Prenatal High Estradiol Exposure Induces Sex-Specific and Dietarily Reversible Insulin Resistance Through Decreased Hypothalamic INSR

Abstract

An adverse intrauterine environment may induce adult disease in offspring, but the mechanisms are not well understood. It is reported that fresh embryo transfer (ET) in assisted reproductive technology leads to high maternal estradiol (E₂), and prenatal high E₂ exposure increases the risk of organ disorders in later life. We found that male newborns and children of fresh ET showed elevated fasting insulin and homeostasis model of assessment for insulin resistance index (HOMA-IR) scores. Male mice with high prenatal estradiol exposure (HE) grew heavier than control mice and developed insulin resistance; they also showed increased food intake, with increased orexigenic hypothalamic neuropeptide Y (NPY) expression. The hypothalamic insulin receptor (INSR) was decreased in male HE mice, associated with elevated promoter methylation. Chronic food restriction (FR) in HE mice reversed insulin resistance and rescued hypothalamic INSR expression by correcting the elevated Insr promoter methylation. Our findings suggest that prenatal exposure to high E₂ may induce sex-specific metabolic disorders in later life through epigenetic programming of hypothalamic Insr promoter, and dietary intervention may reverse insulin resistance by remodeling its methylation pattern.

Intrauterine development is a crucial process in early life that can influence future health. The developmental origin of health and disease theory raised by British epidemiologist David Barker points to a “programming” effect of early life events on adulthood (1) and provides a clue for the etiological study and prevention of chronic diseases.

Assisted reproductive technology (ART), used for the treatment of infertility, has led to >5 million births worldwide (2). ART would generate high serum estradiol (E₂) via ovarian stimulation. In women undergoing fresh embryo transfer (ET), in comparison with frozen ET and spontaneous conception (SC), the elevated E₂ lasts throughout early pregnancy (3). Prenatal exposure to high E₂ during early pregnancy leads to increased risk of low birth weight, dyslipidemia, and thyroid and cardiovascular dysfunction (3–7), suggesting that this abnormal intrauterine hormonal environment may impair the health of offspring.

Mice from in vitro fertilization (IVF) showed increased glucose intolerance and insulin resistance (8–10), and children conceived by IVF presented higher fasting glucose levels and lower peripheral insulin sensitivity (9). Because the mechanism remains unclear, we designed the current study to explore whether high maternal E₂ levels may play a role.

The hypothalamus-centered glucoregulatory system contributes to maintaining glucose homeostasis (11). The hypothalamic arcuate nucleus (ARC)–paraventricular nucleus (PVN) feeding network releases neuropeptide Y (NPY) and proopiomelanocortin (POMC), playing a major role in feeding regulation (12). Overexpressed hypothalamic NPY contributes to development of obesity and insulin resistance (13, 14). Direct administration of insulin into the brain reduces ARC NPY levels (15, 16). Reduction in insulin binding to insulin receptors (INSRs) in ARC leads to increased expression of NPY and its release into the PVN (17). In addition, chronic reduction of INSR in the hypothalamus produces glucose intolerance and may contribute to the development of type 2 diabetes (18).

In the present study, we compared the metabolism-related characteristics of offspring of SC, frozen ET, and fresh ET. We monitored the glucose metabolism of mice with prenatal high E₂ exposure and examined the hypothalamic genes regulating food intake and insulin sensitivity. The reversal effect on metabolic disorders of food restriction (FR) and its mechanism were discovered. Our findings will help us study insulin resistance originating in early life and may be useful for preventing metabolic disorders in fresh ET offspring.

Material and Methods

Human sample collection

We enrolled newborns and children aged 3 to 6 years, who were conceived by SC, frozen ET, and fresh ET, between October 2013 and June 2015. The newborns (SC, 114 male and 109 female; frozen ET, 43 male and 50 female; fresh ET, 25 male and 25 female) were delivered by cesarean section at Women’s Hospital, Zhejiang University. Characteristics of their mothers were obtained from the inpatient database. Umbilical cord blood from the newborns was taken during cesarean delivery. The children in the frozen ET (56 male and 68 female) and fresh ET (117 male and 100 female) groups were born to mothers who underwent ART in the reproductive medicine center of the Women’s Hospital, Zhejiang University, whereas children in the SC group (85 male and 70 female) were randomly chosen from the kindergarten physical examination program. Children and their parents were interviewed to collect essential information, and the children underwent physical examination to ensure consistent growth and development before a fasting blood sample was collected. All the newborns and children were full-term singletons without prenatal congenital malformation or fetal distress (Apgar score <7 at 5 minutes). Children born of mothers with gestational diabetes mellitus or other severe pregnancy complications (e.g., hypertension, intrahepatic cholestasis, abnormal thyroid function) or a family history of diabetes mellitus were excluded. Serum glucose and insulin concentrations of the blood were detected by a biochemical analyzer (Abbott). The homeostasis model of assessment for insulin resistance index (HOMA-IR) was calculated as glucose concentration (mmol/L) × insulin concentration (mIU/L)/22.5.

This study was approved by the Ethics Committee of the Women’s Hospital, Zhejiang University. The consents for the newborns and children were obtained from their parents.

Animal model

Male and female C57BL/6 mice (8 weeks of age) were paired (one male to two females) overnight, and the next day was declared day 0.5 of pregnancy (E0.5) if a vaginal plug was present in the morning. The pregnant mice were randomly assigned to receive gavage of 100 μg/kg/d estradiol valerate (Sigma) dissolved in corn oil or an equal volume of pure corn oil from E5.5 to E11.5. Maternal blood was collected from the angular vein on E11.5 at 3 hours after gavage to measure the serum E₂ concentration. The pups were weighed once per week after birth and were weaned at the age of 3 weeks. All animal protocols were approved by the Institutional Animal Care and Use Committee of Shanghai Jiao Tong University (permit no. A2016016).

Tolerance tests for glucose and insulin

Intraperitoneal glucose tolerance tests (GTTs) and insulin tolerance tests (ITTs) were performed at 3, 8, 12, 24, and 32 weeks after birth. For the GTT, mice were fasted overnight for 16 hours and injected intraperitoneally with glucose at 2 g/kg body weight. For the ITT, mice were fasted for 6 hours and injected intraperitoneally with insulin at 1 U/kg body weight. Blood samples were obtained from the tail vein at 0, 30, 60, and 120 minutes after injection both in GTT and ITT, and glucose was measured with an automatic glucometer (Roche). The area under the curve (AUC) was calculated as an index of glucose and insulin tolerance.

Food intake and food restriction

Mice used to monitor food intake were housed individually, and for each mouse the intake amount was determined daily for 1 week to calculate the average daily intake. The food given was preweighed, and the food left over was weighed 24 hours later. For FR, mice were given 75% of average daily food intake, which lasted for 8 weeks.

Enzyme-linked immunosorbent assay

Blood used for enzyme-linked immunosorbent assays (ELISAs) was obtained from the angular vein with capillary glass tubes. The serum E₂, dihydrotestosterone (DHT), insulin, and leptin levels were analyzed by an E₂ ELISA kit (Cusabio), a DHT ELISA kit (Cusabio), an insulin ELISA kit (Crystal Chem), and a leptin ELISA kit (Crystal Chem), respectively. Liver homogenate of fetal mice was made to perform ELISA assays with an alpha-fetoprotein ELISA kit (Cusabio).

Tissue immunofluorescence

Fetal and adult brain tissue was analyzed by immunofluorescence microscopy. Pregnant mice at E18.5 were euthanized, and fetal mice were dissected from the uterus and euthanized by decapitation. The fetal brains were removed and fixed in 4% paraformaldehyde (PFA) for 24 hours before infiltration with 20% to 30% sucrose. Sex determination of fetal mice was performed with SRY polymerase chain reaction (PCR) of DNA extracted from tail tips (19). The SRY primers are listed in Supplemental Table 1. Adult mice under anesthesia were transcardially perfused with 4% PFA, and their brains were then removed and fixed in 4% PFA for 4 hours before infiltration with 20% to 30% sucrose. Brain sections of 20 μm were made with a freezing microtome (Leica). Brain sections were blocked with 5% bovine serum albumin/0.3% Triton X-100 for 1 hour at room temperature. Incubation of primary antibodies lasted overnight at 4°C, followed by reaction with secondary antibodies at room temperature for 2 hours and counterstaining with 4′,6-diamidino-2-phenylindole. Primary antibodies were rabbit anti-NPY [1:3000; Cell Signaling Technology, catalog no. 11976; Research Resource Identifier (RRID): AB_2716286], rabbit anti-POMC (1:200; Cell Signaling Technology, catalog no. 23499; RRID: AB_2716565), and rabbit anti-INSR (1:100; Abcam, catalog no. ab131238; RRID: AB_11155955). Secondary antibodies were anti-rabbit Alexa Fluor 594 (1:200; Invitrogen, catalog no. A11012; RRID: AB_141359) and anti-rabbit Alexa Fluor 488 (1:200; Invitrogen, catalog no. A11008; RRID: AB_143165). The stained cells in brain sections were quantified in Image J (National Institutes of Health) software.

Quantitative real-time polymerase chain reaction and Western blotting

Quantitative real-time polymerase chain reaction (qPCR) was performed as previously reported (4). Primers are listed in Supplemental Table 1.

The Western blotting was performed as previously reported (4). The antibodies are mouse antiβ-actin (1:1000; Cell Signaling Technology, catalog no. 3700; RRID: AB_2242334), mouse anti-estrogen receptor α (ERα) (1:500; Abcam, catalog no. ab66102; RRID: AB_1140015), mouse anti-estrogen receptor β (ERβ) (1:1000; Abcam, catalog no. ab288; RRID: AB_303379), rabbit anti-androgen receptor (AR) (1:100; Abcam, catalog no. ab74272; RRID: AB_1280747), and rabbit anti-INSR (1:1000; Abcam, catalog no. ab131238; RRID: AB_11155955). Secondary antibodies were horseradish peroxidase–linked anti-rabbit immunoglobulin G (1:5000; Cell Signaling Technology, catalog no. 7074; RRID: AB_2099233) and horseradish peroxidase–linked anti-mouse immunoglobulin G (1:5000; Cell Signaling Technology, catalog no. 7076; RRID: AB_330924).

Bisulfite genomic sequencing PCR

The hypothalami of E18.5 and 32-week male offspring mice were dissected, and the parts close to the median line were carefully isolated under a stereoscope. Genomic DNA was extracted with the QIAamp DNA Mini Kit (Qiagen). Bisulfite was converted with the EpiTect Bisulfite Kit (Qiagen). The bisulfite-treated DNA was amplified, and the purified PCR products were cloned with the pMD18-T Vector System (TaKaRa). The primers are listed in Supplemental Table 1. The sequence obtained by cloning was analyzed with BiQ Analyzer (Max-Planck-Institut Informatik).

Statistical analysis

Data are presented as the mean ± standard error of the mean (SEM) and were analyzed by the Statistical Package for Sciences Software, version 17.0 (IBM). An unpaired Student t test was used for comparisons between two groups, a one-way analysis of variance (ANOVA) was used for comparisons between three groups, and a χ² test was performed to compare categorical data and the methylation percentage. P < 0.05 was considered statistically significant.

Results

Male newborns and children resulting from fresh ET show elevated blood insulin and HOMA-IR scores

Umbilical blood from SC, frozen ET, and fresh ET male newborns showed similar glucose levels, but the insulin levels and HOMA-IR scores were significantly higher in the fresh ET group [Fig. 1(a)]. However, the girls showed no significant differences for either index in the three groups [Fig. 1(b)]. The characteristics of these newborns are listed in Table 1. Still, boys from the fresh ET group presented elevated fasting insulin and HOMA-IR scores [Fig. 1(c)], whereas girls showed no significant differences between the three groups [Fig. 1(d)]. The characteristics of these children based on brief physical assessments are listed in Table 2.

Figure 1.

Comparison of glucose and insulin levels in offspring resulting from SC, frozen ET, and fresh ET. (a) Umbilical cord blood glucose, insulin, and HOMA-IR scores among male SC, frozen ET, and fresh ET newborns. (b) Umbilical cord blood glucose, insulin, and HOMA-IR scores among female SC, frozen ET, and fresh ET newborns. (c) Fasting blood glucose, insulin, and HOMA-IR scores among male SC, frozen ET, and fresh ET children. (d) Fasting blood glucose, insulin, and HOMA-IR scores among female SC, frozen ET, and fresh ET children. Error bars represent the SEM. Significance was determined by one-way ANOVA. *P < 0.05; **P < 0.01. ns, not significant.

Table 1.

Characteristics of Newborns in the SC, Frozen ET, and Fresh ET Groups

Characteristic	SC	Frozen ET	Fresh ET	P, Fresh ET vs SC	P, Fresh ET vs Frozen ET	P, Frozen ET vs SC
Male
n	114	43	25
Birth weight, g	3489 ± 37.90	3498 ± 60.47	3466 ± 76.56	0.96	0.94	0.99
Birth body length, cm	49.56 ± 0.09	49.64 ± 0.26	50.00 ± 0.16	0.40	0.67	0.98
Gestational age, wk	38.79 ± 0.08	38.87 ± 0.15	38.54 ± 0.18	0.44	0.33	0.87
Maternal age, y	31.45 ± 0.40	32.58 ± 0.60	32.76 ± 0.54	0.30	0.98	0.25
Apgar score	10.00 ± 0.00	10.00 ± 0.00	10.00 ± 0.00	na	na	na
Umbilical glucose, mmol/L	3.82 ± 0.06	3.83 ± 0.10	4.13 ± 0.36	0.27	0.38	0.10
Umbilical insulin, mIU/L	5.35 ± 0.19	4.70 ± 0.40	6.50 ± 0.49	0.01	<0.01	0.24
HOMA-IR score	0.91 ± 0.04	0.79 ± 0.07	1.20 ± 0.13	0.01	<0.01	0.28
Female
n	109	50	25
Birth weight, g	3328 ± 83.90	3364 ± 63.88	3303 ± 33.02	0.96	0.65	0.63
Birth body length, cm	49.73 ± 0.06	49.57 ± 0.16	49.67 ± 0.24	0.98	0.96	0.89
Gestational age, wk	38.81 ± 0.08	38.70 ± 0.14	38.61 ± 0.23	0.60	0.92	0.77
Maternal age, y	32.02 ± 0.38	32.84 ± 0.62	32.60 ± 0.61	0.79	0.97	0.45
Apgar score	9.98 ± 0.02	9.93 ± 0.05	10.00 ± 0.00	0.97	0.78	0.68
Umbilical glucose, mmol/L	3.83 ± 0.05	3.50 ± 0.17	3.51 ± 0.24	0.24	0.99	0.07
Umbilical insulin, mIU/L	5.33 ± 0.19	5.31 ± 0.35	5.90 ± 0.84	0.46	0.51	0.10
HOMA-IR score	0.93 ± 0.04	0.85 ± 0.08	1.12 ± 0.17	0.21	0.09	0.70

Data are presented as the mean ± SEM; significance was determined by one-way ANOVA.

Abbreviation: na, not applicable.

Table 2.

Characteristics of Children in the SC, Frozen ET, and Fresh ET Groups

Characteristic	SC	Frozen ET	Fresh ET	P, Fresh ET vs SC	P, Fresh ET vs Frozen ET	P, Frozen ET vs SC
Male
n	85	56	117
Age, y	4.59 ± 0.10	4.74 ± 0.12	4.81 ± 0.08	0.15	0.90	0.49
Weight, kg	18.83 ± 0.43	19.28 ± 0.44	19.24 ± 0.37	0.73	1.00	0.77
Height, cm	109.8 ± 0.86	110.9 ± 0.97	111.4 ± 0.67	0.31	0.93	0.66
BMI, kg/m2	15.48 ± 0.17	15.59 ± 0.21	15.39 ± 0.16	0.92	0.72	0.92
Blood pressure, mm Hg
Systolic pressure	97.89 ± 0.94	98.36 ± 1.31	98.21 ± 0.94	1.00	1.00	0.98
Diastolic pressure	56.16 ± 0.66	57.30 ± 1.11	54.64 ± 0.85	0.76	0.15	0.87
Heart rate, bpm	97.67 ± 1.39	95.28 ± 1.67	94.61 ± 1.18	0.61	0.95	0.77
Birth weight, g	3448 ± 42.38	3462 ± 45.97	3394 ± 38.11	0.63	0.86	0.42
Birth body length, cm	50.31 ± 0.14	50.05 ± 0.13	50.24 ± 0.16	0.95	0.74	0.60
Gestational age, wk	39.05 ± 0.19	38.71 ± 0.12	38.71 ± 0.09	0.35	1.00	0.40
Maternal age at birth, y	29.66 ± 0.38	30.88 ± 0.44	30.77 ± 0.34	0.08	0.98	0.13
ART type				na	0.96	na
IVF	na	39 (70%)	81 (69%)
ICSI	na	17 (30%)	36 (31%)
Delivery mode				0.97	0.81	0.85
Vaginal delivery	10 (12%)	6 (11%)	14 (12%)
Cesarean delivery	75 (88%)	50 (89%)	103 (88%)
Fasting glucose, mmol/L	4.83 ± 0.06	4.79 ± 0.05	4.87 ± 0.04	0.95	0.83	0.71
Fasting insulin, mIU/L	4.00 ± 0.17	4.00 ± 0.26	5.00 ± 0.15	0.02	<0.01	0.77
HOMA-IR score	0.84 ± 0.04	0.88 ± 0.06	1.15 ± 0.04	0.04	0.02	0.83
Female
n	70	68	100
Age, y	4.74 ± 0.10	4.78 ± 0.08	4.83 ± 0.08	0.74	0.96	0.90
Weight, kg	18.19 ± 0.35	18.60 ± 0.49	19.48 ± 0.82	0.35	0.53	0.96
Height, cm	108.8 ± 0.72	110.0 ± 0.81	110.0 ± 1.20	0.69	1.00	0.76
BMI, kg/m2	15.30 ± 0.18	15.26 ± 0.25	15.15 ± 0.16	0.59	0.59	1.00
Blood pressure, mm Hg
Systolic pressure	97.71 ± 1.37	101.60 ± 1.08	98.78 ± 1.32	0.95	0.60	0.68
Diastolic pressure	55.57 ± 0.44	56.41 ± 1.03	56.91 ± 0.88	0.84	0.65	1.00
Heart rate, bpm	98.21 ± 1.88	99.67 ± 1.21	98.48 ± 1.41	1.00	0.98	0.98
Birth weight, g	3326 ± 45.11	3402 ± 55.91	3311 ± 41.04	0.97	0.50	0.44
Birth body length, cm	50.07 ± 0.08	50.20 ± 0.10	50.10 ± 0.19	1.00	0.70	0.66
Gestational age, wk	39.23 ± 0.12	38.88 ± 0.11	38.74 ± 0.11	0.33	0.34	0.79
Maternal age at birth, y	30.68 ± 0.37	31.62 ± 0.44	30.88 ± 0.37	0.94	0.79	0.63
ART type				na	0.75	na
IVF	na	46 (68%)	70 (70%)
ICSI	na	22 (32%)	30 (30%)
Delivery mode				0.50	0.21	0.63
Vaginal delivery	8 (11%)	8 (12%)	15 (15%)
Cesarean delivery	62 (89%)	60 (88%)	85 (85%)
Fasting glucose, mmol/L	4.93 ± 0.07	5.01 ± 0.04	4.95 ± 0.04	0.93	0.98	0.99
Fasting insulin, mIU/L	4.04 ± 0.23	4.08 ± 0.35	4.34 ± 0.19	0.21	0.32	0.90
HOMA-IR score	0.99 ± 0.05	0.92 ± 0.09	1.00 ± 0.05	0.44	0.50	0.48

Data are presented as the mean ± SEM or percent; significance was determined by one-way ANOVA or χ² test.

Abbreviations: BMI, body mass index; ICSI, intracytoplasmic sperm injection; na, not applicable.

Prenatal exposure to high E₂ induces insulin resistance in mice

To verify whether high maternal E₂ can induce abnormal glucose metabolism in offspring, we created a mouse model of high E₂ prenatal exposure. Pregnant mice were gavaged with 100 μg/kg/d estradiol valerate in corn oil or with corn oil alone for control from E5.5 to E11.5 [Fig. 2(a)] (6), whose offspring were defined as the high prenatal estradiol exposure (HE) and negative control (NC) groups, respectively. Serum ELISA detection at E11.5 revealed a substantially elevated maternal E₂ concentration in the HE group [Fig. 2(b)]. Both male and female offspring in the HE group had lower weights during the first 2 weeks after birth, followed by a period of catch-up growth. However, the weight of HE male mice exceeded that of the NC male mice from 20 weeks, which did not happen in female mice [Fig. 2(c) and 2(d)]. We carried out GTTs and ITTs to monitor the glucose metabolism function [Fig. 2(e)–2(l) and Supplemental Fig. 1]. The HE male mice showed an increase in the AUC of GTT at 12 weeks after birth [Fig. 2(g)]; the difference was greater at 24 weeks and was accompanied by the appearance of an increase in the ITT AUC [Fig. 2(h) and 2(l)], indicating an impairment of glucose and insulin tolerance. However, the female offspring did not show any differences in their GTT or ITT results between the two groups [Supplemental Fig. 1(a)–1(h)]. The fasting blood of HE male mice at 24 weeks showed a significant increase in insulin and HOMA-IR score [Fig. 2(n) and 2(o)], but no differences in glucose or leptin were observed [Fig. 2(m) and 2(p)].

Figure 2.

Weight observations of the HE mouse model and the glucose metabolism tests of male offspring. (a) Schematic of the method used to generate the HE mouse model. (b) Maternal serum E₂ at E11.5. (c) Body weight of male offspring. (d) Body weight of female offspring. (e, f, g, and h) Left: GTT of male offspring at (e) 3, (f) 8, (g) 12, and (h) 24 weeks after birth. Right: AUC from the GTTs in arbitrary units (AU). (i, j, k, and l) Left: ITT of male offspring at (i) 3, (j) 8, (k) 12, and (l) 24 weeks after birth. Right: AUC from the ITTs. (m) Fasting glucose, (n) insulin, (o) HOMA-IR scores, and (p) fasting leptin of 24-week male offspring. Error bars represent the SEM; some SEMs are not visible because of their low values. Significance was determined by Student t test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. ns, not significant.

Prenatal exposure to high E₂ leads to increased NPY and decreased INSR in hypothalamus

Feeding behavior plays an essential role in metabolic homeostasis, so we observed the food intake of male offspring during growth. The HE mice tended to eat more from 8 weeks after birth compared with the NC group, although the difference was not significant until 20 weeks [Fig. 3(a)], which coincided with the body weight curve [Fig. 2(c)]. qPCR showed significantly increased messenger RNA (mRNA) expression of Npy (orexigenic) in the HE group at 24 weeks, whereas Pomc (anorectic) levels did not differ between the HE and NC groups [Fig. 3(b)]. Immunostaining showed that NPY-positive cells were significantly increased in the ARC and PVN in the HE hypothalami, although the number of POMC-positive cells was similar between HE and NC mice [Fig. 3(c)–3(e)]. We analyzed a series of genes involved in insulin signaling, leptin signaling, and glucose sensing by qPCR. Only Insr was significantly decreased in hypothalami from 24-week male HE mice [Fig. 3(f)]. Western blotting and immunofluorescence were used to verify this finding [Fig. 3(g) and 3(h)]. INSR is predominantly expressed in ARC and PVN within hypothalamus (20). It is of note that INSR staining was reduced in the ARC and PVN of HE hypothalami [Fig. 3(h)], in contrast to NPY expression [Fig. 3(c)]. We then examined male fetal brains for INSR expression. The mRNA and protein levels of Insr in E18.5 hypothalami were also decreased in the HE group [Fig. 3(i) and 3(j)], and INSR staining was weaker across the entire hypothalamus [Fig. 3(k)]. These results indicate that early life exposure to high E₂ exerts a negative effect on hypothalamic INSR expression, which is strong enough to last until adulthood.

Figure 3.

Food intake of male offspring and their hypothalamic expression of food intake–related neuropeptides and INSR. (a) Daily amount of food intake at 3, 8, 16, 20, and 24 weeks after birth of male offspring. (b) qPCR of Npy and Pomc expression in 24-week male hypothalami. β-Actin was included as an internal control. (c) Representative immunofluorescence images of NPY and POMC in 24-week male ARC and PVN of hypothalami; scale bar: 100 μm. (d and e) Quantification of (d) NPY-positive and (e) POMC-positive cells in the ARC and PVN (n = 5 mice per group). (f) qPCR of genes involved in insulin signaling, leptin signaling, and glucose sensing in hypothalami from 24-week male HE and NC mice. β-Actin was included as an internal control. (g) Left: Western blot of INSR in 24-week male hypothalami. β-Actin was included as an internal control. Right: quantification of staining density relative to β-actin. (h) Left: representative immunofluorescence images of INSR in the ARC and PVN from 24-week male hypothalami; scale bar: 100 μm. Right: quantification of INSR-positive staining in the ARC and PVN (five mice per group). (i) Quantitative qPCR of Insr in hypothalami from E18.5 male HE and NC mice. β-Actin was included as an internal control. (j) Top: Western blot of INSR in E18.5 male hypothalami. β-Actin was included as an internal control. Bottom: quantification of staining density relative to β-actin. (k) Left: representative immunofluorescence image of INSR in E18.5 male hypothalami; scale bar: 200 μm. Right: quantification of INSR-positive staining (five mice per group). Error bars represent the SEM. Significance was determined by Student t test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. ns, not significant.

Chronic FR reverses insulin resistance and rescues hypothalamic INSR

As mentioned earlier, HE offspring presented increased food intake with decreased insulin sensitivity, so we tried to explore whether limiting the daily food supply would correct this metabolic disorder. Some of the 24-week male HE mice were randomly chosen to be given food equivalent to 75% of their original daily intake for 8 weeks. Mice that underwent such FR lost weight [Fig. 4(a)]. GTTs and ITTs were then performed to examine the effect of FR. The AUCs of both the GTTs and ITTs were reduced in the HE-FR group, to levels close to those of the NC group [Fig. 4(b) and 4(c)]. No difference was found in fasting glucose, and the fasting insulin and HOMA-IR score recovered to normal levels in the HE-FR group [Fig. 4(d)–4(f)]. In addition, we found that fasting leptin decreased after FR [Fig. 4(g)]. The hypothalamic INSR mRNA and protein expression both increased in the HE-FR group [Fig. 4(h) 4(i)]. However, Npy expression was not affected by FR [Fig. 4(j)]. Immunofluorescence showed increased INSR with unchanged NPY upon FR in ARC and PVN [Fig. 4(k)–4(m)]. Because decrease of INSR expression was detected in both fetal and adult HE hypothalami and could be rescued by dietary intervention, we tried to explore whether epigenetic alteration was involved. One CpG island was found in the Insr promoter, and two pairs of primers were designed to cover the 21 CpG sites [Fig. 4(n)]. However, no DNA methylation was detected in CpG sites 9 to 21 of all samples (data not shown). The bisulfite genomic sequencing PCR showed a higher percentage of methylation in 4 CpG sites (sites 1, 2, 6, and 8) from the E18.5 Insr promoter [Fig. 4(o)] and 4 CpG sites (sites 1, 2, 3, and 8) from the 32-week Insr promoter [Fig. 4(p)] of male HE hypothalami compared with NC group. Among them sites 1, 2, and 8 were consistently highly methylated. It is worth noting that the methylation pattern was altered by chronic FR, with a decreased methylation percentage in sites 2, 7, and 8 in the HE-FR mice compared with the HE group [Fig. 4(p)].

Figure 4.

Effect of FR on metabolic alterations and related genes in HE offspring. (a) Body weight of male NC and HE mice and in HE mice after FR (HE-FR). (b) Left: GTTs of 32-week male NC, HE, and HE-FR offspring; *NC vs HE, ^#HE-FR vs HE. Right: AUC from the GTTs. (c) Left: ITTs of 32-week male NC, HE, and HE-FR offspring; *NC vs HE, ^#HE-FR vs HE; right: AUC from the ITTs. (d) Fasting glucose of 32-week male NC, HE, and HE-FR offspring. (e) Fasting insulin of 32-week male NC, HE, and HE-FR offspring. (f) HOMA-IR scores of 32-week male NC, HE, and HE-FR offspring. (g) Fasting leptin of 32-week male NC, HE, and HE-FR offspring. (h) qPCR of Insr in the hypothalamus of 32-week male NC, HE, and HE-FR offspring. β-Actin was included as an internal control. (i) Left: Western blot of INSR in 32-week male NC, HE, and HE-FR offspring. β-Actin was included as an internal control. Right: quantification of staining density relative to β-actin. (j) qPCR of Npy in the hypothalamus of 32-week male NC, HE, and HE-FR offspring. β-Actin was included as an internal control. (k) Representative immunofluorescence images of INSR and NPY in the ARC and PVN from 32-week male NC, HE, and HE-FR hypothalami; scale bar: 100 μm. (l) Quantification of INSR-positive staining in the ARC and PVN (five mice per group). (m) Quantification of NPY-positive staining in the ARC and PVN (five mice per group). (n) Schematic representation of the CpG island in the Insr promoter and the primers used in bisulfite genomic sequencing PCR. (o) Average methylation ratio in each CpG site of E18.5 male mice (three mice per group). (p) Average methylation ratio in each CpG site of 32-week male mice (three mice per group), *NC vs HE, ^#HE-FR vs HE. Error bars represent the SEM; some SEMs are not visible because of their low values. Significance was determined by one-way ANOVA or χ² test. ^#/*P < 0.05; ^##/**P < 0.01; ^###/***P < 0.001; ^####/****P < 0.0001. ns, not significant; TSS, transcription start site.

Discussion

ART was first performed in 1978 and has been widely and globally used, and its short- and long-term effects on the resulting offspring have attracted increasing attention. Increasing evidence indicates that ART may predispose offspring to increased risk of metabolic syndrome, type 2 diabetes, and cardiovascular disease (21). Fresh ET and frozen ET have been the conventional strategies in ART. Although increasing research about maternal and perinatal outcomes of fresh ET and frozen ET has been carried out, the safety and effects of the two strategies are still under debate. Patients who undergo fresh ET or frozen ET have higher levels of E₂ during follicular growth than do those with SC because of ovarian stimulation. This high level of E₂ continues during the first trimester of pregnancy in fresh ET cycles, whereas the frozen ET cycles are performed under natural hormonal conditions (as for SC).

Maternal E₂ can pass through the placental barrier to affect fetal growth and metabolism (22). We have tested the expression of α-fetoprotein in E18.5 fetal mouse, a protein secreted by fetal liver binding to E₂ to protect the fetus from the maternal estrogenic environment (23). No significant difference was detected between NC and HE fetuses, indicating no influence on α-fetoprotein production by the high maternal E₂ (Supplemental Fig. 2). Most studies have revealed that E₂ increases insulin sensitivity and glucose metabolism in adult humans and in animal models (24–26), but very little is known about how prenatal exposure to high E₂ would affect insulin action and glucose homeostasis during later life. E₂ is able to repress INSR expression in human and mouse cells in a dose- and time-dependent manner through unknown mechanisms (27, 28).

We previously reported that offspring of fresh ET have higher risk of low birth weight than those of frozen ET and SC (3). In the present study, the lower birth weight of fresh ET (Tables 1 and 2 ) compared with frozen ET and SC failed to show statistical significance, which may reflect the small sample size. The clinical results showed that insulin and HOMA-IR scores were significantly higher in the fresh ET group in male rather than female offspring, and this sex difference was confirmed by the mouse model. Both male and female HE mice weighed less than NC mice during the first 2 weeks after birth, which is consistent with the published findings (3), but the HE male offspring developed impaired metabolism, whereas the female offspring did not. It seems that male offspring were more vulnerable to prenatal high E₂ exposure, although more analysis is needed to reveal the mechanism of this sex-specific effect. The clinical findings showed signs of insulin resistance in the newborns, whereas HE male offspring developed glucose and insulin intolerance in adulthood. This discrepancy may be caused by the different sensitivities of the human and mouse fetuses to E₂ exposure. Moreover, GTTs and ITTs are needed to confirm when insulin resistance develops in human offspring.

Based on the published literature (3, 29) and our clinical experience, we found that mean maternal E₂ during early pregnancy in fresh ET is approximately twice that in frozen ET and SC, up to four times as high. According to the literature, the blood concentration of E₂ peaks at 3 hours after oral administration of estradiol valerate in rodents (30). The maternal serum E₂ level tested 3 hours after gavage with 100 μg/kg estradiol valerate in the HE group increased to four times that in the NC group, which is similar to the maximum E₂ increase in mothers undergoing fresh ET compared with frozen ET and SC during early pregnancy. We have tried a higher dose of ≥200 μg/kg/d, and miscarriage occurred frequently, indicating that excessively high maternal E₂ may exert developmental toxicity. We also tried subcutaneously imbedding an estradiol tablet in pregnant mice, expecting to achieve a consistent release of E₂ rather than a pulsed mode, which is more similar to the situation during human pregnancy. However, the imbedding surgery consistently induced miscarriage, even at a low E₂ release amount of 10 μg/kg/d, possibly because of vulnerability during early pregnancy to surgical procedures, anesthesia, or direct increase in blood E₂ levels without a hepatic first-pass effect. Although the phenotype of HE offspring proved that estradiol valerate gavage is effective, we would prefer to develop a better method of producing consistently elevated E₂ levels without inducing miscarriage.

The neurons in the hypothalamus are able to sense and respond to peripheral signals to regulate metabolism of the whole body. Prenatal development of the hypothalamus is plastic and is sensitive to an adverse maternal environment, and thus negative metabolic outcomes may be initiated during early developmental stages (31). Hypothalamic neurogenesis begins during early pregnancy, with the maturation of the ARC-PVN projections being completed before birth and 3 weeks after birth in humans and mice, respectively (32, 33). Thus the process of neurogenesis, during which neural stem cells differentiate to functional neurons (such as NPY and POMC neurons), occurs in a high-E₂ environment in the fresh ET fetus.

Hypothalamic insulin signaling has a major role in maintaining peripheral glucose homeostasis. Decreasing INSR in the hypothalamus causes hyperphagia and insulin resistance (34), and restoring insulin signaling in the hypothalamus increases the glycemic response to insulin in rats with diabetes (35). Our results present evidence that fetal programming leads to alterations in glucose metabolism in the adult via attenuation of hypothalamic insulin signaling. Although the insulin-resistant phenotype occurred in adult HE offspring, increased Insr promoter methylation was detected in the fetus. The development of diabetes or prediabetes is known to be a chronic process that can take a long time to progress from a genotype to a phenotype. Indeed, glucose intolerance in mice occurs 3 to 4 months after knockdown of Insr in the hypothalamus (18). We speculate that a compensatory function might delay the appearance of metabolic alterations and that a “second hit,” such as a high-fat diet, may lead to an earlier occurrence of these changes.

Despite the neuropeptides regulating appetite, sex hormones may also be involved in modulating eating behavior (36). We tested AR, and ERα, and ERβ in 24-week male hypothalami and compared their expression in females. Serum E₂ and DHT levels were also examined. However, neither male nor female mice showed differences between the NC and HE groups, except that male mice presented stronger AR staining with higher DHT levels, and female mice presented stronger ERα and ERβ staining with higher E₂ levels (Supplemental Fig. 3). These results revealed that the increased food intake of male HE mice resulted mainly from increased hypothalamic NPY rather than a stronger masculinization by sex hormones.

Studies have proved that FR can improve insulin sensitivity (37, 38) and reverse obesity induced by early-life overnutrition (39). Weight loss can increase insulin sensitivity by upregulating INSR (40, 41). HE mice that underwent FR recovered from insulin resistance, and the hypothalamic expression of INSR was rescued, confirming the existence of INSR-related insulin signaling impairment in HE offspring. We note that although increased hypothalamic insulin activity is recognized to downregulate NPY, the NPY level in ARC and PVN did not decrease with FR in our experiment. In fact, numerous studies have proved that FR or calorie restriction leads to increased hypothalamic NPY (42), which results from hunger signals such as decreased peripheral insulin and leptin (43). We speculate that the decreased levels of insulin and leptin in HE-FR mice relative to HE group played a role in upregulating NPY, which counteracted the downregulation by recovered hypothalamic insulin sensitivity, thus resulting in similar levels of NPY in HE and HE-FR mice. These findings suggest that although FR improves metabolism in HE mice, this intervention should be continued over the long term because the hypothalamic NPY remained higher than normal, and weight regain may result from the still-increased appetite when the diet intervention stops.

We found prenatal high E₂ exposure altered promoter methylation of hypothalamic Insr, and a published study also indicates that early life exposure to E₂ has lifelong effects on DNA methylation (44), although the specific mechanism awaits further investigation. The methylation analysis showed three CpG sites of Insr promoter that were highly methylated in both fetal and adult HE mice (sites 1, 2, and 8), for which FR induced two (sites 2 and 8) to levels approaching those of the NC group. The other site that was demethylated (site 7) upon FR also contributed to the recovered INSR level. In fact, as a reversible biological signal except during early development, the DNA methylation machinery is active throughout life and sensitive to environmental or dietary factors (45). Consistent with our findings, adult dietary restriction has already been proved to induce alterations in DNA methylation correlated with metabolism (46). These results demonstrate that FR exerts a remodeling effect on fetal-programmed epigenetic modulation, which throws light on prevention and intervention of early-life originated adult diseases.

Conclusion

To out knowledge, our study demonstrates for the first time that (1) male human offspring of fresh ET present higher risk of insulin resistance, (2) prenatal exposure to high E₂ leads to sex-specific insulin resistance via elevated methylation of hypothalamic Insr promoter in mice, and (3) chronic food restriction reverses insulin resistance by correcting abnormal Insr promoter methylation. This discovery gives insight into epigenetics-mediated reversibility of fetal-programmed adult metabolic disorders, provides a reference for ART strategy options in terms of maternal E₂ levels, and suggests the necessity of metabolic monitoring and dietary management of fresh ET offspring.

Abbreviations:

 
• ANOVA

  analysis of variance

 
• AR

  androgen receptor

 
• ARC

  arcuate nucleus

 
• ART

  assisted reproductive technology

 
• AUC

  area under the curve

 
• DHT

  dihydrotestosterone

 
• E₂

  estradiol

 
• ELISA

  enzyme-linked immunosorbent assay

 
• ERα

  estrogen receptor α

 
• ERβ

  estrogen receptor β

 
• ET

  embryo transfer

 
• FR

  food restriction

 
• GTT

  glucose tolerance test

 
• HE

  high prenatal estradiol exposure

 
• HOMA-IR

  homeostasis model of assessment for insulin resistance index

 
• INSR

  insulin receptor

 
• ITT

  insulin tolerance test

 
• IVF

  in vitro fertilization

 
• mRNA

  messenger RNA

 
• NC

  negative control

 
• NPY

  neuropeptide Y

 
• PCR

  polymerase chain reaction

 
• PFA

  paraformaldehyde

 
• POMC

  proopiomelanocortin

 
• PVN

  paraventricular nucleus

 
• qPCR

  quantitative real-time polymerase chain reaction

 
• SC

  spontaneous conception

 
• SEM

  standard error of the mean.

Acknowledgments

The authors thank Zhao-Wen Yan (Department of Pathological Teaching and Research, School of Medicine, Shanghai Jiao Tong University), Lei Cai, Chao Yang (Institutes of Bio-X, Shanghai Jiao Tong University), and Fan Yang (Shanghai Research Center for Model Organisms) for excellent technical assistance.

Financial Support: This work was supported by Special Fund for the National Key Research and Development Plan Grant 2017YFC1001303 (to H.-F.H.); National Natural Science Foundation of China Grants 81490742, 31471405, 81661128010 (all to H.-F.H.), and 81671456 (to X.-M.L.); Special Fund for Science and Technology Innovation of Shanghai Jiao Tong University Grant YG2014ZD08 (to H.-F.H.); and Doctoral Innovation Fund of the School of Medicine, Shanghai Jiao Tong University Grant BXJ201744 (to H.-H.W.).

Author Contributions: H.-H.W. and H.-F.H. designed experiments, researched data, wrote and edited the manuscript, and obtained funding to support the research. C.-L.Z. and M.L. researched data and wrote and edited the manuscript. M.H. and Q.Y. researched data and edited the manuscript. J.-X.L. designed experiments and edited the manuscript. J.L. and Y.-T.W. researched data. X.-M.L. and J.-Z.S. designed experiments and edited the manuscript. X.-M.L. obtained funding to support the research. H.-F.H. is the guarantor of this work and, as such, had full access to all study data and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Disclosure Summary: The authors have nothing to disclose.

Author notes

These authors contributed equally to this work.