Regulation of Maternal Behavior and Offspring Growth by Paternally Expressed Peg3

See allHide authors and affiliations

Science  09 Apr 1999:
Vol. 284, Issue 5412, pp. 330-334
DOI: 10.1126/science.284.5412.330


Imprinted genes display parent-of-origin–dependent monoallelic expression that apparently regulates complex mammalian traits, including growth and behavior. The Peg3 gene is expressed in embryos and the adult brain from the paternal allele only. A mutation in the Peg3 gene resulted in growth retardation, as well as a striking impairment of maternal behavior that frequently resulted in death of the offspring. This result may be partly due to defective neuronal connectivity, as well as reduced oxytocin neurons in the hypothalamus, because mutant mothers were deficient in milk ejection. This study provides further insights on the evolution of epigenetic regulation of imprinted gene dosage in modulating mammalian growth and behavior.

Maternal and paternal genomes contribute unequally to development (1) through the monoallelic expression of imprinted genes that affect embryonic and placental development, as well as behavior in mice (2). It is possible that complex behavioral output of the central nervous system (CNS) might be a common function of a disparate group of neurally expressed imprinted genes such asMest and Peg3. In this context, we showed previously that parthenogenetic (PG: duplicated maternal genome) and androgenetic (AG: duplicated paternal genome) cells contribute unequally to the formation of the CNS in chimeric mice (3). PG cells contribute more to the cortex and striatum, whereas AG cells contribute to the hypothalamus. The imprinted gene Mest of paternal origin is expressed in the hypothalamus and functions in regulating growth and maternal behavior (4). Another paternally expressed gene, Peg3, of unknown function is expressed in a variety of embryonic meso-endodermal tissues, in the hypothalamus and the adult brain (5–7). The putative Peg3 protein contains 12 C2H2-type zinc-finger motifs and two proline or acidic amino acid–rich repeat domains (5, 6), suggesting its involvement in DNA-binding and protein-protein interactions, respectively. Recently,Peg3 (also known as Pw1) was implicated as a partner protein to TRAF2 (8) and in the tumor necrosis factor (TNF) signaling pathway.

To study the function of Peg3 in vivo, we mutated the gene by insertion of a βgeo selection cassette into its 5′ coding exon using gene targeting (Fig. 1A) (9). The heterozygous embryos that inherited thePeg3 β geo mutation from the paternal germ line (designated +/–) showed no detectable wild-typePeg3 mRNA (Fig. 1B) (10). However, they showed appropriate β-galactosidase (β-Gal) expression (5,6) (Fig. 1C). In adults, β-Gal expression was localized in the brain (Fig. 1D), pituitary, and adrenal medulla. The +/– mice were smaller (11) but otherwise fertile, healthy, and normal in their general behavior. As expected, maternal transmission of the mutation (designated –/+) had no phenotypic effects and showed no β-Gal activity throughout development (Fig. 1C). Thus, the mutantPeg3 β geo locus was imprinted exactly as the endogenous wild-type locus.

Figure 1

Generation and characterization ofPeg3 β geo mutant mice. (A) The wild-type Peg3 locus with its open reading frame spans from exon 3 to 9 (black boxes). The mutantPeg3 β geo locus was generated by insertion of a βgeo selection cassette containing an internal ribosome entry site (green box),lacZ-neo fusion gene (blue arrow), and SV40 polyadenylation (red oval) into exon 5. H, Hind III; B, Bgl II; Xb, Xba I; Xh, Xho I. Probes used for Southern and Northern blot analyses are shown as open and grey boxes, respectively. (B) Northern blot analysis of mRNA (2 μg/lane) from the wild-type (+/+), paternal heterozygote (+/–), maternal heterozygote (–/+), and homozygote (–/–) embryos at 12.5 dpc. GAPDH, glyceraldehyde-3-phosphate dehydrogenase. (C) β-Gal staining of sagittal planes of the +/– and –/+ embryos at 12.5 dpc (23).(D) β-Gal staining of coronal brain cryosections from the +/– adult mice. High expression was observed in the neuronal regions known to be crucial for maternal behavior (indicated by arrows) (15). CX, cortex; MPOA, medial preoptic area; PVN, paraventricular nucleus; MA, medial amygdala; BNST, bed nucleus of the stria; HP, hippocampus. Scale bars, 1.0 mm.

A distinct behavioral phenotype became evident from interbreeding of +/– heterozygotes. Very few first litters (8%) of mutant mothers (+/–) grew to weaning age, compared with those nursed by wild-type females (83%). As the progeny from mutant females and wild-type males (+/– × +/+) also failed to survive, this suggested that the genotype of the father was not relevant for their survival. More importantly, as the progeny inherited the active paternal Peg3 allele and the silent maternalPeg3 β geo mutant allele, they should develop as normal adults. Their failure to thrive showed the existence of a maternal nurturing defect. Nevertheless, mutant females improved their nurturing ability by the third parturition because most of them (7 of 10, 70%) cared for their young, although they remained maternally impaired (see below).

Because the newborn rodents are deaf, blind, and immobile, the mother normally builds a nest, gathers her pups together, and keeps them warm by crouching over them. We found that the mutant primiparous mother failed to exhibit any of these maternal responses. To validate these observations, we first tested the response of the postpartum mothers toward three newborn pups (Fig. 2). Mutant mothers took 11 times longer to retrieve and 8 times longer to build a nest and, unlike wild-type mothers, never crouched over their pups in the 15 min of testing. The inability to find the pups was not a factor for the impaired maternal response, as the mutant mothers sniffed the pups as quickly as wild-type mothers (14.0 ± 5.4 s compared with 8.8 ± 4.2 s; P > 0.1).

Figure 2

Impaired maternal behavior in thePeg3 β geomutant females (24). The wild-type (grey bars) and +/– mutant (black bars) primiparous postpartum (PP), multiparous diestrus (ME), and diestrus virgin (V) females were examined for maternal response toward pups (n = 6). (A) Latency of pup retrieval, (B) time spent to retrieve all pups, (C) latency of nest building, and (D) latency of crouching over pups were recorded. The data represent the mean ± SEM. *,P < 0.05; **, P < 0.01 by Mann-Whitney U test.

The rapid onset of maternal behavior seen in postpartum females is induced by both the hormonal priming during pregnancy and the stimulus from contact with pups at parturition (12). Sensory exposure to pups can also induce maternal behavior in virgin females, although this response is less immediate (13). To determine whether the Peg3 β geo mutation affects maternal behavior independently of pregnancy and parturition, we examined virgin and multiparous diestrus females for maternal behavior (Fig. 2). All of these mutants, with or without maternal experience, spent a significantly longer time than control animals in retrieving all the pups and in exhibiting nest-building activity. Thus, maternal behavior induced by pup exposure was also affected by the mutation.

The neural circuit responsible for maternal behavior is complex because it involves multisensory stimulation in the female after exposure to the pups. High levels of Peg3 expression were present in the hypothalamic nuclei, including the medial preoptic area (MPOA), as well as in the medial amygdala, bed nucleus of the stria terminalis, hippocampus (see Fig. 1D), and olfactory bulb (14). Lesion analysis previously showed that several of these regions are crucial for maternal behavior (15). Thus, the neural expression pattern of Peg3 is consistent with its role in regulating maternal behavior. However, no obvious structural anomalies were observed in the brain sections of the mutants, and β-Gal expression matched that of Peg3 RNA in wild-type females. We also examined expression of the fosB gene in the hypothalamic MPOA area because it is activated rapidly by exposure of virgin females to pups (16). Furthermore, a null mutation of this fosB gene causes impaired maternal behavior in both postpartum and virgin females (16). However, our Peg3 β geo mutant virgin females did show appropriate fosB induction in the MPOA after their exposure to pups, suggesting that the lack of maternal behavior is independent of fosB activation.

Despite the smaller litter size ofPeg3 β geo females, the surviving (nonmutant) progeny gained less weight compared with progeny of wild-type mothers during the first 3 weeks (P < 0.002) (Fig. 3A, left). However, their weight did catch up after weaning (P = 0.7449) (Fig. 3A, right). The preweaning deficiency in weight gain could result from either impaired maternal response or a defect in lactation in the mutant females (or from both). To test this, we measured both maternal behavior and the weight gain by the pups after the separation of the mothers and pups for 2 hours (Fig. 3B). The mutant mothers were slower in adopting the crouching posture compared with wild-type mothers (15.3 min compared with 8.7 min; n = 6, P < 0.004), but the pups of both groups attached to the nipples after 1 hour. In the control group, the pup's weight increased by 1.8 ± 0.5 and 3.2 ± 0.25 mg after 6 and 24 hours, respectively. By contrast, the pups suckled by the mutant mothers gained no weight after 6 hours and only 0.98 ± 0.2 mg after 24 hours (P< 0.01). The reduced weight gain in the latter suggested a defect in lactation in the mutant mothers.

Figure 3

Reduced weight gain in pups nursed by the mutant mothers. (A) Body weight (mean ± SEM) of progeny (n = 12) from +/+ female × +/+ male crosses (open circles) and of progeny (n = 12) from +/– female × +/+ male crosses (closed circles) before (left) and after (right) weaning. (B) (left) The latency of crouching behavior of the wild-type (grey bar) and +/– mutant (black bar) mothers (n = 6). (right) The weight gain observed in pups with the wild-type mothers (n = 46, open circles) and in pups with the mutant mothers (n = 26, closed circles) after a 2-hour separation.

To determine the underlying cause of lactational defect, we first examined the mammary glands of the mutant mothers but found them to be histologically normal both at prepartum and postpartum. Next, we examined oxytocin neurons because milk ejection is controlled by oxytocin released from the hypothalamic paraventricular (PVN) and supraoptic nuclei in response to the suckling stimulus; a mutation in the oxytocin gene abolishes milk ejection (17). We found that the postpartum mutant mothers had reduced oxytocin-positive neurons in the hypothalamus compared with the wild-type females (Fig. 4). The total number of oxytocin-positive neurons in the mutant females was 2984 ± 209 compared with 4496 ± 252 in the controls (n = 5, P < 0.02). The difference was significant even after allowing for the lower body weight and smaller PVN size in mutant females (Fig. 4). The reduced weight gain of pups could therefore be explained by an insufficient oxytocin surge.

Figure 4

Reduced oxytocin neurons in postpartum mutant females (25). Immunohistochemical staining of oxytocin neurons in the hypothalamus of postpartum wild-type and mutant females: (A) in the MPOA and supraoptic nucleus (SON) and (B) in the PVN. OX, optic chasma. Scale bars, 0.3 mm. (C and D) The number of oxytocin (OT) neurons in the hypothalamus (Hypothal.) relative to body weight and (C) in the PVN relative to the size of the nucleus (D), in the wild-type (grey bars) and mutant (black bars) females. The data are expressed as ±SEM. **, P < 0.01.

In rodents, central oxytocin synthesis increases at parturition, and central infusions of this hormone stimulate a rapid maternal behavior (18), but this behavior is inhibited by the hormone antiserum or antagonists (19). Thus, the behavioral and neuroendocrine responses in this study have oxytocin as a common component for both the peripheral and central events that are impaired in the Peg3 β geomutants. The reduced number of oxytocin-producing neurons may have impaired the neuronal coupling and synchrony, which is required for a bolus of oxytocin release at postpartum to achieve milk let-down (20). The involvement of Peg3 in the TNF signaling pathway affecting cell survival and proliferation (8) could account for these decreases in oxytocin neurons in the PVN and other hypothalamic neurons. However, this major endocrine dysfunction cannot explain the behavioral phenotype fully, because Peg3 mutant mothers conceived normally, gave birth, and had normal development of mammary glands during lactation.

There are now two imprinted genes, Peg3 andMest (4), of different structural classes implicated in the behavioral function of the CNS. This suggests that maternal behavior may be a specifically selected function through imprinting of neurally expressed genes. The most widely accepted hypothesis for genomic imprinting is that of “parental conflict,” in which the parental genomes compete to regulate intrauterine embryonic growth through different sets of imprinted genes (21). Maternal behavior is a powerful motivation that is essential for the survival of offspring, which is governed by the hypothalamus. However, for females to maximize their lifetime reproductive success, it is necessary for progression to the next pregnancy to happen in the shortest possible time. While nursing the young, female rodents can mate at postpartum estrus and allow embryonic development to continue to the blastocyst stage when they enter into diapause (22). At the end of lactation, these embryos implant and continue development. Lactation and pregnancy (involving embryonic growth) demand considerable maternal energy resources; these two demands cannot be met simultaneously. Because the paternal interest is best served by prolonged care and feeding of his progeny through maternal lactation, there are asymmetrical parental interests involved. This could explain why the paternal genome may have acquired the ability to regulate maternal behavior through imprinted genes such asMest and Peg3.

  • * Present address: Columbia University, College of Physicians and Surgeons, Department of Genetics and Development, New York, NY 10032, USA.

  • To whom correspondence should be addressed. E-mail: as10021{at}


Stay Connected to Science

Navigate This Article