Excessive Glucocorticoids During Pregnancy Impair Fetal Brown Fat Development and Predispose Offspring to Metabolic Dysfunctions

Stress, including anxiety and depression, has become a serious health concern in women during pregnancy (1,2). National population-based studies reported that 15.2% of pregnant women have anxiety disorder and 37% and 17% of pregnant women have depressive symptoms and severe depressive disorders, respectively (1,2). Glucocorticoids (GC) are major hormones responsive to stress, and the maternal stress during pregnancy significantly increases fetal exposure to GC (3,4). The activity of 11β-hydroxysteroid dehydrogenase, an enzyme limiting maternal GC across the placental barrier, was also reduced by maternal stress (5), further exaggerating excessive exposure of fetuses to GC (6,7).

Programming effects of hyper–intrauterine GC increase susceptibility of offspring to not only neurophysiological and psychological disorders but also the risk of childhood and adult obesity (8–11). Overweight and obesity, due to excessive accumulation of white adipose tissue, are closely associated with type 2 diabetes, cardiovascular diseases, and several cancers (12,13). Different from white adipose tissue, brown adipose tissue (BAT) dissipates energy through thermogenic activity of uncoupling protein 1 (UCP-1), preventing obesity and type 2 diabetes (12–14). Mitochondria are key organelles in maintaining BAT thermogenic functions and energy expenditure, and UCP-1 is anchored in inner mitochondrial membrane (12–14). Peroxisome proliferator–activated receptor γ coactivator-1α (Ppargc1a) (protein PGC-1a) is a master regulator of mitochondrial biogenesis (15,16), and its transcriptional inhibition decreases mitochondrial density and UCP-1 protein, contributing to cold sensitivity and obesity (17). Recently, hyper–DNA methylations have been identified in the Ppargc1a promoter in pancreas (18) and skeletal muscle (19) of patients with type 2 diabetes. Adiposity, glucose tolerance, and insulin resistance induce epigenetic modulations, which reduce Ppargc1a expression and mitochondrial biogenesis across tissues (18,19). Currently, little information is available on changes and functional roles of Ppargc1a promoter DNA methylation in BAT function of obese and diabetic mice or humans with obesity and diabetes. Because of the BAT importance in energy expenditure and glucose homeostasis (12,13), revealing the link between offspring BAT phenotypes and DNA methylation of Ppargc1a promoter in responding to maternal hyper-GC during pregnancy is of key interest.

Previous evidence suggests that excessive GC in circulation increase the risks of obesity in adult mice (20). Similarly, patients with Cushing syndrome, due to chronic hypercortisolism, are also characterized with insulin resistance and central obesity (20). Hyper-GC were reported to inhibit BAT thermogenesis, but a recent study reversibly showed that excessive GC activate human BAT thermogenic activity (21). The inconsistency in reports underscores the knowledge deficiency in the direct GC exposure on adult BAT thermogenesis. Moreover, the effects of maternal hyper-GC exposure during pregnancy on fetal BAT development and long-term energy expenditure and thermogenesis in adult offspring remain unexamined. In this study, pregnant mice were administrated with synthetic GC dexamethasone (DEX) during the last trimester; synthetic GC can readily cross the placenta, imitating maternal stress–induced fetal hyper-GC exposure as previously reported (22–24). In addition, DEX injection is used for >70% of women at risk of preterm delivering (25). The development of BAT is initiated around mid-gestation and followed by rapid brown adipogenic differentiation in the last trimester (12,13). Targeting this most active stage of brown adipogenesis (12–14), we hypothesized that the intrauterine hyper-GC exposure has substantial impacts on fetal BAT development, leading to changes of postnatal developmental trajectory and functions. We further hypothesized that maternal hyper-GC exposure epigenetically inhibits Ppargc1a expression, persistently reducing its expression and BAT energy expenditure and thermogenesis in offspring.

Wild-type female C57BL/6J mice (The Jackson Laboratory, Bar Harbor, ME) at 4 months of age were mated with wild-type male mice fed a regular diet (10% energy from fat, D12450H; Research Diets, New Brunswick, NJ). Successful mating was determined by the presence of a copulation plug in the vagina and marked as embryonic day 0.5 (E0.5). At E14.5, pregnant mice were randomly separated into two groups intraperitoneally administrated with PBS (as control) or 0.1 mg/kg body wt DEX (D4902; Sigma-Aldrich) daily until E20.5. After birth, the litter size was standardized to six and the maternal mice together with their offspring were housed at room temperature (22°C) until weaning (postnatal day 21 [P21]). After weaning, male offspring in control and DEX groups were further randomly assigned to a chow (10% energy from fat) or obesogenic (high-fat diet [HFD]) (60% energy from fat; D12492) diet until 4 months of age. At 4 months old, a portion of offspring were acclimated to thermoneutrality (30°C) for 4 weeks to examine BAT-independent thermogenic changes due to maternal DEX exposure (26–28). At P0, P21, and 4 months, offspring were euthanized by CO₂ and brown fat (BAT) was collected for analyses. Due to the small size of neonatal BAT, neonates from the same litter were pooled. Each dam (each pregnancy) was treated as an experimental unit. All animal studies were conducted in a Washington State University animal facility accredited by the Association for Assessment and Accreditation of Laboratory Animal Care. Animal protocol was approved by the Washington State University Institute of Animal Care and Use Committee.

FACS was performed as previously described (14). BAT was digested in 0.2% collagenase type II (285 units/mg) in DMEM-free FBS medium for 30 min at 37°C. Tissues debris was filtered through 40-µm strainers, and the filter solution was centrifuged at 400g for 5 min. After centrifugation, BAT stromal vascular fractions were blocked in 1% BSA for 15 min on ice. Cells were washed by PBS and incubated in conjugated CD45 (phycoerythrin/Cy7; BioLegend) and PDGFRa (allophycocyanin; BioLegend) in the dark at 4°C for 1 h. Cells were washed with PBS and sorted in a SY3200 Cell Sorter (Sony Biotechnology, San Jose, CA). Flow cytometry data were analyzed using FlowJo software (Treestar). Gates were established using fluorescence minus controls.

DNA Methylation Immunoprecipitation (MeDIP) analysis was performed as previously described (14). BAT was digested in protease K solution. Total DNA was isolated from solution containing Tris-phenol, chloroform, and isoamyl alcohol. Isolated DNA was dissolved into Tris-EDTA buffer followed by sonication (30% power, 10 s on/off for 5 min) on ice. Average 300–500 base pair DNA fragment size was obtained and verified by electrophoresis in 2% agarose gel. Denatured DNA (2 mg) containing RNase A was incubated with 2 μg antibody against 5-methylcytosine (5mC) (Zymo Research, Irvine, CA) or anti-mouse IgG (Thermo Fisher Scientific) overnight at 4°C. The DNA-antibody solution was further incubated with precleaned protein A magnetic beads (Cell Signaling Technology) at 4°C for 1 h. Washed beads were incubated in digestion buffer to recover precipitated DNA at 65°C for 3 h. Recovered DNA was quantified by quantitative PCR (qPCR). Primers for MeDIP qPCR are listed in Supplementary Table 1.

Genomic DNA was converted by bisulfite using EZ DNA Methylation-Direct Kit (D5021; Zymo Research). Converted genomic DNA was used as a template to amplify the target sequence. GC-purified biotinylated primers were targeted to the Ppargc-1a proximal promoter (19,29), including the following sequence: 5′attttttttttcctctctctctaagcgttacttcactgaggcagagggccccttggagtgacg3′. Pyrosequencing and data analyses were performed by EpigenDx (Hokinton, MA).

Total protein was isolated in lysis buffer as previously described (14). Lysate was precleaned with protein A magnetic beads (Cell Signaling Technology) at 4°C for 1 h. GC receptor (GR) primary antibody (2 μg) (Cell Signaling Technology) was added to lysate containing 500 μg total proteins, and the mixture was gently rotated overnight at 4°C. After that, precleaned protein A magnetic beads were mixed into lysate and rotated for 1 h at room temperature. Samples were followed with 3× cold PBS washing, and proteins on beads were eluted for immunoblotting analyses of GR, DNMT3b (Santa Cruz Biotechnology), DNMT3a (Cell Signaling Technology) and DNMT1 (Cell Signaling Technology).

Chromatin immunoprecipitation assay (ChIP) analyses were performed as previously described (14). Briefly, protein and DNA were cross-linked in 1% formaldehyde followed by addition of 125 mmol/L glycine at room temperature. Grounded tissues were suspended in cold ChIP lysis buffer containing protease inhibitor (Thermo Fisher Scientific). Samples were sonicated to shear DNA fragments to average length 300–500 base pairs. DNA (30 μg) was used as inputs and the same amount of DNA was incubated with 3 μg antibodies against GR (Cell Signaling Technology), DNMT3b (Santa Cruz Biotechnology), or negative control IgG (Thermo Fisher Scientific) at 4°C overnight. Precleaned protein A magnetic beads (40 μL) (Cell Signaling Technology) were further added into the solution and rotated at 4°C for 3 h. Precipitated DNA was recovered from washing beads in low salts, high salts, and LiCl solutions. After washing with elution buffer, recovered supernatant was treated with RNase A (10 mg/mL) and proteinase K (20 mg/mL). Isolated DNA was dissolved into Tris-EDTA buffer and quantified by qPCR. Primers in ChIP analyses are listed in Supplementary Table 1.

TRIzol reagent (Invitrogen) was used to extract total RNA according to the manufacturer’s instruction (14). RNA expression was standardized to 18s rRNA. Primer sequences are listed in Supplementary Table 1.

Western blot was performed to analyze target proteins using UCP-1 (Cell Signaling Technology), IRS1 (Cell Signaling Technology), IRS1-T612 (Thermo Fisher Scientific), AKT (Cell Signaling Technology), AKT-S473 (Cell Signaling Technology), β-tubulin (Cell Signaling Technology), PGC-1a (Proteintech), PRDM16 (Thermo Fisher Scientific), VDAC (Cell Signaling Technology), and Cytochrome c (Cell Signaling Technology) antibodies (14).

Brown adipogenic induction was conducted as previously described (14). PDGFRa + brown progenitors were isolated from BAT using a manual magnetic cell separation system (Miltenyi Biotec, Bergisch Gladbach, Germany). Isolated brown progenitors were cultured in DMEM with 10% FBS. After confluence, brown adipogenic induction was induced in medium containing 0.1 μg/mL insulin, 0.5 mmol/L isobutylmethylxanthine, 1 μmol/L DEX, 125 μmol/L indomethacin, and 1 nmol/L T3 for 3 days followed by 0.1 μg/mL insulin and 1 nmol/L T3 for 2 days (14).

Transmitting electronic microscopy (TEM) was performed as previously described (14). Imaging was recorded in TEM FEI Technai G2 20 Twin (200 kv LaB6).

Immunostaining was performed to detect mitochondria using MitoSpy Green FM (BioLegend) (14). EVOS XL Core Imaging System was used to obtain images. Cellular lipid droplets were stained using 60% of oil red O (14). The absorbance of oil red O was measured at 492 nm in a Synergy H1 microplate reader (BioTek Instruments, Winooski, VT).

During the analyses, mice were caged individually in the metabolic chamber with ad libitum water and their respective diets. Comprehensive Lab Animal Monitoring System (CLAMS) (Columbus Instruments) was used for an indirect open-circuit calorimetry measurement. Data were normalized according to ANCOVA energy expenditure guidelines (30,31).

A FLIR E6 infrared thermal camera (FLIR Systems, Wilsonville, OR) was used to measure surface body temperature (14). After mice were removed from nests, the surface temperature was immediately obtained using a double-blind protocol. During capture of thermal images, animal behavior and scanning distance were controlled to avoid artificial effects. For reduction of variances, rectal body temperature was also measured using a highly precise electronic thermometer (TH-5 Thermalert Clinical Monitoring Thermometer; Physitemp Instruments).

After 6 h fasting with ad libitum water, mice were intraperitoneally injected with 2 g/kg body wt d-glucose. Glucose concentration was measured in blood obtained from tail tip at 0, 15, 30, 60, and 120 min after administration using a Contour glucose monitor (Bayer).

After 6 h of fasting, insulin was measured using a mouse insulin ELISA (ALPCO). Insulin resistance was calculated according to the following formula: HOMA of insulin resistance = (fasting blood glucose mg/dL × 0.055) × (fasting insulin μU/mL)/22.5.

For assessment of hypothalamic-pituitary-adrenal axis activity, offspring at 4 months of age were intraperitoneally administrated with DEX (20 μg/kg) and blood was collected after 2 h. Corticosterone concentration in serum was measured with an ELISA kit (ADI-900; Enzo Life Sciences, Farmingdale, NY) according to the manufacturer’s instructions.

Cells were incubated with 6 mg/mL 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide (MTT) (Sigma-Aldrich) for 4 h. Formazan was dissolved in 100 μL DMSO, and the absorbance was measured at 540 nm using a microplate reader.

Cells were incubated in 1 μmol/L BrdU (Sigma-Aldrich) for 24 h followed by BrdU (Cell Signaling Technology) immunostaining.

Results were presented as means ± SEM. All statistical analyses were performed using SAS, version 9.4 (SAS Institute, Cary, NC). Unpaired two-tail Student t test was applied for two-group comparison. One-way ANOVA was used in comparisons for multiple groups. Each dam (pregnancy) was used as an individual replicate. Metabolic chamber data were analyzed by National Institute of Diabetes and Digestive and Kidney Diseases minimal-change chronic pancreatitis ANCOVA multiple linear regression models (30,31). Significant differences are shown in figures as P < 0.05, P < 0.01, P < 0.001, and P < 0.0001.

The data sets generated and analyzed during this study are available from the corresponding author upon reasonable request.

Pregnant mice were intraperitoneally administrated with DEX during the last trimester of pregnancy, and DEX-treated maternal mice displayed similar food intake, body weight, and blood glucose compared with placebo mice at room temperature (22°C) (Supplementary Fig. 1A–C). Regardless of feeding with chow or HFD, DEX offspring at 4 months of age had similar calorie intake and body mass (Fig. 1A and Supplementary Fig. 1D) but showed higher glucose tolerance, fasting insulinemia, and reduced insulin sensitivity compared with control offspring (Fig. 1B–D and Supplementary Fig. 1E). Metabolically, the use of ANCOVA with total body mass or lean mass as a covariate revealed that DEX offspring robustly exhibited lower O₂ consumption, CO₂ respiration, and energy expenditure (30,31) (Fig. 1E–G and Supplementary Fig. 1F–N). The difference of energy expenditure between control and DEX offspring was exaggerated when body fat effects were accounted for (Fig. 1E–G and Supplementary Fig. 1J–N), showing presence of metabolic dysfunctions in thermogenic fat. Consistently, DEX offspring had lower BAT mass (Fig. 1H) and thermogenic protein UCP-1 content (Fig. 1I and J), indicating blocked thermogenic function. In agreement with the hyperinsulinemia, BAT of DEX offspring had impaired insulin signaling as indicated by the reduced tyrosine phosphorylation of IRS-1 (T612) and serine phosphorylation of AKT (S473) (Fig. 1I, K, and L). Such reduction was further confirmed following insulin stimulation (Supplementary Fig. 2A). Maternal GC exposure also decreased expression of genes involved in fatty acid oxidization and BAT thermogenesis in offspring BAT (Fig. 1M). The reduced energy expenditure and fatty acid oxidation in BAT contributed to lipid accumulation in gonadal and inguinal fat and increased total fat mass in DEX offspring regardless of chow or HFD feeding (Fig. 1N and O and Supplementary Fig. 2B). Moreover, muscle mass tended to be lower in DEX offspring, which was exacerbated by HFD feeding (Supplementary Fig. 2C and D).

To further test the mediatory roles of BAT thermogenesis in energy metabolic dysfunctions of DEX offspring, we acclimated offspring to thermoneutrality (30°C) to block BAT thermogenic activation (26–28). Consistent with previous reports (26–28), thermoneutrality blunted BAT activation as indicated by similar UCP-1 protein content between control and DEX offspring (Supplementary Fig. 3A). No significant difference was observed for calorie intake and body mass (Supplementary Fig. 3B and C), but the difference in glucose, insulin sensitivity, and energy expenditure between control and DEX offspring was ameliorated under thermoneutrality (Supplementary Fig. 3D–L), showing that BAT thermogenic inactivation mediates metabolic dysfunctions in DEX offspring observed at room temperature.

After offspring were subjected to cold challenge (4°C) for 2 days, the difference of BAT mass became even greater between control and DEX offspring receiving either chow or HFD (Fig. 2A), showing the reduced BAT thermogenic plasticity. DEX offspring also had lower interscapular surface and core body temperatures (Fig. 2B–D), which were in agreement with inactivation of brown adipogenic and thermogenic proteins, including PGC-1a, PRDM16, and UCP-1 (Fig. 2E and Supplementary Fig. 4A–C). Accordingly, the mitochondrial density was also reduced due to maternal DEX exposure (Fig. 2F). Taken together, these data showed that maternal hyper-GC impaired offspring BAT energy expenditure and thermogenesis, contributing to metabolic dysfunctions and cold intolerance in adult offspring.

Due to the importance of mitochondria in mediating thermogenesis and energy expenditure (15,16), we further analyzed mitochondrial density and structure in offspring BAT and found that brown adipocytes had larger sizes of lipid droplets but reduced mitochondria number and cristae length and density in DEX offspring fed with both chow and HFD (Fig. 3A–E). Consistent with the reduced mitochondria number, BAT of DEX offspring also showed substantial decreases of mtDNA content (Fig. 3F) and mitochondrial biomass indicators VDAC and CYTO-C (Fig. 3G and Supplementary Fig. 4D and E). PGC-1a centrally regulates mitochondrial biogenesis and UCP-1 thermogenesis, which was also profoundly reduced in DEX offspring BAT (Fig. 3G and H and Supplementary Fig. 4F), in alignment with the reduced expression of mitochondrial biogenic nuclear genes (Fig. 3I). Moreover, mitochondrial encoding genes involved in oxidative phosphorylation and respiration chain were also reduced in DEX offspring (Fig. 3I), showing that maternal GC exposure impaired Ppargc1a expression and mitochondrial biogenesis in offspring BAT.

Because brown progenitors are responsible for BAT plasticity, we further isolated BAT stromal vascular cells and sorted for brown progenitors using well-characterized surface markers Lin⁻ CD45⁻/PDGFRa⁺ by FACS (32,33). BAT of DEX offspring had profoundly reduced brown progenitor density (Fig. 4A and B). Brown progenitors of DEX offspring also had the lower Ppargc1a expression (Fig. 4C and D) and mitochondrial biomass as indicated by CYTO-C, VDAC, and mtDNA contents (Fig. 4C and E–G). Immunostaining using mitochondrial tracker robustly displayed lower mitochondrial density (Fig. 4H and I), verifying that the brown progenitors of DEX offspring had reduced capacity of mitochondrial biogenesis; progenitor plasticity is required for maintaining BAT thermogenic activation in responses to cold and other stimuli (33–35). In addition, isolated brown progenitors in DEX offspring BAT had reduced NAD(P)H-dependent oxidoreductase enzyme activity in MTT assay (Fig. 4J), and less incorporation of thymidine analog BrdU into nuclear DNA (Fig. 4K and L), suggesting lower proliferative capacity. Following a standard brown adipogenic induction, brown progenitors of DEX offspring also displayed a reduced capacity to differentiate into brown adipocytes based on oil red O staining (Fig. 4M and N). Taken together, our data showed that maternal GC exposure decreased Ppargc1a expression, mitochondrial biogenesis, and brown adipogenesis of progenitors in offspring BAT.

Ppargc1a is a master regulator of mitochondrial biogenesis, which was inhibited in BAT and brown progenitors in DEX offspring. We hypothesized that such inhibition was due to DNA methylation in the Ppargc1a promoter, which is known to regulate its expression (18,19). Indeed, we noticed the proximal promoter of Ppargc1a harbors CpG and noncanonical CpG methylation sites (Fig. 5A and Supplementary Fig. 5A). To analyze, we used MeDIP analysis and found that 5mC was higher in the Ppargc1a promoter in both BAT and brown progenitors of DEX offspring at 4 months of age compared with control offspring (Fig. 5B and C). Intriguingly, the elevated DNA methylation was also observed in BAT and brown progenitors of DEX offspring at weaning (Supplementary Fig. 5B and C) and further traced back to neonates at P0 (Fig. 5D and E), showing that higher DNA methylation in the Ppargc1a promoter of DEX offspring BAT had a fetal origin. However, DNA hypermethylations were not observed in the promoters of unrelated genes in BAT and brown progenitors of DEX neonates, including Nrf1, Pax3, and Zfp423 (Supplementary Fig. 5D and E), suggesting locus-specific DNA methylation in gene promoters.

To identify methylation sites in the Ppargc1a proximal promoter, bisulfite pyrosequencing was performed (19,29) (Fig. 5F). BAT of DEX offspring did not only display higher methylation of CpG sites; intriguingly, hypermethylation was also observed in non-CpG sites (Fig. 5G and H). Similar patterns were also revealed from brown progenitors in DEX offspring (Fig. 5G and H), further highlighting that non-CpG methylation was likely involved in the regulation of Ppargc-1a transcriptional activity.

In alignment with the higher DNA methylation in the Ppargc1a promoter, Ppargc1a expression was also profoundly lower in BAT and brown progenitors of DEX neonates (Fig. 6A–C). In agreement, BAT of DEX neonates had elevated GR expression but reduced contents of UCP-1, VDAC, and CYTO-C, as well as mitochondria number and cristae density (Fig. 6B–D and Supplementary Fig. 6A and B). Consistent with GR activation in BAT, neonates displayed higher concentration of corticosterone, which was further confirmed in offspring at 4 months of age as assessed by GC suppression test (Supplementary Fig. 6C). Considering the critical role of neonatal BAT in maintaining body temperature, the reduced BAT thermogenesis explained the hypothermia in DEX neonates (Fig. 6E), which correlated with reduced survival of offspring in the first postnatal week (Supplementary Fig. 6D). Brown progenitors isolated from DEX neonates also had lower contents of mitochondrial biomass protein markers, mitochondrial density, and mtDNA (Fig. 6C and F and Supplementary Fig. 6E). Consistently, the population of brown progenitors was also reduced in DEX neonatal BAT (Fig. 6G and H). Furthermore, the brown progenitors had reduced capacity undergoing brown adipogenic differentiation in vitro (Supplementary Fig. 6F), supporting the suppressed BAT mass in DEX neonates (Fig. 6I). Similar data regarding to the Ppargc1a transcriptional inactivation and reduced BAT mitochondrial biogenesis and thermogenesis were also observed in DEX offspring at weaning (Supplementary Fig. 7A–H). Collectively, the DNA methylation of Ppargc1a promoter was inversely correlated with transcriptions in both BAT and brown progenitors in offspring (Fig. 6J and K). These data support the notion that the reduced Ppargc1a expression resulted from the abnormal gaining of DNA methylation in the Ppargc1a promoter during fetal BAT development of DEX mothers.

To identify the underlying mechanisms leading to the locus-specific increase in DNA methylation of Ppargc1a promoter, we analyzed DNA binding elements and identified a potential GR-responsive binding element as previously described (36,37) (Supplementary Fig. 5A). Using GR ChIP analyses, we observed that the maternal DEX exposure increased GR binding to the Ppargc1a promoter in both neonatal BAT and brown progenitors (Fig. 7A). In GR coimmunoprecipitation analyses, DNMT3b, but neither DNMT3a nor DNMT1, physically interacted with GR (Fig. 7B and C). Accordingly, maternal GC exposure enhanced the interaction between GR and DNMT3b (Fig. 7B and C), suggesting that GR might increase the recruitment of DNMT3b into the Ppargc1a promoter for de novo DNA methylation in BAT and brown progenitors. Using DNMT3b ChIP analyses, we confirmed that the DEX administration increased DNMT3b binding to the Ppargc1a promoter in both BAT and brown progenitors (Fig. 7D). Increased GR interaction with DNMT3b was also observed in BAT of DEX offspring at 4 months of age (Supplementary Fig. 7I). Taken together, these data showed that Ppargc1a, as a GR target gene, was subjected to the enhanced DNA methylation in its promoter during fetal BAT development, potentially involving the recruitment of the GR/DNMT3b complex to the Ppargc1a promoter, which persistently suppresses its expression and impairs mitochondrial biogenesis and fetal BAT development, programming postnatal offspring to metabolic dysfunctions.

The intrauterine environment has profound effects on fetal development and birth outcome, which can program metabolic health of offspring, including obesity and type 2 diabetes (4,8–11,38). Antenatal stress, depression, and preterm delivery are prevalent in women during pregnancy, leading to fetal exposure to substantial levels of GC in the uterus (6,7). Maternal stress and antenatal GC administration are associated with outcomes of low birth weight (39–42), which is also a strong indicator for obesity and type 2 diabetes risks (43), but mechanisms of maternal hyper-GC linking to offspring obesity remain poorly understood. The present findings represent, to our best knowledge, the first report that the maternal GC exposure during pregnancy impaired offspring BAT thermogenesis, which was correlated with glucose and insulin resistance, obesity, and cold sensitivity. The finding highlighted the importance of maintaining maternal psychological health and discretion using exogenous GC during pregnancy in order to protect offspring from metabolic dysfunctions (3). Mechanistically, our study also disclosed that the maternal GC exposure persistently inhibited Ppargc1a expression in BAT and brown progenitors, leading to impaired mitochondrial biogenesis (15,16). We also found that DEX elicited excessive DNA methylation in both canonical and noncanonical CpG sites in the Ppargc1a proximal promoter (19). Furthermore, the DNMT3b and GR complex was identified to mediate the accretion of DNA methylation in the Ppargc1a promoter during fetal BAT development. The discovery provides a potential target to reduce the risks of obesity and metabolic dysfunctions in offspring born from mothers who suffered from hyper-GC conditions.

In this study, we administered pregnant mice with 100 μg/kg DEX in the third trimester, which was in the range of GC concentrations in maternal circulation under stress or during preterm delivering as previously described (44,45). This dose was >10-fold lower than the DEX administration in nonpregnant mice for anti-inflammatory therapy or pathological GC secretion in patients with an adrenal tumor such as in Cushing syndrome (46). The low dose and a relatively short duration of treatment might explain the lack of difference in food intake and body weight of maternal mice between treatments. However, we observed a profound GR activation in both BAT and brown progenitors of DEX neonates, suggesting that the low dose is sufficient to induce corresponding changes in fetuses. Therapeutic DEX administration to mothers with preterm delivery improves lung maturation of fetuses, but it also increases risks of low birth weight (40,41). As for programming effects, under HFD, adult DEX offspring displayed insulin resistance, impaired glucose and fatty acid oxidization, and adiposity (8,14). Notably, DEX offspring fed with a normal calorie diet also displayed metabolic dysfunctions and adiposity, showing the negative effects of maternal hyper-GC on offspring health regardless of postnatal diets. For validation of the mediatory roles of BAT thermogenesis in reducing energy expenditure of DEX offspring at room temperature, adult offspring were also acclimated to thermoneutrality and exposed to acute cold, respectively (26–28). Under thermoneutrality, the reduction of energy expenditure in DEX offspring was substantially blunted. Meanwhile, DEX adult offspring, regardless of postweaning diet, displayed severe cold sensitivity and reduced brown thermogenesis under acute cold challenge, demonstrating that maternal hyper-GC exposure significantly impaired offspring BAT thermogenic activity, which mediated reduced energy expenditure and metabolic dysfunctions of their offspring. Of note, physical activity was not directly measured in offspring; thus, the potential effects of physical activity on offspring energy expenditure could not be excluded. On the other hand, we observed lower metabolic rates of DEX offspring during both nighttime (physically active) and daytime (inactive), suggesting physical activity was not a major factor contributing to metabolic dysfunction in DEX offspring.

Through regulating mitochondrial biogenesis, PGC-1a transduces a variety of cellular functions (15,16), facilitating differentiation of progenitors and stem cells (34,35). Mitochondrial biogenesis activates energy expenditure and heat production of brown adipocytes (12). Sufficient energy provided from tricarboxylic cycle is also required for the biosynthetic platform during proliferation and differentiation of progenitors (47). In the absence or limitation of mitochondrial biogenesis, the proliferation and differentiation of progenitors are not sustained, suppressing tissue growth and regeneration (47). Maternal GC exposure reduced Ppargc1a expression in BAT of adult offspring, impairing mitochondrial biogenesis and thermogenesis. The detrimental effects were further observed in the sorted brown progenitors with use of well-characterized markers (32,33), explaining impaired BAT plasticity under cold stimulation. Brown progenitors are essential for BAT plasticity in response to environmental stimuli, including hormones and cold (13,33). Intriguingly, the impaired mitochondrial biogenesis and Ppargc1a transcription were further traced back to neonates. The neonatal offspring born from DEX dams displayed hypothermia and reduced BAT mass and thermogenic activity. The BAT thermogenic inactivation increases postnatal death in responsive to severe cold (12). Collectively, we revealed that maternal hyper-GC impaired fetal BAT development, which had persistent effects in impairing offspring BAT thermogenic function and plasticity.

Epigenetic regulation during fetal development exerts long-term programming impacts on gene expression, which may shape tissue developmental trajectory and lifelong health of offspring (4). We identified that BAT and brown progenitors of DEX offspring had higher DNA methylation in the Ppargc1a promoter. In the meantime, the hypermethylation was also observed in non-CpG sites, supporting a previous report that noncanonical CpG methylation may play a critical role in pathogenic Ppargc1a inactivation in patients with diabetes (19). The DNA methylations in the Ppargc1a promoter durably suppress Ppargc1a expression and mitochondrial biogenesis (18,19). Excessive DNA methylation in the Ppargc1a promoter was also reported in the muscle of patients with diabetes, causing impaired mitochondrial biogenesis and muscle atrophy (19). In this study, we also noticed the hyper–DNA methylation of Ppargc1a promoter in the muscle of DEX offspring (data not shown). As the energy expenditures in muscle and BAT are highly dependent on mitochondrial biogenesis, the Ppargc1a promoter methylation contributes to the pathogenesis of obesity and type 2 diabetes (18,19). Maternal hyper-GC exposure also programmed offspring hypothalamic-pituitary-adrenal axis as indicated by high cortisol in circulation and in response to GR activation, which might contribute to gaining of DNA methylation from elevated interaction between GR and DNMT3b in offspring BAT and brown progenitors (22). A similar interaction was also reported in neurons (48,49). Taken together, results indicate that the GC exposure enhanced the GR binding to the regulatory region of the Ppargc1a promoter, likely serving as a dock for DNMT3b recruitment, locally driving the de novo DNA methylation in the transcriptional regulatory regions of corresponding genes (48,49). These findings deepen our mechanistic understanding of maternal hyper-GC exposure in reducing energy expenditure and increasing susceptibility to metabolic dysfunctions of offspring (19). Of note, in this study, we only examined male offspring. Because the negative effects of excessive GC exposure during pregnancy have been well-documented to predispose offspring to obesity and metabolic dysfunctions in both sexes of mice and humans (8–11), biological changes identified in male offspring should be applicable to females.

In conclusion, we discovered that the maternal hyper-GC exposure during pregnancy substantially inhibits fetal BAT development, which was associated with persistent elevation of DNA methylation in the Ppargc1a promoter and increased risks of obesity and metabolic dysfunctions in offspring. These findings underscored that the prevalent maternal physiological stress or pharmaceutic GC administration during pregnancy impaired metabolic health of offspring, which may contribute to the obesity epidemics in the modern society.

Acknowledgments. The authors thank Franceschi Microscopy & Imaging Center for help with electronic microscopy imaging.

Funding. This work was funded by the National Institutes of Health (NIH R01-HD067449).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. Y.-T.C. and M.D. developed the concept, designed experiments, and interpreted the data. Y.-T.C., Y.H., J.S.S., X.-D.L., and J.M.d.A. conducted experiments and collected data. Y.-T.C., Y.H., and Q.-Y.Y. analyzed data. Y.-T.C. and M.D. prepared the manuscript. Y.-T.C., M.-J.Z., and M.D. made revisions to the manuscript. All authors approved the final content. Y.-T.C. and M.D. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.