Effects of maternal obesity in an ovine model on metabolic outcomes in F2 adults and F3 neonates

Abstract

Accumulating evidence suggests that indications of metabolic syndrome can be inherited through the germline as a result of maternal obesity. We hypothesized that diet-induced maternal obesity during gestation would program metabolic consequences for multiple generations of offspring, even when first, second, and third generation offspring (F1, F2, F3, respectively) were fed only to requirements. Control (CON) and obese (OB) ewes (generation 0; F0) were bred to a single ram to produce the first generation of offspring (F1). From 60 d prior to conception through term, CONF0 ate 100% National Research Council recommendations (NRC), while OBF0 ewes ate 150% NRC. All F1, F2, and F3 ate 100% NRC after weaning. All mature F1 ewes were bred to a single ram to generate CONF2 (n=6) and OBF2 (n=10). All mature F2 ewes were bred to a single ram to produce CONF3 (n=6) and OBF3 (n=10). OBF2 ewes exhibited greater (P<0.0001) plasma cortisol than CONF2 throughout gestation. A glucose tolerance test at 90% gestation revealed OBF2 ewes had higher (P<0.05) insulin response with similar glucose, resulting in greater (p<0.05) insulin resistance. OBF3 neonates had similar weight, lean mass, and body fat mass to CONF3 neonates. These data suggest that multigenerational programming of adverse metabolic phenotypes occur in association with F0 maternal obesity, yet adiposity may return to CON levels in F3 neonates.

1. Introduction

Obesity is commonly understood as an imbalance of energy in versus energy out, leading to the accumulation of energy stores in the form of adipose tissue. However, we also know that individuals differ substantially in their susceptibility to obesity (1). Therefore, current research is focused on unveiling mechanisms other than energy balance that drive individual differences in weight gain, adiposity, and obesity. Significant progress has been made in understanding a relationship between unfavorable developmental exposures and future disease risk in a phenomenon referred to as the developmental origins of health and disease (DOHaD). The DOHaD hypothesis postulates that altered developmental environments, including those associated with maternal obesity, drive adaptive developmental responses in the fetus (2). Although initially beneficial, fetal adaptations in response to maternal obesity ultimately increase susceptibility to disease, obesity, or related risk factors (2), in a mechanism commonly known as fetal programming (3).

The fetal programming of indications for metabolic syndrome have been well characterized in rodents (4-7), sheep (8-10), non-human primates (11,12), and epidemiologic models (13,14). Additionally, rodent and sheep models have provided evidence that maternal obesity programs metabolic dysregulation in second generation offspring as well (15-18). Rodent studies in particular have established a solid foundation for the understanding of transgenerational metabolic programming effects, and are well represented in the literature (19-24). Those previous studies, and many others, have provided a fundamental understanding of transgenerational programming, allowing us to design an ovine model that is more analogous to a human pregnancy considering rodents’ exhibit different temporal patterns of development, are polytocous, and give birth to altricial offspring. Evidence for transgenerational programming is lacking in precocial, monotocous species, with similar size and temporal patterns of development to humans, making an ovine model ideal for progressing this field of research (25,26).

Importantly, we sought to determine transgenerational programming effects in this study, in addition to multigenerational effects, as previously defined (see review, (27)). Briefly, multigenerational programming consists of phenotypic changes observable in first (F1) and second (F2) generation offspring, as the developing F1 conceptus and its associated germ cells (future F2) are directly exposed to the altered uterine environment in OBF0 ewes. Therefore, hormonal dysregulation and/or hypercaloric environment from maternal obesity results in a direct exposure of OBF1 and OBF2. Since the third generation (F3) is never directly exposed to the zero generation (F0), any phenotypic alterations observed in the F3 are the result of an indirect, transgenerational mechanism. This is in contrast to paternal programming effects, in which transgenerational programming is observed in the F2, since they are never directly exposed to the original insult.

We also sought to produce evidence to help determine whether transgenerational traits, if present, were programmed epigenetically or via serial programming. Assessing the endocrinology of the F2 ewes during gestation provides evidence to address this question, as serial programming is defined as the propagation of a phenotype resulting from repeated exposure to maternal endocrine dysregulation during gestation (28). Absence of endocrine dysregulation in the F2 pregnancy would therefore suggest programming mechanisms are not due to serial programming, but instead epigenetic or other mechanisms (29).

Using our well-established ovine model for diet-induced maternal obesity during gestation (30), we have previously demonstrated multigenerational programming of metabolic syndrome. Specifically, we have previously reported increased appetite, increased adiposity, hyperglycemia, hypercortisolemia, and insulin resistance in both first- and second-generation offspring of obese ewes at multiple different developmental timepoints including prenatal (31), neonatal (8,16), and after maturity (10,15). Interestingly, postnatal phenotypes were often latent until offspring were exposed to a metabolic stress. Throughout numerous studies in this model we have not only identified a variety of indications for metabolic syndrome in OB offspring, but have also demonstrated that they can be exposed at multiple different life-stages. In the current study, we hypothesized that diet-induced maternal obesity during gestation would program adverse metabolic phenotypes in multiple generations of offspring even when F1, F2, and F3 offspring were fed only to requirements, demonstrating transgenerational fetal programming.

2. Materials and methods

2.1. Animals

All animal procedures were approved by the Animal Care and Use Committee at the University of Wyoming and conducted in AALAC accredited facilities. Multiparous Rambouillet/Colombia cross ewes were bred to a single ram and randomly selected to be fed either a control (CON) diet consisting of 100% National Research Council (NRC) recommendations (NRC, 2007) or an obese (OB) diet of 150% NRC from 60 d prior to conception through term. Experimental diets consisted of pelleted rations (88% dry matter, 17.4% crude protein, 93.9% in vitro dry matter digestibility), followed by open access to high-quality alfalfa hay, supplemented with cracked corn during lactation, to meet NRC recommendations for a lactating ewe. All ewes consumed 100% of their pelleted diet each day. All first generation (F1) ewe lambs were housed together and fed 100% NRC recommendations from weaning through life. To obtain the F2 generation, all F1 ewes were bred to a single ram at 18-19 mo of age and maintained on 100% NRC diets. F2 ewe lambs were housed together, maintained on 100% NRC requirements, and bred to a single ram upon reaching 18-19 mo of age. F2 ewe (CONF2; n=6, OBF2; n=10) weights were recorded every two weeks throughout gestation, and their offspring (third generation; F3) weights were recorded within an hour after birth by a single, trained technician. We established a large flock of ewes and carefully selected breeders to continue the experimental lineage while controlling for factors such as sex and birth-type as only singleton or first-born twin ewes were used for F1 and F2 breeding stock. Only first-born twins from a single litter were analyzed in the F3, resulting in n=6 CONF3 (2 male, 4 female) and n=10 OBF3 (5 male, 5 female). Each F3 lamb corresponds to a unique F0 ewe (Figure 1).

Figure 1.

Flow diagram for animals used to generate the F3 generation. Unique rams were used for breeding each new generation. Each F0 female (n=16) corresponds to a single F2 female (n=16), and a single F3 male (n=7) or female (n=9).

2.2. Blood collection

Fasted F2 gestational blood samples were taken at 0700 each month of gestation (d 30, 60, 90, 120, 150) into standard 10 ml vacutainer tubes (BD, Franklin Lakes NJ, USA) for serum and heparinized plasma collection. Heparinized samples were centrifuged immediately after collection for 15 min at 2,500 × g and plasma was aliquoted into sample tubes and frozen at −80° C until utilized. Serum samples were allowed to clot overnight at 4° C and processed similarly to plasma samples the following day. Plasma and serum samples were analyzed to determine concentrations of glucose, insulin, and cortisol.

Intravenous glucose tolerance test (IVGTT)

At 0.9 of gestation (d 135; term 150 d) an IVGTT was performed on 12 randomly selected F2 ewes (n=6 CONF2, n=6 OBF2). All animals were fasted for 24 hr prior to the IVGTT. Six hours prior to the IVGTT, jugular catheters (Abbocath, 14 ga, Abbott Laboratories, North Chicago, IL, USA) were placed in all ewes as previously described (Shasa et al., 2015). Baseline blood samples were collected at −15, −5 and 0 min prior to intravenous administration of dextrose (50% dextrose, Vedco Inc., St Joseph, MO, USA). Bolus administration of dextrose occurred at time point 0, at the dosage of 0.25g/kg of body weight. Blood samples were then collected at 2, 5, 10, 15, 30, 60, and 120 min after dextrose administration. Blood samples were collected, processed as described above, and analyzed for glucose and insulin concentrations.

2.3. Dual-energy X-ray absorptiometry

Dual-energy X-ray absorptiometry (DEXA, GE Lunar Prodigy™ 8743, Madison, WI) scans were performed in newborn lambs on d1 of life as previously described (16). All F3 lambs were immobilized by wrapping in a large towel and scanned. A single experienced technician, blinded to the animal grouping, performed all DEXA scans to determine body fat percentage, lean mass, and fat mass.

2.4. Biochemical assays

Plasma glucose concentrations were determined in triplicate via validated colorimetric hexokinase assays (Liquid Glucose Hexokinase Reagent Set, Point Scientific, Inc., Canton MI, USA) in a 96 well plate as previously described (32). Mean intra-assay and inter-assay coefficient of variation (CV) were 5.89% and 5.41%, respectively. Insulin concentrations were determined in duplicate via a radioimmunoassay (RIA) kit (Porcine insulin kit, Millipore Corporation, Billerica, MA, USA) per manufacturer instructions, with intra-assay and inter-assay CV of 2.58% and 8.05%, respectively, a sensitivity of 1.611 μIU/mL, and a cross-reactivity of 106 ± 4% (33). Serum cortisol concentrations were determined with ImmuChem RIA kits (MP Biomedicals, Solon, OH, USA) per manufacturer instructions, with intra-assay CV of 3.07% an inter-assay CV of 11.7%, and sensitivity of 10 ng/mL (33).

2.5. Statistical analyses

Data were analyzed using the mixed procedure of SAS, (SAS Institute’s Inc., Cary, NC, USA) using an unstructured repeated measures statement when appropriate, including data in figures 2-4. Male and female F3 within the same treatment group were pooled and analyzed using the GLM procedure in SAS to control for sex (Table 1.). Significant differences were determined with p<0.05. Glucose and insulin (AUC) values were determined using GraphPad Prism (GraphPad Software Inc., La Jolla, CA, USA). Data are presented as means ± SEM.

Figure 2.

(A.) Biweekly weights of CONF2 (n=6) and OBF2 (n=10) ewes over gestation. Data are presented as means and standard error of the mean (SEM). (B.) Total gestational weight gain for CONF2 (n=6) and OBF2 (n=10) ewes. ^a,b Means with different letters differ (p<0.05).

Figure 4.

Plasma glucose and insulin dynamics for CONF2 (n=6) and OBF2 (n=10) ewes at late gestation (d 135: 90%) during an intravenous glucose tolerance test (IVGTT). Data are presented as means and standard error of the mean (SEM). Area under the curve (AUC) is shown as a bar graph on the top right for each group. (A.) Insulin dynamics. (B.) Glucose dynamics. *means differ (p<0.05) within a time point between groups.

Table 1.

Neonatal F3 dual-energy X-ray absorptiometry scan results for CONF3 (n=6) and OBF3 (n=10). Data are presented as mean ± standard error of the mean (SEM). *Means with asterisks differ (p<0.05) within a measurement.

	CONF3	OBF3
Body Fat (%)	7.4 ± 1.7	9.2 ± 1.3
Fat Mass (g)	442 ± 102	505 ± 80
Lean Mass (g)	5550 ± 466	4937 ± 364

3. Results

3.1. F2 gestational length, body weights, and average daily gain

Gestational length was similar between CONF2 and OBF2 ewes (153.1 ± 0.8 vs. 152.3 ± 0.7 d, respectively). F2 weights did not differ by treatment group at any time throughout gestation (Figure 2A.). However, F2 gestational weight gain was greater (p<0.05) in OBF2 than CONF2 ewes (24.7 ± 1.4 vs. 19.8 ± 1.4 kg, respectively, Figure 2B.), as was percent weight gain for OBF2 compared to CONF2 ewes (29.7 ± 2.1 vs. 22.5 ± 2.1 % gain, respectively). Average daily gain was greater (p<0.05) in OBF2 than CONF2 ewes (0.162 ± 0.009 vs. 0.129 ± 0.009 kg/d).

3.2. Endocrine and metabolic changes.

No treatment differences were found between baseline fasted serum concentrations of insulin or glucose at any point of F2 gestation (Figures 3A & 3B). Serum cortisol was greater (p<0.0001) in OBF2 ewes in comparison to CONF2 ewes throughout gestation, averaging 22.0 ± 1.8 and 9.5 ± 2.4 ng/mL, respectively (Figure 3C). During the late gestation (d135, 90%) IVGTT, fasted glucose and insulin values were similar between CONF2 and OBF2 ewes prior to dextrose administration (40 ± 14 vs 43 ± 14 mg/dL and 1.8± 3.9 vs 4.8 ± 3.2 μIU/mL, respectively). After bolus administration of glucose, OBF2 ewes exhibited greater (p<0.05) plasma concentrations of insulin at 5 and 10 min, and greater (p<0.05) glucose concentrations at 10 min (Figure 4A and 4B.). Insulin AUC during the IVGTT was higher (p<0.05) in OBF2 relative to CONF2 ewes (809 ± 125 vs. 446 ±137 μIU/mL*min, respectively), while glucose concentrations were similar between groups (13397 ± 895 vs. 14198 ± 895 mg/dL*min, for OBF2 and CONF2, respectively).

Figure 3.

Monthly plasma analyses for CONF2 (n=6) and OBF2 (n=10) ewes over gestation. Data are presented as means and standard error of the mean (SEM). (A.) Fasted F2 plasma insulin through gestation. (B.) Fasted F2 plasma glucose through gestation. (C.) Fasted F2 plasma cortisol through gestation. *means differ (p<0.05) between groups at a time point.

3.3. F3 body composition

DEXA scans failed to expose any treatment differences in % body fat, fat mass, or lean mass.

4. Discussion

Similar to previous studies in this model (see review, (34)), these data show that diet-induced maternal obesity prior to and throughout gestation in the F0 induces multigenerational programming of adverse metabolic phenotypes. Importantly, these programming effects are observed even though all subsequent generations (F1-F3) ate 100% NRC, and are therefore not on an obesogenic diet. Specifically, we have demonstrated insulin/glucose dysregulation, increased adiposity, hypercortisolemia and hyperphagia in OBF1 and OBF2 ewes at later stages of life (10,15,16,31). However, adverse phenotypes were latent in mature OBF1 and OBF2 until subjected to a metabolic stress such as an ad libitum feeding trial, where animals were allowed unlimited access to a pelleted diet for 12 weeks. In the feeding trial studies, OB offspring demonstrated increased appetite, increased weight gain, and endocrine dysregulation relative to controls (9,10). Interestingly, the current data suggest that gestation is a sufficient metabolic stressor that can expose the programmed metabolic phenotypes in OBF2 females similar to what we have observed after the feeding trials, as gestating OBF2 ewes demonstrated insulin/glucose dysregulation, hypercortisolemia, and increased wait gain over gestation. Similar findings were observed previously in OBF1 ewes at mid-gestation (Table 2).

Table 2.

Generational comparisons of programmed traits from obese lineages relative to control lineages.

Ewe gestation	F0 ewes		F1 ewes		F2 ewes
fasted glucose	+	[31]	+	[16]	+
fasted insulin	+	[31]	+	[16]	+
plasma cortisol	+	[33]	NR	[N/A]	+
GTT insulin (late gestation, d135)	+	[8,31]	+	[16]	+
GTT glucose (late gestation, d135)	+	[31]	+	[16]	+
weight	+	[8]	ND	[16]	+
Neonatal lambs	F1 lambs		F2 lambs		F3 lambs
birth weight	ND	[8]	ND	[16]	ND
fat mass	NR	[N/A]	+	[16]	ND

“+” indicates an increase (p < 0.05) and “−” indicates a decrease (p < 0.05) in OB relative to controls, while “ND” indicates no difference reported and “NR” indicates values were not previously reported. References are provided for each finding within the same column. Data for F2 ewes and F3 neonatal lambs are drawn from the present study.

The elevated gestational cortisol observed in OBF2 ewes suggests that indications of metabolic syndrome may also be programmed in OBF3. Elevated glucocorticoids during development have been identified as a primary mechanism of fetal programming (See review: (35)), and models of gestational hypercortisolemia show similar metabolic outcomes in offspring to maternal obesity models (36-39). Specifically, Long and colleagues investigated fetal programming effects of excess cortisol by administering synthetic glucocorticoids to ewes at late gestation which resulted in multigenerational programming outcomes similar to those observed with maternal obese models (40). These data are supported by previous findings in rodent models reporting transgenerational programming from synthetic glucocorticoid exposure. In studies assessing both the maternal and paternal lineage, programming effects were observed transgenerationally but seemingly resolved by the F3 in the paternal line (41). Our data also correspond to multigenerational programming studies from rodent overnutrition, where adverse metabolic phenotypes were observed through the F2 generation (42).

The observed patterns of hypercortisolemia support a possible a serial programming mechanism, where elevated OBF0 cortisol adversely programs OBF1 (40). The OBF1 then exhibits hypercortisolemia in response to pregnancy, therefore inducing serial programming of the OBF2 (16). As demonstrated by the current study, OBF2 pregnancies are also characterized by hypercortisolemia, continuing this pattern and predisposing OBF3 offspring to adverse metabolic programming. In addition, F2 ewes exhibited increased weight gain over gestation, which has been previously reported as a risk factor for adverse outcomes in offspring (43). Together, the hypercortisolemia, glucose/insulin dysregulation, and increased weight gain exhibited in OBF2 ewes during gestation support the presence of a serial programming mechanism. Importantly, this does not rule out other potential mechanisms of transgenerational programming, such as epigenetics.

Body compositions were similar between CONF3 and OBF3 neonates. Similar body composition between F3 treatment groups could be attributed to either a normalization of phenotype as the pedigree has progressed, the lack of programming as OBF3 were not directly exposed to the OBF0 gestational environment, or the possibility that programmed phenotypic differences will not emerge until OBF3 are exposed to a metabolic stressor/challenge. Importantly, OBF1 and OBF2 offspring also demonstrated similar body weight at birth, and still went on to develop adverse metabolic phenotypes. Since OBF3 were exposed to elevated OBF2 maternal cortisol throughout gestation, we suspect that some programmed phenotypes will remain latent and emerge later in life. Future studies should be directed at elucidating the phenotypes observed following ad libitum feeding, pregnancy, or other metabolic stresses, in OBF3.

It is important to consider the strengths and limitations of this study. As for the strengths, this is a long-term study that required significant time and resources to accomplish. Further, it was completed in a large, precocial specie, with similar size and temporal development to humans, making it ideally suited for translation to future studies and clinical considerations. Also, breeding strategies were implemented so that F3 numbers reflect the original F0. That is, for every F3 lamb represented in this paper, there is a sole F0 correspondent. Every effort was made to remove or reduce confounding factors such as litter selection, birth type (singleton, twin, etc.,), and so on. However, this resulted in limitations that should not be overlooked. The F3 number is low and the distribution between sexes was not equal.

In summary, this study provides evidence that gestation is a significant metabolic stressor that can unveil adverse metabolic phenotypes in OBF2 ewes, even when these ewes are maintained on a control diet throughout life. Elevated circulating cortisol and insulin resistance during OBF2 gestation provides evidence for serial programming mechanisms that can additionally program adverse phenotypes in the OBF3. It is important to consider that this study only assessed limited F3 phenotype during the neonatal period, and that there may be evidence of endocrine dysregulation at other levels (tissue, cellular, molecular, etc.) that have yet to be identified as they are outside the scope of the current paper. Future studies with a focus on OBF3 phenotypes in response to metabolic stressors are important to determine the severity of transgenerational programming from maternal obesity.

Highlights.

• Maternal obesity programs diminishing metabolic phenotypes over 3 generations.

• Gestation exposes latent phenotypes in second generation offspring of obese ewes.

• Endocrine dysregulation in 2nd generation pregnancy suggests serial programming.