Voluntary activation of the sympathetic nervous system and attenuation of the innate immune response in humans

Abstract

Significance
Hitherto, both the autonomic nervous system and innate immune system were regarded as systems that cannot be voluntarily influenced. The present study demonstrates that, through practicing techniques learned in a short-term training program, the sympathetic nervous system and immune system can indeed be voluntarily influenced. Healthy volunteers practicing the learned techniques exhibited profound increases in the release of epinephrine, which in turn led to increased production of anti-inflammatory mediators and subsequent dampening of the proinflammatory cytokine response elicited by intravenous administration of bacterial endotoxin. This study could have important implications for the treatment of a variety of conditions associated with excessive or persistent inflammation, especially autoimmune diseases in which therapies that antagonize proinflammatory cytokines have shown great benefit.

Full text

Voluntary activation of the sympathetic nervous
system and attenuation of the innate immune
response in humans
Matthijs Kox
a,b,c,1
, Lucas T. van Eijk
a,c
, Jelle Zwaag
a,c
, Joanne van den Wildenberg
a,c
, Fred C. G. J. Sweep
d
,
Johannes G. van der Hoeven
a,c
, and Peter Pickkers
a,c
a
Intensive Care Medicine,
b
Anesthesiology,
c
Nijmegen Institute for Infection, Inflammation and Immunity, and
d
Laboratory Medicine, Radboud University
Medical Centre, Geert Grooteplein 10, 6500 HB, Nijmegen, The Netherlands
Edited by Tamas L. Horvath, Yale University School of Medicine, New Haven, CT, and accepted by the Editorial Board March 14, 2014 (received for review
December 5, 2013)
Excessive or persistent proinflammatory cytokine production plays
a central role in autoimmune diseases. Acute activation of the
sympathetic nervous system attenuates the innate immune re-
sponse. However, both the autonomic nervous system and innate
immune system are regarded as systems that cannot be voluntar-
ily influenced. Herein, we evaluated the effects of a training
program on the autonomic nervous system and innate immune
response. Healthy volunteers were randomized to either the
intervention (n=12) or control group (n=12). Subjects in the
intervention group were trained for 10 d in meditation (third
eye meditation), breathing techniques (i.a., cyclic hyperventilation
followed by breath retention), and exposure to cold (i.a., immer-
sions in ice cold water). The control group was not trained. Sub-
sequently, all subjects underwent experimental endotoxemia (i.v.
administration of 2 ng/kg Escherichia coli endotoxin). In the inter-
vention group, practicing the learned techniques resulted in inter-
mittent respiratory alkalosis and hypoxia resulting in profoundly
increased plasma epinephrine levels. In the intervention group,
plasma levels of the anti-inflammatory cytokine IL-10 increased
more rapidly after endotoxin administration, correlated strongly
with preceding epinephrine levels, and were higher. Levels of
proinflammatory mediators TNF-α, IL-6, and IL-8 were lower in
the intervention group and correlated negatively with IL-10 levels.
Finally, flu-like symptoms were lower in the intervention group. In
conclusion, we demonstrate that voluntary activation of the sympa-
thetic nervous system results in epinephrine release and subsequent
suppression of the innate immune response in humans in vivo.
These results could have important implications for the treatment
of conditions associated with excessive or persistent inflammation,
such as autoimmune diseases.
LPS
|
cathecholamines
|
cortisol
The innate immune system is crucial to our survival, but ex-
cessive or persistent proinflammatory cytokine production
can result in tissue damage and organ injury, such as in auto-
immune diseases. Biological therapies that antagonize proin-
flammatory cytokines or their receptors are very effective and
have revolutionized the treatment of autoimmune diseases, such
as rheumatoid arthritis and inflammatory bowel disease (1, 2).
However, these drugs are expensive and have serious side
effects (3, 4). Therefore, innovative therapies aimed at lim-
iting inflammatory cytokine production in a more physiolog-
ical manner are warranted.
Acute activation of the sympathetic nervous system attenuates
inflammation via activation of β2-adrenoreceptors by catechol-
amines, exemplified by the fact that (nor)epinephrine attenuates
lipopolysaccharide (LPS)-induced TNF-αrelease in vitro (5, 6)
and short-term infusion of epinephrine limits production of
proinflammatory cytokines in vivo during experimental endo-
toxemia (i.v. administration of LPS in healthy volunteers) (7). In
addition, as part of a stress response, increased levels of
catecholamines are often accompanied by elevations of the well-
known immunosuppressive hormone cortisol [via activation of
the hypothalamic–pituitary–adrenal (HPA) axis] (8, 9).
Next to exogenous (i.e., pharmacological or electrical) mod-
ulation of the autonomic nervous system (ANS), endogenous
stimulation of ANS activity may also limit the inflammatory
response, but the ANS is generally regarded as a system that
cannot be voluntarily influenced. However, results from a re-
cently performed case study on a Dutch individual, who holds
several world records with regard to withstanding extreme cold,
suggest otherwise (10). It was shown that this individual was able
to voluntarily activate the sympathetic nervous system through
a self-developed method involving meditation, exposure to cold,
and breathing techniques. This resulted in increased catechol-
amine and cortisol release and a remarkably mild innate immune
response during experimental endotoxemia compared with more
than 100 subjects who previously underwent experimental endo-
toxemia. In the present study, we investigated the effects of his
training program (see Movie S1 for an impression) on sympa-
thetic nervous system parameters and the innate immune response
in healthy male volunteers during experimental endotoxemia in
a randomized controlled fashion.
Significance
Hitherto, both the autonomic nervous system and innate im-
mune system were regarded as systems that cannot be vol-
untarily influenced. The present study demonstrates that,
through practicing techniques learned in a short-term training
program, the sympathetic nervous system and immune sys-
tem can indeed be voluntarily influenced. Healthy volunteers
practicing the learned techniques exhibited profound increases
in the release of epinephrine, which in turn led to increased
production of anti-inflammatory mediators and subsequent
dampening of the proinflammatory cytokine response elicited
by intravenous administration of bacterial endotoxin. This
study could have important implications for the treatment of
a variety of conditions associated with excessive or persistent
inflammation, especially autoimmune diseases in which ther-
apies that antagonize proinflammatory cytokines have shown
great benefit.
Author contributions: M.K., L.T.v.E., J.G.v.d.H., and P.P. designed research; M.K., L.T.v.E.,
J.Z., and J.v.d.W. performed research; M.K. analyzed data; F.C.G.J.S. supervised the cate-
cholamine analysis; J.G.v.d.H. cosupervised the conduct of the study; P.P. supervised the
conduct of the study; and M.K., L.T.v.E., F.C.G.J.S., J.G.v.d.H., and P.P. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission. T.L.H. is a guest editor invi ted by the
Editorial Board.
1
To whom correspondence should be addressed. E-mail: matthijs.kox@radboudumc.nl.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1322174111/-/DCSupplemental.
www.pnas.org/cgi/doi/10.1073/pnas.1322174111 PNAS
|
May 20, 2014
|
vol. 111
|
no. 20
|
7379–7384
IMMUNOLOGY

Results
Baseline characteristics of subjects that underwent experimental
endotoxemia in both groups were similar (Table 1).
Cardiorespiratory Parameters, Temperature, and Symptoms. In the
control group, arterial blood gas parameters pCO
2
,pO
2
, pH,
bicarbonate, lactate, and oxygen saturation were normal and did
not substantially change during endotoxemia (Fig. 1 A–F). In
contrast, in trained individuals, practicing the learned breathing
techniques resulted in an immediate and profound decrease of
pCO
2
and bicarbonate, and an increase in pH (reaching up to
7.75 in individual subjects; Fig. 2 and Movie S2), indicating acute
respiratory alkalosis, which normalized quickly after cessation
of the breathing techniques. Mean pO
2
remained virtually un-
altered in trained subjects, whereas lactate levels were signifi-
cantly elevated, but not to clinically relevant levels. A significant
decrease in oxygen saturation was observed in the trained group
during practicing of the breathing techniques (Fig. 1F). Mini-
mum oxygen saturation levels in each cycle of hyper/hypo-
ventilation (after cessation of breathing for several minutes)
typically dropped to around 50% in trained individuals for a
short period (∼10 s; Fig. 2 and Movie S2). Heart rate and mean
arterial blood pressure (MAP) showed a pattern typical for
endotoxemia in the control group: a gradual decrease in MAP
and a compensatory rise in heart rate after LPS administration
(Fig. 1 Gand H). In the trained group, heart rate increased after
commencing the breathing techniques and normalized earlier
compared with the control group, whereas MAP decreased
during the breathing techniques and thereafter followed the
same pattern as in the control group. LPS administration resul-
ted in fever, with a maximum temperature increase in the control
group of 1.9 ±0.2 °C (mean ±SEM), whereas this increase was
less pronounced and normalized earlier in the trained group
(Fig. 1I). Self-reported symptoms (nausea, headache, shivering,
and muscle and back pain on a six-point Likert scale) peaked
1.5 h after LPS administration in both groups, but were attenu-
ated in the trained individuals compared with the control group
(reduction of 56% in peak levels; Fig. 1J).
Catecholamine and Cortisol Levels. Plasma epinephrine levels (Fig.
3A) increased sharply 1 h after LPS administration and peaked at
T=1.5 h in the control group. In trained subjects, baseline
epinephrine levels were significantly higher compared with the
control group (mean ±SEM: 1.02 ±0.22 vs. 0.35 ±0.06 nmol/L,
P=0.007) (unpaired Student ttest). After starting practicing the
learned breathing techniques, epinephrine levels further in-
creased in this group and peaked just before administration of
LPS (mean ±SEM: 2.08 ±0.37 nmol/L at T=0 h, with in-
dividual subjects reaching up to 5.3 nmol/L) and remained ele-
vated until cessation of the breathing techniques. In contrast
to epinephrine, norepinephrine and dopamine levels remained
within the reference range throughout the experiment (Fig. 3 B
and C). Norepinephrine levels were similar between groups
during the breathing period, although trained subjects displayed
higher levels at baseline and after cessation of the breathing
techniques. In contrast, dopamine levels were slightly lower in
trained individuals during the breathing techniques but were
similar between groups before and afterward. There were no
differences in serum levels of the stress hormone cortisol be-
tween the groups before or during the period in which the
trained group practiced their techniques; however, levels nor-
malized more quickly in trained individuals (Fig. 3D).
Leukocyte Counts. Total leukocyte counts in both groups showed
the typical endotoxemia-induced biphasic pattern with an initial
leukopenia followed by leukocytosis (Fig. S1A). Leukocyte con-
centrations were markedly higher in trained individuals. 30 min
after start of the breathing techniques (T=0 h), an increase in
lymphocytes was observed in trained individuals, which was not
present in the control group (Fig. S1B). Concentrations of neu-
trophils and monocytes were similar between groups at this early
time point, but were distinctly higher in the trained group at later
time points (Fig. S1 Cand D).
Plasma Cytokines. Plasma concentrations of proinflammatory cyto-
kines TNF-α, IL-6, and IL-8, and the anti-inflammatory cytokine
IL-10 all markedly increased after LPS administration in both
groups (Fig. 4). However, in trained individuals, TNF-α,IL-6,
and IL-8 levels were significantly attenuated, whereas the IL-10
response was greatly augmented compared with the control
group (TNF-α, IL-6, and IL-8 levels 53%, 57%, and 51% lower;
IL-10 levels 194% higher). Furthermore, IL-10 levels in the
trained group increased sharply early after LPS administration
Table 1. Subject demographic characteristics
Parameter Trained group, n=12 Control group, n=12 Pvalue
Age, y 24 (19–27) 22 (19–27) 0.43
Height, cm 181 (172–190) 185 (179–189) 0.30
Weight, kg 75 (58–92) 78 (65–91) 0.25
BMI, kg/m
2
23 (19–26) 23 (20–27) 0.98
HR, beats/min 60 (41–80) 61 (40–75) 0.88
MAP, mmHg 92 (82–113) 94 (78–105) 0.89
Parameters were measured during screening visit. BMI, body mass index;
HR, heart rate; MAP, mean arterial blood pressure. Data are presented as
median (range). Pvalues were calculated using Mann–Whitney utest.
B
pCO2
2
4
6
kPaApH
74
7.6
7.8
12
14
16
CpO2
kPa
oxygen saturation
100
%
F
0
2
p<0.0001
-1 0 1 2 3 4 5 6 7 8
7.2
7.4
p<0.0001
-101234567 8
bicarbonate
25
mmol/L
D
10
12
-1 0 1 2 3 4 5 6 7 8
3
mmol/L
Elactate
p<0.55
88
92
96
p<0.0001
10
15
20
p<0.0001
0
1
2
p<0.0003
88
-1 0 1 2 3 4 5 6 7 8
temperature
38
39
oCI
HR
80
100
bpm
GMAP
100
120
mmHg
H
10
-101234567 8 0-1 0 1 2 3 4 5 6 7 8
36
37
38
-1 0 1 2 3 4 5 6 7 8
time (hours post-LPS)
p=0.001
t
AU
J
40
60
80
p<0.0001
-1 0 1 2 3 4 5 6 7 8 60
80
100
p<0.0001
-1 0 1 2 3 4 5 6 7 8
time (hours post-LPS) time (hours post-LPS)
symptom score
4
8
12
AU
p<0.0001 control
trained
time (hours post-LPS)
0-1 0 1 2 3 4 5 6 7 8
time (hours post-LPS)
Fig. 1. Cardiorespiratory parameters, temperature, and symptoms during
experimental endotoxemia in control and trained subjects. (A) Carbon di-
oxide partial pressure (pCO
2
) in arterial blood. (B) Oxygen partial pressure
(pO
2
) in arterial blood. (C) pH in arterial blood. (D) Bicarbonate (HCO3
−
)in
arterial blood. (E) Lactate in arterial blood. (F) Oxygen saturation measured
by pulse oximetry. (G) Heart rate (HR). (H) Mean arterial pressure (MAP). (I)
Temperature. (J) Score of self-reported symptoms. Data are expressed as
mean ±SEM of 12 subjects per group. Gray box indicates period in which the
trained subjects practiced their learned breathing techniques. Pvalues be-
tween groups were calculated using repeated measures two-way analysis of
variance (ANOVA, interaction term). AU, arbitrary units; bpm, beats per minute.
7380
|
www.pnas.org/cgi/doi/10.1073/pnas.1322174111 Kox et al.

(at T=1 h) and peaked 1 h before the peak observed in the
control group. In line with previous reports (11), plasma levels
of the proinflammatory cytokine IL-1βwere barely detectable
during human endotoxemia. Concentrations were below the
detection limit (3.9 pg/mL) in all but four subjects (two in each
group, showing very low concentrations (4–6 pg/mL) at one to
three time points with no apparent kinetics over time). Con-
centrations of the anti-inflammatory cytokine TGF-βshowed no
kinetics after administration of LPS and were not different be-
tween groups (Fig. S2A). We also measured plasma concen-
trations of leptin, an adipokine that exerts proinflammatory
activity. At baseline (T=−1 h), there was a trend toward lower
levels of leptin in the trained group compared with the control
group (mean ±SEM: 3.36 ±0.55 vs. 4.99 ±0.74 ng/mL, P=0.09,
unpaired Student ttest), which remained apparent at all sub-
sequent time points (Fig. S2B). Leptin kinetics showed a biphasic
pattern with an initial modest decrease followed by a gradual
increase in both groups. However, there were no differences be-
tween groups over time.
Correlation Analyses. As depicted in Fig. 5A, there was a strong
positive correlation (r
s
=0.82, P=0.001) between epinephrine
levels in the trained group at T=0 h (30 min after commencing
the breathing techniques) and the early increase in IL-10 levels
at T=1 h, which was not present in the control group (r
s
=0.18,
P=0.571). Furthermore, there were significant inverse correla-
tions between levels of the anti-inflammatory cytokine IL-10 at
T=1 h and peak levels of the proinflammatory mediators TNF-α
(at T=1.5 h), IL-6 (at T=2 h), and IL-8 (at T=2 h) in the
trained group (Fig. 5 B–D). In the control group, no such inverse
correlations between IL-10 and proinflammatory cytokines were
observed. In fact, we found significant positive correlations be-
tween preceding TNF-αand IL-6 levels on the one hand and IL-10
levels at later time points (TNF-α
T=1
vs. IL-10
T=2
:r
s
=0.59,
P=0.045 and IL-6
T=1.5
vs. IL-10
T=2
:r
s
=0.60, P=0.039).
Discussion
Herein, we show that a short-term training program and prac-
ticing breathing techniques learned during this training program
results in release of epinephrine, induction of early anti-
inflammatory IL-10 production, and consequently attenuation
of the proinflammatory innate immune response during experi-
mental human endotoxemia. Also, trained individuals experi-
enced fewer endotoxemia-associated flu-like symptoms, and a
more swift normalization of fever and cortisol levels, which are
likely the result of the attenuated proinflammatory response.
This study demonstrates that the in vivo innate immune re-
sponse can be voluntarily influenced in a nonpharmacological
20
te (bpm)
A
hyper-
venti-
lation
retention
hyper-
venti-
lation
retention
hyper-
venti-
lation
retention
0
10
Respiratory rat
100
)
60
80
40
O2Saturation (%)
B
80
100
120
Heart rate (bpm)
40
C
60
H
100
150
MAP (mmHg)
D
tilation
ntion
tilation
ntion
tilation
ntion
0200 400 600
Time (s)
50
M
100 300 500 700
Start
End
hyperven
End reten
End
hyperven
End reten
End
hyperven
End reten
pH 7.40 7.66 7.44 7.67 7.46 7.75 7.50
pC02 (kPa) 4.49 2.11 4.01 2.03 3.76 1.69 3.48
pO2 (kPa) 165 220 56 229 48 226 34
pO2 (kPa) 16.5 22.0 5.6 22.9 4.8 22.6 3.4
HCO3-
(mmol/l) 20.9 18.0 20.3 17.6 20.2 17.4 20.4
Lactate
(mmol/l) 0.69 0.86 0.69 1.03 0.77 1.16 0.91
Fig. 2. Cardiorespiratory and biochemical changes during cyclic hyperven-
tilation and breath retention in a representative subject of the trained
group. (A) The respiratory rate alternately increased to around 20 breaths
per minute (bpm) for several minutes, and then dropped to zero during
voluntary breath retention. These cyclic changes in respiration resulted in
profound changes in (B) oxygen saturation, (C) heart rate, and (D) mean
arterial pressure. The data depicted were sampled from the monitor every
10 s. At the end of each hyperventilation phase and breath retention phase,
an arterial blood sample was drawn for arterial blood gas analysis, of which
the results are listed in the table below D. The cycles of hyper/hypo-
ventilation in this particular subject can be viewed in Movie S2.
Ae
p
ine
p
hrine
nmol/L nore
p
ine
p
hrine
nmol/L
B
control trained
pp
2
3
pp
2
3
p=0.01
p<0.0001
1
2
1
2
00
-1 0 1 2 3 4 5 6 7 8 -1 0 1 2 3 4 5 6 7 8
cortisol
1.5
μmol/L
dopamine
0.3
nmol/L
CD
p
<0.0001
p
=0.001
1
01
0.2
p
p
0
0.5
0
0
.
1
0
-1 0 1 2 3 4 5 6 7 8
time (hours post-LPS)
-1 0 1 2 3 4 5 6 7 8
time (hours post-LPS)
Fig. 3. Plasma cathecholamine concentrations and serum cortisol concen-
trations during experimental endotoxemia in control and trained subjects.
(A) Plasma epinephrine. (B) Plasma norepinephrine. (C) Plasma dopamine.
(D) Serum cortisol. Data are expressed as mean ±SEM of 12 subjects per
group. Gray box indicates period in which the trained subjects practiced
their learned breathing techniques. Pvalues between groups were calcu-
lated using repeated measures two-way analysis of variance (ANOVA, in-
teraction term).
Kox et al. PNAS
|
May 20, 2014
|
vol. 111
|
no. 20
|
7381
IMMUNOLOGY

manner through voluntary activation of the sympathetic nervous
system. In accordance with the data of our control group, human
endotoxemia in itself has been shown previously to result in in-
creased levels of epinephrine (12). However, in trained individ-
uals epinephrine levels were already profoundly increased 30
min after start of practicing the breathing techniques, before LPS
administration. Epinephrine levels in trained individuals were
even higher than those reported in a recent study in which acute
stress elicited by a bungee jump was found to suppress cytokine
production by leukocytes ex vivo stimulated with LPS (13).
As norepinephrine, dopamine, and cortisol levels were not in-
creased in the training group, it appears that the techniques
predominantly result in stimulation of the sympathetic input to
the adrenal medulla, because this is the most abundant source of
epinephrine in the body and epinephrine-producing chromaffin
cells in the adrenal medulla are much more abundant than those
producing norepinephrine (14).
The observed potentiating effects on anti-inflammatory IL-10
production as well as the attenuation of proinflammatory cyto-
kine levels are in agreement with a previously performed study,
where epinephrine was i.v. administered before LPS in healthy
volunteers and resulted in early and increased IL-10 production
(7), and with studies showing that pretreatment with IL-10 results
in attenuation of the proinflammatory response in healthy volun-
teers (15, 16). In the training group, strong inverse correlations
between IL-10 levels at an early time point and later-occurring
peak levels of the proinflammatory mediators were found,
whereas in the control group the opposite was found: positive
correlations between preceding levels of proinflammatory
mediators with the later-occurring peak levels of IL-10. These
findings indicate that the proinflammatory response drives IL-10
production in the control group, whereas the epinephrine-in-
duced early increase in IL-10 production inhibits proinflam-
mation in the trained group. The early increases in lymphocytes
and subsequent higher concentrations of circulating neutrophils
in the training group compared with the control group can also
be attributed to the elevated epinephrine levels found in trained
individuals, as catecholamines induce leukocytosis characterized
by an initial lymphocytosis followed by an increase of other sub-
populations (17). Furthermore, similar changes in leukocyte
counts were previously observed during voluntary hyperventila-
tion (18). Our study is limited by the fact that we did not measure
specific leukocyte subtypes such as CD3, CD4, and CD8 num-
bers as well as B cells, dendritic cells, and natural killer (NK)
cells, some of which have been shown to be specifically altered by
catecholamines and/or stress (19, 20).
It appears that mainly the breathing techniques used by the
trained individuals account for the increase in epinephrine and
subsequent attenuation of the inflammatory response. A limita-
tion of our study design is that it does not allow the identification
of the particular component of the practiced breathing exercises
that results in increased epinephrine levels. Furthermore, the
effect of the length of the training and the length of propensity
for altered responses after training has yet to be determined.
However, the effects on epinephrine are likely a consequence of
both the hyperventilation phase and hypoxia due to breath re-
tention, as both have been demonstrated to increase epinephrine
levels (18, 21–24). The hyperventilation-induced increase in
epinephrine was shown to be dependent on decreased levels of
bicarbonate, as hyperventilation combined with bicarbonate
infusion (resulting in hypocapnia and alkalosis, but normal
bicarbonate levels) nullified epinephrine increase (24). In
TNF-α
400
500
pg/mL AUC TNF-α
8
10
control trained
control trained
p
=0.02
AB
0
100
200
300
400
IL-6
500
pg/mL
0
2
4
6
8
10
AUC IL-6
p
CD
-1 0 1 2 3 4 5 6 7 8
0
100
200
300
400
IL-8pg/mL
0
2
4
6
8
AUC IL-8
p=0.01
EF
-1 0 1 2 3 4 5 6 7 8
0
100
200
300
400
500
600
0
2
4
6
8
10
12
p=0.004
-1 0 1 2 3 4 5 6 7 8
IL-10
0
200
400
600
800
1000
pg/mL
0
4
8
12
16
20
AUC IL-10
p=0.01
H
G
1
0
1
2
3
4
5
6
7
8
-
1
0
1
2
3
4
5
6
7
8
time (hours post-LPS)
Fig. 4. Plasma cytokine concentrations during endotoxemia in control and
trained subjects. (A,C,E, and G) Median values of pro- (TNF-α, IL-6, and IL-8)
and anti-inflammatory (IL-10) cytokines (n=12 per group). (B,D,F,andH)
Median ±interquartile range of area under curve (AUC) of pro- (TNF-α, IL-6,
and IL-8) and anti-inflammatory (IL-10) cytokines (n=12 per group; unit:
×10
4
pg/mL·h). Pvalues were calculated using Mann–Whitney utests.
800
1000
TNF-α T=1.5 hours (pg/mL)
800
1000
IL-10 T=1 hours (pg/mL)
r
=
0.82
r
=-
0.71
AB
400
600
800
400
600
800
r0.82
p=0.001
r
0.71
p=0.01
0
200
400
0
200
400
IL
8T 2h ( / L)
0
0 1 2 3 4 5 6
0
epinephrine T=0 hours (nmol/L)
IL
6T 2h ( / L)
0 200 400 600 800 1000
IL-10 T=1 hours (pg/mL)
800
1000
IL
-
8T
=
2h
ours
(
pg
/
m
L)
800
1000
IL
-
6T
=
2h
ours
(
pg
/
m
L)
r=-0.71
001
r=-0.59
p=0 045
D
C
400
600
400
600
p=
0
.
01
p=0
.
045
0
200
0
200
0 200 400 600 800 1000
IL-10 T=1 hours (pg/mL)
0 200 400 600 800 1000
IL-10 T=1 hours (pg/mL)
Fig. 5. Correlations in trained individuals. (A) Correlation between peak
plasma levels of epinephrine (at T=0 h) and plasma levels of the anti-
inflammatory cytokine IL-10 at T=1h.(B) Correlation between plasma levels
of the anti-inflammatory cytokine IL-10 at T=1 h and peak plasma levels of
the proinflammatory cytokine TNF-α(at T=1.5 h). (C) Correlation between
plasma levels of the anti-inflammatory cytokine IL-10 at T=1 h and peak
plasma levels of the proinflammatory cytokine IL-6 (at T=2 h). (D) Corre-
lation between plasma levels of the anti-inflammatory cytokine IL-10 at T=1h
and peak plasma levels of the proinflammatory cytokine IL-8 (at T=2h).
Rand Pvalues were calculated using Spearman correlation.
7382
|
www.pnas.org/cgi/doi/10.1073/pnas.1322174111 Kox et al.

concordance, in the present study, bicarbonate levels were sig-
nificantly lower in the trained subjects during practicing of the
breathing techniques compared with control subjects. The at-
tenuated cytokine response is unlikely to be a direct result from
low pCO
2
and high pH levels because hypocapnic alkalosis, as
opposed to hypercapnic acidosis (25), is not associated with anti-
inflammatory effects. Therefore, epinephrine is the most prob-
able intermediate factor (7). Nevertheless, it cannot be ruled out
that other elements of the training, apart from practicing the
breathing exercises, ultimately affected the LPS-induced in-
nate immune response. For instance, the exposition to extreme
cold and subsequent rewarming during the training sessions might
have resulted ischemic preconditioning and/or release of dan-
ger associated molecular patterns (DAMPs), which could re-
sult in a tolerant state toward a subsequent LPS challenge.
It remains to be determined whether the results of this study
using an acute model of inflammation in healthy volunteers can
be extrapolated to patients with chronic autoimmune diseases.
For instance, chronic stress might be harmful in these conditions
due to induction of proinflammatory mediators (26), whereas
bouts of short-term stress, similar to the effects of the training
intervention described in this study, may be beneficial due to
immunosuppressive effects (26). Of interest, the in vivo anti-
inflammatory potential in humans of biologics currently used in
the treatment of rheumatoid arthritis was first established in
proof-of-principle human endotoxemia studies (27, 28), illus-
trating the relevance of the model to investigate novel therapies
for this type of disease.
In conclusion, the present proof-of-principle study demon-
strates that the sympathetic nervous system and immune system
can be voluntarily influenced through practicing techniques that
are relatively easy to learn within a short time frame. It therefore
could have important implications for the treatment of a variety
of conditions associated with excessive or persistent inflammation,
especially auto-immune diseases.
Materials and Methods
Subjects. This parallel randomized controlled study was registered at
ClinicalTrials.gov as NCT01835457. After approval by the local ethics com-
mittee of the Radboud University Nijmegen Medical Centre (CMO 2012/455),
30 healthy, nonsmoking, Dutch male volunteers were included in the trial.
All subjects provided written informed consent and experiments were in
accordance with the Declaration of Helsinki, including current revisions, and
Good Clinical Practice guidelines. Subjects were screened before the start of
the experiment and had a normal physical examination, electrocardio-
graphy, and routine laboratory values. Exclusion criteria were: febrile illness
during the 2 wk before the endotoxemia experiment, taking any pre-
scription medication, history of spontaneous vagal collapse, practicing or
experience with any kind of meditation, or participation in a previous trial
where LPS was administered. The subjects were randomly allocated to the
trained group (n=18) or the control group (n=12) by the opening of
a sealed envelope prepared by a research nurse not involved in the study.
After having fulfilled the training program, 12 of the 18 trained subjects were
randomly assigned to participate in the experimental endotoxemia experi-
ments (further explained in Study Design and Training Procedure below).
Three subjects in the control group that underwent endotoxemia on the
same day and received LPS from the same ampoule were excluded from the
trial and replaced. Their symptoms, temperature rise, hemodynamic re-
sponse, and cytokine response were inconsistent with having received an
adequate dose of 2 ng/kg LPS. Batchwise determination of cytokine levels
revealed exceptionally low levels in all three subjects: Their peak cytokine
response (TNF-αand IL-6) was less than half of that of the lowest recorded in
a cohort of 112 healthy male subjects that previously underwent experi-
mental endotoxemia (10) and peaked at atypical time points (subject 1, TNF-
α=39 pg/mL at 4 h after LPS administration and IL-6 =27 pg/mL at 4 h after
LPS administration; subject 2, TNF-α=32 pg/mL at 3 h after LPS adminis-
tration and IL-6 =31 pg/mL at 3 h after LPS administration; and subject 3,
TNF-α=9 pg/mL at 2 h after LPS administration and IL-6 =7 pg/mL at 3 h
after LPS administration). Therefore, a endotoxin dose administration error
was assumed and the subjects were replaced.
Study Design and Training Procedure. The study was sequentially conducted in
two identical blocks, each consisting of nine subjects in the trained group (of
which six finally participated in the endotoxemia experiments, further
explained below) and six subjects in the control group. This design was chosen
to minimize the bias due to differences in the interval between the end of
the training period and the endotoxemia experiments. As the aim of our
study was to investigate the effects of the training intervention on the innate
immune response in a standardized model of systemic inflammation, we
did not assess the effects of the training intervention on immune system
parameters in the absence of endotoxemia. A schematic overview of the
study design (one block) is depicted in Fig. S3. The trained group was trained
by Dutch individual Wim Hof and three trainers who previously received an
instructor course by Wim Hof to become a trainer. A medical doctor of the
study team (L.T.v.E.) and the principal investigator (M.K.) were present
during all training sessions (in Poland and in The Netherlands), and during
the experimental endotoxemia experiments. The first 4 d of the training
program took place in Poland and were most intensive. The program con-
sisted of three main elements: meditation, exposure to cold, and breathing
techniques (see Movie S1 for an impression of the training program).
i) Meditation, so-called “third eye meditation,”a form of meditation in-
cluding visualizations aimed at total relaxation.
ii) During the training, subjects voluntarily exposed themselves to cold in
several ways: standing in the snow barefoot for up to 30 min and lying
bare chested in the snow for 20 min; daily dipping/swimming in ice-cold
water (0–1 °C) for up to several minutes (including complete sub-
mersions); and hiking up a snowy mountain (elevation: 1,590 m) bare
chested, wearing nothing but shorts and shoes at temperatures ranging
from −5to−12 °C (wind chill: −12 to −27 °C).
iii) Breathing techniques, consisting of two exercises. In the first exercise
subjects were asked to hyperventilate for an average of 30 breaths. Sub-
sequently, the subjects exhaled and held their breath for ∼2–3 min (“re-
tention phase”). The duration of breath retention was entirely at the
discretion of the subject himself. Breath retention was followed by
a deep inhalation breath, that was held for 10 s. Subsequently a new
cycle of hyper/hypoventilation began. The second exercise consisted of
deep inhalations and exhalations in which every inhalation and exhala-
tion was followed by breath holding for 10 s, during which the subject
tightened all his body muscles. These two breathing exercises were also
performed during the endotoxemia experiments. Additional element of
the training program consisted of strength exercises (e.g., push-ups and
yoga balance techniques).
After returning from Poland, the subjects practiced the techniques they
learned daily by themselves at home (2–3 h/d; cold exposure was achieved
through taking cold showers) until the endotoxemia experiment day (5–9d
later). In addition, a final group training took place and at the end of this
day, six of the nine trained subjects (in each block) were randomly selected
for participation in the endotoxemia experiments, using the sealed envelope
method. This selection was performed to allow for subject replacement in
case of an adverse event or illness in one of the trained subjects selected
for the endotoxemia experiments. The selected subjects practiced in a final
training session led by Wim Hof on the day before the endotoxemia ex-
periment day. Wim Hof was present to coach the subjects during the
endotoxemia experiment days during the 3 h that the subjects in the trained
group practiced the learned techniques. The control group did not undergo
any training procedures throughout the study period.
Experimental Human Endotoxemia. Subjects refrained from caffeine- or
alcohol-containing substances 24 h before the start of the experiment, and
food 10 h before the start of the endotoxemia experiment. The experiments
were performed at the research unit of the intensive care department. The
procedures on the endotoxemia experiment day are depicted in Fig. S4.
Purified lipopolysaccharide (LPS, US Standard Reference Endotoxin Escherichia
coli O:113) obtained from the Pharmaceutical Development Section of the
National Institutes of Health,supplied as a lyophilized powder, was recon-
stituted in 5 mL saline 0.9% for injection and vortex mixed for at least 20 min
after reconstitution. The LPS solution was administered as an i.v. bolus in-
jection at a dose of 2 ng/kg body weight in 1 min at T=0 h. A cannula was
placed in an antecubital vein to permit infusion of 0.9% NaCl solution; the
subjects received 1.5 L 0.9% NaCl during 1 h starting 1 h before endotoxin
infusion (prehydration) as part of our standard endotoxemia protocol (29),
followed by 150 mL/h until 6 h after endotoxin infusion and 75 mL/h until
the end of the experiment. The radial artery was cannulated using a
20-gauge arterial catheter (Angiocath; Becton Dickinson) and connected
to an arterial pressure monitoring set (Edwards Lifesciences) to allow the
Kox et al. PNAS
|
May 20, 2014
|
vol. 111
|
no. 20
|
7383
IMMUNOLOGY

continuous monitoring of blood pressure and blood sampling. Heart rate
(three-lead electrocardiogram), blood pressure, respiratory rate, and oxygen
saturation (pulse oximetry) data were recorded from a Philips MP50 patient
monitor every 30 s by a custom in-house–developed data recording system,
starting 1 h before administration of LPS until discharge from the intensive
care unit 8 h after LPS administration. Body temperature was measured
using an infrared tympanic thermometer (FirstTemp Genius 2; Sherwood
Medical). LPS-induced flu-like symptoms (headache, nausea, shivering, muscle
and back pain) were scored every 30 min on a six-point Likert scale (0 =no
symptoms, 5 =worst ever experienced), resulting in a total score of 0–25.
Thirty minutes before LPS administration (T=−0.5 h), subjects in the
trained group started the first breathing technique (hyper/hypoventilation
cycles, see Movie S2) until T=1 h, followed by the second breathing tech-
nique (deep inhalation and exhalation in combination with tightening
muscles) until T=2.5 h. Afterward, the subjects stopped practicing all of the
techniques. The control group did not practice any techniques throughout
the endotoxemia experiment day.
Blood Gas Parameters. Blood gas parameters were analyzed in lithium heparin
anticoagulated arterial blood using CG4+cartridges and a point-of-care i-STAT
blood gas analyzer (Abbott).
Catecholamines. Blood was collected into chilled lithium-heparin tubes and
were immediately placed on ice and centrifuged at 2,000 ×gfor 10 min at
4 °C after which plasma was stored at −80 °C until analysis. Plasma norepi-
nephrine, epinephrine, and dopamine concentrations were measured using
routine analysis methods also used for patient samples (HPLCy with fluoro-
metric detection, as described previously) (30).
Cortisol. Blood was collected in serum-separating tubes and was allowed to
clot at room temperature for a minimum of 30 min. Subsequently, samples
were centrifuged at 2,000 ×gfor 10 min at 4 °C, after which serum was
stored at −80 °C until analysis. Cortisol levels were determined using a
routine analysis method also used for patient samples (electrochemiluminescent
immunoassay on a Modular Analytics E170 (Roche Diagnostics).
Leukocyte Counts and Differentiation. Analysis of leukocyte counts and dif-
ferentiation was performed in EDTA anticoagulated blood using routine
analysis methods also used for patient samples (flow cytometric analysis on
a Sysmex XE-5000.
Plasma Cytokines. EDTA anticoagulated blood was centrifuged immediately
at 2,000 ×gfor 10 min at 4 °C after which plasma was stored at −80 °C until
analysis. Concentrations of TNF-α, IL-6, IL-8, and IL-10 were measured using
a simultaneous Luminex assay according to the manufacturer’s instructions
(Milliplex; Millipore). IL-1β, TGF-β, and leptin were measured using ELISAs
according to the manufacturer’s instructions (IL-1βand TGF-β, Quantikine
and leptin, Duoset; both R&D Systems).
Calculations and Statistical Analysis. Data are represented as median and
interquartile range/range or mean and SEM based on their distribution
(calculated by the Shapiro–Wilk test). Statistical tests used are indicated in
the figure/table legends or text. Spearman’s correlation was used. A Pvalue
of <0.05 was considered statistically significant. Statistical calculations were
performed using Graphpad Prism version 5.0 (GraphPad Software).
ACKNOWLEDGMENTS. The authors thank K. Mostard, K. Pijper, D. Bernard,
and E. Hof for help during the training sessions; the Radboud University
Nijmegen Sports Centre for providing space for training days in The
Netherlands; and the research nurses of the Radboud University Medical
Centre Intensive Care Unit for help during the endotoxemia experiments.
This study was supported by a Serendipity Grant from Reumafonds (www.
reumafonds.nl).
1. Buch MH, Emery P (2011) New therapies in the management of rheumatoid arthritis.
Curr Opin Rheumatol 23(3):245–251.
2. Cheifetz AS (2013) Management of active Crohn disease. JAMA 309(20):2150–2158.
3. Hansel TT, Kropshofer H, Singer T, Mitchell JA, George AJ (2010) The safety and side
effects of monoclonal antibodies. Nat Rev Drug Discov 9(4):325–338.
4. Marra CA, Bansback N, Anis AH, Shojania K (2011) Introduction to economic modeling
for clinical rheumatologists: Application to biologic agents in rheumatoid arthritis.
Clin Rheumatol 30(Suppl 1):S9–S18.
5. Siegmund B, Eigler A, Hartmann G, Hacker U, Endres S (1998) Adrenaline enhances
LPS-induced IL-10 synthesis: Evidence for protein kinase A-mediated pathway. Int J
Immunopharmacol 20(1–3):57–69.
6. van der Poll T, Jansen J, Endert E, Sauerwein HP, van Deventer SJ (1994) Noradren-
aline inhibits lipopolysaccharide-induced tumor necrosis factor and interleukin
6 production in human whole blood. Infect Immun 62(5): 2046–2050.
7. van der Poll T, Coyle SM, Barbosa K, Braxton CC, Lowry SF (1996) Epinephrine inhibits
tumor necrosis factor-alpha and potentiates interleukin 10 production during human
endotoxemia. J Clin Invest 97(3):713–719.
8. Alvarez SM, et al. (2007) Low-dose steroid alters in vivo endotoxin-induced systemic
inflammation but does not influence autonomic dysfunction. J Endotoxin Res 13(6):
358–368.
9. van der Poll T, Barber AE, Coyle SM, Lowry SF (1996) Hypercortisolemia increases
plasma interleukin-10 concentrations during human endotoxemia—a clinical research
center study. J Clin Endocrinol Metab 81(10):3604–3606.
10. Kox M, et al. (2012) The influence of concentration/meditation on autonomic nervous
system activity and the innate immune response: A case study. Psychosom Med 74(5):
489–494.
11. van Eijk LT, et al. (2007) Gender differences in the innate immune response and
vascular reactivity following the administration of endotoxin to human volunteers.
Crit Care Med 35(6):1464–1469.
12. Sayk F, et al. (2008) Endotoxemia causes central downregulation of sympathetic va-
somotor tone in healthy humans. Am J Physiol Regul Integr Comp Physiol 295(3):
R891–R898.
13. van Westerloo DJ, et al. (2011) Acute stress elicited by bungee jumping suppresses
human innate immunity. Mol Med 17(3–4):180–188.
14. Wong DL (2003) Why is the adrenal adrenergic? Endocr Pathol 14(1):25–36.
15. Kumar A, et al. (2005) Interleukin-10 blunts the human inflammatory response to
lipopolysaccharide without affecting the cardiovascular response. Crit Care Med
33(2):331–340.
16. Pajkrt D, et al. (1997) Attenuation of proinflammatory response by recombinant
human IL-10 in human endotoxemia: Effect of timing of recombinant human IL-10
administration. J Immunol 158(8):3971–3977.
17. Benschop RJ, Rodriguez-Feuerhahn M, Schedlowski M (1996) Catecholamine-induced
leukocytosis: Early observations, current research, and future directions. Brain Behav
Immun 10(2):77–91.
18. Staubli M, Vogel F, Bartsch P, Fluckiger G, Ziegler WH (1994) Hyperventilation-
induced changes of blood cell counts depend on hypocapnia. Eur J Appl Physiol Occup
Physiol 69(5):402–407.
19. Huang CJ, Webb HE, Garten RS, Kamimori GH, Acevedo EO (2010) Psychological stress
during exercise: Lymphocyte subset redistribution in firefighters. Physiol Behav
101(3):320–326.
20. Schedlowski M, et al. (1993) Catecholamines induce alterations of distribution and
activity of human natural killer (NK) cells. J Clin Immunol 13(5):344–351.
21. Laderach H, Straub W (2001) Effects of voluntary hyperventilation on glucose, free
fatty acids and several glucostatic hormones. Swiss Med Wkly 131(1–2):19–22.
22. Mantysaari M, et al. (2011) Unaltered blood coagulation and platelet function in
healthy subjects exposed to acute hypoxia. Aviat Space Environ Med 82(7):699–703.
23. Oltmanns KM, et al. (2006) Acute hypoxia decreases plasma VEGF concentration in
healthy humans. Am J Physiol Endocrinol Metab 290(3):E434–E439.
24. Krapf R, Caduff P, Wagdi P, Staubli M, Hulter HN (1995) Plasma potassium response to
acute respiratory alkalosis. Kidney Int 47(1):217–224.
25. Ijland MM, Heunks LM, van der Hoeven JG (2010) Bench-to-bedside review: Hyper-
capnic acidosis in lung injury–from ’permissive’to ’therapeutic’.Crit Care 14(6):237.
26. Cutolo M, et al. (2000) The hypothalamic-pituitary-adrenal and gonadal axes in
rheumatoid arthritis. Ann N Y Acad Sci 917:835–843.
27. Granowitz EV, et al. (1993) Hematologic and immunomodulatory effects of an in-
terleukin-1 receptor antagonist coinfusion during low-dose endotoxemia in healthy
humans. Blood 82(10):2985–2990.
28. Suffredini AF, et al. (1995) Effects of recombinant dimeric TNF receptor on human
inflammatory responses following intravenous endotoxin administration. J Immunol
155(10):5038–5045.
29. Dorresteijn MJ, et al. (2005) Iso-osmolar prehydration shifts the cytokine response
towards a more anti-inflammatory balance in human endotoxemia. J Endotoxin Res
11(5):287–293.
30. Willemsen JJ, et al. (1995) Highly sensitive and specific HPLC with fluorometric de-
tection for determination of plasma epinephrine and norepinephrine applied to ki-
netic studies in humans. Clin Chem 41(10):1455–1460.
7384
|
www.pnas.org/cgi/doi/10.1073/pnas.1322174111 Kox et al.