The impact of voluntary exercise on mental health in rodents: A neuroplasticity perspective

Article (PDF Available)inBehavioural Brain Research 192(1):42-60 · October 2008with105 Reads
DOI: 10.1016/j.bbr.2008.03.014 · Source: PubMed
There is growing interest in the effects of voluntary wheel running activity on brain and behaviour in laboratory rodents and their implications to humans. Here, the major findings to date on the impact of exercise on mental health and diseases as well as the possible underlying neurobiological mechanisms are summarised. Several critical modulating factors on the neurobehavioural effects of wheel running exercise are emphasized and discussed--including the amount of wheel running, sex and strain/species differences. We also reported the outcome of an empirical investigation of the impact of wheel running exercise on the expression of both cognitive and non-cognitive phenotypes in a triple (3 x Tg-AD) transgenic mouse model for Alzheimer's disease (AD). Clear sex- and paradigm-specific effects of exercise on the genetically determined phenotypes are illustrated, including the efficacy of wheel running activity in attenuating the sex-specific cognitive deficits. It is concluded that the wheel running paradigm represents a unique environmental manipulation for the investigation of neurobehavioural plasticity in terms of gene-environment interactions relevant to the pathogenesis and therapies of certain neuropsychiatric conditions.
Behavioural Brain Research 192 (2008) 42–60
Contents lists available at ScienceDirect
Behavioural Brain Research
journal homepage:
Research report
The impact of voluntary exercise on mental health in rodents:
A neuroplasticity perspective
Susanna Pietropaolo
, Yan Sun
, Ruixi Li
, Corinne Brana
, Joram Feldon
, Benjamin K. Yee
Laboratory of Behavioural Neurobiology, ETH Zurich, Schorenstrasse 16, CH-8603 Schwerzenbach, Switzerland
Department of Anatomy, Histology and Embryology, Shanghai Medical College, Fudan University, Shanghai 200032, PR China
article info
Article history:
Received 20 November 2007
Received in revised form 6 March 2008
Accepted 13 March 2008
Available online 21 March 2008
Sex differences
Alzheimer’s disease
There i s growing interest in the effects of voluntary wheel running activity on brain and behaviour in
laboratory rodents and their implications to humans. Here, the major findings to date on the impact of
exercise on mental health and diseases as well as the possible underlying neurobiological mechanisms
are summarised. Several critical modulating factors on the neurobehavioural effects of wheel running
exercise are emphasized and discussed including the amount of wheel running, sex and strain/species
differences. We also reported the outcome of an empirical investigation of the impact of wheel running
exercise on the expression of both cognitiveand non-cognitive phenotypes in a triple (3×Tg-AD)transgenic
mouse model for Alzheimer’s disease (AD). Clear sex- and paradigm-specific effects of exercise on the
genetically determined phenotypes are illustrated, including the efficacy of wheel running activity in
attenuating the sex-specific cognitive deficits. It is concluded that the wheel running paradigm represents
a unique environmental manipulation for the investigation of neurobehavioural plasticity in terms of
gene-environment interactions relevant to the pathogenesis and therapies of certain neuropsychiatric
© 2008 Elsevier B.V. All rights reserved.
1. Introduction
1.1. Animals models and the investigation of behavioural
A broad definition of “behavioural plasticity” was formulated by
William James (1890) over a century ago as the ability of human
behaviour to undergo meaningful changes (see [31]). This early
concept of plasticity, which initially focuse d on human psychol-
ogy, has been developed to refer to the capacity of a system to
achieve new functions by transforming its constituent elements
and/or its internal connectivity network according to environmen-
tal constraints [150]. A number of procedures have been developed
for the experimental investigation of environmental influences on
neurobehavioural plasticity in laboratory animals [159]. Amongst
these, manipulations of the housing conditions and of the availabil-
ity of physical exercise are known to exert multiple effects on brain
and behaviour [41,169,205].
1.1.1. Enriched and impoverished environments
The enriched rearing procedure was experimentally defined
and validated during the second half of the last century by
Corresponding author. Tel.: +41 44 655 74 17; fax: +41 44 655 72 03.
E-mail address: (B.K. Yee).
Rosenzweig and colleagues. Environmental enrichment is typically
implemented in rats or mice by housing them in groups within
large cages typically equipped with toys, tunnels, running wheels
and occasionally additional nesting materials [162]. In compari-
son to con-specifics reared in standard cages, rodents reared under
enriched environment exhibited enhanced cognitive abilities in
both spatial and non-spatial memory tests [119,161] and enhanced
hippocampal neurogenesis [17,195]. In contrast, rearing in social
isolation can induce the emergence of psychotic-like symptoms
[68,202], signs of cognitive dysfunction [49,125] and reduced hip-
pocampal neurogenesis [49,125,187].
Notably, the brain and behavioural effects of environmental
enrichment as well as of social isolation are characterized by a
common critical period confined largely to the juvenile phase
of neurodevelopment [162]. Hence, the majority of the existing
studies have applied the isolation or enriched rearing procedures
immediately after weaning.
1.1.2. Physical exercise
The brain and behavioural effects of physical exercise in rodents
have been mainly investigated by employing two different experi-
mental paradigms: (i) exposing animals to daily or weekly sessions
of forced running on a treadmill and (ii) allowing free access to
a running wheel. Our review will focus exclusively on the effects
and characteristics of voluntary wheel running in rats and mice.
0166-4328/$ see front matter © 2008 Elsevier B.V. All rights reserved.
S. Pietropaolo et al. / Behavioural Brain Research 192 (2008) 42–60 43
Although beneficial effects of forced exercise have been describe d
in rodents especially on motor functions [108], treadmill run-
ning may induce physiological symptoms of chronic stress, such
as adrenal hypertrophy, elevated basal corticosterone levels or
immunosuppression [138]. In contrast to voluntary wheel running,
treadmill training reportedly exerts null [7] or detrimental effects
on cognitive abilities [13]. Furthermore, treadmill performance and
voluntary wheel running do not correlate in mice [120] or in rats
[113], supporting the view that they may represent two distinct
forms of exercise.
It has been described for a long time that rats [74,92,103,141]
and mice [59,60,103,166] in captivity spontaneously and intensively
perform wheel running. The possibility of voluntarily exploiting the
running wheel as well as to control the amount of running activity
seems to be of high relevance; mice placed in motorized wheels
were found to operate a switch to stop a rotating wheel, and to
restart it [102,103].
1.1.3. Wheel running
Converging lines of evidence strongly indicate that wheel run-
ning can exert powerful impacts on brain and behaviour in rodents.
In rats two nights of free access to a running wheel are sufficient
to increase hippocampal BDNF expression [135,143], and a single
week of exercise sufficient to improve memory performance in the
water maze task [199]. With extended training up to 3–4 weeks,
exercise can lead to further marked effects on behaviour and cog-
nition [1,135]. Hence, the majority of the studies employing wheel
running as a tool to investigate neurobehavioural plasticity have
allowed animals to access the wheel for about 3–6 weeks. This pro-
cedure has the advantage not only of maximizing the probability
of observing significant brain and behavioural changes, but also of
obtaining a stable leve l of wheel running activity (see Section 1.2.1).
Unlike environmental enrichment or isolation rearing, wheel
running activity reportedly exerts marked neurobehavioural effects
not only when implemented during the juvenile age [196], but also
in adulthood [12,91,139,191] and senescence [32,172]. Voluntary
physical exercise therefore differs from social isolation and environ-
mental enrichment as its efficacy is largely not restricted to a critical
developmental window in early life. Indeed, most of the existing
studies in both rats [139,191,197,199] and mice [12,91,165,193] initi-
ated wheel running training at adulthood, i.e., around 3–5 months
of age. The reason for the preferential employment of adult ani-
mals is two-fold. First, at adulthood the patterns of wheel running
activity are expressed in a robust and stable manner in both rats
[36,152,153] and mice [99,192]. Second, since individual quantifi-
cation of individual wheel running activity would require single
housing, using adult subjects can circumvent the effects of social
isolation that are expected to have the highest impact when imple-
mented in juvenile animals.
The possibility of quantifying the amount of wheel running
activity represents a further advantage of this environmen-
tal manipulation compared to social isolation or environmental
enrichment. The analysis of wheel running patterns can reveal
potential alterations in the circadian rhythm or in the amount of
running activity which may be critical determinants for the induc-
tion of the effects of exercise. It is indeed possible that the amount
of wheel running activity may critically influence the effects of
exercise on certain behavioural variables [91,165]. Animals display-
ing abnormal levels of wheel running, for example due to genetic
manipulations, might therefore fail to benefit from the access to
a running wheel. At present, the psychological relevance of the
amount of wheel running is however far from being understood,
as it will be discussed in depth in Section 1.2. Nevertheless, the
quantification of wheel running activity can considerably add to the
power of data interpretation in studies of neurobehavioural plastic-
ity. In this regard, it is interesting to note that quantitative analysis
of wheel running patterns has recently been included as a simple
and reliable procedure to phenotype genetically-modified mouse
lines [38].
1.2. Wheel running activity: characteristics and functions
Although voluntary wheel running is typically associated with
laboratory and pet rodents, it has been described in a wide variety
of species, including the domestic hen [132] and the red fox [104].
Rodents are highly motivated to perform wheel running, as sug-
gested by studies in which a cost (e.g., in the form of an operant
demand or the administration of an aversive stimulus) is imposed
for its access. Both rats [27] and mice [179] are found to b e moti-
vated to gain access to areas containing a wheel as well as to unlock
it [26,100]. Indeed, voluntary wheel running in rodents is often
performed with prodigious intensity. The distance run in 24 h can
reach about 40 km in rats [166] and 20 km in mice [103]. A large
number of studies conducted in the last century have provided
us with the detailed quantitative and qualitative analysis of wheel
running activity in a number of rodent species (for a review see
1.2.1. Patterns of wheel running activity
All rodents, with the exception of hamsters, undergo an initial
phase of gradual increase in wheel running activity in the first 4–6
weeks before stabilization [29]. This initial phase is somewhat more
extended in rats (circa 6–7 weeks, [79,92,139]
) than in mice (about 4
weeks, [37,60,71,110]). Captive laboratory mice [37,59,109] and rats
[74,107,136] also exhibit a circadian pattern in their running activ-
ity. They usually begin running at lights off with the peak running
period occurring 1–4 h after lights off. Running gradually decreases
after lights are turned on and is negligible during the light phase.
Besides these common characteristics, however, the fine pattern of
wheel running activity can vary considerably between sexes and
across strains (see below).
1.2.2. Genetic factors affecting wheel running performance
In recent years, there has been considerable interest in eluci-
dating the genetic contributions to exercise performance. Human
studies have identified genetic polymorphism in association with
individual difference in the propensity to engage in physical exer-
cise [14,84,85]. This is in line with the observation that sex and
strain dif ferences can be readily observed in the qualitative and
quantitative expression of wheel running activity in rodents.
More vigorous running (in terms of amount as well as speed) by
the female sex has been described in several strains of rats [52,98]
and mice [122,163]. In female rats [8] and mice [72,109,164], the
levels of voluntary physical activity are found to be modulated
by estrogen levels. Physical exercise is enhanced in the proestrous
phase [3,72,109,164], and dramatically reduced following estrogen
deprivation [8].
Differences among strains in wheel running have also been
reported in both mice [9,37,122] and rats [97,107,181],notonlyin
the amount of running, but also in the rhythmicity of the activ-
ity patterns [97,107,181]. Such inter-strain variability in mice may
stem from variations in the expression of genes involved in cardiac
function and hemodynamics [120]. It has also been hypothesized
that such variability may be related to specific psychophysiologi-
cal traits persisting amongst strains despite selective inbreeding.
However, strain specific patterns of wheel running activity in mice
do not seem to predict consistent differences in behavioural or psy-
chological traits which include explorative [65,149] and cognitive
behaviour [65]. Yet, correlation between individual differences in
running activity and specific behaviour may also be possible within
44 S. Pietropaolo et al. / Behavioural Brain Research 192 (2008) 42–60
a given strain (see Section 1.2.4), which certainly deserves further
1.2.3. Behavioural characteristics of genetically determined high
A useful approach to investigate the relevance and impact of
differing levels of wheel running consists of employing animals
selectively bred for high wheel running performance. After about
20 generations, this can generate mice with a three-fold increase in
wheel running activity relative to non-selected controls, yet they
display the same qualitative running pattern [16,70,72,164,165].
However, such prodigious runners did not differ from controls
in terms of water maze learning or hippocampal BDNF content
[165]. More surprisingly, they did not even differ in open field
activity levels [16], although they were more active than con-
trols in the home cage [164]. This indicates that the propensity
of performing wheel running and open field activity may rep-
resent genetically dissociable behavioural traits [180]. This is in
line with the complementary observation that mice selectively
bred for high locomotor activity in the open field test did not
differ in their wheel running levels [39]. In that study however,
mice were exposed to the wheel running activity only for 2 days.
The possibility that a different outcome might emerge with a
longer period of exposure certainly warrants further investigation.
Indeed, the precise relationship between the amount of home cage
wheel running and open field locomotor activity levels remains
1.2.4. Correlational analyses
While the selective breeding strategy may be useful in character-
izing the genetic influence on wheel running activity and possible
associated traits, it is less readily able to capture the relationship
between the amount of wheel running and behavioural traits in
the normal population. In this regard, non-genetic factors may play
a more substantial role. As an alternative approach, correlational
analyses between individual differences in the amount of wheel
running and specific behavioural traits within a random sample
of subjects taken from a single mouse or rat strain can be carried
out. This approach is favoured by the presence of large individual
differences in the levels of wheel running activity that even exist
within an inbred mouse strain. Harri et al. [89] reported that wheel
running varied from 5 to 200 km/day in a sample of 17 mice of the
C57BL/6 strain.
However, very few studies have adopted the correlational
approach and those have been limited to few behavioural vari-
ables. In rats wheel running activity over 6 weeks does not correlate
with shuttle boxescape behaviour or conditioned freezing response
[79]. No correlation between running levels and water maze per-
formance has been reported in outbred female mice [165].
We have investigated the relationship between the amount of
wheel running activity and measures of emotionality, exploratory
activity, sensori-motor processing and spatial memory in male
C57BL/6 mice. To this end, 21-day-old mice received access to home
cage running wheels for 5 weeks and were then evaluated for
anxiety, open field locomotor activity, prepulse inhibition of the
acoustic startle response and reference memor y learning in the
water maze. The only behavioural variable that correlated signif-
icantly with home cage running activity was open field locomotor
activity, as indexed by the total distance travelled over the 60-min
testing session (see Fig. 1).
The lack of extensive correlation was surprising given that most
of the behaviours considered in our study are known to be altered
by voluntary home cage running when compared with sedentary
(standard laboratory housing) controls (see Section 1.3). The lack
of correlation therefore suggests that the precise levels of wheel
running exercise may not be a critical determinant for the emer-
gence of an effect of home cage voluntary exercise. On the other
hand, the amount of voluntary exercise has been reported to cor-
relate positively with neurogenesis levels in the dentate gyrus of
adult mice [91,165], although there appeared to be a threshold such
that the correlation was not found in mice selected for high levels
of running due to a ceiling effect [165]. This suggests that there
is a limit to the amount of neurogenesis that wheel running is
able to induce [165]. The correlation between the amount of exer-
cise and its effects on the brain also appears to depend on sex:
a positive correlation between running activity and hippocampal
BDNF levels was found in male [142,148], but not in female [165]
Finally, it has been shown that in male rats running speed cor-
relates positively with the discharge rate of selected hippocampal
neurons (in about 10% of recorded CA1 cells), which have been iden-
tified as “wheel neurons” [34]. However, it should be noted that
Fig. 1. Wheel running levels and their correlation with locomotor activity in the open field. Twenty-nine na
ıve male C57BL/6 mice were allowed access to a running wheel in
their home cage from weaning to 5 weeks before being subjected to the elevated plus maze test of anxiety, open field locomotor activity, prepulse inhibition of the acoustic
startle response and reference memory learning in the water maze. All tests were conducted in the dark phase, and the experimental procedures and data analysis were
similar to those described in Section 2.2. Pre-testing wheel running activity during the dark phase gradually increased during the first 3 weeks and remained stable afterwards,
as typically seen in mice and rats (A). The running levels during the light phase were maximal in the first week and then quickly reduce d to a consistently low level of activity
(B). Amongst all the behavioural variables examined, open field locomotor activity (indexed by the total distance travelled in 60 min) was the only variable that correlated
significantly with home cage running activity (p < 0.0005, d.f. = 27). This correlative relationship was strictly related to activity recorded in the dark (C) but not in the light
phase (D). The solid line in (C) and (D) represents the fitted linear regression line.
S. Pietropaolo et al. / Behavioural Brain Research 192 (2008) 42–60 45
the animals in that study were not tested in the home cage and
were trained to run on the wheel to obtain a reward. Nonetheless,
these data provide further support for the existence of a marked
inter-dependence between running activity and hippocampal neu-
rophysiology. This highlights the role of exercise in modulating
mental health, from cognition to emotion.
1.3. The neurobehavioural effects of wheel running: physical
exercise and mental health
Over the past decades, a large number of studies have demon-
strated the benefits of exercise on brain function in humans.
Physical activity significantly reduces the risk of developing major
depression disorders in healthy subjects [189], and attenuates the
clinical symptoms in depressed patients [43,129]. Furthermore,
both prospective [117] and retrospective [63] studies have indicated
that physical exercise is associated with reduced risk of dementia
and Alzheimer’s disease (AD) in later life, and is effective in improv-
ing a variety of behavioural and cognitive symptoms in AD patients
Although it is generally accepted that the beneficial effects
of physical activity on mental function stem primarily from its
effect on the brain rather than on general physical health [30],
only recently have researchers begun to unravel the mechanisms
underlying the neurobiological effects of exercise. Animal models
of voluntary exercise have played a crucial role in corroborating
and providing neurobiological explanations for the positive clinical
outcomes observed in humans.
1.3.1. The effects of exercise on the intact brain: the role of
neurotrophic factors
Although some exceptions have been reported [154,188], one
of the main behavioural effects of physical exercise in rodents is
improved cognitive function as revealed in a number of studies
assessing both spatial (rats: [1,44,197,199] mice: [87,114,126,193])
and non-spatial memory (rats, [53]; mice, [172]). Furthermore,
reduced anxiety- and depression-like behaviours have also been
reported following wheel running exercise in rats [45,79,80,211] as
well as in mice [12,182].
Most of the notable exercise-induced brain changes that may
potentially be linked to the behavioural outcomes of exercise
have been observed in the hippocampus [30] a highly plas-
tic brain area with both cognitive and emotional functions. In
an attempt to identify the molecules implicated in the b eneficial
effects of exe rcise, Tong et al. [191] have examined genes expres-
sion in the rat hippocampus following 3 weeks of physical exercise
and reported changes in the expression of about 90 genes. These
genes were mostly involved in neurogenesis and synaptic function,
such as vesicle recycling, synaptic growth or neurotrophic factor
Neurotrophic factors involved in neuronal survival and differ-
entiation as well as synaptic plasticity [124,130] are considered
to be the most likely candidates in mediating the effects of exer-
cise on brain and behaviour. Expression of several neurotrophins
including nerve growth factor (NGF) [143], fibroblast growth fac-
tor 2 (FGF-2) [75,76] and brain-derived neurotrophic factor (BDNF)
[8,142,143,170] has been shown to increase in the rat hippocampus
only after a few days of exercise. In particular, exercise-induced
upregulation of BDNF appears to be more robust and long last-
ing compared to the other neurotrophins, suggesting its potentially
major role in mediating the long-term effects of exercise. Indeed,
a relationship between hippocampal BDNF and learning [86,133]
as well as with emotionality [21,209] has been proposed. Further-
more, not only is BDNF central to the modulation of emotive and
cognitive functions, but it is also related to energy homeostasis
[105,127] the other main physiological process that is reliably
affected by physical exercise. The BDNF signalling system is sug-
gested to be the fundamental link by which exercise affects energy
metabolisms and modulates neuroplasticity implicated in cogni-
tive and behavioural functions [198]. The pivotal role of BDNF is
supported by the demonstration that exercise-induced improve-
ment in spatial memory can be abolished by quenching the action
of endogenous BDNF with TrkB-IgGs [197].
The continual genesis of new neurons in the dentate gyrus of
adult mice [193,194] and rats [58] may also underlie the exercise-
induced enhancement in learning and memory [193,194] as well
as its anti-depressant effects [51,57]. Here, BDNF may also play a
major role, since a positive relationship between BDNF adminis-
tration and neurogenesis in the dentate gyrus has been described
[177]. One clear implication of these findings is the application of
exercise to prevent, ameliorate, or counter the structural and func-
tional deterioration implicated in brain trauma, aging, stress and
neurodegenerative diseases.
1.3.2. The effects of exercise on the damaged brain Focal brain injury. Voluntary exercise can be beneficial to
a variety of central nervous system injuries, including cerebral
ischemia [4,126], lesions of the brain [81] and of the spinal cord
[54,55], either before or after the damage has occurred. These ben-
eficial effects have been attributed to enhanced neuroprotection
or angiogenesis [126] and/or reduced apoptosis and edema [108].
Neurotrophins are potential candidates in mediating some of these
protective effects of exercise on brain injury. There is evidence to
suggest that exercise-induced reduction in brain infarct volume
after focal ischemia in rats is accompanied by an increase in BDNF
expression [4]. Aging. Experiments in rats [134] and mice [87,172] have
largely confirmed the observations in humans [111,117] that phys-
ical exercise can counteract the cognitive decline in senescence.
In rats, exercise reportedly reduces membrane lipid peroxidation
and oxidative DNA damage in the brain [160], and hippocam-
pal expression of mitochondrial proteins [191], suggesting that
the anti-aging effects of exercise may be mediated by reduced
oxidative stress [77]. These effects can also be linked to BDNF,
which is efficacious against oxidative stress and free radical
production [23]. Furthermore, there is some evidence to sug-
gest that brain BDNF content decreases as a function of age,
especially in the hippocampus, which may suggest that under-
activity of the BDNF signalling system (BDNF synthesis, content
and its receptors) may contribute to the emergence of cogni-
tive impairments observed in senescence ([131,137,176]; see also
[131,137]). Finally, the anti-aging effects of physical exercise could
also be related to its efficacy in countering the marked reduc-
tion in hippocampal neurogenesis which is associated with aging
[112]. Stress-induced alterations. Physical activity can mitigate
the harmful consequences of acute stress exposure on brain
and behaviour. Hence, exercise can prevent the expression
of depression-like behaviours following exposure to uncon-
trollable stress [45,79]. Possible neural mediators of exercise’s
anti-depressant and stress-protective effects include the seroton-
ergic and noradrenalinergic systems [45,79]. In addition, exercise
increases the brain and peripheral expression of the protective cel-
lular stress protein (e.g., the heat shock protein 72) in rats [146] as
well as in humans [115].
It is well known that stress can precipitate the onset and exac-
erbate the severity of many neuropsychiatric conditions, including
depression and anxiety [24,47], as well as cognitive dysfunction
46 S. Pietropaolo et al. / Behavioural Brain Research 192 (2008) 42–60
[15,174]. One hypothesis is that exercise attenuates the stress
response and thereby provides some protection against the dele-
terious effects of stress on emotionality [6,171] and cognition [62].
Increased susceptibility to stress is also associated with a higher
risk of developing Alzheimer’s disease [40,206]. Hence, failure in
stress adaptation may play a role in the pathogenesis of this major
neurodegenerative disorder, and exercise may protect against it. Alzheimer’s disease. In line with the stress adaptation
model, the “cognitive reserve hypothesis” has been formulated in
an attempt to explain the preventive/protective effects of physical
exercise. This hypothesis, which was originally proposed in the con-
text of dementia [186,210], posits that physical activity enhances
the reserve capacity of the brain to react to pathological insults.
Accordingly, individuals with a higher cognitive reserve fostered by
an active lifestyle would tolerate and withstand the development
of brain pathology predisposed by genetic factors [62].
A major impetus to the elaboration and development of the
reserve hypothesis can be attributed to the recent advent of genetic
mouse models of familial AD. Typically, AD pathology in these
mouse lines is induced by the mutation of single or multiple genes
encoding for the amyloid-precursor protein (APP), presenilins (PS1
or 2) or tau protein [56,96].
Early evidence indicating that wheel running may have an
impact on the development of AD pathology in specific trans-
genic mouse models of the disease was derived indirectly, i.e., from
studies involving enriched environments in which running wheels
were included [5,94,118]. These studies have shown that enriched
environment improved spatial memory in transgenic AD mice
[5,94,207], and enhanced BDNF levels as well as neurogenesis in the
hippocampus [207]. Although not without exceptions [5,94,207],
enriched environment was reported to reduce -amyloid pathol-
ogy in transgenic AD mice, and this effect was more pronounced in
mice spending more time running on the wheels [118]. This led to
the suggestion that wheel running activity may assume a major
role in the resulting beneficial effects of environmental enrich-
ment on AD pathology [94,118]. These findings therefore provide
the impetus to investigate the protective effects of voluntary exer-
cise as such on the development of AD-like pathology. This is further
encouraged by the fact that the efficacy of enriched environment
seems to depend greatly on its application in early post-weaning
life, whereas physical exercise remains highly effective even when
applied later in life (see Section 1.1.3).
Interestingly, one recent study has shown that 9-month expo-
sure to enrichment, but not to exercise alone, enhanced water maze
learning in 11-month-old APP-23 transgenic mice, although neither
manipulation affected plaque load in the brain [207]. It appears that
the lack of an exercise effect in this study could be related to the
fact that grouped (instead of single) housing was adopted. This is
because Adlard and colleagues showed that 5 months of exposure
to exercise (from 1 month of age under single housing) was suf-
ficient to improve water maze learning and suppress -amyloid
pathology in the TgCRND8 mouse line [2]; and the reduction of
-amyloid pathology was as effective as that observed following -
amyloid immunization in the same mouse line [95]. Similarly, with
single caging, even brief exposure to exercise for a few weeks at an
advanced age (when AD pathology had already established) was
also effective in ameliorating cognitive performance in the Tg2576
mouse model [145]. Although the contribution of other procedural
differences (age and genotype of the subjects, housing conditions)
to the discrepant outcomes between studies cannot be entirely
excluded, it is possible that exposure to the running wheel under
grouped housing conditions may not be ideal. This may lead to com-
petition amongst cage-mates for access to the wheel and thereby
result in insufficient training in some of the animals.
The existing studies on exercise and transgenic mouse lines
related to AD are however characterized by certain caveats. First,
few studies have provided quantification of wheel running activity
and the quantification included was often of short duration [145]
or limited to the transgenic group only [2]. Hence, it is uncertain if
the transgenic mice under investigation differ from control mice in
their basal wheel running activity (in terms of both amount and pat-
tern). As seen in wild-type mice (see Section 1.2), the actual amount
of wheel running activity may influence the efficacy of exercise in
modifying the phenotypes of AD-mouse lines.
Second, most of the studies on wheel running and AD-related
mouse lines have exclusively utilized female mice [2,207]. Hence,
another uncertainty concerns the possible presence of sex dif-
ference in the efficacy of exercise to alter AD-like pathology in
these transgenic mice. The lack of data in this respect is surpris-
ing given that sex difference in the severity and onset of brain
and behavioural AD-pathology [20,25,175,201] as well as in the
response to environmental manipulations [83,156,212] have been
previously described in mice.
Third, the scope of current studies on exercise and its abil-
ity to suppress the emergence of AD-related symptoms has been
mainly limited to the cognitive domain. Yet, the evaluation of the
non-cognitive symptoms in relation to AD is equally warranted,
especially, due to the possible inter-dependence between cognitive
and non-cognitive AD symptoms reported in humans [33,140,183]
2. The impact of voluntary wheel running on the
emergence of AD-like symptoms: a study in the 3×TG-AD
mouse model
2.1. Aims and experimental design
Here, we present an experimental investigation of the patterns
and the behavioural effects of wheel running activity in a triple-
transgenic mouse model of AD (3×Tg-AD) harbouring PS1
, and tau
transgenes in comparison to wild-type con-
trol mice. This mouse model is characterized by the age-dependent
emergence of two critical neuropathological hallmarks of the AD,
i.e., -amyloid plaques and neurofibrillary tau tangles [147].At6
months of age, intracellular -amyloid immunoreactivity appears
in the hippocampus and the amygdala, in the absence of tau pathol-
ogy [147].
At this age, in the absence of physical impairments 3×Tg-AD
mice reportedly displayed mild deficits in water maze learn-
ing [10,11,19,25,69,147], although the robustness of this cognitive
phenotype has been questioned by previous data. Mutant mice
also showed impaired performance in inhibitory avoidance tests
[10,19,25,147], but they were comparable to wild-type controls in
object recognition memor y [25] and in the acquisition of Pavlo-
vian conditioned freezing [155]. Furthermore, the non-cognitive
phenotype of the 3×Tg-AD mouse line has been extensively char-
acterized at multiple ages in a previous study [155]. At 6 months
of age, 3×Tg-AD animals showed normal emotionality in the ele-
vated plus maze and open field, but they displayed marked signs
of enhanced reactivity to aversive stimuli, as demonstrated by the
enhanced acoustic startle response, and this latter finding may also
explain the improvement observed in the cued version of the water
Yet, it is not clear whether the various phenotypes of the 3×Tg-
AD model might b e expressed differentially between sexes. A
lack of sex differences have been described in the water maze
and inhibitory avoidance at 6 months of age in some reports
[10,11,19,69,147], while another study from the same group have
also reported the presence of more severe deficits in female
S. Pietropaolo et al. / Behavioural Brain Research 192 (2008) 42–60 47
3×Tg-AD mice in both tests [25]. On the other hand, no differ-
ences between sexes have been described in open field behaviour
of 3×Tg-AD mice at around 5 months of age [144]. Indeed,
little is known about sex differences in the non-cognitive phe-
notypic profile of the 3×Tg-AD mouse line, since an extensive
behavioural characterization has thus far been performed only in
males [155].
The aims of the present study were: (i) to evaluate sex dif-
ferences in both the non-cognitive and cognitive phenotype of
the 3×Tg-AD mouse line, (ii) to investigate whether 3 months
of exposure to voluntary exercise at adulthood would attenuate
the AD-symptoms, and, (iii) to assess the potential dependency
on sex of these exercise effects. Here, we employed a 2 × 2 × 2
factorial design including the factors genotype, sex and run-
ning in animals that had been subjected to either running or
sedentary conditions from 3 months of age (i.e., at the pre-
pathological stage for the mutants). Animals housed in cages
equipped with a locked wheel were employed here as sedentary
controls [73,91,139,154] in order to avoid the confounding effects
due to the mere presence of the wheel (See Section 3.4). Multiple
behavioural assays were conducted to examine anxiety, spon-
taneous locomotor activity, sensori-motor response and spatial
learning in order to characterize the impact of voluntary exercise
on behaviour against genetic disposition to develop AD-like pathol-
ogy in the 3×Tg-AD mice. In addition, -amyloid immunoreactivity
was also assessed in the hippocampus and amygdala to evaluate
potential effects of exercise and sex on the severity of AD-brain
2.2. Methods
2.2.1. Subjects
The subjects were 36 mutant 3×Tg-AD and 34 wild-type mice of matched genetic
background. Subjects of both sexes were used.
Both the mutant and control wild-type mouse lines were originally generated
and maintained by LaFerla and colleagues, and a full descriptions of their generation
has been reported previously [147]. Briefly, founder mutant 3×Tg-AD mice were
generated by microinjection of human APP
and tau
transgenes in single-cell
embryos obtained from PS1
homozygous knockin mice (PS1-KI). Founder mice
were backcrossed to PS1-KI mice. Because the APP and tau transgene co-integrate at
the same site, the 3×Tg-AD mice all have the same genetic background. Mice with
the identical genetic background of 129/C57Bl6 hybrids (based on the initial PS1-KI
mouse line) were used as controls. Separate breeding pairs of homozygous 3×Tg-
AD mice and the corresponding wild-type controls were obtained directly from Dr.
Frank LaFerla (University of California, Irvine). They were maintained and bred in the
SPF (specific-pathogen-free) facility in the Laboratory of Behavioral Neurobiology
(Swiss Federal Institute of Technology, Zurich); and the offspring were used as the
experimental subjects in the present report. Genotype status was re-confirmed by
PCR analysis of DNA isolated from tail tips collected after animals were sacrificed.
The animals were weaned on postnatal day (PND) 21 and housed in unisexual
groups of littermates in standard Makrolon cages (Tecniplast, Milan, Italy), measur-
ing 41 cm × 20 cm × 19 cm. Sawdust bedding (Schill AG, Muttenz, Switzerland) was
renewed weekly. Food chow (Provimi Kliba SA, Kaiseraugst, Switzerland) and water
bottles were provided on the stainless steel cage top.
Here, we refrained from determining the oestrous cycle of the female mice and
to avoid (i) stressing the animals due to repeated vaginal smears, and (ii) introducing
a confounding factor with sex since the male mice could not be equally treated.
2.2.2. Maintenance and housing
At 3 months of age, mice were transferred to a separate temperature (22 ± 1
and humidity (55 ± 5%) controlled room. They were caged singly in standard
Makrolon cages identical to those described above, except that they were equipped
with either a running or locked wheel [73,91,139,154]. We ensured that members
from a given litter were randomly allocated between the two housing conditions
(i.e., either running or sedentary), in order to minimize possible confounding due to
litter effects [213]. The number of subjects in each experimental group comprised
male wild-type: 8 sedentary, 9 running; male mutant: 9 sedentary, 10 running;
female wild-type: 10 sedentary, 7 running; female mutant: 9 sedentary, 8 running.
The animals were kept on an ad lib feeding condition and maintained under
a 12:12 h reversed light–dark cycle (lights off: 07:00–19:00 h). The ambient light
intensity in the animal vivarium during the light phase was 250 lux. Animals were
left undisturbed for 3 months prior to testing except for the renewal of sawdust
bedding, performed every 10–14 days.
2.2.3. Installation of running-wheel
A plastic running wheel with a solid back and a solid running surface, measuring
10 cm in diameter and 6 cm deep (Rolf C. Hagen Corp., Mansfield, MA) was mounted
beneath the metal lid of each cage. In half of the cages, the wheel was locked to pre-
vent it from rotating. In the remaining cages, the rotating drum remained free, and
the animals could freely perform running exercise. Mice housed in cages equipped
with a locked wheel might explore the inside of the drum, and sometimes preferred
to nest under the immobile wheel [73,91,139,154].
2.2.4. Assessment of pre-testing wheel running activity
Movement of the rotating wheel was detected by a sensor mounted on the cage
top responsive to the magnetic field created by four magnets fixed on the outer sur-
face of each running wheel. Each passage of a magnet under the sensor generated
an electric pulse, which was transmitted to a PC controlled by an in-house program
(by Peter Schmid, ETH Zurich). The impulses were quantified into 30-min bins, and
stored for subsequent analysis. The record began at 20:00 h on the first day of dif-
ferential housing and lasted for 90 days, until the day before the commencement
of behavioural testing. Days in which sawdust in the cages was renewed (6 days
in total over the 90-day pre-testing period) were excluded from the analysis. The
data set therefore comprised a total of 84 days. Running data were analysed from
the beginning of the dark phase of the first day of differential housing until the end
of the light period of the last pre-testing day; the first night of habituation to the
differential housing conditions was excluded.
2.2.5. Behavioural procedures
Behavioural testing commenced when the animals were 180 days old. Mice were
maintained in the corresponding housing conditions throughout the entire experi-
mental period, which lasted for about a month. Five tests were included in this study:
(i) elevated plus maze test of anxiety, (ii) open field test of spontaneous locomotor
activity and spatial exploration, (iii) startle reactivity, (iv) spatial reference mem-
ory in the water maze, and (v) spontaneous alternation in the Y-maze. The animals
were approximately 180, 183, 190, 200 and 218 days old at the start of the five tests,
respectively. When possible, tests that relied mainly on observations of spontaneous
behaviour were conducted first in order to minimize possible undesirable transfer
effects [67]. Behavioural tests were always carried out during the dark phase of the
cycle. All manipulations described here had been approved by the Cantonal Veteri-
nary Authority of Zurich, and were in accordance to the directives of the European
Union (86/609/EEC). Elevated plus maze test. The construction and dimensions of the elevated
plus maze have been fully described elsewhere [154,156]. A digital camera was
mounted above the maze. Images were captured at a rate of 5 Hz and transmitted
to a PC running the Ethovision (Version 3.1, Noldus Technology, The Netherlands)
tracking system.
To begin a trial, the mouse was gently placed in the central square with its head
facing one of the open arms. It was allowed to explore freely and undisturbed for
5 min.
Two anxiety-related measures were calculated: percent time in open
arms = time in open arms/time in all arms × 100%, and percent entries into open
arms = number of entries into open arms/number of entries into open and enclosed
arms × 100%. In addition, the total distance travelled in the entire maze surface (i.e.,
arms and central platform) was recorded. Open field Test. Details of the four identical open fields used here have been
fully described before [154,156]. The subjects were tested in squads of four counter-
balancing for sex and housing conditions. They were gently placed in the centre of
the appropriate arena and allowed to explore undisturbed for 60 min. Afterwards,
they were returned to the home cage, the arenas were cleansed with water and dried
prior to the next squad. Digital images were captured at a rate of 5 Hz by a camera
mounted above the open fields and were transmitted to a PC running the Ethovi-
sion (Version 3.1, Noldus Technology, The Netherlands) tracking system. Locomotor
activity was indexed by the distance travelled recorded in consecutive 10-min bins. Assessment of the acoustic startle reflex. A set of four acoustic startle cham-
bers for mice (SR-LAB, San Diego Instruments, San Diego, CA, USA) was used. A full
description of the apparatus has been provided elsewhere [154,156]. Acoustic startle
reflex was assessed during a session lasting for approximately 30 min. The subjects
were presented with a series of discrete acoustic white noise stimuli against a con-
stant 65 dB
background noise. The acoustic stimuli varied randomly amongst 10
intensities: 69, 73, 77, 81, 85, 90, 95, 100, 110 and 120 dB
, and lasted either 20 or
40 ms in duration. The test began when the animals were placed inside a Plexiglas
enclosure, and allowed to acclimatise to the apparatus and background noise for two
minutes before the first acoustic stimulation. The first six trials consisted of acous-
tic stimuli of the highest intensity only (120 dB
, three trials with 40 ms and three
with 20 ms stimulus duration). These trials served to stabilize the animals’ startle
response, and were analysed separately. Subsequently, the animals were presented
with five blocks of discrete test trials. Each block consisted of 20 pulse-alone trials,
48 S. Pietropaolo et al. / Behavioural Brain Research 192 (2008) 42–60
one for each stimulus intensity and stimulus duration. All trials were presented in
a pseudorandom order, with an average inter-trials interval of 13 s. Spatial reference memory in the water maze. The dimensions of the water
maze have been described before [154]. The maze was located in the middle of
a quiet room enriched with several distal cues. On each experimental day, mice
were habituated to the experimental room: they were individually housed in stan-
dard Makrolon cages (26 cm × 21 cm × 14 cm in size) and left undisturbed for about
10 min before testing began. A digital camera was mounted above the water maze,
capturing images at a rate of 5 Hz and transmitting the data to a PC running the
Ethovision tracking system. On each trial the latency and distance swum to reach
the platform (visible or hidden) were recorded. Performance on probe tests was
evaluated by percent time spent and distance swum in the target quadrant in com-
parison to the remaining quadrants. Average swim speed on each trial was also
computed and submitted to statistical analysis.
On days 1 and 2, the animals were tested in the visually-cued task across four
consecutive trials with a 1-min ITI. The platform surface (measuring 14 cm in diame-
ter) was 0.5 cm above the water level, and was made visible by a black disk mounted
directly above it. This served to habituate the animals to swimming in the pool and
to escaping from the water by climbing onto the platform. It also assessed possible
non-cognitive alterations related to swimming such as visual abilities, and general
motivation to escape from the pool. The platform was positioned in different loca-
tions across the four trials: in the centre of the pool and in the centre of the three
quadrants not serving as location of the hidden platform in the subsequent reference
memory task. The starting point for releasing the mouse in the pool was constant
across trials within a day, but changed from day 1 to 2. Platform locations as well
as the starting position were as far as possible counterbalanced with respect to all
between-subject factors. A trial ended when the animals escaped onto the platform
or when 60 s had elapsed, at which time the animal was guided to the platform by
the experimenter. The animal was allowed to spend 30 s on the platform. Afterwards
it was placed into a waiting cage for further 30 s prior to commencement of the next
On days 3–8, the animals were trained to locate the escape platform, which was
now hidden under the water surface and remained in a constant location (in the
middle of one of the quadrants, 22 cm off the maze wall). Across the four trials in a
day, the start positions varied amongst N, E, S and W in a pseudorandom sequence.
Otherwise, the testing procedures were identical to those described above.
On day 9, in addition to the four trials of hidden-platform training, two 30-s
probe tests were conducted: 1.5 h before and 1.5 h after hidden-platform training.
The mice were returned to the home cage for the 1.5 h intervals. In the probe test,
the platform was removed from the watermaze, and the animal was released into
the quadrant opposite to the one in which the platform was previously located (i.e.,
the “target” quadrant).
On day 10, the animals underwent another 30-s probe test, as described above.
A similar procedure employing multiple probe tests had been employed by previous
studies carried out in the 3×Tg-AD model [10,11,69]. Spontaneous alternation in the Y-maze. Spontaneous alternation was
assessed in a grey, wooden Y-maze, elevated 80 cm from the floor and located in
the middle of a room containing a variety of extramaze cues. Mice were habituated
to the experimental room as described for the water maze experiment. The walls
of the maze were 10 cm high and 1 cm thick; each arm was 50 cm long and 10.5 cm
wide and the floor of the maze was covered with sawdust bedding. A digital camera
was mounted above the maze, capturing images at a rate of 5 Hz and transmitting
the data to a PC running the Ethovision tracking system.
Mice were assigned two arms (start and familiar arm) to which they were
exposed during the first phase of the test (sample phase). The remaining third arm
constituted the novel arm during the second phase (test phase). Allocation of arms
(start, familiar and novel) was counterbalanced within each experimental group.
During the sample phase, access to the novel arm was blocked by a grey wooden
door, 12 cm high, 7.5 cm wide and 0.5 cm thick. Mice were placed at the end of the
start arm and allowed to explore freely both the start and the other unblocked arm
for 5 min before being removed and returned to the waiting cage. Once the mouse
had left the start arm, timing of the 5 min sample phase period began. After 90 s in
the waiting cage, the test phase began. During this phase, the door was removed and
all three arms were unblocked. To begin the test phase, mice were placed at the end
of the start arm and allowed to explore the entire maze for 2 min once they had left
the start arm. In the interval between the exposure and the test phase the sawdust
from each arm was mixed and randomly distributed in order to overshadow possible
olfactory cues.
2.2.6. Assessment of ˇ-amyloid pathology Immunohistochemistry. Animals were deeply anesthetized with an overdose
of sodium pentobarbital (Nembutal
; 40 mg/kg, i.p.) and perfused transcardially
with ice-cold saline. The brains were removed in toto, and then bisected into two
hemispheres. The left hemisphere was immediately frozen and the right hemi-
sphere was post-fixed for 24 h by immersion into a cold fixative (0.15 M phosphate
buffer with 4% paraformaldehyde and 15% saturated picric acid solution, pH 7.4).
After fixation, the right hemisphere underwent microwave-assisted processing as
previously described [64]. Following cryoprotection with 30% sucrose solution,
free-floating coronal sections (40 m thick) were cut on a freezing microtome
and eight series of about 20 interleaved sections were sampled from bregma
+2.58 to 4.16 mm and stored at 20
C in cryoprotectant solution until further
Immunohistochemistry was performed on one series of brain sections, which
was randomly selected per animal, with the mouse anti-human -amyloid pro-
tein as primary antibody (APP, Chemicon, Temecula CA, 1:500). The free-floating
sections were rinsed in PBS three times for 10 min and incubated for 1 h in PBS
containing 5% normal goat serum (NGS) and 0.3% Triton X-100 at room tempera-
ture. Sections were incubated overnight at 4
C in the primary antibody diluted in
PBS containing 2% NGS and 0.3% Triton X-100 (antibody buffer). Sections were then
washed with PBS and incubated with the biotinylated secondary antibody (Jackson
ImmunoResearch Laboratories Inc., Pennsylvania, USA) diluted 1:500 in antibody
buffer for 1 h at room temperature. After three washes in PBS, sections were incu-
bated in Vectastain ABC Kit (Vector Laboratories Inc., California, USA) diluted in PBS,
for 1 h at room temperature, and washed three times in 0.1 M Tris–HCl buffer, pH 7.4.
Immunoreactivity was visualized by standardized DAB method: sections were incu-
bated with 1.25% 3,3-diaminobenzidine (DAB, Fluka, Buchs, Switzerland) and 0.08%
in 0.1 M Tris–HCl, pH 7.6 for 2–15 min, and then washed three times in PBS.
Sections were then mounted on gelatine-coated slides, dried overnight, dehydrated
and coverslipped with Eukitt
(Kindler GmbH & Co, Freiburg, Germany). Evaluation of -amyloid immunoreactive cells in the hippocampus and amyg-
dala. -amyloid immunoreactivity was mainly observed in pyramidal neurons, as
suggested by the morphological evaluation of the -amyloid-immunoreactive (ir)
cells, and it was mainly localized in the cytoplasm. -amyloid-ir cells were detected
mainly in CA1 and CA3 of the dorsal and ventral hippocampus and in the basolateral
nucleus of the amygdala.
The -amyloid-ir cells were counted in the basolateral nucleus of the amyg-
dala using live microscopy in three non-adjacent sections per animal from bregma
0.82 to 1.82 mm. Areas were delineated as described by the Paxinos and Franklin
mouse brain atlas (2001) with a 5× objective lens (N.A. 0.075) [151] and -amyloid-
positive cells were counted with a 20× objective lens (N.A. 0.75). Volumetric analysis
was performed with the aid of the image analysis software StereoInvestigator (ver-
sion 6.50.1, Microbrightfield, Colchester, VT, USA). The volumes were reported as
and were estimated according to the formula V = A × t
× 1/ssf, where
A = the summed areas of the delineated areas (computed with Stereo Investiga-
tor), t
= the nominal section thickness of 40 m, and ssf = the section sampling
fraction (1/8).
Stereological estimation of the number of -amyloid-ir cells was performed in
the hippocampus and the cell number was estimated in CA1 and CA3 hippocampal
subfields using the optical fractionator method [18,82]. Four non-adjacent sections
for the dorsal (from bregma 1.34 to 2.30 mm) and two for the ventral hippocam-
pus (from bregma 2.92 to 3.64 mm) were analysed with the aid of the image
analysis software StereoInvestigator (version 6.50.1; Microbrightfield, Colchester,
VT). The following sampling parameters were use d: a fixed counting frame measur-
ing 30 m × 30 m; and a sampling grid size of 99.7 m × 55.1 m. The counting
frames were placed randomly by the software at the intersections of the grid within
the outlined structure of interest. The cells were counted following the unbiased
sampling rule [93] using a 20× objective lens (N.A. 0.75) and included in the mea-
surement when they came into focus within the optical dissector (height, 10 m). For
the analysis, the total number of estimated -amyloid-ir neurons per unit volume
was evaluated.
2.2.7. Statistical analysis
All data were analysed by parametric analysis of variance (ANOVA) with geno-
type, sex and running as the between-subject factors. Additional within-subjects
factors (e.g., trials, days) were also included as determined by the nature of the
dependent variables under consideration. Supplementary restricted analyses were
also conducted to assist data interpretation whenever appropriate. Data from startle
reactivity were ln-transformed to better conform to the assumptions of paramet-
ric ANOVA. All statistical analyses were carried out using SPSS
13.0 for Windows
(Release 13.0.1, SPSS Inc. Chicago IL, USA) implemented on a PC running the Microsoft
Windows XP SP2 operating system.
2.3. Results
2.3.1. Assessment of pre-testing wheel running activity
As expected, all mice preferentially engaged in wheel running
exercise in the dark phase (Fig. 2). Running activity was consider-
ably higher in the dark than in the light phase, although the animals
clearly also performed wheel running in the light period as well.
This preference was seen in mice of both genotypes (Fig. 2A) and
both sexes(Fig. 2B). While weeklylevels of activity in the light phase
S. Pietropaolo et al. / Behavioural Brain Research 192 (2008) 42–60 49
Fig. 2. Assessment of wheel running activity. Running activity was recorded as
the number of impulses obtained from four magnets glued on the surface of each
running wheel and detected by a sensor on the ceiling of the cage. Impulses were col-
lected in 30-min bins and analysed in 12-h blocks across 12 pre-testing weeks (data
from 6 out of the 90 pre-testing days were excluded from the analysis because of the
occurrence of sawdust renewal). The patterns of wheel running did not qualitatively
differ between genotypes (A) and sexes (B). All animals performed wheel running
mostly during the dark period, where the activity levels gradually increased to reach
a peak around week 3 and was stable from week 4 onwards (A and B, left panel).
Genotype and sex differences were detected in the amount of wheel running activ-
ity displayed during the dark periods of the first week. Error bars represent ± S.E.M.
*p < 0.05. Data were obtained from: 9 male wild-type; 10 male mutant; 7 female
wild-type; 8 female mutant.
were relatively stable across the observation period, running in the
dark phase underwent 3 weeks of steady increase before stabilizing
from week 4 onwards.
A2× 2 × 2 × 12 (genotype × sex × dark–light phases × weeks)
ANOVA of weekly running activity revealed a significant effect of
dark–light phases [F(1,30) = 11.46, p < 0.0005], and its interaction
Fig. 3. Anxiety-like behaviour in the elevated plus maze. The effect of running on
anxiety-like behaviour differed between male and female mice, but it was indepen-
dent of genotype. Error bars represent ± S.E.M. *p < 0.05. The numbers of subjects
in each experimental group were: male wild-type: 8 sedentary, 9 running; male
mutant: 9 sedentary, 10 running; female wild-type: 10 sedentary, 7 running; female
mutant: 9 sedentary, 8 running.
with weeks [F(11,330) = 7.15, p < 0.001]. No other interactions with
dark–light phases attained statistical significance [all Fs < 1].
Next, the running activity in the two phases was separately
analysed. A 2 × 2 × 12 (genotype × sex × weeks) ANOVA restricted
to the light phase yielded only a significant ef fect of weeks
[F(11,330) = 3.10, p = 0.001], suggesting that there were nonethe-
less significant weekly fluctuations. No other effects of interaction
reached statistical significance.
An equivalent ANOVA restricted to the dark phase also yielded
a significant effect of weeks [F(11, 330) = 6.91, p < 0.0005], which
was further accompanied by a significant interaction with sex
[F(11,330) = 2.38, p < 0.01].
Additional analyses restricted to each week yielded a clear effect
of sex [F(1,30) = 6.21, p < 0.05] only in the first week, and in the
same analysis the main effect of genotype also emerged signifi-
cantly [F(1,30) = 12.46, p = 0.001]. Hence, although there was limited
overall difference in the pattern of running activity across weeks
between the two genotypes [in the dark phase, genotype × weeks:
F(11,330) = 1.10, p = 0.4], the mutant mice were significantly less
engaged in wheel running activity in the initial week of adaptation.
2.3.2. Elevated plus maze
Running exercise produced a clear effect on the expression of
anxiety-like behaviour in the elevated plus maze (Fig. 3). This effect
was however bidirectional depending solely on sex. Female runners
Fig. 4. Locomotor activity in the open field test. Locomotor habituation was observed in all animals, but it was more pronounced in male compared to female mice (A). A
reduction in the overall activity levels was detected in mutant compared to wild-type animals, but only in the female sex (B). Error bars represent ± S.E.M. *p < 0.05. The
numbers of subjects in each experimental group were: male wild-type: 8 sedentary, 9 running; male mutant: 9 sedentary, 10 running; female wild-type: 10 sedentary, 7
running; female mutant: 9 sedentary, 8 running.
50 S. Pietropaolo et al. / Behavioural Brain Research 192 (2008) 42–60
were less anxious than female sedentary controls, whereas male
runners appeared more anxious than male sedentary controls. At
the same time, performance in the four sedentary control groups
(see Fig. 3) appeared relatively comparable.
These impressions were confirmed by a 2 × 2 × 2 (geno-
type × sex × running) ANOVA of the percent entries into the
open arms which yielded a significant sex × running interaction
[F(1,62) = 1.13, p < 0.005]. Separate two-way ANOVAs restricted to
either sex revealed a significant effect of running in females
[F(1,30) = 6.73, p < 0.05], and a tendency in the males [F(1,32) = 3.82,
p = 0.06]. In these restricted analyses, neither the effect of genotype
nor its interactions was significant [all Fs < 1].
Parallel analysis of the alternative anxiety measure of percent
time spent in the open arms yielded a highly similar pattern of
results, although the critical sex × running interaction failed to
reach statistical significance [F(1,62) = 3.27, p = 0.08]. In addition,
analysis of locomotor activity indexed by distance travelled in the
entire maze surface did not yield any significant main effect or
2.3.3. Open field test
All animals displayed locomotor habituation to the open field,
as demonstrated by a decrease in the distance moved across
time bins. The habituation effect appeared more pronounced in
male than in female mice (Fig. 4A), and this was independent
of genotype or running conditions. These impressions were con-
firmed by a 2 × 2 × 2 × 6 (genotype × sex × running × 10-min bins)
ANOVA, which yielde d a significant effect of bins [F(5,310) = 235.82,
p < 0.0005] and a significant sex × bins interaction [F(5,310) = 5.35,
p < 0.001].
The ANOVA also revealed an effect of genotype on the over-
all levels of open field locomotor activity, which was dependent
on sex. Mutant mice were overall less active than wild-type con-
trols, but this effect was preferentially observed in females (Fig. 3B).
This impression was supported by an ANOVA of the total distance
moved which revealed a significant genotype × sex interaction
[F(1,62) = 7.42, p < 0.01]. Restricted analysis further confirmed the
presence of a significant effect of genotype in female [F(1,30) = 7.79,
p < 0.01], but not in male mice [F < 1].
2.3.4. Assessment of the acoustic startle reflex
Mutant mice showed a clear elevation in their acoustic star-
tle reactivity compared to wild-type controls, regardless of sex
or running condition (Fig. 5). The expression of this genotype
effect was dependent on the intensity but not the duration
of the acoustic stimulus. The reactivity profiles of the 3×Tg-
AD and wild-type animals began to diverge at the stimulus
intensity of 85 dB
; thereafter the mutant mice consistently
showed a stronger startle reaction. These conclusions were sup-
portedbya2× 2 × 2 × 2 × 10 (genotype × sex × running × stimulus
duration × stimulus intensity) ANOVA of the (ln-transformed)
reactivity scores, which yielded a significant effect of genotype
[F(1,62) = 85.58, p < 0.0001], of stimulus intensity [F(9,558) = 434.15,
p < 0.0001], and of their interaction [F(9,558) = 55.27, p < 0.0001].
The genotype × running × intensity interaction also achieved sta-
tistical significance [F(9,558) = 2.56, p < 0.01], suggesting that the
intensity-dependent elevation of startle reactivity in the 3×Tg-AD
mice was somewhat modified by exercise. Restricted analyses indi-
cated that in the running condition the reactivity of the mutants
was already higher than controls at the lowest stimulus intensity.
2.3.5. Water maze Cued task (days 1–2). As shown in Fig. 6A, performance
in terms of escape latency clearly improved over days in all ani-
mals. However, 3×Tg-AD mice showed shorter escape latencies
Fig. 5. Acoustic startle reactivity. The intensity of startle reaction (in arbitrary units)
was ln-transformed and expressed here as a function of stimulus intensity, col-
lapsed across the two stimulus-duration conditions of 20 and 40 ms. Mutant mice
of both sexes displayed enhanced startle reactivity compared to wild-type controls
and this phenomenon was not attenuated by running. In the running group the
startle response of mutant animals differed from wild-type controls not only at and
above 85 dB
, but also at the lowest levels of the stimulus intensity. Error bars repre-
sent ± S.E.M. *p < 0.05. Data were obtained from: 18 wild-type (8 male and 10 female)
and 18 mutant (9 male and 9 female) sedentary mice; 16 wild-type (9 male and 7
female) and 18 mutant running mice (10 male and 8 female).
on both days. This genotype effect was equally observed in both
sedentary and running mice of both sexes. A 2 × 2 × 2 × 2 × 4
(genotype × sex × running × days × trials) ANOVA of the escape
latency confirmed the above impressions, revealing a significant
effect of genotype [F(1,62) = 29.25, p < 0.0001], days [F(1,62) = 69.01,
p < 0.0001], and trials [F(3,45) = 13.16, p < 0.01]. Parallel analyses
performed on the dependent measure of path length yielded an
identical pattern of results and statistical outcomes. In addition,
separate analysis of swim speed did not yield any significant effect. Acquisition of spatial reference memory (days 3–9). Days 3–8 (before probe test 1). Performance over
the first 6 days of hidden-platform training (i.e., prior to the
first probe test) is depicted in Fig. 6A and B expressed as
a function of 2-day blocks. This readily suggests that perfor-
mance improved over days. This improvement appeared to be
less marked in the mutants, because they were performing bet-
ter than controls in the first training block (Fig. 6B).A2× 2 × 2 × 3
(genotype × sex × running × blocks) ANOVA of escape latency con-
firmed the above impressions, revealing a significant effect of
blocks [F(2,124) = 14.91, p < 0.0001], and its interaction with geno-
type [F(2,124) = 6.24, p < 0.005].
The impact of exercise on water maze performance was oppo-
site between the two sexes as shown in Fig. 6C. This shows
that in the females exercise reversed the mild genotype-induced
impairment in reference memory acquisition observed in the
sedentary condition. Indeed, this reversal led to enhanced per-
formance in the running female mutants relative to running
female wild type controls. These conclusions were supported
by the emergence of a significant genotype × running interaction
[F(1,62) = 4.56, p < 0.05] in the overall analysis. Although the interac-
tion genotype × sex × running failed to reach statistical significance
[F(1,62) = 3.28, p = 0.08], four separate genotype × blocks analyses
restricted to each of the four (sex × running) conditions were per-
formed to seek additional statistical evaluation of the genotype
effect. In female runners, a significant genotype effect [F(1,13) = 5.27,
p < 0.05] emerged, which was accompanied by a nearly significant
genotype × blocks interaction [F
(2,26) = 2.73, p = 0.06]. In female
sedentary controls, only the genotype × blocks interaction showed
a tentative trend towards significance [F(2,34) = 2.73, p = 0.08]. In
S. Pietropaolo et al. / Behavioural Brain Research 192 (2008) 42–60 51
Fig. 6. Reference memory in the water maze. Cue d task (A). The performance of 3×Tg-AD mice was significantly better compared to that of wild-type controls on both testing
days (A) and this effect was independent of sex and running conditions. Acquisition of a reference memory task (B and C). The first 6 days of training, i.e., before the first
probe test, were separately analysed in 2-day blocks. Performance improved with training and this effect was less pronounced in mutant mice who displayed lower escape
latency already on the first day-block (B). An acquisition impairment was observed in mutant sedentary females, but this was reverted by running (C). Probe tests (D–F).
Three separate 30-s probe tests were performed. The first two were carried out on day 9, 1.5 h before and after the last four trials of training. Probe 3 was performed 24 h after
the last training session. Running improved performance in the mutant mice during the first two probe tests. This phenomenon was however only observed in the females
in Probe 1 (D), while it was detected in mutant mice of both sexes on Probe 2 (E). *p < 0.05. Error bars represent ± S.E.M. The numbers of subjects in each experimental group
were: male wild-type: 8 sedentary, 9 running; male mutant: 9 sedentary, 10 running; female wild-type: 10 sedentary, 7 running; female mutant: 9 sedentary, 8 running.
contrast, no significant effect of genotype or of its interaction with
blocks wasdetected in male mice, either in the sedentary or running
conditions [Fs < 1].
Parallel analyses of path length yielded a highly similar pattern
of results as described above. Separate analysis of swim speed data
did not reveal any group difference. Day 9. The performance of the animals did
not differ on day 9, as suggested by a 2 × 2 × 2 × 4 (geno-
type × sex × running × trials) ANOVA of escape latency revealing
no significant effect of sex, genotype, running or of their interaction
[Fs < 1]. This indicated that performance was largely comparable
amongst groups by this day of training. The mean (±S.E.M.) escape
latency (in sec) for each group was: 6.1 ± 3.8 (male wild-type
sedentary), 4.7 ± 2.7 (male 3×Tg-AD sedentary), 10.4 ± 2.7 (male
wild-type running), 5.7 ± 2.5 (male 3×Tg-AD running), 5.6 ± 2.5
(female wild-type sedentary), 10.7 ± 2.7 (female 3×Tg-AD seden-
tary), 4.5 ± 3.0 (female wild-type running), 3.6 ± 2.8 (female
3×Tg-AD running). Probe tests. Altogether three separate probe tests
were conducted: (i) 1.5 h prior to acquisition training on day 9, (ii)
1.5 h after day 9 acquisition training, and (iii) 24 h after day 9 acqui-
sition training; they were separately analysed by evaluating the
percentage of time spent in the four quadrants. In addition, the per-
centage of time spent in the target quadrant (previously occupied
by the escape platform) was analysed on its own for each probe test
(Fig. 6D–F). In each probe test, we evaluated the number of annular
crossings and the latency to the first annular crossing. An annular
crossing was scored whenever the swim path crossed into the area
previously occupied by the platform.
Analysis of the preference for the target quadrant: As depicted
in Fig. 6D, in the first probe test running enhanced spatial search
in the target quadrant in the female mutant mice in compari-
son to female mutants maintained in the sedentary condition. The
direction of the genotype effect in the females was thus reversed
from a relative impairment in the sedentary condition towards an
enhancement in the running condition. This impression was con-
firmed by the 2 × 2 × 2 × 4 (genotype × sex × running × quadrants)
ANOVA of percent time per quadrant, which yielded a significant
effect of quadrants [F(3,186) = 6.50, p < 0.0001] and of the inter-
action genotype × running × quadrants [F(3,186) = 2.59, p < 0.05].
Furthermore, the 2 × 2 ×
2 (genotype × sex × running) ANOVA of
percent time spent in the target quadrant yielded a significant
interaction genotype × running [F(1,62) = 5.25, p < 0.05]. Although
the interaction sex × genotype × running did not reach statistical
significance [F(1,62) = 1.61, p = 0.21], separate analyses designed to
evaluate the effect of running in each (genotype × sex) group were
carried out. These revealed a significant effect of running only in
mutant females [F(1,15) = 4.97, p < 0.05], and not in the other three
groups [Fs < 1].
As depicted in Fig. 6E, a marked overall preference for the target
quadrant was also observed in the second probe test [main effect
of quadrants: F(3,186) = 11.96, p < 0.0001]. Again, running appeared
52 S. Pietropaolo et al. / Behavioural Brain Research 192 (2008) 42–60
Table 1
Additional evaluation of probe performance indices: annular crossings and latency to the first annular crossing
Sex Genotype Running Annular crossings Latency to the first annular crossing (s)
Probe 1 Probe 2 Probe 3 Probe 1 Probe 2 Probe 3
Sedentary 3.3 ± 0.7 3.0 ± 0.6 3.5 ± 0.7 11.5 ± 2.5 7.8 ± 2.3 9.5 ± 2.4
Running 3.3 ± 0.7 3.0 ± 0.5 3.6 ± 0.7 6.8 ± 2.4 8.4 ± 2.2 6.4 ± 2.3
Sedentary 3.8 ± 0.7 3.0 ± 0.5 3.3 ± 0.7 8.2 ± 2.4 5.9 ± 2.2 8.3 ± 2.3
Running 3.7 ± 0.6 3.9 ± 0.5 2.5 ± 0.6 9.8 ± 2.3 7.5 ± 2.1 9.3 ± 2.2
Sedentary 3.3 ± 0.6 3.5 ± 0.5 3.5 ± 0.6 12.3 ± 2.3 9.8 ± 2.1 11.3 ± 2.2
Running 2.6 ± 0.8 3.0 ± 0.6 3.3 ± 0.8 9.2 ± 2.7 9.1 ± 2.5 5.5 ± 2.6
Sedentary 3.2 ± 0.7 3.3 ± 0.5 3.2 ± 0.7 10.1 ± 2.4 8.0 ± 2.2 9.6 ± 2.3
Running 4.5 ± 0.7 4.9 ± 0.6 5.3 ± 0.7 4.1 ± 2.5 3.9 ± 2.3 2.7 ± 2.5
The analysis of the number of annular crossings and of the latency to the first annular crossing yielded a pattern of results which was highly similar to that obtained from
the evaluation of the percent time spent in the target quadrant (see Fig. 6D–F). Running appeared to selectively improve the performance of female mutants on Probes 1
and 2, although no effect of running, genotype, sex or their interactions reached the levels of statistical significance. Once again, no difference was observed on Probe 3. The
numbers of subjects in each experimental group were: male wild-type: 8 sedentary, 9 running; male mutant: 9 sedentary, 10 running; female wild-type: 10 sedentary, 7
running; female mutant: 9 sedentary, 8 running. Data are expressed as mean ± S.E.M.
to improve performance in the mutant mice, but this effect was
now observed in both male and female mice. However, a 2 × 2 × 2
(genotype × sex × running) ANOVA of percent time spent in the
target quadrant failed to reveal a significant genotype × running
interaction[F(1,62) = 2.70, p = 0.11]. Nonetheless, we conducted sep-
arate analyses designed to evaluate the effect of running in each
(genotype × sex) group as describe d above. These revealed a signif-
icant effect of running only in mutant [male: F(1,17) = 5.19, p < 0.05;
female; F(1,15) = 4.75, p < 0.05], and not in wild-type controls of
either sex [Fs < 1]. This provided some tentative support for a pref-
erential effect of running in the mutant mice, among which those
reared in the sedentary condition were again performing poorly.
As depicted in Fig. 6F, no difference in the animals’
performance was detected between genotype, sex or run-
ning condition by the third probe test. A 2 × 2 × 2 × 4
(genotype × sex × running × quadrants) ANOVA of percent
time per quadrant yielded only a significant effect of
quadrants [F(3,186) = 16.31, p < 0.0001]. No other effects
achieved or approached statistical significance. The 2 × 2 × 2
(sex × genotype × running) ANOVA of the time spent in the target
quadrant did not reveal any significant effect. The same impression
was obtained even when the effect of running was assessed
separately in each (genotype × sex) group as described above.
Analysis of annular crossings and escape latencies: Although the
genotype × sex × running ANOVAs demonstrated that no effect
reached the levels of statistical significance, the results of the anal-
ysis of the number of the annular crossings and of the escape
latency (see Table 1) were highly similar to those obtained from the
evaluation of the percent time spent in the target quadrant. Run-
ning appeared to selectively improve the performance of female
mutants on Probes 1 and 2, while no difference was observed on
Probe 3.
Additional analyses: In addition, the above analyses were
repeated by focusing on the first 15 s of each probe test, and the
outcomes were essentially unchanged. Furthermore, parallel anal-
yses were performed on the dependent measure of percent distance
swum per quadrant and these yielded highly similar outcomes
as described above. Finally, analyses of the swim speed revealed
that mutant mice tended to swim faster than wild-type controls,
but only during probe 2 [genotype effect: F(1,62) = 4.02, p < 0.05].
The mean (±S.E.M.) values of swim speed expressed in cm/s
were: Probe 1: 21.05 ± 0.67 (wild-type), 22.64 ± 0.64 (mutants);
Probe 2: 21.25 ± 0.53 (wild-type), 22.73
± 0.51 (mutants); Probe 3:
21.31 ± 1.74 (wild-type), 22.51 ± 0.72 (mutants).
2.3.6. Spontaneous alternation in the Y-maze
During the sample phase, mice of all groups did not differ in
the exploration of the maze, and they all spent an equal amount of
time in each of the unblocked arms (arms effect and its interactions:
Fs < 1; data not shown).
During the test phase, all animals displayed a clear preference
for the novel arm, but this preference appeared to be less marked
in the sedentary female mutants (Fig. 7). These conclusions were
supported by a 2 × 2 × 2 × 3 (genotype × running × sex × arms)
Fig. 7. Spontaneous alternation in the Y-maze. During the test phase all animals displayed a clear preference for the novel arm, but this preference tended to be weaker in
the sedentary female mutants [genotype effect in sedentary females: p = 0.06]. The numbers of subjects in each experimental group were: male wild-type: 8 sedentary, 9
running; male mutant: 9 sedentary, 10 running; female wild-type: 10 sedentary, 7 running; female mutant: 9 sedentary, 8 running. ST, start arms; NOV, novel arm; FAM,
familiar arm.
S. Pietropaolo et al. / Behavioural Brain Research 192 (2008) 42–60 53
ANOVA of time spent in each arm, revealing a significant arm
effect [F(2,124) = 15.42, p < 0.0001]. Although the interaction geno-
type × running × sex × arms failed to reach statistical significance
[F(2,124) = 2.02, p = 0.14], the separate analysis of the time spent
in the novel arm revealed that the effect of genotype was nearly
significant in sedentar y females [F(1,17) = 4.05, p = 0.06], but it was
far from reaching statistical significance in running females [F < 1],
sedentary males [F < 1] or running males [F(1,17) = 1.13, p = 0.30].
The analysis of the proportion of time spent in the novel arm
(time spent in the novel arm divided by the time spent in the
three arms) yielded a similar pattern of results, with the interac-
tion genotype × running × sex approaching statistical significance
[F(1,62) = 3.39, p = 0.07], thus lending some support for the results
described above.
2.3.7. Assessment of ˇ-amyloid pathology
The brain samples of some subjects had to be excluded from
the analyses due to damage incurred during processing. The
number of subjects included in the final analyses of -amyloid
immunoreactivity in the dorsal hippocampus (dHPC), ventral
(vHPC) hippocampus, and basolateral amygdala (BLA) was as fol-
lows: male sedentary: 9 dHPC, 6 vHPC, 7 BLA; male running: 7 dHPC,
Fig. 8. Quantification of -amyloid pathology in 3×Tg-AD mice. As expected, -
amyloid immunoreactivity was extensively observed in the hippocampus of mutant
mice. In both sedentary and running conditions, the -amyloid pathology appeared
to be more pronounced in female compared to male mutants. This sex difference
was more apparent in the dorsal hippocampus (dHPC), where it was observed in
both CA1 (A) and CA3 (B) subfields. In the ventral hippocampus (vHPC) a significant
sex difference was detected in CA3 (D), but not in CA1 (C). The number of subjects
included in each brain analysis was: dHPC: 16 male (9 sedentary and 7 running), and
14 female (7 sedentary and 7 running); vHPC: 13 male (6 sedentary and 7 running),
13 female (6 sedentary and 7 running). Error bars represent ± S.E.M. *p < 0.05.
7 vHPC, 10 BLA; female sedentary: 7 dHPC, 6 vHPC, 8 BLA; female
running: 7 dHPC, 7 vHPC, 8 BLA.
In both sedentary and running conditions, the -amyloid
pathology in the hippocampus appeared to be more pronounced
in the female compared to the male mutants (Fig. 8). This sex dif-
ference was more pronounced in the dorsal hippocampus where it
was observed in both CA1 [sex effect: F(1,26) = 4.86, p < 0.05; Fig. 8A]
and CA3 [sex effect: F(1,26) = 6.01, p < 0.05; Fig. 8B]. In the ven-
tral hippocampus, a similar sex difference was detected, but this
only reached statistical significance in CA3 [F(1,22) = 7.75, p < 0.05;
Fig. 8D], not in CA1 [F <1;Fig. 8C].
No differences were detected in the basolateral amygdala
between sexes [F(1,29) = 1.29 p = 0.27] or between running con-
ditions [Fs < 1 for main effect and the interaction]. The mean
(±S.E.M.) values of -amyloid-ir cells/mm
for each group
were: 24,057.60 ± 2364.42 in male sedentary, 23,688.59 ± 1978.22
in male running, 26,088.23 ± 2211.72 in female sedentary,
26,639.67 ± 2211.72 in female running.
Representative sections are illustrated in Fig. 9 to illustrate the
results obtained from the quantitative analysis. Hippocampal -
amyloid pathology appeared to be more pronounced in female
compared to male mice in the dorsal CA1 (9A) and CA3 (9B)
subfields. A similar sex difference was observed in the ventral hip-
pocampus, although it was more readily detectable in CA3 (9D) than
in CA1 (9C).
3. Discussion
The present study demonstrated clear sex differences in the
behavioural and brain phenotypes of the 3×Tg-AD mouse line.
Furthermore, our results showed that home cage voluntary exer-
cise was effective in modifying behaviour and cognition, but these
effects were dependent on the animals’ sex and genotype. In the
following sections, the full implications of the present data set are
discussed and its relevance to the general context of neural plastic-
ity examined.
3.1. The behavioural phenotype of the 3×Tg-AD mouse line: sex
First, the present study confirmed and extended the results
obtained from our previous behavioural investigation in the
3×Tg-AD model [155]. It replicated the presence of pronounced
non-cognitive abnormalities in the cued water maze task and
startle reaction tests in male mutant mice, and the absence of a
clear deficit in spatial memory. Furthermore, a similar behavioural
phenotype was observed in female 3×Tg-AD animals, but it also
included hypoactivity in the open field and a slight deficit in the
acquisition of the water maze task. This behavioural phenotype
was accompanied by the expected AD-like brain pathology, with
extensive deposition of intracellular -amyloid in hippocampal and
amygdalar neurons. Moreover, hippocampal -amyloid pathology
appeared to be more severe in female compared to male mutants.
The presence of more severe cognitive deficits in female 3×Tg-
AD mice is in line with previous results from Clinton et al. [25].
These authors attributed the memory deficits to the presence of
corticosterone response to water maze training observed in female
mutants only. However, these differences were only observed at
9 months, but not at 6 months of age. Furthermore, it is possible
that the stress response of female mutants may be a conse-
quence rather than the cause of the more severe impairment
observed in the water maze performance. This interpretation is
supported by the present data from the Y-maze, since mild sex-
specific cognitive deficits were detected also in this task which
54 S. Pietropaolo et al. / Behavioural Brain Research 192 (2008) 42–60
Fig. 9. -amyloid pathology in hippocampal areas. As suggested by the quantitative evaluation of -amyloid immunoreactivity, hippocampal -amyloid pathology appeared
to be more pronounced in female compared to male mice. This sex difference was observed in the dorsal CA1 (A) and CA3 (B) areas. A similar sex difference was also observed
in the ventral hippocampus, although it was mainly detected in CA3 (D), while it was weak in CA1 (C). Microphotographs were obtained from one male and one female
non-running mutants. Scale bar = 100 m. SO, stratum oriens; SP, stratum pyramidale; SR, stratum radiatum; SL, stratum lucidum.
does not involve the stressful experience of exposure to the
Furthermore, it can be hypothesized that the more pronounced
hippocampal -amyloid pathology observed in female mutants
may contribute to the emergence of the water maze deficits. The
more severe brain pathology of female animals is in agreement with
previous data from the same [144] and other genetic AD-mouse
models [20,175,201]. However, Clinton et al. [25] reported no sex
differences in the -amyloid pathology of 3×Tg-AD mice. Yet, the
employment here of a more sensitive method for the quantification
S. Pietropaolo et al. / Behavioural Brain Research 192 (2008) 42–60 55
of the -amyloid-ir cells, viz., stereological counting, may account
for the discrepancy between our results and theirs. The sex differ-
ences observed in this study in the severity of the brain pathology
are particularly relevant since epidemiological studies indicate that
women are more susceptible to AD than men [61,66]. It has been
suggested that the underlying mechanism for this sex difference
may involve the protective effects of testosterone: testosterone
reportedly reduced the -amyloid production [78] and more severe
AD pathology was observed in gonadectomised male 3×Tg-AD mice
The apparent lack of deficits in water maze learning in the male
3×Tg-AD contrasts previous reports on the same mouse line at
this age [10,11,19,25,69,147]. This discrepancy might be attributed
to some key procedural differences. These include differences in
the temperature of the water maze and in the testing procedures,
i.e., whether the visual cued task was performed before acquisi-
tion training and whether water maze testing was conducted in the
light or dark phase. Of course, the delineation of the precise rela-
tive contributions of these various test parameters would certainly
require additional experiments, but the fact that these variables
can affect water maze performance (even in control wild-type ani-
mals) is well acknowledged [185,200]. Here, the water temperature
of was about 25
C, which is common in mouse water maze exper-
iments [22,35,204,208], but is lower compared to the one used in
most of the previous studies with 3×Tg-AD mice which is 26–29
[10,11,19,25,69,147]. In the present study, we also adopted the more
conventional approach of performing the cued task at the begin-
ning [22,200], and of testing the animals during the active (dark)
phase of the cycle [208].
Such procedural dif ferences may also contribute to the discrep-
ancy between the present and previous results in the cued water
maze: no differences between wild-type and 3×Tg-AD mice have
been described previously by others [10,11,19,25,147], while a better
performance was observed here in mutant mice of both sexes. The
relative order of the cued task vs. the spatial training (with hidden
platform) might especially affect the outcome of the water maze
experiment. While we adopted the more conventional approach in
performing the cued task in the beginning [22,200], previous stud-
ies had opted to conduct the cued task 24 h after acquisition training
The importance of this procedural difference was highlighted
by a comparison between two studies in the Tg2576 mouse model.
King et al. [106] reported a deficit while Westerman et al. [203]
failed to find any difference on the cued task [106,203]. In mice,
Vorhees and Williams [200] also suggested that the cued task
had helped to minimize the tendency to search for alternative
escape routes, thus improving the chances of successful learning in
the hidden-platform task [22,200]. Critically, the cued task allows
one to unmask potential differences in relevant sensori-motor and
motivational factors that may affect water maze performance. Had
this been conducted at the end, either the presence or the absence
of an effect would not have been easily interpretable.
The lack of cognitive deficits in male mutants was observed here
not only in the water maze test, but also in another hippocampus-
dependent task, namely, the Y-maze [173]. This is suggestive of
intact hippocampal function despite the presence of -amyloid
pathology. The lack of cognitive deficits may represent an important
limitation of the 3×Tg-AD mouse model, because of the obvi-
ous relevance of these symptoms to the mnemonic dysfunction
characteristic of AD patients. Yet, the expression of AD-like brain
pathology is not invariably accompanied by cognitive deficits in
other mouse models [90,106,158]; (see also for a review [48])
as well as in humans [42,101,184]. In contrast to the absence
of a pronounced cognitive phenotype, the 3×Tg-AD mice model
showed pronounced non-cognitive symptoms in the cued water
maze task and startle reaction tests which are highly relevant to
AD–resembling the enhanced responsiveness to aversive stimuli
reported in human patients [121,178].
3.2. The behavioural effects of physical exercise in the 3×Tg-AD
mouse line
Our data here demonstrate that the effects of wheel running
exercise on behavioural dysfunction in the 3×Tg-AD mouse model
are largely limited to the cognitive domain, and are critically depen-
dent on sex.
First, wheel running attenuated the memory deficits observed
in female 3×Tg-AD mice, without affecting the performance in the
female controls. Although a similar observation has been reported
in another transgenic AD-mouse line [145], the lack of effects of
wheel running on spatial memory in wild-type animals was some-
what unexpected according to previous studies (see Section 1.3.1). It
is possible that the specific hybrid genetic background of our wild-
type mice may be less sensitive to the effects of exercise than other
pure wild-type strains. Our data may also suggest that the positive
effect of wheel running on cognition can be more readily demon-
strated against the background of AD-like brain pathology. One
suggestion that may predict this outcome is that exercise specif-
ically attenuates AD-like pathology, for example by promoting
amyloid degradation, as observed following environmental enrich-
ment [118]. Yet, our data did not lend support to this hypothesis.
Exercise did not reduce the number of -amyloid-ir cells in the
hippocampus or amygdala, i.e., it did not affect one of the major
hallmarks of AD pathology that has been proposed as a key causal
element of the disease [88]. Interestingly, previous studies have
reported that the effects of environmental manipulations on the
cognitive deficits of transgenic AD mice may not require a reduction
in the -amyloid pathology. Jankowsky et al. [94,214] demon-
strated that environmental enrichment improved the water maze
performance in AD mice and concomitantly (and paradoxically)
increased -amyloid load. However, it is still possible that exercise
may indirectly affect the severity of AD-brain pathology through
other mechanisms, e.g., reduced inflammation or elevation of BDNF
levels. Further investigations are warranted to specifically address
these issues that are highly relevant to the -amyloid hypothesis
of AD [186].
In contrast to what was observed in the water maze, running
affected anxiety levels in both mutant and wild-type animals,
suggesting that it may act on emotionality through different mech-
anisms other than attenuating AD-related pathology. The effects of
exercise on anxiety may be mediated by enhanced neurogenesis or
BDNF expression, as mentioned in Section 1.3.1. Overall, our data
suggest a stronger impact of physical exercise in the female sex,
as also seen with another form of home cage activity such as grid
climbing [156].
Wheel running did not affect the non-cognitive symptoms dis-
played by mutant animals of both sexes. One possible speculative
explanation for this null result is that exercise is not able to
attenuate the AD-brain pathology underlying the non-cognitive
symptoms, either because of its severity or b ecause of its specific
neurobiological bases. It is also possible that the exercise expo-
sure in our study was applied too late to counter the emergence of
non-cognitive symptoms. At 3 months of age, i.e., when exposure
to exercise was initiated in the present study, 3×Tg-AD mice have
been found to be free from any cognitive alteration [10], but little
is known about the presence or absence of non-cognitive pheno-
types at this age [69]. Hence, further investigations are necessary to
assess whether the alterations in acoustic startle reactivity or per-
formance in the cued task version of the water maze are already
present in the 3×Tg-AD mice at 3 months of age.
56 S. Pietropaolo et al. / Behavioural Brain Research 192 (2008) 42–60
3.3. Does the amount of wheel running activity matter?
Is it possible that differences in the amount of wheel running
activity explain the sex- and genotype-dependent effects of exer-
cise on water maze performance? Here, we reported that female
mice initially displayed higher running levels compared to males,
as commonly described (see Section 1.2.2). Yet, this phenomenon
disappeared around week 4 and the levels of running activity during
the last 8 weeks were highly similar between sexes. It is therefore
unlikely that the higher responsiveness of females to exercise here
was due to their higher levels of running activity. Interestingly, the
sex-specific differences in the amount of wheel running were not
fully confirmed by the results obtained in the open field test, in
which female mice showed similar initial levels of locomotor activ-
ity, but displayed slower habituation. Hence, our data do not allow
us to draw a clear conclusion about the relationship between wheel
running and open field activity, a controversial issue that has been
discussed in Sections 1.2.3 and 1.2.4.
Differences in the amount of wheel running activity also cannot
readily explain the preferential efficacy of exercise in affecting cog-
nition in the 3×Tg-AD mice. The overall amount of wheel running
and the circadian rhythm of mutant animals were highly compa-
rable to wild-type controls, although their running activity levels
were initially lower. Nonetheless, this initial difference between
3×Tg-AD mice and wild-type controls suggests that the quantifi-
cation of wheel running activity for 1–3 weeks may represent a
sensitive measure for the phenotypic characterization of other AD-
mouse lines [38].
3.4. Beyond physical exercise: the psychological relevance of
wheel running
It has been suggested that the protective effects of the home
cage availability of a running wheel on cognitive function may not
be entirely attributable to the running activity itself [180]. Learn-
ing for example, which is known to improve hippocampal function
[114,195,205] could play a major role in mediating the effects of
wheel running exposure.Indeed, the presence of “a learning phase”,
lasting over weeks, that is necessary to stabilize wheel running
levels in both rats and mice (see Section 1.2.1) supports the view
that wheel running is a spontaneous yet acquired behaviour thus
implying some forms of training or practice. Play could also repre-
sent an important factor in explaining the effects of wheel running
exercise, since it is commonly acknowledged that rodents actively
exploit a running wheel in a manner that may resemble certain
play-like activities, e.g., clinging to the wheel and allowing them-
selves to be repeatedly carried around it for several revolutions [50].
Both learning and play are intrinsic properties of home cage wheel
running that cannot be easily disentangled. In contrast, it is pos-
sible to exclude that the effects of wheel running activity are due
to the mere presence of the wheel, i.e., as a minor form of envi-
ronmental enrichment. Although cognitive enhancing effects have
been demonstrated in mice housed with a locked wheel compared
to no-wheel controls [145], behavioural and brain differences have
also been describ ed when running animals were compared with
locked-wheel controls [73,91,139,154]. The use of the locked wheel
may therefore constitute a more appropriate and stringent strategy
to identify the unique effects o f physical exercise by controlling for
the potentially “enrichment-like” effects of wheel running [73].
3.5. Wheel running and standard housing conditions for
laboratory rodents
There is another methodological-theoretical issue that concerns
the sedentary controls employed in animal models of physical exer-
cise. In contrast to humans, sedentary rodents are typically housed
in conditions that constrain spontaneous activities. Hence, it has
been hypothesised that behavioural differences between running
and sedentary rodents may reflect the abnormal status of seden-
tary controls rather than enhanced abilities in the running animals
[46]. It should be noted that this potential problem of abnor-
mally “impoverished” controls is not unique to exercise research;
its relevance has been pointed out already with other housing
manipulations, e.g., environmental enrichment [162]. Nonethe-
less, this interpretative problem of the standard controls does
not undermine our capability of investigating the neurobiolog-
ical mechanisms through which environmental manipulations
exert their effects on behaviour. Furthermore, this interpretative
issue of the sedentary controls in exercise research may turn
out to be of practical advantage when translated to the mod-
ern Homo sapiens. According to some authors, the amount of
physical activity humans get in today’s age is way below the lev-
els of physical exertion that they are genetically predisposed to
sustain [28], so that the normal levels of activity of most mod-
ern humans should actually be considered “sub-optimal” [198]
as well as those of captive rodents kept in standard laboratory
According to some authors, the impoverished environment pro-
vided by standard caging conditions could be responsible not only
for the observed effects, but also for the expression of wheel run-
ning activity per se. In other words, wheel running would be a
result of the captive environment that animals are forced to inhabit
[103,167], as well as several other stereotypic behaviours.A rebound
effect similar to that observed following prevention of stereotypic
activities [128] has indeed been observed when access to a wheel
was reinstated af ter a period of deprivation. Hence, running often
occurs in bursts of much greater intensity than prior to depriva-
tion [123]. Yet, wheel running has also b een described in complex
and semi-natural environments in both rats [162] and mice [179],
thus suggesting that it is not entirely an artefact of impoverished
standard conditions.
3.6. Conclusions
Wheel running activity is a spontaneously acquired form of
physical (or perhaps also psychological) exercise that is robustly
and consistently expressed by laboratory rodents. Compared to
other commonly employed environmental manipulations (e.g.,
enrichment or social isolation procedures), home cage voluntary
wheel running enjoys a unique advantage in being able to provide
an easily quantifiable measure to index the degree of engagement in
individual subjects. Hence, it represents a useful tool to investigate
neurobehavioural plasticity in rodents, though some methodolog-
ical issues such as its quantification, the choice of appropriate
controls and the inclusion of subjects of both sexes must be care-
fully considered in the experimental design and data interpretation.
As illustrated by the experiments presented here, the recent avail-
ability of genetically-modified mouse lines allows us to employ the
wheel running paradigm to investigate possible gene–environment
interactions and their sex dependency relevant to the aeti-
ology and pathogenesis of neuropsychiatric and degenerative
disorders. Yet, further investigations are warranted to help eluci-
dating more precisely the neural and psychological mechanisms
underlying the effects of wheel running on genetically modified
The present study was supported by a Swiss National Sci-
ence Foundation (SNF) 3100A0-100309/1 grant to Joram Feldon,
S. Pietropaolo et al. / Behavioural Brain Research 192 (2008) 42–60 57
with additional support by the Swiss Federal Institute of Technol-
ogy and the National Center for Competence in Research (NCCR):
Neural Plasticity and Repair. The authors thank Prof. Frank M.
LaFerla (University of California, Irvine, USA) for kindly provid-
ing the animals to generate the subjects of the present study. The
authors thank Elisabeth Weber for assisting in the processing of
brain samples, Anita B
uttiker for her assistance in experimentation
and data analysis and Peter Schmid for his technical assistance,
especially in the quantification of wheel running activity. The
authors are grateful to the animal technicians for their caring of
the animals, and Dr Frank Bootz for his veterinary expertise and
[1] Adlard PA, Perreau VM, Engesser-Cesar C, Cotman CW. The timecourse of
induction of brain-derived neurotrophic factor mRNA and protein in the rat
hippocampus following voluntary exercise. Neurosci Lett 2004;363:43–8.
[2] Adlard PA, Perreau VM, Pop V, Cotman CW. Voluntary exercise decreases
amyloid load in a transgenic model of Alzheimer’s disease. J Neurosci
[3] Anantharaman-Barr HG, Decombaz J. The effect of wheel running and
the estrous cycle on energy expenditure in female rats. Physiol Behav
[4] Ang ET, Wong PT, Moochhala S, Ng YK. Neuroprotection associated with
running: is it a result of increased endogenous neurotrophic factors? Neu-
roscience 2003;118:335–45.
[5] Arendash GW, Garcia MF, Costa DA, Cracchiolo JR, Wefes IM, Potter H. Environ-
mental enrichment improves cognition in aged Alzheimer’s transgenic mice
despite stable beta-amyloid deposition. Neuroreport 2004;15:1751–4.
[6] Babyak M, Blumenthal JA, Herman S, Khatri P, Doraiswamy M, Moore K, et al.
Exercise treatment for major depression: maintenance of therapeutic benefit
at 10 months. Psychosom Med 2000;62:633–8.
[7] Barnes CA, Forster MJ, Fleshner M, Ahanotu EN, Laudenslager ML, Mazzeo
RS, et al. Exercise does not modify spatial memory, brain autoimmunity, or
antibody response in aged F-344 rats. Neurobiol Aging 1991;12:47–53.
[8] Berchtold NC, Kesslak JP, Pike CJ, Adlard PA, Cotman CW. Estrogen and exer-
cise interact to regulate brain-derived neurotrophic factor mRNA and protein
expression in the hippocampus. Eur J Neurosci 2001;14:1992–2002.
[9] Billat VL, Mouisel E, Roblot N, Melki J. Inter- and intrastrain variation in mouse
critical running speed. J Appl Physiol 2005;98:1258–63.
[10] Billings LM, Oddo S, Green KN, McGaugh JL, LaFerla FM. Intraneuronal Abeta
causes the onset of early Alzheimer’s disease-related cognitive deficits in
transgenic mice. Neuron 2005;45:675–88.
[11] Billings LM, Green KN, McGaugh JL, LaFerla FM. Learning decreases A beta*56
and tau pathology and ameliorates behavioral decline in 3×Tg-AD mice. J
Neurosci 2007;27:751–61.
[12] Binder E, Droste SK, Ohl F, Reul JM. Regular voluntary exercise reduces
anxiety-related behaviour and impulsiveness in mice. Behav Brain Res
[13] Blustein JE, McLaughlin M, Hoffman JR. Exercise effects stress-induced anal-
gesia and spatial learning in rats. Physiol Behav 2006;89:582–6.
[14] Bouchard C, Chagnon M, Thibault MC, Boulay MR, Marcotte M, Cote C, et al.
Muscle genetic variants and relationship with performance and trainability.
Med Sci Sports Exerc 1989;21:71–7.
[15] Bremner JD. Does stress damage the brain? Biol Psychiatry 1999;45:797–805.
[16] Bronikowski AM, Carter PA, Swallow JG, Girard IA, Rhodes JS, Garland Jr T.
Open-field behavior of house mice selectively bred for high voluntary wheel
running. Behav Genet 2001;31:309–16.
[17] Brown J, Cooper-Kuhn CM, Kempermann G, Van Praag H, Winkler J, Gage FH,
et al. Enriched environment and physical activity stimulate hippocampal but
not olfactory bulb neurogenesis. Eur J Neurosci 2003;17:2042–6.
[18] Buckmaster PS, Jongen-Relo AL. Highly specific neuron loss preserves lateral
inhibitory circuits in the dentate gyrus of kainate-induced epileptic rats. J
Neurosci 1999;19:9519–29.
[19] Caccamo A, Oddo S, Billings LM, Green KN, Martinez-Coria H, Fisher A, et al.
M1 receptors play a central role in modulating AD-like pathology in transgenic
mice. Neuron 2006;49:671–82.
[20] Callahan MJ, Lipinski WJ, Bian F, Durham RA, Pack A, Walker LC. Augmented
senile plaque load in aged female beta-amyloid precursor protein-transgenic
mice. Am J Pathol 2001;158:1173–7.
[21] Castren E. Neurotrophins as mediators of drug effects on mood, addiction, and
neuroprotection. Mol Neurobiol 2004;29:289–302.
[22] Chen G, Chen KS, Knox J, Inglis J, Bernard A, Martin SJ, et al. A learning deficit
related to age and beta-amyloid plaques in a mouse model of Alzheimer’s
disease. Nature 2000;408:975–9.
[23] Cheng B, Mattson MP. NT-3 and BDNF protect CNS neurons against
metabolic/excitotoxic insults. Brain Res 1994;640:56–67.
[24] Chorpita BF, Barlow DH. The development of anxiety: the role of control in the
early environment. Psychol Bull 1998;124:3–21.
[25] Clinton LK, Billings LM, Green KN, Caccamo A, Ngo J, Oddo S, et al. Age-
dependent sexual dimorphism in cognition and stress response in the
3×Tg-AD mice. Neurobiol Dis 2007;28:76–82.
[26] Collier G, Hirsch E. Reinforcing properties of spontaneous activity in the rat. J
Comp Physiol Psychol 1971;77:155–60.
[27] Collier GH, Johnson DF, Cybulski KA, McHa
oe CA. Activity patterns in rats
(Rattus norvegicus) as a function of the cost of access to four resources. J Comp
Psychol 1989;104:53–65.
[28] Cordain L, Gotshall RW, Eaton SB, Eaton 3rd SB. Physical activity, energy
expenditure and fitness: an evolutionary perspective. Int J Sports Med
[29] Cornish ER, Mrosovsky N. Activity during food deprivation and satiation of six
species of rodent. Anim Behav 1965;13:242–8.
[30] Cotman CW, Berchtold NC. Exercise: a behavioral intervention to enhance
brain health and plasticity. Trends Neurosci 2002;25:295–301.
[31] Cotman CW, Engesser-Cesar C. Exercise enhances and protects brain function.
Exerc Sport Sci Rev 2002;30:75–9.
[32] Cui L, Hofer T, Rani A, Leeuwenburgh C, Foster TC. Comparison of lifelong and
late life exercise on oxidative stress in the cerebellum. Neurobiol Aging 2007.
[33] Cummings JL. Cognitive and behavioral heterogeneity in Alzheimer’s disease:
seeking the neurobiological basis. Neurobiol Aging 2000;21:845–61.
[34] Czurko A, Hirase H, Csicsvari J, Buzsaki G. Sustained activation of hippocam-
pal pyramidal cells by ‘space clamping’ in a running wheel. Eur J Neurosci
[35] D’Hooge R, De Deyn PP. Applications of the Morris water maze in the study of
learning and memory. Brain Res Brain Res Rev 2001;36:60–90.
[36] Dawson KA, Crowne DP, Richardson CM, Anderson E. Effects of age on noctur-
nal activity rhythms in rats. Prog Clin Biol Res 1987;227B:107–10.
[37] De Bono JP, Adlam D, Paterson DJ, Channon KM. Novel quantitative phenotypes
of exercise training in mouse models. Am J Physiol Regul Integr Comp Physiol
[38] de Visser L, van den Bos R, Spruijt BM. Automated home cage observations
as a tool to measure the effects of wheel running on cage floor locomotion.
Behav Brain Res 2005;160:382–8.
[39] DeFries JC, Hegmann JP, Halcomb RA. Response to 20 generations of selection
for open-field activity in mice. Behav Biol 1974;11:481–95.
[40] Deshmukh VD, Deshmukh SV. Stress-adaptation failure hypothesis of
Alzheimer’s disease. Med Hypotheses 1990;32:293–5.
[41] Diamond MC. Response of the brain to enrichment. An Acad Bras Cienc
[42] Dickson DW, Crystal HA, Mattiace LA, Masur DM, Blau AD, Davies P, et al.
Identification of normal and pathological aging in prospectively studied non-
demented elderly humans. Neurobiol Aging 1992;13:179–89.
[43] Dimeo F, Bauer M, Varahram I, Proest G, Halter U. Benefits from aerobic
exercise in patients with major depression: a pilot study. Br J Sports Med
[44] Ding Q, Vaynman S, Akhavan M, Ying Z, Gomez-Pinilla F. Insulin-like growth
factor I interfaces with brain-derived neurotrophic factor-mediated synaptic
plasticity to modulate aspects of exercise-induced cognitive function. Neuro-
science 2006;140:823–33.
[45] Dishman RK, Renner KJ, Youngstedt SD, Reigle TG, Bunnell BN, Burke KA, et al.
Activity wheel running reduces escape latency and alters brain monoamine
levels after footshock. Brain Res Bull 1997;42:399–406.
[46] Dishman RK, Berthoud HR, Booth FW, Cotman CW, Edgerton VR, Fleshner MR,
et al. Neurobiology of exercise. Obesity (Silver Spring) 2006;14:345–56.
[47] Dlugosz LJ, Hocter WJ, Kaiser KS, Knoke JD, Heller JM, Hamid NA, et al. Risk fac-
tors for mental disorder hospitalization after the Persian Gulf War: U.S. Armed
Forces, June 1, 1991–September 30, 1993. J Clin Epidemiol 1999;52:1267–78.
[48] Dodart JC, Mathis C, Bales KR, Paul SM. Does my mouse have Alzheimer’s
disease? Genes Brain Behav 2002;1:142–55.
[49] Dong H, Goico B, Martin M, Csernansky CA, Bertchume A, Csernansky JG.
Modulation of hippocampal cell proliferation, memory, and amyloid plaque
deposition in APPsw (Tg2576) mutant mice by isolation stress. Neuroscience
[50] Drickamer LC. Chemosignals and activity of wild stock house mice, with a
note on the use of running wheels to assess activity in rodents. Behav Proc
[51] Duman RS, Nakagawa S, Malberg J. Regulation of adult neurogenesis by antide-
pressant treatment. Neuropsychopharmacology 2001;25:836–44.
[52] Eikelboom R, Mills R. A microanalysis of wheel running in male and female
rats. Physiol Behav 1988;43:625–30.
[53] Eisenstein SA, Holmes PV. Chronic and voluntary exercise enhances learning of
conditioned place preference to morphine in rats. Pharmacol Biochem Behav
[54] Engesser-Cesar C, Anderson AJ, Basso DM, Edgerton VR, Cotman CW. Volun-
tary wheel running improves recovery from a moderate spinal cord injury. J
Neurotrauma 2005;22:157–71.
[55] Engesser-Cesar C, Ichiyama RM, Nefas AL, Hill MA, Edgerton VR, Cotman CW, et
al. Wheel running following spinal cord injury improves locomotor recovery
and stimulates serotonergic fiber growth. Eur J Neurosci 2007;25:1931–9.
[56] Eriksen JL, Janus CG. Plaques, tangles, and memory loss in mouse models of
neurodegeneration. Behav Genet 2007;37:79–100.
[57] Ernst C, Olson AK, Pinel JP, Lam RW, Christie BR. Antidepressant effects of exer-
cise: evidence for an adult-neurogenesis hypothesis? J Psychiatry Neurosci
58 S. Pietropaolo et al. / Behavioural Brain Research 192 (2008) 42–60
[58] Farmer J, Zhao X, van Praag H, Wodtke K, Gage FH, Christie BR. Effects of volun-
tary exercise on synaptic plasticity and gene expression in the dentate gyrus
of adult male Sprague-Dawley rats in vivo. Neuroscience 2004;124:71–9.
[59] Festing MF, Greenwood R. Home cage wheel activity recording in mice. Lab
Anim 1976;10:81–5.
[60] Festing MF. Wheel activity in 26 strains of mouse. Lab Anim 1977;11:257–8.
[61] Fratiglioni L, Viitanen M, von Strauss E, Tontodonati V, Herlitz A, Win-
blad B. Very old women at highest risk of dementia and Alzheimer’s
disease: incidence data from the Kungsholmen Project, Stockholm. Neurology
[62] Fratiglioni L, Paillard-Borg S, Winblad B. An active and socially inte-
grated lifestyle in late life might protect against dementia. Lancet Neurol
[63] Friedland RP, Fritsch T, Smyth KA, Koss E, Lerner AJ, Chen CH, et al. Patients
with Alzheimer’s disease have reduced activities in midlife compared with
healthy control-group members. Proc Natl Acad Sci USA 2001;98:3440–5.
[64] Fritschy JM, Weinmann O, Wenzel A, Benke D. Synapse-specific localization of
NMDA and GABA(A) receptor subunits revealed by antigen-retrieval immuno-
histochemistry. J Comp Neurol 1998;390:194–210.
[65] Fuller JL. Ef fects of the albino gene upon behaviour of mice. Anim Behav
[66] Gao S, Hendrie HC, Hall KS, Hui S. The relationships between age, sex, and
the incidence of dementia and Alzheimer disease: a meta-analysis. Arch Gen
Psychiatry 1998;55:809–15.
[67] Genn RF, Tucci SA, Thomas A, Edwards JE, File SE. Age-associated sex differ-
ences in response to food deprivation in two animal tests of anxiety. Neurosci
Biobehav Rev 2003;27:155–61.
[68] Geyer MA, Ellenbroek B. Animal behavior models of the mechanisms under-
lying antipsychotic atypicality. Prog Neuropsychopharmacol Biol Psychiatry
[69] Gimenez-Llort L, Blazquez G, Canete T, Johansson B, Oddo S, Tobena A, et al.
Modeling behavioral and neuronal symptoms of Alzheimer’s disease in mice:
a role for intraneuronal amyloid. Neurosci Biobehav Rev 2007;31:125–47.
[70] Girard I, McAleer MW, Rhodes JS, Garland Jr T. Selection for high voluntary
wheel running increases speed and intermittency in house mice (Mus domes-
ticus). J Exp Biol 2001;204:4311–20.
[71] Girard I, Garland Jr T. Plasma corticosterone response to acute and chronic
voluntary exercise in female house mice. J Appl Physiol 2002;92:1553–61.
[72] Girard I, Swallow JG, Carter PA, Koteja P, Rhodes JS, Garland T. Maternal-care
behavior and life-history traits in house mice (Mus domesticus) artifi-
cially selected for high voluntary wheel running activity. Behav Processes
[73] Gobbo OL, O’Mara SM. Exercise, but not environmental enrichment, improves
learning after kainic acid-induced hippocampal neurodegeneration in associ-
ation with an increase in brain-derived neurotrophic factor. Behav Brain Res
[74] Goff ML, Finger FW. Activity rhythms and adiurnal light–dark control. Science
[75] Gomez-Pinilla F, Dao L, So V. Physical exercise induces FGF-2 and its mRNA in
the hippocampus. Brain Res 1997;764:1–8.
[76] Gomez-Pinilla F, So V, Kesslak JP. Spatial learning and physical activity con-
tribute to the induction of fibroblast growth factor: neural substrates for
increased cognition associated with exercise. Neuroscience 1998;85:53–61.
[77] Goto S, Radak Z, Nyakas C, Chung HY, Naito H, Takahashi R, et al. Regular
exercise: an effective means to reduce oxidative stress in old rats. Ann N Y
Acad Sci 2004;1019:471–4.
[78] Gouras GK, Xu H, Gross RS, Greenfield JP, Hai B, Wang R, et al. Testosterone
reduces neuronal secretion of Alzheimer’s beta-amyloid peptides. Proc Natl
Acad Sci USA 2000;97:1202–5.
[79] Greenwood BN, Foley TE, Day HE, Campisi J, Hammack SH, Campeau S, et al.
Freewheel running prevents learned helplessness/behavioral depression: role
of dorsal raphe serotonergic neurons. J Neurosci 2003;23:2889–98.
[80] Greenwood BN, Strong PV, Dorey AA, Fleshner M. Therapeutic effects of exer-
cise: wheel running reverses stress-induced interference with shuttle box
escape. Behav Neurosci 2007;121:992–1000.
[81] Griesbach GS, Hovda DA, Molteni R, Wu A, Gomez-Pinilla F. Voluntary exercise
following traumatic brain injury: brain-derived neurotrophic factor upregu-
lation and recovery of function. Neuroscience 2004;125:129–39.
[82] Gundersen HJ, Bagger P, Bendtsen TF, Evans SM, Korbo L, Marcussen N, et al.
The new stereological tools: disector, fractionator, nucleator and point sam-
pled intercepts and their use in pathological research and diagnosis. APMIS
[83] Guo M, Wu CF, Liu W, Yang JY, Chen D. Sex difference in psychological
behavior changes induced by long-term social isolation in mice. Prog Neu-
ropsychopharmacol Biol Psychiatry 2004;28:115–21.
[84] Hagberg JM, Moore GE, Ferrell RE. Specific genetic markers of endurance per-
formance and VO
. Exerc Sport Sci Rev 2001;29:15–9.
[85] Hagberg JM, Zmuda JM, McCole SD, Rodgers KS, Ferrell RE, Wilund KR, et al.
Moderate physical activity is associated with higher bone mineral density in
postmenopausal women. J Am Geriatr Soc 2001;49:1411–7.
[86] Hall J, Thomas KL, Everitt BJ. Rapid and selective induction of BDNF expression
in the hippocampus during contextual learning. Nat Neurosci 2000;3:533–5.
[87] Harburger LL, Nzerem CK, Frick KM. Single enrichment variables differen-
tially reduce age-related memory decline in female mice. Behav Neurosci
[88] Hardy JA, Higgins GA. Alzheimer’s disease: the amyloid cascade hypothesis.
Science 1992;256:184–5.
[89] Harri M, Lindblom J, Malinen H, Hyttinen M, Lapvetelainen T, Eskola S, et al.
Effect of access to a running wheel on behavior of C57BL/6J mice. Lab Anim
Sci 1999;49:401–5.
[90] Holcomb LA, Gordon MN, Jantzen P, Hsiao K, Duff K, Morgan D. Behavioral
changes in transgenic mice expressing both amyloid precursor protein and
presenilin-1 mutations: lack of association with amyloid deposits. Behav
Genet 1999;29:177–85.
[91] Holmes MM, Galea LA, Mistlberger RE, Kempermann G. Adult hippocampal
neurogenesis and voluntary running activity: circadian and dose-dependent
effects. J Neurosci Res 2004;76:216–22.
[92] Holmquest DL, Retiene K, Lipscomb HS. Random light and wheel running.
Science 1966;154:159.
[93] Howard CV, Reed MG. Unbiased stereology. Oxford: Garland Science/BIOS
Scientific Publishers; 2005.
[94] Jankowsky JL, Melnikova T, Fadale DJ, Xu GM, Slunt HH, Gonzales V, et al.
Environmental enrichment mitigates cognitive deficits in a mouse model of
Alzheimer’s disease. J Neurosci 2005;25:5217–24.
[95] Janus C, Pearson J, McLaurin J, Mathews PM, Jiang Y, Schmidt SD, et al. A
beta peptide immunization reduces behavioural impairment and plaques in
a model of Alzheimer’s disease. Nature 2000;408:979–82.
[96] Janus C, Westaway