Neurotoxicology and Teratology 32 (2010) 362372

Contents lists available at ScienceDirect

Neurotoxicology and Teratology

j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / n e u t e r a

Moderate developmental undernutrition: Impact on growth and cognitive function

in youth and old age
M.E. Gilbert a,b,, R. MacPhail a, J. Baldwin c, V.C. Moser a, N. Chernoff a
a
Toxicity Assessment Division, U.S. Environmental Protection Agency, Research Triangle Park, NC, USA
b
Department of Psychology, University of North Carolina, NC, USA
c
Meredith College, Raleigh, NC, USA

a r t i c l e i n f o a b s t r a c t

Article history: Low weight at birth is a common adverse developmental effect reported in human populations and animal
Received 25 June 2009 toxicity studies. Epidemiological evidence links low birth weight to a syndrome of metabolic changes that
Received in revised form 29 October 2009 increase later risk for obesity, type 2 diabetes, hypertension, and cardiovascular disease. The present study
Accepted 21 December 2009
used a four-treatment cross-over experimental design to evaluate the selective impact of early nutritional
Available online 31 December 2009
deciency on metabolism and brain function across the lifespan of male Sprague Dawley rats. Undernutrition
was induced prenatally by restricting maternal food intake to 50% of ad lib from GD3 to birth. Postnatal
Keywords:
Undernutrition undernutrition was induced by fostering pups at birth to nave dams in large (n = 16) litters as opposed to
Prenatal stress small (n = 8) control litters. Body weights were monitored in the early neonatal period, in early adulthood
Brain beginning at 5 months and through to senescence at 21 months of age. In contrast to recent reports, no
Cognition increase in the prevalence of obesity was seen in animals born to food restricted dams and reared under ad
Development lib feeding conditions. Behavioral tests of locomotion, learning and memory were performed in young,
Learning and memory middle-aged, and aged animals. No effects of pre or postnatal nutritional history were detected. Age-
Neurotoxicity dependent reductions in locomotor activity were detected, as well as decits in spatial learning as measured
Obesity
in the Morris water maze and in context fear conditioning. These ndings indicate that moderate fetal
Nutritional programming
undernutrition followed by neonatal adequate nutrition does not appear to lead to obesity or neurological
dysfunction in young adulthood or old age.
Published by Elsevier Inc.

1. Introduction undernutrition-induced reduction in numbers of pancreatic cells [29];

hypertension stemming from alterations in glucocorticoid hormones
Low weight at birth in relation to gestational age is a common (see [28]); and cardiovascular disease, particularly associated with low
adverse developmental effect reported in human populations [21]. The birth weight populations experiencing normal nutrition postnatally
most common cause of low birth weight globally is maternal [3,4,7,31,42].
undernutrition [25]. Prenatal and postnatal undernutrition is increas- Many of these observations have been interpreted in the context of
ingly recognized as a critical factor in both human health and in standard environmental programming, which describes the process where
laboratory animal bioassays in the regulation of environmental environmental insults during critical phases of development lead to
chemicals and pharmaceuticals. Impaired fetal growth and subsequent permanent changes in physiology and metabolism of the adult offspring
low birth weight are major predictors of pre- or postnatal morbidity and (see reviews [19,27]). Maternal nutritional status is one of the major
have long-term clinical consequences. Increasing epidemiological factors implicated in environmental programming [8]. Animal studies
evidence links low birth weight to a syndrome of metabolic changes specically designed to examine the impact of nutritional programming
(see [46,18,31,47]). These adverse health conditions include increased on the development of the metabolic syndrome have reported
risk of developing obesity when embryo/fetal undernutrition occurred hyperphagia, obesity, and insulin resistance in adult offspring following
during the rst half of pregnancy [44]; type 2 diabetes resulting from in utero moderate [13,14] and severe (e.g., [56]) nutritional deprivation.
The impact of fetal programming on increased risk for development of
neurological disease has also been postulated, with nutritional factors
implicated in the etiology of schizophrenia, anxiety disorders, adult
Corresponding author. Toxicity Assessment Division, Neurotoxicology Branch (MD- depression and suicide [41,52,58,60]. Similarities in the neurobehavioral
B105-05), National Health and Environmental Effects Research Laboratory, 109 TW
Alexander Drive, U.S. Environmental Protection Agency, Research Triangle Park, NC
and neurochemical manifestations of offspring nutritionally stressed in
27711, USA. Tel.: + 1 919 541 4394; fax: +1 919 541 4849. utero, offspring of dams exposed to dexamethasone, and offspring of
E-mail address: [Link]@[Link] (M.E. Gilbert). dams exposed to restraint stress suggest that fetal exposure to maternal

0892-0362/$ see front matter. Published by Elsevier Inc.

doi:10.1016/[Link].2009.12.008
 M.E. Gilbert et al. / Neurotoxicology and Teratology 32 (2010) 362372 363

glucocorticoids may underlie some of the nutritionally-mediated effects dent group of dams (see [10]). Following these preliminary studies, a
on brain and behavior [26,39,58]. total of 96 timed pregnant dams were obtained on GD3 (sperm positive
Body weight reductions in dams and offspring are also a frequent is GD0). On arrival 62 dams were allowed free access to food throughout
observation in standardized testing protocols designed to assess the pregnancy (Con), 34 were provided with daily food allotments equal to
developmental toxicity of environmental contaminants [11,49]. This is 50% of the ad lib consumption levels (Treat). On the day of birth,
an especially critical factor in the regulatory process since the Lowest offspring were separated by sex and randomly distributed to a group of
Observed Adverse Effect Level (NOAEL) in standard rat developmental 58 control mothers (fed ad lib throughout pregnancy). Postnatal
toxicology bioassays is determined by weight reductions alone in 84% undernutrition was accomplished by cross-fostering pups to untreated
and 71% of the maternal and fetal groups respectively [11]. Given the control mothers in litters of 8 (_Con) or 16 (_Treat) using equal numbers
consequences of reduced birth weight in the regulatory process, coupled of males and females, a manipulation that has proved to be effective in
with an increased interest in the long-term effects of in utero and early altering nutritional status and weight gain in offspring [10]. This
postnatal exposure to xenobiotics, the present study investigated the generated 4 study groups, the treatment name before the underscore
potential impact of pre and/or postnatal alterations in nutritional status signifying in utero condition, the name following the underscore the
in rats on metabolism and brain function throughout life. We chose a postnatal condition:
model of moderate prenatal undernutrition, reducing ad lib food intake
by 50%. Postnatal undernutrition was induced by fostering large litters of Treatment Prenatal maternal group Postnatal offspring group
16 pups to nave dams at birth. We included a group in which nutritional Con_Con Ad lib food 8/litter
status was compromised throughout the pre and postnatal period, to Treat_Con Undernutrition 8/litter
cover the period of late gestation in human as rodents at birth are Con_Treat Ad lib food 16/litter
premature relative to humans, the last trimester of pregnancy Treat_Treat Undernutrition 16/litter
corresponding to the rst two postnatal weeks in rat.
Any postnatal undernutrition manipulation in nursing pups is The cross-fostered litters were weighed on PD1, 7, 14, 21 and
challenging as extra-nutritional factors such as maternal care, hypother- weaned to ad lib feeding conditions on PN22 with Purina 5001 chow.
mia, hormonal environment, nursing competition and social interactions One to two males from each litter housed 2/cage by experimental group
may contribute to the overall cognitive effects. However, in the present were used for neurobehavioral studies. Adult animals were weighed
study, the increase in litter size was moderate compared to other models once monthly beginning at 5-months of age. A subset of animals was
(16 vs 18 or 20 pups [45,53,55]), and fostering to nave dams eliminated sacriced after behavioral testing at 6 months of age, and a second at
concerns of reduced milk yield observed under conditions of lactational 22 months of age for neurogenesis studies to be reported elsewhere.
food restriction [34]. This model produces a moderate level of undernu- Different animals were used at each age, but the same animals were
trition and behavioral decits observed in offspring can be primarily assessed on motor activity and fear conditioning at the early ages
attributed to nutritional over environmental factors (see [12,53]). (b12 months of age), and motor activity, fear conditioning, and water
One of the tenets of environmental programming is that adaptive maze in older animals, in that order.
responses signaled in utero through changes in gene proles produce
a phenotype that is matched to the anticipated environment [20]. A 2.3. Motor activity
mismatch between the prenatal to postnatal environment is predicted
to have the most untoward effects, as programming is maladaptive to Motor activity was assessed using six photocell devices (Motron
the existing conditions. Therefore, the present study utilized a four- Electronic Motility Meter, Motron Produkter, Stockholm, Sweden) in
treatment cross-over experimental design to evaluate the selective independent groups of animals at 3, 13, 18 and 24 months of age. Each
impact of nutritional programming on metabolic and brain function, device had a 5 8 matrix of photodetectors in the platform that was
including a group where undernutrition spanned the pre- and early illuminated by a single overhead incandescent (30 W) lamp; move-
postnatal period. In this paradigm, nutritional status of the dams and ments that occluded these detectors were recorded as horizontal
pups remained consistent during fetal and neonatal life (either activity. An array of six photoemitters and detectors was also placed
adequate or inadequate pre- and postnatally) or, the nutritional 16.5 cm above the platform to record vertical activity (rearing). Each
conditions were reversed between the pre- and postnatal periods. It device was housed in a sound- and light-attenuating ventilated cubicle
was hypothesized that a change in nutritional conditions from the in a dedicated test room. A clear plexiglas chamber (33 21 26 cm)
pre- to the postnatal period would lead to obesity and neurological with removable lid was placed over each platform to contain the rat
dysfunction and that these effects would be exacerbated with age. during testing. Motor activity was recorded in ve successive 6-min
intervals during a single 30-min session. All testing occurred between
2. Methods 0700 and 1200 h on the same day. Order of testing was counterbalanced
across dose groups. Motor activity data were statistically analyzed using
2.1. Animals a two-way ANOVA, with age at testing and treatment condition as
between-subject factors. Separate two-way ANOVAs were conducted
Time-pregnant Sprague Dawley CD rats were obtained from Charles for horizontal activity and for vertical activity.
River Breeding Laboratory (Raleigh, NC). Animals were housed
individually in polycarbonate cages on pine shaving bedding in an 2.4. Fear conditioning
AAALAC certied facility. Animal rooms were maintained at 22 C, with
relative humidity of 4060%, and unless otherwise stated, a 12 h light: Trace fear conditioning to both cue and context was assessed in
dark (7:00 am:7:00 pm) photoperiod. BioServ (Frenchtown, NJ) 500-mg independent groups of animals at 4, 12 and 18 months of age. Animals
food pellets were used throughout gestation and lactation until pups were were adapted to a reverse light:dark cycle (lights on at 8:00 pm, off at
weaned. 8:00 am) for two weeks so that training and testing could occur during
the active phase of their diurnal cycle. Two different chambers were
2.2. Pre- and postnatal undernutrition used, one for training and context testing, and a different chamber in a
different room for cue testing. The conditioning chamber (Habitest,
A restricted diet consisting of 50% of the daily calculated food Coulbourn Instruments, Allentown, PA) had aluminum side walls with
consumption in the dam during gestation was determined based upon Plexiglas front and back walls, an aluminum ceiling, and a standard grid
food consumption monitored daily from GD1 to GD21 in an indepen- oor consisting of parallel steel rods (5 mm diameter and 15 mm
364 M.E. Gilbert et al. / Neurotoxicology and Teratology 32 (2010) 362372

spacing). The dimensions of the chamber were 31 25 29 cm into the tank, facing the wall, and allowed to circumnavigate in search of
(length width height). A small-animal shock generator (H13-16, the escape platform for a maximum of 60 s. On each day the start points
Coulbourn Instruments) delivering a scrambled footshock (1 mA, 0.5 s used for each trial varied in a pseudo-random sequence such that no two
duration) served as the unconditioned stimulus (US). Conditioning trials on the same day commenced from the same starting quadrant.
boxes were enclosed in a sound-attenuating chamber. The conditioning Latency to reach the escape platform was recorded and the animals
chamber was sprayed with Windex cleaner to provide a distinct permitted 15 s on the platform before removal from the tank. If an
olfactory cue and was wiped down with this cleaner between each animal failed to locate the platform within 60 s, it was guided by the
animal. On day 1, animals were placed in the conditioning chamber that experimenter, placed upon it for 15 s, and assigned a latency score of
was dimly illuminated by a single house light. The training procedure 60 s for that trial.
consisted of a 2-min baseline period followed by onset of a 15 s
compound light+ tone conditioning stimulus (CS). The light compo- 2.5.1. Probe trials
nent of the CS (1.5 cm round disk, 18 lx) was centered on the left wall of A series of probe trials was conducted on trial 1 of test days 3, 6, 9, 12
the test box. The tone component of the CS came from a Sonalert (86dB), and 19 in which the platform was removed and animals were allowed to
located 5 cm from the top and 2.5 cm from the right side of the back wall swim freely for 60 s. The platform was reinserted at the end of this
of the chamber. To increase task difculty, a non-contingent distractor period and animals were permitted to rest on it for 15 s to avoid
stimulus comprised of a dimly ashing light placed on the chamber extinction of the search behavior. The percentage of time the animals
wall opposite the cue light and elevated to the top of the chamber was spent in each quadrant of the pool was recorded for each probe trial.
randomly presented throughout the training episode. Activity was Escape latencies on probe trial days were based on data from the 2nd
monitored by an infrared motion detector (Colbourn Instruments) trial only.
mounted on the ceiling of each test chamber. Reduction of activity was
taken as the measure of fear-induced learning. 2.5.2. Cue learning and swim speed
For trace fear conditioning, a 30-s interval was interposed between Visual and motor competence was assessed to determine if pre- or
CS offset and US onset during training. Training consisted of 2 CSUS postnatal undernutrition inuenced performance in the water maze
pairings with a 3 min intertrial interval. The following day animals task, independent of cognitive function. Cue testing was conducted on
were placed into the training chamber and activity counts within a 5-min day 13 and again at the end of testing on day 20. Under cue conditions,
test period recorded as a measure of conditioning to context. Animals the platform was placed in the opposite quadrant and was visible above
were returned to holding cages in a dimly-lit room for 1 h. Conditioning to the water level with a glove identical to that worn by the experimenter
cue was assessed by placing animals in a different test chamber in a suspended above it. Animals were placed in the pool as previously
different brightly lit room with the sound-attenuating enclosure doors left described and latency to reach the visible platform was recorded on two
open. The chamber was of similar dimension as the training one but was consecutive trials. Swim speed was determined using a videotracking
equipped with black and white vertical stripes on the back and two side system of pathlength (HVS Image, Buckingham, UK) from a videotape
walls and a smooth white plexiglas oor. Unlike the training/context made of the rst probe trial where animals swam freely for 60 s in the
chamber, this chamber was cleaned between animals with water to absence of the escape platform. For this trial animals were marked with
maintain distinct olfactory environments. As with training and context a black marker to permit optimal performance of the tracking device
testing, activity was monitored by a ceiling-mounted infrared motion when tapes were scored.
detector identical to that used in the training box. CS presentation was
identical to that used during training but no shock (US) was delivered. 2.6. Morris water maze acquisition at 20 months

2.5. Morris water maze acquisition at 13 months Under the original testing protocol, evidence of learning was not
readily apparent in either latency or probe data in 13 month-old animals
Spatial learning was assessed in the Morris water maze in two (see below). As such it was not prudent to assess older animals in the
independent sets of animals at 13 (1314 months) and 20 (18 same way. In addition, in older animals, it is possible that motoric
20 months) months of age who had previously been tested on fear competence may be challenged and could possibly inuence task
conditioning. Animals were placed back to normal 12 h:12 h light:dark performance. So, to assess learning ability in older animals the task was
cycle for 1 month prior to testing. Data are also described for a group of modied to increase the probability of learning. We utilized a data
nave 3-month-old Sprague Dawley rats for comparison (see below). acquisition system for which videotracking of pathlength was available
One male offspring from each treatment group was selected from each (HVS Image, Buckingham, UK). A smaller diameter pool was used
litter to evaluate spatial learning. The water maze consisted of a white (1.35 m vs 1.7 m diameter) which dramatically reduced the search area
circular galvanized tank with a diameter of 1.7 m, lled with tap water (2.7 m2 to 1.43 m2) and facilitated acquisition of the task [57]. To
adjusted to 24 C and made opaque by the addition of white powdered facilitate tracking the interior of the tub was black and black powdered
paint. Four locations around the edge of the pool were dened as start paint was added to obscure the platform. The room in which the maze
points, and divided the pool into 4 equally sized quadrants. A white was placed was also smaller (3 3.3 m) and the tub was placed in the
circular acrylic escape platform, 10 cm in diameter, was placed 2 cm corner of the room, approximately 0.3 m from adjacent walls. The
below the surface of the water in the middle of one quadrant. The pool experimenter was positioned distally to the maze beside a laboratory
was placed in the center of a 3 3.9 m room containing invariant spatial cart holding a computer and recording equipment. The overhead
stimuli including a door, a computer and VCR on a cart, and four posters uorescent light remained off during testing, primary lighting provided
of different shapes and colors mounted on each of the four walls. A video by two sconce lamps on each of the two adjacent walls closest to the
camera suspended from the ceiling above the middle of the tank maze. Additional visual cues included a door, sink and benchtop, and
permitted the observer to follow the animal's behavior on a video- distinct black geometric shapes on three walls. Squads of 4 animals
monitor placed in one corner of the room. Latency to reach the platform remained in the room on the benchtop during the test session.
and time spent in each quadrant were recorded on a personal computer Pathlength as well as trial latency were recorded. Additional alterations
using custom-designed software. On the day prior to acquisition testing, in protocol included 2 trials/day as in the 13-month group, but for 5
all animals were acclimated to the task by individually placing them in consecutive days (MondayFriday) followed by 4 days the following
the pool and allowing a 60-s free swim with no platform present. week for a total of 9 acquisition days. No probe trials were arranged
Thereafter, animals were tested for 2 daily trials, each trial separated by during acquisition, but rather the rst trial on the 10th day was a probe
approximately 35 min for 20 consecutive days. Animals were placed in which no platform present and the animals swam freely for 60 s. The
 M.E. Gilbert et al. / Neurotoxicology and Teratology 32 (2010) 362372 365

second trial on the nal day constituted the cued trial where the body weight monitoring began at 5 months and continued once
platform was placed in the opposite quadrant and the water level monthly until 21 months of age (Fig. 1B). Both postnatally undernour-
lowered to make it visible above the water surface. ished groups continued to be of lower body weight than animals of the
Con_Con condition. Body weight in the adult offspring of the Treat_Con
3. Results group was intermediate, slightly lower than Con_Con at the majority of
sampling times. The sample size across all four-treatment conditions
3.1. Body weights and health status dropped over the sampling period as animals died or were humanely
euthanized due to failing health. As such, it was not possible to evaluate
Body weight was monitored in dams during gestation and in body weight over time so separate statistical analyses were performed
offspring at birth (PN1), PN7 PN14 and PN22. These data were for body weight at 5, 9, 13 and 20 months of age. Animals from the
previously reported in Chernoff et al. [10] but are included here for Con_Con condition were heavier than all other treatment groups
completeness (Fig. 1). Dams undernourished throughout pregnancy throughout the experiment. Signicantly lower body weight was seen in
(Treat) weighed signicantly less than their ad lib fed counterparts postnatally undernourished animals (Con_Treat and Treat_Treat)
(Con) the day after parturition (306.2 2.9 vs 217.1 1.7 g, p b 0.05). relative to animals in Con_Con and Treat_Con conditions at three of
Birth weights were lower in pups born to undernourished mothers these four ages (see Fig. 1B, p b 0.05).
(6.02 0.06 vs 6.25 0.06 g, p b 0.001). On PN7 prenatally undernour-
ished pups (Treat_Con) were equivalent in body weight to the Con_Con 3.2. Motor activity
group, and both postnatally undernourished groups (Con_Treat and
Treat_Treat) were lower in weight and similar to each other. This Motor activity was assessed in independent groups of rats at 3, 13, 18
pattern continued until weaning on PN22 (see Fig. 1A, p b 0.05). Adult and 21 months of age. Motor activity decreased with age in all treatment
groups (Fig. 2). For horizontal activity, the decrease due to age was
signicant [F(3,186)=16.95, pb 0.0001], from 3 to 21 months horizontal
activity decreased by approximately 40%. For vertical activity, the decrease
due to age was also signicant [F(3,186)=90.08, pb 0.001] with a greater
decrease of 85% in vertical activity from 3 to 21 months. There was no
statistically reliable effect of pre or postnatal treatment condition on
motor activity at any age, and no treatmentage interaction.

3.3. Trace fear training

Trace fear conditioning was examined in independent groups of rats

at 5, 12 and 18 months of age. A reduction in activity count was used as
the measure of learning. Activity counts decreased and remained below
baseline levels upon delivery of the rst footshock, and a further
reduction to less than 1015% of baseline was seen in response to the
second footshock in all treatment groups at all ages (data not shown).

3.4. Fear conditioning to context

Context learning was examined 24 h after training by returning

the animals to the same test chamber in which they had been shocked
on the previous day. Activity counts were recorded over a 5-min
period and no stimulus cues were delivered. Mean counts over 5 min
did not differ among treatment groups (data not shown). Learning is
evidenced by a signicant drop in activity from the 2-min baseline
period prior to shock delivery during training (Fig. 3A) to the rst
2 min on the context test day (Fig. 3B). No signicant differences were
detected in baseline activity counts during training as a function of
Age [F(2, 133) = 1.45, p N 0.23], but signicant Treatment [F(3, 133) =
15.16, p N 0.0001] and Age Treatment [F(6,133) = 5.16, p b 0.0001]
effects were detected. Inspection of Fig. 3A reveals that these statistical
differences were driven by a curiously low baseline activity counts in the
two postnatal treatment groups (Con_Treat and Treat_Treat) at
5 months of age relative to control (Con_Con) and prenatal (Treat_Con)
groups which were not evident at any subsequent age. A similar, but
non-signicant, pattern is seen in mean activity counts during context
testing, with lower counts evident in these two groups and limited to
the 5 month age group. The dotted lines in Fig. 3 represent the mean
activity level across all 4 treatment groups for the 5-month group. To
Fig. 1. Body weight pre- or postnatal undernutrition. Mean (SE) body weights in
preweanling and adult rats. (A) Signicant reductions in body weight were evident at PN7, permit an evaluation of context learning in the face of differing baseline
14 and 22 in animals undernourished in the postnatal period (Con_Treat and Treat_Treat) activity levels, the mean counts over the rst 2 min of context testing are
relative to controls (Con_Con) and animals undernourished in the prenatal period only expressed as a function of each animal's baseline activity counts during
(Treat_Con). (B) Mean (SE) weight in adult offspring throughout the lifespan. training (Fig. 3C). No differential effect of treatment on context learning
Postnatally undernourished animals (Con_Treat and Treat_Treat) weighed less than
Con_Con animals at 3 of the 5 timepoints selected for analysis. No differences in body
was evident, but clear differences as a function of age were revealed.
weight were detected between controls and the prenatally undernourished group Strong learning is evident in the younger animals as counts were
(Treat_Con). p b 0.05. reduced by more than 60% relative to those seen prior to shock delivery.
366 M.E. Gilbert et al. / Neurotoxicology and Teratology 32 (2010) 362372

Fig. 2. Motor activity is reduced in aging animals. A) Motor activity assessed in adult
male offspring at 3, 13, 18 and 21 months of age was not altered by pre- or postnatal
undernutrition, but mean ( SE) activity counts over a 30 min testing period for (A)
horizontal and (B) vertical (i.e. rearing) was signicantly decreased as a function of
age.

Older animals exhibit a learning decit as their reduction in activity was

signicantly less at 12 and 18 months relative to 5 months of age
(Fig. 3C). This conclusion is supported by the results of ANOVA in which
a signicant effect of Age [F(2,133) = 11.18, p b 0.0001] was detected in
the absence of any Treatment or Age Treatment effects (both
ps N 0.38). Mean contrast tests support these conclusions and revealed
signicant differences between ages, 5 vs 12 months and 5 vs
18 months, but no differences between 12 and 18 months (p b 0.05).
Higher activity accounts suggestive of impaired learning can be seen in
the prenatally undernourished group (Treat_Con) at 18 months of age
(Fig. 3B and C), but this difference did not reach statistical signicance.

3.5. Fear conditioning to cue

Fig. 3. Context fear learning is impaired in aging animals. (A) Mean (SE) activity counts
Conditioning to cue was evaluated after completion of context in the 2 min baseline period during training. Activity counts were signicantly lower in the
two postnatally undernourished groups. The dotted line represents the mean activity level
testing in a separate room under conditions that were distinct from
across treatment groups at 5 months. Baseline activity counts were similar to this mean at
training and context testing. The stimulus presentation protocol was 12 and 18 months. ANOVA revealed signicant main Age Treatment interaction that was
similar to that used during training but in the absence of shock and due to lower counts in these two groups at 5 months of age. (B) Mean (SE) activity
distractor stimuli. Mean activity counts in 30 s were calculated for counts in the rst 2 min of context testing. Increased activity counts were detected across
the 2-min baseline period, during CS presentation, and the interval all Age groups, regardless of Treatment. (C) Context learning expressed as percent of
training baseline. A signicant effect of Age in the absence of a Treatment or
corresponding to the trace interval between CS and shock presen-
Age Treatment interaction conrms impaired context learning in 12 and 18 month old
tation in training, and for an additional 30 s after completion of the animals relative to animals 4 months of age.
trace interval. Baseline activity counts were increased with increases
in age (mean counts/30 s = 58, 68, 83 for the 5, 12 and 18 month ages in the postnatally undernourished group at 5 months of age were not
respectively, p b 0.003) but no treatment-related differences were seen in baseline counts that preceded cue testing. Cue learning is
detected (p N 0.88). The reduction in baseline counts seen in training presented in Fig. 4A, and activity counts during the Cue, Trace, and
 M.E. Gilbert et al. / Neurotoxicology and Teratology 32 (2010) 362372 367

Post-Trace intervals are expressed as a function of mean baseline suggestive of an impairment in cue learning at 5 (Fig. 4A, p N 0.10) and
counts during cue testing. All animals regardless of treatment or age 18 (Fig. 4C, p N 0.08) months of age, this was not supported by statistical
showed a decline in activity in response to CS presentation in a novel analysis. No differences in activity counts were detected as a function of
context indicating they had learned the cue and its association with Age (p N 0.15) or Treatment (p N 0.17), or Age Treatment (p N 0.30).
footshock. These conclusions are supported by results from ANOVA Together these data suggest equivalent acquisition of the trace fear
where a signicant effect of State is seen [F(2, 240) = 72.25, learning to cue in animals from all treatment conditions and across the
p b 0.0001]. Although higher activity counts in the prenatally under- lifespan. Neither was context learning impacted by pre- or postnatal
nourished treatment group (Treat_Con) during the trace interval are undernutrition, but context learning was impaired as a function of age.
This decit is not attributable to decreases in activity levels in older
animals as baseline activity counts during training and cue testing
revealed increases rather than decreases in activity in the older animals.
Neither is it likely that decreases in sensitivity to shock underlie this
decit as reductions in activity counts during training were similar
across treatment groups and age, and no impairments were detected in
cue learning.

3.6. Morris water maze learning 1314 months

Unfortunately, we were unable to evaluate water maze learning in

rats younger than 13 months in the present study. However we have
included, for comparison, water maze performance in a group of nave
rats of the same strain run under similar conditions. At 13 months of age,
very limited evidence of learning was apparent over 12 days of training
in any treatment groups, so training was extended for 7 days. A cue trial
was conducted at the end of day 12 and once again on day 20. Latency
data are presented for the rst 9 days of training in Fig. 5A to allow for
direct comparison to performance in 20-month-old animals (Fig. 5B).
Despite extended training in 13-month animals, search latencies were
still signicantly longer than under cue conditions. As no treatment-
related differences were detected (p N 0.05) data were collapsed across
treatment group and contrasted to performance of younger animals
(Fig. 5C). Clear age-dependent decits are evident in the 13-month
group relative to nave controls at 3-months of age. Comparable
latencies are seen between groups on the cued version of the task. Swim
speed was calculated on the rst probe trial and did not differ between 3
and 13 months of age (19.520.4 cm/s).
Probe data presented in Fig. 6A are consistent with a lack of any
apparent learning until the 12th training day. Time spent in the correct
quadrant remained below 30% (chance performance= 25%) until the
12th training day and indicative of minimal learning. Although a slightly
lower search time was evident in the Treat_Con group, it was not
statistically signicant (p N 0.31) and no treatment-related differences in
probe trial performance were evident [F(3, 37) = 0.34, p N 0.79]. This
pattern in 13-month-old animals is quite distinct from that observed at
3 months of age (Fig. 6B) where a mean of N30% time in the correct
quadrant was attained by day 6 of training and maximal levels of N40%
were evident by day 19.

3.7. Morris water maze learning 1820 months

Given the lack of mastery of this task in animals at 13 months of age,

the procedure was modied for older animals. Maze learning was
conducted using a smaller diameter maze in a smaller room with more
salient visual cues, and only one probe trial at the end of testing (see
Methods). A recording system that permitted analysis of pathlength in
addition to trial latency was utilized to determine if motor function was
compromised in older animals. Under these conditions more robust
learning was evident in both latency and pathlength analyses. Latency
data for 20-month-old animals (Fig. 5B) reveal signicant reductions in
latency across trials (p b 0.0001), but no treatment-related differences
were seen in rate of acquisition (p N 0.25). Similar results are derived
from an analysis of pathlength (data not shown). A single probe trial was
assessed on the nal day of testing and all groups showed evidence of
Fig. 4. Cue fear learning is not impacted by age or pre- or postnatal undernutrition. (A)
Mean percent of baseline activity ( SE) activity in a novel chamber. Activity was
learning (Fig. 6C). A slight reduction in the amount of time spent in the
markedly suppressed during the 30-s Trace interval and the immediate 30-s Post-Trace correct quadrant in group Treat_Con is reminiscent of that seen in this
period. No effects of Age, Treatment, or Age Treatment interaction were detected. treatment group at 13 months of age (Fig. 6A), but also failed to reach
368 M.E. Gilbert et al. / Neurotoxicology and Teratology 32 (2010) 362372

statistical signicance (p N 0.10). Data from the probe trial are consistent
with the acquisition ndings revealing strong evidence of learning and
no treatment-related impairments. Increased latencies under cued trial
conditions may reect disruption of performance due to a change in task
contingency as the platform was visible above the water surface and was
moved to the opposite quadrant from the 9 previous testing days. Mean
swim speed ranged from 16.5 cm/s to 20.0 cm/s and did not differ
among the treatment groups [F(3,59) = 2.01, p N 0.12].
Probe trial data permit an evaluation of learning ability as a function
of age independent of differences in testing protocols and conditions. As
no treatment-related differences were observed in either age group,
search time on acquisition day 12 (3 and 13 month group, trial 23) and
acquisition day 10 (20 month group, trial 19) data were collapsed across
treatment condition. Analysis of percent time in correct quadrant
indicates poorer learning at 13 months relative to 3 months of age when
testing occurred in the same maze (Fig. 6D). Signicant augmentation in
learning was evident in older animals run in the smaller maze at
20 months of age [F(2,141) = 16.22, p b 0.0001]. Mean contrast tests
revealed signicant decits in quadrant dwell time on this probe trial in
13-month relative to 3-month animals, and relatively enhanced dwell
time in 20-month animals run in the smaller version of the maze
(p b 0.05). A similar pattern is also seen when these data are described as
the number of animals in each group spending 30% of search time in
the correct quadrant (60% vs 40% vs 85%). These ndings suggest that
training conditions utilized at 3 and 13 months were more difcult to
master than those incorporated at 20 months of age, that 13-month old
animals were impaired relative to 3 months of age, and that potential for
age-induced impairment of visual or motor function cannot account for
age-dependent decits.

4. Discussion

A cross-over experimental design was used to examine the impact

of pre- or postnatal undernutrition on the development of obesity and
the behavior across the lifespan of male Sprague Dawley rats. No
evidence of pre- or postnatal undernutrition-induced obesity or
neurological impairment was obtained. Body weights were monitored
from birth until 21 months of age. Motor activity was assessed as it is a
widely used test of neurotoxicity and aging. Hippocampal-dependent
learning (spatial learning and trace fear conditioning) was assessed
because of the proposed role of glucocorticoids in fetal programming
[59], the high density of glucocorticoid binding sites in this brain
region [8], and the vulnerability of the hippocampus to the aging
process [40]. Age-related alterations in motor activity, spatial
learning, and context fear conditioning were observed, but no
differential impact of nutritional history was uncovered.

4.1. Nutritional programming and development of obesity?

Undernutrition reduced maternal body weight gain during

pregnancy, and pups born to undernourished dams were smaller at
birth. By PN7, pups fostered to large litters weighed less than those
fostered to smaller litters, regardless of their in utero nutritional
history, a pattern that was maintained until weaning. As adults, the
animals that experienced both pre- and postnatal undernourished
Fig. 5. Spatial learning acquisition at 3, 13 and 20 months. (A) 13 Months mean
( SE) latency to reach the submerged platform in the large maze. Only 9 of the environments (Treat_Treat) weighed less than controls and this
19 days of training are depicted to facilitate comparison with performance of differential was maintained until senescence. Body weights for
20 month old animals (B). 20 Months mean ( SE) latency to reach the submerged adult animals from the pre- and postnatally switched conditions
platform in the smaller maze. No treatment-related differences in acquisition were (Con_Treat and Treat_Con) were intermediate but generally not
observed but signicantly greater learning is evident in 20 relative to 13 month old
animals (see methods for procedural differences associated with aging). (C). Mean
statistically different from controls. As such, under the conditions of
( SE) latency in 13 month (collapsed across Treatment) and nave 3-month old SD the present study, no evidence of nutritional programming of
rats assessed in the same maze used for the 20-month-old rats and similar testing obesity was obtained. These results stand in contrast to previous
conditions. reports of excessive weight gain in animals exposed in utero to an
impoverished nutritional environment and subsequently reared under
ad lib or hypercaloric feeding conditions [13,14,56].
 M.E. Gilbert et al. / Neurotoxicology and Teratology 32 (2010) 362372 369

Fig. 6. Probe trials in spatial learning at 3, 13 and 20 months. (A) Mean ( SE) percent time in correct quadrant on days 3, 6, 9, 12 and 19 of acquisition. No treatment-related
differences in probe trial performance were detected. Signicant evidence of learning was not detected until day 19 of training. (B) In contrast to performance at 13 months of age,
mean ( SE) percent time in correct quadrant on similar test days in nave 3 month old animals show signicant evidence of learning by the 2nd probe trial on day 6 of acquisition.
(C) A single probe trial assessed on the nal day of acquisition in 20-month old animals assessed on a smaller version of the maze shows signicant evidence of learning in all
treatment groups. (D) Probe trial performance collapsing across treatment groups permits direct comparison of learning as a function of age. Animals tested at 13 months of age
spent less time and fewer of them reached criterion performance (30% time in the correct quadrant) compared to animals tested under similar conditions at 3 months of age. Using a
smaller maze and presumably more salient cues, robust learning was readily induced in older animals and no evidence of treatment-related impairments was observed.

The reasons for these disparate ndings may involve the con- 4.2. The impact of undernutrition on brain development
stitution of the postweaning diet (see Vickers et al., 2000 [56]) and/or
the degree of undernutrition. Maternal undernutrition induced in the The present ndings also appear inconsistent with previous reports
present study was less than that used by Vickers et al. (2000) (i.e., 50% in the literature on brain dysfunction associated with developmental
vs 70% restriction), but equal to that of studies of Desai (2005, 2007) undernutrition. Animal models indicate that nutritional deprivation
who also reported excessive weight gain. In the same strain of rat and during fetal development leads to permanent reductions in brain size
an identical experimental design, Desai et al. (2005, 2007) reported and neurochemistry [15,35,48]; disruptions of learning [36,50], and
preweaning body weight proles similar to those observed here. impairments in synaptic plasticity [2,23,51]. In the majority of these
Consistent with our ndings the equivalent of our Treat_Treat studies, protein content of the diet is reduced from an adequate amount
condition produced signicantly lighter animals across the rst (25% casein) to an isocaloric but protein decient one (68% casein)
9 months of age, the latest time sampled by Desai and colleagues throughout the gestational period (see review by [36]). Restricting
[13,14]. Although an approach toward control body weights is evident nutrition in this manner and limiting it to the prenatal period does not
at 9 months, the present data indicate that the body weight decit reduce body weight in the dams, or brain or body weight in the
remains as signicant differences between these two groups were still offspring, yet is accompanied by signicant cognitive impairments.
evident at 20 months of age. In sharp contrast to our ndings of In contrast, undernutrition induced by severe prenatal caloric
similar or slightly lower body weights throughout adulthood in restriction (N70% of ad lib feeding) produces decits in body weight
Treat_Con animals, Desai et al. [13] report excessive weight gain and gain in the dams and lower birth weights in the pups, a pattern similar
obesity in animals nutritionally deprived in utero and placed under ad to that observed in the present study. This paradigm also reduces
lib feeding conditions at birth. However, the consistency in body hippocampal volume in adult offspring, but surprisingly, to a lesser
weight patterns reported here with previous ndings of Smart et al. degree than protein malnutrition despite the more severe neonatal
[50] using a similar cross-over design is worthy of note. phenotype [43]. Spatial learning was compared between these two
370 M.E. Gilbert et al. / Neurotoxicology and Teratology 32 (2010) 362372

undernutrition models in mice on a radial arm maze. Impairments 4.3.3. Trace fear conditioning
were reported in both caloric restriction and protein malnutrition Fear conditioning also failed to show impairments due to nutritional
models, but the pattern of effects was distinct and the impairment history in young, middle-aged, or aged. Moreover, while there was no
less severe with caloric restriction [43]. In the present study, no evidence of an age-dependent impairment in the conditioning to cue,
treatment-induced disruption was revealed in the performance of there was a clear age-dependent decline in context conditioning. Both
two hippocampal-dependent tasks, Morris water maze spatial 12 and 18-month old animals were impaired in context conditioning
learning, and trace fear conditioning. These tasks differ in several relative to their 5-month counterparts. This decit cannot be attributed
ways from the radial arm maze task, and could be less sensitive to to differential sensitivity to shock as all age groups showed a similar
disruption. Additionally, the degree of prenatal undernutrition suppression of ongoing activity with shock presentation during the
induced in the present series of studies was less (50% vs 70% food training phase (see also [37]). Additionally, no age-dependent declines
restriction). Postnatal undernutrition induced by varying litter size is were evident in cue learning indicating that visual and auditory cues
also less severe than maternal food restriction on a number of brain were equally salient to animals of all ages. Neither can the decrement in
indices despite equivalent reductions in somatic growth [53]. In context learning be attributed to an overall reduction in activity levels.
studies on developmental toxicity carried out for environmental Although locomotor activity was reduced with age in 30-min sessions
regulatory purposes, body weight loss in dams and pups is often (Fig. 3), the activity counts during the brief baseline period for fear
reported [11]. The current data suggest that reduced food intake and/or conditioning were not different among age groups. Moyer and Brown
resultant body weight loss of either the dam or offspring, as is often [37] also found age-related decits in context conditioning but no
associated with developmental exposure to environmental chemicals change in baseline activity determined for one minute before training. It
does not necessarily induce neurotoxicity as assessed by these standard appears that extremely short test periods cannot differentiate age-
behavioral testing paradigms. One implication of these ndings is that related changes in motor activity (see also [30]).
cognitive decits observed in response to exposure to a toxicant that are In the present study, a selective impairment of context relative to cue
also coupled with lower birth weight and reduced postnatal weight gain learning in older animals suggests a particular vulnerability of
should not be necessarily attributed to nutritional deciencies, but hippocampal function to the aging process. The trace fear paradigm is
rather, may reect a direct effect of the toxicant on brain development. distinct from delay fear conditioning as it introduces a brief 30-s interval
between presentation of the initially neutral CS and the aversive US. In
4.3. Aging: a risk factor for cognitive decline? delay conditioning, where the CS and US overlap, learning is dependent
upon amygdala function [46], but by imposing a delay between CS and
It was hypothesized that cognitive impairment associated with a US, the hippocampus is engaged in addition to the amygdala [32].
moderate level of nutritional deciency may be exacerbated and Animals with hippocampal damage cannot perform either the cued or
more readily detected with the onset of cognitive decline that the context version of the trace fear conditioning task [32]. Aged animals
normally accompanies aging. As such different groups of animals are impaired on trace but not delay conditioning [33,37]. The present
were assessed at different ages. Pre- or postnatal nutritional history data suggest that contextual learning may be more sensitive to
did not differentially impact the behavior of aged animals using disruption than trace cue conditioning in aged animals. This conclusion
standard tests of motor activity and learning. However, age- is supported by pharmacological studies using the muscarinic antago-
dependent declines in motor activity and learning were evident as nist, scopolamine, which impairs acquisition of fear conditioning to
described below. context more readily that cued fear conditioning [1]. Loss of function in
forebrain cholinergic pathways has been documented in aged rat as well
as monkey brain [16,38] and may contribute to the selective
4.3.1. Motor activity vulnerability to context learning in the present study.
Age-related decreases were obtained in motor activity. Older In summary we have investigated the behavioral effects
animals displayed less horizontal and vertical activity during a 30-min associated with a moderate level of undernutrition during develop-
test period than did younger rats. These ndings are consistent with ment. The resulting decits in body weight and growth were
previous reports in the literature (e.g., [9,22]; see also [24]). Aging also designed to mimic those obtained frequently in standardized
had a bigger impact on vertical activity (rearing) than on horizontal developmental toxicity testing studies in the regulatory arena. We
activity (locomotion). The larger decrease in vertical activity with age found no evidence that this degree of undernutrition altered motor
may have been due to increasing body weight, decreased hindlimb activity or learning carried out over the life span. Moreover, we did
strength, or some combination of these and other factors (see, for not nd evidence of nutritional programming that would lead to
example, [54]). Pre- or postnatal treatment condition, however, had no obesity or a negative impact on cognition in young or aged animals.
effect on either horizontal or vertical activity, nor did treatment alter the However, we did obtain age-dependent impairments in motor
age-related decrease in motor activity. activity and in some domains of cognitive function. Together, these
ndings indicate that a mismatch between the fetal and neonatal
4.3.2. Spatial learning nutritional environments does not universally lead to obesity or
Age-dependent impairments in spatial learning in the Morris neurological dysfunction in adulthood.
water maze have been widely reported [9,17]. In the present study,
decits were evident at 13 months relative to the performance of rats
Conict of interest statement
at 3 months of age. These decits could not be attributed to motor or
sensory dysfunction as swim speed and cued test performance were
The authors declare that there are no conicts of interest.
similar for the two age groups. Probe trials interspersed throughout
the training period and the incidence of animals attaining the
learning criterion also showed clear differences between 3 and Acknowledgements
13 months of age. Aged animals (1822 months of age), however,
exhibited clear evidence of learning when assessed in a smaller maze This document has been subjected to review by the National Health
with more salient visual cues. These latter ndings indicate that and Environmental Effects Research Laboratory and approved for
motor and sensory competence were maintained at this advanced publication. Approval does not signify that the contents reect the
age and suggest that task difculty contributed to the learning views of the Agency, nor does mention of trade names or commercial
decits obtained in rats at 13 months of age. products constitute endorsement or recommendation for use.
 M.E. Gilbert et al. / Neurotoxicology and Teratology 32 (2010) 362372 371

The authors wish to thank Drs. Christopher Lau, Christopher Newland [27] S.C. Langley-Evans, Nutritional programming of disease: unravelling the mech-
anism, J. Anat. 15 (2009) 3651.
and Diane Miller for critical reviews of an earlier version of this
[28] C.M. Law, A.W. Shiell, Is blood pressure inversely related to birth weight? The
manuscript. The assistance of Michael Gage, Pam Phillips, Kathy McDaniel, strength of evidence from a systematic review of the literature, J. Hypertens. 14
Willard Anderson, Jackie Farmer, Kimberly Jarema, and Donna Kronstadt (1996) 935941.
for various phases of this project is gratefully acknowledged. [29] H.O. Lithell, P.M. McKeigue, L. Berglund, R. Mohsen, Lithell, D.A. Leon, Relation of
size at birth to non-insulin dependant diabetes and insulin concentrations in men
aged 5060 years, BMJ 312 (1996) 406410.
References [30] J.F. Marshall, Sensorimotor disturbances in the aging rodent, J. Gerontol. 37 (1982)
548554.
[31] C.N. Martyn, D.J. Barker, Reduced fetal growth increases risk cardiovascular disease,
[1] S.G. Anagnostaras, S. Maren, J.R. Sage, S. Goodrich, M.S. Fanselow, Scopolamine and Health Rep. 6 (1994) 4553.
Pavlovian fear conditioning in rats: dose-effect analysis, Neuropsychopharmacology [32] M.D. McEchron, H. Bouwmeester, W. Tseng, C. Weiss, J.F. Disterhoft, Hippocampect-
21 (6) (Dec 1999) 731744. omy disrupts auditory trace fear conditioning and contextual fear conditioning in the
[2] K.B. Austin, J. Bronzino, P.J. Morgane, Prenatal protein undernutrition affects rat, Hippocampus 8 (6) (1998) 638646.
synaptic potentiation in dentate gyrus of rats in adulthood, Brain Res. 394 (1986)
[33] M.D. McEchron, A.Y. Cheng, M.R. Gilmartin, Trace fear conditioning is reduced in
267273.
the aging rat, Neurobiol. Learn. Mem. 82 (2) (Sep 2004) 7176.
[3] D.J. Barker, Intrauterine programming of coronary heart disease and stroke, Acta
Paediatr. Suppl. 423 (1997) 178182. [34] M.K. McGuire, A.W. Littleton, K.J. Schulze, K.M. Rasmussen, Pre- and postweaning
[4] D.J. Barker, The long-term outcome of retarded fetal growth, Schweiz. Med. food restrictions interact to determine reproductive success and milk volume in
Wochenschr. 129 (1999) 189196. rats, J. Nutr. 125 (9) (Sep 1995) 24002406.
[5] D.J. Barker, The origins of the developmental origins theory, J. Intern. Med. 261 (5) [35] D.J. Mokler, O.I. Torres, J.R. Galler, P.J. Morgane, Stress-induced changes in
(May 2007) 412417. extracellular dopamine and serotonin in the medial prefrontal cortex and dorsal
[6] D.J. Barker, P.M. Clark, Fetal undernutrion and disease in later life, Rev. Reprod. 2 hippocampus of prenatally malnourished rats, Brain Res. 1148 (2007) 226233.
(1997) 105112. [36] P.J. Morgane, D.J. Mokler, J.R. Galler, Effects of prenatal protein malnutrition on the
[7] D.J. Barker, C. Osmond, S.J. Simmonds, G.A. Wield, The relation of small head hippocampal formation, Neurosci. Biobehav. Rev. 26 (4) (Jun 2002) 471483.
circumference and thinness at birth to death from cardiovascular disease in adult [37] J.R. Moyer Jr, T.H. Brown, Impaired trace and contextual fear conditioning in aged
life, BMJ 306 (1993) 422426. rats, Behav. Neurosci. 120 (3) (Jun 2006) 612624.
[8] C.E. Bertram, M.A. Hanson, Prenatal programming of postnatal endocrine responses [38] A.H. Nagahara, T. Bernot, R. Moseanko, L. Brignolo, A. Blesch, J.M. Conner, A. Ramirez,
by glucocorticoids, Reproduction 124 (4) (Oct 2002) 459467. M. Gasmi, M.H. Tuszynski, Long-term reversal of cholinergic neuronal decline in aged
[9] C.S. Carter, C. Leeuwenburgh, M. Daniels, T.C. Foster, Inuence of calorie restriction on non-human primates by lentiviral NGF gene delivery, Exp. Neurol. 215 (2009)
measures of age-related cognitive decline: role of increased physical activity, J. Gerontol. 153159.
A Biol. Sci. Med. Sci. 64 (2009) 850859. [39] C.R. Neal Jr, G. Weidemann, M. Kabbaj, D.M. Vzquez, Effect of neonatal
[10] N. Chernoff, M.I. Gage, T.E. Stoker, R. Cooper, M.E. Gilbert, E.H. Rogers, Effects of dexamethasone exposure on growth and neurological development in the adult
maternal undernutrition in the rat: intergenerational pup growth and viability, age at rat, Am. J. Physiol. Regul. Integr. Comp. Physiol. 287 (2) (Aug 2004) R375R385.
reproductive maturity, and onset of reproductive senescence, Reprod. Toxicol. 28 [40] O.G. Nilsson, F.H. Gage, Anticholinergic sensitivity in the aging rat septohippocampal
(2009) 489494. system as assessed in a spatial memory task, Neurobiol. Aging 14 (5) (SepOct 1993)
[11] N. Chernoff, E.H. Rogers, M.I. Gage, B.M. Francis, The relationship of maternal and 487497.
fetal toxicity in developmental toxicology bioassays with notes on the biological
[41] M.G. Opler, E.S. Susser, Fetal environment and schizophrenia, Environ. Health Perspect.
signicance of the no observed adverse effects level, Reprod. Toxicol. 25 (2008)
113 (9) (Sep 2005) 12391242.
192202.
[12] A. da Silva Hernandes, A.L. Franolin-Silva, C.T. Valadares, M.T. Fukuda, S.S. Almeida, [42] C. Osmond, D.J. Barker, Fetal, infant, and childhood growth are predictors of coronary
Effects of different malnutrition techniques on the behavior of rats tested in the heart disease, diabetes, and hypertension in adult men and women, Environ. Health
elevated T-maze, Behav. Brain Res. 162 (2) (Jul 30 2005) 240245. Perspect. 108 (Suppl 3) (Jun 2000) 545553.
[13] M. Desai, D. Gayle, J. Babu, M.G. Ross, Programmed obesity in intrauterine growth- [43] S.C. Ranade, A. Rose, M. Rao, J. Gallego, P. Gressens, S. Mani, Different types of
restricted newborns: modulation by newborn nutrition, Am. J. Physiol. Regul. Integr. nutritional deciencies affect different domains of spatial memory function checked
Comp. Physiol. 288 (1) (Jan 2005) R91R96. in a radial arm maze, Neuroscience 152 (4) (Apr 9 2008) 859866.
[14] M. Desai, D. Gayle, G. Han, M.G. Ross, Programmed hyperphagia due to reduced [44] G.-P. Ravelli, Z.A. Stein, M.W. Susser, Obesity in young men after famine exposure
anorexigenic mechanisms in intrauterine growth-restricted offspring, Reprod. Sci. ion utero and early infancy, New England J. Med. 295 (1976) 349353.
14 (4) (May 2007) 329337. [45] F. Remmers, M.F. Schreuder, R.J. Gemke, H.A. Delemarre-van de Waal, Energy intake
[15] J. Dobbing, J. Sands, Vulnerability of developing brain. IX. The effect of nutritional and resting energy expenditure in adult male rats after early postnatal food
growth retardation on the timing of the brain growth-spurt, Biol. Neonate 19 (4) restriction, Br. J. Nutr. 99 (5) (May 2008) 11491156.
(1971) 363378. [46] S.M. Rodrigues, G.E. Schafe, J.E. LeDoux, Molecular mechanisms underlying emotional
[16] W. Fischer, K.S. Chen, F.H. Gage, A. Bjrklund, Progressive decline in spatial learning learning and memory in lateral amygdala, Neuron. 44 (2004) 7591.
and integrity of forebrain cholinergic neurons in rats during aging, Neurobiol. Aging [47] T.J. Roseboom, J.H. van der Meulen, A.C. Ravelli, C. Osmond, D.J. Barker, O.P. Bleker,
13 (1) (JanFeb 1992) 923. Effects of prenatal exposure to the Dutch famine on adult disease in later life: an
[17] M. Gallagher, R. Burwell, M. Burchinal, Severity of spatial learning impairment in overview, Mol. Cell. Endocrinol. 185 (12) (Dec 20 2001) 9398.
aging: development of a learning index for performance in the Morris water maze, [48] L.N. Rotta, D.N. Leszczinski, A.M. Brusque, P. Pereira, L.F. Brum, C.W. Nogueira, M.E.
Behav. Neurosci. 107 (4) (Aug 1993) 618626. Frizzo, M.L. Perry, D.O. Souza, Effects of undernutrition on glutamatergic parameters
[18] P.D. Gluckman, M.A. Hanson, Maternal constraint of fetal growth and its consequences, in the cerebral cortex of young rats, Physiol. Behav. 94 (4) (Jul 5 2008) 580585.
Semin. Fetal Neonatal Med. 9 (5) (Oct 2004) 419425.
[49] B.A. Schwetz, M.W. Harris, Developmental toxicology: status of the eld and
[19] P.D. Gluckman, M.A. Hanson, Developmental and epigenetic pathways to obesity: an
contribution of the National Toxicology Program, Environ. Health Perspect. 100
evolutionary-developmental perspective, Int. J. Obes. (Lond.) 32 (Suppl 7) (Dec 2008)
(1993) 269282.
S62S71.
[20] P.D. Gluckman, K.A. Lillycrop, M.H. Vickers, A.B. Pleasants, E.S. Phillips, A.S. Beedle, [50] J.L. Smart, J. Dobbing, B.P. Adlard, A. Lynch, J. Sands, Vulnerability of developing
G.C. Burdge, M.A. Hanson, Metabolic plasticity during mammalian development is brain: relative effects of growth restriction during the fetal and suckling periods
directionally dependent on early nutritional status, Proc. Natl. Acad. Sci. U.S.A. 104 on behavior and brain composition of adult rats, J. Nutr. 103 (9) (Sep 1973)
(31) (Jul 31 2007) 1279612800. 13271338.
[21] E.H. Hamilton, A.M. Minino, J.A. Martin, K.D. Kochanek, D.M. Strobino, B. Guyer, [51] R. Soto-Moyano, L. Valladares, W. Sierralta, H. Prez, M. Mondaca, V. Fernndez,
Annual summary of vital statistics: 2005, Pediatrics 119 (2007) 345360. H. Burgos, A. Hernndez, Mild prenatal protein malnutrition increases alpha2C-
[22] M.A. Herbert, G.A. Gerhardt, Normal and drug-induced locomotor behavior in aging: adrenoceptor density in the cerebral cortex during postnatal life and impairs
comparison to evoked DA release and tissue content in Fischer 344 rats, Brain Res. neocortical long-term potentiation and visuo-spatial performance in rats,
797 (1998) 4254. J. Neurochem. 93 (5) (Jun 2005) 10991109.
[23] A. Hernndez, H. Burgos, M. Mondaca, R. Barra, H. Nez, H. Prez, R. Soto-Moyano, [52] C. Thompson, H. Syddall, I. Rodin, C. Osmond, D.J. Barker, Birth weight and the risk
W. Sierralta, V. Fernndez, R. Olivares, L. Valladares, Effect of prenatal protein of depressive disorder in late life, Br. J. Psychiatry 179 (Nov 2001) 450455.
malnutrition on long-term potentiation and BDNF protein expression in the rat [53] J. Tonkiss, C.A. Cohen, S.B. Sparber, Different methods for producing neonatal
entorhinal cortex after neocortical and hippocampal tetanization, Neural Plast. 2008 undernutrition in rats cause different brain changes in the face of equivalent somatic
(2008) 646919. growth parameters, Dev. Neurosci. 10 (3) (1988) 141151.
[24] D. Ingram, Age-related decline in physical activity: generalization to non-humans, [54] J.C.L. Tou, C.E. Wade, Determinants affecting physical activity levels in animal
Med. Sci. Sports Exerc. 32 (2000) 16231629. models, Exp. Biol. Med. 227 (2002) 587600.
[25] G.V. Iyengar, P.P. Nair, Global outlook on nutrition and the environment: meeting the [55] E. Velkoska, T.J. Cole, R.G. Dean, L.M. Burrell, M.J. Morris, Early undernutrition leads
challenges of the next millennium, Sci. Total Environ. 249 (2000) 331346. to long-lasting reductions in body weight and adiposity whereas increased intake
[26] J.I. Koenig, G.I. Elmer, P.D. Shepard, P.R. Lee, C. Mayo, B. Joy, E. Hercher, D.L. Brady, increases cardiac brosis in male rats, J. Nutr. 138 (9) (Sep 2008) 16221627.
Prenatal exposure to a repeated variable stress paradigm elicits behavioral and [56] M.H. Vickers, B.H. Breier, W.S. Cuteld, P.L. Hofman, P.D. Gluckman, Fetal origins of
neuroendocrinological changes in the adult offspring: potential relevance to hyperphagia, obesity, and hypertension and postnatal amplication by hyperca-
schizophrenia, Behav. Brain Res. 156 (2) (Jan 30 2005) 251261. loric nutrition, Am. J. Physiol. Endocrinol. Metab. 279 (1) (Jul 2000) E83E87.
372 M.E. Gilbert et al. / Neurotoxicology and Teratology 32 (2010) 362372

[57] C.V. Vorhees, M.T. Williams, Morris water maze: procedures for assessing spatial [59] L.A. Welberg, J.R. Seckl, Prenatal stress, glucocorticoids and the programming of
and related forms of learning and memory, Nat. Protoc. 1 (2) (2006) 848858. the brain, J. Neuroendocrinol. 13 (2001) 113128.
[58] I.C. Weaver, N. Cervoni, F.A. Champagne, A.C. D'Alessio, S. Sharma, J.R. Seckl, S. [60] T.Y. Zhang, C. Parent, I. Weaver, M.J. Meaney, Maternal programming of individual
Dymov, M. Szyf, M.J. Meaney, Epigenetic programming by maternal behavior, differences in defensive responses in the rat, Ann. N. Y. Acad. Sci. 1032 (Dec 2004)
Nat. Neurosci. 7 (8) (Aug 2004) 847854. 85103.