Impact of embryo culture and day of embryo transfer on the cardiometabolic health of adult mice

Abstract

Assisted reproductive technologies (ARTs) have rapidly evolved since being introduced in 1978. However, many of the procedures used clinically lack a basis of long-term studies to ensure their safety. Though in vitro fertilization (IVF) is largely safe, follow-up studies have shown that IVF-conceived children may show signs of altered fat deposition, increased fasting glucose, and increased blood pressure. These results are, however, limited by a low number of patients and different ART variables (different lengths of embryo culture or different types of culture media). Results of studies using animal models have confirmed many of these results and shown that more stressful culture conditions result in disrupted adult phenotypes. Presently, it is unclear if culture conditions such as duration and day of transfer might affect adult health. To investigate the hypothesis that the length of embryo culture could affect adult phenotype, we generated mice by IVF and transferred them at the cleavage stage (IVF_8C group) or at the blastocyst stage (IVF_BL group) and studied adult phenotype. Results were compared to those obtained with naturally conceived animals flushed out of the uterus and transferred to the recipient (FB group). We found sexual dimorphic effects with male mice showing a more severe phenotype. Male offspring resulting from cleavage stage transfer showed altered glucose handling, left cardiac dysfunction, and shorter lifespan, while male offspring post blastocyst transfer showed reduced locomotor activity. Female mice showed a milder phenotype, particular for female offspring generated by transfer at the cleavage stage.

Longer embryo culture results in adult male mice displaying lower locomotor activity, while shorter embryo culture gives rise to adult male mice showing mild glucose intolerance, left cardiac dysfunction, and shorter lifespan. Female mice are less affected.

Graphical Abstract

Graphical Abstract.

Introduction

Substantial evidence indicates that embryonic or fetal stress may profoundly influence the long-term health of offspring. The Developmental Origins of Health and Disease hypothesis posits that discrepancies between intrauterine and postnatal environments can predispose offspring to chronic diseases in adulthood [1]. Epidemiological studies on humans have demonstrated that maternal undernutrition, such as during famine, is associated with the onset of metabolic and cardiovascular disorders in adulthood [2]. Animal models have confirmed this phenomenon as maternal caloric restriction, exposure to a high-fat diet, or other intrauterine stressors give rise to offspring with hypertension or diabetes later in life [3, 4]. Notably, this sensitivity extends both to the neonatal and the preimplantation period [5, 6]. Given that preimplantation embryo manipulation is routinely performed in assisted reproductive technology (ART) to treat infertility, understanding the long-term health effects of exposure of embryos to non-physiological condition is important.

Since the birth of the first ART baby in 1978, over 12 million children have been born by ART worldwide. Over the past four decades, ART techniques have evolved significantly, enhancing fertilization rates, embryo quality, and pregnancy outcomes [7]. A notable clinical strategy that has improved live birth rates is the transfer of blastocyst-stage embryos as opposed to cleavage-stage embryos [8]. The timing of blastocyst transfer is considered physiologically optimal as it closely mimics natural implantation, theoretically enhancing synchrony between the uterine lining and the embryo [9]. However, because not all embryos reach the blastocyst stage, continued clinical use of cleavage-stage embryo transfers is necessary [10]. As of today, patients with fewer than five embryos are often offered cleavage embryo transfer to reduce the risk of cycle cancellation and the possibility that no embryo survives to the blastocyst stage in vitro. Varying length of culture generates a challenge because extended laboratory culture (5–6 days versus 3 days) could result in epigenetic changes that could cause changes in the long-term health of offspring.

Studies using animal models show that embryo culture may impact embryo development, gene expression [11, 12], mitochondrial function [13] embryo metabolism [14] epigenetic marks, and imprinting via DNA methylation [15, 16] Additionally, we and others have demonstrated that in vitro fertilization (IVF) and embryo culture result in mouse offspring with altered growth and glucose intolerance [17–20]. However, the specific long-term health consequences for offspring following embryo transfer at 3 days versus 5 days remain unknown. To date, Aljahdali et al. conducted the sole study indicating that the duration of embryo culture alters adult metabolic responses in a sexually dimorphic manner in mice. Adult male mice conceived via IVF and cultured to the blastocyst stage prior to embryo transfer exhibited significantly higher systolic blood pressure and increased angiotensin-converting enzyme activity in serum and lung compared to naturally mated mice and mice conceived by IVF and transferred at the two-cell stage. Furthermore, male mice conceived via IVF showed impaired glucose tolerance in both short- and long-term culture groups, as evidenced by slower recovery compared to naturally mated mice [21]. Given that human embryos are not normally transferred at the two-cell stage, we investigated how transfer at the cleavage (post-zygotic genome activation) or blastocyst stage could affect adult phenotype. Further, unlike previous studies that primarily examined cardiovascular outcomes, we evaluated a broader range of health indicators, including energy expenditure, glucose metabolism, ¹³C flux studies, and cardiac function, providing a more comprehensive assessment of long-term effects.

Materials & methods

Ethical approval and cohort generation

All experiments were reviewed and approved by the Institutional Animal Care and Use Committee of the University of California San Francisco. Animals were housed in a clean barrier facility in a room maintained under a constant 12-h light/dark cycle with a room temperature setting of 21°C–23°C. Mice were provided with nesting material in cages and had free access to standard chow and tap water. In vitro fertilization was performed as previously described [14]. Briefly, female C57BL6 mice were superovulated using an injection series consisting of 5 IU of pregnant mare serum gonadotropin (PMSG) followed by 5 IU of human chorionic gonadotropin (HCG) 48 h later. The following morning, approximately 12–14 h later, C57BL6 male mice were euthanized by CO₂ asphyxiation. The cauda epididymis was dissected and placed in a 0.9 ml drop of Human Tubal Fluid (Sigma Aldrich) medium overlayed with oil. The tissue was slashed at several locations to allow for sperm to swim out and capacitate in culture for 1 h. Female mice were euthanized by CO₂ asphyxiation, and the oviducts were dissected and placed in 50 μl drops of HTF to collect the cumulus–oocyte complexes from the ampullae. Gametes were co-incubated in a fresh drop of HTF medium overlayed with oil for 4–6 h and placed in a humidified incubator under 5% CO₂ and 20% O₂. Fertilized zygotes were quickly washed with K⁺ Simplex Optimized Medium supplemented with amino acids (KSOM+AA) to minimize debris carryover prior to being placed in fresh 50 μl drops of KSOM+AA, then returned to the incubator for embryo culture until the planned embryo transfer day.

To generate the cohort of animals used in this study, embryo transfer was performed at two different time points of preimplantation embryo development: 8-cell (IVF_8C) and blastocyst (IVF_BL) stages. Prior to the transfer, pseudopregnant recipient mothers were prepared by mating CF-1 females to CD-1 vasectomized male mice. IVF_8C embryos were collected after 48 h post-IVF and transferred in the oviducts of a recipient female, 10 embryos per side. For the IVF_BL group, late cavitating embryos of similar morphology were transferred to the uterus of a recipient female, 10 blastocysts per uterine horn. The control group of animals (flushed blastocyst, FB) were produced by first superovulating C57BL6 females with the same injection series described above and placing up to two females in a cage of one C57BL6 male immediately after HCG injection. At 3.5 days post coitum, the mated females were euthanized, and the uterus was dissected and flushed with HTF media to collect blastocysts produced by natural mating. Twenty blastocysts were collected and transferred to the uterus of a recipient female CF-1 mouse, 10 per uterine horn. This process allows for the control of both the superovulation and embryo transfer procedures. Pups were expected at E19.5 and weaned after 21 days. Litters were separated by sex and placed in a cage holding a maximum of five mice. The total number of animals generated for this study is as follows: Male FB n = 12, IVF_8C  n = 16, IVF_BL  n = 18 and Female FB n = 21, IVF_8C  n = 13, IVF_BL  n = 12. The total number of litters generated for this study is reported in Supplemental Table 1.

Postnatal growth, Intraperitoneal Glucose Tolerance Test, and blood collection

Postnatal weights were measured every week starting from week 3, post-weaning, until week 62 to observe any later-stage physiological changes. A glucose tolerance test (GTT) was performed at week 35. Prior to the start of the test, animals were fasted for 6 h. Baseline measurement was performed by measuring whole blood collected from the tail vein and using a handheld glucometer. Prior to the start of the experiment, up to 50 μl of whole blood was collected into a tube using a glass capillary. To start the experiment, a bolus of 1 mg/g glucose was injected intraperitoneally, and glucose levels were measured at 15, 30, 60, and 120 min post-injection. Additionally, whole blood was collected 15 min post-injection. Animals were immediately provided with free access to food and water immediately after the last measurement.

Body composition analyses

Animal body compositions were measured by echoMRI as previously described [22]. Mice were placed in a cylindrical plastic restraint, and movement was limited by gently adding a weighted plastic insert on top of the animal. The restraint apparatus was then loaded into a magnetic resonance imaging machine, and the mice were briefly exposed to a low-intensity electromagnetic field to measure fat and lean mass.

Comprehensive Laboratory Animal Monitoring System

Metabolic studies were conducted at the UCSF Mouse Metabolism Core. Mice were single-housed for 1 week for acclimatization to experimental settings. The 16-chamber Comprehensive Lab Animal Monitoring System (CLAMS; Columbus Instruments, Inc.) was used to measure metabolic parameters. Mice were placed individually in a chamber and monitored for over 5 days. Data were recorded in 1-h bins and used to calculate the following metabolic parameters including oxygen consumption (VO₂), carbon dioxide production (VCO₂), and respiratory exchange rate, as well as food intake and locomotor activity. Data are presented as activity during light hours (7:00 p.m. to 6:59 a.m.) and the total average over a 24-h period. Due to a technical error discovered post-run, all animals were exposed to light during dark hours across all the CLAMS experiments used for this analysis. Additionally, CLAMS chambers are equipped with infrared photocell technologies that create beams of light that count as a movement when broken by an animal passing through. Three categories of movements are quantified by the system: X_TOT (total number of X-axis breaks), X_AMB (total number of ambulatory breaks in the X axis), and Z_TOT (total number of vertical motions, also known as rearing).

Blood pressure measurements

A non-invasive tail cuff system (SC1000, Hatteras Instruments, Inc., Grantsboro, NC, USA) was employed to measure systolic and diastolic blood pressures (BPs). Since only male mice were found to present differences in metabolic phenotypes, female mice were excluded from BP studies. Mice were positioned in a metal animal holder on a heated specimen platform maintained at 35°C–37°C. BP was detected using a sensor system that monitors changes in tail blood flow. For each mouse, 5 preliminary and 10 experimental measurements were recorded at each time point. The BP and heart rate values were derived by averaging 10 experimental measurements. To minimize stress-induced variations in BP, mice underwent a 3-day training period prior to the commencement of the experiment. Baseline BP measurements were subsequently obtained over two consecutive days across all three animal groups (FB, n = 10; IVF_8C, n = 9; IVF_BL, n = 10).

Echocardiography

Ultrasonography studies were conducted using the Vevo 2100 system (FUJIFILM VisualSonics, Inc., 3080 Yonge Street, Suite 6100, Box 66, Toronto, ON, Canada), a specialized preclinical ultrasound device designed for laboratory animals. The scanning frequency was set to 40 MHz, with an acquisition rate of 20 frames per second. Multiple videos were recorded during each scanning session, capturing at least five cardiac cycles per video. Each examination included both axial and longitudinal scans in B-mode and M-mode. Three cardiac cycles were analyzed per sequence. Axial measurements of the left ventricle were performed at the level of the papillary muscle base, while longitudinal scans were conducted at the aortic outflow tract, where the mitral valve was visible. These anatomical landmarks ensured the reproducibility of the examination protocol.

Procedures involving live animals were conducted in accordance with institutional ethical committee guidelines. For ultrasonography, mice were anesthetized using isoflurane in an air mixture (5% for induction, 2% for maintenance). The airflow was set to 0.8 L/min, and the specimens were placed on a heated platform to maintain their body temperature within a physiological range. Animals were positioned in dorsal recumbency after confirming the appropriate level of anesthesia. The skin was shaved, and fur was removed using depilatory cream to eliminate interference between the ultrasonographic gel and the animal’s skin. The warmed gel was applied to the skin, and the scanning probe was positioned on top. The heart rate was monitored throughout the imaging process to ensure adequate anesthesia and to minimize variability in results due to differences in cardiac frequency. Upon completion of the scan, animals were returned to their cages for recovery.

Intraperitoneal Glucose Tolerance Testing with U-¹³C-glucose and ¹⁴C-2-deoxyglucose isotopic tracers & terminal tissue collection

Commencing at 7:00, animals were fasted for 6 h and weighed prior to the start of experiments. For the purpose of studying glucose flux, a 20% solution of isotopically labeled U-¹³C-glucose (M + 6, CLM 1396–10, Cambridge Isotope Laboratories, Inc., Tewksbury, MA) was prepared fresh each time the experiment was performed. A dosage volume equivalent to 1 mg/g of 20% ¹³C-glucose was aliquoted into a clean tube for each animal and the injection volume, and for the purpose of determining tissue glucose uptake, ¹³C-glucose was combined with 13 μCi of 2-[¹⁴C(U)]-deoxy-D-glucose ([¹⁴C]2DG, NEC720A250UC, Revvity, Inc., Waltham, MA). Prior to injection, baseline glucose levels from whole blood obtained from the tail vein were measured using a handheld glucometer and compatible test trips (AccuCheck Guide Me, Roche Diagnostics). Glucose levels were measured at t = 10, 20, and 30 min post-injection. Immediately after the final measurement, mice were euthanized by CO₂ asphyxiation and cervical dislocation, and the following tissues of interest were collected for metabolite analysis: brain, heart, liver, muscle, gonadal fat, and blood. Serum was separated from whole blood by centrifugation, collected into clean tubes, and frozen. Dissected tissues were placed in vials, snap frozen in liquid nitrogen, and stored at −80°C until ready to be studied. Serum insulin was measured by radioimmunoassay.

Metabolite extraction, derivatization, and gas chromatography–mass spectrometry

Plasma (50 μl of plasma) and tissue (30–50 mg tissue) metabolites were isolated using a biphasic methanol/water/chloroform extraction. Norvaline (20 μl of 5 mM) was added to each sample as an internal standard for metabolite quantification. The polar layer of the extract was isolated using a fine-tipped pipette and air-dried overnight for storage at −80°C prior to derivatization. Polar metabolites from plasma and tissue extracts were converted to their methoxime tert-butylsilyl derivatives using MtBSTFA+1% TBDMCS (catalog 1–270,144-200, Regis Technologies). Calibration standards with known amounts of each metabolite were prepared and derivatized simultaneously with the extracted samples for absolute quantification of metabolite abundances. Derivatized samples were injected onto an HP-5 ms column (catalog 19091S-433, Agilent Technologies) in an Agilent 7890B gas chromatograph paired with an Agilent 5977A mass spectrometer. Data were acquired in scan mode, and metabolites were identified through a comparison of mass spectra using a previously generated standard library. The accuracy of mass isotopomer distribution measurements was validated through a comparison of the theoretical and experimental values of unenriched control samples. Radioactivity of [¹⁴C]2DG and [¹⁴C]2DG-6-P in tissues was determined as previously described [23]. Tissue [¹⁴C]2DG-6-P levels were normalized to the brain [¹⁴C]2DG-6-P for each mouse.

Beta cell isolation and in vitro secretion assay

Beta cells were isolated from the exocrine pancreas of 65-week-old female mice from FB and IVF_8C as previously reported [19]. Mice were provided to the UCSF Islet Production Core, and experiments were performed according to standard procedures [24]. Briefly, animals were sacrificed by cervical dislocation, and the pancreas and common bile duct were exposed. The pancreas is inflated through the common bile duct with a 30 g needle and a 5 ml syringe containing a concentration of 0.8 mg/ml Collagenase P. The whole pancreas was dissected out of the body, collected into a clean vial, and incubated in a 37°C water bath for up to 17 min. The vial was shaken gently by hand to digest the tissue. The pancreas digest was placed on a mesh screen and washed several times with washing buffer to collect loose islets that are detached from the tissue matrix. Isolated islet cells in washing buffer are transferred into 50 ml conicals and spun gently for a few seconds with the centrifuge promptly stopped once it reached a speed of 1000 RPM. The supernatant is removed and resuspended in Ficoll with 1.108 density to begin density gradient separation. Ficoll densities of 1.096, 1.069, and 1.037 are carefully placed above the 1.108 layer and spun for 15 min at 1800 RPM in 4°C. A band of islet tissues forms between the 1.069 and 1.096 layers, and the supernatant is removed until the islet layer can be carefully recovered into a clean conical tube and washed three times with washing buffer. Thirty islets of similar sizes were collected and incubated in either 2.8 or 28 mM glucose for 1 h each and placed in lysis buffer to measure insulin levels by enzyme-linked immunosorbent assay (ELISA).

Statistics

Data are separated by sex and presented as the mean ± SD. A one-way ANOVA was used for statistical analysis, as appropriate. Tukey post hoc corrections were applied to test for differences between groups when a one-way ANOVA was significant. Since variation in litter size can profoundly affect postnatal development [25], we used only recipients that give birth to four to nine mice. Second, we used the pups as opposed to the litter as a unit of comparison [26], given the great cost of the experimental procedures and because this approach has been done by us [18, 19] and others [27, 28] in past experiments. Supplemental Table 4 indicates how the results would have changed by analysis by litters.

For calorimetry experiments, ANCOVA was performed using the CalR software (https://calrapp.org/) as this method statistically detaches the influence of a continuous variable, such as body mass, on the analysis of indirect calorimetry data [29]. For radioactive tracer experiments, values greater or less than 2.0 times the standard deviation of the group were considered outliers and removed from the analysis. To determine differences in animal lifespan, a Kaplan–Meier analysis was performed to generate a survival curve. Since our analysis compares three datasets, if the curve was found to be significantly different, we performed a pairwise comparison between any two datasets to determine additional statistical significance via the Gehan–Breslow–Wilcoxon Test. All analyses were performed using Prism 10.2.2, and a P-value of <0.05 was used to determine statistical significance (GraphPad Software, Inc., Boston, MA).

Results

Male IVF_8C offspring have higher weight and display subtle differences in glucose metabolism compared to IVF_BL and FB

Offspring generated following transfer at the eight-cell stage (IVF_8C) display an increase in weight compared to other groups of mice. Male IVF_8C mice had higher weight compared to control FB males from week 14 until week 23 and from week 36 to 40 (Figure 1A). Total body composition showed no differences in fat or lean mass in male mice (Supplemental Figure S1A and C). Tissue weight at sacrifice was found not to be different in either sex between any groups (Supplemental Table 2). Glucose tolerance test at 35 weeks of age showed no difference between IVF-generated mice and control mice. However, IVF_BL showed signs of insulin resistance compared to IVF_8C, having higher fasting glucose (IVF_BL 208.8 ± 46.68 vs. IVF_8C 167.5 ± 20.58, P < 0.05) as well higher glucose following GTT (Figure 1B), with higher AUC (AUC IVF_8C = 27 099 ± 11 806 vs. AUC IVF_BL = 40 287 ± 10 315; P < 0.05 Figure 1C). There were no differences in insulin levels measured at fasting and 15 min between any groups of mice (Figure 1D). At 60 weeks, male mice did not exhibit any differences in glucose tolerance (Figure 1E and F).

Figure 1.

Effect of length of culture on growth and glucose tolerance in adult male mice. (A) Growth curves of male offspring from the FB (n = 12 pups, 6 litters), IVF_8C (n = 16 pups, 7 litters), and IVF_BL (n = 22 pups, 6 litters) groups. Body weight (g) was measured weekly. (B) A glucose tolerance test (GTT) performed at week 35 suggests IVF_8C males (n = 12) exhibit lower glucose levels from baseline to 60 min compared to controls (n = 12) and IVF_BL (n = 19). Different superscripts indicate significantly different comparisons. (C) The area under the curve of the male GTT was found only to be different between the IVF groups where IVF_8C (n = 10) was ~30% lower than IVF_BL (n = 19). (D) Male Insulin levels performed at 35 weeks show borderline blunted insulin response in IVF_8C at baseline and 15 min post-glucose administration. (E, F) Terminal glucose tolerance performed at 60 weeks does not show differences between groups according to point comparisons or AUC. (G–I) Glucose uptake index in muscle (G), fat (H), and heart (I) tissues of male offspring from the FB, IVF_8C, and IVF_BL groups. Glucose uptake is expressed relative to brain uptake for each mouse. Male IVF_8C (n = 5) mice exhibited a higher glucose uptake in muscle tissue compared to FB (n = 8, P = 0.08) and IVF_BL (n = 10, P < 0.05) offspring. Data are presented as mean ± SD. ^*P < 0.05, ^**P < 0.01.

Flux studies with ¹³C-glucose showed that IVF_8C mice had a higher glucose uptake in muscle compared to IVF_BL mice (0.47 vs. 0.15, P < 0.05, Figure 1G) and FB mice (0.18, P < 0.05) but not in adipose (Figure 1H) or cardiac tissues (Figure 1I). Labeling of secondary metabolites (alanine and lactate) was not different between the groups (Supplemental Figure S2), suggesting no difference in their metabolism.

No group differences were found in the estimate of glycolytic flux (Figure 2A, APE lactate: APE glucose ratio) or pyruvate oxidation (Figure 2B, ratio of APE lactate: APE citrate).

Figure 2.

¹³C flux studies: effect of length of embryo culture on glycolysis and pyruvate oxidation. (A, C) The ratio of lactate atom percent excess (APE) to glucose APE, a measure of glycolysis, in various tissues of male FB (n = 7), IVF_8C (n = 6), and IVF_BL (n = 10) groups (A) and female offspring in FB (n = 9), IVF_8C (n = 7), and IVF_BL (n = 8) groups. Each group derived from at least 4 litters. No significant differences were observed in either male or female mice between IVF groups and control FB.

Female IVF_8C mice exhibit increased body weights and altered glucose metabolism later in life compared to control mice

Female IVF_8C mice weighed significantly more than FB starting at week 36 (Figure 3A). Female IVF_8C exhibited nearly 16% higher fat compared to IVF_BL females (47% vs. 31%, P < 0.05 Supplemental Figure S1B) and no difference in lean mass content (Supplemental Figure S1D). Female mice did not show significant differences among any groups in glucose tolerance (Figure 3B and C) or insulin (Figure 3F) levels at 35 weeks of age.

Figure 3.

Effect of length of culture on growth and glucose tolerance in adult female mice. (A) Female IVF_8C offspring (n = 13 pups; 6 litters) displayed increased body weights compared to FB (n = 21 pups, 7 litters) and IVF_BL (n = 12 pups, 6 litters) beginning at 36 weeks. (B) No differences were found in female GTT or (C) glucose AUC between FB (n = 18) and IVF groups (IVF_8C  n = 13 and IVF_BL = 12). (D) Terminal GTT performed at 60 weeks demonstrates that IVF_8C females (n = 7) show lower baseline glucose but (E) higher glucose AUC compared to controls (n = 9). (F) Insulin levels of IVF_8C female mice also showed a similar borderline blunted insulin response seen in male mice. (G) More pancreatic β-cells were isolated from 65-week-old IVF_8C female mice (n = 3 and 1 litter) compared to FB (n = 5 and 2 litters). (H) However, unlike FB islets, IVF_8C islets did not demonstrate a significant increase in insulin production following glucose stimulation. (I–K) Glucose uptake in muscle (I), fat (J), and heart (K) tissues of female mice showed no statistical differences between the groups. Data are presented as mean ± SD. ^*P < 0.05, ^**P < 0.01.

At 60 weeks, female IVF_8C and IVF_BL mice had significantly lower fasting glucose levels compared to control female mice; IVF_8C female mice also had greater glucose AUC compared to control mice (IVF_8C = 7645 ± 1917 vs. FB = 5811 ± 1884, P < 0.05, Figure 3D and E). To further test for the possibility that the glucose intolerance was driven by impaired insulin secretion in IVF offspring, we isolated beta cells from five control and three IVF_8C female mice at 60 weeks of age. We found that IVF_8C mice had a higher number of beta cells isolated per mouse (325 vs. 230. Figure 3G, P < 0.05) and the isolated beta cells of IVF_8C had overall higher baseline insulin secretion and a non-statistically significant increase in insulin post-glucose exposure (Figure 3H), similar to what we had found in a prior cohort of IVF mice [19].

Regarding flux studies, neither female IVF_8C and IVF_BL mice showed any differences in glucose uptake in muscle (Figure 3I), adipose (Figure 3J), or cardiac tissues (Figure 3K) compared to FB females. As in males, labeling of secondary metabolites (alanine and lactate) was not different (Supplemental Figure S2E and F). Female mice also did not show any differences in glycolytic (Figure 3C) or pyruvate oxidation flux (Figure 3D) among the tissues tested.

Male mice generated by IVF show reduced ejection fraction compared to controls.

Cardiac function was examined by echocardiography. IVF male mice (both IVF_8C, 49.97 ± 5.38 and IVF_BL, 49.14 ± 7.93) showed reduced ejection fraction compared to control (61.83 ± 10.21, P < 0.05, Figure 4A) and reduced left ventricle shortening (IVF_8C = 25.22 ± 3.14, IVF_BL = 24.90 ± 5.0 vs. FB = 33.27 ± 7.31, P < 0.05, Figure 4D) and larger internal diameters of ventricles in systole and diastole (Supplemental Figure S3A and D). Left ventricular volume was borderline increased only in IVF_8C males in both systole (P = 0.05) and diastole (P = 0.06, Supplemental Figure S3B and E).

Figure 4.

Cardiac function and blood pressure are affected in male IVF_8C mice. (A) Ejection fraction (EF) for males and (B) females. Fractional shortening (FS) in (D) male and (E) female offspring from the FB, IVF_8C, and IVF_BL groups. Male IVF_8C offspring exhibited significantly lower EF and FS compared to FB, indicating impaired cardiac function. No significant differences in EF or FS were observed between female groups. (C) Systolic blood pressure (SBP) and (F) diastolic blood pressure in male mice. Male IVF_8C mice had significantly higher SBP compared to FB. FB n = 10 (4 litters), IVF_8C  n = 5 (4 litters), IVF_BL  n = 10 (3 litters) and females: FB n = 10 (5 litters), IVF_8C  n = 9 (5 litters), IVF_BL  n = 10 (5 litters). Data are presented as mean ± SD. ^*P < 0.05, ^**P < 0.01 compared to FB.

Female mice do not express any differences in cardiovascular phenotypes between groups (Figure 4B and E).

Because only male mice exhibited echocardiographic differences, we measured blood pressure only in male mice. No differences in blood pressure were noted, with the exception that IVF_8C male mice had lower systolic blood pressure compared to IVF_BL (102 vs. 110 mm/Hg, respectively, P < 0.05) mice (Figure 4C).

Male IVF mice transferred at the blastocyst stage show decreased locomotor activity

Body weight was used as a covariate for the analysis of calorimetry studies. While body weight had a significant effect on multiple parameters, there was no significant interaction between body weight and fertilization method on oxygen consumption, carbon dioxide production, energy expenditure, or food intake in either male (Supplemental Figure S4) and female (Supplemental Figure S5) mice (Supplemental Table 3).

Male mice showed no differences in total horizontal cage movements (X_TOT, Figure 5A); however, IVF_B males moved less by showing a 45% reduction in X_AMB (i.e., total number of ambulatory breaks in the X-axis IVF_BL = 82.72 ± 32.2; IVF_8C = 153.0 ± 61.0; FB = 182.5 ± 75.6, P < 0.05 for FB vs. IVF_BL,  Figure 5C) and 21% decrease in rearing motion (Z_TOT, i.e., total number of vertical motions, IVFBL = 11.9 ± 4.4 IVF_8C = 46.2 ± 36.1; FB = 58.0 ± 30.4, P < 0.05 for FB vs. IVF_BL  Figure 5E) compared to FB and IVF_8C mice.

Figure 5.

Locomotor activity in male and female offspring. (A, B) A total number of axis (X_TOT) breaks is not different across groups in either males (A) or females (B). Data are binned and averaged for light, dark, and full 24-h periods. (C, D) Ambulatory activity indicated by consecutive movements across the cage suggests male IVF_BL mice (n = 7, 3 litters) do not explore the cage environment as much as control mice (n = 7, 3 litters), (C) and this behavior is not exhibited by females (D). (E, F) Rearing behavior in which animals stand on their hind legs (Z_TOT) is significantly reduced in male IVF_BL mice compared to control (E) and is not different in females (F). Different superscripts indicate a significantly different comparison. Data are presented as mean ± SD. FB n = 7 (3 litters), IVF_8C  n = 6 (2 litters), IVF_BL  n = 7 (3 litters) and females: FB n = 10 (3 litters), IVF_8C  n = 7 (4 litters), IVF_BL  n = 7 (4 litters).

There were no differences in locomotor activities in female mice (Figure 5B, D, and F).

Male mice generated by IVF and transferred at the cleavage stage showed reduced lifespan

Animals were sacrificed after 60 weeks, and survival analysis was evaluated. IVF_8C male mice showed significantly decreased survival (P < 0.05) compared to FB and BL, with 6/16 surviving to week 60 (compared to 19/22 IVFBL and 11/11 FB, Figure 6A). Female mice did not display any notable differences in survival probability between groups (Figure 6B).

Figure 6.

Effect of IVF and culture conditions on survival of male and female mice.

(A) Kaplan–Meier survival curves for male offspring from the FB, IVF_8C, and IVF_BL groups. Male IVF_8C mice exhibited significantly reduced survival compared to FB and IVF_BL. (B) No significant difference in survival was observed between female mice in any of the groups. Different superscripts denote significantly different comparisons, P < 0.05. Males FB (n = 12; 6 litters) IVF_8C  n = 16 (7 litters), IVF_BL  n = 22 (6 litters) and females: FB (n = 21; 6 litters), IVF_8C (n = 13; 7 litters), IVF_BL (n = 12; 6 litters).

Discussion

Given the continuous increase in the use of ART, it is crucial to monitor offspring health and understand how different embryonic stresses may lead to adult onset of diseases. Given that human embryos are routinely transferred at the cleavage or blastocyst stage [30], we aimed to study the effects of different lengths of embryo culture (cleavage vs. blastocyst stage) on long-term cardiometabolic health in mouse offspring.

Overall, we found that embryos generated by IVF show mild variation in phenotype compared to in vivo conceived mice (i.e., FB group) and these differences were more pronounced in male mice. More specifically, mice generated by IVF compared to control show: (1) mild growth and glucose metabolism differences (slight increase in weight with age and altered glucose tolerance), (2) evidence of left cardiac dysfunction, (3) decreased activity, and surprisingly, (4) male offspring generated after cleavage embryo transfer had shorter lifespan. Subgroup analysis limited to mice generated by IVF showed that cleavage-stage embryo transfer had a more severe phenotype.

The first notable finding is that mice conceived by IVF showed mild growth and glucose metabolism differences compared to control mice. Overall, both male and female offspring generated after cleavage embryo transfer (IVF_8C) showed more significant changes and gained weight more rapidly compared to controls and IVF_BL mice, with male mice starting at 14 weeks of age and females at later ages (36 weeks and later).

Interestingly, control mice lost more weight following the GTT performed at 35 weeks (Figure 1A) than IVF-conceived mice. While mice can lose up to 15% of weight due to fasting and blood loss following GTT [31], it was surprising that the IVF-conceived mice showed a less severe response to this type of stress. While we cannot explain the result, based on a two-hit model [32], we would have predicted the opposite, with IVF-conceived mice losing more weight.

Regarding glucose metabolism, IVF-conceived mice had a tendency for glucose intolerance, more evident in the IVF_8C mice of both sexes. In particular, the length of embryo culture did not appear to affect GTT response in IVF-conceived male mice compared to controls at 35 or 60 weeks of age. However, male IVF_BL mice showed reduced glucose tolerance compared to IVF_8C mice only at 35 weeks of age. IVF_8C female mice were glucose intolerant at 60 weeks compared to control mice only. Insulin measurements were not different at 35 weeks, although, interestingly, IVF_8C mice of both sexes showed a decline of insulin levels at 15 min of an oral glucose tolerance test (OGTT) compared to fasting, which bordered statistical significance. This drop in insulin levels at 15 min indicates likely a faster peak of insulin secretion post-glucose stimulation or an overall decreased ability to secrete insulin. The functional studies in isolated beta cells in female mice (Figure 3) provide some guidance. The larger beta cell mass in IVF_8C females at 65 weeks of age and the lack of significant increase in insulin secretion post-high-glucose exposure compared to control suggest signs of beta cell exhaustion. This result is similar to what we had found in female IVF mice in a prior cohort [19]. Loss of normal beta cell function is central to the pathogenesis of type II diabetes. Interestingly, early life stress has been linked to beta cell failure as a high-fat diet given to female mice during gestation resulted in beta cell failure and the onset of diabetes in the offspring [33], likely because nutrient overloading may trigger an inflammatory response resulting in beta cell dysfunction or death [34, 35].

¹³C-glucose/¹⁴C-DG flux studies showed that the skeletal muscle of IVF_8C males had significantly higher glucose uptake compared to both FB and IVF_BL males.

Previously, we showed that animals conceived by IVF demonstrate glucose intolerance compared to in vivo conceived mice in both inbred [19] and outbred [18] mouse models when embryos were cultured with Whitten medium, a more suboptimal culture condition. Our new findings expand on these data and show that even when using an optimal medium, the difference in length of embryo culture results in adult offspring showing phenotypic differences [36–38].

The difference in phenotype in different studies can be explained by different environmental conditions present in a particular cohort of mice or by the stain, sex, and age at which the experiments were performed [39].

Second, we found that male mice conceived following cleavage embryo transfer (IVF_8C) showed signs of left ventricular dysfunction compared to control mice, with reduced ejection fraction, likely secondary to reduced left ventricle shortening and larger internal diameters of ventricles in systole and diastole. Of note, we previously found similar results in an outbred cohort of mice [18].

Importantly, it has been observed that ART-conceived children show an increase in atrial and ventricular septal defects [40] and current clinical guidelines recommend performing echocardiograms in IVF-conceived pregnancies [41].

The mechanisms responsible for cardiac dysfunction in IVF mice and humans are difficult to pinpoint. Pathological left ventricular remodeling can be secondary to heart disease or inflammation [42]. The extent to which the left ventricle is altered may correspond to increased left ventricle volumes and decreased ejection fraction [43]. Although the systolic blood pressure measured by tail cuff showed a lower blood pressure in IVF_8C males compared to IVF_BL males only, lower blood pressure is correlated to decreased ejection fraction [44]. Of note, our past study on DNA methylation and histone modifications in the inner cell mass of IVF-derived mouse embryos revealed activation of the “cardiac hypertrophy” pathway [45].

Third, we found that IVF mice transferred at the blastocyst stage showed reduced locomotor activity. In particular, male IVF_BL showed a 50% decrease in consecutive ambulatory movement (X_AMB) and a 20% reduction in rearing movement (Z_TOT) compared to control mice. Of note, Ecker et al. showed that mouse offspring generated after embryo culture from the two-cell stage to the blastocyst stage had decreased locomotor activity in the open maze test, and they concluded that these mice had reduced anxiety [46]. Our data, although describing behavior in the physiologic environment of a cage, similarly suggest a reduction in overall exploratory behavior.

Alarmingly, IVF_8C male mice showed reduced lifespan compared to the other two groups of male mice. This finding was surprising. Previously, while one study found no differences in lifespan in IVF-conceived mice [47], Rexhaj et al. found that male IVF mice exposed to high-fat diet had a significantly lower lifespan compared to in vivo mice or IVF mice not exposed to a high-fat diet [48]. Interestingly, Rexhaj et al. had also transferred embryos at the cleavage stage. The significance of this finding will need to be confirmed by larger future studies.

Finally, we observed a clear sexual dimorphic effect with male mice generated by IVF showing more phenotypic differences. The finding of sexual dimorphic effects following preimplantation or prenatal stress is well described [49, 50]. For example, following GTTs, female mice show fewer changes than males [51, 52]. In humans, embryo culture or exposure to endocrine disruptors and nutrient restriction limited to the preimplantation period result in sex-specific differences in trophectodermal cell number, gene expression, and placentation [53]. In sheep, a maternal diet low in B vitamins and methionine results in obese and insulin-resistant male offspring [54].

The molecular mechanisms underlying these changes are unknown, but it is hypothesized that differences in gene expression in embryos and fetuses or exposure of offspring to different levels of maternal androgens may be contributors to this phenomenon [55].

Among the unique advantages of our model are the long follow-up in animals and the extensive phenotypic studies performed. Our study has some limitations. First, we performed embryo culture using 20% oxygen due to technical limitations experienced during the pandemic. In the past, we cultured embryos in both 5% and 20% oxygen tension. This is relevant because 5% oxygen has been determined to be beneficial for embryo culture [56]. Second, due to a technical error during the CLAMS run, mice from all groups were exposed to light also during the planned dark hours. Indeed, circadian rhythm is important for regulating metabolic processes and energy [57]. However, since all animals were equally exposed to the lighting issue, we believe that the data still provide valuable information. We also did not use an in vivo control group because we found in past experiments that the litter size following unassisted conceptions is much larger with resulting lower birth weight, a fact that can greatly affect postnatal development [18]. Further, the use of FB control allowed us to account for the effect of superovulation as well as control for the embryo transfer procedures, both of which have been shown to affect adult mouse physiology [58, 59]. Another limitation is that we did not measure blood pressure in female mice. Blood pressure was not measured because of the limited availability of the blood pressure cuff system and the fact that female mice had a less severe phenotype. Regarding statistical analysis, we used the pups as a unit of comparison rather than the litter, given the great cost of the experimental procedures and because this approach has been done by us [18, 19] and others [27, 28]. Lastly, we were unable to perform autopsies in mice that died unexpectedly during the performance of the study, due to partial cannibalizations or because the carcasses were discarded by the veterinary team as soon as they were identified. It is entirely possible that the animals who experienced early death had more severe phenotypes and that the surviving animals we tested later in life do not appropriately reflect the health of the whole cohort (survivor bias). In fact, aging studies are difficult to perform in mice specifically because of the difficulty of predicting a suitable endpoint to understand age effects on outcomes [60].

The results of our study bear similarity to the work of Aljahdali et al. who, however, compared the effects of embryo transfer at the two-cell stage vs. blastocyst stage in mouse offspring [21]. These authors also found a more significant effect on male mice; the only phenotype observed in female mice was heavier weight in the blastocyst embryo transfer compared to the two-cell transfer and the in vivo group. Male mice transferred at the blastocyst stage had higher systolic blood pressure as well as higher serum and lung angiotensin-converting enzyme activity, whereas shorter culture resulted in males with higher insulin levels and increased liver lipid size and accumulation than naturally mated animals. While these results are very relevant, since two-cell embryo transfer is not routinely performed in humans, our experiments offer more clinically relevant results, given that we mimicked the stage of embryo transfers performed during human IVF treatment (eight-cell vs. blastocyst stage). Second, our control group (FB) was designed to control for superovulation, litter size, and embryo transfer, parameters that can affect fetal and postnatal growth. Lastly, we used inbred mice (C57BL6) while different authors used mixed strain (C57BL6xCBA). It is well known that different strains of mice have different responses to similar stimuli: for example, CBA mice show less exploratory behavior, less sociability, and slower habituation to acoustic startle [61].

Human data regarding the long-term outcome of offspring generated after the cleavage stage or blastocyst embryo transfer are limited. However, it is highly likely that IVF conditions and procedures have an influence on adult health. For instance, though the cumulative live birth rates between cleavage and blastocyst-stage transfers are not different, patients who receive a fresh blastocyst transfer show a higher live birth rate and require fewer embryo transfers to result in a birth [62]. However, blastocyst transfer was also found to increase the risk for placenta previa and preterm birth when compared to patients who received a cleavage transfer [63].

The molecular underpinnings of the effect of the duration of culture remain elusive, but it is likely that epigenetic changes secondary to different lengths of embryo culture are responsible for our findings [64]. It is possible that a mismatch between the embryo’s predicted environment and the actual environment experienced in postnatal life is responsible for the observed adult metabolism [65].

In conclusion, in vitro fertilization (IVF) and embryo culture for different lengths of time predispose offspring to adverse cardiovascular and metabolic outcomes, particularly evident in male mice and following cleavage embryo transfer. Future studies following the adult health of human offspring should consider embryo culture length as an important variable to follow.

Author contributions

PR designed the study. PR, RS, SL, RO, XL, OM, and LL performed experiments. RS and PR analyzed the data. RS prepared the figures, tables, and supplemental information. RS, SL, RO, XL, GB, and PR wrote and edited the manuscript.

Data availability

Data are available upon request.