Short-term cold acclimation improves insulin sensitivity in patients with type 2 diabetes mellitus

Abstract

Cold exposure may be a potential therapy for diabetes by increasing brown adipose tissue (BAT) mass and activity. Here we report that 10 d of cold acclimation (14-15 °C) increased peripheral insulin sensitivity by ∼43% in eight type 2 diabetes subjects. Basal skeletal muscle GLUT4 translocation markedly increased, without effects on insulin signaling or AMP-activated protein kinase (AMPK) activation and only a minor increase in BAT glucose uptake.

Full text

© 2015 Nature America, Inc. All rights reserved.
B r i e f c o m m u n i c at i o n s
nature medicine 1
Cold-induced adaptive thermogenesis has gained considerable interest
due to the ‘rediscovery’ of BAT in adult humans in 2009 (refs. 1–3). In
response to these findings, activation of human BAT has been proposed
as a potential treatment for type 2 diabetes (T2D) because such activation
leads to the dissipation of energy as heat, which is expected to not only
increase energy expenditure but also boost the oxidation of triglycerides
and glucose as substrates. Thus it is expected that patients with T2D
treated in this way would display not only weight loss but also improve-
ments in lipid and glucose homeostasis
4,5
. Indeed, in rodents enhanced
BAT activity has been shown to prevent the development of obesity and
diabetes under obesogenic conditions
6
. We
7
and others
8,9
have recently
shown that chronic cold exposure (cold acclimation) leads to increases
in both BAT quantity and activity in humans, suggesting that BAT can
be recruited upon cold acclimation. Here we investigated whether cold
acclimation for 10 d could be used to recruit BAT and improve glucose
homeostasis in eight individuals with T2D (Supplementary Table 1).
Body weight and clinical blood parameters were unaffected by cold accli-
mation (Supplementary Table 2), and thyroid-stimulating hormone
(TSH) concentrations decreased while free thyroxine (T4) concentra-
tions increased after cold acclimation (Supplementary Table 2).
As a measure of BAT activity, we determined acute cold-induced
2-deoxy-2-[
18
F]fluoro--glucose ([
18
F]FDG) uptake by positron emission
tomography–computed tomography (PET-CT) scanning (Fig. 1a,b). Upon
cold acclimation, [
18
F]FDG uptake (standardized uptake values (SUV)
max
(data not shown) and SUV
mean
) in the supraclavicular BAT region
increased in all subjects (Fig. 1c). Also, CT radio density (in Hounsfield
units) of the BAT depot increased in 7 out of 8 subjects after cold acclima-
tion (Fig. 1d), suggesting that triglyceride content was reduced, indicative
of BAT activation
10
. However, we noted that even after cold acclimation
the observed [
18
F]FDG uptake values in BAT in the T2D patients in our
study were very low compared with those in the literature, which were for
young, healthy subjects
7,11
. We also measured cold-induced [
18
F]FDG
uptake in several muscle groups from the cold-acclimated patients. Skeletal
muscle [
18
F]FDG uptake tended to increase in most of the investigated
muscle groups, and the resulting average skeletal muscle [
18
F]FDG uptake
was indeed increased after cold acclimation (Fig. 1e,f).
We next examined whether glucose uptake in other adipose tissue
depots increased upon cold acclimation, which could indicate ‘beig-
ing’ (conversion of white adipose tissue (WAT) into adipose tissue
containing UCP1-expressing adipocytes). However, [
18
F]FDG uptake
was unaffected by cold acclimation in WAT depots, such as subcutane-
ous and visceral fat (Fig. 1e). We also performed microarray analysis
in subcutaneous WAT biopsies from the human subjects taken before
and after 10 d of cold acclimation and compared it to WAT obtained
from cold-acclimated mice. No signs of beiging could be detected in
the human WAT samples, and most genes identified as beiging mark-
ers were unaffected by the cold acclimation (Fig. 1g). Together these
results indicate that after cold acclimation glucose uptake specifically
increases in the supraclavicular BAT depot upon acute cold exposure,
but BAT activity remains very low in these individuals.
We previously showed that activation of BAT by acute mild cold
exposure correlates with cold-induced nonshivering thermogenesis
(NST)
7,12
and that NST increases upon cold acclimation in lean, healthy
subjects
7
. Here we found that after, but not before, cold acclimation,
acute mild cold exposure increased energy expenditure compared
to energy expenditure measured in the thermoneutral zone (TNZ)
(Fig. 1h), indicating increased capacity for NST upon cold acclimation.
To investigate adaptation to the cold, we determined subjective ratings
for sensation, thermal comfort and shivering. All ratings were improved
at day 10 of cold acclimation compared to day 3, which was the first
day with 6 h of cold exposure (Supplementary Fig. 1). Subjective
ratings of hunger, fullness and satiety before and after lunch did not
differ between day 10 and day 3 (data not shown). Mean skin and core
temperatures, skin blood flow and blood pressure during cold exposure
were unaffected by cold acclimation (Supplementary Table 3).
To investigate whether cold acclimation improved glucose homeo-
stasis in individuals with T2D, we used hyperinsulinemic-euglycemic
clamps, the gold-standard technique, to determine insulin sensitivity
before and after cold acclimation. Of note, clamps were performed
Short-term cold acclimation
improves insulin sensitivity in
patients with type 2 diabetes mellitus
Mark J W Hanssen
1
, Joris Hoeks
1
, Boudewijn Brans
2
,
Anouk A J J van der Lans
1
, Gert Schaart
3
, José J van den Driessche
1
,
Johanna A Jörgensen
1
, Mark V Boekschoten
4
, Matthijs K C Hesselink
3
,
Bas Havekes
5
, Sander Kersten
4
, Felix M Mottaghy
2,6
,
Wouter D van Marken Lichtenbelt
1,7
& Patrick Schrauwen
1,7
Cold exposure may be a potential therapy for diabetes by
increasing brown adipose tissue (BAT) mass and activity. Here
we report that 10 d of cold acclimation (14–15 °C) increased
peripheral insulin sensitivity by ~43% in eight type 2 diabetes
subjects. Basal skeletal muscle GLUT4 translocation
markedly increased, without effects on insulin signaling or
AMP-activated protein kinase (AMPK) activation and only a
minor increase in BAT glucose uptake.
1
Department of Human Biology, School of Nutrition and Translational Research in Metabolism, Maastricht University Medical Center, Maastricht, the Netherlands.
2
Department of Nuclear Medicine, Maastricht University Medical Center, Maastricht, the Netherlands.
3
Department of Human Movement Sciences, School of Nutrition
and Translational Research in Metabolism, Maastricht University Medical Center, Maastricht, the Netherlands.
4
Nutrition, Metabolism and Genomics group, Division
of Human Nutrition, Wageningen University, Wageningen, the Netherlands.
5
Department of Internal Medicine, Division of Endocrinology and Metabolism, School of
Nutrition and Translational Research in Metabolism, Maastricht University Medical Center, Maastricht, the Netherlands.
6
Department of Nuclear Medicine, University
Hospital Rheinisch-Westfälische Technische Hochschule (RWTH) Aachen, Aachen, Germany.
7
These authors contributed equally to the study. Correspondence should
be addressed to P.S. (p.schrauwen@maastrichtuniversity.nl).
Received 7 January; accepted 27 May; published online 6 July 2015; doi:10.1038/nm.3891

© 2015 Nature America, Inc. All rights reserved.
2 nature medicine
B r i e f c o m m u n i c at i o n s
at thermoneutrality. Insulin sensitivity was markedly increased after
cold acclimation as indicated by on average a 43% increase in the glu-
cose infusion rate during the clamp (Fig. 2a). In fact, the increase in
insulin sensitivity even exceeded the improvements in insulin sensitiv-
ity that are seen upon long-term exercise training
13
, which is generally
considered to be the best strategy to prevent or treat T2D. Improved
whole-body insulin sensitivity was mainly accounted for by a marked
increase in insulin-stimulated glucose disposal (Fig. 2b), reflecting
improvement of peripheral insulin sensitivity. Basal endogenous
glucose production (EGP) was not affected by cold acclimation,
but insulin-induced suppression of EGP, reflecting hepatic insulin
sensitivity, tended to improve upon cold acclimation (Fig. 2c and
Supplementary Table 4). After cold acclimation nonesterified fatty
acid concentrations were lower during the clamp (Fig. 2d), which indi-
cated improved adipose tissue insulin sensitivity
14
. Metabolic flexibility,
the ability to shift from predominantly fat oxidation in the fasted state
to glucose oxidation upon insulin stimulation, also tended to
increase (data not shown), most likely as a reflection of improved
insulin-stimulated glucose uptake into skeletal muscle
15
.
Because the increase in insulin sensitivity was much more pronounced
than anticipated and the activity of BAT remained very low in the treated
individuals, we performed detailed analyses of skeletal muscle biopsies
obtained after an overnight fast, both before and after 10 d of cold accli-
mation. As mitochondrial function is a main determinant of insulin
sensitivity in individuals with T2D
16
, we first examined whether cold
acclimation altered skeletal muscle mitochondrial oxidative capacity
and/or leak respiration. However, none of the mitochondrial respiration
states were affected by cold acclimation (Supplementary Fig. 2a–e).
Oxidative phosphorylation complexes and PGC-1α protein content,
markers of mitochondrial content and biogenesis, respectively, were
unaffected by cold acclimation (Supplementary Fig. 2f,g). Skeletal mus-
cle fat content was also unaffected by cold acclimation (Supplementary
Fig. 2h). It was recently suggested that sarcolipin may regulate futile Ca
2+
cycling in skeletal muscle and contribute to nonshivering thermogen-
esis
17
. However, the protein contents of sarcolipin and the sarcoendo-
plasmic reticulum calcium-transporting ATPase (Serca) isoform Serca2
were unchanged upon cold acclimation (Supplementary Fig. 2i,j),
although sarcolipin protein expression correlated with NST after cold
acclimation (NST (J s
−1
kg
−1
), r = 0.893, P = 0.007).
Skeletal muscle insulin sensitivity can be more directly affected
by enhancement of the insulin-signaling cascade or by activation
of AMPK
18
. We therefore examined biopsies taken before the
hyperinsulinemic-euglycemic clamp to determine whether insulin
signaling in skeletal muscle was affected by cold acclimation. Notably,
phosphorylation of the serine-threonine kinase AKT at both Thr308 (data
not shown) and Ser473 (Fig. 2e), as well as total AKT (data not shown),
were unaffected by cold acclimation. Furthermore, phosphorylation
of the Rab GTPase–activating protein AS160 was also unaffected by
cold acclimation (Fig. 2f). We subsequently examined whether altered
AMPK activation could contribute to the enhanced insulin sensitivity.
However, total AMPK content (data not shown), phosphorylation
of AMPK (p-AMPK; from 1,600 ± 408 to 1,598 ± 124 AU, n = 7,
P > 0.05; data are mean ± s.e.m.) and p-AMPK/AMPK ratio (Fig. 2g)
were unaffected by cold acclimation. In skeletal muscle, GLUT1 and
GLUT4 are both involved in glucose uptake. Notably, while GLUT1
protein content was unaffected by cold acclimation (from 0.72 ± 0.11
to 0.63 ± 0.08 AU, n = 7, P > 0.05; data are mean ± s.e.m.), we observed
that total GLUT4 protein content tended to increase after 10 d of cold
acclimation (Fig. 2h). Subsequently, we performed immunostain-
ing of GLUT4 on muscle biopsy sections taken before the clamp to
investigate the subcellular distribution of GLUT4 within the muscle
cells. Whereas before cold acclimation GLUT4 was evenly distributed
over the muscle cell, the pattern was clearly different after cold accli-
mation, with pronounced staining of GLUT4 at the cell membrane
(Fig. 2i)—indicative of translocation of GLUT4 to the membrane to
facilitate glucose uptake—occurring even under basal conditions. We
quantified GLUT4 localization, which revealed an on average 60% enrich-
ment of GLUT4 in cell membranes after cold acclimation (Fig. 2j).
The observation of increased GLUT4 in the membrane is consistent
with the higher [
18
F]FDG uptake in skeletal muscle after cold acclimation
(Fig. 1e,f). Therefore, our data suggest that cold acclimation leads to an
enrichment of GLUT4 at the sarcolemma, which may facilitate the uptake
of glucose. This GLUT4 translocation could not be explained by AMPK
activation or improved insulin signaling. Also, subjects did not report
Figure 1 Glucose uptake in the supraclavicular BAT region before and after
cold acclimation and WAT beiging. (a) [
18
F]FDG-PET images after acute cold
exposure in one representative individual of the eight in the study before
(left) and after (right) cold acclimation. Black arrows indicate supraclavicular
BAT activity. (b) Transversal CT, PET and PET-CT fusion slices of the
supraclavicular region showing [
18
F]FDG uptake in BAT locations (white
arrows) in the same individual as in a. (c,d) Cold-induced glucose uptake
(expressed as SUV
mean
) (c) and CT radio density (in Hounsfield units) (d) in
the supraclavicular BAT region in all individual subjects. (e) Cold-induced
glucose uptake in the supraclavicular BAT region, subcutaneous (WAT
subc
)
and visceral WAT (WAT
visc
), liver, brain and upper-body skeletal muscle (SM).
n = 8, *P < 0.05; error bars, mean ± s.e.m. (f) Cold-induced glucose
uptake in individual skeletal muscle groups used to calculate average SM
uptake in e. Er. spinae, erector spinae; Lev. scapulae, levator scapulae.
n = 8; error bars, mean ± s.e.m. (g) Gene expression changes of selected
genes involved in thermogenesis and/or beiging, or suggested as marker
genes for beige adipose tissue, in subcutaneous WAT of human subjects with
T2D (right column) and subcutaneous WAT from mice subjected to either
10 d (left column; GSE51080) or 7 d (middle column; GSE13432) of cold
exposure (~5 °C). P values represent intensity-based moderated t-statistic
(IBMT)-regularized paired t-test raw P values. (h) Energy expenditure in
thermoneutral conditions and upon acute cold exposure. TNZ, thermoneutral
zone. n = 8; comparisons between data before and after cold acclimation
or between thermoneutral and mild cold conditions were analyzed using
Wilcoxon signed-rank tests; *P < 0.05; error bars, mean ± s.e.m.
g
Tmem26
Gene P value
Mouse WAT (10 d)
Mouse WAT (7 d)
Human WAT
Expression
Cited1
Ppargc1a
Fgf21
Tbx1
Shox2
Hoxc8
Hoxc9
Elovl3
Ucp1
Cidea
Prdm16
Ppara
Pparg
Dio2
Tnfrsf9
Signal log
ratio
–4
–2
–1
–0.5
0
0.5
1
2
4
0.06
0.12
0.70
0.001
0.38
0.75
0.15
0.65
0.72
0.53
0.24
0.98
0.88
0.56
0.06
0.11
Low
Low
Low
Low
Low
Low
Low
Low
Low
Low
High
Low
Moderate
Moderate
Moderate
Moderate
Before
a
b
Before After
CT
PET
PET-CT
After
d
c
SUV
mean
Hounsfield units
2.5
1.5
0.5
0.4
0.3
0.2
–100
–120
–140
–160
–180
Before After
Before After
*
*
e
Before
After
BAT
SM
WAT
subc
WAT
visc
Liver
Brain
*
*
SUV
mean
8.0
4.0
1.0
0.5
0
f
1.5
1.2
0.9
0.7
0.6
0.5
0.4
0.3
Deltoid
Biceps
Triceps
Er. spinae
Scalene
Lev. scapulae
Psoas major
SUV
mean
Before
After
TNZ
Mild cold
h
*
1.1
1.0
0.9
0.8
Energy expenditure
(J s
–1
kg
–1
)
Before After

© 2015 Nature America, Inc. All rights reserved.
nature medicine 3
B r i e f c o m m u n i c at i o n s
a
40
*
GIR (µmol kg
–1
min
–1
)
30
20
10
0
Before After
25
*
GIR (µmol kg
–1
min
–1
)
Before After
20
15
10
5
0
–60
% ∆EGP (µmol kg
–1
min
–1
)
Before
After
–70
–80
–90
–100
10
Basal EGP (µmol kg
–1
min
–1
)
c
Before After
8
6
4
2
0
*
j
Before After
GLUT4 intensity
(sarcolemma/cytosol)
2.0
1.5
1.0
0.5
0
150
*
NEFA (µmol/l)
d
Before After
100
50
*
R
d
(µmol kg
–1
min
–1
)
b
Before After
25
NOGD
CHO
ox
20
15
10
5
0
i
Subject 1 Subject 3Subject 2
Before
After
g
Before
After
Before
After
Before After
AMPK
p-AMPK
p-AMPK/AMPK
0.8
0.6
0.4
0.2
0
e
Before After Before After
Before
After
p-AKT
(Ser473)
p-AKT (Ser473)
2.0
1.5
1.0
0.5
0
f
Before After Before After
Before After
p-AS160
p-AS160 (AU)
2.5
2.0
1.5
1.0
0.5
0
h
Before After Before After
Before After
GLUT4
GLUT4 (AU)
1.2
1.0
0.8
0.6
overt shivering in the final days of the cold acclimation and enhanced
skeletal muscle GLUT4 translocation after cold acclimation was observed
in the overnight-fasted, thermoneutral state. Therefore, alternative mecha-
nisms need to be explored: it has recently been proposed that β-adrenergic
stimulation is able to activate GLUT4 translocation
19
, and daily cold
exposure may increase sympathetic activity. Alternatively, it has recently
been shown that BAT may release endocrine factors that can engage
other metabolic tissues
20
, and it would be worth exploring if some of these
endocrine factors have insulin-sensitizing effects on skeletal muscle.
In conclusion, our findings indicate that cold acclimation for 10 d has
very marked positive effects on whole-body and skeletal muscle insulin
sensitivity and thereby provide a new avenue to improve the metabolic
health of patients with T2D.
METHODS
Methods and any associated references are available in the online
version of the paper.
Accession codes. Gene Expression Omnibus: microarray data have
been deposited with accession number GSE67297.
Note: Any Supplementary Information and Source Data files are available in the
online version of the paper.
ACKNOWLEDGMENTS
We thank K. Jardon, E. Broeders, D. Van Moorsel, K. Jansen, M. Visser,
R. Hensgens and R. Wierts (Maastricht University Medical Center) for assistance
during the experiments and H. Aydeniz, E. Kornips, J. Stegen, W. Sluijsmans,
L. Donselaar (Maastricht University Medical Center), W. Wickenhagen (VU
University Medical Center Amsterdam) and M. Ackermans (Academic Medical
Center Amsterdam) for assistance with the biochemical analyses. The technical
support of P. Schoffelen, L. Wouters and M. Souren (Maastricht University Medical
Center) is highly appreciated. This work was supported by the EU FP7 project
DIABAT (HEALTH-F2-2011-278373 to W.D.v.M.L.) and by the Netherlands
Organization for Scientific Research (NWO) (TOP 91209037 to W.D.v.M.L). J.H. is
supported by an NWO Vidi grant for innovative research (grant 917.14.358).
AUTHOR CONTRIBUTIONS
M.J.W.H. was responsible for study design, data acquisition and data
analysis and wrote the manuscript. J.H. contributed to the study design,
data acquisition and data analysis. B.B. contributed to data analysis.
A.A.J.J.v.d.L. contributed to the study design and data acquisition.
J.J.v.d.D., M.V.B., M.K.C.H. and S.K. contributed to data acquisition
and data analysis. G.S., J.A.J. and B.H. contributed to data acquisition.
F.M.M., W.D.v.M.L. and P.S. contributed to the study design and
interpretation of data. All authors contributed to the critical revision
of the manuscript and approved the final version.
COMPETING FINANCIAL INTERESTS
The authors declare no competing financial interests.
Reprints and permissions information is available online at http://www.nature.com/
reprints/index.html.
1. van Marken Lichtenbelt, W.D. et al. N. Engl. J. Med. 360, 1500–1508
(2009).
2. Virtanen, K.A. et al. N. Engl. J. Med. 360, 1518–1525 (2009).
3. Cypess, A.M. et al. N. Engl. J. Med. 360, 1509–1517 (2009).
4. Bartelt, A. et al. Nat. Med. 17, 200–205 (2011).
5. Stanford, K.I. et al. J. Clin. Invest. 123, 215–223 (2013).
6. Townsend, K.L. & Tseng, Y.H. Trends Endocrinol. Metab. 25, 168–177
(2014).
7. van der Lans, A.A. et al. J. Clin. Invest. 123, 3395–3403 (2013).
8. Yoneshiro, T. et al. J. Clin. Invest. 123, 3404–3408 (2013).
9. Blondin, D.P. et al. J. Clin. Endocrinol. Metab. 99, E438–E446 (2014).
10. Ouellet, V. et al. J. Clin. Invest. 122, 545–552 (2012).
11. Yoneshiro, T. et al. Obesity (Silver Spring) 19, 13–16 (2011).
12. Hanssen, M.J. et al. Diabetologia 58, 586–595 (2015).
13. Meex, R.C. et al. Diabetes 59, 572–579 (2010).
14. Storlien, L., Oakes, N.D. & Kelley, D.E. Proc. Nutr. Soc. 63, 363–368
(2004).
15. Galgani, J.E. et al. Diabetes 57, 841–845 (2008).
16. Schrauwen-Hinderling, V.B. et al. Diabetologia 50, 113–120 (2007).
17. Bal, N.C. et al. Nat. Med. 18, 1575–1579 (2012).
18. Ruderman, N.B., Carling, D., Prentki, M. & Cacicedo, J.M. J. Clin. Invest. 123,
2764–2772 (2013).
19. Sato, M. et al. Diabetes 63, 4115–4129 (2014).
20. Wang, G.X. et al. Nat. Med. 20, 1436–1443 (2014).
Figure 2 Insulin sensitivity and skeletal muscle GLUT4 localization
before and after cold acclimation. (a) Individual data (left) and group
mean ± s.e.m. (right) for glucose infusion rate (GIR), corrected for body
weight, during hyperinsulinemic-euglycemic clamp; n = 8. (b) Insulin-
stimulated glucose disposal (R
d
), as a combination of glucose oxidation
(CHO
ox
) and nonoxidative glucose disposal (NOGD); n = 8. (c) Basal
(non–insulin-stimulated) endogenous glucose production (EGP; left)
and insulin-induced suppression of EGP (% ∆EGP) (right); n = 8.
(d) Plasma nonesterified fatty acid (NEFA) concentrations during the
hyperinsulinemic-euglycemic clamp; n = 8. (e–h) Western blot data for
phosphorylation of AKT at Ser473 (e), phosphorylation of AS160 (f),
ratio of phosphorylated AMPK (p-AMPK) and AMPK (g) and GLUT4
(P = 0.063) (h), including representative blots of two individuals
with T2D, in non–insulin-stimulated skeletal muscle biopsies; n = 7.
(i) Representative images (of 3–5 images per individual) of GLUT4
immunostaining on 5-µm-thick fresh-frozen non–insulin-stimulated skeletal
muscle tissue sections from three individuals in the study. Scale bars,
60 µm. (j) Quantification of average GLUT4 staining intensities at the
sarcolemma and the cytosol (n = 7). Throughout, comparisons between
data before and after cold acclimation were analyzed using Wilcoxon
signed-rank tests. *P < 0.05; error bars, mean ± s.e.m.

© 2015 Nature America, Inc. All rights reserved.
nature medicine
doi:10.1038/nm.3891
ONLINE METHODS
Participants. Eight overweight male individuals with T2D (age 59.3 ± 5.8 years,
body weight 92.0 ± 11.7 kg, BMI 29.8 ± 3.2 kg/m
2
, body fat 26.2 ± 4.0%)
were included in this study. We performed a power calculation with insulin
sensitivity as primary outcome. Based on an expected difference in insulin
sensitivity of 20%, with a power of 0.80 and an α of 0.05 (2-sided), we
calculated that 11 subjects would be needed (G*Power 3.1 software, F. Faul,
E. Erdfelder, A.G. Lang and A. Buchner, University of Trier, Trier, Germany;
http://www.gpower.hhu.de). Since the effects on insulin sensitivity were much
more pronounced than anticipated in the power calculation, we terminated
the study upon completion of all tests in eight subjects, in agreement with our
external monitoring board. All subjects were screened for medical history, and
exclusion criteria included uncontrolled hypertension, active cardiovascular
disease, liver or kidney dysfunction, smoking and use of insulin, beta-blockers
or other medication known to interfere with BAT function. Individuals
included in the study were diagnosed with T2D, which was well controlled
(HbA1c 50.1 ± 6.6 mmol/mol), at least 1.5 years before the start of the study.
All subjects used metformin, and four subjects also used sulfonylurea agents
(see Supplementary Table 1 for detailed subject characteristics). Studies were
performed between January and September 2014.
Study approval. The study was approved by the Ethics Committee of Maastricht
University Medical Center and all participants provided written informed con-
sent. Procedures were conducted according to the principles of the Declaration
of Helsinki. Trial registration number: NTR4319.
Study design. Body composition was determined by dual X-ray absorptiometry
(Discovery A, Hologic, Bedford, MA, USA). Subjects then started the 10-d cold
acclimation intervention. Before and after 10 d of cold acclimation, an individu-
alized cooling protocol was performed, immediately followed by [
18
F]FDG-
PET–CT scanning (Gemini TF PET-CT, Philips, Eindhoven, the Netherlands)
for quantification of BAT activity, as described previously
7
. On a separate day, an
abdominal subcutaneous fat biopsy and a muscle biopsy from the vastus lateralis
muscle
21
was taken after an overnight fast. Subsequently, peripheral insulin
sensitivity was assessed by a 3-h hyperinsulinemic-euglycemic clamp, performed
at thermoneutrality. During the 10-d cold acclimation protocol, subjects were
exposed to an environmental temperature of 14–15 °C for 10 consecutive days:
2 h on day 1, 4 h on day 2, and 6 h on days 3 through 10. On day 11 after the
start, cold-induced BAT activity measurements were repeated (~2.5 h in mild
cold). On the following day (day 12), subjects were again exposed to an envi-
ronmental temperature of 14–15 °C for 6 h, and this was followed by the second
hyperinsulinemic-euglycemic clamp on day 13. Subjects were instructed to
maintain their normal antidiabetic medication use throughout the whole accli-
mation period, except for the days on which the clamps were performed.
Subjects consumed standardized meals on the evenings before each
experimental day and were asked to refrain from heavy exercise at least 48 h
before each of these measurements.
Cold acclimation. During cold acclimation, subjects were dressed in shorts and
T-shirts and remained sedentary while staying in the cold room. Food intake
in the cold room was kept constant and subjects were instructed not to change
their normal dietary regime outside the cold room. Hunger and satiety ques-
tionnaires were completed before and after lunch. At selected time points, blood
pressure was monitored, VAS scales on sensation, thermal comfort and shivering
were completed and incremental AUCs (iAUC) were calculated to determine
subjective responses during cold acclimation
7
.
Individualized cooling and PET-CT imaging. Individualized cooling protocols
were performed after a 4 h fasting period. Subjects were wrapped in a water-
perfused suit (ThermaWrap Universal 3166, MTRE Advanced Technologies Ltd,
Yavne, Israel) and were measured at thermoneutrality for 45 min, after which
they were cooled until a temperature just above their shivering point and mea-
sured for an additional 30 min at this temperature. Core and skin temperatures,
heart rate, skin perfusion and energy expenditure were measured continuously
and blood pressure was measured every 15 min during the cooling protocol, as
described previously
7
. Nonshivering thermogenesis (NST) was calculated as the
absolute increase (corrected for body weight) in energy expenditure upon acute
mild cold exposure above basal metabolic rate (measured at thermoneutral-
ity). Subsequently, 74 MBq of [
18
F]FDG was injected intravenously. One hour
after injection the PET-CT scanning protocol started with a low-dose CT scan
(120 kV, 30 mAs). Directly hereafter, a static PET scan (6 to 7 bed positions,
6 min per bed position) was performed
22
. All PET-CT scans were analyzed using
PMOD software (version 3.0, PMOD Technologies, Zurich, Switzerland) by both
the researcher (M.J.W.H.) and an experienced nuclear medicine physician (B.B.).
Fixed volumes of interest (VOIs) were selected in the supraclavicular adipose
tissue region (between −10 and −180 Hounsfield units [HU]), subcutaneous and
visceral WAT, liver, brain and in the deltoid, biceps, triceps and erector spinae
muscles, as described previously
23
. Additionally, 2.67cm
3
VOIs were placed
in the scalene, levator scapulae and psoas major muscles, and average uptake
in these 7 muscle groups is presented as average skeletal muscle activity. The
VOIs were used to compare [
18
F]FDG uptake (calculated as SUV
mean
) and HU
between these tissues and between scans before and after cold acclimation.
Hyperinsulinemic-euglycemic clamp. Hyperinsulinemic-euglycemic
(40 mU m
−2
min
−1
) clamps were performed according to DeFronzo
24
, with
primed infusion of [6,6-
2
H
2
]glucose (0.04 mg kg
−1
min
−1
). Steele’s single-pool
non–steady state equations
25
were used to calculate rates of glucose appear-
ance (R
a
), glucose disposal (R
d
), nonoxidative glucose disposal (NOGD; mainly
reflecting glycogen synthesis) and endogenous glucose production (EGP), as
previously described
26
. Basal and insulin-stimulated substrate oxidation was
measured by indirect calorimetry (Omnical, IDEE, Maastricht, the Netherlands)
and calculated according to Frayn
27
.
Blood analysis. Blood samples were collected before and during the clamp
(in the basal and insulin-stimulated states). Plasma metabolites were deter-
mined according to standard procedures. Plasma concentrations of glucose
(ABX Glucose HK CP Radiometer, Horiba ABX) and nonesterified fatty acids
(NEFA-HR set, Wako Chemicals) were determined on a Cobas FARA centrifugal
spectrophotometer (Roche Diagnostica). Plasma insulin concentrations were
quantified using an immunometric assay (Advia Centaur, Siemens Diagnostics).
Serum TSH and T4, and plasma catecholamines, were analyzed as described
previously
7
.
WAT microarray analysis. Abdominal subcutaneous WAT was rinsed from
blood, snap frozen in melting isopentane and stored at −80 °C until analyzed.
Gene expression analysis was performed by microarray and compared to
publically available WAT microarray data sets from cold-acclimated mice.
RNA was purified from human fat biopsies using TRIzol (Life Technologies,
Calsbad, CA, USA) followed by an additional round of purification with
RNeasy Minikit columns (Qiagen, Venlo, the Netherlands). RNA quality was
assessed using RNA 6000 nanochips on the Agilent 2100 bioanalyzer (Agilent
Technologies, Amsterdam, the Netherlands). Purified RNA (100 ng) was labeled
with the Affymetrix WT PLUS reagent kit (Affymetrix, Santa Clara, CA, USA)
and hybridized to an Affymetrix Human Gene 1.1 ST array plate (Affymetrix,
Santa Clara, CA, USA). Hybridization, washing and scanning were carried out
on an Affymetrix GeneTitan platform according to the manufacturer’s instruc-
tions. Arrays were normalized using the robust multiarray average method
28,29
.
Probe sets were defined according to Dai et al.
30
. In this method probes are
assigned to Entrez IDs as a unique gene identifier. The P values were calcu-
lated using an intensity-based moderated t-statistic (IBMT)
31
. The microarray
data have been submitted to the Gene Expression Omnibus (accession number
GSE67297). Expression changes in the subcutaneous white fat depot of mice
subjected to either 10 d or 7 d of cold exposure (~5 degrees) were extracted
from publicly available microarray data sets (GSE51080 and GSE13432) using
the analysis pipeline described above.
Ex vivo skeletal muscle mitochondrial respiration. A portion of the muscle
biopsy was directly frozen in melting isopentane and stored at −80 °C until
assayed. Another portion (~30 mg) was immediately placed in ice-cold pres-
ervation medium (BIOPS, OROBOROS Instruments, Innsbruck, Austria) and
used for the preparation of permeabilized skeletal muscle fibers (~2.5 mg wet
weight)
32
, which were analyzed for mitochondrial oxidative capacity using an

© 2015 Nature America, Inc. All rights reserved.
nature medicine
doi:10.1038/nm.3891
oxygraph (OROBOROS Instruments) according to Hoeks et al.
33
. In separate
experiments, mitochondrial leak respiration was measured as the residual
respiration following addition of 1 µg/ml of the ATP synthase–inhibitor
oligomycin, using pyruvate (5 mM) as a substrate (in the presence of 4 mM
malate). All oxygen consumption measurements were performed in quadru-
plicate. One muscle biopsy failed after cold acclimation; therefore, values for
7 subjects are presented.
Muscle biopsy analyses. Protein expression was determined by western
blotting according to standard procedures. Primary antibodies (AKT,
phospho-AKT(Ser473), phospho-AKT (Thr308), phospho-AS160, phospho-
AMPK, AMPK(Thr172) and GLUT1; diluted 1:1,000) were all from Cell
Signaling Technology, Bioké, Leiden, the Netherlands and were all detected
using a horseradish peroxidase–conjugated secondary swine anti-rabbit IgG
antibody (DAKO, Glostrup, Denmark; diluted 1:2,000) and measured using
enhanced chemiluminescence (Pierce, Thermo Scientific, Rockford, IL, USA).
Primary antibodies directed against GLUT4 (Santa Cruz, BioConnect,
Huissen, the Netherlands) UCP3, OXPHOS-cocktail, Serca2 ATPase (all from
Abcam, BioConnect); PGC1α (Calbiochem, Darmstadt, Germany), sarcolipin
(Millipore, Schiphol-Rijk, the Netherlands) and α-sarcomeric actin (loading
control; Sigma, Zwijndrecht, the Netherlands) were detected using appropriate
secondary antibodies conjugated with IRDye680 or IRDye800 and detected with
the Odyssey Near Infrared System (Licor, Westburg, Leusden, the Netherlands).
Intramyocellular lipid (IMCL) content was determined in fresh cryosections
(5 µm) by Oil red O staining combined with fibertyping and immunolabeling
of the basal membrane marker laminin, as described previously
34
.
For GLUT4 imaging, double immunofluorescence assays were performed
on 5-µm-thick fresh-frozen tissue sections, which were fixated for 15 min with
3.7% formaldehyde in PBS and then treated for 5 min with 0.5% Triton X-100 in
PBS. Sections were incubated overnight at 4 °C with a mix of primary antibodies
directed to GLUT4 (Santa Cruz, BioConnect, Huissen, the Netherlands) and
laminin (Sigma, Zwijndrecht, the Netherlands). After three washing steps with
PBS, Alexa Fluor555– and Alexa Fluor488–conjugated secondary antibodies
were incubated for 45 min at room temperature. After a final washing step with
PBS, sections were mounted in Mowiol. Images were observed using a Nikon
E800 fluorescence microscope with NIS-elements Imaging Software (Nikon
Europe BV, Amsterdam, the Netherlands) and were captured with identical
exposure time and gain settings in ‘before’ and ‘after’ conditions. Without any
adjustments with respect to color intensity, brightness or contrast, RGB-stacked
images were quantified using the Plot Profile tool in ImageJ. Thus, we measured
the intensity of GLUT4-dependent signals (16 bits) throughout the sections.
Measured data on intensity were used to generate overlying plots of GLUT4 and
laminin. The GLUT4-derived staining intensity at the membrane was quantified
at multiple locations per individual (13.7 ± 2.9 locations before and 15.4 ± 3.2
locations after cold acclimation, respectively) in randomly chosen cross-
sections of muscle biopsies, and was divided over the mean GLUT4 intensity in
10 pixels located in the cytosol of the very same cell. Thus, a score >1.0 implied
that relatively more GLUT4 was detected in the membrane than in cytosolic
regions and hence reflected GLUT4 translocation.
Statistical analysis. Statistical analyses were performed with PSAW Statistics
20.0 for MAC (SPSS). Nonparametric paired-sample Wilcoxon signed-rank tests
were used to compare findings before and after cold acclimation and between
thermoneutral and mild cold conditions. Spearman rank correlations were
used to identify correlations between variables. P values <0.05 were considered
statistically significant.
21. Bergström, J., Hultman, E. & Roch-Norlund, A.E. Scand. J. Clin. Lab. Invest. 29,
231–236 (1972).
22. Vosselman, M.J. et al. Diabetes 61, 3106–3113 (2012).
23. Vosselman, M.J. et al. Am. J. Clin. Nutr. 98, 57–64 (2013).
24. DeFronzo, R.A., Tobin, J.D. & Andres, R. Am. J. Physiol. 237, E214–E223
(1979).
25. Steele, R. Ann. NY Acad. Sci. 82, 420–430 (1959).
26. Phielix, E. & Mensink, M. Physiol. Behav. 94, 252–258 (2008).
27. Frayn, K.N. J. Appl. Physiol. 55, 628–634 (1983).
28. Bolstad, B.M., Irizarry, R.A., Astrand, M. & Speed, T.P. Bioinformatics 19, 185–193
(2003).
29. Irizarry, R.A. et al. Nucleic Acids Res. 31, e15 (2003).
30. Dai, M. et al. Nucleic Acids Res. 33, e175 (2005).
31. Sartor, M.A. et al. BMC Bioinformatics 7, 538 (2006).
32. Phielix, E. et al. Diabetes 57, 2943–2949 (2008).
33. Hoeks, J. et al. Diabetes 59, 2117–2125 (2010).
34. Koopman, R., Schaart, G. & Hesselink, M.K. Histochem. Cell Biol. 116, 63–68
(2001).