Behavioral changes induced by Toxoplasma infection
of rodents are highly specific to aversion of cat odors
Ajai Vyas*†, Seon-Kyeong Kim‡, Nicholas Giacomini*, John C. Boothroyd‡, and Robert M. Sapolsky*§
*Department of Biological Sciences, Stanford University, Stanford, CA 94305; and ‡Department of Microbiology and Immunology and §Departments
of Neurology and Neurological Sciences and of Neurosurgery, Stanford University School of Medicine, Stanford, CA 94305
Edited by Bruce S. McEwen, The Rockefeller University, New York, NY, and approved February 14, 2007 (received for review September 21, 2006)
The protozoan parasite Toxoplasma gondii blocks the innate aversion
ators, are also important for conditioned or learned fear and
of rats for cat urine, instead producing an attraction to the phero-
unconditioned anxiety (12–17). Furthermore, behaviors pertaining
mone; this may increase the likelihood of a cat predating a rat. This
to both anxiety and learned fear appear related to those pertain-
is thought to re?ect adaptive, behavioral manipulation by Toxo-
ing to innate defensive reactions against predators (such as the
plasma in that the parasite, although capable of infecting rats,
aversion to cat urine among rodents). For example, the exploration
reproduces sexually only in the gut of the cat. The ‘‘behavioral
models widely used to measure anxiety in laboratory rodents are
manipulation’’ hypothesis postulates that a parasite will speci?cally
based on the assumption of a conflict between defense (from
manipulate host behaviors essential for enhancing its own transmis-
interspecific and intraspecific threats) and foraging (18, 19). Simi-
sion. However, the neural circuits implicated in innate fear, anxiety,
larly, fear conditioning measures adaptive memory related to
and learned fear all overlap considerably, raising the possibility that
stimuli associated with danger, which is therefore capable of
Toxoplasma may disrupt all of these nonspeci?cally. We investigated
inducing defensive behaviors (15). In view of the shared neural and
these con?icting predictions. In mice and rats, latent Toxoplasma
behavioral substrates, it is possible that Toxoplasma not only
infection converted the aversion to feline odors into attraction. Such
disrupts aversion to cat urine but also compromises learned fear and
loss of fear is remarkably speci?c, because infection did not diminish
anxiety in general. This is in disagreement with the hypothesis that
learned fear, anxiety-like behavior, olfaction, or nonaversive learning.
effects of parasitic infection are highly specific.
These effects are associated with a tendency for parasite cysts to be
In this study we sought to resolve these conflicting predictions.
more abundant in amygdalar structures than those found in other
We confirmed that Toxoplasma infection in rodents blocks the
regions of the brain. By closely examining other types of behavioral
aversion to cat pheromones. Additionally, we demonstrated the
patterns that were predicted to be altered we show that the behav-
specificity of behavioral modifications by Toxoplasma by examining
ioral effect of chronic Toxoplasma infection is highly speci?c. Overall,
a broad range of behaviors concerning anxiety and learned fear.
this study provides a strong argument in support of the behavioral
Finally, we attempted to extend behavioral observations with the
manipulation hypothesis. Proximate mechanisms of such behavioral
localization of Toxoplasma tissue cysts in the brain.
manipulations remain unknown, although a subtle tropism on part of
the parasite remains a potent possibility.
Health Status of Control and Infected Animals Was Comparable. Body
weight gain of control and infected animals was comparable at the
end of experiment [supporting information (SI) Fig. 7], indicating
the relative benign nature of Toxoplasma infection (20). In vivo
he ‘‘behavioral manipulation’’ hypothesis states that a par-
bioluminescent imaging in mice revealed that parasites could no
asite can alter host behavior specifically to increase its own
longer be detected in peripheral tissues at 1 month after infection
transmission efficiency (1, 2). After an acute infection, the
(Fig. 1). Hence, at this point the acute phase had resolved to a
protozoan parasite Toxoplasma gondii latently persists in the
chronic infection, further characterized by the presence of parasite
brain for the life of an infected host, offering an opportunity to
cysts in the brain (rats: 194
28 cysts per brain, n
6; mice: 80
study the behavioral manipulation hypothesis (3). Toxoplasma
13.4 cysts per brain, n
8). None of chronically infected animals
reproduces sexually in a two-species life cycle (4). The sexual
exhibited obvious discomfort or ‘‘sickness behavior’’ at 30 days after
phase of its reproduction occurs in the feline intestine, from
infection. We started our behavioral experiments at this stage.
which highly stable oocysts are excreted in the feces. Grazing
animals, including rodents, can then ingest these oocysts. In
Infection Reduced the Normal Aversion to Cat Odor and Converted It
these hosts, Toxoplasma forms cysts and persists in the central
into a Mild Attraction. Male Long-Evans rats. To examine aversion to
nervous system. The life cycle is completed when a cat eats an
odors, animals were released into a circular arena, and two oppos-
infected animal. Recent reports indicate that the parasite blunts
ing quadrants were laced with either bobcat or rabbit urine.
the innate aversion of rats for the urine of cats, converting this
Uninfected control animals exhibited robust aversion to bobcat
aversion to an attraction (5), although it does not interfere with
urine, as measured by the low occupancy in the bobcat quadrant
energetically costly behaviors related to mating success and
1.6%) compared with the rabbit quadrant (23.0
social status (6). These findings agree with the behavioral
giving the occupancy ratio of the ‘‘bobcat’’ versus ‘‘bobcat plus
manipulation hypothesis, which predicts that parasites will alter
rabbit’’ quadrant of 0.345
0.044 (random occupancy would yield
only behaviors that are beneficial to their transmission while
leaving other behaviors intact.
Several studies have investigated the innate fear of laboratory
Author contributions: A.V., S.-K.K., J.C.B., and R.M.S. designed research; A.V., S.-K.K., and
rodents toward cat odors (7–11). These studies have delineated a
N.G. performed research; A.V. and S.-K.K. contributed new reagents/analytic tools; A.V. and
neuroanatomical circuit comprising the medial hypothalamic zone
N.G. analyzed data; and A.V., S.-K.K., J.C.B., and R.M.S. wrote the paper.
and associated forebrain structures. These forebrain inputs corre-
The authors declare no con?ict of interest.
spond to those emanating from the ventral hippocampus and
This article is a PNAS Direct Submission.
septum on one hand (septohippocampal pathway) and the medial
†To whom correspondence should be addressed. E-mail: email@example.com.
and basolateral amygdala on the other (amygdalar pathway). In-
This article contains supporting information online at www.pnas.org/cgi/content/full/
terestingly, the medial amygdala, basolateral amygdala, and ventral
hippocampus, implicated above in mediating innate fear to pred-
© 2007 by The National Academy of Sciences of the USA
April 10, 2007
www.pnas.org cgi doi 10.1073 pnas.0608310104
Infection caused a transient increase in luminescent signals emanating from parasites. The series of images re?ects the spread of luminescent signal at
successive days after infection. Photon ?ux is coded by a range of pseudocolors (lowest, blue; highest, red).
a ratio of 0.5) (Fig. 2A). Infection significantly abolished such
both control and infected rats strongly avoided a quadrant con-
aversion and in fact caused a mild attraction (bobcat: 16.6
taining the towel/collar, as compared with sham towel/collar quad-
2.0%; relative occupancy ratio
rant (occupancy ratio: control
0.001; 77% increase). A clear leftward shift in the
cumulative frequency distribution of distance from bobcat urine
Female BALB/c mice. Similar to rats, uninfected mice exhibited a
indicated that infected animals spent relatively more time nearer
marked aversion to bobcat urine quadrant (occupancy ratio:
the bobcat urine than controls (Fig. 2B). Representative scatter
8) (Fig. 3A). This aversion was abolished by
plots, demonstrating the location of a control and an infected
infection (ratio: 0.435
0.05). Aversion to cat
animal relative to odor sources during a trial, are shown in Fig. 2 C
collar was measured by calculating the occupancy ratio in two
and D, respectively. Locomotor activity during the test was com-
bisects of a rectangular arena containing either a worn or an
parable between control and infected animals (P
0.5), in terms
unworn cat collar. Whereas control mice showed a moderate
aversion to the cat collar (ratio of time spent in cat bisect
of both total distance traveled and number of progression segments.
5) (Fig. 3B), infected mice exhibited an attraction
A towel and a collar permeated with cat odor apparently
0.05). Hence, infection
represented a more aversive stimulus than did bobcat urine, in that
significantly blunted avoidance behavior in both rats and mice.
Effect of Infection on Aversion to Cat Odors Was Specific, Leaving a
Variety of Fear-Related Behaviors Intact. Infection did not influence
locomotion and anxiety in an open-field arena in rats. Total distance
traveled (Fig. 4A) and median maximal speed during progression
(SI Table 2) were not significantly different between infected
and uninfected groups (P
0.4). Both groups preferred being
away from the center of arena (Fig. 4B) and avoided active
progression while in the center (SI Table 2). Likewise, both
groups undertook a similar number of progression episodes
during the trial (P
0.5) (Fig. 4C).
Infection did not influence fear conditioning or its extinction in rats. Foot
shocks delivered during the training caused a substantial and similar
increase in freezing in both control (n
18) and infected (n
0.6) (Fig. 5, Training). Control animals exhibited signif-
icant freezing when placed in a conditioned context (P
Infection abolished aversion to bobcat urine in rats, instead produc-
ing an attraction. (A) In a circular arena, infection increased the ratio of
occupancy in the quadrant laced with bobcat urine versus that with rabbit
urine. The ordinate depicts relative occupancy in the bobcat quadrant relative
to total occupancy in the bobcat plus rabbit quadrant. The dotted line depicts
the chance level. Control, n
10; infected, n
10. *, P 0.01 relative to control
(Student’s t test). (B) Infected rats spent more of their time nearer to bobcat
Infection abolished aversion to bobcat urine and cat collar in mice. (A)
urine compared with control animals, as illustrated in the leftward shift in the
Infection increased the ratio of occupancy in the bobcat quadrant versus the
cumulative frequency plot of distance from bobcat urine during the trial. The
rabbit quadrant. The dotted line depicts the chance level. Control, n
50% mark (dotted line) for the total n represents the median values. *, P
13. *, P
0.05 relative to control (Student’s t test). (B) Infection
0.001 (Student’s t test). (C and D) Representative scatter plots depicting the
enhanced the occupancy of mice in a bisect of a rectangular arena containing
location of a control rat (C) and an infected rat (D) during trial, with respect
a collar worn by a cat relative to the other half containing a sham collar.
to bobcat and rabbit urine.
5; infected, n
Vyas et al.
April 10, 2007
Table 1. Infection did not affect spatial memory in rats as
measured in the Morris water maze
Occupancy in target
Data show the preference for the target quadrant during the probe trial
conducted immediately after and 1 day after training.
Infection did not affect anxiety-like behavior of rats in the open-?eld
arena. Control, n
15; infected, n
15. Ordinates depict the total distance
tion of cued food relative to total consumption: control
traveled during a 30-min trial (A), time spent in the center (inner two-thirds of
0.7) (SI Fig. 8B).
the arena; the dotted line represents the chance level) relative to the total
duration of the trial (B), and the number of progression segments (C).
Parasites Could Be Detected in a Variety of Brain Regions, with a Trend
Toward Higher Densities in the Amygdalar Region. Bioluminescent
(Fig. 5, Context). No statistically significant difference was observed
signals emanating from Toxoplasma could be detected in wide range of acute
between control and experimental groups (P
0.3) (Fig. 5,
brain sections. Coronal brain slices (1,000 m) were obtained across
Context). When tested for cue conditioning, both control and
the entire brain (23 slices per brain). Photon flux emitted by
infected rats spent a similar amount of time freezing (Fig. 5, Tone).
luciferase-expressing parasites in infected brains (n
When presented with thirty successive tones 2 days after testing
measured. To control for background bioluminescence, slices from
(Fig. 5, Extinction), time spent in freezing by control and infected
rats infected with a Prugniaud strain lacking the luciferase expres-
animals dropped steadily with no significant difference between the
3) served as negative control. Control and infected
animals did not differ in terms of photon flux emanating from brain
Infection did not affect neophobia toward food of novel scent in mice.
slices derived from olfactory bulb and adjoining levels (levels 1–3).
Control mice consistently avoided food with a novel scent (pro-
But, beginning immediately posterior to that (level 4 and more),
photon flux was significantly higher in infected brain tissue than in
portion of scented food relative to total consumption
0.01, Student’s t test and the Mann–Whitney U test)
15). This neophobia was not affected by infection
(Fig. 6). Hence, parasites were present across a wide range of
0.3). Both control and
coronal levels of the brain.
infected mice consumed similar amounts of food during the trial
Tissue cysts could be detected in several brain regions but were more
0.7). SI Fig. 8A depicts the proportion of novel food consumed
prevalent in amygdalar structures. Bioluminescence imaging ascer-
by control and infected mice.
tained that parasites could be detected across the entire brain (Fig.
Infection did not affect the hippocampal-dependent learning. Infection
6A). We then examined the numerical density of the cysts in
did not influence spatial learning in rats.
different brain regions important for defensive behaviors. Cysts
During training, latency to find the hidden platform dropped
could be observed in all regions to a varying extent (mean numerical
significantly in both control (n
8) and infected (n
0.01 cyst per cubic millimeter). An ANOVA with
0.001). ANOVA did not reveal significant differences between
brain regions as source of variance did not reach statistical signif-
experimental groups (P
0.5). Both groups (P
0.5) spent more
0.15) (Fig. 6B). As the next step of analysis, we divided
time in the target quadrant during a probe trial conducted in the
the nine brain regions into three groups: amygdalar areas (medial
absence of a platform immediately after the training (day 0 in Table
and basolateral amygdala), nonamygdalar areas previously shown
1), whereas this preference dropped down to chance level 1 day
to be responsive to cat odor (olfactory bulbs, prefrontal cortex,
afterward (Table 1). SI Fig. 9 depicts performance of control and
ventral hippocampus, periaqueductal gray, and hypothalamus) and
infected animals during the training and at the probe trials.
areas unresponsive to cat odor (caudate putamen and dorsal
Infection did not affect social transmission of food preference in mice.
hippocampus). Interestingly, amygdalar structures showed a higher
Both control (n
4) and infected (n
6) animals preferred cued
tissue cyst density (0.16
0.05 cyst per cubic millimeter) (Fig. 6C)
food smelled previously on the breath of their cagemates (propor-
than in nonresponsive nonamygdalar structures (0.06
per cubic millimeter) (P
0.05, Tukey’s post hoc test) or responsive
nonamygdalar structures (0.09
0.01 cysts per cubic millimeter)
0.05). No significant differences were found between nonre-
sponsive nonamygdalar structures and responsive nonamygdalar
structures. Cyst density did not differ significantly between medial
and basolateral amygdala.
It has been reported that Toxoplasma infection in wild rodents
not only reduces an aversion to predator odors, but also,
surprisingly, results in the development of an actual attraction
(5). We have replicated this finding in laboratory rats and mice,
and we have investigated the specificity of these effects in terms
of defensive behaviors and also the distribution of parasites in
Infection did not affect freezing during various phases of fear
the central nervous system. We report that effects of infection on
conditioning in rats. The ordinate depicts the percentage of time spent
innate aversion to cat pheromones is remarkably specific, be-
in freezing. Control, n
18; infected, n
0.2 (Student’s t test). Panels
cause infection did not diminish learned fear and anxiety-like
(from left to right) represent freezing during training, strength of condi-
tioning to context, strength of cued conditioning, and extinction of cued
behavior, although these behaviors are closely related to the
innate fear of predator pheromones. Similarly, olfaction and
www.pnas.org cgi doi 10.1073 pnas.0608310104
Vyas et al.
compromised, as indicated by intact spatial learning in the Morris
water maze (rats) and social transmission of food preference (mice).
Moreover, we demonstrate that olfaction remains intact after
infection, because infected mice retained an aversion to food with
a novel odor and were capable of learning from olfactory cues
provided by their social group. This is especially important because
behavioral tasks measuring innate fear have mainly relied on odors.
In addition, both an earlier report (5) and data presented in this
report show that the response of infected animals to urine of a
mammalian nonpredator, namely rabbit, remains unchanged.
Collectively, these observations argue against the possibility that
the observed behavioral changes reflect a side effect of generic
Innate fear of cat pheromones shares considerable neurobiolog-
ical substrates with learned fear and anxiety. Hence, it could be
reasonably argued that these effects reflect a generic reduction in
aversive responses of infected animals. We report here that, al-
though Toxoplasma infection blocks the aversion toward predator
odors, it does not compromise learned fear or anxiety. Hence, the
parasite affects only the part of defensive reaction that is important
for its transmission, namely innate aversion to cat odors, while
leaving other aspects of defensive behaviors intact. Furthermore, it
is not merely the case that Toxoplasma attenuates the aversion to cat
urine; instead, rats and mice develop an actual attraction to the
pheromones. This argues against a passive behavioral effect in-
duced by generic malaise or by reduction in a wide spectrum of
defensive behaviors. Moreover, infected animals retain aversion to
dog urine, a mammalian predator not important for the sexual life
cycle of the parasite. Hence, Toxoplasma infection specifically
Localization of tissue cysts in the brain. (A) Detection of parasites in
abolishes aversion to feline odors.
acute brain slices using bioluminescence imaging revealed the presence of
Prior reports indicate that wild-caught rats with naturally occur-
parasites along a broad range of coronal levels. The ordinate depicts the total
photon ?ux (number of photons per second; section thickness
ring Toxoplasma parasites are more active and exhibit smaller
Animals infected with Pru strain parasites lacking the luciferase gene served
latencies to consume novel food compared with rats without an
as controls in this experiment. Negative control, n
3; infected, n
infection (3, 6). In contrast, we did not see differences in locomotion
Representative luminescence of eight successive sections (Bregma
(in rats) or the response to novel food (in mice). The reason for this
6.30 mm) derived from an infected animal. A threshold was
discrepancy is not clear, but it could readily reflect differences
previously determined by using brain slices of control animals. All pixels
between wild-born and laboratory-bred animals.
registering photon ?ux more than the threshold are depicted. (B) Tissue cysts
were observed in a variety of brain regions examined. (C) Determination of
Support for Manipulation Hypothesis. The core of parasitism is the
tissue cyst density using the physical dissector method showed that, among
ability of an organism to exploit its host. According to the manip-
the brain regions analyzed, amygdalar areas (medial and basolateral amyg-
dala) had higher cyst density compared with nonamygdalar areas either
ulation hypothesis, a parasite may be able to alter the behavior of
responsive to or nonresponsive to cat odor (n
its host for its own selective benefit (1, 3). Such selective behavioral
change is proposed to increase reproductive success of the parasite,
usually by enhancing its transmission efficiency. Alternatively, such
nonaversive learning were not affected. In chronically infected
effects could just be side effects of host sickness or even a fortuitous
brains, parasite cysts were randomly distributed over the entire
by-product of the possibility that infection induces hosts to under-
brain. However, we observed that cyst density in amygdalar
take greater risks in pursuit to meet higher energy demands.
structures was 2-fold higher than nonamygdalar structures.
Sequential parasitism akin to Toxoplasma, where a parasite
moves through multiple hosts to complete its life cycle, provides a
Specific Behavioral Effects. How specific are these effects? Three
classic potential of host manipulation. Loss of aversion to cat odors
alternative explanations could be proposed. First, that these be-
in infected rodents could cause an increase in predation rates and
havioral changes are specific to innate aversion of rodents to the cat
increase the transmission of Toxoplasma to cats, a necessary con-
pheromones. Alternatively, these effects could reflect a general
dition for its sexual reproduction. It would be difficult to reconcile
compromise in defensive behaviors such that infected rodents are
specific behavioral effects with the alternative hypothesis that the
less fearful to a range of aversive stimuli. An even broader
behavioral effects are by-products of host sickness or adaptation. A
possibility will be that these effects are merely side effects of generic
direct experimental observation of increased transmission effi-
sickness in the host.
ciency would require an ethically tenuous comparison of predation
A number of findings argue against these effects being purely a
rates between control and infected animals. Nevertheless, mathe-
reflection of generic sickness. Previous studies elegantly show that
matical models describing a similar prey–predator–parasite system
neither social status nor the ability to compete for mates is altered
demonstrate that even a small selective increase in susceptibility of
in infected rodents (3, 6), both behaviors that are energetically
an infected prey population would be sufficient to cause a signifi-
expensive and require appropriate function of the limbic system.
cant increase of parasitic load in predator populations (21).
Moreover, in our studies, control and infected rats had comparable
growth rates during the course of the experiment. Infected mice
Proximate Mechanisms of Innate Aversion to Cat Pheromones. Re-
were similar in weight to control mice at the time of testing and ate
cently, a dedicated circuitry underlying defensive reactions to pred-
a comparable amount of food during behavioral tasks requiring
ator odor has been suggested based on evidence from immediate-
food consumption. Additionally, nonaversive learning, dependent
early gene activation, lesion studies, and neural connectivity
on the hippocampus and presumably important for survival, was not
between brain regions responsive to the cat odor (7–11). This
Vyas et al.
April 10, 2007
circuitry comprises mainly two forebrain pathways impinging on the
variability of the cyst distribution between individuals and the intact
medial hypothalamic zone, namely the septohippocampal pathway
amygdala-dependent behaviors like fear conditioning. Nonetheless,
consisting of the ventral hippocampus and the amygdalar pathway
subtle tropism remains a potent possibility and should be investi-
consisting of the medial and basolateral amygdala. Many compo-
gated further. Toxoplasma infection also causes neuromodulatory
nents of this circuit either overlap or run parallel to those brain
changes (e.g., in the noradrenergic and dopaminergic systems) (31).
regions thought to be important for other domains of emotionality,
Hence, alteration of neuromodulation also provides an important
such as learned fear and/or generalized anxiety (13–17, 22). In light
avenue for such behavioral manipulation. The enigma with this
of these shared features, it seems logical to propose that a manip-
scenario is how specificity is achieved as a consequence of broad
ulation that will cause a complete attenuation of innate aversion to
the cat odors will also have effects on unconditioned anxiety and
A recent study investigating Toxoplasma infection in the context
conditioned fear. Hence, the specificity of these effects does present
of schizophrenia is also noteworthy here (32). It has been postulated
a surprise. Interestingly, lack of such cross-stimulus generality has
that Toxoplasma has some degree of causal relation to schizophre-
also been demonstrated by specific effects of lesions in brain regions
nia (33). This postulate rests on the positive relationships between
comprising the medial hypothalamic zone defensive system (7). For
the prevalence of Toxoplasma antibodies and the development of
example, a lesion of dorsal premammillary nucleus is known to
schizophrenia. A recent article reports that loss of fear to predator
affect antipredator defensive reactions without changing response
odor in infected rats can be reversed by treatment with the
to the foot shock. These findings should further point to a dedicated
antipsychotic drugs haloperidol and valproic acid (32). This study
neurocircuitry mediating antipredator defensive behaviors.
provides one example of the value of integrating behavioral effects
Several studies have investigated the effects of parasitic infections
of Toxoplasma in models of emotional and psychiatric conditions.
on murine defensive behaviors. For example, mice infected with the
In short, data presented in this article provide evidence that the
nematode Heligmosomoides polygyrus exhibit a reduced magnitude
behavioral effects of latent Toxoplasma infection in rodents are very
of analgesia after exposure to a weasel odor (23). Similarly, mice
specific. This specificity, in turn, provides a strong argument in
infected with the protozoa Eimeria vermiformis exhibit decreased
support of the behavioral manipulation hypothesis. Proximate
avoidance of cat odor when compared with uninfected controls
mechanisms of such behavioral manipulations will require future
(24). Specificity of these behavioral effects has not been investigated
studies of a very extensive nature, although a subtle tropism for the
in either of the cases, although mice infected with E. vermiformis
amygdala on the part of the parasites remains a potent possibility.
also show reduced spatial performance in nonaversive Morris
water-maze task (25). Mice infected with the nematode Toxocara
Materials and Methods
canis exhibit a variety of behavioral changes ranging from impaired
Animals. Male Long-Evans rats (8 weeks old, three per cage;
motor performance and impaired learning to decreased neophobia
Charles River Laboratories, Wilmington, MA) and female BALB/c
(26). In this case behavioral changes do not demonstrate a speci-
mice (7 weeks old, five per cage; The Jackson Laboratory, Bar
ficity of a degree comparable with that reported in this article.
Harbor, ME) were used. Animals from these sources tested sero-
Although this specificity is exciting in its own right, it also raises the
logically negative for Toxoplasma. The Stanford University Admin-
possibility that, at some level of analysis, neural substrates of innate
istrative Panel for Laboratory Animal Care reviewed and approved
aversion to felines can be completely dissociated from substrates of
other aspects of fear and defensive reactions. In other words, it
points to a set of neural mechanisms or substrates that is dedicated
only to the processing of cat odors.
Toxoplasma Culture. We used a Prugniaud strain genetically mod-
What are the proximate mechanisms? Several potential mecha-
ified to constitutively express firefly luciferase (under tubulin
nisms should be systematically weighed and tested. A small inven-
promoter) and green fluorescent protein (under GRA2 promoter)
tory of potential mechanisms could include internalization of a few
(S.-K.K. and J.C.B., unpublished data). Parasites were maintained
olfactory receptors important for cat odor detection, alteration in
as tachyzoites by passage in human foreskin fibroblast monolayers.
emotional valence of cat odor by brain rewiring, preferential
localization of Toxoplasma to brain areas in septohippocampal or
Infection and Experimental Groups. Infected fibroblasts were sy-
amygdalar regions, preferential localization to areas projecting to
ringe-lysed by using a 27-gauge needle to release tachyzoites.
septohippocampal and amygdalar systems, uniform distribution but
Animals were either infected with tachyzoites (106 for rats and 400
some brain regions being more sensitive to neuronal damage by
for mice i.p.) or mock-infected with sterile PBS. All behavioral
Toxoplasma, and diffusible substances emanating from infected
experiments were conducted between 4 and 5 weeks after infection.
cells or from Toxoplasma. These possibilities can be broadly divided
in two categories, those requiring that Toxoplasma selectively or at
In Vivo Bioluminescent Imaging in Mice. Anesthetized mice (2%
least preferentially infect a set of brain areas and those not requiring
inhaled isoflurane) were injected with luciferin (150 mg/kg i.p.).
such tropism. We observe the presence of Toxoplasma cyst in a wide
After 10 min photon flux from parasite luciferase was measured
variety of brain regions in both mice and rats. Hence, selective
over 5 min by a cooled CCD camera (IVIS 100; Xenogen, Cran-
localization of Toxoplasma in the brain is not indicated. However,
the density of cysts in the medial and basolateral amygdala is almost
double that in other structures like hippocampus, olfactory bulbs,
Open-Field Exploration in Rats. Exploration in a circular arena
and prefrontal cortex. This supports the case of a subtle tropism.
75 cm) was recorded for 30 min (10 Hz). A ceramic plate,
Interestingly, the amygdala is widely interconnected with a variety
kept in homecage for at least 2 weeks, was placed in one quadrant
of different brain regions (15, 22, 27–29). Hence, a subtle tropism
and served as homebase. NIH Image software was used to generate
to the amygdala does have the potential to influence innate fear via
spatial coordinates. Software for Exploration of Exploration was
specific modulation of relevant pathways. Indeed, a recent study
used to calculate endpoints. Center was defined as the inner
reports that development of conditioned aversion in rat pups to an
two-thirds of the arena.
odor depends on corticosterone acting on an intact amygdala (30).
Moreover, the presence of the mother during conditioning dimin-
Aversion to Bobcat Urine in Rats and Mice. The arena described
ishes both corticosterone secretion and amygdala activation, result-
earlier was used, and the home plate was put in one quadrant. Two
ing in development of a net attraction rather than aversion to the
similar ceramic plates, with 20 undiluted drops of bobcat urine (Leg
odor (30). On the other hand, a more parsimonious account of
Up Enterprises, Lovell, ME) or rabbit urine (from a local animal
observations reported here will have to take into account the high
facility), were placed in quadrants adjoining the home quadrant.
www.pnas.org cgi doi 10.1073 pnas.0608310104
Vyas et al.
Designation of quadrants was randomized across trials. Animal
Spatial Learning in Rats. Rats were trained in a circular water maze
location was tracked during a 60-min trial (10 Hz).
75 cm) to find a submerged platform. Training
consisted of six successive blocks of three trials (trial duration
Aversion to a Cat Towel and a Collar in Rats. The design described
seconds, interblock duration
10 min). Latency to reach platform
above was used, except for the nature of predator odor. The
was recorded. Retention of spatial memory was tested during probe
experiment was conducted in a circular arena (radius
75 cm). A
trials; one immediately after termination of training and another
cat collar worn for at least 1 month and placed over a cotton towel
24 h later. During the probe trial the platform was removed, and
kept with a cat for 48 h served as a source of predator odor.
occupancy in the target quadrant was recorded over 1 min.
Occupancy in the cat odor quadrant was calculated over a 60-min
trial relative to sham stimuli placed in the opposing quadrant.
Sequence of Behavioral Experiments. The following sequence was
used for rats: open-field exploration, aversion to bobcat urine, and
Aversion to a Cat Collar in Mice. The cat collar and a sham collar were
aversion to cat collar and towel. Animals used for fear conditioning
placed on two opposing sides of a rectangular Plexiglas box (47.5
and the Morris water maze were na?¨ve to other tests. The following
20 cm). The location of the animal was recorded (0.33 Hz)
sequence was used for mice: avoidance of novel food, social
transmission of food preference, and aversion to bobcat urine.
over a trial of 15 min. A habituation trial without collars preceded
Aversion to cat collar was measured in a separate set of animals.
the testing. The cat collar was collected during the summer season
after at least 1 month of contact with the cat.
Bioluminescent Imaging in Acute Rat Brain Slices. Brains were har-
vested from rats under deep anesthesia. Coronal sections were
Fear Conditioning in Rats. Rats were conditioned in two modified
obtained across rostrocaudal axis by using a precision brain matrix
observation chambers (30
40 cm; MedAssociates, St.
1,000 m; Braintree Scientific). Sections were
Albans, VT). A load-cell platform recorded locomotor activity of
incubated in individual wells of a 24-well plate with 500 l of PBS
rats by measuring chamber displacement. Freezing was quantified
containing luciferin (1.5 mg/ml) and ATP (2 mM) for 30 min. Total
as the endpoint and was defined as the cessation of all movements
photon flux was determined over 15 min.
Conditioning consisted of two successive presentations of audi-
Determination of Cyst Density in Brain Slices. Brains were harvested
tory tones (5 kHz, 70 dB, 10 seconds, intertrial duration
from animals after transcardial perfusion with 4% paraformalde-
seconds) coterminating with foot shock (1 mA, 1 second). The next
hyde. Subsequent to cryoprotection in 30% sucrose, 40- m-thick
day, the strength of conditioning to context was measured by
coronal sections were obtained by using rotary cryotome and
placing animals in the same spatial and olfactory context for 15 min.
stained with hematoxylin and eosin. Cyst density was determined in
Animals were subsequently tested for cued auditory conditioning
a stereologically unbiased manner by using a physical dissector
by measuring freezing in response to a continuous tone (5 min, 5
approach. Brain regions previously known to be responsive to either
kHz, 70 dB) in a different context. The next day, rats were presented
cat odor or anxiety-provoking stimuli were individually analyzed. In
with 30 successive auditory tones (5 kHz, 70 dB, 10 seconds,
view of the low number of cysts present, data from these regions
50 seconds) to measure the extinction of cued
were subsequently pooled, and differences in density of cyst in
amygdalar and nonamygdalar areas were investigated.
Avoidance of Novel Food and Social Transmission of Food Preference
Statistical Analysis. Student’s t test was performed to determine
in Mice. Animals were habituated to eat powdered chow for 3 days
statistical significance. Values are reported as mean
and were subsequently food-deprived for 12 h. To measure neo-
Wherever appropriate, analysis of variance was conducted, and only
phobia to scented food, mice were presented with two plastic trays
the significant differences were further analyzed by the post hoc
containing either scented (2% coriander wt/wt) or unscented chow,
Student t test. Resultant F values and P values are listed in SI Table
and consumption was recorded. To measure social transmission of
3. Differences in cyst density were analyzed by one-way ANOVA
food preference, two ‘‘demonstrator’’ mice from a cage were
with brain regions as the source of variances. Significant differences
allowed to consume scented food for 1 h (2% cocoa or 1%
were further analyzed by Tukey’s post hoc test.
cinnamon, randomly chosen) and returned to the home cage in
We are grateful to Drs. Chris Contag and Timothy Doyle for help with
contact with three other cagemates. After 10 min, the other three
the bioluminescence imaging. This work was supported by grants from
mice were separated and presented with two plastic trays containing
the National Institutes of Health (Grants MH70903 and AI41014), the
food with either the cued scent or a novel scent, and consumption
Moody Foundation, the Ellison Medical Foundation, and the National
Alliance for Research on Schizophrenia and Depression.
1. Thomas F, Adamo S, Moore J (2005) Behav Processes 68:185–199.
17. Vazdarjanova A, Cahill L, McGaugh JL (2001) Eur J Neurosci 14:709–718.
2. Klein SL (2005) Behav Processes 68:219–221.
18. Hendrie CA, Weiss SM, Eilam D (1996) Pharmacol Biochem Behav 54:13–20.
3. Webster JP (2001) Microbes Infect 3:1037–1045.
19. Mechiel Korte S, De Boer SF (2003) Eur J Pharmacol 463:163–175.
4. Dubey JP (1998) Int J Parasitol 28:1019–1024.
20. Welch WJ (1990) J Am Stat Assoc 85:693–698.
5. Berdoy M, Webster JP, Macdonald DW (2000) Proc Biol Sci 267:1591–1594.
21. Vervaeke M, Davis S, Leirs H, Verhagen R (2006) Parasitology 132:893–901.
6. Berdoy M, Webster JP, Macdonald DW (1995) Parasitology 111:403–409.
22. Fanselow MS, Poulos AM (2005) Annu Rev Psychol 56:207–234.
7. Blanchard DC, Canteras NS, Markham CM, Pentkowski NS, Blanchard RJ (2005)
23. Kavaliers M, Colwell DD, Perrot-Sinal TS (1997) Brain Res 766:11–18.
Neurosci Biobehav Rev 29:1243–1253.
24. Kavaliers M, Colwell DD (1995) Parasitology 111:257–263.
8. Canteras NS, Chiavegatto S, Valle LE, Swanson LW (1997) Brain Res Bull 44:297–
25. Kavaliers M, Colwell DD (1995) Parasitology 110:591–597.
26. Holland CV, Cox DM (2001) J Helminthol 75:125–135.
9. Dielenberg RA, Hunt GE, McGregor IS (2001) Neuroscience 104:1085–1097.
27. Pitkanen A, Pikkarainen M, Nurminen N, Ylinen A (2000) Ann NY Acad Sci
10. Markham CM, Blanchard DC, Canteras NS, Cuyno CD, Blanchard RJ (2004)
Neurosci Lett 372:22–26.
28. Usunoff KG, Itzev DE, Rolfs A, Schmitt O, Wree A (2006) Anat Embryol (Berlin)
11. McGregor IS, Hargreaves GA, Apfelbach R, Hunt GE (2004) J Neurosci 24:4134–
12. Bannerman DM, Rawlins JN, McHugh SB, Deacon RM, Yee BK, Bast T, Zhang WN,
29. Nathan SV, Griffith QK, McReynolds JR, Hahn EL, Roozendaal B (2004) Ann NY
Pothuizen HH, Feldon J (2004) Neurosci Biobehav Rev 28:273–283.
Acad Sci 1032:179–182.
13. Adamec RE, Blundell J, Burton P (2005) Neurosci Biobehav Rev 29:1225–1241.
30. Moriceau S, Sullivan RM (2006) Nat Neurosci 9:1004–1006.
14. Dayas CV, Buller KM, Day TA (1999) Eur J Neurosci 11:2312–2322.
31. Stibbs HH (1985) Ann Trop Med Parasitol 79:153–157.
15. LeDoux J (2003) Cell Mol Neurobiol 23:727–738.
32. Webster JP, Lamberton PH, Donnelly CA, Torrey EF (2006) Proc Biol Sci 273:1023–
16. Pawlak R, Magarinos AM, Melchor J, McEwen B, Strickland S (2003) Nat Neurosci
33. Torrey EF, Yolken RH (2003) Emerg Infect Dis 9:1375–1380.
Vyas et al.
April 10, 2007