Main

Restricted perinatal growth is an independent risk factor for the development of type 2 diabetes, hypertension, and coronary heart disease (1,2). The primary animal models of programmed adult disease induce fetal growth restriction through maternal nutrient restriction or uterine artery ligation (3). The development of diabetes and obesity in these models has been most dramatic when rodent pups are cross-fostered to well-nourished dams and then weaned to postnatal high-fat diets (4,5). These studies are consistent with epidemiologic data linking excessive weight gain in infancy with adult metabolic syndrome (6). Although these data suggest that restriction of neonatal nutrition might improve adult metabolic health, a comprehensive approach is essential when considering the long-term consequences of perinatal dietary alterations (7).

Preterm delivery increases the risk of impaired growth during a critical window of brain development. By 36-wk postmenstrual age, ∼90% of premature infants have growth failure (weight less than the 10th percentile of reference fetuses) (8). This perinatal growth failure is associated with permanent reductions in adult weight and height (9). It follows that premature delivery greatly increases the risk of neurodevelopmental impairment, insulin resistance, and adult hypertension (10–12). As parallel epidemics of prematurity and obesity unfold, there is a critical need for research-based neonatal nutritional recommendations that seek to optimize short-term health and long-term outcomes.

We previously evaluated the physiologic consequences of natural perinatal growth restriction in isogenic mice (13). We sought to expand on this model by evaluating the long-term consequences of neonatal catch up growth and investigating the programming of diet-induced obesity. We hypothesized neonatal catch up growth would improve the neurobehavioral and cardiovascular outcomes of low-birth weight mice, but increase their susceptibility to high-fat diet-induced obesity.

METHODS

Animal model.

The investigation was approved by the University of Iowa Animal Care and Use Committee and conforms to National Institutes of Health guidelines. Adult C57Bl/6J mice (Jackson Laboratory, Bar Harbor, ME) were bred for 2 wk and then moved into individual cages. Dams delivered naturally, fostered their own pups, and were maintained on standard rodent chow (4 kcal/g, 6% of energy as fat; 7013; Harlan Teklad, Madison, WI) throughout pregnancy and lactation. Litters were never culled. To lessen the risk of rejection, nests were undisturbed for 24 h after delivery, and initial pup weights were obtained on postnatal d 2. From d 2 to 20, pups were marked with indelible ink and daily weights were recorded. Following World Health Organization guidelines, growth restriction was identified by a gender-specific weight below the 10th percentile (14). Fetal or neonatal growth restriction was, therefore, defined by a weight less than the 10th percentile for the colony on postnatal d 2 or 20, respectively. Appropriate growth was defined by a weight within one SD of the colony mean. Pups were weaned to individual cages and standard rodent chow on postnatal d 21. Mice were continuously recruited into this study from a birth cohort spanning from June 2007 to October 2007. Each study group consisted of mice from at least five different litters, with no more than two mice obtained from a given litter.

Baseline endocrine studies.

At 20 wk, feed intake was measured by recording initial feed weight and subtracting the amount remaining 3 wk later. Glucose and insulin tolerance tests were then performed after a 3-h fast, as previously described (13). For glucose tolerance tests, 20% dextrose was administered by intraperitoneal injection (2 g/kg). After at least 48 h of recovery, insulin tolerance tests were performed by intraperitoneal injection of regular human insulin (0.75 U/kg; Humulin-R, Eli Lilly, Indianapolis, IN).

Neurocardiovascular studies.

After the initial endocrine tests, tail cuff systolic blood pressure (SBP) and heart rate were recorded for 5 consecutive days, as previously described (13). Unconditioned anxiety was measured by placing mice in a 40.6 × 40.6 cm open field (San Diego Instruments, San Diego, California) (15). Infrared beam breaks were recorded for 5 min on 2 consecutive days. To further evaluate anxiety-associated thigmotaxis (tendency to remain along the chamber's walls), center activity was defined as infrared beam breaks occurring inside of a 7.62 cm border. Barnes maze testing was then used to assess spatial learning and memory. The Barnes maze exploits the instinct of small rodents to seek escape to a sheltered environment when placed in an open arena under aversive illumination. The maze consists of a circular platform (92 cm diameter) with 20 equidistant holes around the perimeter. One hole leads to an escape box that is consistently oriented with respect to visuo-spatial cues throughout the room. Each mouse was given four trials per day, with an inter-trial interval of 1 h. The initial training day was followed by four consecutive test days to evaluate spatial learning. Mice were retested 2 wk later to assess memory consolidation and long-term retention. A video camera recorded each session. Escape latency (time taken to find and enter the escape box) was recorded by observers blinded to group assignment.

High-fat diet.

After completion of the baseline studies, all mice were switched to a hypercaloric, high-fat diet (5 kcal/g, 45% of energy as fat; D12451; Research Diets, New Brunswick, NJ). The caloric composition of the diet included 45% fat (39% lard and 6% soybean oil), 35% carbohydrates (18% sucrose, 10% maltodextrin, and 7% corn starch), and 20% protein (19.7% casein and 0.3% l-cystine). During the first week of high-fat diet administration, an exercise wheel (35 cm circumference; Thoren Caging Systems, Hazleton, PA) was placed inside each cage to evaluate voluntary activity. Exercise wheel revolutions were digitally recorded over the ensuing 72 h. Twenty weeks later, glucose and insulin tolerance tests were repeated. Insulin levels were measured on plasma collected from fasting mice immediately before glucose tolerance testing, as previously described (13).

Indirect calorimetry.

While receiving the high-fat diet, mice were housed in individual, airtight chambers warmed to thermoneutrality (30°C) by a circulating water bath. Oxygen consumption (VO2) and carbon dioxide production (VCO2) were calculated based on the gas composition flowing into and away from the closed-circuit during the constant flow of air (AEI Technologies oxygen analyzer S-3A and carbon dioxide analyzer CD-3A). To approximate basal metabolic rate, VO2 was calculated at thermoneutrality, after mice had fallen asleep in the midst of the light cycle. RQ was calculated by dividing VCO2 by VO2.

Data analysis.

All values are presented as mean ± SE. Statistical comparisons were performed by 2-tailed t test or two-way ANOVA with Holm Sidak testing, as appropriate. A value of p < 0.05 was considered significant. All analyses were performed using SigmaStat 3.0 (SPSS Inc., Chicago, IL).

RESULTS

Establishment of the model.

Median litter size was seven pups with an interquartile range of six to nine pups. There were slightly more male than female pups on d 2 (53% male, p = 0.31 by χ2 analysis). Newborn mouse weights were normally distributed (Fig. 1A; Shapiro-Wilk test, W = 0.99). Female mice weighed significantly less than male mice on d 2 (female: 1.76 ± 0.02 g; male: 1.82 ± 0.02 g; p < 0.05) and d 20 (female: 8.1 ± 0.1 g; male: 8.4 ± 0.1 g; p < 0.05). Litter size influenced both d 2 pup weight (R = 0.80, p < 0.05) and neonatal weight gain (Fig. 1B; R = 0.94, p < 0.05). Based on the initial 178 male mice, growth restriction (weight below the 10th percentile) was defined by weights below 1.56 g on d 2 and below 7.1 g on day 20. Based on the initial 156 female mice, the corresponding 10th percentile weights were 1.50 g on d 2 and 6.8 g on d 20. Of the mice with fetal growth restriction, 50% of the males and 35% of the females achieved catch up growth.

Figure 1
figure 1

Isogenic male â–ª and female â—‹ C57BL/6 mice had normally distributed birth weights (A, n = 334 mice, W = 0.99). Neonatal weight gain was inversely proportional to litter size (B, n = 57 litters, R = 0.94). Fetal or neonatal growth restriction was defined by a weight below the 10th percentile at 2 d (C) or 20 d (D). Using this definition, mice were categorized as having appropriate fetal and neonatal growth (AA, â–ª, n = 15 males and 14 females), restricted fetal growth but appropriate neonatal growth (RA, , n = 6 males and 8 females), or restricted fetal and neonatal growth (RR, â–¡, n = 7 males and 6 females). *p < 0.01 vs control.

Full size image

By definition, mice with combined fetal and neonatal growth restriction (RR mice) had significantly reduced weights on d 2 and 20. In contrast, mice with fetal growth restriction followed by accelerated neonatal growth (RA mice) had significant reductions in only d 2 weight (Fig. 1C and D). Based on the observed catch up rates and the definition of growth restriction (weight less than the 10th percentile), 5% of males were RR, 5% of males were RA, 6.5% of females were RR, and 3.5% of females were RA. Mice were continuously recruited until the final cohorts of RR and RA mice were established (six to eight mice per group). For every mouse recruited into the RR and RA groups, mice with appropriate fetal and neonatal growth (AA mice) were retained (15 males and 14 females). For most litters, AA pups were fostered along side RA and/or RR pups.

Adult size and caloric intake.

RR male and female mice had persistent growth failure (Fig. 2). The reductions in adult weight (decrease of 31% in males and 13% in females), length (decrease of 7% in males and 5% in females), and brain weight (decrease of 9% in males and 5% in females) were associated with a significant reduction in feed intake (kcal/d) for both male and female RR mice (Fig. 2D). Consistent with asymmetric growth restriction, brain weight normalized to body weight (mg/g) was significantly greater in adult male and female RR mice (RR male 18.2 ± 0.7 versus AA male 14.7 ± 0.8; RR female 21.0 ± 0.7 versus AA female 19.2 ± 0.4). When also corrected for current body weight, adult feed intake (expressed as kcal/g/d) was not influenced by perinatal growth restriction. RA and AA mice had similar weights, lengths, and caloric intakes. Female, but not male, RA mice had a significant reduction in adult brain weight (Fig. 2C).

Figure 2
figure 2

Adult weight (A), length (B), brain weight (C), and caloric intake (D) were measured among AA (â–ª, n = 15 males and 14 females), RA (, n = 6 males and 8 females), and RR (â–¡, n = 7 males and 6 females) mice. *p < 0.05 and **p < 0.01 vs AA.

Full size image

Phenotypes on standard chow.

Although perinatal growth restriction did not alter fasting glucose levels or glucose tolerance at 4 mo, male RA mice had a significant reduction in insulin sensitivity (Fig. 3). Male and female RR mice, as well as female RA mice were hypertensive (Fig. 4A). Perinatal growth patterns did not significantly alter adult heart rates (Fig. 4B). Male RA mice had decreased activity when placed into an open field (Fig. 4C). Male and female RR mice were also less active in the open field, whereas female RA mice had skewed distribution of movement along the periphery of the field (Fig. 4D).

Figure 3
figure 3

AA (â–ª, n = 15 males and 14 females), RA (, n = 6 males and 8 females), and RR (â–¡, n = 7 males and 6 females) mice received standard feed for 20 wk. Glucose levels were then measured while fasting (A) for intraperitoneal glucose tolerance tests (B), or insulin tolerance tests (C, male; D, female). *p < 0.05 vs AA.

Full size image
Figure 4
figure 4

Tail cuff systolic blood pressure (A) and heart rate (B) were recorded among AA (▪, n = 15 males and 14 females), RA (, n = 6 males and 6 females), and RR (□, n = 7 males and 6 females) mice. To measure generalized anxiety, these mice were then placed into a 40.6 × 40.6 cm open-field and infrared beam breaks were recorded for 5 min (C). Beam breaks into the center were normalized by total activity (D). *p < 0.05 and **p < 0.01 vs AA.

Full size image

After open-field testing, four of the eight female RA mice developed open sores from excessive grooming (an anxiety-linked behavior) and meaningful data could no longer be obtained from that study population. During Barnes maze training, male RR mice escaped from the aversive environment much faster than controls (Fig. 5A). This anxiety driven behavior hampered definitive testing of subsequent learning in male mice (Fig. 5B). Female RR mice had significantly impaired learning (Fig. 5C) and memory, evidence by their delayed escape from the Barnes maze 2 wk after training (Fig. 5D).

Figure 5
figure 5

After open-field testing, four of the eight female RA mice developed open sores from excessive grooming, and meaningful data could no longer be obtained from that study population. All other AA (â–ª, n = 8 males and 9 females), RA (, n = 6 males), and RR (â–¡, n = 6 males and 6 females) mice were placed onto a brightly illuminated Barnes maze and escape time was measured (A). Male (B) and female (C) mice were then trained for 5 consecutive days, and visuo-spatial memory was evaluated 14 d later (test day 19, D). *p < 0.05 and **p < 0.01 vs AA.

Full size image

Phenotypes on high-fat chow.

Male mice with neonatal catch up growth (RA mice) had fasting hyperglycemia, hyperinsulinemia, glucose intolerance, and insulin resistance while receiving a high-fat diet (Fig. 6). Female mice with perinatal growth failure (RR mice) had significantly improved glucose tolerance (Fig. 6C) without significant alterations in insulin sensitivity (Fig. 6D).

Figure 6
figure 6

After AA (▪, n = 14 males and 14 females), RA (, n = 6 males), and RR (□, n = 7 males and 6 females) mice were switched to high-fat feed for 20 wk, glucose (A) and insulin (B) levels were measured while fasting. Mice then received intraperitoneal glucose (2 g kg−1; C) or insulin (0.75 units kg−1; D, squares males and circles females) and glucose levels were monitored. *p < 0.05 vs AA.

Full size image

While on the high-fat diet, male and female RR mice gained significantly less weight than controls (Fig. 7A). In male RR mice, this protection from diet-induced obesity was associated with a significant increase in exercise and basal metabolic rate (resting VO2) (Fig. 7). Although male RA mice did not have excessive weight gain on a postnatal high-fat diet, they did have a reduction in voluntary exercise that was consistent with open field testing (Fig. 7B). Respiratory quotients were not significantly influenced by perinatal growth patterns (Fig. 7D).

Figure 7
figure 7

Weight gain during the 20 wk of high-fat diet administration was measured among AA (â–ª, n = 14 males and 14 females), RA (, n = 6 males), and RR (â–¡, n = 7 males and 6 females) mice (A). During the initial week of high-fat diet administration, exercise wheels were placed into the cages to record voluntary activity (B). By indirect calorimetry, resting VO2 (C), and RQ (D) were then determined. *p < 0.05 and **p < 0.01 vs AA.

Full size image

DISCUSSION

Over the past decade, investigations into the developmental origins of adult disease have raised awareness of the multifaceted complications of an adverse intrauterine environment. Although demonstration that fetal growth restriction increases the risk of adult disease has prompted interest in the promotion of perinatal growth, studies have shown neonatal over-nutrition may exacerbate, rather than prevent, the evolution of adult disease (16,17). In contrast, studies in male mice have suggested neonatal malnutrition may prolong lifespan and prevent the development of diet-induced obesity (18–20). We expanded upon these studies in demonstrating that natural perinatal growth restriction provides protection from diet-induced metabolic syndrome, but at the expense of neurocardiovascular health.

Mice with combined RR mice remained small as adults. This reduced adult size is consistent with data showing premature and small for gestational age term infants have permanent reductions in adult height and weight (9). Our data showing that perinatal growth restriction leads to adult hypertension are consistent with data from the Helsinki birth cohort showing hypertension strongly associates with impaired growth through the first 4 y of life (1). The blood pressure normalization we demonstrated following catch up growth in male mice differs from the findings of Singhal et al. (17), which showed that neonatal overnutrition increases diastolic blood pressure among low-birth weight term infants. Apart from potential species differences, comparisons are hampered by the limited caloric intake data obtained in our study, the clinical trial's limited follow-up (51%) and a significantly enriched proportion of males in the clinical trial's overnutrition population. Further studies with adequate power to demonstrate sex-specific effects are needed to better define the role of neonatal catch up growth in the programming of hypertension.

Controlling for confounding variables, reduced neonatal brain growth has been consistently associated with long-term intellectual and behavioral impairment (21,22). Our data strengthen this association in showing genetically identical growth restricted mice raised in identical postnatal environments have permanent reduction in brain weight and cognitive performance. Interestingly, the decline in brain growth and cognitive performance among growth-restricted mice was strongly paralleled by the presence of tail cuff hypertension. Given the physical restraint and handling required with tail cuff blood pressure measurements, it is possible the observed hypertension is a further expression of programmed anxiety.

We used two separate behavioral tests to evaluate the presence of anxiety. Mice with perinatal growth restriction displayed markedly decreased movement when placed within an open field chamber. This behavior seemed to be related to anxiety, rather than a general reduction in movement, as growth restricted male mice had significantly increased activity on exercise wheels, and they escaped more rapidly from the stress-evoking Barnes maze. Related to these results, Vickers et al. (23) have shown both maternal undernutrition and postnatal overnutrition decrease adult rat movement in an anxiety-provoking behavioral apparatus. Further studies are necessary to evaluate the roles of fetal and postnatal growth on the programming of locomotor activity under a variety of environmental conditions.

Intriguingly, neonatal catch up growth normalized Barnes maze performance for male mice, whereas catch up growth accentuated the expression of anxiety in female mice. In concert with the brain weight data, these data suggest female mice have an earlier neurodevelopmental window of susceptibility, such that later interventions, including neonatal catch up growth, are less likely to modify female mouse neurocognitive performance. Considering the relative immaturity of the mouse brain at the time of delivery (roughly analogous to the developmental stage of the third trimester human fetus), these data are consistent with clinical data showing premature male infants are at twice the risk of growth failure-associated neurodevelopmental impairment (24). This sexually dimorphic neonatal window of vulnerability (or reversibility) may also underlie the greater reduction in adult size seen in male RR mice fostered in large litters, and the higher rate of postnatal catch up growth seen in male mice fostered in small litters. The etiologies for this sexual dimorphism in growth trajectory and resultant susceptibility to neurocardiovascular disease are not known.

Consistent with an expanding body of literature, catch up neonatal growth led to insulin resistance among low-birth weight male mice. These mice had marked insulin resistance without significant hyperglycemia on standard chow, suggesting that the loss in insulin sensitivity precedes the development of diet-induced hyperglycemia. This speculation is further supported by the 4-fold increase in fasting insulin among these mice, as well as clinical data showing that among premature infants, neonatal growth velocity is inversely associated with adult insulin sensitivity (6). With data supporting a role for early exercise in the prevention of insulin resistance and diabetes (25), the reduction in exercise seen among the diabetes-prone male RA mice merits further consideration as a target for early intervention.

Our data are consistent with previous findings in male offspring of undernourished rats (5,26). Maternal starvation to 30% of ad libitum intake throughout pregnancy reduced offspring weight through adulthood and programmed tail cuff hypertension and hyperinsulinemia (5). Similarly, profound maternal malnutrition (60% protein restriction) throughout pregnancy and lactation persistently decreased the body weight of adult offspring (26). As demonstrated in our study, postnatal hypercaloric nutrition amplified these programmed phenotypes. Intriguingly, both of these studies showed prenatal undernutrition followed by neonatal catch up growth significantly increases retroperitoneal fat pad weight in male rats, suggesting a possible mechanism for the insulin resistance we observed in RA male mice.

Although the degree of self-injury among female RA mice made definitive testing in this gender impossible, our baseline data support a greater predisposition for male mice to programmed diabetes. This finding is consistent with data supporting a sexually dimorphic susceptibility to diabetes among inbred C57BL/6 male mice (27,28), and enhanced diabetes susceptibility among protein malnourished male, but not female, rats (29).

Among the factors that regulate postnatal growth, we evaluated caloric intake, resting energy expenditure, and voluntary activity. Perinatal growth restriction seemed to reset a number of homeostatic mechanisms, including a reduction in caloric intake. In addition to this reduced intake, perinatal growth restricted male mice had increased exercise wheel utilization and increased resting energy expenditure. Although the perinatal origins of this altered resting energy expenditure have not been described, the trend seen toward increased heart rate might be contributory.

By allowing pups to wean naturally to standard rodent feed, our study uniquely mirrored the typical growth patterns seen in other naturally growth restricted populations. We demonstrated that naturally occurring perinatal growth restriction leads to hypertension and behavioral alterations but overrides genetic predisposition and protects the same mice from metabolic syndrome when a high-fat diet is introduced in adulthood. These findings support previous studies demonstrating that accelerated postnatal growth may be associated with detrimental effects on adult metabolic control, but postnatal growth impairment negatively influences neurocardiovascular outcomes. Further studies are needed to identify strategies that facilitate brain growth while simultaneously protecting the survivors of adverse intrauterine exposures from increasingly harsh postnatal environments. Identification of candidate genes and genetic polymorphisms holds the promise of tailored pharmacologic and nutritional interventions to better match disease susceptibilities, perinatal growth, and the prevailing environment.