Physiology & Behavior 104 (2011) 934–941

Contents lists available at ScienceDirect

Physiology & Behavior

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

Aerobic exercise improves hippocampal function and increases BDNF in the serum of
young adult males
Éadaoin W. Griffin a, c, Sinéad Mullally b, c, Carole Foley a, Stuart A. Warmington a, d,
Shane M. O'Mara b, c, Áine M. Kelly a, c,⁎
a
Department of Physiology, School of Medicine, University of Dublin, Trinity College, Dublin 2, Ireland
b
School of Psychology, University of Dublin, Trinity College, Dublin 2, Ireland
c
Trinity College Institute of Neuroscience, University of Dublin, Trinity College, Dublin 2, Ireland
d
Centre for Physical Activity and Nutrition Research, School of Exercise and Nutrition Sciences, Deakin University, Burwood 3125, Victoria, Australia

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

Article history: Physical activity has been reported to improve cognitive function in humans and rodents, possibly via a brain-
Received 29 October 2009 derived neurotrophic factor (BDNF)-regulated mechanism. In this study of human subjects, we have assessed
Received in revised form 1 June 2011 the effects of acute and chronic exercise on performance of a face–name matching task, which recruits the
Accepted 6 June 2011
hippocampus and associated structures of the medial temporal lobe, and the Stroop word–colour task, which
does not, and have assessed circulating concentrations of BDNF and IGF-1 in parallel. The results show that a
Keywords:
Exercise
short period of high-intensity cycling results in enhancements in performance of the face–name matching,
Learning but not the Stroop, task. These changes in cognitive function were paralleled by increased concentration of
BDNF BDNF, but not IGF-1, in the serum of exercising subjects. 3 weeks of cycling training had no effect on
Hippocampus cardiovascular fitness, as assessed by VO2 scores, cognitive function, or serum BDNF concentration. Increases
in fitness, cognitive function and serum BDNF response to acute exercise were observed following 5 weeks of
aerobic training. These data indicate that both acute and chronic exercise improve medial temporal lobe
function concomitant with increased concentrations of BDNF in the serum, suggesting a possible functional
role for this neurotrophic factor in exercise-induced cognitive enhancement in humans.
© 2011 Elsevier Inc. All rights reserved.

1. Introduction performance in a map recognition task [8] and in the Stroop word–
colour task [9]. However, Grego et al. [8] also showed that prolonged
The benefits that physical activity confers on cardiovascular health exercise leading to fatigue compromises cognitive function. It has
are well known, while recent evidence has also demonstrated the been suggested that intense exercise may facilitate aspects of
ability of exercise to promote brain health. The evidence that cognitive function in a manner dependent on an individual's
physically active older people, particularly those that have been cardiovascular fitness [10]. A recent meta-analysis indicates that
active throughout their lifespan, are at decreased risk of developing cognitive performance may be enhanced or impaired depending on
Alzheimer's disease and other forms of dementia relative to their when, relative to an acute exercise bout, performance is measured,
sedentary counterparts [1–3] strongly suggests that exercise may be a the type of cognitive task selected, and the type of exercise performed
powerful protective strategy against age-related neurodegenerative [11].
decline. In addition to its neuroprotective actions, exercise enhances Evidence available from animal studies provides some insight into
cognitive function in elderly people and slows the progression of the mechanisms by which exercise may enhance cognition. In rodent
dementia-related cognitive symptoms [4–6]. Thus, exercise may models, exercise has consistently been shown to enhance learning
reduce the risk of developing dementia or ameliorate cognitive and persistently upregulate expression of brain-derived neurotrophic
impairment already present in those suffering from neurodegenera- factor (BDNF) in the hippocampus [12–16]. The indisputable
tive decline. importance of the hippocampus in learning and memory and the
Moreover, exercise may also enhance cognitive function in young, role of BDNF in mediating hippocampal synaptic plasticity are well
healthy, adults. High impact running has been shown to improve established [17–20]; while additional evidence indicates a role for
vocabulary learning [7], while cycling has been shown to improve insulin-like growth factor (IGF-1) in mediating the cognitive effects of
exercise [21–23]. Interestingly, serum BDNF concentration has
⁎ Corresponding author at: Trinity College Institute of Neuroscience, University of
repeatedly been reported to increase following exercise in humans
Dublin, Trinity College, Dublin 2, Ireland. Tel.: + 353 18963794; fax: + 353 16793545. [9,24–26] (for review see [27]), while IGF-1 responses to exercise are
E-mail address: [email protected] (Á.M. Kelly). more variable [28–30].

0031-9384/$ – see front matter © 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.physbeh.2011.06.005
 É.W. Griffin et al. / Physiology & Behavior 104 (2011) 934–941 935

Here, we have investigated the effect of acute exercise and aerobic a

Session 1: Week 1
exercise training on cognitive function in young, sedentary men.
Cognitive tests (trial 1)
Given the evidence from animal models that hippocampal function is
particularly responsive to exercise intervention, we assessed the
impact of acute exercise and aerobic exercise training on performance Rest Acute exercise*
in a face–name matching task that recruits the hippocampus, and also
on circulating concentrations of BDNF and IGF-1, in an attempt to CON EX
investigate the possible causal links between increased availability of Cognitive tests (trial 2)
these growth factors and enhancements in cognitive function. *Blood sampling @ time 0, 30, 60, 90min

2. Materials and methods

2.1. Participants A-EX3, C-EX3 A-EX5, C-EX5

The experimental protocol was approved by the Ethical Committee

for Research Involving Human Participants, Faculty of Health Sciences,
Session 2: Week 3 Session 2: Week 5
Trinity College Dublin. Forty-seven healthy male students volun-
Cognitive tests (trial 3) Cognitive tests (trial 3)
teered to participate (age, height, weight: 22 ± 2 yrs, 180 ± 7 cm, 82 ±
11 kg respectively, mean ± SD). All subjects were sedentary (not
Rest Acute exercise* Rest Acute exercise*
involved in any regular physical training) prior to commencement of
the study, and each received a routine medical examination before
providing written informed consent in accordance with the declara- CON A-EX3 C-EX3 CON A-EX5 C-EX5
tion of Helsinki. Exclusion criteria included any contraindications to Cognitive tests (trial 4) Cognitive tests (trial 4)
intense exercise discovered during the medical examination, intake of *Blood sampling @ time 0, 30, 60, 90min *Blood sampling @ time 0, 30, 60, 90min
prescription medication, history of neurological problems, pre-
existing injuries, smoking and intake of recreational drugs. All
subjects were required to fast for 2 h and to refrain from consumption b
of caffeine for 12 h prior to testing. CON: (n=15) No acute exercise, no aerobic training
EX: (n=32) Acute exercise in week 1, then EX group splits;
2.2. Experimental protocol A-EX3: (n=5) Acute exercise in weeks 1 & 3; 3 weeks sedentary
C-EX3: (n=9) Acute exercise in weeks 1 & 3; 3 weeks aerobic training
A-EX5: (n=9) Acute exercise in weeks 1 & 5; 5 weeks sedentary
Participants were allocated to an exercise group (EX) or a C-EX5: (n=9) Acute exercise in weeks 1 & 5; 5 weeks aerobic training
sedentary control group (CON; Fig. 1b). During the first testing
session all participants performed a set of cognitive tasks, including
Fig. 1. Outline of the experimental protocol. (a) Sessions 1 and 2 assessed the effects of
the face–name matching task and the Stroop task. Following this, the acute exercise on cognitive function. During session 1, subjects were divided into CON
EX group completed a graded exercise test, which served as the acute and EX groups and underwent a cognitive testing trial (trial 1), following which EX
exercise bout, while CON participants had a 30 min rest (Fig. 1a). completed an acute exercise bout while CON rested for a corresponding 30 min period.
Blood samples were collected from the EX group throughout the Then all subjects completed a second cognitive testing trial (trial 2). Blood samples
were collected from EX at times 0, 30, 60 and 90 min, with acute exercise occurring
testing session, in order to assess the effect of acute exercise on serum
between times 30 min and 60 min. An overview of the subject groups is shown in (b).
BDNF and IGF-1 concentrations. A baseline blood sample was taken at Baseline measures of cognitive function and serum BDNF concentration were obtained
0 min, followed by the first set of cognitive tests. Another blood during session 1. The effect of both 3 and 5 weeks of chronic exercise then was assessed
sample was taken at 30 min, prior to the graded exercise test (30 min by subdividing the EX group. C-EX3 and C-EX5 completed 3 and 5 weeks of aerobic
rest period for CON). A third, post-acute exercise, blood sample was training respectively, prior to session 2, while A-EX3 and A-EX5 remained sedentary for
the corresponding 3-week or 5-week period prior to session 2. The CON group
taken at 60 min, the second set of cognitive tests was completed and a
completed session 2 after a 3 or 5-week interval and were sedentary both during the
final blood sample was drawn at 90 min (Fig. 1a). interval and during the testing session.
Two chronic exercise protocols were used to assess the effects of
both a 3-week and a 5-week aerobic training intervention on
cognition and on serum concentrations of BDNF and IGF-1. These
aerobic training programmes were identical in all aspects except for
duration. Following the first testing session the EX group was split was set at 75 W and increased by 50 W increments every 3 min, until
into the subgroups: C-EX3, C-EX5, A-EX3 & A-EX5 (Fig. 1a). The 9 min, and subsequently by 25 W increments each min until volitional
chronic-exercise subgroups, C-EX3 and C-EX5, completed 3 and exhaustion was reached. The subject wore a facemask throughout the
5 weeks of aerobic training respectively. The acute-exercise sub- test in order to collect expired air, which was analysed for volume and
groups; A-EX3 and A-EX5, remained sedentary for the corresponding gas composition using an online system (Metalyser, Cortex Biophysik,
3 and 5-week intervals and exercised only when performing the Leipzig, Germany) and from which VO2 max was determined as a
graded exercise test. All subgroups repeated the testing session measure of aerobic fitness. Subjects wore a heart rate monitor (Polar,
(session 2, as in Fig. 1a) following the appropriate 3 or 5-week Oulu, Finland) throughout the test and were verbally encouraged to
interval. The CON group remained sedentary both during the testing continue cycling until heart rate (beats∙min − 1) approached the age
sessions and during the 3-week or 5-week interval between sessions. predicted maximum (220 – age (yrs)).

2.3. Acute exercise

2.4. Cognitive testing
The acute exercise protocol consisted of a graded exercise test
(GXT) to volitional exhaustion, performed on a stationary cycle Subjects were seated before a computer screen and were tested
ergometer (Lode Excalibur, Groningen, Netherlands), to establish using the face–name matching task, followed immediately by the
maximal oxygen consumption rate (VO2 max). The initial workload Stroop word–colour task.
936 É.W. Griffin et al. / Physiology & Behavior 104 (2011) 934–941

2.5. Face–name matching task 2.9. Analysis of BDNF & IGF-1

The face–name task was adapted from a paradigm by Zeineh et al. Serum concentrations of BDNF (Emax® Immunoassay system;
[31] and included encoding, distractor and recall phases. During the Promega Corporation, Madison, WI, USA) and IGF-1 (Human IGF-1
encoding block the subject was presented with images of ten DuoSet ELISA Development kit; R&D Systems Europe, U.K.) were
unfamiliar faces paired with names. Each was presented in sequence assessed by ELISA. In the case of BDNF, 96-well plates were coated
for 3.5 s via a computer screen display. A 40 s distractor task was with anti-BDNF monoclonal antibody (1:1000 dilution in carbonate
performed between each encoding and recall phase to prevent rote coating buffer; 50 μl) and incubated overnight at 4 °C. Plates were
rehearsal of the face–name associations. During the distractor task, a washed with Tris-buffered saline-Tween 20 wash buffer (TBS-T;
fixation cross was presented on the screen at random intervals (0.5 to 150 mM NaCl, 50 mM Tris–HCl, and 0.05% v/v Tween 20, pH 7.4) using
3.5 s), and was replaced for 300 ms on screen by a black circle; the an automated plate washer (Columbus Plus, Tecan, Austria), and
subject was required to press a response button on the keypad when blocked with block and sample buffer (100 μl) for 1 h at room
presented with a black circle. The recall block consisted of a temperature. Plates were washed with TBS-T and samples and
randomized presentation of the previously-viewed faces but without standards were added (50 μl) and incubated for 2 h at room
the paired names. Subjects were requested to recall the correct name temperature on an automated plate shaker. After a further five
and communicate it verbally to the experimenter. washes with TBS-T, anti-Human BDNF polyclonal antibody was added
Subjects viewed the same face–name combinations four times per (1:500 dilution in 1X block buffer; 50 μl) and incubated for 2 h at
task, providing a maximum possible score of 40. With each subject room temperature. Plates were rinsed five times with TBS-T, anti-IgY
eventually completing two testing sessions comprising two cognitive HRP conjugate (1:200 dilution in 1X block buffer; 50 μl) was added
trials, four different series of faces and names were used during the and the plates incubated for 1 h at room temperature on a plate
study, with no subject repeating the cognitive assessment on the same shaker. After a final wash with TBS-T, TMB One solution (50 μl) was
series of faces. Results are presented as the number of pairs recalled. added and plates incubated for 30 min on a plate shaker. The reaction
was stopped with 1 N HCL (50 μl) and the absorbance of samples and
2.6. Stroop word–colour task standards were read at 450 nm using a plate reader (Sunrise basic,
Tecan, Austria).
The Stroop word–colour task consisted of 2 trials of equal duration For the IGF-1 ELISA, 96-well plates were coated with capture
with a short (b1 min) inter-trial rest period. Participants were antibody (mouse anti-human IGF-1, 1:180 dilution in PBS; 80 μl) and
presented with a series of colour words (red, yellow, green, blue) incubated overnight at room temperature. Plates were washed and
on a computer screen. Words were presented in the same (congruent) blocked with block buffer (5% Tween 20, 5% Sucrose in PBS). Samples
or different (incongruent) coloured font. Words appeared on screen and standards were added (50 μl) and incubated for 2 h at room
for 1.3 s. Subjects were required to inhibit their automatic response to temperature. Plates were washed, incubated with detection antibody
read the word stimulus presented, and instead to report the font (biotinylated goat anti-human IGF-1, 1:180 dilution in reagent
colour in which words were presented by pressing colour-coded diluent; 80 μl) for 2 h at room temperature, and reacted with
buttons on a response pad. The protocol consisted of frequent Streptavidin-HRP (1:200 dilution in reagent diluent; 80 μl) for
congruent stimuli with randomized infrequent incongruent stimuli, 20 min. The reaction was stopped with 1 N H2SO4 (50 μl). The
in order to maximize cognitive interference. Results are presented as absorbance of samples and standards were read at 450 nm, standard
percentage response accuracy. Tasks were programmed and run using curves were constructed for each plate and concentrations of BDNF
E-Prime version 1.1 software (Psychology Software Tools, Pittsburgh and IGF-1 in the samples were extrapolated from the curves.
Tools, Pittsburgh, USA).
2.10. Statistical analysis

2.7. Aerobic training Statistical analyses were performed using Graphpad Prism 5 for
Mac OSX. Data are expressed as mean ± standard deviation (SD).
The chronic-exercise groups, C-EX3 and C-EX5 were required to Group n numbers are indicated in Fig. 1b. Where n numbers differ
attend the lab for supervised aerobic cycle training three times per from Fig. 1b, it is due to the removal of outliers (values greater than
week until session 2. Training was performed on a stationary cycle two standard deviations outside the mean) and is clearly indicated in
ergometer for between 30 and 60 min per session. The workload and the results. A possible limitation of our study is that a priori power
duration of the exercise were increased gradually until subjects could calculations were not completed to determine optimal sample size.
maintain a workload estimated to require 60% VO2 max for 60 min. For analysis of the face–name task and the Stroop task, two-way
Heart rate was monitored during each training session as a second repeated measures analysis of variance (ANOVA) were used to assess
method of ensuring training intensity was near to the expected both the effect of trial (the repeated measure) and the effect of group.
workload and as such that the training was sub-maximal and Where a significant difference occurred, Bonferroni post hoc analyses
progressive over the entire training period (3 or 5 weeks). were performed. For the serum analysis, no blood samples were taken
from the CON group, hence one-way repeated measures ANOVA with
2.8. Blood sampling post hoc Newman–Keuls were used to analyse serum BDNF changes
over time for session 1. Two-way repeated measures ANOVA with post
During each testing session venous blood samples were obtained hoc Bonferroni were used to analyse BDNF concentrations for session
at t = 0, 30, 60 and 90 min via an indwelling catheter located in a 2, to assess both the effect of time (the repeated measure) and group.
forearm vein. Following each sample collection the catheter was A value of p b 0.05 was considered to be significant.
flushed with sterile saline (NaCl 0.9% (w/v)) to prevent clot formation
within the catheter, while the catheter was cleared of saline prior to 3. Results
each sample collection. Samples were collected into coagulant-free
6 ml vaccutainer specimen tubes, incubated for 20 min at room 3.1. Acute exercise selectively enhanced cognitive function
temperature to allow clotting, then centrifuged at 5000 rpm for
20 min. The resulting supernatant was removed and stored at − 80 °C Acute exercise induced an enhancement in hippocampal-
for later analysis of the serum concentration of IGF-1 and BDNF. dependent memory, as assessed by the face–name task (Fig. 2a).
 É.W. Griffin et al. / Physiology & Behavior 104 (2011) 934–941 937

3.2. Effect of acute exercise on serum BDNF and IGF-1

Acute exercise in the form of a graded exercise test (GXT) induced an

increase in serum BDNF concentration (Fig. 3a). There was a significant
difference between timepoints (p = 0.0126, F(3,93) = 3.813; one-way
repeated measures ANOVA and post hoc Newman–Keuls, n = 32). Post
hoc analysis revealed a significant increase in serum BDNF concentration
immediately post-GXT (60 min: 1294.0 ± 820.5 pg∙ml− 1), relative to
baseline (0 min: 974.7 ± 741.6 pg∙ml− 1, p b 0.05) and relative to the
30 min timepoint (1004.0 ± 694.1 pg∙ml− 1, p b 0.05). At 90 min (ie.
30 min post-exercise) serum BDNF was not significantly different to
baseline, 30 min or to 60 min (90 min: 1254.0 ± 871.4 pg∙ml − 1).
Results are expressed as mean± SD. Acute exercise did not alter
serum IGF-1 concentration (Fig. 3b) although the degree of inter-
individual variation was large. Statistical analysis was by one-
way repeated measures ANOVA and post hoc Newman–Keuls (0 min:
1087.0 ± 1780.0 pg∙ml− 1, 30 min: 1020.0 ± 1721.0 pg∙ml− 1, 60 min:
1041.0 ± 1718 pg∙ml− 1, 90 min: 1084.0 ± 1744 pg∙ml− 1; n = 28). Re-
sults are expressed as mean± SD.

3.3. Effect of training on aerobic fitness

Aerobic fitness, as assessed by the VO2 max test, was altered by the
aerobic training paradigms (Fig. 4). There was a significant effect of
training (p = 0.0061, F(1,14) = 10.42; two-way repeated measures
ANOVA and post hoc Bonferroni; C-EX3 n = 9; C-EX5 n = 7). Post hoc
analysis revealed a significant increase in post-training VO2 max
scores relative to pre-training values in the C-EX5 group (pre-
training: 39.70 ± 6.74 ml∙min − 1 ∙kg − 1 , post-training: 49.26 ±
4.97 ml∙min − 1∙ kg − 1, p b 0.05) indicating that 5 weeks of training
enhanced aerobic fitness. 3 weeks of aerobic training had no effect
on VO2 max scores in the C-EX3 group (pre-training: 44.69 ±

Fig. 2. Effect of an acute exercise bout on cognitive function. (a) Acute exercise
enhanced performance of the face–name task. The total number of face–name pairs
recalled in trial 2 was greater than trial 1, *p b 0.05, ***p b 0.001 (CON n = 13, EX n = 30).
Acute exercise did not alter performance of the Stroop word–colour task in either
congruent (b) or incongruent (c) trials (CON n = 14, EX n = 28). Statistical analysis
by two-way repeated measures ANOVA and post hoc Bonferroni. Data expressed as
mean ± SD.

There was a significant effect of trial (p b 0.0001, F(1,41) = 19.42; two-

way repeated measures ANOVA and post hoc Bonferroni; CON n = 13;
EX n = 30), indicating that performance in trial 2 was greater than
trial 1. There was also a significant effect of group, (p = 0.0219,
F(1,41) = 5.681), indicating that exercise altered task performance. Post
hoc analysis revealed that while CON scores increased across trial,
(trial 1: 12.15 ± 3.13 pairs recalled, trial 2: 17.08 ± 6.85 pairs recalled,
p b 0.05), EX improved in the performance of the task to a greater
extent, (trial 1: 16.03 ± 5.77 pairs recalled, trial 2: 21.20 ± 7.01 pairs
recalled, p b 0.001). This suggests that, although familiarization with
the task may have resulted in an improved score, the acute-exercise
intervention resulted in an enhancement in the performance of this
task. Results are expressed as mean ± SD of the total number of pairs
recalled across the 4 recall blocks, hence 40 is the maximum possible
score. Acute exercise did not alter performance of the Stroop word– Fig. 3. Effect of acute exercise on serum BDNF and IGF-1 concentrations. (a) Acute
exercise (graded exercise test; GXT) induced an increase in serum BDNF concentration,
colour task in either congruent trials (Fig. 2b) or incongruent trials *p b 0.05 relative to 0 min, +p b 0.05 relative to 30 min (n = 32). (b) Acute exercise did
(Fig. 2c). Statistical analysis was by two-way repeated measures not alter serum IGF-1 concentrations (n = 32). Statistical analysis by one-way repeated
ANOVA (CON n = 13, EX n = 29). measures ANOVA and post hoc Newman–Keuls. Data expressed as mean ± SD.
938 É.W. Griffin et al. / Physiology & Behavior 104 (2011) 934–941

greater that pre-training. Post hoc analysis revealed that there was a
significant increase in face–name performance in the C-EX5 group
(pre-training: 13.6 ± 4.4 pairs, post-training: 21.1 ± 6.2 pairs,
p b 0.05), while the CON and A-EX5 groups remained unchanged.
Results are expressed as mean ± SD.

3.5. Effect of training on serum BDNF response to acute exercise

3 weeks of aerobic training had a significant effect on the serum BDNF

response to acute exercise (Fig. 6a). There was a significant effect of time
(p =0.0075, F(3,33) =4.725; two-way repeated measures ANOVA with
Fig. 4. Effect of aerobic training on VO2 max scores. 3 weeks of aerobic training did not post hoc Bonferoni; A-EX3 n = 5, C-EX3 n = 9) and a significant
affect fitness, as assessed by VO2 max scores. 5 weeks of aerobic training significantly interaction (p = 0.0109, F(3,33) = 4.349). Post hoc analysis revealed
increased VO2 max scores, *p b 0.05 relative to pre-training values (C-EX3 n = 9, C-EX5
an increase in the A-EX3 group immediately post-exercise (60 min:
n = 7). Statistical analysis by two-way repeated measures ANOVA and post hoc
Bonferroni. Data expressed as mean ± SD. 1226.0 ± 957.6 pg∙ml − 1) relative to 0 min (339.0 ± 389.0 pg∙ml − 1,
pb 0.01) and relative to 30 min (531.1 ±585.0 pg∙ml− 1, pb 0.05). This
increase was sustained at 30 min post exercise (90 min: 1059.0 ±
10.36 ml∙min − 1∙kg − 1, post-training: 47.65 ± 8.01 ml∙min − 1∙kg − 1). 1032.0 pg∙ml− 1, pb 0.01). However, serum BDNF concentration did not
Results are expressed as ml oxygen consumed per min per kg body change from baseline in the C-EX3 group (0 min: 898.3±796.3 pg∙ml− 1,
mass, mean ± SD. 30 min: 882.4 ±821.9 pg∙ml− 1, 60 min: 878.8± 500.6 pg∙ml− 1, 90 min:
955.7 ± 560.7 pg∙ml− 1) indicating that 3 weeks of aerobic training
3.4. Effect of training on cognitive function altered the profile of the serum BDNF response to acute exercise.
5 weeks of aerobic training also has a significant effect on the
Hippocampal function, as assessed by the face–name task, was not serum BDNF response to acute exercise (Fig. 6b). There was a
altered by 3 weeks of aerobic training (Fig. 5a). Statistical analysis was significant effect of time (p = 0.0001, F(3,42) = 8.973; two-way
by two-way repeated measures ANOVA (CON n = 15, A-EX3 n = 5, C- repeated measures ANOVA with post hoc Bonferroni; A-EX5 n = 8,
EX3 n = 9). However, 5 weeks of aerobic training enhanced perfor-
mance in the face–name task (Fig. 5b). There was a significant effect of
training (p = 0.0172, F(1,28) = 6.414; two-way repeated measures
ANOVA with post hoc Bonferroni; CON n = 15, A-EX5 n = 8, C-EX5
n = 8) indicating that the number of pairs recalled post-training was

Fig. 6. Effect of aerobic training on serum BDNF response to acute exercise. (a) 3 weeks
of aerobic training had a significant effect on the serum BDNF response to acute
exercise. There was a significant increase in serum BDNF in the A-EX3 group
immediately post acute-exercise (60 min) which was sustained at 90 min, **p b 0.01
relative to 0 min, +p b 0.05 relative to 30 min. However, serum BDNF concentration did
not change from baseline in the C-EX3 group (A-EX3 n = 5, C-EX3 n = 9). (b) 5 weeks of
aerobic training had a significant effect on the serum BDNF response to acute exercise.
There was a significant increase in serum BDNF in the A-EX5 group immediately post
Fig. 5. Effect of aerobic training on performance of the face–name task. (a) 3 weeks of acute-exercise (60 min) which was sustained at 90 min, ***p b 0.001 relative to 0 min,
aerobic training did not affect performance of the face–name task (CON n = 15, A-EX3 **p b 0.01 relative to 0 min, ++p b 0.01 relative to 30 min, +p b 0.05 relative to 30 min.
n = 5, C-EX3 n = 9). (b) 5 weeks of aerobic training enhanced performance of the face– There was also a significant increase in the C-EX5 group in response to acute exercise,
name task. There was a significant increase in face–name performance in the C-EX5 however this did not occur until 30 min post acute-exercise (90 min), *p b 0.05 relative
group, *p b 0.05 relative to pre-training value (CON n = 15, A-EX5 n = 8, C-EX5 n = 8). to 0 min, ++p b 0.01 relative to 30 min (A-EX5 n = 6, C-EX5 n = 8). Statistical analysis
Statistical analysis by two-way repeated measures ANOVA and post hoc Bonferroni. by two-way repeated measures ANOVA with post hoc Bonferroni. Data expressed as
Data expressed as mean ± SD. mean ± SD.
 É.W. Griffin et al. / Physiology & Behavior 104 (2011) 934–941 939

C-EX5 group n = 8). Post hoc analysis revealed a significant increase in increases in basal BDNF concentrations were not found in the chronic
the A-EX5 group immediately post-exercise (60 min: 1485.0 ± analysis, it may be presumed that the increase in serum BDNF
1110.1 pg∙ml − 1 ) relative to 0 min (608.4 ± 538.9 pg∙ml − 1 , reported here is also transient.
p b 0.001) and relative to 30 min (741.8 ± 512.3 pg∙ml − 1, p b 0.01). The source of the BDNF increase remains unclear. Evidence
This increase was sustained at 90 min (1326.4 ± 1121.0 pg∙ml − 1, indicates that the brain is a major, but not the sole contributor to
p b 0.01). There was also a significant increase in BDNF concentration circulating BDNF [32] and platelets also represent a likely source of
in the C-EX5 group, but this did not occur until 90 min (1345.0 ± serum BDNF, as has consistently been reported [36,37]. In this context,
650.9 pg∙ml − 1, p b 0.05 relative to 0 min: 778.4 ± 458.2 pg∙ml − 1, and reports of the ability of BDNF to cross the blood-brain barrier may be
p b 0.01 relative to 30 min: 602.7 ± 340.5 pg∙ml − 1) indicating that of relevance, with movement of BDNF from brain to blood said to
5 weeks of chronic exercise has altered the temporal profile of the occur via bulk flow associated with the reabsorption of the
serum BDNF response to acute exercise. Results are expressed mean ± cerebrospinal fluid [38]. It has also been suggested that exercise
SD. transiently increases the permeability of the blood brain barrier as
demonstrated by an increase in the extravasation of Evans blue
4. Discussion albumin into the brain following 30 min of forced swim exercise in
rats [39] or post-exercise increases in serum S100β concentration in
Here we present evidence that an acute bout of intense aerobic humans [40].
exercise (in the form of a graded exercise test) selectively improves The acute exercise bout did not alter serum IGF-1 concentrations,
performance in a hippocampal-dependent learning task in parallel although the degree of inter-individual variation was large. This result
with increased BDNF concentration in the serum. We also demon- is in broad agreement with the literature. It has been reported that
strate that 5 weeks, but not 3 weeks, of aerobic training improves one hour of treadmill running had no effect on serum IGF-1
performance in hippocampal learning and alters the serum BDNF concentrations in the rat, although increased uptake of serum IGF-1
response to acute exercise. We have recently shown that intracer- into the brain was reported [21], and this has been causally linked
ebroventricular injection of exogenous BDNF protein mimics exercise- with learning enhancements and the induction of BDNF in the brain
induced enhancements in hippocampal-dependent learning in the rat post-exercise. Data from human subjects indicates that the IGF-1
[16] and although circulating BDNF originates both from central and response to exercise is highly variable and may depend on exercise
peripheral sources [27], evidence suggests that the brain is a major intensity, duration and modality. While IGF-1 concentration in the
source of circulating BDNF, both at rest and during exercise, in serum has been reported to increase following short bouts of cycling
humans [32]. We propose that exercise-induced enhancements in [41,42] and high-intensity running [43] other studies have reported
cognitive function in human subjects may be stimulated via a BDNF- no increase in serum IGF-1 following running bouts of varying
linked mechanism. durations and intensities [44–46].

4.1. Acute exercise 4.2. Chronic exercise

An acute bout of exercise induced an enhancement in cognitive A comparison of VO2 max scores revealed that 5 weeks of aerobic
function, as shown by the improvement in face–name task perfor- training increased maximal oxygen consumption rates relative to pre-
mance. This is in agreement with previous studies which suggest that training values, indicating a significant increase in cardiovascular
intense acute exercise enhances learning and memory as assessed by fitness, while 3 weeks of aerobic training did not improve aerobic
a language-learning model [7], and the Stroop task [9]. Face capacity. As we observed no changes in performance of the Stroop
recognition has been shown to recruit the right medial-temporal task following acute exercise, we assessed the impact of training on
lobe (MTL), as evidenced by the inability of patients with right the face–name task alone. 5 weeks of aerobic training enhanced face–
amygdalo-hippocampectomy to recognise previously viewed faces name task performance, while 3-weeks of aerobic training had no
[33]. It has also been repeatedly demonstrated, using high-resolution effect on MTL-dependent learning and memory. This suggests a
functional magnetic resonance imaging (fMRI) acquisition and positive correlation between aerobic fitness and this learning task.
analysis methods, that the face–name association task used in the Taken together these results indicate that both acute exercise and
present study engages the hippocampus [31] and nearby MTL cortical 5 weeks of aerobic training enhance MTL-dependent learning. We
areas, including the amygdala, parahippocampal cortex, perirhinal hypothesized that the mechanism by which physical activity
cortex and entorhinal cortex [34]. Hence, in the present study, the enhances MTL-dependent cognition may be linked with BDNF. An
acute exercise-induced cognitive enhancement appears to be selec- increase in serum BDNF concentration with acute exercise has been
tively MTL-dependent. To our knowledge, this is the first evidence for demonstrated and was associated with a concomitant enhancement
an acute exercise-induced enhancement in function of these MTL in the performance of the face–name task. Furthermore, this serum
structures in humans. BDNF response to acute exercise was shown to be reproducible in
We observed no change in performance of the Stroop word–colour session 2 at both the 3 week and 5 week timepoints. We also
task, which recruits the anterior cingulate cortex and other frontal monitored IGF-1 following chronic training and saw no change at
regions [35], after an acute exercise bout. In contrast to the present any timepoint (data not shown). This is consistent with the lack of
study, Ferris and colleagues [9] have reported post-exercise improve- effect of acute exercise on IGF-1 concentration that we have reported
ments in the performance of the Stroop word and colour tests. herein.
However, given that no control, non-exercising, groups were included There is strong evidence from studies of both human and animal
in the study design, it is possible that the improvements they subjects that alterations in BDNF concentration may have functional
observed were a result of a practice effect. consequences for cognition. Missense polymorphisms in BDNF in
In agreement with the literature, the serum analysis revealed an human subjects are linked with decreased synaptic abundance in the
acute exercise-induced increase in BDNF concentration in sedentary hippocampus and poor performance in memory tasks [47]. The
young men. According to a recent review, 69% of studies in healthy binding of BDNF to its TrkB receptor mediates plastic changes
human subjects reported a ‘mostly transient’ increase in peripheral involved in recognition memory in sheep [48], while exercise has
BDNF concentration following acute exercise [27]. In the present shown to enhance object recognition learning in association with an
study, acute exercise induced an increase in BDNF that had not quite increased concentration of BDNF in the dentate gyrus of young rats
returned to baseline at 30 min post-exercise, however given that [15,16]. Evidence from the literature suggests that BDNF can facilitate
940 É.W. Griffin et al. / Physiology & Behavior 104 (2011) 934–941

neurotransmitter release and enhance synaptic transmission [49,50], [8] Grego F, Vallier JM, Collardeau M, Rousseu C, Cremieux J, Brisswalter J. Influence of
exercise duration and hydration status on cognitive function during prolonged
leading to the hypothesis that the acute exercise-induced enhance- cycling exercise. Int J Sports Med 2005;26(1):27–33.
ment in hippocampal function may be mediated by the actions of [9] Ferris LT, Williams JS, Shen CL. The effect of acute exercise on serum brain-
BDNF on synaptic transmission. derived neurotrophic factor levels and cognitive function. Med Sci Sports Exerc
2007;39(4):728–34.
Neither 3 weeks nor 5 weeks of aerobic training had any effect on [10] Tomporowski PD. Effects of acute bouts of exercise on cognition. Acta Psychol
basal BDNF concentration. Furthermore, following 3 weeks of aerobic (Amst) 2003;112(3):297–324.
training, the ability of acute exercise to increase serum BDNF [11] Lambourne K, Tomporowski P. The effect of exercise-induced arousal on cognitive
task performance: a meta-regression analysis. Brain Res 2010;1341:12–24.
concentration was abolished. Similarly, 5 weeks of aerobic training [12] Chen HI, Lin LC, Yu L, Liu YF, Kuo YM, Huang AM, et al. Treadmill exercise enhances
altered the profile of the BDNF response to acute exercise, in that the passive avoidance learning in rats: the role of down-regulated serotonin system in
BDNF increase was delayed until 30 min post-exercise. These results the limbic system. Neurobiol Learn Mem 2008;89(4):489–96.
[13] Farmer J, Zhao X, van Praag H, Wodtke K, Gage FH, Christie BR. Effects of voluntary
indicate that aerobic training is affecting the induction of increased
exercise on synaptic plasticity and gene expression in the dentate gyrus of adult
serum BDNF by acute exercise, altering the temporal profile of the male Sprague–Dawley rats in vivo. Neuroscience 2004;124(1):71–9.
acute-exercise effect. It is unclear whether this is a result of an [14] Gobbo OL, O'Mara SM. Exercise, but not environmental enrichment, improves
alteration in the mechanism of BDNF release. It is possible that the learning after kainic acid-induced hippocampal neurodegeneration in associa-
tion with an increase in brain-derived neurotrophic factor. Behav Brain Res
effect of chronic exercise on cognition may be mediated by an 2005;159(1):21–6.
alternative mechanism involving BDNF. BDNF infusion has been [15] O'Callaghan RM, Ohle R, Kelly AM. The effects of forced exercise on hippocampal
shown to induce neurogenesis in rats [51]. It has been demonstrated plasticity in the rat: a comparison of LTP, spatial- and non-spatial learning. Behav
Brain Res 2007;176(2):362–6.
that hippocampal neurogenesis also occurs in the dentate gyrus of [16] Griffin EW, Bechara RG, Birch AM, Kelly AM. Exercise enhances hippocampal-
adult humans [52] and that aerobic exercise training can increase the dependent learning in the rat: evidence for a BDNF-related mechanism.
size of the anterior hippocampus in older adults in parallel with Hippocampus 2009;19(10):973–80.
[17] Gooney M, Shaw K, Kelly A, O'Mara SM, Lynch MA. Long-term potentiation and
increased concentration of BDNF in the serum [53]. Potential spatial learning are associated with increased phosphorylation of TrkB and
functional consequences of adult neurogenesis must occur as long- extracellular signal-regulated kinase (ERK) in the dentate gyrus: evidence for a
term adaptations, rather than acute benefits, due to the time required role for brain-derived neurotrophic factor. Behav Neurosci 2002;116(3):455–63.
[18] Hennigan A, Callaghan CK, Kealy J, Rouine J, Kelly AM. Deficits in LTP and
for newly-generated neurons to mature and become integrated into a recognition memory in the genetically hypertensive rat are associated with
network [54]. While this provides an intriguing potential mechanism decreased expression of neurotrophic factors and their receptors in the dentate
underlying the effects of exercise training on cognition, experimental gyrus. Behav Brain Res 2009;197(2):371–7.
[19] Hennigan A, O'Callaghan RM, Kelly AM. Neurotrophins and their receptors: roles
limitations inherent in assessing the cellular mechanisms mediating
in plasticity, neurodegeneration and neuroprotection. Biochem Soc Trans 2007;35
cognition in humans means that investigation of such a hypothesis is (Pt 2):424–7.
currently unfeasible. [20] Tyler WJ, Alonso M, Bramham CR, Pozzo-Miller LD. From acquisition to
The results presented here provide evidence for a link between consolidation: on the role of brain-derived neurotrophic factor signaling in
hippocampal-dependent learning. Learn Mem 2002;9(5):224–37.
acute exercise and cognitive function. Acute exercise has been shown [21] Carro E, Nunez A, Busiguina S, Torres-Aleman I. Circulating insulin-like growth
to increase serum BDNF and selectively improve MTL-dependent factor I mediates effects of exercise on the brain. J Neurosci 2000;20(8):2926–33.
memory. Hence, BDNF is proposed as a mediator of the cognitive [22] Ding Q, Vaynman S, Akhavan M, Ying Z, Gomez-Pinilla F. Insulin-like growth factor
I interfaces with brain-derived neurotrophic factor-mediated synaptic plasticity to
enhancements described, possibly through its reported role in short- modulate aspects of exercise-induced cognitive function. Neuroscience
term mechanisms underlying synaptic plasticity. Furthermore, it has 2006;140(3):823–33.
been shown that while the 3-week training programme was [23] Trejo JL, Llorens-Martin MV, Torres-Aleman I. The effects of exercise on spatial
learning and anxiety-like behavior are mediated by an IGF-I-dependent
insufficient to improve aerobic fitness or augment memory test mechanism related to hippocampal neurogenesis. Mol Cell Neurosci
performance, the 5-week chronic exercise programme resulted in 2008;37(2):402–11.
enhanced fitness scores and improvements in MTL-dependent [24] Gold SM, Schulz KH, Hartmann S, Mladek M, Lang UE, Hellweg R, et al. Basal serum
levels and reactivity of nerve growth factor and brain-derived neurotrophic factor
cognition. A role for BDNF in these improvements in central nervous to standardized acute exercise in multiple sclerosis and controls. J Neuroimmunol
system function is tentatively suggested. 2003;138(1–2):99–105.
[25] Rojas Vega S, Struder HK, Vera Wahrmann B, Schmidt A, Bloch W, Hollmann W.
Acute BDNF and cortisol response to low intensity exercise and following ramp
Acknowledgments incremental exercise to exhaustion in humans. Brain Res 2006;1121(1):59–65.
[26] Tang SW, Chu E, Hui T, Helmeste D, Law C. Influence of exercise on serum brain-
derived neurotrophic factor concentrations in healthy human subjects. Neurosci
Funded by the Irish Research Council for Science, Engineering and
Lett 2008;431(1):62–5.
Technology (Embark Initiative) and Faculty of Engineering, Mathe- [27] Knaepen K, Goekint M, Heyman EM, Meeusen R. Neuroplasticity – exercise-induced
matics and Science, Trinity College Dublin. response of peripheral brain-derived neurotrophic factor: a systematic review of
experimental studies in human subjects. Sports Med 2010;40(9):765–801.
[28] Abellan R, Ventura R, Pichini S, Di Giovannandrea R, Bellver M, Olive R, et al. Effect
References of physical fitness and endurance exercise on indirect biomarkers of recombinant
growth hormone misuse: insulin-like growth factor I and procollagen type III
[1] Abbott RD, White LR, Ross GW, Masaki KH, Curb JD, Petrovitch H. Walking and peptide. Int J Sports Med 2006;27(12):976–83.
dementia in physically capable elderly men. Jama 2004;292(12):1447–53. [29] Kraemer WJ, Aguilera BA, Terada M, Newton RU, Lynch JM, Rosendaal G, et al.
[2] Larson EB, Wang L, Bowen JD, McCormick WC, Teri L, Crane P, et al. Exercise is Responses of IGF-I to endogenous increases in growth hormone after heavy-
associated with reduced risk for incident dementia among persons 65 years of age resistance exercise. J Appl Physiol 1995;79(4):1310–5.
and older. Ann Intern Med 2006;144(2):73–81. [30] Nguyen UN, Mougin F, Simon-Rigaud ML, Rouillon JD, Marguet P, Regnard J.
[3] Rovio S, Kareholt I, Helkala EL, Viitanen M, Winblad B, Tuomilehto J, et al. Leisure- Influence of exercise duration on serum insulin-like growth factor and its binding
time physical activity at midlife and the risk of dementia and Alzheimer's disease. proteins in athletes. Eur J Appl Physiol Occup Physiol 1998;78(6):533–7.
Lancet Neurol 2005;4(11):705–11. [31] Zeineh MM, Engel SA, Thompson PM, Bookheimer SY. Dynamics of the
[4] Cassilhas RC, Viana VA, Grassmann V, Santos RT, Santos RF, Tufik S, et al. The hippocampus during encoding and retrieval of face–name pairs. Science
impact of resistance exercise on the cognitive function of the elderly. Med Sci 2003;299(5606):577–80.
Sports Exerc 2007;39(8):1401–7. [32] Rasmussen P, Brassard P, Adser H, Pedersen MV, Leick L, Hart E, et al. Evidence for a
[5] Heyn P, Abreu BC, Ottenbacher KJ. The effects of exercise training on elderly release of brain-derived neurotrophic factor from the brain during exercise. Exp
persons with cognitive impairment and dementia: a meta-analysis. Arch Phys Med Physiol 2009;94(10):1062–9.
Rehabil 2004;85(10):1694–704. [33] Crane J, Milner B. Do I know you? Face perception and memory in patients with
[6] Stevens J, Killeen M. A randomised controlled trial testing the impact of exercise selective amygdalo-hippocampectomy. Neuropsychologia 2002;40(5):530–8.
on cognitive symptoms and disability of residents with dementia. Contemp Nurse [34] Kirwan CB, Stark CE. Medial temporal lobe activation during encoding and
2006;21(1):32–40. retrieval of novel face–name pairs. Hippocampus 2004;14(7):919–30.
[7] Winter B, Breitenstein C, Mooren FC, Voelker K, Fobker M, Lechtermann A, et al. [35] Leung HC, Skudlarski P, Gatenby JC, Peterson BS, Gore JC. An event-related
High impact running improves learning. Neurobiol Learn Mem 2007;87(4): functional MRI study of the stroop color word interference task. Cereb Cortex
597–609. 2000;10(6):552–60.
 É.W. Griffin et al. / Physiology & Behavior 104 (2011) 934–941 941

[36] Rosenfeld RD, Zeni L, Haniu M, Talvenheimo J, Radka SF, Bennett L, et al. [46] Koistinen H, Koistinen R, Selenius L, Ylikorkala Q, Seppala M. Effect of marathon
Purification and identification of brain-derived neurotrophic factor from human run on serum IGF-I and IGF-binding protein 1 and 3 levels. J Appl Physiol
serum. Protein Expr Purif 1995;6(4):465–71. 1996;80(3):760–4.
[37] Yamamoto H, Gurney ME. Human platelets contain brain-derived neurotrophic [47] Egan MF, Kojima M, Callicott JH, Goldberg TE, Kolachana BS, Bertolino A, et al. The
factor. J Neurosci 1990;10(11):3469–78. BDNF val66met polymorphism affects activity-dependent secretion of BDNF and
[38] Pan W, Banks WA, Fasold MB, Bluth J, Kastin AJ. Transport of brain-derived human memory and hippocampal function. Cell 2003;112(2):257–69.
neurotrophic factor across the blood-brain barrier. Neuropharmacology [48] Broad KD, Mimmack ML, Keverne EB, Kendrick KM. Increased BDNF and trk-B
1998;37(12):1553–61. mRNA expression in cortical and limbic regions following formation of a social
[39] Sharma HS, Cervos-Navarro J, Dey PK. Increased blood-brain barrier permeability recognition memory. Eur J Neurosci 2002;16(11):2166–74.
following acute short-term swimming exercise in conscious normotensive young [49] Jovanovic JN, Czernik AJ, Fienberg AA, Greengard P, Sihra TS. Synapsins as
rats. Neurosci Res 1991;10(3):211–21. mediators of BDNF-enhanced neurotransmitter release. Nat Neurosci 2000;3(4):
[40] Watson P, Black KE, Clark SC, Maughan RJ. Exercise in the heat: effect of fluid 323–9.
ingestion on blood-brain barrier permeability. Med Sci Sports Exerc 2006;38(12): [50] Xu B, Gottschalk W, Chow A, Wilson RI, Schnell E, Zang K, et al. The role of brain-
2118–24. derived neurotrophic factor receptors in the mature hippocampus: modulation of
[41] Cappon J, Brasel JA, Mohan S, Cooper DM. Effect of brief exercise on circulating long-term potentiation through a presynaptic mechanism involving TrkB.
insulin-like growth factor I. J Appl Physiol 1994;76(6):2490–6. J Neurosci 2000;20(18):6888–97.
[42] Schwarz AJ, Brasel JA, Hintz RL, Mohan S, Cooper DM. Acute effect of brief low- and [51] Scharfman H, Goodman J, Macleod A, Phani S, Antonelli C, Croll S. Increased
high-intensity exercise on circulating insulin-like growth factor (IGF) I, II, and IGF- neurogenesis and the ectopic granule cells after intrahippocampal BDNF infusion
binding protein-3 and its proteolysis in young healthy men. J Clin Endocrinol in adult rats. Exp Neurol 2005;192(2):348–56.
Metab 1996;81(10):3492–7. [52] Eriksson PS, Perfilieva E, Bjork-Eriksson T, Alborn AM, Nordborg C, Peterson DA,
[43] Kraemer RR, Durand RJ, Acevedo EO, Johnson LG, Kraemer GR, Hebert EP, et al. et al. Neurogenesis in the adult human hippocampus. Nat Med 1998;4(11):
Rigorous running increases growth hormone and insulin-like growth factor-I 1313–7.
without altering ghrelin. Exp Biol Med (Maywood) 2004;229(3):240–6. [53] Erickson KI, Voss MW, Prakash RS, Basak C, Szabo A, Chaddock L, et al. Exercise
[44] Banfi G, Marinelli M, Roi GS, Colombini A, Pontillo M, Giacometti M, et al. Growth training increases size of hippocampus and improves memory. Proc Natl Acad Sci
hormone and insulin-like growth factor I in athletes performing a marathon at U S A 2011;108(7):3017–22.
4000 m of altitude. Growth Regul 1994;4(2):82–6. [54] Kempermann G, Wiskott L, Gage FH. Functional significance of adult neurogenesis.
[45] Jahreis G, Hesse V, Schmidt HE, Scheibe J. Effect of endurance exercise on Curr Opin Neurobiol 2004;14(2):186–91.
somatomedin-C/insulin-like growth factor I concentration in male and female
runners. Exp Clin Endocrinol 1989;94(1–2):89–96.