Placental protection of the fetal brain during short-term food deprivation

Abstract

The fetal genome regulates maternal physiology and behavior via its placenta, which produces hormones that act on the maternal hypothalamus. At the same time, the fetus itself develops a hypothalamus. In this study we show that many of the genes that regulate placental development also regulate the developing hypothalamus, and in mouse the coexpression of these genes is particularly high on embryonic days 12 and 13 (days E12–13). Such synchronized expression is regulated, in part, by the maternally imprinted gene, paternally expressed gene 3 (Peg3), which also is developmentally coexpressed in the hypothalamus and placenta at days E12–13. We further show that challenging this genomic linkage of hypothalamus and placenta with 24-h food deprivation results in disruption to coexpressed genes, primarily by affecting placental gene expression. Food deprivation also produces a significant decrease in Peg3 gene expression in the placenta, with consequences similar to many of the placental gene changes induced by Peg3 mutation. Such genomic dysregulation does not occur in the hypothalamus, where Peg3 expression increases with food deprivation. Thus, changes in gene expression brought about by food deprivation are consistent with the fetal genome's maintaining hypothalamic development at a cost to its placenta. This biased change to gene dysregulation in the placenta is linked to autophagy and ribosomal turnover, which sustain, in the short term, nutrient supply for the developing hypothalamus. Thus, the fetus controls its own destiny in times of acute starvation by short-term sacrifice of the placenta to preserve brain development.

The hypothalamus undergoes its major development in the mouse during embryonic days 11–14 (days E11–14) (1), which also is the period when placental vasculature becomes established and proliferation of the hormone-producing giant cells occurs (2). The hormones produced by the placenta act on the maternal hypothalamus to silence sexual behavior and reproduction, increase maternal feeding, and prime the brain for postpartum maternal care by activating gene expression for synthesis of oxytocin and its receptors (3, 4). The mammalian placenta is linked to the evolution of genomic imprinting (5, 6); studies, primarily with mice, have shown that a number of imprinted genes are expressed in both the developing hypothalamus and placenta and influence neuroendocrine function (7). Necdin (8) and Magel2 (9) are two such imprinted genes whose dysregulation results in Prader–Willi syndrome characterized by a failure to suckle and development of a voracious appetite in early childhood. Necdin-deficient (10) and paternally expressed gene 3 (Peg3)-deficient (11) mice have a reduction in oxytocin neurons which produce hormones that are important for maternal care, milk letdown, and parturition. The maternally imprinted gene, mesoderm-specific transcript (Mest), like Peg3, regulates hypothalamic and placental development, and Mest-deficient mice are poor mothers that fail to show placentophagia (12). The fetal hypothalamus and placenta thus are synchronized in their development, providing a template for evolutionary selection pressures to operate ensuring the adult hypothalamus of the following generation is programmed to respond optimally to signals from the fetal placenta of that generation. In this study, and based on the findings of the growth phenotype produced by Peg3 deletions (3, 13, 14), we determined optimal timing for placental and hypothalamic coexpressed gene changes. Based on this timing, we deprived mothers of food for 24 h, a relatively mild stressor but one likely to be encountered in natural environments and one likely to produce strong selection pressures for maternal and fetal adaptations in pregnancy. We further investigated the consequences of 24-h starvation on hypothalamic and placental gene changes and how these changes relate to coexpressed gene changes brought about by the Peg3 mutation.

Results

Hypothalamus and Placental Gene Coexpression.

The ability of the adult maternal hypothalamus to respond optimally to placental signals requires some form of developmental hypothalamic preprogramming. To address how this programming occurs, we investigated the developing hypothalamus and placenta and found that genes that significantly (P < 0.05–0.01) change their expression over days E11, E12, and E13, and thus are likely to be engaged in development, were coexpressed synchronously in both hypothalamus and placenta and increased from 9% of all gene changes (on days E11–12) to 44% (on days E12–13) (Fig. 1 A and B). The percentage of changes in gene expression restricted exclusively to the hypothalamus during this same period remained relatively stable, at 34% of all gene changes (on days E11–12) and 40% (on days E12–13), whereas significant changes in the percentage of gene expression exclusively in the placenta decreased from 57% of all gene changes (on days E11–12) to 16% (on days E12–13) (Fig. 1 A and B). Thus, a developmental bias occurs from days E12–13 in the changing patterns of gene transcription from genes that change their expression uniquely in the hypothalamus or the placenta to genes that are common to and synchronized for changed expression in both the developing hypothalamus and placenta (Table S1).

Fig. 1.

(A) Changes in gene expression in the developing placenta and hypothalamus show increasing synchronization for expression in these tissues over days E11–13. (B) Peg3 inactivation particularly targets coexpressed genes, suppressing some and activating others; 24-h starvation virtually eliminates hypothalamus/placental coexpression, particularly targeting the placenta at a level fourfold that of the placenta.

Food Deprivation for 24 h Disrupts Placental but Not Hypothalamic Gene Expression.

We next examined changes in gene expression following 24-h food deprivation (on days E12–13) that completely disrupted coexpressed gene transcription, primarily by targeting the placenta (Fig. 1B). Food deprivation also exerted a fourfold increase in the number of genes significantly affected in the placenta compared with the hypothalamus. Moreover, food deprivation increased the expression of the imprinted gene Peg3 in the hypothalamus 1.68-fold (P > 1.56 × 10⁻⁷) but decreased its expression in the placenta 1.35-fold (P > 3.15 × 10⁻⁹). Because the mutation of this gene produces both a growth and feeding phenotype (15), we further examined the expression of genes in the placenta and hypothalamus when the Peg3 gene was deleted. This analysis showed that inactivation of Peg3 changes 56% of the coexpressed genes (Fig. 1B). Peg3 inactivation induced 202 changes in gene expression in the placenta; 99% of these same genes were changed by food deprivation, and their direction of changed expression was 100% concordant for the two conditions (Table S2). However, the number of genes that changed their expression in the hypothalamus in response to both 24-h food deprivation and to Peg3 inactivation was relatively low (19%) (Fig. 1B). Moreover, with starvation, the majority of these coexpressed genes that were common to the Peg3 deletion were up-regulated in the hypothalamus (64%) but were down-regulated in the placenta (60%). Thus, in the context of placental development, the transcriptional effects of food deprivation are similar to those of Peg3 inactivation. This result is not surprising, because starvation decreases placental Peg3 expression. Conversely, the hypothalamic increase in Peg3 expression would seem to protect this region of the brain from gene changes consequent on food deprivation.

The placental gene pathways that are changed significantly by food deprivation and Peg3 inactivation include oxidative phosphorylation, mitochondrial dysfunction, protein ubiquitination, and ubiquinone dysfunction, all of which are relevant to autophagy and cell death (Fig. 2A). The remaining placental gene changes associated with 24-h food deprivation but not associated with Peg3 inactivation included gene changes that inhibit cellular growth and proliferation, enhance lipid metabolism, proteolysis, and protein synthesis, and enhance cell cycle progression and cell death. Thus, both sets of placental gene changes (Peg3-dependent and starvation-dependent) present a common theme, with starvation enhancing nutrient supply through placental cell death and autophagy (Fig. 2 A and B).

Fig. 2.

Pathway analysis. The functionally defined pathways activated by 24 h of maternal food deprivation alone (Upper) and those activated by changes in gene expression that are common to Peg3 inactivation and 24 h of maternal food deprivation (Lower).

Placental Autophagy.

Mammals survive starvation by activating proteolysis and lipolysis in many different tissues (16). The ubiquitin–proteosome system is primarily responsible for increased protein breakdown during starvation through the rapid degradation and ubiquitylation of the protein, which then is recognized by ubiquitin receptors that direct it to proteosomes (17). In this study, 24-h maternal starvation induced significant up-regulation of 10 proteosome genes and several ubiquitinylating genes in the placenta, none of which were up-regulated in the hypothalamus. Ubiquitin-dependent proteolysis can be antagonized by deubiquitination, and whereas the gene coding for ubiquitin-dependent proteolysis (Dub2a) was up-regulated significantly in the hypothalamus, it was down-regulated in the placenta. Ubiquitination is the hallmark of protein degradation by the 26S proteosome (18), which was up-regulated in the placenta, and a molecular link between ubiquitination and autophagy is the autophagy-specific modifier Gaba A associated protein (Lc3/Gabarap) (19, 20), which also was up-regulated significantly in the placenta.

Little is known about the target proteins destined for major degradation when rapid adaptation is required. However, in yeast mature ribosomes are degraded rapidly by autophagy upon nutrient starvation (21). This ribophagy requires catalytic activity of a specific ubiquitin protease, a member of which is up-regulated in the placenta on starvation (22, 23). More notable is the up-regulation of gene expression for 152 different ribosomal proteins involving every chromosome except the Y chromosome (Table S3). All these genes are up-regulated in the starved placenta, whereas only a single riboprotein gene (Rp31) increases expression in the hypothalamus on starvation [a possible fail-safe mechanism because Rp31 maintains cellular levels of zinc (24)].

Adaptations in the placenta to meet nutrient supply also involve transporter proteins, 14 of which show significantly changed transcription in the placenta after 24-h maternal starvation. Transporter proteins are integral membrane proteins that catalyze active transfer (e.g., amino acids) across the plasma membrane of the syncytiotrophoblast (25, 26). However, the taurine transporter (Slc6a6), choline transporter (Slc44a), and the bicarbonate transporter (Slc4a7) are all up-regulated in the developing hypothalamus but are down-regulated in the placenta on maternal starvation. Transporter proteins on the mitochondrion plasma membrane (Slc25a5 and Slc25a17) concerned with mitochondrial energy regulation are up-regulated in the placenta.

Hypothalamic Protection.

An important consequence of 24-h maternal starvation is the availability of nutrient calories for the fetus because of the increasing demands, especially from the developing brain, placed on fetal energetic capacity. Energy is supplied by increasing the levels of blood sugar and through mitochondrial oxidation and phosphorylation of the cellular bioenergetic system. When calories are abundant, ATP and acetyl-CoA are produced, which are important in the brain for chromatin phosphorylation and acetylation, opening nuclear DNA for transcription and replication (27). The placenta thus appears to provide the first line of defense against fetal starvation by decreasing its own demands for glucose [insulin-like growth factor 1 receptor and insulin-like growth factor 2 receptor (Igf2r) are down-regulated significantly; insulin-like growth factor binding protein 4 (Igfbp4) and insulin-like growth factor binding protein 7 (Igfbp7) are up-regulated significantly], but the hypothalamus maintains its normal function with increases in Atp (Atp2bp, Atp6vla, Atp312, and Atp7a are significantly increased), acyl-CoA dehydrogenase, and acyl-CoA synthetase (P > 0.001).

Thus, during 24-h maternal starvation the developing hypothalamus is sustained in part by increasing Peg3 expression in the hypothalamus. Down-regulation of Peg3 expression in the hypothalamus would impair maternal feeding and fetal growth of the next generation, maternal care, nest building, and milk letdown for the next generation, and delayed puberty and impaired reproductive success of the next generation (15). The number of genes disrupted by starvation in the hypothalamus is less than one-quarter the number of genes disrupted in the placenta, with no evidence for autophagy or ribosomal turnover in the hypothalamus. Moreover, with starvation the main categories of gene changes recorded in the hypothalamus were concerned with normal nervous system development, growth proliferation, and cell-cycle progression. Changes in genes involved in the development of the nervous system included those concerned with neural survival, axon myelination, and axon guidance [semaphorin3a (Sema3a), semaphorin3b (Sema3b), nurophilin (Nrp1), atrophin (Atn), fukutin (Fktn), glypicans 3 (Gpc 3) and 6 (Gpc 6), and sox determining region Y (sox4) all of which were up-regulated. Other genes were concerned with angiogenesis glucose 6 phosphate isomerase (slit6) was up-regulated, and receptor activity modifying protein 2 (ramp2) was down-regulated] and hypothalamus and pituitary development [ubiquitin protein ligase (Ube3a) was up-regulated, and wnt oncogene analog 5 (Wnt5) was down-regulated]. Genes that were down-regulated included homeobox genes [alx homeobox 1 (Alx1), lens intrinsic membrane protein 2 (Lim2), pou class 3 homeobox 2 (Pou3f2), and pou class 3 homeobox 3 (Pou3f3)] concerned with neural tube, anterior neural plate, and neural stem cells, all of which are ventricular zone neural precursors. Taken together with the down-regulation of the mitochondrial genome and somatostatin, this result suggests that an energy-saving and growth-slowing process in hypothalamic development occurs on starvation with no evidence of gene-expression pathology.

Discussion

The hypothalamus and placenta are coadapted through the imprinted genes that are coexpressed in both tissues. Imprinted genes are known to be a key regulator of placental and hypothalamic development (7, 28), and transcriptional silencing produces deficits in maternalism and placental growth, which are functionally coadapted (13). Thus, when Peg3 transcription is inactivated in the hypothalamus of the pregnant mother carrying wild-type offspring, the functional phenotypic outcomes are remarkably similar to those that occur when the same gene is inactivated selectively in the developing placenta and fetal hypothalamus in a wild-type mother (14). This similarity in outcome illustrates the importance of synchronized expression of imprinted Peg3 in the hypothalamus and placenta for functional convergence of these phenotypes. Moreover the developmental time period for the maximal number of coexpressed placental/hypothalamic genes (days E12–13) was coincident with maximal disruption to coexpression changes brought about by the Peg3 mutation. Interestingly, starvation desynchronized the expression of Peg3 by increasing its expression in the hypothalamus and decreasing its expression in the placenta. The outcome is very different for the two tissues, with placental authophagy, particularly of ribosomal proteins, producing nutrients to sustain the energy demands of the neural tissue. No autophagic genes are activated in the hypothalamus, which continues to express genes that regulate neural development including ubiquitin protein ligase E3a (Ube3a), the imprinted gene implicated in Angleman's syndrome. A number of epigenetically regulated imprinted genes are expressed in the placenta and in the hypothalamic regions that control maternal physiology and behavior (14). Moreover, all these maternally imprinted genes have their epigenetic mark (the differentially methylated region) reprogrammed in the female germline when the imprints are removed by active demethylation between days E10.5 and E12.5 of oocyte development (29). This time period coincides with the timing for placental/hypothalamic developmental programming of the fetus. Thus, maternal imprinting extends its influence across three generations, the maternal generation when the adult mother's hypothalamus responds to placental signals, the next generation when the developing hypothalamus is genetically coadapted for placental signaling, and a third generation when the key regulatory genes responsible have their imprint reprogrammed in the developing female germline.

CpG islands, which represent the epigenetic mark for methylation of the imprint control region, are susceptible to environmental insults including low oxygen supply (30), assisted reproductive manipulation (31), and acute starvation. This study reveals changes to imprinted gene expression in the placenta [Peg3, Igf2r, and small nucleolar RNA 115 (Snord115)] as well as 41 gene changes with acute starvation stress which normally are classified under neurological disease, including schizophrenia histidine triad nucleotide binding protein 1 (Hint1) and Strathmin1 (Stmn1). Clearly further work is required with these results in mind, but because placental tissue is readily available at birth, knowledge that these genes show coexpression in the brain could provide a useful screen for neural gene identity. A recent study provides support for maternal–fetal relationships influencing brain development and function (32). This study has shown a direct role for placental metabolic pathways in modulating fetal brain development by the production of serotonin and indicates that maternal–placental–fetal interactions could underlie the pronounced impact of serotonin on long-lasting mental health outcomes.

Methods

Animals.

C57BL/6J mice were used throughout. Because we required embryos at accurately defined ages (days E11, E12, and E13), males were placed with wild-type females for one 8-h period only, and the presence of a vaginal plug was used to indicate a successful mating. Mice were housed indoors on a 12-h reversed light/dark cycle with ad libitum access to food and water. All mice were killed between 2:00 and 3:00 PM by cervical dislocation. Individual embryos were examined under a dissecting microscope, and their developing hypothalami were excised. Placentae also were removed, and their weights were recorded. To eliminate the possibility that an abnormal embryo with a pathologically extremely high or low growth rate might bias our results, embryos whose placental weight deviated from a mean value by >10% were not used in this study.

Eight mouse embryos were used for each group. Animal experiments were performed in accordance with the U.K. Animals (Scientific Procedures) Act 1986 and European Economic Community (EEC) directive 86/609/EEC.

Starvation.

Female mice were mated and left undisturbed until 2:00 PM on day E12. Then food was removed, and the mice were starved until 2:00 PM on day E13, when they were killed.

Peg3 Inactivation.

Because Peg3 is maternally imprinted and hence paternally expressed, maternal transmission of Peg3 has no genotypic effects. Therefore, male heterozygous Peg3^+/− mice were mated with wild-type females. With this method, a gravid uterus contains a mixture of both Peg3^+/− mutant and wild-type embryos. Our Peg3^+/− mutant mice contain a marker gene that induces β-galactosidase expression in cartilage (11), so embryos were genotyped by placing cartilage overnight in PBS (1 mg/mL) containing 4 mM potassium ferrocyanide, 4 mM potassium ferricyanide, 2 mM magnesium chloride, and 1 mM EGTA (all from Sigma). Subsequent blue staining in cartilage therefore indicated Peg3^−/− inactivation.

RNA Extraction and Microarray Hybridization.

Placentae/developing hypothalami were homogenized in 500 μL of RLT lysis buffer (RNeasy kit, Qiagen) and were stored at −80 °C until required. RNA then was extracted from the stored, homogenized lysate using the standard protocol for animal tissues supplied with the RNeasy Mini (hypothalamus) or Midi (placenta) RNA Extraction Kit. Extracted RNA was dissolved in water (30 μL for hypothalami and 60 μL for placentae) and stored at −80 °C. After extraction, total RNA was assessed for purity and quality by Nanodrop spectrophotometer and Agilent 2100 Bioanalyser. All samples used for array analysis were 260/280 (2.05–2.13) and RIN (9.9–10). For each sample, 200 ng of total RNA was amplified and labeled using the Whole-Transcript Sense Target protocol described in the Whole-Transcript Sense Target Labeling Assay Manual, v. 4 (Affymetrix). Briefly, total RNA is converted to cDNA and then is linearly amplified using an Eberwine procedure to create an antisense cRNA, library which then is converted to single-strand sense cDNA that is fragmented and end-labeled before hybridization. The amplified targets were hybridized to Mouse Gene ST 1.0 Arrays (Affymetrix) overnight and were scanned using a Gene-Chip Scanner 3000 7 G. Data files were extracted from the image files automatically by Gene-Chip Command Software v. 2 (Affymetrix), and the Adobe .cel extension (CEL) file format subsequently was used for analysis.

Microarray Analysis.

Analysis was performed using Genespring GX 10.0.2 software (Agilent). Data were summarized and normalized using the Robust Multiarray Analysis algorithm, and individual probe sets were filtered to incorporate probe sets to those between the 20th and 100th percentile. All gene-expression data are reported as fold change. Genes were considered to be transcribed differentially (i.e., starved vs. unstarved) if they were significantly up- or down-regulated by at least 1.5-fold relative to the earlier developmental period. Statistical analysis was performed using a one-way ANOVA followed by a Tukey highly significant differences post hoc test. To exclude high false-discovery rates that can be produced using multiple statistical comparisons on very large (20,000) datasets, data were adjusted using the Benjamini–Hochberg false-discovery rate multiple testing correction method.

Starvation.

The following comparisons were performed: unstarved (control) vs. starved at day E13 and starved vs. Peg3^+/−. Genes were divided into those whose changes in expression were confined to the brain, those whose changes in expression were confined to the placenta, and those whose changes in expression occurred in both brain and placenta.

Peg3 Inactivation.

The following comparisons were performed: (i) wild-type brain on day E12 vs. wild-type brain on day E13; (ii) wild-type placentae on day E12 vs. wild-type placentae on day E13; (iii) Peg3^+/− brain on day E12 vs. Peg3^+/− brain on day E13; (iv) Peg3^+/− placentae on dayE12 vs. Peg3^+/− placentae on day E13. These gene sets were used as a template, to derive (i) genes whose expression was suppressed by Peg3^+/− inactivation (genes that were transcribed differentially in the wild-type brain or placenta between days E12 and E13 but were not transcribed differentially over the same period in the Peg3^+/− brain or placenta) and (ii) genes induced by Peg3^+/− inactivation (genes that were not transcribed differentially in the wild-type mouse brain and placenta between days E12 and E13 but were transcribed differentially in the Peg3^+/− brain or placenta). In each case the genes were categorized as genes whose changes in expression were confined to the hypothalamus, genes whose changes in expression were confined to the placenta, and genes whose changes in expression occurred in both hypothalamus and placenta.

Ingenuity pathway analysis software (Ingenuity Systems, www.Ingenuity.com) was used to identify biological function and assign genes to a range of known metabolic or signaling pathways. Genes from an individual dataset that met the P value cutoff of 0.005 (using a Fischer's exact test) were considered for pathway analysis.