Assisted Reproductive Technologies Predispose to Insulin Resistance and Obesity in Male Mice Challenged With a High-Fat Diet

Abstract

Assisted reproductive technology (ART) alters glucose homeostasis in mice and humans, but the underlying mechanisms are incompletely understood. ART induces endothelial dysfunction and arterial hypertension by epigenetic alteration of the endothelial nitric oxide synthase (eNOS) gene. In eNOS-deficient mice, insulin resistance is related to impaired insulin stimulation of muscle blood flow and substrate delivery and defective intrinsic skeletal muscle glucose uptake. We therefore assessed glucose tolerance, insulin sensitivity (euglycemic clamp), insulin stimulation of muscle blood flow in vivo, and muscle glucose uptake in vitro in male ART and control mice fed a normal chow (NC) or challenged with a high-fat diet (HFD) during 8 weeks. Glucose tolerance and insulin sensitivity were similar in NC-fed animals. When challenged with a HFD, however, ART mice developed exaggerated obesity, fasting hyperinsulinemia and hyperglycemia, and a 20% lower insulin-stimulated glucose utilization than did control mice (steady-state glucose infusion rate (GIR), 51.3 ± 7.3 vs 64.0 ± 10.8 mg/kg/min, P = 0.012). ART-induced insulin resistance was associated with defective insulin stimulation of muscle blood flow, whereas intrinsic skeletal muscle glucose uptake was normal. In conclusion, ART-induced endothelial dysfunction, when challenged with a metabolic stress, facilitates glucose intolerance and insulin resistance. Similar mechanisms may contribute to ART-induced alterations of the metabolic phenotype in humans.

The past decade has witnessed an exponential growth of the use of assisted reproductive technologies (ARTs), and children born by these techniques now make up 2% to 5% of births in developed countries (1). There is abundant epidemiological evidence suggesting that pathological events during early life predispose to premature cardiometabolic morbidity and mortality (2–4). In line with these findings, ART that implicates manipulation and culture of the embryo during a critical phase predisposes to premature vascular aging and arterial hypertension in animals and humans (5–7). In ART mice, this problem is related, at least in part, to altered DNA methylation of the endothelial nitric oxide synthase (eNOS) gene promoter resulting in decreased eNOS expression and, in turn, impaired nitric oxide (NO) synthesis in the vasculature (8). Studies in genetically engineered animals and in humans indicate that eNOS not only plays a key role for cardiovascular homeostasis, but also for the regulation of glucose homeostasis (9, 10). Consistent with this concept, mice generated by in vitro fertilization have recently been shown to display increased fasting glucose levels and impaired glucose tolerance, but the underlying mechanisms remain unclear (11–14).

In mice, eNOS knockout animals display insulin resistance and arterial hypertension, whereas eNOS heterozygous mice are predisposed to exaggerated high-fat diet (HFD)–induced insulin resistance and arterial hypertension (10, 15). In eNOS-deficient mice, impairments of insulin stimulation of muscle blood flow and substrate delivery to skeletal muscle and stimulation of intrinsic skeletal muscle glucose uptake contribute to insulin resistance (10, 15). We speculated that a similar mechanism contributes to insulin resistance in ART mice. We therefore assessed glucose tolerance, insulin sensitivity, and insulin stimulation of muscle blood flow in vivo together with intrinsic skeletal muscle glucose uptake in vitro in ART and control mice. To accentuate potential differences of glucose homeostasis between the two groups, assessments were performed in animals fed a normal chow or challenged with a HFD during 8 weeks.

Methods

All protocols were approved by the Institutional Animal Care and Use Committee of the University of Bern, Switzerland. Throughout the study period, mice were housed with lights on from 7:00 am to 7:00 pm and had access to food and water ad libitum. Starting at the age of 4 weeks, mice were fed either a normal chow (NC) diet (SAFE, Épinay-sur-Orge, France) or a HFD (fat, 78%; proteins, 22%; carbohydrates, 0%; SAFE) for 8 weeks.

The homeostasis model assessment index

To assess insulin sensitivity, we calculated the homeostasis model assessment (HOMA) index and performed intraperitoneal insulin tolerance tests (IPITTs). The HOMA index was calculated as the product of the fasting plasma insulin level (μU/mL) and the fasting plasma glucose level (mmol/L), divided by 22.5. Glucose (glucose Trinder kit 100; Sigma Diagnostics, St. Louis, MO) and insulin (enzyme-linked immunosorbent assay; Mercodia, Uppsala, Sweden) plasma concentrations were measured after a 6-hour fast in blood samples obtained by retro-orbital puncture under anesthesia with isoflurane (Piramal Healthcare, Mumbai, India).

Glucose tolerance test

A glucose tolerance was assessed with the intraperitoneal glucose tolerance test (IPGTT). Mice were fasted for 6 hours and injected intraperitoneally (IP) with glucose (1 g/kg). Blood was collected for glucose determination from the tip of the tail at given time points before and every 15 minutes after the injection.

Insulin tolerance test

The IPITTs were performed after a 6-hour fasting period. Insulin (0.5 U/kg, Actrapid HM; Novo Nordisk Pharma, Küsnacht, Switzerland) was injected IP, and plasma glucose was measured on the tail tip every 15 minutes during 1 hour after insulin injection.

Hyperinsulinemic euglycemic clamp studies

The glucose turnover rate was assessed after 8 weeks of dietary treatment during a hyperinsulinemic euglycemic clamp in freely moving mice following a 6-hour fast as described previously (10). Briefly, 3 days before the study, mice were anesthetized with isoflurane, and an indwelling catheter was inserted into the vena cava through the femoral vein, sealed under the back skin, and exteriorized and glued at the back of the neck. On the day of the clamp, insulin (18 mU/kg/min, a dose known to suppress hepatic glucose production) (15) was infused into the femoral vein for 2 hours. Euglycemia (5.5 mmol/L) was maintained by periodically adjusting a variable infusion of 15% glucose. The steady-state glucose infusion rate (GIR) was calculated as the mean of the values obtained every 5 minutes during the last 30 minutes of the clamp.

Insulin stimulation of muscle blood flow in vivo

A 90-minute hyperinsulinemic euglycemic clamp study was performed as described previously. During the clamp, muscle blood flow was measured in anesthetized mice (5% isoflurane inhalation for the induction, followed by 1.5% to 2% for the maintenance of anesthesia), using a laser Doppler probe (PeriFlux System 5000, probe no. 403; Perimed, Järfälla, Sweden) placed directly onto the hindlimb skeletal musculature and stabilized with a micromanipulator, as described previously (10). During the entire study, the body temperature was maintained at 37°C with a temperature control unit (FHC, Bowdoin, ME). The blood flow signal was recorded on a personal computer using specific data acquisition software (PowerLab 400; ADInstruments, Dunedin, New Zealand).

Intrinsic muscle glucose uptake in vitro

After cervical dislocation, the soleus muscles were rapidly isolated, tied separately to silk threads by the tendons, and immersed for 15 minutes into an incubation medium [Krebs–Ringer bicarbonate (pH 7.3) supplemented with 1% bovine serum albumin (fraction V, pH 7.0) and 2 mM sodium pyruvate]. Under an atmosphere containing 5% CO₂ and 95% O₂, the muscles were then incubated in the medium with or without 10 nM insulin for 60 minutes at 37°C. Thereafter, the muscles were immersed for 20 minutes in the incubation medium supplemented with [³H]2-deoxyglucose (0.1 mM, 0.5 µCi/mL). During this immersion, the [³H]2-deoxyglucose is metabolized and accumulates as [³H]2-deoxyglucose-6-phosphate. To stop the reaction, the muscles were immersed in ice-cold saline buffer, washed for 30 minutes, and then dissolved in 1 M NaOH at 55°C for 60 minutes. An aliquot of the extract was neutralized with 1 M HCl, spun down, and the [³H]-labeled radioactivity was counted in the presence of a scintillation buffer. Sample aliquots were used for protein determination.

NOS activity in aorta and skeletal muscle

Tissues samples were suspended in radioimmunoprecipitation assay buffer (Amresco, Solon, OH) supplemented with a protease inhibitor cocktail (mammalian, Amresco). Seventy microliters and 10 μL/mg tissue were used for aorta and skeletal muscle, respectively. Tissues were incubated 15 minutes on ice before being mechanically lysed using a Qiagen TissueLyser for 3 minutes at a frequency of 30 Hz. Samples were incubated 15 minutes on ice before being centrifuged 10 minutes at 14,000 rpm at 4°C. The supernatant was harvested and centrifuged (10 minutes at 14,000 rpm at 4°C) before quantification of supernatant protein using a bicinchoninic acid protein assay (5). NOS activity was quantified using an ultrasensitive colorimetric assay for NOS (NB78; Oxford Biomedical Research, Rochester Hills, MI) as described previously (16).

Blood chemical analysis

Glucose (glucose Trinder kit 100; Sigma Diagnostics) and insulin (enzyme-linked immunosorbent assay; Mercodia) plasma concentrations were measured in the fed state in conscious mice, between 1:00 and 3:00 pm, after a 6-hour fast (n = 10 to 20 mice for each group). The interassay and intra-assay coefficients of variation for insulin were 3.2% and 4.0%, respectively (n = 17), and for glucose were 9.7% and 9.5%, respectively (n = 12).

In vitro fertilization and embryo culture

ART mice were generated as previously described (5). Briefly, 8- to 12-week-old female FVB mice were superovulated by IP injection of 5 IU (0.1 mL) pregnant mare serum gonadotropin (Intervet, Zürich, Switzerland), followed 50 hours later by an IP injection of 5 IU (0.1 mL) of human chorionic gonadotropin (hCG; Intervet, Zürich, Switzerland). Fourteen hours after hCG, cumulus–oocyte complexes were recovered from oviducts in human tubal fluid (HTF) medium (Irvine Scientific, Santa Ana, CA) supplemented with 5 mg/mL human serum albumin (HSA; Irvine Scientific). Spermatozoa were collected from the cauda epididymis of 10- to 14-week-old FVB mice and capacitated for 60 minutes in HTF/HSA medium at 37°C under a humidified atmosphere of 6% CO₂ in air. Oocytes were inseminated 14 hours after hCG with 10⁶ spermatozoa in HTF/HSA medium for 4 hours at 37°C and 6% CO₂. Eggs were then transferred to 25-µL drops of G1 medium (Vitrolife, Göteborg, Sweden) covered with paraffin oil (Irvine Scientific). The embryo culture was conducted up to the blastocyst stage in sequential G1 and G2 media (Vitrolife) pre-equilibrated at 37°C and 6% CO₂. Embryos were kept in the G2 medium for 48 hours before the transfer to pseudopregnant females.

Embryo transfer

NMRI (and, for some studies, FVB) females of at least 6 weeks of age were placed with vasectomized males to mate 2.5 days prior to embryo transfer. The morning after mating, females were checked for the presence of a vaginal plug. On the transfer day, pseudopregnant females were anesthetized by IP injection of xylazine (15 mg/kg) and ketamine (100 mg/kg). Seven to 10 embryos were inserted into each fallopian tube.

Control FVB mice were generated by mating animals from our colony. Females were culled at weaning. Males were separated and put together (one per litter) with mice originating from other litters (four to five mice per cage) and then randomly assigned to the different tests. Particular care was taken to mix mice from different litter sizes. Ten- to 14-week-old male mice were used for the studies.

Statistical analysis

Data were analyzed using the GraphPad Prism 5 (GraphPad Software, La Jolla, CA). Statistical analysis was done with analysis of variance (one- or two-way) for between-group comparisons. Post hoc analyses were performed using the Bonferroni posttest for multiple comparisons. Two-tailed t tests were used for single comparisons. A P value of <0.05 was considered to indicate statistical significance. Unless otherwise indicated, data are given as mean ± standard deviation (SD).

Results

Table 1 shows embryo development rate in ART mice, as well as litter size and sex ratio in ART and control mice.

Table 1.

Embryo Development, Litter Size, Sex Ratio, and Weight in ART and Control Mice

	ART (NMRI): 28 Litters	ART (FVB): 15 Litters	Control: 20 Litters	P
				ART (NMRI) vs ART (FVB)	ART (NMRI) vs Control	ART (FVB) vs Control
Two cell rate, % (SD)	67.6 (14.7)	69.9 (10.6)	—	0.56
Blastocyst rate, % (SD)	57.4 (16.8)	53.0 (8.7)	—	0.26
Litter size, n/mother (range)	5.0 (3-11)	7.1 (3–11)	8.1 (5–11)	0.001	<0.0001	0.042
Sex ratio, M:F	0.95:1.05	1.08:0.92	0.97:1.03	0.23	0.92	0.32
Weight at 7 days, g (SD)	5.0 (0.8)	4.9 (0.8)	4.7 (0.4)	0.61	0.11	0.45

—, not applicable.

Effects of ART on fasting glucose and insulin levels, HOMA index, and body weight

In NC-fed mice, fasting plasma glucose concentration was significantly higher in ART than in control mice [10.2 ± 0.9 vs 8.6 ± 1.5 mmol/L, P < 0.01, ART vs control, Fig. 1(a)], whereas fasting plasma insulin concentration was comparable in the two groups [10.1 ± 3.5 vs 10.7 ± 3.3 µU/mL, P = 0.66, ART vs control, Fig. 1(b)]. In HFD-fed animals, fasting plasma glucose [10.8 ± 1.6 vs 9.5 ± 1.2 mmol/L, P = 0.04, ART vs control, Fig. 1(a)] and fasting plasma insulin levels [14.8 ± 6.2 vs 9.2 ± 4.7 μU/mL, P < 0.01, ART vs control, Fig. 1(b)] were significantly higher in ART than in control mice. The HOMA index was comparable in NC-fed animals, but when challenged with a HFD, it was significantly increased in ART compared with control mice [7.0 ± 2.7 vs 4.2 ± 2.5, P = 0.02, ART vs control, Fig. 1(c)].

Figure 1.

Fasting plasma glucose (a) and insulin concentration (b) and HOMA index (c) in 12-week-old male mice conceived by ARTs or naturally conceived control (Ctrl) mice fed with a NC diet or HFD for 8 weeks. Data represent mean ± SD for at least eight mice per group.

Figure 2 shows that starting from week 9.5, body weight in ART mice was significantly higher (all P < 0.05) than in control mice.

Figure 2.

Body weight in ART mice generated using NMRI or FVB gestational carriers and in control mice fed with a NC diet or a HFD. P < 0.05 control vs ART (NMRI) and ART (FVB). Data represent mean ± SD.

Effects of ART on glucose tolerance

To further characterize the effects of ART on glucose homeostasis, we performed IPGTTs. Figure 3(a) shows that in NC-fed animals, glucose tolerance was similar in the two groups [area under the curve (AUC) IPGTT, 1161.3 ± 124.8 vs 1181.6 ± 156.5 min⋅mmol/L, P = 0.78, ART vs control]. Whereas HFD, as expected, impaired glucose tolerance in both groups (P < 0.001, NC vs HFD), glucose intolerance was significantly more severe in ART than in control mice [AUC IPGTT, 2188.7 ± 32.6 vs 1938.9 ± 177.6 min⋅mmol/L, P = 0.013, ART vs control, Fig. 3(a)].

Figure 3.

Plasma glucose concentration during IPGTTs (a), decrease of the plasma glucose concentration during IPITTs (b), and GIRs during hyperinsulinemic euglycemic clamp studies (c) in 12-week-old male mice conceived by ARTs or naturally conceived control (Ctrl) mice fed with a NC diet or HFD for 8 weeks. Data represent mean ± SD for at least eight mice per group. (a) *P < 0.05 ART HFD vs control HFD. (b) ***P < 0.001 ART HFD vs all the others groups. (c) *P < 0.05 ART HFD vs all the others groups.

Effects of ART on insulin sensitivity

To more directly assess the effects of ART on insulin sensitivity, we performed IPITTs and hyperinsulinemic clamp studies.

The insulin-induced decrease of plasma glucose concentration was comparable in NC-fed animals. In contrast, when fed a HFD, it was significantly smaller in ART than in control mice [analysis of variance P < 0.0001, Fig. 3(b)].

In NC-fed animals, the GIR during the steady-state phase of hyperinsulinemic euglycemic clamp studies was comparable in ART and control mice [80.6 ± 13.9 vs 79.9 ± 19.1 mg/kg/min, P = 0.91, ART vs control, Fig. 3(c)]. In contrast, when fed with a HFD, the GIR needed to maintain euglycemia was roughly 20% lower in ART than in control mice [51.3 ± 7.3 vs 64.0 ± 10.8 mg/kg/min, P = 0.012, ART vs control, Fig. 3(c)].

To directly test for the potential effects on our results, including issues that may be different during gestation and/or early care between NMRI and FVB gestational carriers, we performed clamp studies in ART mice generated in FVB recipient mothers. The GIR needed to maintain euglycemia in ART mice generated in FVB gestational carriers was not different from the one observed in ART mice generated in NMRI recipient mothers [57.9 ± 9.1 vs 51.3 ± 7.3 mg/kg/min, P = 0.17, ART (FVB) vs ART (NMRI)], and, most importantly, it was significantly lower than the GIR needed to maintain euglycemia in control mice [57.9 ± 9.1 vs 74.1 ± 16.3 mg/kg/min, P < 0.01, ART (FVB) vs control].

Plasma glucose concentration during the steady-state phase of the clamp studies and plasma insulin concentration at the end of the clamp were comparable in all conditions (data not shown).

Effects of ART on vascular and skeletal muscle NOS activity

To test the effects of ART on NOS activity, we measured vascular and skeletal muscle NOS activity. In the aorta, NOS activity was significantly lower in ART mice than in control mice [Fig. 4(a)], whereas in skeletal muscle tissue it was comparable in all groups [Fig. 4(b)].

Figure 4.

NOS activity in (a) aorta and (b) skeletal muscle tissue in ART and control mice fed with a NC diet or HFD. Data represent mean ± SD for at least five mice per group.

Effects of ART on insulin stimulation of skeletal muscle blood flow in vivo and muscle glucose utilization in vitro

To test whether ART alters insulin stimulation of skeletal muscle perfusion and substrate delivery, we measured hindlimb muscle blood flow during clamp studies. In NC-fed animals, insulin stimulation of muscle blood flow was similar in the two groups (AUC 10,740.6% ± 1267.9% vs 11,175.9% ± 1669.2%, P = 0.27, ART vs control, Fig. 5), whereas in HFD-fed animals, this effect was significantly impaired in ART compared with control mice [AUC 8915.3% ± 811.5% vs 10,387.4% ± 1362.6%, P < 0.01, ART vs control, Fig. 5(a)].

Figure 5.

AUC of insulin stimulation of muscle blood flow (a) during 90-minute hyperinsulinemic euglycemic clamp studies. (b) Basal (BL) and insulin-stimulated (Ins) 2-deoxyglucose (2-DG) uptake in soleus muscle in vitro harvested from 12-week-old male mice conceived by ARTs or naturally conceived control (Ctrl) mice fed with a NC diet or HFD for 8 weeks. Data represent mean ± SD for at least seven mice per group.

To study the effects of ART on muscle glucose uptake in the absence of confounding effects of muscle perfusion, we measured glucose uptake in isolated skeletal muscle preparations. The basal and insulin-stimulated intrinsic skeletal muscle glucose uptake in vitro was comparable in NC-fed and HFD-fed ART and control mice [Fig. 5(b)].

Discussion

ART has allowed millions of infertile couples to have children. However, this success may have come at a price, because evidence is accumulating indicating that ART causes premature vascular aging, arterial hypertension, and glucose intolerance in mice and humans (5, 6, 12). Although the underlying mechanisms are incompletely understood, there is evidence that ART-induced alterations of the cardiovascular phenotype are related to altered eNOS function, a gene that is also important for the regulation of glucose homeostasis (5, 6, 10, 17–19). In the present study, we show that when fed a NC diet, ART mice had normal insulin sensitivity. When challenged with a HFD for 8 weeks, however, ART mice developed exaggerated obesity, glucose intolerance, and insulin resistance. Insulin-resistance appeared to be primarily related to impaired endothelium-dependent insulin stimulation of blood flow and substrate delivery to skeletal muscle tissue rather than altered intrinsic skeletal muscle glucose uptake.

In this study, we assess insulin sensitivity in ART mice using the gold standard method (20), the hyperinsulinemic euglycemic clamp with an insulin infusion rate (18 mU/kg/min) known to be sufficiently high for detecting differences in peripheral insulin resistance in insulin-resistant mouse models (10, 15). We found that when challenged with a HFD, the GIR needed to maintain euglycemia was roughly 20% lower in ART than in control mice. Moreover, and in line with these findings, IP insulin administration induced a smaller decrease of the plasma glucose concentration in ART mice than in control mice. We previously found that in mice fed with a HFD, during insulin infusion used in the present clamp studies, hepatic glucose production is completely suppressed (15), indicating that the lower GIR in ART mice is accounted for by decreased glucose uptake in peripheral tissues. In a previous clamp study, Donjacour et al. (13), using much lower insulin infusion rates (4 mU/kg/min) that resulted in steady-state GIRs of ∼20 mg/kg/min, found no difference in insulin sensitivity between ART and control mice. Whereas this low rate of insulin infusion appears to suppress hepatic glucose production in mouse models, and thus can be used to assess this variable, the low GIR needed to maintain euglycemia during such low-dose insulin infusion may not be sensitive enough for assessing differences in peripheral insulin resistance in mice (13). Although we did not find detectable alteration of insulin sensitivity in NC-fed FVB ART mice, Chen et al. (12) reported altered glucose homeostasis in NC-fed Bl6 ART mice, and, most importantly, with regard to clinical relevance, these authors reported insulin resistance in young apparently healthy ART subjects under a controlled normal diet. Thus, it appears that ART-induced alterations of glucose homeostasis may be more easily detectable in humans than in FVB mice, because they do not depend on HFD feeding.

The present findings shed light on the mechanisms underlying ART-induced insulin resistance. In mice, ART causes vascular dysfunction that is related to epigenetically altered expression and function of the eNOS gene in the vasculature (5, 18). Accordingly, in this study, we found that vascular NOS activity was significantly lower in ART mice compared with controls. There is abundant evidence that defective endothelial NO synthesis, by altering mitochondrial biogenesis and facilitating visceral fat gain (21), and by altering insulin signaling in endothelial cells (22) leading to impaired insulin stimulation of blood flow and substrate delivery to skeletal muscle tissue, causes insulin resistance in humans and mice (10, 15, 17, 23). In line with these observations, we found that in ART mice, HFD-induced exaggerated insulin resistance was associated with impaired insulin stimulation of muscle blood flow. Interestingly, when fed a NC diet, insulin sensitivity and insulin stimulation of muscle blood flow were normal in ART mice, indicating that ART-induced alteration of eNOS function was not sufficient to trigger a detectable change of glucose homeostasis under these conditions. This finding is in accordance with previous observations showing normal insulin sensitivity and insulin stimulation of muscle blood flow in mice with partial eNOS deletion (eNOS^+/− mice) (15). Taken together, these findings further support the concept that a partial impairment of endothelial NO synthesis is sufficient to maintain normal insulin sensitivity under usual conditions. During metabolic stress (HFD), however, eNOS deficiency amplifies a pathological mechanism and leads to exaggerated insulin resistance (15). Moreover, and consistent with previous results in normal and eNOS^+/− mice (15, 24), the low-carbohydrate HFD did not alter basal and insulin-mediated muscle glucose uptake in vitro in control and ART mice. Finally, skeletal muscle tissue NOS activity was comparable in ART and control mice. Taken together, these findings suggest that ART-induced alterations in skeletal muscle tissue did not contribute importantly to exaggerated metabolic insulin resistance in HFD-fed ART mice in vivo.

In this study, HFD also induced glucose intolerance in ART mice, as evidenced by higher fasting plasma glucose concentration and impaired glucose tolerance after IP glucose administration. These data are in line with previous data in ART mice generated in a different mouse strain (C57BL/6J) (11). Taken together, the present and these earlier data (11–15) demonstrate that ART induces comparable alterations of glucose tolerance in different mouse strains. It is noteworthy that the alterations of glucose homeostasis in these studies were observed in offspring of healthy mice generated by ART, ruling out any confounding effects of factors such as sterility or older age of the parents or transmission of unrecognized cardiometabolic risk factors as possible causes of the altered metabolic phenotype in the offspring, further demonstrating that ART per se is the main culprit of this alteration. In this context, although ovarian stimulation does not appear to markedly alter the vascular phenotype in the offspring (5, 6), there is evidence that it may alter epigenetic marks in oocytes (25) and alter glucose metabolism in HFD-fed female mice (11).

An additional important observation of our studies was that HFD induced more severe obesity in ART mice than in control mice. The underlying mechanisms are not known. Although gain-of-function studies in which NO has been increased pharmacologically or genetically show that NO may have antiobesogenic effects (26–28), eNOS deficiency per se does not appear to alter body weight in mice, as evidenced by comparable body weight in eNOS knockout and control mice (10), as well as in eNOS^+/− and control mice fed with a HFD (15). Alternatively, it is conceivable that ART alters the regulation of genes involved in body weight control (29), alterations that either per se or in conjunction with altered eNOS function may facilitate HFD-induced obesity in ART mice. Of note, ART mice of another mouse strain (C57BL/6J), although displaying glucose intolerance, did not develop exaggerated obesity when fed a HFD (12). This may be related to differences in the ART technique used, different composition of the diet or differences of the susceptibility to diet-induced obesity between mouse strains.

There are some additional features and limitations that need to be mentioned. First, the mouse strain of the gestational carrier (NMRI) and the ART mice (FVB) was different, and by issues that may be different during gestation or early care could have influenced the results. The present observation that insulin resistance in ART mice was comparable when using FVB mice as gestational carriers suggests that early life environmental differences related to strain differences between carrier and offspring did not contribute importantly to the results. Second, we studied only male mice, because the aim was to test whether the ART-induced impairment of vascular function in male ART mice causes insulin resistance. In females, estrogens may attenuate the effects of fetal programming of vascular dysfunction (30). Third, litter size in ART mice was smaller than in in vivo controls. This difference, although statistically significant, was relatively small and indeed comparable to the difference in litter size between ART (litter size 6.2 ± 2.2) and flushed blastocysts control mice (litter size 8.1 ± 3.0) used as appropriate controls in the Donjacour et al. (13) study. Finally, and most importantly, in our additional studies using FVB gestational carriers, the litter size (7.0 ± 1.8) was similar to the control group (8.1 ± 1.8). Thus, these small differences in litter size are unlikely to have had a major impact on the interpretation of the present data. Fourth, in the present study embryo culture was performed in 20% oxygen. The data of Chen et al. (12) showing insulin resistance in in vitro fertilization mice generated using embryo culture under 5% oxygen, together with our present observation in in vitro fertilization mice generated using 20% oxygen, could be consistent with the concept that oxygen concentration may not be the main culprit in the pathogenesis of ART-induced alteration of glucose homeostasis in mice.

In conclusion, we found that ART mice are susceptible to exaggerated HFD-induced insulin resistance, glucose intolerance, and obesity. Our data suggest that there exists an important interaction between ART-induced epigenetic alterations in the embryo and environmental factors later in life in the regulation of glucose homeostasis. In humans, ART now accounts for 3% to 5% of births in developed countries. Altered glucose metabolism and insulin resistance have recently been reported in humans conceived by ART under energy balanced and HFD overfeeding conditions (12), suggesting that a similar mechanism may be operational in humans. Moreover, there is evidence for decreased vascular NO bioavailability in ART children (31). We speculate that, under a metabolic stress, such as the one represented by a Western-type diet, this alteration may facilitate the development of insulin resistance, glucose intolerance, and obesity in the human ART population.

Abbreviations:

 
• ART

  assisted reproductive technology

 
• AUC

  area under the curve

 
• eNOS

  endothelial nitric oxide synthase

 
• GIR

  glucose infusion rate

 
• hCG

  human chorionic gonadotropin

 
• HFD

  high-fat diet

 
• HOMA

  homeostasis model assessment

 
• HSA

  human serum albumin

 
• HTF

  human tubal fluid

 
• IP

  intraperitoneal(ly)

 
• IPITT

  intraperitoneal insulin tolerance test

 
• IPGTT

  intraperitoneal glucose tolerance test

 
• NC

  normal chow

 
• NO

  nitric oxide

 
• NOS

  nitric oxide synthase

 
• SD

  standard deviation.

Acknowledgments

We are indebted to Caroline Mathieu and Pierre Dessen for their invaluable technical assistance.

This work was supported by the Swiss National Science Foundation, the Placide Nicod Foundation, the Swiss Society of Hypertension, and the Swiss Society of Cardiology.

Disclosure Summary: The authors have nothing to disclose.

Author notes

*

These authors contributed equally to this study.

Address all correspondence and requests for reprints to: Emrush Rexhaj, MD, Departments of Cardiology and Clinical Research, Inselspital, University Hospital Bern, CH-3010 Bern, Switzerland. E-mail: [email protected].