Abstract
Maternal caloric restriction during late gestation reduces birth weight, but whether long-term adverse metabolic outcomes of intra-uterine growth retardation (IUGR) are dependent on either accelerated postnatal growth or exposure to an obesogenic environment after weaning is not established. We induced IUGR in twin-pregnant sheep using a 40% maternal caloric restriction commencing from 110 days of gestation until term (∼147 days), compared with mothers fed to 100% of requirements. Offspring were reared either as singletons to accelerate postnatal growth or as twins to achieve standard growth. To promote an adverse phenotype in young adulthood, after weaning, offspring were reared under a low-activity obesogenic environment with the exception of a subgroup of IUGR offspring, reared as twins, maintained in a standard activity environment. We assessed glucose tolerance together with leptin and cortisol responses to feeding in young adulthood when the hypothalamus was sampled for assessment of genes regulating appetite control, energy and endocrine sensitivity. Caloric restriction reduced maternal plasma glucose, raised non-esterified fatty acids, and changed the metabolomic profile, but had no effect on insulin, leptin, or cortisol. IUGR offspring whose postnatal growth was enhanced and were obese showed insulin and leptin resistance plus raised cortisol. This was accompanied by increased hypothalamic gene expression for energy and glucocorticoid sensitivity. These long-term adaptations were reduced but not normalized in IUGR offspring whose postnatal growth was not accelerated and remained lean in a standard post-weaning environment. IUGR results in an adverse metabolic phenotype, especially when postnatal growth is enhanced and offspring progress to juvenile-onset obesity.
Introduction
There is increasing evidence to support the early life programing of adult obesity, type 2 diabetes, and hypertension. The influence of prenatal environment depends on organ-specific windows of susceptibility with some, but not all, outcomes linked to mechanisms affecting the size at birth (Barker 1997, Roseboom et al. 2000). In large mammals, including sheep, pigs, and humans, chronic caloric restriction throughout late gestation results in intra-uterine growth retardation (IUGR) (Roseboom et al. 2000, Symonds et al. 2009), which contrasts with suboptimal maternal nutrition in earlier gestation, which does not influence birth weight (Roseboom et al. 2000, Bispham et al. 2003, Sharkey et al. 2009).
During diet-induced IUGR, maternal homeostasis is altered, affecting the metabolic environment in which the fetus develops (Tygesen et al. 2008). It is possible that these metabolic adaptations influence fetal growth independently of changes in the feto-maternal endocrine environment. Not only nutrients fulfill energetic requirements but also a range of lipids, non-esterified fatty acids (NEFAs), and amino acids act as signaling molecules and could therefore influence epigenetic processes linked to the long-term regulation of metabolic function (McMillen & Robinson 2005). In this study, we examined whether a reduction in maternal food intake in late pregnancy, leading to changes in maternal metabolic homeostasis and birth weight, is essential in programing adult predisposition to the following characteristics: i) altered body composition, ii) central and peripheral insulin resistance, iii) the regulation of food intake, and iv) postprandial and postabsorptive endocrine responses to feeding. Furthermore, the extent to which the impact of IUGR on each of these adaptations is dependent on postnatal growth patterns is not known. This is important as many human studies indicate that the long-term impact of reduced birth size, in both term and preterm infants, can be dependent on early postnatal growth (Singhal et al. 2003, Stettler et al. 2003). In sheep, the relative importance of enhanced postnatal growth on long-term outcomes has not been widely examined, although, when combined with IUGR, accelerated postnatal growth differentially affects energy sensing within the stomach and hypothalamus (Sebert et al. 2011). The post-weaning environment is an additional factor that appears to determine the magnitude of the phenotypic response to alterations in maternal diet in pregnancy in rodents (Desai et al. 2007). In these studies, the metabolic and related effects in young adult offspring who were nutritionally manipulated in utero are minimal unless adiposity has been promoted. The extent to which similar changes in body composition also apply to large mammals has not been investigated.
Each organ has set a developmental trajectory and therefore they are not all similarly affected by IUGR. The hypothalamus is particularly sensitive to environmental stresses in early life and plays a central role in the regulation of energy homeostasis (Adam et al. 2008). Cortisol can influence blood pressure as a consequence of regulating gene expression of arginine vasopressin (AVP) and corticotropin-releasing hormone (CRH), which when suppressed, acts through negative feedback, to reduce cortisol secretion from the adrenals through decreased adrenocorticotropic hormone action (Lightman 2008). Food intake is regulated by changes in the plasma concentration of markers of energy status, including insulin, leptin, and glucose (Schwartz et al. 2000). These factors determine the action of neurotransmitters in the arcuate nucleus of the hypothalamus, especially neuropeptide Y (NPY) and pro-opiomelanocortin (POMC), which have antagonistic actions in signaling peripheral energy status to other hypothalamic nuclei, the cortico-limbic system, and the brain stem, which ultimately determine food intake and physical activity (Schwartz et al. 2000). Critically, the fetal hypothalamus shows an orexigenic response by increased NPY signaling to maternal nutrient restriction in late gestation (Warnes et al. 1998) and an anorexigenic response by increased POMC signaling to maternal and early postnatal overnutrition (Muhlhausler et al. 2006). However, whether these adaptations persist into adulthood is not known. In addition, while organogenesis and the developmental maturation of the hypothalamus in altricial species are particularly sensitive to the late gestational nutritional environment (Adam et al. 2008), hypothalamic maturation continues after birth when it is particularly responsive to the postnatal energetic environment (Paus 2010). Whether specific changes in the early postnatal environment during key windows, i.e., immediately after birth and at weaning, modulate the long-term molecular adaptation of the hypothalamus to IUGR has received no attention. This study, therefore, not only examines the effects of maternal caloric restriction during late pregnancy on maternal homeostasis but also tests the hypothesis that the adverse effects of IUGR are dependent on the postnatal energetic environment and concomitant differences in peripartum or post-weaning growth. We investigated the effects of: i) IUGR followed by an accelerated postnatal growth combined with a low-activity obesogenic environment after weaning and compared with those offspring born to mothers fed to requirements throughout pregnancy and with the same postnatal treatment, ii) differing postnatal growth rates on adult IUGR offspring submitted to obesogenic conditions, and finally, and iii) a differing energetic environment on IUGR offspring submitted to regular postnatal growth rate.
Materials and methods
Animals and experimental design
All animal procedures were performed in accordance with the UK Animals (Scientific Procedures) Act 1986 with approval from the Local Ethics Committee of the University of Nottingham. The experimental nutritional intervention has previously been described in detail (Sebert et al. 2011). In brief, 28 Bluefaced Leicester cross Swaledale twin-bearing sheep (ovis) were individually housed at 100 days of gestation (dGA) and, at 110 dGA, randomly allocated to the experimental groups (Fig. 1). All pregnancies continued normally until term (∼145±1 days) and produced heterozygous twins. They included a control group (C, n=9) that were fed to requirements through pregnancy (i.e., from 0.46 MJ/kg×BW0.75 at 110 dGA, increasing to 0.72 MJ/kg×BW0.75 at 130 dGA), while the remaining 19 mothers were calorie restricted (R) and were pair-fed to 60% of control intake, based on their body weight. All mothers were individually weighed once a week before feeding in order that their total food requirements could be adjusted. From birth, the offspring born to C mothers were then reared to promote accelerated (A) early postnatal growth (CA, n=8, four males and four females), accomplished by only one twin being reared by its mother. The offspring born to R mothers were reared to promote an accelerated (RA n=9, two males and seven females) or a regular (RR, n=17) early postnatal growth rate, accomplished by being reared together as twins. After weaning, all offspring were kept in a low-activity environment until 17 months of age in order to promote obesity (O, six animals on 19 m2, fed on straw nuts and a micronutrient supplement ad libitum) with the exception of nine RR offspring that were kept in a normal-physical-activity environment, in order to remain lean (RRL, n=9, five males and four females, six animals on 1125 m2, could access grass and a micronutrient supplement ad libitum; RRO n=8; two males and six females; Fig. 1). Discrepancies between the total number (n) of mothers and offspring are due to additional offspring for independent intervention groups for the twins that were removed from their mother on the first day of birth. This included formula-reared twins within the CAO and RAO offspring, which were not included in this study. The numbers of twin-bearing mothers entered into the study for each nutritional group were predicted to be sufficient to produce enough numbers of male and female offspring for each of the postnatal intervention groups. However, due to the uneven distribution of males and females born to R mothers, there were fewer male offspring available than anticipated. The resulting groups permit us to draw comparisons between animals with and without IUGR (RAO vs CAO) and, within those with IUGR, to investigate the effects of early postnatal growth (RAO vs RRO) and post-weaning environment (RRO vs RRL).
Timing of samplings and in vivo challenges
Maternal blood sampling
At 130 dGA, jugular venous blood samples (5 ml) were collected from the ewes in the morning, before, and 2 h after, feeding. Venous blood was collected into heparinized or K+EDTA-coated tubes and the plasma was immediately separated by centrifugation (2500 g×10 min at 4 °C) and stored at −80 °C until analysis.
Offspring blood sampling
Venous blood samples (prepared and stored under identical conditions as described earlier) were collected after an overnight fast (≥18 h) at both 7 and 16 months of age. Jugular catheters were inserted by percutaneous venipuncture 1–2 days before sampling. Additional blood samples were collected at 16 months of age following the presentation of a mix of high- and low-energy-dense feed (3 kg straw nuts, 8.5 MJ/kg and 800 g concentrate pellets, 12.5 MJ/kg) to study the postabsorptive and postprandial response at 2, 4, 8, and 24 h after feeding.
Determination of insulin sensitivity
Glucose tolerance tests (GTTs) were undertaken on all offspring at 7 and 16 months of age in which jugular vein catheters had been previously inserted and the area under the curve (AUC) calculated. Animals were fasted overnight (≥18 h) and injected intravenously with 0.5 g/kg glucose. Glucose and insulin concentrations were measured in plasma samples before and at 10, 20, 30, 60, 90, and 120 min, after the i.v. glucose (Gardner et al. 2005). The homeostatic model assessment for insulin resistance (HOMA-IR) index was calculated by multiplication of glucose (mmol/l) and insulin (μg/l) concentrations measured in fasted plasma (Wallace et al. 2004).
Determination of body composition, physical activity, and food intake at 16 months of age
Total body fat was determined when the animal was sedated (i.m. injection of 1.5 mg/kg ketamine with 0.1 mg/kg xylazine) and scanned in a transverse position using a Lunar DPX-L (fast-detail whole body smartscan, GE Healthcare, Little Chalfont, Buckinghamshire, UK). The level of spontaneous physical activity in adulthood in their respective environments was determined using uniaxial accelerometers (Actiwatch; Linton Instrumentation, Diss, Norfolk, UK). Average total food intake was measured in 24 h intervals over a 10-day period with all animals kept in individual pens and they could access feed, straw nuts (8.5 MJ/kg), and concentrate pellets ad libitum (12.5 MJ/kg).
Postmortem procedures and hypothalamic collection
At 17 months of age, all offspring were killed by electrical stunning and exsanguination after an overnight fast. The entire hypothalamus was dissected according to anatomic landmarks (Sebert et al. 2009), snap frozen, and stored at −80 °C until analyzed. The use of entire hypothalamus allows analysis of the entire hypothalamic response but cannot be extrapolated to responses that would require nuclei-specific analyses.
Laboratory analysis
Plasma metabolites and hormones
Plasma glucose, triglycerides, and NEFAs were measured by colorimetric assays (Randox, Crumlin, Co. Antrim, UK). Insulin was assayed using an ovine specific ELISA assay (Mercodia, Diagenics Ltd, Milton Keynes, Buckinghamshire, UK). Leptin (Delavaud et al. 2000) and cortisol (DPC coat-a-count, Siemens, Camberley, UK) were determined by RIA.
Analysis of the plasma metabolome
Fasted heparin-treated plasma samples taken from mothers at 130 dGA were analyzed for a wide spectrum of metabolites by liquid chromatography coupled to high-resolution mass spectrometry (LC–HRMS). Plasma was defrosted on ice and filtered by centrifugation (Nanosep Omega, Pall, Port Washington, NY, USA) to remove high-molecular-weight species, proteins in particular (over 10 kDa). Metabolomic LC–HRMS profiles were acquired from 15 μl of each filtered serum sample using an Agilent 1200 HPLC system equipped with a 150×2.1 mm Uptisphere HDO-C18 column with 3 μm particle size (Interchim, Montluçon, France) coupled to a high-resolution LTQ-Orbitrap hybrid mass spectrometer (Thermo Fisher Scientific, Bremen, Germany) fitted with an electrospray source operated in the positive ion mode. The detailed conditions applied both for the HPLC separation and mass spectrometric signal acquisition have been described previously (Courant et al. 2009, Alexandre-Gouabau et al. 2011). Quality control standards and samples were randomly included five times into the sequence of injection.
Metabolomic data processing
Open-source XCMS software (Smith et al. 2006) was used for non-linear alignment of the generated raw data and automatic integration and extraction of the signal intensities measured for each mass-retention time ((m/z; rt)) feature constituting these metabolomic fingerprints, which each represent one ion. The XCMS parameters were implemented with the algorithm ‘match-filter’ using default settings except for the interval of m/z value for peak picking which was set at 0.1, the noise threshold set at six, the group bandwidth set at ten, and the minimum fraction set at 0.5 as described previously (Robinson & Williamson 1980). After XCMS processing, the signal abundances observed for identical ions in two groups of samples were statistically analyzed, and annotation and the subsequent identification of putative metabolites of interest were achieved using an in-house reference databank (Tygesen et al. 2008).
Gene expression measurements
Offspring hypothalami were homogenized and RNA isolated, using the RNeasy Plus kit (Qiagen). An aliquot of 4 μg of RNA was reverse transcribed with the High Capacity RNA-to-cDNA kit (Applied Biosystems). The resulting cDNA was amplified in a real-time thermocycler (Quantica, Techne, Burlington, NJ, USA) using a SYBR green system in the Taq polymerase reaction mix (ABsolute blue QPCR SYBR green, Thermo Scientific, Epsom, UK). Specificity of primers was confirmed by sequencing PCR product (Supplementary Information Table 1, see section on supplementary data given at the end of this article). Hypothalamic gene expression was assessed for the following pathways: i) orexigenic neurotransmitters: NPY and agouti-related peptide (AgRP); ii) insulin and leptin signaling: protein tyrosine phosphatase non-receptor type 1 (PTP1B (PTPN1)), suppressor of cytokine signaling 3 (SOCS3), insulin receptor (IR (INSR)), and leptin receptor (obRb (LEPR)); iii) intracellular energy signaling: AMP-activated kinase (AMPKA2 (PRKAA2)), mammalian target of rapamycin (mTOR), and fat mass and obesity-associated gene (FTO); and iv) cortisol regulation: glucocorticoid receptor (GCR (NR3C1)), CRH, and AVP. rRNA 18S showed a stable expression and was used as a housekeeping gene. Gene expression was calculated using the 2−ΔΔCt method (Livak & Schmittgen 2001).
Statistical analysis
Metabolomic data
All multivariate data analyses and modeling were performed using the SIMCA-P+software (v 12, Umetrics, Inc., Umeå, Sweden) on log-transformed (van den Berg et al. 2006) and Pareto-scaled (Cloarec et al. 2005) data as described previously (Alexandre-Gouabau et al. 2011). The susceptibility of the metabolic phenotypes of the mothers to caloric restriction in late pregnancy was assessed using a supervised method, partial least squares discriminant analysis (PLS-DA), which was applied to the transformed dataset to reveal the potentially existing discrimination between sample groups to be compared within the dataset, and to point out the variables more importantly involved in this discrimination. PLS-DA was combined with a multivariate preprocessing filter called orthogonal signal correction (OSC). By removing within-class variability and confounders that may interfere with chemometric analysis, such as LC–MS technical variability, OSC can significantly improve PLS-DA performance, yielding a better discrimination of the clusters (Wagner et al. 2006). The quality of the generated OSC-PLS-DA model was classically evaluated by several goodness-of-fit parameters and criteria including: R2 (X), the proportion of the total variance of the dependent variables that is explained by the model; R2 (Y), defining the proportion of the total variance of the response variable (i.e., the class of the samples) explained by the model; and the predictive ability parameter Q2 (Y), which was calculated by a seven-round internal cross-validation of the data. In addition, a permutation test (n=100) was carried out to validate, and test, the degree of over-fitting for OSC-PLS-DA models. The score values from OSC-PLS-DA were subjected to ANOVA to test the model and the validation was considered successful with P<0.01. The variables that discriminate the metabolic signatures most significantly were pinpointed by their loadings on PLS-DA.
For non-metabolomic outcomes
Statistical analysis of the data was performed using the PASW statistics software (v 17.02, IBM, Chicago, IL, USA). The Kolmogorov–Smirnov tests were realized on every parameter analyzed to determine the Gaussian distributions of the variables. The influence of maternal nutrition (CAO vs RAO), early postnatal growth (RAO vs RRO), and obesogenic environment (RRL vs RRO) were determined, according to parametric distribution, using ANOVA with a pairwise a priori test or the Mann–Whitney U tests. Of the metabolites and hormones measured over 24 h, changes over time were tested using a paired t-test. Data are expressed as mean values with their standard errors. To address the limitations of multiple testing, statistical trend was accepted with a 95% CI (P<0.05) and significance was accepted with a 99% CI (P<0.01). Correlations were tested with non-parametric Spearman's test and slope of the correlation was reported on the linear fit. Each variable was tested for sex. Body weight and fat mass are known to differ, in absolute scale, between male and female sheep (Bloor et al. 2013) thus sex-specific Z-score transformation was used before analyses. Specifically, we saw no indication of a difference in male and female offspring in glucose homeostasis, which is consistent with earlier studies (Gardner et al. 2005). Moreover, comparison for each variable between groups for females only demonstrated similar outcomes, although without reaching statistical significance, hence data for each sex were combined for further analyses and greater statistical power.
Results
Mothers
Relative to their weight at the beginning of the caloric restriction, R mothers gained less weight up to term compared with C mothers (Fig. 2). At 130 dGA, plasma glucose was reduced in fasted R mothers but there was a greater increase after feeding (Table 1). Plasma NEFA concentrations were higher in fasted R mothers but did not differ between groups after feeding, while plasma triglycerides, cortisol, insulin, and leptin were unaffected by maternal diet (Table 1). Metabolomic analysis showed a specific biological signature associated with the calorie-restricted mothers, with a strong overall difference between the groups (OPLS-DA model of all the 2629 (m/z, rt) features detected, using two latent factors for maternal plasma metabolomic profiles (describing 43% of variable information); C mothers n=7 and R mothers n=17; validation parameters: R2X (cum)=0.434, R2Y (cum)=0.999, Q2 (cum)=0.994, permutation test (n=100) with R2 intercept=0.331 and Q2 intercept=−0.333, ANOVA P value=9.9×10−19), and of the 2629 detected features constituting these metabolomic profiles, 133 differed significantly (P<0.01). Of these, 95 were upregulated with a fold change (expressed as a ratio of the mean abundance in R group compared with the mean abundance in the C group) of >1.4 and only seven were downregulated with a fold change of <0.71. Owing to species-specific technical constraints, only five of these compounds could be precisely identified (Table 2) as phenylalanine, tryptophan, and three forms of O-acetyl-carnitine, which were all upregulated in R mothers.
Effect of maternal diet in late gestation commencing at 110 days of gestation on plasma endocrine and metabolic characteristics. Plasma was sampled from mothers at 130 days of gestation and concentrations of metabolites and hormones, determined immediately before and 2 h after feeding. Twin-pregnant sheep were either fed to requirements (C, n=9) or pair-fed to 60% of that amount from 110 days of gestation (R, n=19). Fold change of each feature is reported as mean±s.e.m. of the abundance for calorie-restricted mothers relative to controls
Variables | Control | Restricted |
---|---|---|
Glucose (mmol/l) | ||
Fasted | 3.6±0.6 | 2.3±0.2* |
2H-fed | 4.3±0.9 | 7.0±0.4† |
Change | 1.4±1.0 | 4.7±0.5† |
NEFAs (mmol/l) | ||
Fasted | 0.34±0.05 | 0.85±0.08‡ |
2H-fed | 0.74±0.12 | 0.65±0.09 |
Change | 0.50±0.05 | −0.31±0.12‡ |
Triglycerides (mmol/l) | ||
Fasted | 0.32±0.01 | 0.33±0.01 |
2H-fed | 0.30±0.02 | 0.32±0.01 |
Change | −0.02±0.03 | −0.01±0.02 |
Fasted insulin (μg/l) | 0.3±0.01 | 0.3±0.01 |
Fasted leptin (ng/ml) | 1.5±0.3 | 1.0±0.1 |
Fasted cortisol (nmol/l) | 17±3 | 22±3 |
*P<0.05; †P<0.01; ‡P<0.001.
Effect of maternal diet in late gestation commencing at 110 days of gestation on plasma endocrine and metabolic characteristics. Plasma was sampled from mothers at 130 days of gestation and metabolites in prefeeding samples that were identified by metabolomic fingerprinting to have significantly changed with maternal diet (Mann–Whitney U test, P<0.01). Twin-pregnant sheep were either fed to requirements (C, n=9) or pair-fed to 60% of that amount from 110 days of gestation (R, n=19). Fold change of each feature is reported as mean±s.e.m. of the abundance for calorie-restricted mothers relative to controls
Metabolites (LC–HRMS) | M | Fold change | P |
---|---|---|---|
O-acetyl-carnitine [M+H]+ | 203.11 | 1.8±0.2 | 0.0003 |
O-acetyl-carnitine [M+Na] | 203.11 | 2.7±0.4 | 0.0008 |
O-acetyl-carnitine [2M+H]+ | 203.11 | 2.9±0.3 | 0.0011 |
Tryptophan [M-NH3+H]+ | 204.09 | 2.9±0.3 | 0.0011 |
Phenylalanine [M+H]+ | 165.08 | 1.2±0.1 | 0.0085 |
M, monoisotopic mass; all compounds were identified using authentic standards.
Offspring
The primary characteristics of each offspring group over the study are summarized in Table 3. Offspring of R mothers were smaller at birth and, when subjected to an intervention of accelerated early postnatal growth (RA), gained weight faster before weaning than either CA or RR groups, suggesting that nutrient restriction during late pregnancy did not diminish milk production. By 7 months of age, body weight was similar between groups. At 16 months of age, as expected, physical activity was higher in those offspring kept in an unrestricted environment (RRL vs RRO) but did not differ between those maintained within an obesogenic environment. RRL animals were further smaller and consumed less feed each day.
Influence of fetal intrauterine growth restriction, accelerated postnatal growth, and obesity on offspring body weight and on adult body composition, physical activity, and food intake as measured at 16 months of age. Offspring of C and R mothers were subjected to an accelerated (CA, n=8; RA, n=9) or regular growth (RR, n=17) during lactation. After weaning, offspring were then exposed to an obesogenic environment (O) with the exception for a subset of RR, which remained lean (RRL, n=9; RRO, n=8). Values are expressed as mean±s.e.m.
Maternal diet | Control | Nutrient restricted | ||
---|---|---|---|---|
Growth from birth to weaning | Accelerated | Restricted | ||
Phenotype from weaning | Obese | Lean | ||
Group | CAO | RAO | RRO | RRL |
Birth weight (kg) | 5.0±0.2‡ | 4.0±0.1§ | ||
Weight gain, 0–81 days (kg) | 6.6±0.3‡ | 7.5±0.5§ | 6.1±0.3‡ | |
Weight 7 months (Z-score) | 0.44±0.35 | 0.27±0.31 | −0.07±0.16 | −0.60±0.33 |
Weight 17 months (Z-score) | 0.65±0.17 | 0.37±0.25 | 0.38±0.32‡ | −1.24±0.12§ |
Relative body fat (Z-score) | 0.04±0.29 | 0.37±0.38 | 0.12±0.34 | −0.52±0.31 |
Physical activity (counts/24 h) | 121±13 | 166±16 | 178±54‡ | 550±40§ |
Food intake (MJ/kg per day) | 0.32±0.01 | 0.32±0.01 | 0.29±0.04* | 0.26±0.02† |
Fasted TG (mg/dl) | 0.18±0.02* | 0.12±0.03† | 0.12±0.02 | 0.16±0.02 |
Fasted NEFAs (mmol/l) | 0.65±0.06 | 0.49±0.07 | 0.48±0.08 | 0.38±0.05 |
Significant differences between groups are represented by different superscripts, * vs †P<0.05; ‡ vs §P<0.01. NEFAs, non-esterified fatty acids; TG, triglycerides.
Insulin sensitivity
At 7 months of age, the glucose AUC during the GTT was higher in obese animals compared with lean animals (RRO vs RRL, Fig. 3 and Table 4). RAO offspring showed twice the insulin response to a standard glucose challenge in comparison to RRO. By 16 months of age, glucose AUC did not differ between groups but the insulin response during the GTT was significantly higher in RAO compared with CAO (P<0.05), as was the HOMA-IR, an index of insulin resistance, which was also higher in RAO than in the RRO offspring (P<0.01).
Influence of fetal intrauterine growth restriction, accelerated postnatal growth, and obesity on the onset of insulin resistance in the offspring. Plasma glucose and insulin responses to an i.v. glucose injection at 7 (i.e., puberty) and 16 months (i.e., young adulthood) of age. Offspring of C and R mothers were subjected to an accelerated (CA, n=8; RA, n=8) or regular growth (RR, n=15) during lactation. After weaning, offspring were then exposed to an obesogenic environment (O) with the exception for a subset of RR, which remained lean (RRL, n=8; RRO, n=7). Values are expressed as mean±s.e.m.
CAO | RAO | RRO | RRL | |
---|---|---|---|---|
At 7 months of age | ||||
Glucose AUC (mmol/l) | 1167±46 | 1070±74 | 1252±93* | 1015±77† |
Insulin AUC (μg/l) | 45±12 | 67±13* | 37±7† | 23±6 |
HOMA-IR | 0.61±0.05 | 0.66±0.07 | 0.52±0.08 | 0.70±0.21 |
At 16 months of age | ||||
Glucose AUC (mmol/l) | 1302±36 | 1226±60 | 1215±78 | 1148±46 |
Insulin AUC (μg/l) | 53±14* | 94±14† | 67±15 | 22±5 |
HOMA-IR | 0.92±0.08‡ | 1.17±0.05§ | 1.00±0.06‡ | 0.99±0.03 |
Significant differences between groups are represented by different superscripts, * vs †P<0.05; ‡ vs §P<0.01.
Effect of feeding on plasma profiles of leptin and cortisol
Before the feeding challenge, plasma leptin was higher in RAO than in RRO (Fig. 4A). In the RAO group, plasma leptin initially declined on feeding (P<0.05), to gradually increase between 8 and 24 h after feeding (P<0.05). This effect was not observed in any other group, as plasma leptin remained unchanged. Plasma glucose and insulin differed between RRO and RRL animals during the 24 h of measurements (P<0.05) but not between groups raised in an obesogenic environment (Fig. 4B and C). Plasma cortisol both peaked 4 h after feeding and was highest in RAO offspring compared with CAO offspring, a difference that persisted until at least 8 h after feeding (Fig. 4D).
Hypothalamic gene expression
Expression of appetite regulatory genes was unchanged, while AMPKA2, mTOR, and FTO were all higher in RAO groups compared with CAO groups (Table 5). A statistically significant negative correlation between anorexigenic circulating hormones insulin and leptin and the expression of orexigenic genes NPY and AgRP was observed in RAO and RRO groups, but not in CAO or RRL groups. Taken together, these different relationships suggest a potential change in insulin and leptin sensitivity within the hypothalamus after IUGR (Table 6). Gene expression of both AVP and CRH was higher in RAO offspring when compared with CAO. Expression of NPY was three times higher and that of PTP1B, AMPKA2, mTOR, and GCR was lower in the RRL offspring when compared with RRO. Importantly, postnatal growth rate (RAO vs RRO) did not have any effect on hypothalamic gene expression in any of the pathways investigated.
Effect of maternal caloric restriction, accelerated postnatal growth, and juvenile-onset obesity on the regulation of energy balance and endocrine sensitivity in the hypothalamus of young adults. Offspring of C and R mothers were subjected to an accelerated (CA, n=5; RA, n=8) or regular growth (RR, n=15) during lactation. After weaning, offspring were then exposed to an obesogenic environment (O) with the exception for a subset of RR, which remained lean (RRL, n=8; RRO, n=7). Values are expressed as mean±s.e.m. and n=5–8 per time point
Pathways | Genes | CAO (n=5) | RAO (n=8) | RRO (n=7) | RRL (n=8) | Effect of maternal diet | Effect of postnatal growth | Effect of obesity |
---|---|---|---|---|---|---|---|---|
Orexigenic neurotransmitters | NPY | 3.0±0.7 | 4.4±0.9 | 2.9±0.4 | 10.5±2.5 | NS | NS | P=0.019 |
AGRP | 1.2±0.5 | 1.9±0.3 | 1.5±0.3 | 1.6±0.3 | NS | NS | NS | |
Insulin and leptin signaling | PTP1B | 1.3±0.2 | 1.9±0.2 | 1.9±0.3 | 1.2±0.2 | (P=0.051) | NS | P=0.031 |
SOCS3 | 5.9±1.1 | 8.5±1.3 | 9.3±2.0 | 7.4±1.4 | NS | NS | NS | |
IR | 0.7±0.1 | 1.1±0.2 | 1.4±0.3 | 0.9±0.2 | NS | NS | NS | |
OBRb | 6.5±0.6 | 7.0±0.8 | 6.7±1.3 | 7.5±0.9 | NS | NS | NS | |
Intracellular energy signaling | AMPKA2 | 2.5±0.2 | 4.1±0.4 | 4.3±0.5 | 3.0±0.4 | P=0.008 | NS | P=0.014 |
MTOR | 1.2±0.1 | 1.9±0.2 | 2.2±0.2 | 1.3±0.1 | P=0.014 | NS | P=0.003 | |
FTO | 8.0±0.7 | 12.1±1.1 | 12.3±0.9 | 9.4±1.1 | P=0.016 | NS | (P=0.053) | |
Cortisol regulation | GCR | 1.1±0.1 | 1.5±0.1 | 1.6±0.2 | 1.1±0.2 | NS | NS | P=0.038 |
CRH | 0.7±0.1 | 1.2±0.2 | 1.1±0.2 | 1.0±0.2 | P=0.048 | NS | NS | |
AVP | 1.5±0.2 | 2.8±0.3 | 2.4±0.6 | 2.7±0.6 | P=0.007 | NS | NS |
Statistical significance for the effect of maternal diet (i.e., CAO vs RAO), accelerated postnatal growth (i.e., RAO vs RRO), and obesity (i.e., RRO vs RRL). NS, not significant; NPY, neuropeptide Y; AGRP, agouti-related peptide; PTP1B, protein tyrosine phosphatase, non-receptor type 1; SOCS3, suppressor of cytokine signaling 3; IR, insulin receptor; obRb, leptin receptor, long form; AMPKA2, AMP-activated protein kinase α2; mTOR, mammalian target of rapamycin; FTO, fat mass and obesity-associated gene; GCR, glucocorticoid receptor; CRH, corticotropin-releasing hormone; AVP, arginine vasopressin.
Correlations between plasma insulin and leptin concentrations and hypothalamic gene expression for NPY and AgRP (2−ΔΔCt) at 16 months of age. Offspring of C and R mothers were subjected to an accelerated (CA, n=5; RA, n=8) or regular growth (RR, n=15) during lactation. After weaning, offspring were then exposed to an obesogenic environment (O) with the exception for a subset of RR, which remained lean (RRL, n=8; RRO, n=7.)
Group | NPY | AGRP | |||||
---|---|---|---|---|---|---|---|
Slope | Spearman's ρ | P value | Slope | Spearman's ρ | P value | ||
Insulin | CAO | −0.02 | −0.50 | 0.39 | −47.3a | −0.72 | 0.19 |
RAO | −0.0015 | −0.21 | 0.61 | −9.33a | −0.45 | 0.26 | |
RRO | −0.0026 | −0.52 | 0.25 | −14.7a | −0.64 | 0.12 | |
RRL | −0.01 | −0.41 | 0.42 | −15.0a | −0.12 | 0.83 | |
Leptin | CAO | −19.8a | −0.48 | 0.39 | −5.89b | −0.74 | 0.19 |
RAO | −6.9a | −0.88 | 0.004† | −1.93b | −0.71 | 0.047* | |
RRO | −1.7a | −0.04 | 0.90 | −3.71b | −0.79 | 0.036* | |
RRL | +40.2a | 0.33 | 0.42 | −2.37b | −0.31 | 0.46 |
*P<0.05; †P<0.01. Slope is expressed as a×10−5 and b×10−6. AgRP, agouti-related peptide; NPY, neuropeptide Y.
Discussion
We have established that the long-term adverse outcomes of IUGR on insulin sensitivity can be dependent on exposure to accelerated early postnatal growth together with an obesogenic post-weaning environment. Accelerated early postnatal growth and post-weaning obesity following IUGR resulted in central resistance to insulin and leptin and was accompanied by an upregulation of gene expression for markers primarily recruited in energy sensing. In an absence of adult obesity, the detrimental effects of IUGR appeared to be much less pronounced. We have, therefore, indicated the important association between raised plasma insulin and in utero programed changes of hypothalamic sensitivity observed previously following juvenile-onset obesity (Sebert et al. 2011).
Both acute and chronic reductions in maternal food intake in late gestation stimulate maternal catabolism resulting in hypoglycemia, ketoacidosis (Herrera & Amusquivar 2000, Tygesen et al. 2008), and an increased lipolysis (Symonds et al. 1989). In this study, caloric restriction over the same period not only induced fasting hypoglycemia but also was accompanied by a more pronounced rise in plasma glucose levels immediately after feeding. These substantial fluctuations in maternal plasma glucose are likely to be paralleled within the fetus, and thus possibly resetting metabolic homeostasis. We, therefore, propose that the metabolic stimuli following maternal nutrient restriction in late gestation not only promotes NEFA oxidation (Symonds et al. 1989) but also stimulates protein catabolism as indicated by raised plasma acetylcarnitine identified in the metabolomic analysis.
This is the first study to analyze the maternal metabolomic response to caloric restriction in any species. Given the substantial dichotomy in the maternal metabolic profiles with maternal nutrient restriction or free access to food, this study suggests that the source of energy available to the fetus may be a primary determinant of long-term energy homeostasis in the offspring, especially when subsequently exposed to an obesogenic environment. The brain is dependent on the availability of glucose and ketone bodies (Robinson & Williamson 1980) and this switch in energy source may be essential to hypothalamic plasticity. Although insulin resistance following maternal caloric restriction can be exacerbated further with age (Kongsted et al. 2014), we observed an effect of both postnatal growth rate and a clear influence of exposure to an obesogenic environment at 16 months of age.
Despite higher insulin and leptin concentrations, IUGR offspring raised in an obesogenic environment (RAO vs CAO) did not exhibit alterations in gene expression for orexigenic neurotransmitters such as NPY and AgRP and correlations between plasma leptin with NPY and AgRP suggest a blunted response in RAO when compared with CAO, i.e., early-onset hypothalamic resistance to leptin (Schwartz & Baskin 2013). No reduction in gene expression for insulin and leptin receptors was found, which could have suggested a potential mechanism. Whether these effects are mediated through changes in downstream signaling is yet to be confirmed. We were unable to detect any significant changes in expression of PTP1B or SOCS3, suggesting that further mechanistic studies are required.
Lean IUGR offspring (RRL vs RRO) were characterized as exhibiting reduced hypothalamic gene expression for PTP1B, but the abundance of the orexigenic neurotransmitter NPY was raised, reflecting a high central sensitivity to insulin and leptin, as expected in animals of normal body weight (Ahmad et al. 1997). Glucose homeostasis and the hormonal response to feeding in RRL were similar to CAO offspring. Taken together, these findings indicate a degree of maladaptation, as lean IUGR individuals would be expected to exhibit lower plasma concentrations of fasted metabolites and hormones and show a smaller response to those challenges than obese animals, at least in terms of NEFAs, insulin, and leptin (Sebert et al. 2009). One hypothalamic outcome of IUGR was increased expression of genes involved in energy sensing, which were also higher in the offspring reared within an obesogenic, compared with a lean environment (RRO vs RRL). In the lean IUGR group, the expression for those genes was reduced to values very similar to obese controls, even though FTO is known to be more highly expressed in obese sheep than lean sheep (Sebert et al. 2010). This further suggests that IUGR has a long-term effect, which is not fully corrected with exposure to a high-activity environment. However, these assumptions will need to be tested further with a more appropriate control group and in both male and female offspring.
IUGR also resulted in raised gene expression for hypothalamic genes involved in cortisol regulation, i.e., CRH and AVP which, when combined with the higher plasma cortisol response to feeding observed in the obese IUGR group subject to an accelerated postnatal growth rate, may be indicative of reduced negative feedback control (Lightman 2008). The same higher expression of AVP and CRH was observed in the obese IUGR animals subjected to a slower postnatal growth rate and lean IUGR animals, which both had a lower cortisol response to feeding. Therefore, we did not see a similar loss of negative feedback in these latter offspring. This difference in cortisol regulation is novel and requires further investigation. It has recently been described that female sheep with juvenile-onset obesity have elevated plasma cortisol concentrations (Bloor et al. 2013), a difference not found in this study.
All offspring raised in an obesogenic environment became equally obese irrespective of their in utero diet, and this may reflect the more physiological, long-term exposure we adopted to induce this condition. Both twin and singleton pregnancies are common in sheep, leading to differences in birth weight and post-weaning growth (Hancock et al. 2012). Only twin-bearing mothers were selected for this study, hence it is not possible to ascertain whether similar interventions designed to affect postnatal growth rates would lead to identical outcomes in singleton offspring. However, our study demonstrates that both the postnatal and post-weaning environments are important determinants of long-term outcomes following IUGR. To date, there are no large animal studies, which have looked at the developmentally exacerbated effects of adult-onset obesity together with the extent to which all symptoms of the metabolic syndrome become manifest. This is due to a number of practical considerations that include the extended time period required, well beyond the 3 year time frame of most project grant awards and the very high cost of such studies. In addition, the sex of the offspring is not predictable in naturally conceived pregnancies. A study designed to analyze the biological interaction between the sex of the offspring and the outcomes of fetal programing would clearly require a much larger number of mothers to reach the appropriate number of male and female offspring. Given the current limitations and knowledge, our present data support the evidence that some long-term impacts of fetal programing are common to both sexes. However, future studies that are able to include sufficient numbers of males and females are warranted to analyze further the effect of the sex of the offspring and its interaction with the fetal and postnatal environments.
In conclusion, in sheep, manipulation of the maternal metabolic status alone, without significant changes in maternal plasma insulin, leptin, and cortisol, is sufficient to have long-term consequences for the offspring's health. The adverse phenotype of IUGR is enhanced by accelerated postnatal growth and exposure to an obesogenic environment in juvenile life.
Supplementary data
This is linked to the online version of the paper at http://dx.doi.org/10.1530/JOE-14-0600.
Declaration of interest
The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.
Funding
This work was supported by the European Union Sixth Framework Program for Research and Technical Development of the European Community, the Early Nutrition Programming Project FOOD-CT-2005-007036, the Nutricia Research Foundation, and the Nottingham Respiratory Biomedical Research Unit. S P S was also supported by a Wellcome Trust Value-in-People award.
References
Adam CL, Findlay PA, Chanet A, Aitken RP, Milne JS & Wallace JM 2008 Expression of energy balance regulatory genes in the developing ovine fetal hypothalamus at midgestation and the influence of hyperglycemia. American Journal of Physiology. Regulatory, Integrative and Comparative Physiology 294 R1895–R1900. (doi:10.1152/ajpregu.00163.2008)
Ahmad F, Azevedo JL, Cortright R, Dohm GL & Goldstein BJ 1997 Alterations in skeletal muscle protein–tyrosine phosphatase activity and expression in insulin-resistant human obesity and diabetes. Journal of Clinical Investigation 100 449. (doi:10.1172/JCI119552)
Alexandre-Gouabau M-C, Courant F, Le Gall G, Moyon T, Darmaun D, Parnet P, Coupé B & Antignac J-P 2011 Offspring metabolomic response to maternal protein restriction in a rat model of intrauterine growth restriction (IUGR). Journal of Proteome Research 10 3292–3302. (doi:10.1021/pr2003193)
Barker DJ 1997 Fetal nutrition and cardiovascular disease in later life. British Medical Bulletin 53 96–108. (doi:10.1093/oxfordjournals.bmb.a011609)
Bispham J, Gopalakrishnan GS, Dandrea J, Wilson V, Budge H, Keisler DH, Broughton Pipkin F, Stephenson T & Symonds ME 2003 Maternal endocrine adaptation throughout pregnancy to nutritional manipulation: consequences for maternal plasma leptin and cortisol and the programming of fetal adipose tissue development. Endocrinology 144 3575–3585. (doi:10.1210/en.2003-0320)
Bloor ID, Sébert SP, Saroha V, Gardner DS, Keisler DH, Budge H, Symonds ME & Mahajan RP 2013 Sex differences in metabolic and adipose tissue responses to juvenile-onset obesity in sheep. Endocrinology 154 3622–3631. (doi:10.1210/en.2013-1207)
Cloarec O, Dumas M-E, Craig A, Barton RH, Trygg J, Hudson J, Blancher C, Gauguier D, Lindon JC & Holmes E et al. 2005 Statistical total correlation spectroscopy: an exploratory approach for latent biomarker identification from metabolic 1H NMR data sets. Analytical Chemistry 77 1282–1289. (doi:10.1021/ac048630x)
Courant F, Pinel G, Bichon E, Monteau F, Antignac J-P & Le Bizec B 2009 Development of a metabolomic approach based on liquid chromatography–high resolution mass spectrometry to screen for clenbuterol abuse in calves. Analyst 134 1637. (doi:10.1039/b901813a)
Delavaud C, Bocquier F, Chilliard Y, Keisler DH, Gertler A & Kann G 2000 Plasma leptin determination in ruminants: effect of nutritional status and body fatness on plasma leptin concentration assessed by a specific RIA in sheep. Journal of Endocrinology 165 519–526. (doi:10.1677/joe.0.1650519)
Desai M, Babu J & Ross MG 2007 Programmed metabolic syndrome: prenatal undernutrition and postweaning overnutrition. American Journal of Physiology. Regulatory, Integrative and Comparative Physiology 293 R2306–R2314. (doi:10.1152/ajpregu.00783.2006)
Gardner DS, Tingey K, Van Bon B, Dandrea J, Keisler DH, Stephenson T & Symonds ME 2005 Programming of glucose-insulin metabolism in adult sheep after maternal undernutrition. American Journal of Physiology. Regulatory, Integrative and Comparative Physiology 289 R947–R954. (doi:10.1152/ajpregu.00120.2005)
Hancock SN, Oliver MH, McLean C, Jaquiery AL & Bloomfield FH 2012 Size at birth and adult fat mass in twin sheep are determined in early gestation. Journal of Physiology 590 1273–1285. (doi:10.1113/jphysiol.2011.220699)
Herrera E & Amusquivar E 2000 Lipid metabolism in the fetus and the newborn. Diabetes/Metabolism Research and Reviews 16 202–210. (doi:10.1002/1520-7560(200005/06)16:3<202::AID-DMRR116>3.0.CO;2-#)
Kongsted AH, Tygesen MP, Husted SV, Oliver MH, Tolver A, Christensen VG, Nielsen JH & Nielsen MO 2014 Programming of glucose-insulin homoeostasis: long-term consequences of pre-natal versus early post-natal nutrition insults. Evidence from a sheep model. Acta Physiologica 210 84–98. (doi:10.1111/apha.12080)
Lightman SL 2008 The neuroendocrinology of stress: a never ending story. Journal of Neuroendocrinology 20 880–884. (doi:10.1111/j.1365-2826.2008.01711.x)
Livak KJ & Schmittgen TD 2001 Analysis of relative gene expression data using real-time quantitative PCR and the 2−ΔΔCT method. Methods 25 402–408. (doi:10.1006/meth.2001.1262)
McMillen IC & Robinson JS 2005 Developmental origins of the metabolic syndrome: prediction, plasticity, and programming. Physiological Reviews 85 571–633. (doi:10.1152/physrev.00053.2003)
Muhlhausler BS, Adam CL, Findlay PA, Duffield JA & McMillen IC 2006 Increased maternal nutrition alters development of the appetite-regulating network in the brain. FASEB Journal 20 1257–1259. (doi:10.1096/fj.05-5241fje)
Paus T 2010 A primer for brain imaging: a tool for evidence-based studies of nutrition? Nutrition Reviews 68 (Suppl 1) S29–S37. (doi:10.1111/j.1753-4887.2010.00327.x)
Robinson AM & Williamson DH 1980 Physiological roles of ketone bodies as substrates and signals in mammalian tissues. Physiological Reviews 60 143–187.
Roseboom TJ, van der Meulen JH, Osmond C, Barker DJ, Ravelli AC, Schroeder-Tanka JM, van Montfrans GA, Michels RP & Bleker OP 2000 Coronary heart disease after prenatal exposure to the Dutch famine, 1944-45. Heart 84 595–598. (doi:10.1136/heart.84.6.595)
Schwartz MW & Baskin DG 2013 Leptin and the brain: then and now. Journal of Clinical Investigation 123 2344–2345. (doi:10.1172/JCI69346)
Schwartz MW, Woods SC, Porte D, Seeley RJ & Baskin DG 2000 Central nervous system control of food intake. Nature 404 661–671. (doi:10.1038/35007534)
Sebert SP, Hyatt MA, Chan LLY, Patel N, Bell RC, Keisler D, Stephenson T, Budge H, Symonds ME & Gardner DS 2009 Maternal nutrient restriction between early and midgestation and its impact upon appetite regulation after juvenile obesity. Endocrinology 150 634–641. (doi:10.1210/en.2008-0542)
Sebert SP, Hyatt MA, Chan LLY, Yiallourides M, Fainberg HP, Patel N, Sharkey D, Stephenson T, Rhind SM & Bell RC et al. 2010 Influence of prenatal nutrition and obesity on tissue specific fat mass and obesity-associated (FTO) gene expression. Reproduction 139 265–274. (doi:10.1530/REP-09-0173)
Sebert SP, Dellschaft NS, Chan LLY, Street H, Henry M, Francois C, Sharma V, Fainberg HP, Patel N & Roda J et al. 2011 Maternal nutrient restriction during late gestation and early postnatal growth in sheep differentially reset the control of energy metabolism in the gastric mucosa. Endocrinology 152 2816–2826. (doi:10.1210/en.2011-0169)
Sharkey D, Gardner DS, Fainberg HP, Sebert S, Bos P, Wilson V, Bell R, Symonds ME & Budge H 2009 Maternal nutrient restriction during pregnancy differentially alters the unfolded protein response in adipose and renal tissue of obese juvenile offspring. FASEB Journal 23 1314–1324. (doi:10.1096/fj.08-114330)
Singhal A, Fewtrell M, Cole TJ & Lucas A 2003 Low nutrient intake and early growth for later insulin resistance in adolescents born preterm. Lancet 361 1089–1097. (doi:10.1016/S0140-6736(03)12895-4)
Smith CA, Want EJ, O'Maille G, Abagyan R & Siuzdak G 2006 XCMS: processing mass spectrometry data for metabolite profiling using nonlinear peak alignment, matching, and identification. Analytical Chemistry 78 779–787. (doi:10.1021/ac051437y)
Stettler N, Kumanyika SK, Katz SH, Zemel BS & Stallings VA 2003 Rapid weight gain during infancy and obesity in young adulthood in a cohort of African Americans. American Journal of Clinical Nutrition 77 1374–1378.
Symonds ME, Bryant MJ & Lomax MA 1989 Lipid metabolism in shorn and unshorn pregnant sheep. British Journal of Nutrition 62 35–49. (doi:10.1079/BJN19890006)
Symonds ME, Sebert SP, Hyatt MA & Budge H 2009 Nutritional programming of the metabolic syndrome. Nature Reviews. Endocrinology 5 604–610. (doi:10.1038/nrendo.2009.195)
Tygesen MP, Nielsen MO, Nørgaard P, Ranvig H, Harrison AP & Tauson A-H 2008 Late gestational nutrient restriction: effects on ewes' metabolic and homeorhetic adaptation, consequences for lamb birth weight and lactation performance. Archives of Animal Nutrition 62 44–59. (doi:10.1080/17450390701780276)
Van den Berg RA, Hoefsloot HCJ, Westerhuis JA, Smilde AK & van der Werf MJ 2006 Centering, scaling, and transformations: improving the biological information content of metabolomics data. BMC Genomics 7 142. (doi:10.1186/1471-2164-7-142)
Wagner S, Scholz K, Donegan M, Burton L, Wingate J & Völkel W 2006 Metabonomics and biomarker discovery: LC–MS metabolic profiling and constant neutral loss scanning combined with multivariate data analysis for mercapturic acid analysis. Analytical Chemistry 78 1296–1305. (doi:10.1021/ac051705s)
Wallace TM, Levy JC & Matthews DR 2004 Use and abuse of HOMA modeling. Diabetes Care 27 1487–1495. (doi:10.2337/diacare.27.6.1487)
Warnes KE, Morris MJ, Symonds ME, Phillips ID, Clarke IJ, Owens JA & McMillen IC 1998 Effects of increasing gestation, cortisol and maternal undernutrition on hypothalamic neuropeptide Y expression in the sheep fetus. Journal of Neuroendocrinology 10 51–57. (doi:10.1046/j.1365-2826.1998.00172.x)