Content uploaded by Susan Jenks
Author content
All content in this area was uploaded by Susan Jenks on Oct 26, 2017
Content may be subject to copyright.
This article appeared in a journal published by Elsevier. The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/authorsrights
Author's personal copy
Behavioural Processes 96 (2013) 27–35
Contents lists available at SciVerse ScienceDirect
Behavioural Processes
journal homepage: www.elsevier.com/locate/behavproc
Ingestion of Mycobacterium vaccae decreases anxiety-related behavior
and improves learning in mice
Dorothy M. Matthewsa,∗, Susan M. Jenksb
aDepartment of Biology, The Sage Colleges, Troy, NY 12208, USA
bDepartments of Biology & Psychology, The Sage Colleges, Troy, NY 12208, USA
article info
Article history:
Received 5 January 2013
Accepted 18 February 2013
Keywords:
Mycobacterium vaccae
Serotonin
Learning
Memory
Anxiety-related behaviors
Old friends hypothesis
abstract
Coevolution of microbes and their hosts has resulted in the formation of symbiotic relationships that
enable animals to adapt to their environments and protect themselves against pathogens. Recent studies
show that contact with tolerogenic microbes is important for the proper functioning of immunoregulatory
circuits affecting behavior, emotionality and health. Few studies have examined the potential influence
of ambient bacteria, such as Mycobacterium vaccae on the gut–brain–microbiota axis. In this preliminary
research, we show that mice fed live M. vaccae prior to and during a maze learning task demonstrated a
reduction in anxiety-related behaviors and maze completion time, when tested at three maze difficulty
levels over 12 trials for four weeks. Treated mice given M. vaccae in their reward completed the maze
twice as fast as controls, and with reduced anxiety-related behaviors. In a consecutive set of 12 maze trials
without M. vaccae exposure, treated mice continued to run the maze faster for the first three trials, and
with fewer errors overall, suggesting a treatment persistence of about one week. Following a three-week
hiatus, a final maze run revealed no differences between the experimentals and controls. Additionally, M.
vaccae-treated mice showed more exploratory head-dip behavior in a zero maze, and M. vaccae treatment
did not appear to affect overall activity levels as measured by activity wheel usage. Collectively, our
results suggest a beneficial effect of naturally delivered, live M. vaccae on anxiety-related behaviors and
maze performance, supporting a positive role for ambient microbes in the immunomodulation of animal
behavior.
© 2013 Elsevier B.V. All rights reserved.
Coevolution of microbes, macrobiotic organisms and their
animal hosts over the past 500 million years has resulted in
the development of some symbiotic relationships that enable
animals to adapt to the ambient environment and protect them-
selves against pathogens (Strachan, 1989; Chakrabarty, 2003;
Tlaskalova-Hogenova et al., 2011; Rook, 2012). Such relation-
ships involve bidirectional signaling between the gastrointestinal
tract and the brain via neural, hormonal and immune interactions
(Grenham et al., 2011). Recent work on communication between
the brain–gut–microbiota axis using rodents (Berick et al., 2011;
Grenham et al., 2011; Bravo et al., 2012), monkeys (Bailey et al.,
2004), pigs (Barnes et al., 2012) and humans (Knowles et al., 2008;
Khani et al., 2012) has deepened our understanding of how such
symbiotic relationships can influence animal behavior. Studies with
germ-free animals allow evaluation of the effects of the micro-
biota on the CNS; antibiotic studies provide insight on how use of
∗Corresponding author at: Department of Biology, The Sage Colleges, 63 First
Street, Troy, NY 12208, USA. Tel.: +1 518 925 4958; fax: +1 518 244 3174.
E-mail addresses: matthd@sage.edu,
dorothymatthews1@gmail.com (D.M. Matthews).
broad-spectrum antibiotics can modulate the microbiome and
affect behavior; infection studies show that enteric pathogens
can induce anxiety-like behaviors in animals; probiotic studies
show beneficial effects on the intestinal tract and improved behav-
iors associated with anxiety-related conditions (see Bravo et al.,
2012). For example, Li and colleagues (2009) reported that alter-
ations in the diversity of enteric bacteria influence memory and
learning in mice, Clarke et al. (2012) found sex-specific regu-
lation of hippocampal serotonin associated with anxiety using
germ-free mice and Bravo et al. (2011) further demonstrated that
Lactobacillus rhamnosus influences emotional behavior in mice
through the GI tract with involvement of the vagus nerve and
gamma-amminobutyric acid (GABA) system. This research pro-
vides evidence about how changes in the gut microbiota can lead
to modification in CNS function with ramifications for behavior.
Homeostatic function and behavior, however, can be influenced
not only by normal and disrupted enteric microbiota associations,
but by organisms present in the ambient environment as well. Rook
and Brunet (2002) have proposed and championed the “old friends”
hypothesis as a way to explain the explosion of allergic, chronic
inflammatory and autoimmune disorders present among people
living in developed nations. They suggest that contemporary urban
0376-6357/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.beproc.2013.02.007
Author's personal copy
28 D.M. Matthews, S.M. Jenks / Behavioural Processes 96 (2013) 27–35
lifestyles have disrupted long established relationships during pre-
natal, neonatal and adulthood with coevolved organisms such as
helminths, soil and water microbes, farm animals and pets that are
typically recognized as harmless by the innate immune system and
induce an anti-inflammatory response. Since allergies are medi-
ated by T helper (TH2) lymphocytes, and autoimmunity is mediated
by T helper (TH1/TH17) lymphocytes, the immune dysregulation
caused by lack of exposure to “old friends” likely involves disrup-
tions not only in innate immunity but to the adaptive immune
system as well. Such dysregulation of immunoregulatory circuits
of the immune system may also potentially affect mood, cogni-
tive function and behavior (Rook et al., 2003, 2012; Raison et al.,
2010; Rook, 2012). These same homeostatic processes are likely
important to the behavioral ecology of all mammals.
A microbe that has been the subject of several studies inves-
tigating the hypothesis that extant nonpathogenic organisms can
improve behavioral health outcomes is Mycobacterium vaccae.M.
vaccae is an aerobic bacterium found in temperate environments
and animals are likely exposed to it through contact with water,
soil and vegetation (Sneath et al., 1986; Gomez et al., 2001; Kazda
et al., 2009). As an aerobe, it cannot colonize the anaerobic GI tract
of animals and is thought of as a transient commensal (Rook and
Brunet, 2005). M. vaccae was used in clinical trials in which termi-
nal lung cancer patients were inoculated with heat-killed M. vaccae.
Treated patients showed improved emotional health and general
cognitive function (O’Brien et al., 2004). These findings led to specu-
lation that an immune response to M. vaccae antigens might involve
a ubiquitous neurotransmitter such as serotonin that plays a role
in mood, arousal and learning (Leussis and Bolivar, 2006; Cools
et al., 2007; Hohmann et al., 2007; Cifariello et al., 2008). Thus, an
immune response to this microbe might positively impact behavior
influenced by emotionality.
Examining this idea in a mouse model, Lowry et al. (2007) tested
the hypothesis that peripheral exposure to M. vaccae antigen causes
a T helper cell response that activates brain serotonergic systems
in mice. Their research demonstrated that mice injected with heat-
killed M. vaccae antigen experienced (1) a TH1 and T regulatory
cell biased immune activation of a subset of serotonergic neurons
located in the dorsal raphae nucleus (DRI) of the brainstem and
that project to the hippocampus and other forebrain regions, (2)
elevated serotonin metabolism in the ventromedial prefrontal cor-
tex, and (3) a reduction in stress-related emotional behavior in the
forced swim test. Prior to this, Hunt et al. (2005) showed that heat-
killed M. vaccae could influence immunocompetence through GI
tract interaction in mice after administration by gavage.
Several researchers document the effect of immunomodula-
tion on cognition and psychiatric disorders (Brynskikh et al., 2008;
Miller, 2010; Yirmiya and Goshen, 2011). Integration of Lowry
et al.’s (2007) findings and recent research on the nature of
brain–gut–enteric microbiota interactions encouraged us to ask:
Could ingestion of M. vaccae alter anxiety behavior and influence
learning in mice? We hypothesized that if M. vaccae decreases
stress response through an immune system activation of serotonin
pathways, then mice that ingest M. vaccae may show superior com-
plex maze performance and fewer anxiety-related behaviors than
control mice.
1. General methods: all experiments
1.1. Animals
For all experiments, male, BALB/c specific pathogen free mice
were obtained from Charles River Laboratories when they were
about 38 days old, housed individually in an isolated animal room
under a 12 h light/dark cycle and at a constant 25 ◦C temperature,
and fed Carolina Biological Supply Company Mazuri rodent pellets
(5663) (ad libitum). This mouse strain was used to maintain con-
sistency with the mice used by Lowry et al. (2007). Each mouse
was placed in an individual polycarbonate cage with a wire bar lid
used to hold the water bottle and feed. Carefresh Natural Premium
pet bedding, obtained from Carolina Biological Supply Company,
was placed directly into the cage allowing the absorption of urine
and the animal to burrow and/or den. To allow the mice to become
acclimated to their new setting, the experiments were started when
mice were 52 days old, and weighted approximately 21–25 g.
1.2. Ethical note
All animal experiments were conducted in accordance with the
2010 US Animal Welfare Act under animal use protocols (#01-2010
and #01-2011) and animal husbandry standard operating proce-
dures approved by the Sage Colleges Institutional Animal Care and
Use Committee. All efforts were made to minimize the number of
animals used and their suffering. At the completion of each study
animals were humanely euthanized using CO2.
1.3. M. vaccae
M. vaccae (15,483) was purchased from the American Tissue Cell
Culture (ATCC) and stored at 5 ◦C until reconstituted. M. vaccae
was grown in nutrient broth for four days at 37 ◦C and stored in
a refrigerator until needed. Aliquots of 0.1 mL, containing approx-
imately 4.5×106CFU/mL (determined by a standard plate count)
was applied to the food vehicle of treatment mice, as appropriate.
1.4. Food vehicles
All mice were denied food for 24 h before administration of food
vehicles. The food vehicle given the experimental mice consisted
of a piece of white Wonder bread (produced by Hostess Brands),
approximately 1 cm ×1 cm, onto which 0.1 mL of M. vaccae was
aseptically pipetted. The bread was coated on the same side with
a thin layer of store brand creamy peanut butter to increase pal-
atability. Control mice received a food vehicle like that given the
experimental animals (1 cm ×1 cm square of white bread coated
with peanut butter), but which lacked M. vaccae. Treatment mice
in experiments 2 and 3 received a food vehicle identical to that
given the control mice, i.e. it lacked M. vaccae.
2. Experiments 1–3: Complex maze experiments
2.1. Methods
2.1.1. Sample
In experiments 1–3, ten mice constituted the treatment group
and eight mice constituted the control group. The same mice were
used through the progression from experiments 1 to 3.
2.1.2. Complex maze
A Hebb–Williams style complex maze was used in this study
(Fig. 1). This type of maze is widely used in measuring spatial learn-
ing tasks and working memory with rodents (Shore et al., 2001;
Parle et al., 2006). This maze operates on appetitive rather than
aversive principles.
The mice were tested in a maze free of bedding or other
materials. The maze was a square Plexiglas box (14 cm high,
45 cm ×45 cm) consisting of five rows, 9 cm wide, with five door
openings, 8 cm wide. The start box was 9 cm ×13cm in size. Eight
turns are required to reach the end point of the unobstructed maze.
Three levels of maze difficulty were used in experiment 1. Each
successive level involved additional turns and openings and longer
Author's personal copy
D.M. Matthews, S.M. Jenks / Behavioural Processes 96 (2013) 27–35 29
Fig. 1. Illustration of the maze layout. Mice were placed in the maze at the start box,
and rewards were placed at the finish of level 1, level 2 or level 3 as appropriate.
(Level 3 finish is shown here.)
maze run length. This was accomplished by closing off pertinent
door openings with Plexiglas barriers that were attached to the top
of the maze walls with small clips. Level 1 total maze run distance
was 54 cm, level 2 was 64cm and level 3 was 136 cm.
For level 1, barriers blocked off the two door openings in the
second row of the maze. This required that a minimum of two turns
to be made to reach the food reward. For level 2, barriers were
removed from the second row of the maze that established level 1,
and a barrier was instead placed in the door opening of the third row
of the maze. This required that a minimum of four turns be made
to reach the food reward. For level 3, all barriers were removed and
the maze was completely unobstructed, requiring a minimum of
eight turns to reach the food reward. Level 3 of the maze was used
in all testing in experiments 2 and 3. After each mouse was tested,
the maze was sanitized with 70% alcohol, rinsed with water and
allowed to dry completely.
2.1.3. Anxiety-related behaviors
Mice demonstrate a variety of anxiety-related behaviors
(Blanchard et al., 2001, 2003; Leussis and Bolivar, 2006; Bailey and
Crawley, 2009; Gould, 2010; Smolinsky et al., 2010). Seven separate
anxiety-related behaviors (Table 1) were observed and scored.
2.1.4. Analysis
Maze trials were videotaped (SONY Handicam DCR-HC85). An
experienced observer blind to the treatment for each mouse assign-
ment scored the anxiety-related behaviors from the tapes. All
statistical analyses used IBM Statistical package for the Social Sci-
ences (SPSS), version 20, and all reported values are means and
Table 1
Ethogram of anxiety-related behaviors.
Behavior pattern Description
1. Defecation Number of fecal boli released per trial
2. Elongation Mouse moves with a low-back, stretched posture or
movement while keeping hind feet stationary; number
of events per trial counted
3. Grooming Mouse uses front paws to rub face and whiskers;
number of events per trial counted
4. Immobilization Mouse remains motionless for three or more seconds;
number of events per trial counted
5. Latency to start Number of seconds mouse spends in start box after
initial placement until hindquarters cross the start box
boundary to the maze
6. Return to start Mouse moves from the start box to other area of the
maze and then returns to start box and remains for at
least a second; number of events per trial counted
7. Wall climbing Mouse puts both front paws on maze wall while on
hind legs; all events within 15 s period were scored as
one event
standard errors of the means (S.E.M.). Comparisons of two inde-
pendent means were made using a two-tailed t-test (P< 0.05).
Comparisons among means in experimental designs with multiple
between subjects factors were analyzed using analysis of variance
(ANOVA, P< 0.05) followed, when appropriate, by post hoc analysis
using pairwise comparisons with Bonferroni corrections. Compar-
isons of within-subjects factors were performed using repeated
measures analysis of variance (P< 0.05) followed, where appropri-
ate, by post hoc pairwise comparisons using Bonferroni corrections.
For experiments 1–3, a different observer rescored 8% of the tests
for an interobserver reliability estimate of 92% for the scored
anxiety-related behaviors. Maze running errors were scored as the
total number of three types of errors: wrong turns moving for-
ward; backtracking; and direct return to start box (Winocur and
Moscovitch, 1990; Devan et al., 2006).
3. Details of individual experiments
3.1. Experiment 1: Effect of M. vaccae
3.1.1. Methods
To determine whether mice that ingested live M. vaccae perform
differently in a maze than control mice, experimental mice (N= 10)
were immunologically primed by placing a food vehicle on the wire
bar lid of their cages on two occasions: 21 days and 7 days prior
to the start of maze testing in experiment 1. Since M. vaccae was
incorporated into the food reward of the experimental mice at the
finish of each maze run, those mice received additional M. vaccae
during the 12 maze trials of experiment 1. Control mice were given
a food vehicle on days −21 and −7 as well, but it lacked M. vaccae.
Likewise, the food rewards at the finish of the maze runs of control
mice lacked M. vaccae.
Maze testing was a repeated measures design at three levels of
maze difficulty. All mice were tested during each trial and all test-
ing occurred on Sunday, Tuesday and Thursday. Four trials were
conducted at level 1, one each on Sunday, Tuesday, Thursday, and
Sunday. Likewise, four trials were conducted at level 2, one each on
Tuesday, Thursday, Sunday and Tuesday. Finally, four trials were
conducted at level 3 of the unobstructed maze, one each on Thurs-
day, Sunday, Tuesday and Thursday. This resulted in 12 trials over
a 4-week period.
Start time was recorded when all four paws of the mouse
touched the floor of the maze. Completion time for the maze was
scored when the mouse ate the food reward for a full three seconds
to ensure commitment to ingestion. The order in which individuals
from the experimental and control groups were tested was alter-
nated each time a mouse was tested. Time to finish the maze was
recorded and demonstrated anxiety-related behaviors were scored.
3.1.2. Results
3.1.2.1. Maze run time. Twelve trials of testing at three maze diffi-
culty levels revealed that mice that ingested M. vaccae completed
the maze twice as fast (X= 55.2 ±10.6 s, N= 10) as control mice
(X= 116.8 ±11.8 s, N= 8). A repeated measures ANOVA showed that
ingestion of M. vaccae had a significant effect on the time it took
for experimental and control mice to complete the maze (ANOVA:
F1,16 = 15.08, P=0.001). A main effect by maze level was observed
as well. Mauchly’s test indicated that the assumption of sphericity
had been violated, 2= 16.3, P< 0.05, therefore degrees of freedom
were corrected using Greenhouse–Geisser estimates of sphericity
(ε= 0.6), (ANOVA: F1.2, 19.2 = 30.26, P= 0.0001). Bonferroni post hoc
tests showed that performance differed at levels 1 and 2 of the
maze (Mdiff = 91.65, 95% CI [49.79, 133.52]), and at levels 1 and 3
of the maze (Mdiff = 95.93, 95% CI [49.79, 142.1.]), but not at levels
2 and 3 of the maze (Fig. 2). The group treatment by maze level
Author's personal copy
30 D.M. Matthews, S.M. Jenks / Behavioural Processes 96 (2013) 27–35
Fig. 2. Comparison of the effect of ingestion of M. vaccae on maze run time at each
level of the complex maze in experiment 1. Experimental mice completed the maze
faster than control mice at each level of the maze with the largest difference in
performance seen at level 1.
interaction was also significant (ANOVA: F1.2,19.2 = 6.96, P= 0.013).
Post hoc comparison of means and confidence interval between the
two groups at each level revealed that while experimental mice
performed better than the control mice at all levels of the maze,
their performance at level 3 was not different than their perfor-
mance at level 2 (Table 2).
3.1.2.2. Anxiety-related behaviors. A mixed measures ANOVA with
Bonferroni corrections was performed for each of the anxiety
behaviors, with the group (experimental [N= 10] versus control
[N= 8]) as the between-subjects independent variable, the maze
level (level 1 versus level 2 versus level 3) as the within-subjects
variable, and the duration or count or the anxiety behaviors as
the dependent variables. For all of the behaviors except return to
start, Mauchly’s tests of sphericity were violated and the degrees
of freedom were corrected using Greenhouse–Geisser estimates of
sphericity (ε). Table 3 presents the main effects. The main effect of
maze level was observed for all seven behaviors. Four behaviors
exhibited treatment (group) effects: elongation, immobilization,
Table 2
Summary of repeated measures ANOVA multiple comparisons of M. vaccae treat-
ment by maze level for maze run time in experiment 1.
Maze Level 1 Level 2 Level 3
X(SEM)
95% CI
X(SEM)
95% CI
X(SEM)
95% CI
Exp 87.9a(23.5) 44.0b(7.4) 33.6b(8.6)
[38.05,
137.7]
[28.25,
59.65]
[15.47,
51.78]
Control 209.1c
(26.3)
69.7a(8.3) 71.5a(9.6)
[153.39,
264.8]
[52.16,
87.27]
[51.17,
91.76]
Note. CI =confidence interval. Different subscript letters indicate statistically signif-
icant differences P< 0.05.
grooming and latency to start. A maze level by treatment inter-
action was shown for three behaviors: immobilization, grooming
and latency to start.
Anxiety-related behaviors for both experimentals and controls
decreased from maze level 1 to maze level 3. Post hoc comparisons
are reported only for the behaviors which had both a significant
group effect and maze level effect. The most common pattern was
a decrease in anxiety-related behaviors between levels 1 and 2,
with no significant differences between levels 2 and 3. Bonferroni
post hoc tests showed that both the experimentals and the con-
trols exhibited significantly fewer immobilizations between maze
level 1 and maze level 2 (Xdiff = 0.70, 95% CI [0.214,1.17], P=0.004),
and maze level 3 (Xdiff = 0.63, 95% CI [0.110,1.153], P= 0.015), which
were not significantly different from each other. Likewise, for
latency to start, both the experimentals and the controls exhib-
ited significantly fewer immobilizations between maze level 1 and
maze level 2 (Xdiff = 63.22, 95% CI [29.90, 96.55], P= 0.0001), which
were not significantly different from each other. This pattern was
the same for the grooming behavior (levels 1–2: Xdiff = 0.38, 95% CI
[0.056, 0.71], P= 0.019; levels 1–3: Xdiff = 0.49, 95% CI [0.07, 0.90],
P= 0.019). For elongation a different pattern emerged: elongations
at level 1 and 3 (Xdiff = 2.92, 95% CI [2.04, 3.81], P= 0.0001) and at
level 2 and 3 (Xdiff = 2.10 95% CI [0.95, 3.25], P= 0.0001) were signif-
icantly different, but not between levels 1 and 2. For return to start,
wall climbing and defecation, the pattern of decline was mixed for
both experimentals and controls in terms of whether a significant
difference from level 1 was observed at level 2 first or not until level
3, and the group effect was not significant for these behaviors.
Table 3
Summary of main effects and interaction effects for anxiety behaviors in experiment 1.
Behavior Main effects Interaction
Level Group
Immobilization F1.17, 18.7 = 11.88 F1,16 = 12.99 F1.17,18.7 = 12.96
P= 0.002 P= 0.002 P= 0.03
Grooming F1.26, 20.11 = 8.99 F1,16 = 11.02 F1.26,20.11 = 6.02
P= 0.005 P= 0.004 P= 0.018
Latency to start F1.01,16.19 = 23.7 F1,16 = 12.99 F1.01,16.19 = 7.49
P= 0.001 P= 0.002 P= 0.014
Elongation F1.51, 24.11 = 22.61 F1,16 = 5.84
P= 0.001 P= 0.03
Return to start F2,32 = 9.38
P= 0.001
Wall climbing F1.48,23.70 = 5.50
P= 0.02
Defecation F1.39,22.31 = 8.6
P= 0.004
Author's personal copy
D.M. Matthews, S.M. Jenks / Behavioural Processes 96 (2013) 27–35 31
Table 4
Summary of repeated measures ANOVA multiple comparisons of M. vaccae treatment by maze level for anxiety behaviors in experiment 1.
Maze Level 1 Level 2 Level 3
Behavior X(SEM) 95% CI X(SEM) 95% CI X(SEM) 95% CI
Immobilization
Exp 0.2a(0.3) 0.0b(0.0) 0.03b(0.1)
[−0.30, 0.75] [−0.8, 0.8] [−0.15, 0.20]
Control 1.3c(0.3) 0.2a(0.0) 0.3a(0.1)
[0.73, 1.9] [0.6, 0.25] [0.6, 0.44]
Grooming
Exp 0.2a,d (0.2) 0.1a,d (0.1) 0.1a(0.1)
[−0.27, 0.57] [−0.11, 0.31] [−0.12, 0.22]
Control 1.1b(0.2) 0.4c(0.1) 0.3a,c,d (0.1)
[0.70, 1.60] [0.18, 0.64] [0.07, 0.44]
Latency to start
Exp 31.5a(17.5) 3.8b, e (2.0) 3.5b(1.8)
[−5.52, 68.57] [−0.34, 8.0] [−0.33, 7.33]
Control 108.2c(19.5) 9.5d(2.2) 8.1d,e(2.0)
[66.80, 149.64] [4.82, 14.15] [3.78, 12.35]
Note. CI =confidence interval. Different subscript letters indicate statistically significant differences P< 0.05.
The treatment by maze level interaction was significant for three
of the behaviors: immobilization, grooming and latency to start
(Table 4). These three behaviors show the pattern of performance
in which controls at maze levels 2 and 3 approximate that of exper-
imentals at maze level 1. The significant interaction indicates that
the difference between the groups in these anxiety-related behav-
iors due to treatment was present at maze level 1 but not at levels 2
or 3 of the maze. Although controls exhibited more returns to start,
wall climbing and defecation than experimentals at each maze
level, these differences were not significant in interaction across
the maze levels. By the third level of the maze, the controls were
showing fewer anxiety-related behaviors, approximating the lower
levels of anxiety behaviors expressed by the experimentals at the
first two levels. The measure of latency to start appeared to be the
best indicator of initial anxiety for the experimentals and the con-
trols, with a large mean differences at maze level 1 compared to
maze levels 2 and 3. While experimentals showed significantly
less hesitation to start than controls at each level, this was the
only behavior in which the controls at both levels 2 and 3 exhib-
ited less of the particular anxiety behavior the experimentals at
level 1.
The patterns of reduced run time and reduced demonstration of
anxiety-related behaviors were similar for both experimentals and
controls across the three levels of the maze. This suggests that as
anxiety-related behaviors decline, running speed improves.
3.1.2.3. Errors. Mice treated with M. vaccae demonstrated fewer
errors (X= 3.7 ±0.6, N= 10) across all three levels during maze runs
than the control mice (X= 4.6 ±0.7, N= 8), but these differences
were not statistically significant.
4. Experiment 2: M. vaccae removal
4.1. Methods
To determine what would happen to complex maze perfor-
mance and anxiety-related behaviors when M. vaccae was no longer
administered, both experimental and control mice were tested only
at level 3 of the maze without M. vaccae in the food reward. To
maintain the same maze testing schedule that was used in experi-
ment 1, experiment 2 began three days following the last test day
of experiment 1. All mice were subsequently tested three times a
week for four weeks, yielding a total of 12 trials. Time to finish the
maze and demonstrated anxiety-related behaviors were recorded
as for experiment 1.
4.2. Results
4.2.1. Maze run time
In the consecutive set of 12 trials at level 3 of the complex maze
without M. vaccae in the food reward, experimental mice contin-
ued to complete the maze twice as fast (X= 21.6 ±10.1 s, N= 10)
as the control mice (X= 47.0 ±11.3 s, N= 8). A repeated measures
ANOVA over the course of the 12 trials, however, revealed that these
differences were not statistically significant. Further analysis of dif-
ferences in the maze performance of experimental (X= 18.8 ±6.3 s,
N= 10) and control mice (X= 50.1 ±7.1 s, N= 8) at trials 1 and 2
(ANOVA: F1= 15.33, P= 0.001) and experimental (X= 26.0 ±6.3 s,
N= 10) and control mice (X= 49.5 ±7.0 s, N= 8) at trials 1, 2 and
3 (ANOVA: F1,2 = 6.92, P= 0.018) revealed that experimental mice
completed the maze faster than control mice and that these differ-
ences were statistically significant. At trials 4–12 of maze testing,
however, statistically significant differences in the maze run time
of the two groups were not observed.
4.2.2. Anxiety-related behaviors
Repeated measures ANOVA for each of the seven anxiety-related
behaviors indicated that experimental (X= 0.1 ±0.04, N= 10) and
control mice (X= 0.3 ±0.05, N= 8) differed significantly from one
another in only one behavior, grooming (F1,16 = 10.73, P= 0.005).
4.2.3. Errors
Experimental mice demonstrated fewer errors (X= 2.2 ±0.6,
N= 10) than control mice (X= 4.1 ±0.7, N=8) during the 12 trials of
maze testing. Analysis of these results show that mice who previ-
ously ingested M. vaccae displayed significantly fewer errors than
control mice (ANOVA: F1,16 = 4.53, P= 0.049). There was no main
effect of trial number, or group by trial interaction. This indicates
that even though experimental mice were not running faster than
control mice over the course of the 12 trials of testing at level 3,
they were making less errors in the maze than the control mice.
5. Experiment 3: Strength of memory
5.1. Methods
To determine how well mice remembered the maze pattern, all
mice were rested for three weeks and one final maze test was con-
ducted at level 3, seven weeks after the experimental mice had last
been exposed to M. vaccae.NoM. vaccae was administered in the
food reward at this time. Time to finish the maze and demonstrated
Author's personal copy
32 D.M. Matthews, S.M. Jenks / Behavioural Processes 96 (2013) 27–35
anxiety-related behaviors were recorded as for experiments 1 and
2.
5.2. Results
The experimental mice completed the maze faster
(X= 12.9 ±3.0 s, N= 10) than control mice (X= 20.0 ±4.6 s, N= 8),
and with fewer anxiety-related behaviors (X= 0.8 ±0.2, N= 10)
than control mice (X= 1.1 ±0.3, N= 8). However, these differences
were not statistically significant. Similarly, the experimental mice
demonstrated fewer errors (X= 1.9 ±0.6, N= 10) during maze runs
than the control mice (X= 2.4 ±0.6, N= 8), but the differences in
errors were not statistically significant.
6. Experiment 4: Elevated Zero Maze
6.1. Methods
To evaluate the effects of M. vaccae treatment on anxiety-related
behaviors in addition to those measured during a complex maze
learning task, an elevated zero maze was employed. The elevated
zero maze (EZM) examines anxiety behaviors based on the premise
that mice have an aversion for open, more illuminated spaces (Jonas
et al., 2010) and allows for exploration uninterrupted by a central
space, such as in the elevated plus maze (Shepherd et al., 1994;
Walf and Frye, 2007; Braun et al., 2011).
6.2. Subjects
Forty-one mice were divided into a control group (N= 11) who
did not receive M. vaccae in their food vehicle and three treat-
ment groups (N= 30) of 10 mice each. Mice in the treatment groups
all received M. vaccae in their food vehicle but differed from one
another in the time of testing in the EZM following their last M.
vaccae treatment: 12 h (N=10), 18 h (N= 10) or 24h (N= 10).
6.3. Elevated zero maze
The maze (Med Associates, St. Albans, VT) consisted of a circular
platform (7.0 cm wide with a 45.5cm inner diameter) that was ele-
vated 64.5 cm above the floor. It was equally divided into two closed
quadrants and two open quadrants. The two closed areas had walls
on both sides that were 20.5 cm in height. The open areas lacked
walls, but were bordered by a narrow lip of clear plastic (0.5 cm
high) to diminish the likelihood of mice falling onto the floor. The
activity of the mice was monitored by an overhead camera and
scored by visual observation of behaviors.
6.4. M. vaccae exposure
Experimental mice were exposed to M. vaccae using the
immunological priming schedule utilized in experiment 1. Experi-
mental mice were given a food vehicle on the wire lid of their cages
which contained M. vaccae on three occasions: 3 weeks before maze
testing, 1 week before maze testing and on the day before EZM test-
ing. Control mice were given a food vehicle on the same schedule
as experimental mice, but which lacked M. vaccae.
6.5. EZM testing protocol
EZM testing occurred in a room separate from the home colony
and under dim light (approximately 70 lux) using a protocol
described by Walf and Frye (2007). Each mouse was transported
to the testing room and placed within the closed area of the EZM
at the boundary to the open area, facing inward. EZM testing lasted
for 5 min. Placement in each of the two closed areas of the maze
was alternated between subjects. After each mouse was tested,
the maze was sanitized with 70% alcohol, and allowed to dry com-
pletely.
We scored three behaviors from the EZM trials: number of
entries into the open maze area, time spent in the open area, and
the number of head dips from the closed area and from the open
area. Head dips are a standard ethological measurement used in
EZM testing indicative of motivation to explore and risk assess-
ment (Shepherd et al., 1994; Bourin et al., 2007; Walf and Frye,
2007). Entry into the open area of the maze was recorded when
all four paws were in the open area. Time spent in the open area
was scored as the percentage of time that mice spent with all four
paws in the open area. Head dips are defined as the mouse looking
over the edge of the maze by arching the neck and pointing the
nose down toward the floor. Head dip from the closed area was
scored when one or more paws remained in the closed area while
the mouse looks over the edge. Head dip from open area was scored
when the mouse had four paws in the open area while looking over
the edge of the maze.
6.6. Analysis
Experiment 4 zero maze trials were videotaped using the same
equipment as in experiments 1–3. An experienced observer who
was blind to the treatment assignment scored the entries into the
open area, time spent in the open area and head dipping behaviors
from the tapes.
6.7. Results
There were no statistically significant differences between
experimentals and controls in the time spent in the open area and in
the number of entries into the open area from the closed area. Mice
did differ in the number of combined head dips (combined open
and closed area head dips) (ANOVA: F3,41 = 3.01, P= 0.042). Post hoc
pairwise comparisons revealed that there was a significant differ-
ence in the head dipping behavior of the 12 h group (X= 13.0 ±1.55,
CI [9.86,16.14]) and the control group (X= 8.50 ±1.62, CI [5.21,
11.79]), but no significant differences between the 18 h and 24 h
groups.
7. Experiment 5: activity testing
7.1. Methods
7.1.1. Subjects
Fifteen mice (8 = treatment, 7 = control) were individually
housed in home cages with standard running wheels. Rotation of
the wheels was recorded by magnetically activated counters (Mini
Mitter, Respironics Company).
7.1.2. M. vaccae exposure
Experimental mice were exposed to M. vaccae at three different
times: on the day before the wheels were released (day zero), on
day 14 and on day 21.
7.1.3. Activity testing protocol
Mice were acclimated for two weeks within the activity cages
with the wheels immobilized. On the day following the first expo-
sure of experimental mice to M. vaccae, the wheels were released.
Running activity was collected from the counters daily for the
next 23 days, and the mean km distance traveled/day was calcu-
lated.
Author's personal copy
D.M. Matthews, S.M. Jenks / Behavioural Processes 96 (2013) 27–35 33
7.1.4. Results
The mean daily distance traveled by the experimental mice
(X= 6.3 ±1.4 km, N= 8) did not significantly differ from the control
mice (X= 6.8 ±1.7 km, N= 7) over the 23 days of activity testing.
8. Discussion
This research shows that ingestion of live M. vaccae prior to and
during a complex maze learning task (experiments 1–3) reduced
maze run time and anxiety-related behaviors in BALB/c mice. Four
of the seven measured anxiety-related behaviors, immobilization,
grooming, latency to start and elongation, were significantly dif-
ferent in the M. vaccae treated group as compared to the control
group (Table 3). These effects do not appear to be due to differ-
ences in generalized activity levels related to treatment with M.
vaccae. Experiment 5 did not show differences in wheel running
activity due to treatment with M. vaccae. Additional evaluation of
anxiety using a standard anxiety-testing maze, EZM, revealed only
differences in one measure: head dipping.
In complex maze experiment 1, mice given M. vaccae showed
superior performance compared to controls with both faster
maze run time and reduced expression of anxiety-like behaviors.
Although maze level effects were significant overall, the primary
difference was seen at level 1 for both running time and anxi-
ety behaviors. The difference in performance of the two groups
of mice was greatest during early exposure to the maze when
the novelty of the task might have been most anxiety-provoking
to these animals. The behavior latency to start reveals hesita-
tion to exit the start box and enter the maze. The experimentals
entered the maze more readily than the controls at every level.
Even though at level 1 both experimentals and controls exhibited
much more hesitation than they did at levels 2 and 3, experimen-
tals were approximately three times less hesitant than controls at
each maze level (Table 4). These results indicate that M. vaccae
treatment may have abated anxiety in the experimentals which
affected both motivation to run through the maze and expression
of anxiety-related behaviors. The other three behaviors for which
there was a significant group effect (immobilization, grooming and
elongation) showed a similar pattern. The experimentals and con-
trols did not differ in errors at each level, and the run time of
the two groups was similar by level 3 despite the fact that the
maze difficulty increased at each maze level. Therefore the per-
formance differences across the levels within each group may be
primarily due to increasing familiarity with the maze. The main
outcome of this experiment is that M. vaccae treated mice showed
superior maze run time and diminished anxiety compared to the
controls, and that difference was most pronounced in early maze
trials.
The reduction of anxiety-related behaviors resulting from M.
vaccae ingestion may have allowed more rapid complex maze
investigation, resulting in reduced run time once the maze is
learned at each maze level. If we consider these results in the con-
text of the findings of Lowry et al. (2007) who demonstrated that
injected M. vaccae antigen stimulates brain serotonin production
and decreases stress-related behaviors in mice for a short period
of time in a forced swim test, then it seems feasible that inges-
tion of live M. vaccae may stimulate serotonin production through
immunological mechanisms. This suggests that ingested M. vac-
cae may upregulate DRI serotonergic neurons that modulate stress
responsive behaviors. Plasticity in the stress response influencing
cognitive behavior can be modulated through a T cell response.
While many studies (Grenham et al., 2011; Bravo et al., 2012; Clarke
et al., 2012) show that microbiota can activate innate and adap-
tive immune mechanisms that influence anxiety and behavior, our
research indicates that ingestion of an ambient bacterium which is
not part of the enteric microbiome may have this same beneficial
effect.
In a consecutive set of 12 trials without M. vaccae treatment
at level 3 of the maze (experiment 2), experimental mice contin-
ued to run the maze faster and with less anxiety-related behaviors
than control mice. This pattern was only statistically significant,
however, during the first three trials of maze testing in this exper-
iment. It appears that while M. vaccae ingestion had an effect on
maze performance of experimental mice for about a week after
M. vaccae removal, this effect was not long lasting. Among the
test behaviors, a significant group difference was only observed
for grooming. This indicates that anxiety-behaviors were no longer
influencing maze performance. However, there were significant
differences in the number of maze navigational errors with con-
trol mice demonstrating two times as many errors as experimental
mice. The experimental mice may have remembered the complex
maze pattern better than the controls for the 4-week duration of
experiment 2. Lastly, following a three-week rest period (experi-
ment 3), a final trial revealed no statistically significant differences
in run time, anxiety-related behaviors or errors between the two
groups. The similarities in maze performance and demonstrated
anxiety-related behaviors may have occurred because, after 25
maze trials, both groups of mice knew the maze pattern equally
well.
Although our study did not aim to elucidate the underlying
neuroimmunological mechanisms potentially responsible for the
results we observed, those mechanisms must interact with several
higher order behavioral systems. M. vaccae treatment may have
increased motivation to run the maze due to a facilitatory interac-
tion with rewarding properties inherent in maze exploration itself,
the action of running, and/or to an acutely rewarding feature of the
M. vaccae in the food vehicle given at the end of each maze trial
in experiment 1. Acquisition of conditioned cues may have been
heightened by the treatment resulting in increased performance
times and anxiety reduction which would then have facilitated
familiarization with the maze. In experiment 1, after every four tri-
als, the maze complexity and distance to complete the maze were
increased. Both the experimental mice and control mice exhibited
fewer anxiety-related behaviors with each subsequent level of the
maze despite increasing complexity. However, the experimentals
continued to exhibit fewer anxiety-related behaviors than the con-
trol mice at each new level of the maze. A further consideration
is the level of stress reactivity exhibited by various mouse strains.
BALB/c is considered to be a stress reactive mouse strain (Palumbo
et al., 2009). M. vaccae treatment might produce a different outcome
relative to the stress profile of the strain.
In order to examine the effect of ingested M. vaccae on anx-
iety behavior in a non-cognitive task, we observed performance
in an elevated zero maze (experiment 4). The experimental and
control groups did not differ in their willingness to enter the
open areas of the maze whether tested 12, 18 or 24 h after the
last exposure to ingested M. vaccae or the placebo. However, the
experimental and control groups did differ in head dip behavior
in the 12 h test. As reviewed in Shepherd et al. (1994), includ-
ing ethological measures of behavior indicative of risk assessment
and exploration add clarity and reliability to the standard meas-
ures of open arm entries and time spent, elucidating anxiolytic
and anxiogenic drug effects. In experiment 4, we observed signif-
icant differences in the exploratory risk assessment behavior of
head dipping. Head dipping in elevated mazes is commonly con-
sidered to be an exploratory movement (Bourin et al., 2007) with
increased head dips being indicative of decreased anxiety (Braun
et al., 2011). Head dipping in the EZM may be related to explo-
ration and risk assessment involved in “looking for an escape route”
behavior. In this way it may be related to behaviors required for
exploring, spatially navigating, and learning a complex maze such
Author's personal copy
34 D.M. Matthews, S.M. Jenks / Behavioural Processes 96 (2013) 27–35
as that used in experiments 1–3. Thus, M. vaccae treatment may
affect the motivational system involved in exploration of novel
and potentially threatening environments, and enhance spatial
memory. Moreover, the results of experiment 5 indicate that the
behavioral differences observed in mice treated with M. vaccae
versus controls in experiments 1–4 were likely not due to effects
on basic activity level. It should be noted, however, that the small
sample size used in this experiment may not have provided enough
power to detect a treatment effect of M. vaccae exposure on activ-
ity. Future research should be conducted to answer this question
more fully.
Raison et al. (2010) speculate that the mammalian microbiome
plays a critical role in the development of the immune system and
the maintenance of human health. Saprophytic mycobacteria are
common in the environment (Kazda et al., 2009) and while they do
not replicate in the gut, were likely to have always been present in
the gastrointestinal tract of our ancestors due to contact with mud
and water (Rook, 2010). Repeated, long-term exposure to M. vac-
cae could serve as an adaptive mechanism for the development of
tolerance responses to stressful situations. Further, such a scenario
dovetails with the old friends hypothesis (Raison et al., 2010). Thus,
by upsetting the long established relationship of our immune sys-
tem to ambient bacteria such as M. vaccae, complex behaviors such
as learning could be negatively affected through the dysregulation
of immunoregulatory responses coupled to the neuromodulation
of emotionality.
While recent research has shed light on the ability of the micro-
biota to influence behavior via neural, hormonal and immune
interactions (Li et al., 2009; Bravo et al., 2011; Clarke et al., 2012), it
is surprising to think that a common ambient microbe may modu-
late anxiety behaviors. Our research provides initial data suggesting
that ingestion of live M. vaccae can reduce anxiety behaviors related
to exploration of novel environments, and exert a previously unre-
ported influence on learning in mice. This effect on behavior was
fast acting, observed by the first maze run trial, but the effect dimin-
ished with M. vaccae removal. This suggests that M. vaccae may
act as a kind of pharmabiotic, inducing short-term physiological
changes affecting behavior.
The impact of exposure to microbes on animal behavior in natu-
ral environments is unknown. We could hypothesize, for example,
that differential exposure to microbes such as M. vaccae may influ-
ence the expression of behavioral phenotypes related to being a
“wanderer” or a “resident” (as in prairie voles, Microtus ochrogaster,
Getz et al., 1993; Solomon and Jacquot, 2002; Ophir et al., 2008)
with implications for spatial exploration related to differential
reproduction. Research that incorporates a behavioral ecologi-
cal perspective on brain–gut–microbe interactions is necessary to
understand the underlying mechanisms that shape the evolution of
those interactions. Our results contribute preliminary evidence of
an adaptively significant behavioral response of mice to M. vaccae
ingestion that could have arisen from the coevolution of mam-
malian neuroimmunological systems and ambient microbes.
Conflict of interest statement
All authors declare that there are no conflicts of interest.
Acknowledgements
This research was supported by a Sage Colleges Faculty Research
Grant. The authors gratefully acknowledge Charles Rivers Labora-
tories for supplying the animals used in this study. We thank the
anonymous reviewers for their careful and constructive review of
the manuscript. We also thank T.H. Reynolds for sharing equipment,
K. Light and V. Bolivar for helpful comments on the research, B.
Elder, M. Grubb, M.J. Matthews and L. Drickamer for their thought-
ful review of versions of the manuscript, K. Jones, S. Statham and J.
Dahlgren for graphical assistance, and R. Spica, J. Spear, A. Mathews,
M.J. Matthews and J. Bonaccorso for technical support.
References
Bailey, K.R., Crawley, J.N., 2009. Anxiety-related behaviors in mice. In: Buccafusco,
J.J. (Ed.), Methods of Behavior Analysis in Neuroscience. , 2nd ed. CRC Press, Boca
Raton, pp. 1–26.
Bailey, M.T., Lubach, G.R., Coe, C., 2004. Prenatal stress alters bacterial colo-
nization of the gut in infant monkeys. J. Pediat. Gastroenterol. Nutr. 38,
414–421.
Barnes, J.L., Hartmann, B., Hoist, J.J., Tappenden, K.A., 2012. Intestinal adaptation is
stimulated by partial enteral nutrition supplemented with the prebiotic short-
chair fructooligosaccharide in a neonatal intestinal failure piglet model. J. Parent.
Enteral Nutr. 36, 524–537.
Berick, P., Denou, E., Collins, J., Jackson, W., Lu, J., Deng, Y., Blennerhasset, P., Macri,
J., McCoy, K.D., Verdu, E.F., Collins, S., 2011. The intestinal microbiota affects
central levels of brain-derived neurotropic factor and behavior in mice. Gastro-
enterology 141, 599–609.
Blanchard, D.C., Griebel, G., Blanchard, R.J., 2003. The mouse defense test battery:
pharmacological and behavioral assays for anxiety-related and panic. Eur. J.
Pharmacol. 463, 97–116.
Blanchard, D.C., Griebel, G., Blanchard, R.J., 2001. Mouse defensive behaviors: phar-
macological and behavioral assays for anxiety and panic. Neurosci. Biobehav.
Rev. 25, 205–218.
Braun, A.A., Skelton, M.R., Vorhees, C.V., Williams, M.T., 2011. Comparison of the
elevated plus and elevated zero mazes in treated and untreated male Sprague-
Dawley rats: effects of anxiolytic and anxiogenic agents. Pharmacol. Biochem.
Behav. 97, 406–415.
Bravo, J.A., Forsythe, P., Chew, M.V., Escaravage, E., Savignac, H.M., Dinan, T.G.,
Bienenstock, J., Cryan, J.F., 2011. Ingestion of a Lactobacillus strain regulates emo-
tional behavior and central GABA receptor expression in a mouse via the vagus
nerve. Proc. Natl. Acad. Sci. U.S.A. 108, 16050–16055.
Bourin, M., Petit-Demouliere, B., Dhonnchadha, B.N., Hascoet, M., 2007. Animal mod-
els of anxiety in mice. Fundam. Clin. Pharmacol. 21, 567–574.
Bravo, J.A., Julio-Pieper, M., Forsythe, P., Kunze, W., Dinan, T.G., Bienenstock, J., Cryan,
J.F., 2012. Communication between gastrointestinal bacteria and the nervous
system. Curr. Opin. Pharmacol. 12, 667–672.
Brynskikh, A., Warren, T., Zhu, J.X., Kipnis, J., 2008. Adaptive immunity affects learn-
ing behavior in mice. Brain Behav. Immun. 22, 861–869.
Chakrabarty, A.M., 2003. Microorganisms and cancer: quest for therapy. J. Bacteriol.
185, 2683–2686.
Cifariello, A., Pompili, A., Gasbarri, A., 2008. 5-HT7 receptor in the modulation of
cognitive processes. Behav. Brain Res. 195, 171–179.
Clarke, G., Grenham, S., Scully, P., Fitzgerald, P., Moloney, R.D., Shanahan, F., Dinan,
T.G., Cryan, J.F., 2012. The microbiome–gut–brain axis during early life regu-
lates the hippocampal serotonergic system in a sex-dependent manner. Mol.
Psychiatry June, http://dx.doi.org/10.1038/mp.2012.77.
Cools, R., Roberts, A., Robbins, T., 2007. Serotonergic regulation of emotion and
behavioural control responses. Trends Cogn. Sci. 12, 31–40.
Devan, B.D., Bowker, J.L., Duffy, K.B., Bharati, I.S., Jimenez, M., Sierra-Mercado Jr.,
D., Nelson, C.M., Spangler, E.L., Ingram, D.K., 2006. Phosphodiesterase inhibition
by sildenafil citrate attenuates a maze learning impairment in rats induced by
nitric oxide synthase inhibition. Psychopharmacology (Berl) 183, 439–445.
Hunt, J.R.F., Marinelli, R., Adams, V.C., Rook, G.A.W., Brunet, L.R., 2005. Intragas-
tric administration of Mycobacterium vaccae inhibits severe pulmonary allergic
inflammation in a mouse model. Clin. Exp. Allergy 35, 685–690.
Getz, L.L., McGuire, B., Pizzuto, T., Hofmann, J.E., Frase, B., 1993. Social organization
of the prairie vole (Microtus ochrogaster). J. Mammal. 74, 44–59.
Grenham, S., Clarke, G., Cryan, J.F., Dinan, T.G., 2011. Brain–gut–microbe communi-
cation in health and disease. Front. Physiol. 2, 1–15.
Gomez, A., Mve-Obiang, A., Vray, B., Rudnicka, W., Shamputa, I.C., Portaels, F., Mey-
ers, W.M., Fonteyne, P.-A., Realini, L., 2001. Detection of phospholipase C in
nontuberculous mycobacteria and its possible role in hemolytic activity. J. Clin.
Microbiol. 39, 1396–1401.
Gould, T.D., 2010. Mood and Anxiety Related Phenotypes in Mice: Characterization
using Behavioral Tests. Humana Press, New York.
Hohmann, C.F., Walker, E.M., Boylan, C.B., Blue, M.E., 2007. Neonatal serotonin.
Depletion alters behavioral responses to spatial change and novelty. Brain Res.
1139, 163–177.
Jonas, I., Schubert, K.A., Reijne, A.C., Scholte, J., Garland Jr., T., Gerkema, M.P.,
Scheurink, A.J., Nyakas, C., van Dijk, G., 2010. Behavioural traits are affected
by selective breeding for increased wheel-running behaviour in mice. Behav.
Genet. 40, 542–550.
Kazda, J., Pavlik, I., Hruska, K., 2009. The Ecology of Mycobacteria: Impact on Animals
and Human’s Health. Springer, London.
Khani, S., Hosseini, H.M., Taheri, M., Nourani, M.R., Imani Fooladi, A.A., 2012. Probio-
tics as an alternative strategy for prevention and treatment of human diseases:
a review. Inflammat. Allergy Drug Targets 11, 79–89.
Knowles, S.R., Nelson, E.A., Palombo, E.A., 2008. Investigating the role of perceived
stress on bacterial flora activity and salivary cortisol secretion: a possible mech-
anism underlying susceptibility to illness. Biol. Psychol. 77, 132–137.
Author's personal copy
D.M. Matthews, S.M. Jenks / Behavioural Processes 96 (2013) 27–35 35
Leussis, M.P., Bolivar, V.J., 2006. Habituation in rodents: a review of behavior neu-
robiology and genetics. Neurosci. Biobehav. Rev. 30, 1045–1064.
Li, W., Dowd, S.E., Scurlock, B., Acosta-Martinez, V., Lyte, M., 2009. Memory and
learning behavior in mice is temporally associated with diet-induced alterations
in gut bacteria. Physiol. Behav. 96, 557–567.
Lowry, C.A., Hollis, J.H., de Vries, A., Pan, B., Brunet, L.R., Hunt, J.R.F., Paton, J.F.R.,
van Kampen, E., Knight, D.M., Evans, A.K., Rook, G.A.W., Lightman, S.L., 2007.
Identification of an immune-responsive mesolimbocortical serotonergic sys-
tem: potential role in regulation of emotional behaviour. Neuroscience 146,
756–772.
Miller, A.H., 2010. Depression and immunity: a role for T cells? Brain Behav. Immun.
24, 1–8.
O’Brien, M.E., Anderson, H., Kaukel, E., O’Byrne, K., Pawlicki, M., von Pawel, J.,
Reck, M., 2004. SRL172 (killed Mycobacterium vaccae) in addition to standard
chemotherapy improves quality of life without affecting survival, in patients
with advanced non-small-cell lung cancer: phase III results. Ann. Oncol. 15,
906–914.
Ophir, A.G., Wolff, J.O., Phelps, S.M., 2008. Variation in neural V1aR predicts sexual
fidelity and space use among male prairie voles in semi-natural settings. Proc.
Natl. Acad. Sci. U.S.A. 105, 1249–1254.
Palumbo, M.L., Zorrilla Zubilete, M.A., Cresmaschi, G.A., Genaro, A.M., 2009. Different
effect of chronic stress on learning and memory in BALB/c and C57BL/6 inbred
mice: involvement of hippocampal NO production and PKC activity. Stress 12,
350–361.
Parle, M., Singh, N., Vasudevan, M., 2006. Regular rehearsal helps consolidation of
long term memory. J. Sports Sci. Med. 5, 80–88.
Raison, C.L., Lowry, C.A., Rook, G.A.W., 2010. Inflammation, sanitation, and conster-
nation. Arch. Gen. Psychiatry 67, 1211–1224.
Rook, G.A.W., 2010. The changing microbial environment, Darwinian medicine and
the hygiene hypothesis. In: Rook, G.A.W. (Ed.), The Hygiene Hypothesis and
Darwinian Medicine. Birkhauser Verlag AG, Basel, pp. 1–28.
Rook, G.A.W., 2012. A Darwinian view of the hygiene or old friends hypothesis.
Microbe 7, 173–180.
Rook, G.A.W., Brunet, L.R., 2002. Give us this day our daily germs. Biologist 49,
145–149.
Rook, G.A.W., Brunet, L.R., 2005. Microbes, immunoregulation and the gut. Gut 54,
317–320.
Rook, G.A.W., Raison, C.L., Lowry, C.A., 2012. Can we vaccinate against depression?
Drug Discovery Today 17, 451–458.
Rook, G.A.W., Martinelli, R., Brunet, L.R., 2003. Innate immune responses to mycobac-
teria and the downregulation of atopic responses. Curr. Opin. Allergy Clin.
Immunol. 3, 337–342.
Shepherd, J.K., Grewal, S.S., Fletcher, A., Bill, D.J., Dourish, C.T., 1994. Behavioural and
pharmacological characterization of the elevated zero-maze as an animal model
of anxiety. Psychopharmacology (Berlin) 116, 56–64.
Shore, D.I., Stanford, L., MacInnes, W.J., Klein, R.M., Brown, R.E., 2001. Of mice and
men: using virtual Hebb–Williams mazes to compare learning across gender
and species. Cognition Affect. Behav. Neurosci. 1, 83–89.
Smolinsky, A.N., Bergner, C.L., LaPorte, J.L., Kalueff, A.V., 2010. Analysis of
grooming behaviour and its utility in studying animal stress, anxiety, and
depression. In: Gould, T.D. (Ed.), Mood and anxiety related phenotypes in
mice: Characterization using behavioural tests. Humana Press, New York, pp.
21–36.
Sneath, P.H.A., Mair, N.S., Sharpe, M.E., Holt, J.G. (Eds.), 1986. Bergey’s Manual of
Systematic Bacteriology, Vol. 2. Williams and Wilkins, Baltimore.
Solomon, N.G., Jacquot, J.J., 2002. Characterization of resident and wandering voles
Microtus ochrogaster. Can. J. Zool. 80, 951–955.
Strachan, D.P., 1989. Hay fever, hygiene and household size. Br. Med. J. 6710,
1259–1260.
Tlaskalova-Hogenova, H., Stepankova, R., Kozakova, H., Hudcovic, T., Vannucci, L.,
Tuckova, L., Rossmann, P., Hrncir, T., Kverka, M., Zakostelska, Z., Klemesova,
K., Pribylova, J., Bartova, J., Sanchez, D., Fundova, P., Borovska, D., Srutkova,
D., Zidek, Z., Schwarzer, M., Drastich, P., Funda, D.P., 2011. The role of gut
microbiota (commensal bacteria) and the mucosal barrier in the pathogenesis
of inflammatory and autoimmune diseases and cancer: contribution of germ-
free and gnotobiotic animal models of human diseases. Cell. Mol. Immunol. 8,
110–120.
Walf, A.A., Frye, C.A., 2007. The use of the elevated plus maze as an assay of anxiety-
related behaviour in rodents. Nat. Protoc. 2, 322–328.
Winocur, G., Moscovitch, M., 1990. Hippocampal and prefrontal cortex contributions
to learning and memory: analysis of lesion and aging effects on maze learning
in rats. Behav. Neurosci. 104, 544–551.
Yirmiya, R., Goshen, I., 2011. Immune modulation of learning, memory, neural plas-
ticity and neurogenesis. Brain Behav. Immun. 25, 181–213.
