Maternal undernutrition during the preimplantation period of rat development causes blastocyst abnormalities and programming of postnatal hypertension

Correlations between low birthweight in humans and susceptibility to a number of adult chronic diseases, including coronary heart disease, stroke, high systolic blood pressure and non-insulin-dependent diabetes mellitus, have been identified (Barker, 1994; O’Brien et al., 1999). The ‘foetal origins of adult disease’ (FOAD) hypothesis proposes that these cardiovascular and related disorders derive from foetal adaptations in utero to maternal undernutrition that permanently alter growth characteristics, and postnatal metabolism and physiology (Barker, 1994; O’Brien et al., 1999). In an animal model of FOAD, female rats fed a low-protein diet (LPD; 9% or 6% casein versus 18% casein control) during pregnancy produced offspring with altered birthweight and raised systolic blood pressure throughout life (Langley and Jackson, 1994; Langley-Evans et al., 1999).

The mechanisms underlying programming are likely to be multifactorial and complex (O’Brien et al., 1999; Langley-Evans et al., 1999). However, further studies have shown that programming of hypertension in the rat LPD model may occur as a result of foetal exposure during mid-late gestation to elevated maternal glucocorticoids following LPD-induced reduction in the activity of the placental regulating enzyme, 11β-hydroxysteroid dehydrogenase type 2 (Edwards et al., 1993; Langley-Evans and Nwagwu, 1998; Langley-Evans et al., 1999; Seckl et al., 1999). One ‘downstream’ consequence of raised glucocorticoid activity has been shown to be increased expression of renin-angiotensin system components, such as angiotensin-converting enzyme, which may lead directly to increased blood pressure (Langley-Evans et al., 1999).

Although these changes in hormonal regulation provide a mechanistic framework for understanding potential programming pathways in later gestation and beyond, we are largely ignorant of when programming may initiate during development and what earlier ‘upstream’ mechanisms may be involved. An early origin for programming is indicated, as maternal LPD fed during just the first third of pregnancy also led to hypertension in male offspring (Langley-Evans et al., 1996). In addition, maternal LPD has been shown to alter glucose consumption and lactate production rates in rat embryos (9.5-10.5 days post-implantation) (Leese and Isherwood-Peel, 1999). There is also evidence that poor nutrition around the time of conception can influence birthweight in the human (Wynn and Wynn, 1988) and ovine (McEvoy et al., 1997), as well as ovine insulin-like growth factor (IGF1) expression in late gestation (Gallaher et al., 1998). Here, we provide evidence that programming may initiate within the preimplantation embryo and propose a cellular mechanism to explain its inception.

Virgin female Wistar rats (bred at University of Southampton Biomedical Facility; 12 hour light cycle, 24°C; 220-250 g, 11-13 weeks old), fed ad libitum from weaning on standard chow diet (CRM(X), 18% protein, Special Diet Services, Cambridge, UK) and water, were naturally mated overnight and vaginal plugs identified the following morning (11 a.m. reference point for day 0 of development). Plug-positive females were then transferred to single cages and fed ad libitum LPD (9% or, for specific experiments, 6% casein; Langley and Jackson, 1994) or control diet (18% casein) until day 5. The composition of these diets is shown in Table 1. The control and LPD diets are balanced for energy intake assuming equivalent consumption rates. For postnatal studies, rats fed LPD were switched to control 18% protein diet at day 4.25 for the remainder of gestation, while control animals remained on this diet. At birth, dams were returned to standard chow diet. Gestation length, birthweight and post-weaning growth rates were determined. Systolic blood pressure was determined at 4 and 11 weeks by tail-cuff plethysmography using an IITC model 229 blood pressure monitor (Linton Instruments, Diss, UK), as described (Langley-Evans et al., 1996), after 1 hour acclimatisation. Recordings were made ‘blind’ (mean of four per rat) by coding animals and if heart rate exceeded 480 beats/minutes (indicative of stress), results were discarded. At 12 weeks, rats were sacrificed and body/organ weights were measured.

Embryos were flushed from the oviducts or uteri of rats using H6 medium plus 4 mg/ml BSA (Sheth et al., 1997). Freshly collected embryos were coded and analysed ‘blind’ for cell numbers within trophectoderm and ICM lineages by differential fluorochrome nuclear labelling (Handyside and Hunter, 1984), following removal of the zona pellucida in acid Tyrodes medium (Sheth et al., 1997), and using a rabbit anti-rat spleen antiserum and guinea pig complement (V. H. Bio, Newcastle-upon-Tyne, UK), both of which were diluted 1:10 in H6 for the partial immunosurgery step. Embryos were processed and examined using a Leitz Diaplan fluorescence microscope as described (Stoddart et al., 1996). The number of apoptotic cells in zona-intact blastocysts was determined by the terminal deoxynucleotidyl transferase (TdT)-mediated dUTP nick-end labelling (TUNEL; Boehringer Mannheim, UK) method (Brison and Schultz, 1997). After fixation, permeabilisation and TUNEL labelling steps, embryos were washed in RNase A (50 μg/ml) for 20 minutes at 37°C in the dark and incubated in propidium iodide (50 μg/ml in RNase A solution; 15 minutes) to counterstain nuclei for mitotic index scoring. Fluorescent labelled cells were examined using a Nikon inverted microscope (×63 objective) linked to a Bio-Rad MRC-600 series confocal imaging system equipped with a krypton-argon laser. Whole embryos were serially sectioned optically, and percentage apoptotic and mitotic cells per embryo determined from stored images. The number of implantation sites in uteri was determined by pontamine sky blue staining (Kramer et al., 1993). Dams at day 4 and 5 of development were anaesthetised by intraperitoneal injection of 0.35 ml Sagatal (Rhone Merieux Ltd, Harlow, UK; 60 mg/ml), the femoral vein was exposed and 0.5 ml of pontamine sky blue (10 mg/ml in 0.85% NaCl) injected into the vessel and allowed to circulate for 15 minutes prior to culling by cervical dislocation. Uterine horns were excised and discrete blue-stained implantation sites counted immediately following rinsing in saline and fixation in 4% formaldehyde and 2% glutaraldehyde in 0.1 M phosphate buffer, pH 7.2. In some experiments, control or experimental rats from the diet regimen described above for postnatal studies, plus rats maintained on 9% LPD during 0-10 days of development, were sacrificed on day 10 by cervical dislocation and decidual masses from uteri were carefully dissected out. The dimensions of decidua and enclosed visceral yolk-sacs were measured using a calibrated eye-piece graticule.

Control and 9% LPD fed rats were withdrawn from food for 2 hours and, immediately following cervical dislocation and dissection, blood was removed by heart puncture. For glucose analysis, 200 μl blood was allowed to clot on ice for 30 minutes before centrifugation (1000 g; 10 minutes) and storage of serum at −80°C. Remaining blood was allowed to clot overnight at 4°C before centrifugation and serum collection. Radioimmunoassay analysis of insulin (Biogenesis, Poole, UK) and IGF1 (Diasorin, USA) was performed. Prior to IGF1 determination, IGF binding proteins were extracted using the acid-ethanol cryo-precipitation method (Breier et al., 1991). Glucose levels were determined by the glucose oxidase method (Sigma, UK). Amino acid analysis was performed with an ABI 420A derivatisation/analysis system (PE Applied Biosystems, Warrington, UK). Nor-leucine was added to each sample as an internal standard.

Student t-test was used to compare differences between experimental and control treatments. Data are expressed as mean±s.e.m. Mann-Whitney U-test was used for embryo dead cell and mitotic index analyses (Table 5).

Age- and weight-standardised rats fed standard chow diet were naturally mated overnight and plug-positive females switched immediately to either control 18% casein diet or LPD 9% casein (Langley and Jackson, 1994; Table 1) only for the duration of preimplantation development (0-4.25 days) before all females were supplied 18% control diet for the remainder of gestation. Food consumption, equivalent energy intake and maternal weight gain during preimplantation and later periods of development were not significantly different between control and LPD-fed animals (Table 2). Similarly, gestational length, litter size and the male:female offspring ratio were unaffected by the LPD regimen (Table 3). However, female birthweight was significantly reduced in maternal LPD-derived offspring and, up to 7 weeks of age, these animals appeared to overcompensate, with significantly heavier weight gain than control females before both groups showed equivalent weight at 11-12 weeks (Table 3). Although birthweight of maternal LPD-derived male offspring was not significantly lower than that of controls, they displayed the same pattern of growth overcompensation during early postnatal development (Table 3). The LPD regimen also induced a significant increase in systolic blood pressure within male offspring measured at both 4 and 11 weeks of age (Table 4). Male offspring at 12 weeks also displayed disproportionate sizes of liver and kidney in relation to their body weight (Table 4).

These data reveal that maternal undernutrition during the preimplantation period of development may influence the programming of birthweight, postnatal growth rate, systolic blood pressure and the relative size of specific organs. Gender-specific sensitivities appear to exist in the programming mechanisms.

The phenotype of preimplantation embryos following maternal 9% LPD from the time of natural mating was investigated to gain insight into early mechanisms of programming. The number of blastocysts collected from uteri at 4 days after natural mating was not affected by maternal LPD (LPD group: 12.5±0.72 per dam; control group: 11.4±1.29 per dam). Embryos collected at day 3.5, 4.0 and 4.25 (correlating approximately with 16-cell morulae, early blastocyst and expanding blastocyst stages, respectively) were analysed by differential nuclear labelling to determine the number of cells within trophectoderm and ICM lineages, the progenitors of postimplantation chorio-allantoic placental and foetal lineages, respectively. At the morula stage, when allocation and segregation of these cell types initiates (Fleming, 1987), cell numbers within trophectoderm and ICM lineages were not affected by maternal LPD (Fig. 1A). However, at the early blastocyst stage, maternal LPD caused a significant reduction (approx. 15%) in the number of cells within the ICM while no effect was evident in the trophectoderm (Fig. 1B). At the expanding blastocyst stage, ICM cell numbers continued to be reduced but, in addition, significantly fewer trophectoderm cells were present after maternal LPD (Fig. 1C).

To determine the basis of reduced blastocyst cell numbers following maternal LPD, apoptotic dead cell index and mitotic index of embryos were analysed using nuclear fluorescent staining (propidium iodide, TUNEL) and serial optical sectioning with confocal microscopy. In these experiments, embryos were collected at day 4.25 (early blastocyst) from dams fed either 9% or 6% LPD, or control diet from day 0. The dead cell index in blastocysts was not affected by the type of maternal diet used (Table 5). However, the mitotic index was significantly reduced in response to 6% but not 9% maternal LPD (Table 5). Nonparametric correlation analysis (Spearman’s rank correlation) showed no significant relationship existed between embryo cell number and either dead cell or mitotic indices for each diet treatment (data not shown). Collectively, these data indicate that preimplantation embryos may have fewer cells after maternal LPD, owing to a slower rate of proliferation rather than to increased apoptosis. The number of uterine implantation sites was also determined following 9% LPD and control diet treatments over days 0-5. No significant difference was observed between these groups (9% LPD, 13.43±0.9, n=7 rats; 18% control, 13.50±1.15, n=6). Similarly, analysis of uteri at three other time points over the period 4.25-5.5 days showed no significant difference between groups (data not shown). These data indicate that, despite having fewer cells, the capacity of embryos to implant following maternal LPD remains unaltered, consistent with the absence of effect on litter size (Table 1).

To determine the consequence of reduced preimplantation lineage sizes on post-implantation development, we examined the effect of maternal 9% LPD during day 0-4.25 on conceptus size at day 10 of development. No significant difference was found in the size of deciduum or enclosed visceral yolk-sac when compared with controls (data not shown). However, if the period of maternal LPD was extended to day 10 of development, this caused a significant reduction in the size of the deciduum (9% LPD: 60.96±0.99 mm³, n=77; control: 73.53±1.06 mm³, n=159; P<0.001) and of the ICM-derived visceral yolk-sac (9% LPD: 6.04±0.21 mm³, n=39; control: 6.72±0.18 mm³, n=78; P<0.05).

To investigate potential mediators of the preimplantation phenotype observed following 0-4 day maternal 9% LPD, we analysed maternal serum levels of glucose (Fig. 2A), insulin (Fig. 2B) and IGF1 during and beyond this period. Although IGF1 levels remained constant (9% LPD 146.76±14.17 ng/ml, n=13; control 144.99±14.17 ng/ml, n=15), at day 4 a significant increase in glucose and reduction in insulin occurred between days 3 and 4, with both returning to control levels by day 6, 2 days after switching back to control diet. Amino acid analysis of maternal serum was conducted at 2 and 4 days after mating and LPD treatment (Table 6). A significant reduction in essential amino acids was evident at both time points after LPD, with six specific amino acids depleted at day 4 (isoleucine, leucine, methionine, proline, threonine and valine).

Using a tightly regulated dietary regimen in the rat model for foetal programming (Langley and Jackson, 1994; Langley-Evans et al., 1999), our study has revealed for the first time that maternal undernutrition during just the preimplantation period of development is sufficient to programme significant changes in postnatal growth and physiology. The brief maternal LPD treatment induced a reduction in birthweight, an apparent overcompensating growth rate post-weaning, an increase in systolic blood pressure and disproportionate growth of specific organs. This remarkable influence and early origin of programming is not reversed, despite the provision of normal control diet for dams throughout the remainder of gestation.

The basis for postnatal programming appeared to reside in the altered protein content of the diet during preimplantation development. Thus, food consumption and energy intake from the isocaloric diet were equivalent in both LPD and control-fed dams throughout preimplantation and later periods of gestation, as was the maternal weight gain. The LPD regimen also had no effect on litter size or, at an earlier stage, the number of blastocysts generated or implantation sites formed. This suggests that maternal protein undernutrition during early development does not compromise foetal viability by inducing pregnancy termination, but rather has more subtle effects on growth-related criteria. Reduced birthweight and abnormal organ/body weight ratios have been observed in a number of studies of undernutrition during gestation (e.g., Langley-Evans and Nwagwu, 1998; Langley-Evans, 2000; Desai et al., 1996; Langley-Evans et al., 1996). The capacity to compensate for low birthweight by subsequent accelerated weight gain has also been identified in previous maternal nutrition studies (Desai et al., 1996; Langley-Evans et al., 1996), illustrating a postnatal ‘window’ for growth regulation. Our data show that undernutrition during the preimplantation period can activate postnatal growth regulation but this does not protect against the onset of hypertension.

Some of the effects of the maternal LPD regimen appeared gender-specific, such as reduced birthweight in female offspring and increased systolic blood pressure and abnormal organ/body mass ratios in male offspring. Natural differences between male and female control animals in some of these parameters (Desai et al., 1996; Langley-Evans et al., 1996; present study) may confer gender-specific susceptibility to programming. If programming initiates in the preimplantation embryo (see later), specific gender differences at this stage of development may also be contributory. Male embryos of different mammalian species proliferate to the blastocyst stage at a faster rate than female embryos (reviewed in Erickson, 1997). Although the molecular basis for this transient selective advantage has not been characterised (‘growth factor Y’, Erickson, 1997), it suggests that male preimplantation embryos have a greater capacity to respond to the maternal environment and may, as a consequence, exhibit heightened sensitivity to specific programming influences.

To investigate the mechanisms responsible for early programming, we first examined the effect of maternal LPD on cell numbers within preimplantation lineages. This study revealed that initially the ICM, and subsequently the trophectoderm lineages of the blastocyst were significantly reduced by maternal undernutrition. In an early study on the mouse, maternal protein undernutrition for 2 or 4 weeks prior to mating and during the preimplantation period adversely affected the rate of fertilisation and blastocyst development (Munoz and Bongiorni-Malave, 1979). The phenotype reported in the present study is less severe and excludes dietary influence on oocyte maturation. We found that the reduced preimplantation lineage sizes following maternal LPD was not accounted for by an increased rate of apoptosis, which remained unaltered by LPD. Rather, our data indicate that the rate of cellular proliferation may be slowed by maternal undernutrition. Although the mitotic index of embryos derived from mothers fed 9% LPD was unaltered compared with controls, a significant reduction in mitotic index was observed in response to the more severe 6% LPD. We suspect that the extent of cell number reduction (approx. 15%) and the short duration of mitosis (approx. 20 minutes) in blastomeres cycling at approx. 12 hour intervals may provide insufficient sensitivity to identify a change in mitotic index following 9% maternal LPD.

The phenotype of reduced cell numbers following maternal LPD is developmentally significant, as sizes of preimplantation lineages have been shown to exert long-term effects throughout gestation, possibly as a consequence of inappropriate stem-cell allocation for normal growth. In normal mouse embryos, cell interactions from the time of ‘compaction’, when E-cadherin adhesion is initiated, play a critical role in regulating the orientation of cell divisions resulting in blastocysts that contain ICM and trophectoderm cell numbers within a relatively narrow range (Fleming, 1987). Experimental studies in mice have shown consistently that artificial reduction in total preimplantation cell numbers reduces the size of egg cylinder stage conceptuses up to day 8 after transfer, delays the timing of gastrulation and morphogenesis, and significantly increases the rate of pregnancy loss (Tam, 1988; Power and Tam, 1993; Hishinuma et al., 1996). However, during later gestation, increased rates of proliferation in surviving embryos can compensate for deficient numbers of preimplantation blastomeres so that foetal size may be restored to the normal level (Power and Tam, 1993; Hishinuma et al., 1996). In addition, culture conditions that caused a reduction in mouse ICM and total blastocyst cell number also reduced their long-term viability after transfer (Lane and Gardner, 1997), and a deficient ICM cell number has been identified as a potential causative component of foetal growth retardation and large placenta, which are characteristic of the inherited BB/E diabetic rat model (Lea et al., 1996).

The long-term influences of deficient preimplantation lineage sizes has broad significance among a range of mammalian species and is implicated in the so-called ‘large offspring syndrome’ in which in vitro manipulated sheep and cattle preimplantation embryos can give rise to abnormally sized lambs and calves after transfer (Thompson et al., 1995; Walker et al., 1996; Leese et al., 1998; Young et al., 1998; Sinclair et al., 1999; Barnes, 2000). In the bovine embryo, in vitro culture conditions appear to retard embryo compaction, which in turn reduces the allocation of cells to the ICM lineage and may lead to abnormal foetal and perinatal growth (Walker et al., 1996; Van Soom et al., 1997).

We also examined maternal serum to investigate whether specific changes in the maternal environment might be responsible for the reduced proliferation of preimplantation embryos following LPD. This analysis revealed a significant increase in glucose and reduction in insulin levels. Elevated glucose levels in vitro have been shown to retard the proliferation of rat blastocyst ICM and trophectoderm cell populations (De Hertogh et al., 1991) and, in the mouse, alter the metabolic state of the embryo (Moley et al., 1996). A number of peptide growth factors, expressed from either or both maternal and embryonic genomes, stimulate proliferation in the preimplantation embryo, which is mediated by signalling through their specific receptors (reviewed in Kane et al., 1997). In relation to our maternal LPD data, insulin and IGF family growth factors have been shown in vitro to stimulate preferentially ICM proliferation in early blastocysts (Kaye et al., 1992; Kane et al., 1997). Insulin is not expressed by the preimplantation embryo and maternally derived insulin is delivered to the ICM by receptor-mediated transcytosis across trophectoderm. Furthermore, in chemically or genetically regulated rat diabetes models in which maternal serum insulin depletion and hyperglycemia are induced, proliferation of ICM or total cell numbers within blastocysts in vivo is inhibited to a similar extent to that reported here (Lea et al., 1996; Pampfer et al., 1997). Moreover, short-term exposure of preimplantation mouse embryos in vitro to insulin caused a long-term increase in foetal growth rate after transfer (Kaye and Gardner, 1999). We conclude, therefore, that the mildly hyperglycemic maternal environment generated by LPD provides the most likely mechanism to explain retarded preimplantation proliferation and reduced ICM and trophectoderm cell numbers.

The maternal serum analysis also revealed that the LPD regimen caused a significant reduction in the essential amino acids, with six specific amino acids being depleted at day 4 (isoleucine, leucine, methionine, proline, threonine and valine). Maternal LPD maintained throughout rat gestation has similarly been shown to reduce significantly maternal serum threonine, indicative of an important role for this amino acid in foetal growth (Rees et al., 1999). It is noteworthy that significant uptake of four of the serum-depleted amino acids following LPD (isoleucine, leucine, methionine and valine) has been detected during blastocyst formation and expansion in vitro (Lamb and Leese, 1994). Preimplantation embryos require uptake of amino acids to enhance protein synthesis for growth from the blastocyst stage (Lamb and Leese, 1994) and suboptimal exposure to amino acids reduced ICM and total cell number in blastocysts and the rate of foetal development after embryo transfer (Lane and Gardner, 1997). Thus, serum amino acid depletion following maternal LPD may, in addition to maternal hyperglycemia, contribute to the mechanism of retarded preimplantation proliferation. Moreover, cells derived from the ICM lineage (F9 embryonal carcinoma cells) have been shown to respond to amino acid deprivation by expression of growth arrest genes (Fleming et al., 1998). These genes are also expressed in the cleavage-stage embryo (Fleming et al., 1997), indicating one pathway by which retarded preimplantation proliferation may lead to longer-term growth restriction.

Our data have revealed that protein undernutrition in the maternal diet during the preimplantation phase of development can have a profound and long-term effect on foetal and postnatal growth, cardiovascular physiology and organ/body mass characteristics in the next generation. The preimplantation embryo is particularly sensitive to epigenetic modifications that may have programming consequences (Reik et al., 1993; Dean et al., 1998), and our data strongly indicate, but do not prove, that it is the preimplantation embryo itself that is programmed. Embryo transfer experiments will help identify whether the maternal environment, even after restoration of normal diet, contributes to programming during postimplantation development. Our data also indicate that mild hyperglycemia and amino acid deficiency generated transiently within maternal serum as a result of dietary restriction may initiate the phenotypic response and act as a maternal mechanism leading to programming. Available evidence indicates that the large offspring syndrome may also be explained as a consequence of metabolic stress to early embryos, raising the need for more caution in the introduction of new technologies to assisted conception in human (Leese et al., 1998). There is also a parallel between the effects on embryos mediated by poor maternal diet in vivo in our study and the stress effects on embryos mediated by in vitro culture which can activate abnormal glycogen metabolism (Edirisinghe et al., 1984), free oxygen radical release (Johnson and Nasr-Esfahani, 1994) and abnormal gene expression patterns (Eckert and Niemann, 1998; Wrenzycki et al., 1999). Thus, the consequences not only of in vitro culture but also maternal diet in vivo during the preimplantation period need to be addressed more rigorously in human and animal reproduction. Although downstream mechanisms responsible for propagating the early programming events into later gestation and postnatal life have yet to be determined, our results do emphasise the importance of periconceptional diet in the control of mammalian development, with clinical implications for long-term healthcare.

We are grateful to the Medical Research Council, Wessex Medical Trust and the University of Southampton Annual Grants and School of Medicine for their financial support. We thank Professors D.J.P. Barker, E.J. Thomas and M.J. Arthur; and Dr S.C. Langley-Evans for their encouragement and assistance during this project; and staff of the Biomedical Facility, University of Southampton, for technical help.