Abstract Background: Children born to diabetic or obese mothers have a higher risk of heart disease at birth and later in life. Using chromatin immunoprecipitation sequencing, we previously demonstrated that late-gestation diabetes, maternal high fat (HF) diet, and the combination causes distinct fuel-mediated epigenetic reprogramming of rat cardiac tissue during fetal cardiogenesis. The objective of the present study was to investigate the overall transcriptional signature of newborn offspring exposed to maternal diabetes and maternal H diet. Methods: Microarray gene expression profiling of hearts from diabetes exposed, HF diet exposed, and combination exposed newborn rats was compared to controls. Functional annotation, pathway and network analysis of differentially expressed genes were performed in combination exposed and control newborn rat hearts. Further downstream metabolic assessments included measurement of total and phosphorylated AKT2 and GSK3β, as well as quantification of glycolytic capacity by extracellular flux analysis and glycogen staining. Results: Transcriptional analysis identified significant fuel-mediated changes in offspring cardiac gene expression. Specifically, functional pathways analysis identified two key signaling cascades that were functionally prioritized in combination exposed offspring hearts: (1) downregulation of fibroblast growth factor (FGF) activated PI3K/AKT pathway and (2) upregulation of peroxisome proliferator-activated receptor gamma coactivator alpha (PGC1α) mitochondrial biogenesis signaling. Functional metabolic and histochemical assays supported these transcriptome changes, corroborating diabetes- and diet-induced cardiac transcriptome remodeling and cardiac metabolism in offspring. Conclusion: This study provides the first data accounting for the compounding effects of maternal hyperglycemia and hyperlipidemia on the developmental cardiac transcriptome, and elucidates nuanced and novel features of maternal diabetes and diet on regulation of heart health. Keywords: PI3K/Akt pathway, mitochondrial biogenesis, cardiovascular disease, maternal diabetes, high fat diet, functional genomics Introduction Cardiovascular disease (CVD) is the leading cause of death in the United States and by 2030 is projected to affect 40.5% of the US population ([35]1). It is critical to identify high-risk populations and implement targeted prevention in order to decrease this growing burden of disease. The pathogenesis of CVD is influenced over time by both hereditary and environmental factors. Mounting evidence shows that these processes may even begin before birth ([36]2, [37]3). Specifically, exposure to excess circulating maternal fuels during critical windows of fetal development increases the lifetime risk of CVD ([38]4–[39]7). Worldwide, there are 21.3 million live births annually to women with hyperglycemia during pregnancy ([40]8). Additionally, 35% of women are obese ([41]9), a co-morbidity that exacerbates this growing problem. Indeed, obese women develop gestational diabetes mellitus (GDM) at 4 times higher odds than non-obese women ([42]10). For these reasons, finding effective, targeted prevention for this growing and readily identifiable population would significantly lower the burden of heart disease over time. While it is increasingly recognized that infants born to diabetic or obese mothers have a higher incidence of heart disease at birth and later in life, prevention is hindered because the underlying mechanisms remain unknown. Importantly, infants exposed to diabetic pregnancy have a higher incidence of cardiac hypertrophy, diastolic dysfunction and impaired myocardial performance that begins in utero and cardiac pathology is similar regardless of diabetes type (pregestational or gestational diabetes) ([43]11). To improve overall outcomes of diabetic pregnancy, the current standard of care is routine screening and efforts to optimize maternal blood sugar levels before and during pregnancy ([44]12). While improved glycemic control, especially alongside enhanced and earlier screening for gestational diabetes has certainly decreased perinatal morbidities ([45]13–[46]16), infants continue to have a higher risk of heart disease even when born to mothers with good glycemic control ([47]11, [48]17–[49]20). Additionally, gestational diabetes predisposes infants to macrosomia and programmed cardiometabolic disease as adults, even if their mother was treated during pregnancy ([50]21–[51]28). This suggests additional under-recognized, targetable risk factors including lipids. Both maternal diabetes and high fat (HF) diet increase circulating lipids above the normal physiologic hyperlipidemia of pregnancy ([52]29). We developed a rat model to determine the individual and compounding effects of maternal diabetes and HF diet on cardiac outcomes in offspring. We found that a triad of maternal hyperglycemia, hyperlipidemia, and fetal hyperinsulinemia led to progressively worsening mitochondrial dysfunction, impaired cellular bioenergetics, and poorer cardiac function in newborn offspring hearts ([53]30, [54]31). We hypothesized that exposure to excess circulating fuels disrupts the in utero gene-environment interaction to program heart disease in the developing fetus, specifically through metabolic and mitochondrial mediated mechanisms. Our previous data using a well-characterized rat model and chromatin immunoprecipitation sequencing (ChIP-Seq) showed that maternal HF diet, especially alongside late-gestation hyperglycemia, caused distinct fuel-mediated epigenetic programming of cardiac metabolism during fetal cardiogenesis ([55]32). The present study used a cardiac systems biology approach that uncovered specific mechanisms underlying cardiometabolic pathology that may serve as potential targets for intervention. Materials and Methods All experimental methods were carried out in agreement with applicable international, national, and institutional guidelines for the care and use of animals (Animal Welfare Act and National Institutes of Health policies) and were approved by the Sanford Research Institutional Animal Care and Use Committee. Sprague Dawley rats (Harlan Laboratories, Indianapolis, IN) were used in all experiments and housed in Sanford Research's Animal Resource Center, a climate-controlled, light-dark cycled facility. Animal Model Characteristics Methods and model characteristics of the four animal groups used in this study have been detailed previously ([56]30, [57]33). Briefly, young adult female rats received either control or HF diet (Teklad, Harlan Laboratories, Madison, WI) for at least 28 days before mating and throughout pregnancy. Gestational day zero (GD0) was determined by a positive vaginal swab for spermatozoa. On GD14, after confirmation of pregnancy through ultrasound, dams received either citrate buffer (0.09 M) or 65 mg/kg of intraperitoneal streptozotocin (Sigma Life Sciences, St. Louis, MO) to induce diabetes in the last third of pregnancy. The model consistently exposes the developing fetus to a triad of maternal hyperglycemia, diet-induced hyperlipidemia, or the combination (glucolipotoxicity) which incites directly proportional levels of fetal hyperinsulinemia, respectively. The timing of diabetes induction in the last 1/3^rd of pregnancy is intentional to exclude confounding from hyperglycemia induced disruption in ovulation, placentation, and organogenesis. Additionally, timing corresponds with peak placental lipid accumulation, fetal pancreatic endocrine function, and more closely translates to pregnancies affected by gestational diabetes in the last trimester. Hyperglycemia was partially controlled with twice daily insulin treatments using regular insulin in the morning and insulin-glargine in the evening to keep non-fasting, whole blood glucose levels in a targeted non-fasting, pre-treatment range of 200–400 mg/dl. This target range was selected to assure an ample hyperglycemic exposure to diabetes-exposed pups, but also avoid significant ketoacidosis or dehydration. Dams that received streptozotocin but did not manifest a fasting blood glucose level ≥ 200 mg/dl were excluded from the study. While in our model all pregnant dams developed physiologic hyperlipidemia of pregnancy, diabetes increased serum triglycerides ~2 fold, diet ~3 fold, and the combination ~5 fold higher than controls leading to corresponding hyperinsulinemia in offspring ([58]30, [59]32–[60]34). Thus, our model of late gestation diabetes plus HF diet translates to poorly controlled gestational diabetes or Type 2 diabetes in the developing offspring with exposure to glucolipotoxicity and fetal hyperinsulinemia. Delivery (~ GD22) yielded postnatal day one (P1) offspring from four distinct groups: exposed to maternal diabetes alone, exposed to maternal HF diet alone, exposed to the combination of both maternal diabetes and HF diet, and control group ([61]Figure 1A). We used all four groups for gene expression analysis with further experiments streamlined to males from control and combination exposed groups. Figure 1. [62]Figure 1 [63]Open in a new tab Maternal diabetes and maternal high fat diet impart distinct cardiac transcriptome signatures in newborn rat offspring. (A) Schematic showing experimental model of exposed groups. Female rats had at least 28 days of either control or high-fat diet prior to breeding. Female diet continued throughout pregnancy. At gestational day (GD) 14 a single injection of citrate buffer (CB) or streptozotocin (STZ) was delivered to a subset of females with high-fat or control diet. At GD22 newborns were delivered and hearts were extracted for study from four exposed groups: controls (green), diabetes exposed (blue), high fat diet exposed (yellow), and combination exposed (orange). (B) Principal components analysis (PCA) plots depicting distinct transcriptome signature among all the exposed groups. Data is visualized here in two plots with different components plotted in each: components 1–3 are plotted on the left (C1 = 25.1%, C2 = 8.9%, and C3 = 4.4%); and components 3–4 are used in PCA plot on right (C2 = 8.9%, C3 = 4.4% and C4 = 5.1%). (C) Class prediction model used individual sample expression signatures to demonstrate with 100% accuracy the identification of each sample to their respective exposure group. Total RNA Isolation and Quantification Hearts were extracted from control, diabetes exposed, HF diet exposed, and combination exposed P1 offspring. Immediately after harvesting, samples were snap frozen in liquid nitrogen, and stored at −80°C until RNA extraction. Each experimental group consisted of a pool of male and female rat hearts. Total RNA was extracted from the whole heart with TRIzol and purified using an affinity resin column (Qiagen RNeasy Mini kit, Germantown, MD) according to manufacturer's protocol. Total RNA concentration was performed using spectrophotometric analysis measurement (abs-emission A260/A280) by NanoDrop 2000 UV-Vis Spectrophotometer (Thermo Fisher Scientific Inc. Waltham, MA). RNA sample integrity was assessed by electropherogram analysis on an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA) and only samples with RNA integrity number (RIN) scores > 8 were used for microarray labeling and hybridization. Microarray Hybridization and Data Analysis Microarray hybridization was performed by the Analytical Genomics Core Facility (Sanford Burnham Medical Discovery Institute, Lake Nona, Orlando, FL) using GeneChip Rat Gene 1.0 ST arrays (Affymetrix, Santa Clara, CA) according to manufacturer's protocol. Briefly, total isolated RNA (100 ng) from each sample was converted to cDNA utilizing SuperScript III First Strand Synthesis Supermix (Invitrogen, Life Technologies Corporation, Carlsbad, CA). Labeled complimentary RNA (cRNA), synthesized and amplified from the double-stranded cDNA template, was fragmented and hybridized onto GeneChip arrays. As a measure of quality control of the fragmented biotin-labeled cRNA, a prior hybridization of a test-3 array was performed and analyzed. GeneChip 3000 scanner (Affymetrix, Santa Clara, CA) was used to scan and quantitatively analyze images of hybridized GeneChip arrays. Intensity values for each probe cell in the arrays were calculated by GeneChip software. Data normalization and analysis was performed using GeneSpring GX 14.01 (Agilent Technologies, Palo Alto, CA). Probe cell intensities were used to calculate an average intensity for each set of probe pairs representing a gene. Quality control (QC) filtering was performed on the normalized intensity values and entities were clustered into four conditions: diabetes, HF diet, combination exposed and controls. Gene expression profiles for each condition were visualized as volcano plots to identify genes significantly upregulated or downregulated in each group. Functional Annotation Analysis Statistically significant gene expression profiles from each comparison were separated into upregulated and downregulated lists for functional annotation, Kyoto Encyclopedia of Genes and Genomes (KEGG), and Reactome pathway enrichment analysis using the Database for Annotation, Visualization and Integrated Discovery (DAVID) Bioinformatics Resources v6.8 ([64]https://david.ncifcrf.gov/, last access on 7/24/2020). To determine over representation or enrichment, the DAVID algorithm employs a modified Fisher's exact test that is incorporated into a score that reports relative priority. Entrez Gene identifiers of differentially expressed genes (structured into three lists: upregulated, downregulated and total genes changing) were submitted to DAVID for functional annotation analysis. RaGene-1_0-st-v1 array gene set was used as background and a high classification stringency was selected to maintain robust groups. Scores were reported for KEGG and Reactome pathways when applicable. Further functional pathway analysis of the upregulated, downregulated and total differentially expressed genes list was done through Reactome pathways database analysis tool (Reactome v69; [65]https://reactome.org/, last access on 07/24/2020). Networks Analysis and Gene Targets Prioritization Ingenuity pathway analysis (IPA; Qiagen, Germantown, MD) was performed to map functional gene networks defined by the quality-filtered transcriptome. Highest priority network scores were determined and all the gene relationships, i.e., functional interactions among genes, were exported from IPA for use in Cytoscape v3.7.1 ([66]https://cytoscape.org/) for further network analysis. Prioritization of gene targets was achieved through graph theory analysis tools within Cytoscape. Molecule Activity Predictor Analysis module in IPA was used to predict activation or inhibition of non-focused neighboring molecules, defined by IPA as molecules not included in the uploaded list of genes/molecules, within the functional network. This prediction analysis is based on the expression of the focused molecules, also known as statistically significant genes, within the network and predicts either upstream and/or downstream activities. Determination of Mitochondrial-Associated Genes MitoCarta 2.0 database (Broad Institute, Cambridge, MA) was used to determine the mitochondrial-associated genes in our list of statistically significant genes. MitoCarta 2.0 is an online repository of 1,158 mammalian (human and mouse) genes encoding proteins where their mitochondrial localization has been validated by various methods. We cross-referenced data from MitoCarta with data from the current study to identify mitochondrial-related genes in our gene expression dataset. The recently updated mouse MitoCarta 2.0 database used in the present analysis can be found at the following website: ([67]https://www.broadinstitute.org/files/shared/metabolism/mitocarta/m ouse.mitocarta.2.0.html). Quantitative RT PCR RNA was extracted from newborn (P1) rat hearts using the RNeasy Fibrous Tissue Mini kit (Qiagen, Germantown, MD) following manufacturer's protocol. RNA integrity was assessed by electropherograms using 2100 BioAnalyzer (Agilent Technologies, Santa Clara, CA) and demonstrated RIN scores of 9.2-10 (average = 9.8). RNA concentration from two groups, control and combination exposed, was measured by Epoch spectrophotometer (BioTek, Winooski, VT). Complementary DNA (cDNA) was synthesized using iScript cDNA Synthesis Kit and T100 Thermal Cycler (Bio-Rad, Hercules, California). Quantitative PCR (qPCR) was performed by TaqMan Gene Expression Assays approach with ABsolute Blue QPCR Mix (Thermo Fisher Scientific, Waltham, MA) using an ABI7500 qPCR system (Thermo Fisher Scientific, Waltham, MA). Beta-2-microglobulin (B2m) was used as the reference gene. Cardiac expression relative to B2m was compared between the control and combination exposed groups (n = 6 males/group). B2m, mitochondrial ribosomal protein L19 (Mrpl19), mitochondrial ribosomal protein S27 (Mrps27), peroxisome proliferator-activated receptor gamma coactivator 1 alpha (Ppargc1a) and fibroblast growth factor receptor 2 (Fgfr2) probe/primer sets were obtained from Thermo Fisher Scientific (Waltham, MA), and death associated protein 3 (Dap3) probe/primer set was obtained from Integrated DNA Technologies (Coralville, IA). Western Blot Analysis Newborn (P1) rat hearts from control and combination exposed males were homogenized and sonicated in RIPA buffer (50 mM Tris (pH 7.5), 150 mM NaCl, 1% Triton X, 0.5% deoxycholate, 0.1% sodium dodecyl sulfate) with cOmplete protease inhibitor cocktail (Roche, Indianapolis, IN) and phosphatase inhibitor cocktail (Sigma-Aldrich, St. Louis, MO). Protein concentrations were measured using the DC Protein Assay kit (Bio-Rad, Hercules, CA) and Cytation 3 Spectrophotometer (BioTek, Winooski, VT). Protein (20 μg) was prepared using Laemmli buffer and reducing agent then subjected to electrophoresis on 4–15% Criterion TGX Gels using Tris/Glycine/SDS buffer (Bio-Rad). MagicMark XP Western Protein Standard (Thermo Fisher Scientific, Waltham, MA) was used to identify band size. Gels were transferred to PVDF membranes using Trans-Blot Turbo Transfer System (Bio-Rad). Membranes were dried, rehydrated in methanol, washed in TBS, blocked in TBS containing 10% Clear Milk Blocking Buffer (Thermo Fisher Scientific) and then incubated overnight at 4°C with primary antibody. After washing in TBS-T, membranes were blocked again and incubated with secondary antibody for 1 h, using goat anti-rabbit IgG-HRP for reference proteins (Southern Biotech, Birmingham, AL) or donkey anti-rabbit IgG IRDye 680RD (LI-COR, Lincoln, NE) for proteins of interest. HRP exposed bands were visualized using Luminata Forte HRP Chemiluminescence Substrate (Thermo Fisher Scientific). Images were captured using a ChemiDoc MP Imaging System (Bio-Rad) and densitometric analysis was done using ImageJ. Optical density (OD) measurements from tested proteins were normalized to primary reference protein β-ACTIN. Voltage-dependent anion channel or porin (VDAC) and translocase of outer mitochondrial membrane 20 (TOMM20), both outer mitochondrial membrane proteins, were used as secondary references.