal (GD18) hepatic metal concentrations.

Table S3. Neonatal (PND22) hepatic metal concentrations.

Table S4. Adult (PND188) hepatic metal concentrations.

Supplemental data 1

Supplemental Data 1. Differentially expressed genes for all group comparisons with a table summary.

Supplemental data 2

Supplemental Data 2. Differentially methylated CpG sites for PND3 samples with table and graph summaries.

Supplemental data 3

Supplemental Data 3. Differentially methylated CpG sites for PND88 samples with table and graph summaries.

Supplemental data 4

Supplemental Data 4. IPA conical pathways linked to in cis genes to dmCpG sites for all treatment groups with table and graph summaries.