Evidence suggests that the prenatal nutritional environment influences the risk of developing obesity, a major health problem worldwide. It is hypothesized that fetal nutrition influences the developing neuroendocrine hypothalamus, the integrative control center for postnatal energy balance regulation. The present aim was to determine whether relevant hypothalamic genes are expressed in midgestation and whether they are nutritionally (glucose) sensitive at this time. Hypothalami from a cohort of 81-day singleton sheep fetuses, with varying glycemia by virtue of maternal dietary and/or growth hormone treatment, were subject to in situ hybridization analysis for primary orexigenic, anorexigenic, and related receptor genes (term = 147 days, n = 24). Neuropeptide Y, agouti-related peptide, proopiomelanocortin (POMC), cocaine- and amphetamine-regulated transcript (CART), and insulin receptor mRNAs were all localized in the hypothalamic arcuate nucleus (ARC) of all fetuses, whereas leptin receptor mRNA was expressed more abundantly in the ventromedial hypothalamic nucleus. ARC expression levels of POMC and CART genes, but none of the other genes, were positively correlated with fetal plasma glucose concentrations. Therefore, key central components of adult energy balance regulation were already present as early as midgestation (equivalent to 22 wk in humans), and two anorexigenic components were upregulated by elevated glycemia. Such changes provide a potential mechanism for the prenatal origins of postnatal energy balance dysregulation and obesity.
- developmental programming
obesity is a major health problem worldwide. The increasingly early onset of obesity in childhood is a particularly undesirable development, and a growing body of epidemiological, clinical and experimental evidence indicates that predisposition to obesity, and associated metabolic disorders, is influenced by the prenatal nutritional environment (for reviews, see Refs. 14, 17, 31, 32, 35). Although the underlying mechanisms remain unresolved, a plausible suggestion is that inappropriate prenatal nutrition may involve programming or modification of the developing neural pathways that regulate energy balance in the brain (17, 18, 21, 25). These pathways lie in the neuroendocrine hypothalamus, the critical integrative control center for energy balance and body weight regulation in the adult (3, 34). Primary targets in the hypothalamic arcuate nucleus (ARC) that respond to nutritional feedback from the periphery are the orexigenic neurons coexpressing neuropeptide Y (NPY) and agouti-related peptide (AGRP), balanced by those coexpressing anorexigenic proopiomelanocortin (POMC) and cocaine- and amphetamine-regulated transcript (CART) (34). The POMC-derived alpha melanocyte-stimulating hormone acts via hypothalamic melanocortin receptors (MC-Rs), while AGRP is an endogenous antagonist at the MC-Rs (34). Energy homeostasis in the free-living individual results from an appropriate balance between the activities of these opposing orexigenic and anorexigenic pathways, achieved largely through leptin and insulin feedback via their receptors (OB-Rb and Ins-R, respectively) also expressed in these neuronal pathways (34). Obesity results from an energy imbalance in favor of the orexigenic pathways and insufficient activation of the anorexigenic pathways. It is unknown whether the prenatal nutritional environment can, in some way, program the inappropriate nutritional sensitivity of these pathways in postnatal life.
Although prenatal nutritional influences on postnatal neuroendocrine gene expression have been reported in the rat (13), there remains a paucity of information on the ontogeny and nutritional sensitivity of hypothalamic energy balance regulatory neuronal pathways in the fetus. Nonetheless, studies in our laboratory first revealed that the normally growing sheep fetus possesses an adultlike hypothalamic appetite-regulatory neural network in late gestation (29). Thus gene expression for NPY, AGRP, POMC, CART, and OB-Rb were present in the ARC at gestational ages 110 and 140 days (term = 147 days). Others have demonstrated NPY-positive projections from the ARC to the paraventricular hypothalamic nucleus (PVN) of 110-day ovine fetuses (38), and late-gestation human and nonhuman primate fetuses (11, 16), but not in rat fetuses in which these projections develop postnatally (10, 11). The fetal sheep is therefore a good experimental model for hypothalamic neuronal development in the human fetus, with similar neuronal maturity at birth and in late gestation in both species.
Unlike the free-living animal, the fetus relies on the transplacental supply of glucose for its nutrition (6). We demonstrated that directly increasing the nutritional status of normally growing late-gestation sheep fetuses by intrafetal glucose infusion (130–140 days) resulted in a significant upregulation of POMC gene expression in the ARC (27). This was an important indication that fetal glucose supply could influence the functional development of the near-term neuroendocrine hypothalamus. We went on to demonstrate that such changes could persist into postnatal life, since increasing fetal nutritional status by maternal overnutrition during the last trimester resulted in upregulation of POMC mRNA in the ARC of 30-day-old offspring (26). Furthermore, these offspring were significantly fatter than controls, lending support to the hypothesis that alterations in fetal hypothalamic gene expression may contribute to fetal programming of postnatal energy balance regulation.
In view of our findings in late gestation, and since structural organization of the brain occurs earlier in gestation, the aim of the present study was to determine whether hypothalamic energy balance regulatory genes are already expressed in midgestation and whether they are nutritionally (glucose) sensitive at this time. We used a cohort of 81-day singleton sheep fetuses with varying glycemia generated in a published study investigating the effects of maternal diet and growth hormone (GH) treatment on nutrient partitioning during adolescent pregnancy (37).
MATERIALS AND METHODS
Fetuses and maternal treatments.
All procedures were licensed under the U.K. Animals (Scientific Procedures) Act 1986 and approved by the Rowett Research Institute Ethical Review Committee. Brains were collected from sheep fetuses generated in a previous study (37). Briefly, recipient growing adolescent ewes had been implanted with singleton embryos, derived from superovulated donors and a single sire, and given either a high (H) or a moderate (M) level of food intake (complete diet). The moderate (control) intake was designed to promote optimum fetal growth in this genotype. From day 35 to 80 of gestation, half of the ewes in each intake group was injected twice daily with recombinant bovine growth hormone (GH), and on day 81, all ewes and fetuses were euthanized by lethal dose of pentobarbital sodium. Maternal and fetal blood samples were obtained immediately before death. The fetus was weighed and dissected to measure major organ weights. The fetal brain was removed intact, snap frozen in isopentane over dry ice, and stored at −80°C. Thus, fetal brains were obtained from the following maternal treatment groups: M (n = 6), H (n = 7), M+GH (n = 5), and H+GH (n = 6).
Maternal and fetal plasma samples were analyzed for glucose and insulin as previously reported (37). Insulin was measured by radioimmunoassay, with detection limit 4 μU/ml and intra-assay coefficient of variation 3.5%. Glucose was measured using an automated Yellow Springs Instrument (Yellow Springs, OH) dual biochemistry analyzer (model 2700).
Coronal cryostat sections (20 μm) of fetal brain were thaw-mounted onto slides double-coated with gelatin and poly-l-lysine, and stored at −80°C. Gene expression for NPY, AGRP, POMC, CART, OB-Rb, and Ins-R was measured by in situ hybridization, using techniques described in detail elsewhere (2, 22). The NPY riboprobe was generated from a rat cDNA (2), the CART probe was generated from a cloned sheep cDNA (5), and AGRP and POMC probes were generated from cloned Siberian hamster cDNAs (24). These heterologous probes have previously been validated on adult sheep brain tissue (1, 2). A riboprobe complementary to fragments of the intracellular domain of OB-Rb was generated from a cloned sheep cDNA (23), and the Ins-R riboprobe was generated from a partial ovine cDNA (4). For each antisense riboprobe, adult sheep hypothalamic sections were included with the fetal brain sections to act as positive controls, and corresponding sense probes were used to verify the specificity of hybridization. Briefly, sections were fixed, acetylated, and hybridized overnight at 58°C using 35S-labeled cRNA probes (1–1·5 × 107 cpm/ml). They were then treated with RNase A, desalted, given a final high-stringency wash (30 min) in 0·5 × SSC at 60°C (Ins-R at 75°C), dried, and apposed for 7–10 days to Hyperfilm β-max (Amersham Pharmacia Biotech UK, Little Chalfont, Bucks, UK). Intensity and total area of hybridization were quantified for the hypothalamic ARC on each autoradiographic image, using the Image-Pro Plus system (Media Cybernetics, Silver Spring, MD). The integrated intensity of the hybridization signal (i.e., the optical density integrated over the total hybridization area) was computed using standard curves generated from 14C autoradiographic microscales (Amersham Pharmacia Biotech, Little Chalfont, Bucks, UK). For each probe, up to six sections spanning the medial hypothalamus were examined from each fetus. For cellular resolution, a selection of sections hybridized to NPY, AGRP, POMC, and CART was also coated with autoradiographic emulsion (LM-1; Amersham), exposed for 12 wk, and stained with Toluidine blue. All reagents were obtained from Sigma (Sigma UK, Poole, Dorset, UK) unless otherwise stated.
ANOVA was used to examine effects of maternal diet, GH treatment, and their interaction (ANOVA; Minitab 14; Minitab, State College, PA). Pearson product-moment correlation analyses and linear regression analyses were used to explore the relationships between variables (Minitab 14). Results are presented as means ± SE, and statistical significance was set at P < 0.05. Because of the fragile nature of the fetal brain sections at this stage of gestation, some values for gene expression were missing or incomplete; the “n” used in each correlation analysis is therefore supplied where appropriate.
Extensive metabolic, endocrine, and morphometric fetal data have already been published for the present treatment groups (37). Mean data and statistical analyses are presented here specifically for the individuals used in the present study (75% of the original fetuses). Fetal body and brain weights did not differ between the groups, but liver weights were increased in the H+GH group (Table 1). There was no clearly visible adipose tissue in the fetuses. Fetal plasma glucose and insulin were increased by diet (H > M; glucose, P < 0.001; insulin, P < 0.01) and by maternal GH treatment (glucose, P < 0.001; insulin, P < 0.03). Maternal glucose and insulin were increased by diet (H > M; both P < 0.001) and by GH treatment (both P < 0.001) (Table 1). Fetal plasma glucose concentration was closely correlated with maternal glucose (r = 0.96), maternal insulin (r = 0.90), and fetal insulin (r = 0.86) (Table 2).
NPY, AGRP, POMC, and CART mRNAs were all localized in the hypothalamic ARC of fetuses in all groups (Fig. 1). Ins-R was localized to the ARC of all fetuses, but OB-Rb showed relatively low hybridization in the ARC of only a few individuals (and was therefore not quantified), with stronger hybridization seen in the ventromedial hypothalamic nucleus of most individuals (Fig. 2). Corresponding sense probes showed no hybridization, and all of the antisense riboprobes hybridized to the ARC of the positive control adult sheep hypothalamic sections.
Mean gene expression levels in the ARC did not differ between the groups for NPY, AGRP, CART, and Ins-R, but the elevated POMC gene expression in H+GH just reached significance (after cube transformation to improve normality, P < 0.045) (Fig. 3).
Correlation/regression analysis across all treatment groups revealed a significant positive relationship between POMC mRNA expression in the ARC and fetal glycemia (n = 19, r = 0.58, P < 0.01, Fig. 4A), with no correlation seen between NPY, AGRP, CART mRNAs, and fetal glycemia (Table 3). When data for the normoglycemic control group (M) were excluded, the correlation between POMC mRNA and glycemia held true (n = 13, r = 0.64, P < 0.02), and a significant relationship between CART mRNA and increased glycemia was revealed (n = 15, r = 0.52, P < 0.05, Table 3, Fig. 4B). Ins-R gene expression did not correlate with fetal plasma concentrations of either glucose (n = 17, r = 0.26, P = 0.31) or insulin (n = 17, r = 0.09, P = 0.73).
This is the first report of gene expression for primary appetite-regulatory genes in the fetal hypothalamic ARC as early as midgestation (second trimester, equivalent to 22 wk in humans). Moreover, expression of the anorexigenic genes POMC and CART in the ARC was increased by elevated glycemia at this stage. Such changes, if they persist, could affect energy balance regulation in later life, providing a potential mechanism for prenatal origins of obesity.
Previously, the earliest that expression of these energy balance regulatory genes has been reported in the ovine ARC was at gestational day 110 (29), equivalent to 30 wk or third trimester in human pregnancies, with nutritional sensitivity reported near term at 130–140 days (27). Thus, they are expressed early in gestation when the fetus receives all of its nutrition from the maternal circulation, in preparation for life after birth when the neonate must take control of its own energy balance. The structural organization of hypothalamus evidently occurs early in fetal life, and it is not implausible that the local nutritional environment would influence the functional development of the energy balance regulatory pathways.
The mothers on the high dietary intake and all those given GH were relatively hyperglycemic and hyperinsulinemic in the present study, with the increases in circulating glucose and insulin particularly marked in the H+GH group (37). The increased circulating glucose crossed the placenta proportionately into the fetal circulation (with r = 0.96), so that the fetuses in all but the control group were also hyperglycemic. GH itself does not cross the ovine placenta (12), and so the changes observed within these fetuses were attributed primarily to the altered glucose supply. Glucose is the primary fetal fuel (6); the high fetal glucose concentrations provided excess nutrition and promoted the hyperinsulinemic state, which was most striking in the H+GH fetuses. However, in spite of the elevated nutritional status, the 81-day fetal body weights were similar in all four groups. This can be explained by the relatively low nutrient demand at this stage because the fetus has attained only about 8% of its final birth weight (37). Nonetheless, while it was apparent that fetal liver weights were increased in the hyperglycemic fetuses, indicating sensitivity of this major organ to nutrient availability, fetal brain weight was not affected. It was, therefore, considered most likely that any changes in gene expression observed within the brain were not due to altered brain growth per se but were attributable to the altered metabolic endocrine milieu in which it was developing.
A feature of our sheep model is the low genotypic variation between the fetuses, since the pregnancies were derived by assisted conception involving a single-sire and a limited number of embryo donors. Consequently, phenotypic differences between the fetuses can be attributed to differences in their nutritional environment during gestation with greater confidence. Increased nutritional status (glucose concentrations) in the 81-day fetuses resulted in increased gene expression in the ARC for POMC. This was consistent with our findings in the near-term fetus (140 days gestation) when intrafetal glucose infusion increased POMC mRNA levels in the ARC (27) and provided evidence for POMC nutritional sensitivity from early prenatal development. It is tempting to speculate that this might program the developing melanocortin pathway by desensitizing it to elevated levels of nutritional feedback, resulting in the defense of a higher body weight and development of obesity in later life. Indeed, we have shown that nutritionally upregulated POMC mRNA persists from fetal to early postnatal life, when it is associated with increased adiposity (26). As a precursor also for ACTH, POMC gene expression would also impact on the functioning of the developing hypothalamo-pituitary-adrenal axis in the fetus, although, in general, it is fetal hypoglycemia and not hyperglycemia that switches on this axis (8). Because fetal cortisol is central to initiation of parturition (19, 20), it is tempting to speculate that upregulated POMC may contribute to an increased incidence of spontaneous preterm birth seen in women suffering from gestational diabetes (15).
The observation that CART gene expression was also upregulated in the more extremely hyperglycemic fetuses was perhaps not surprising given the extensive colocalization of CART with POMC in ARC neurons (1, 9). Nonetheless, the net result was further amplification of anorexigenic pathways, adding to the aforementioned potential desensitizing of the hypothalamus to raised levels of nutritional feedback and putative adverse programming for body weight later in life. Conversely, there appeared to be no reciprocal downregulation of orexigenic NPY and AGRP pathways with increased fetal glycemia, consistent with our findings in late gestation (27). The relative insensitivity of orexigenic pathways in the fetus to overnutrition may reflect the need for a functionally strong orexigenic drive immediately postpartum to ensure optimum survival (33). On the other hand, fetal hypothalamic NPY expression is upregulated by undernutrition in late gestation (38), adding support to the argument that development of orexigenic drive predominates before birth.
The neuropeptide genes investigated here are classically leptin-sensitive and they coexpress with OB-Rb in the ARC of adult sheep (1, 39). In the 81-day fetuses, however, there was negligible OB-Rb gene expression in the ARC and altogether low hypothalamic expression levels. This was perhaps to be expected since circulating fetal plasma leptin concentrations were below the reliable detection limit of our assay, and the percentage of fat in the fetal carcasses was <1% (by chemical analysis; J. M. Wallace, unpublished observations). Leptin mRNA has been reported in perirenal adipose tissue from 90 days gestation in fetal sheep (40), and plasma leptin is directly related to adiposity in late gestation (30). Nonetheless, even when significant measurable circulating leptin concentrations appear in late gestation (140 days), OB-Rb expression levels in the ARC remain relatively low (29) and intrafetal leptin infusion has no effect on POMC, CART, NPY, or AGRP gene expression (28). It appears that OB-Rb expression in the ARC emerges very late in gestation and increases postnatally, perhaps underlying postnatal development of leptin sensitivity in energy balance pathways (28). This is consistent with a recent report of leptin insensitivity in the early postpartum rat brain, which corresponds to the late-gestation human and sheep in terms of neuronal maturity, when leptin treatment had no effect on NPY, AGRP, POMC, or CART gene expression in the ARC (7). Altogether, it seems likely that the differences in gene expression observed in the present study were attributable to the altered fetal nutritional status (glycemia) and not to leptin feedback. Similarly, although the fetuses also had elevated insulinemia, it was considered unlikely that the changes in neuropeptide gene expression were insulin-driven since Ins-R expression was unaffected by increased glucose and insulin concentrations.
Perspectives and Significance
This study has demonstrated in a clinically relevant sheep model that hypothalamic energy balance regulatory genes are already expressed in the developing brain by midgestation and that two anorexigenic genes are sensitive to elevated fetal glycemia at this stage. Similar findings have been reported during late gestation, and altogether, these data have implications for the fetuses of the increasing number of mothers that are obese, overnourished, or suffering from gestational diabetes. Such early changes in anorexigenic pathway activation could influence postnatal energy balance by lowering sensitivity to increased nutritional feedback. It is clearly important to establish whether the in utero alterations to central energy balance regulatory pathways are sustained postnatally and whether the offspring have the predicted increased susceptibility to body weight dysregulation and obesity. This would verify a putative mechanism for the prenatal programming of obesity, providing a potential target for early interventions aimed at preventing development of this disease.
This research was supported by the Scottish Government Rural and Environment Research and Analysis Directorate.
Audrey Chanet was an M.Sc. student at the University of Auvergne, France.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2008 the American Physiological Society