ReportPeroxisome Proliferator-Activated Receptor g2
Controls the Rate of Adipose Tissue Lipid Storage
and Determines Metabolic FlexibilityGraphical AbstractHighlightsd Mice lacking PPARg2 have impaired adipose tissue lipid
storage rate
d Low adipose tissue storage rate leads to metabolic
inflexibility
d Acute hypercaloric challenges can detect impaired adipose
tissue lipid storage rate
d Chronic adipose tissue metabolic inflexibility leads to insulin
resistance with ageVirtue et al., 2018, Cell Reports 24, 2005–2012
August 21, 2018 ª 2018 The Authors.
https://doi.org/10.1016/j.celrep.2018.07.063Authors
Sam Virtue, Kasparas Petkevicius,
Jose´ Maria Moreno-Navarrete, ...,
Albert Koulman,
Jose´ Manuel Ferna´ndez-Real,
Antonio Vidal-Puig
Correspondence
sv234@medschl.cam.ac.uk (S.V.),
ajv22@medschl.cam.ac.uk (A.V.-P.)
In Brief
Virtue et al. describe how the
transcription factor peroxisome
proliferator-activated receptor gamma 2
(PPARg2) regulates lipid storage rate in
adipose tissue. Mice lacking PPARg2
cope when fat storage demands are low,
but acute overfeeding overwhelms
adipose tissue, redirecting lipid to muscle
and causing insulin resistance.
Cell Reports
ReportPeroxisome Proliferator-Activated Receptor g2
Controls the Rate of Adipose Tissue Lipid Storage
and Determines Metabolic Flexibility
Sam Virtue,1,4,* Kasparas Petkevicius,1 Jose´ Maria Moreno-Navarrete,2 Benjamin Jenkins,1 Daniel Hart,1 Martin Dale,1
Albert Koulman,1 Jose´ Manuel Ferna´ndez-Real,2 and Antonio Vidal-Puig1,3,*
1The University of Cambridge Metabolic Research Laboratories, Wellcome Trust–MRC Institute of Metabolic Science,
Cambridge CB2 0QQ, UK
2Biomedical Research Institute of Girona (IDIBGI), CIBERobn Pathophysiology of Obesity and Nutrition, Hospital of Girona
‘‘Dr. Josep Trueta,’’ Avinguda de Franc¸a s/n, and Department of Medical Sciences, Faculty of Medicine, University of Girona, Girona, Spain
3Wellcome Trust Sanger Institute, Hinxton CB10 1SA, UK
4Lead Contact
*Correspondence: sv234@medschl.cam.ac.uk (S.V.), ajv22@medschl.cam.ac.uk (A.V.-P.)
https://doi.org/10.1016/j.celrep.2018.07.063SUMMARY
One understudied function of white adipose tissue
(AT) is its role in postprandial lipid buffering. In this
study, we demonstrate that mice lacking the adipose
tissue-specific transcription factor peroxisome pro-
liferator-activated receptor g2 (PPARg2) exhibit a
defect in their rate of adipose tissue lipid storage.
Impaired adipose tissue storage rate reduces
metabolic flexibility, without compromising fasted
glucose tolerance or insulin sensitivity, even following
prolonged high-fat feeding. However, acutely over-
feeding PPARg2-KO mice caused a 10-fold increase
in insulin levels compared with controls. Although
impaired adipose tissue storage rate did not result
in insulin resistance in young mice, 1-year-old
PPARg2-KO mice developed skeletal muscle insulin
resistance.Ourdata indicate that failedadipose tissue
storagemayoccurprior todefects inglucosehandling
and that overfeeding protocols may uncover genes
controlling adipose tissue storage rate, as opposed
to capacity, and act as a diagnostic test for early-
stage human metabolic disease.
INTRODUCTION
The epidemic of obesity has led to epidemics of diabetes and
cardiovascular disease. An important question is how obesity
leads to metabolic complications. A proposed mechanism link-
ing obesity to metabolic complications is the concept of adipose
tissue (AT) expansion and lipotoxicity. The adipose tissue expan-
sion hypothesis suggests that metabolic complications occur
when an individual’s adipose tissue storage capacity is ex-
ceeded (Virtue and Vidal-Puig, 2010), that preventing adipose
tissue expansion will worsen metabolic complications (Danforth,
2000), whereas unlimited adipose tissue expansion would be
metabolically protective, a concept demonstrated in mice (Kim
et al., 2007).Cell R
This is an open access article undWe previously reported (Medina-Gomez et al., 2005) the
phenotype of mice lacking peroxisome proliferator-activated re-
ceptor g2 (PPARg2). PPARg2 is the adipose tissue-specific iso-
form of PPARg. Surprisingly, despite defective adipogenesis
in vitro, mice lacking PPARg2 had normal adipocyte size on a
chow diet (Medina-Gomez et al., 2005), as well as normal
glucose and insulin tolerance on chow or chronic high-fat
(3 months of feeding) diets. However, crossing the PPARg2-
knockout (KO) mouse onto an ob/ob background (POKOmouse)
led to severe metabolic syndrome. Importantly, defects in insulin
sensitivity in POKO mice occurred before any divergence in
adiposity between genotypes (Medina-Gomez et al., 2007).
The early onset of metabolic impairments in POKO mice raised
the question as to which aspect of leptin deficiency was driving
the phenotypic divergence. Furthermore, why PPARg2-KOmice
on a non-obese background were metabolically healthy was
unresolved.
Although PPARy2 KO mice had normal-sized adipocytes on a
chow diet, they showed molecular markers of impaired adipo-
cyte function (Medina-Gomez et al., 2005). Dysfunctional adi-
pose tissue function has been suggested to lead to impaired
metabolic flexibility in both humans (Frayn, 2002) and rodents
(Asterholm et al., 2012; Asterholm and Scherer, 2010). Metabolic
inflexibility is the inability to switch efficiently between predomi-
nantly using lipid in the fasted state to carbohydrate in the fed
state (Goodpaster and Sparks, 2017; Storlien et al., 2004). In hu-
mans, obesity is characterized by low metabolic flexibility (Stor-
lien et al., 2004). Although direct defects in muscle lipid and
glucose use have been suggested to causemetabolic inflexibility
(Ukropcova et al., 2007), defects in adipose tissue function can
also cause metabolic inflexibility by altering the fluxes of lipids
to muscle (Frayn, 2002). Dysfunctional adipose tissue fails to
store postprandial lipid and redirects it to non-adipose organs.
Conversely, in the fasting state, insufficient release of free fatty
acids (FFAs) by adipose tissue occurs because of impaired lipol-
ysis (McQuaid et al., 2011).
Although multiple studies have investigated metabolic flexi-
bility in mice (Asterholm et al., 2012; Asterholm and Scherer,
2010; Gao et al., 2015; Isken et al., 2010; Jackowski and Leo-
nardi, 2014; Longo et al., 2008; Muoio et al., 2012; Riachieports 24, 2005–2012, August 21, 2018 ª 2018 The Authors. 2005
er the CC BY license (http://creativecommons.org/licenses/by/4.0/).
0 20 40 60 80 100 120D
0
0.5
1
1.5
2
2.5
3
3.5
RD GIR HGP
0
4
8
12
16
0 20 40 60 80 100 120
0
1
2
3
4
5
6
7
uM
ol
/m
ou
se
/m
in
0
2
4
6
8
10
0 20 40 60 80 100 120
A B C
0
20
70
50
40
30
10
S
up
re
ss
io
n 
of
 H
G
P 
(%
)
G
lu
co
se
 (m
M
ol
)
E
xo
 G
lu
co
se
 (m
M
ol
)
E
nd
o 
G
lu
co
se
 (m
M
ol
)
S
up
re
ss
io
n 
of
 F
FA
 (%
)
0
15
35
30
25
20
10
5
WT
PPARγ2 KO
0.7
0.8
0.9
1
1.1
R
E
R
0
0.1
0.2
0.3
0.4
0.7
0.8
0.9
1
1.1
08
:41
11
:41
14
:41
17
:41
20
:41
23
:41
02
:41
05
:41
08
:41
R
E
R
0
2
4
6
8
10
12
0 2 4 6 8
Time (Hours)
TG
 (m
M
ol
/l)
*
*
*
F
K
E
I JH
WT
PPARγ2 KO
Time of day
*AOC
0
2
4
6
8
0 20 40 60 80 100 120
0
4
8
12
16
20
0 20 40 60 80 100 120
0
3
6
9
12
15
0 20 40 60 80 100 120
G
lu
co
se
 (m
M
ol
)
Time (mins)
E
xo
 G
lu
co
se
 (m
M
ol
)
Time (mins) Time (mins)
E
nd
o 
G
lu
co
se
 (m
M
ol
)
Time (mins) Time (mins)
8
* *
*AOC
60
Low High
∆R
E
R
Time (mins)
G
Figure 1. PPARg2-KO Mice Exhibit Altered
Carbohydrate and Lipid Metabolism in the
Fed State
(A–C) Fasted glucose tolerance tests fromWT and
PPARg2-KOmice: (A) total, (B) exogenous, and (C)
endogenous glucose.
(D) Hyperinsulinemic-euglycemic clamps: rate of
disposal (RD), glucose infusion rate (GIR), and
hepatic glucose production (HGP) in the hyper-
insulinemic state (left), suppression of HGP
(middle), and suppression of NEFA from basal to
hyperinsulinemic state (right) (n = 7 WT and n = 9
KO for GTT, n = 8 WT and n = 5 KO for clamps).
(E–G) Metabolic flexibility shown by (E) represen-
tative plot of 24 hr respiratory exchange ratio
(RER) determined by indirect calorimetry, (F)
lowest and highest 10% of RER values for WT and
PPARg2-KO mice, and (G) dRER (n = 8 mice per
group).
(H–J) Fed glucose tolerance tests: (H) total, (I)
exogenous, and (J) endogenous glucose (n = 6WT
and n = 9 KO).
(K) Lipid clearance in PPARg2-KO and WT mice
(n = 9 WT, n = 8 KO).
*p < 0.05, two-tailed Student’s t test. All mice were
4–5 months of age. All data are represented as
mean ± SEM. See also Figure S1.et al., 2004; Vacanti et al., 2014; Vaitheesvaran et al., 2010; Vroe-
grijk et al., 2013), translating murine data to humans is compli-
cated because of their different feeding patterns. Mice eat
multiple small meals per day (Ellacott et al., 2010), and as a
result, the rate at which the adipose tissue of mice is required
to store and release lipid maybe relatively low. In contrast, hu-
mans tend to eat three large meals a day, causing fewer more
acute demands for adipose tissue lipid storage (McQuaid
et al., 2011). Overall, a loss of adipose tissue flexibility may man-
ifest differently in rodents and humans.
The most commonly used model for inducing obesity and
insulin resistance in mice is high-fat diet (HFD) feeding. HFD
causes many mouse strains to consume more calories and
gain weight, but after an initial period of rapid body weight2006 Cell Reports 24, 2005–2012, August 21, 2018gain, usually lasting less than 1 week,
body weight divergence between HFD
and chow-fed mice tends to be relatively
slow. The slow rate of weight gain caused
by established HFD suggests that
chronic HFD may not actually greatly in-
crease demands on adipose tissue stor-
age rate.
In this study, we reinvestigated
PPARg2-KO mice. Although PPARg is
traditionally viewed as a regulator of adi-
pogenesis, PPARg2 is regulated by fast-
ing and refeeding (Vidal-Puig et al.,
1997), suggesting an additional dynamic
role for PPARg2. Supportive of this
concept, PPARg2 is necessary for appro-
priate expression of lipolytic genes in ad-ipocytes (Rodriguez-Cuenca et al., 2012). In this study, we
demonstrate that PPARg2-KOmice have a reduced adipose tis-
sue lipid storage rate and exhibit metabolic inflexibility. In states
in which adipose tissue lipid storage rates are low, mice lacking
PPARg2 have normal insulin and glucose levels. To challenge
adipose tissue storage rate, we exploited the fact that mice
dramatically over-eat when initially switched to an HFD. In
response to 1 day of high-fat feeding (1dHFD), PPARg2-KO
mice failed to store lipid in adipose tissue and redirect it to mus-
cle, causing triglyceride (TG) accumulation and insulin resis-
tance, with PPARg2-KO mice exhibiting 10 times the insulin
levels of controls. Importantly, mice lacking PPARg2 developed
insulin resistance with aging, suggesting that a primary loss of
adipose tissue lipid buffering may be initially tolerated but
020
60
100
Chow  1dHFD
0
0.4
0.8
1.2
1.6
2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
0
10
20
30
40
0
0.4
0.8
1.2
1.6
2
BAT scWAT eWAT Liver 0
2
4
6
8
10
12
FI
 (k
J/
da
y)
dE
E
 (J
/m
in
/m
ou
se
)
dB
W
 (g
)
B
W
 (g
)
Ti
ss
ue
 w
ei
gh
t (
g)
%
 L
iv
er
 F
at
A B C
D E F
**
*
*
WT Chow
KO Chow
WT 1d HFD
KO 1d HFD
0
2
4
6
8
10
12
Chow 1day 1month 0
0.5
1
1.5
2
2.5
3
Chow 1day 1month
0
0.2
0.4
0.6
0.8
1
1.2
1.4
Chow 1day 1month
0
2
4
6
8
10
12
Chow 1day 1month
FF
A 
(m
M
ol
)
TG
 (m
M
ol
)
G
lu
co
se
 (m
M
ol
)
*
* *
G H
I J
WT 1m HFD
KO 1m HFD
In
su
lin
 (u
g/
l)
40
80
Figure 2. PPARg2-KO Mice Exhibited
Impaired Metabolic Responses to Acute
Overfeeding
(A–F) Responses to 1 day high-fat diet (1dHFD): (A)
food intake, (B) change in energy expenditure
(dEE), and (C) change in body weight (dBW) be-
tween chow and after 1dHFD; (D) body weight, (E)
tissue weights, and (F) liver fat percentage after
1dHFD period.
(G–J) Serum biochemistry for (G) insulin, (H) TGs,
(I) FFAs, and (J) glucose after 1dHFD.
N = 8 per group except 1 month HFD KO and food
intake for chow and 1dHFD WT (n = 6). *p < 0.05,
two-tailed Student’s t test. All mice were 4 months
of age. All data are represented as mean ± SEM.
See also Figures S1 and S2.ultimately leads to metabolic disease. Finally, we demonstrate
that, in line with a potential acute metabolic role for PPARg2 in
adipocytes, PPARg2 but not PPARg1 correlates with whole-
body and adipose tissue insulin sensitivity in humans.
RESULTS AND DISCUSSION
PPARg2-KO mice had normal fasting glucose tolerance (Figures
1A–1C) (Medina-Gomez et al., 2005), insulin levels (Figure S1A),
glucose infusion rate (GIR), rate of whole-body glucose disap-
pearance (RD), and hepatic glucose production (HGP) during
a euglycemic-hyperinsulinemic clamp (Figure 1D). However,
PPARg2-KO mice had reduced metabolic flexibility in the fed
state (Figures 1E–1G). On the basis of their amplitude of respira-
tory exchange ratio (RER), PPARg2-KO mice had increased
glucose oxidation in the light phase and increased lipid oxidation
in thedarkphase relative to controls (Figure1F). To investigate the
increased glucose oxidation in the light phase, we performed a
glucose tolerance test (GTT) in the fed state at 10 a.m., incorpo-
rating stable isotopes. Consistent with their RER profile, mice
lacking PPARg2 exhibited increased clearance of exogenous
glucose (Figures 1H–1J). The PPARg2-KO mice also exhibited
reduced RER during the night phase, suggesting that PPARg2
mice may have impaired adipose tissue lipid storage leading toCell Repexcess delivery of lipid to muscle (Randle
et al., 1963), and in line with this concept,
mice lackingPPARg2wereunable toclear
an oral lipid load as effectively as controls,
exhibiting higher TG (Figure 1K) and FFAs
(Figure S1B) during a lipid tolerance test.
In response to chronic HFD, mice lack-
ing PPARg2 have normal carbohydrate
metabolism and adipose tissue mass
(Medina-Gomez et al., 2005). However,
given their compressed RER profiles
and inability to clear lipids, we hypothe-
sized that PPARg2-KO mice may exhibit
a defect in the rate of adipose tissue lipid
storage, rather than total storage capac-
ity. This hypothesis was also supported
by the early onset (4 weeks of age) of hy-perinsulinemia in POKO mice, an age at which ob/ob mice are
already hyperphagic but prior to a divergence in adiposity rela-
tive to controls (Medina-Gomez et al., 2007). To stress adipose
tissue storage rate, we developed an overfeeding challenge by
giving mice a high-fat diet for 1 day (1dHFD). We first character-
ized 1dHFD in wild-type mice. Mice on 1dHFD doubled food
intake (Figure S1C) and increased insulin levels (Figure S1D) to
maintain normoglycemia (Figure S1E). Body weight and subcu-
taneous white adipose tissue (scWAT) mass were increased
(Figures S1F–S1H). Finally, Pparg2 expression in scWAT was
physiologically increased in response to 1dHFD, suggestive of
a role for PPARg2 in responding to overfeeding (Figure S1I).
PPARg2-KO mice placed on HFD for 1 day exhibited similar
food intake (Figure 2A), diet-induced thermogenesis (Figure 2B),
and body weight gain (Figure 2C) to controls, whereas absolute
body weight, BAT weight, liver weight, and liver fat percentage
were increased (Figures 2D–2F).
When subjected to 1dHFD, PPARg2-KO mice exhibited
dramatically increased insulin levels and increased glucose
and FFA levels compared with 1dHFDwild-type controls. Impor-
tantly, prolonged HFD feeding (for either 1 week or 1 month)
normalized the serum biochemistry parameters of PPARg2-KO
mice (Figures 2G–2J, S1L, and S1M), even though PPARg2-KO
mice after 1 month of HFD had greater adiposity (Figures S1Jorts 24, 2005–2012, August 21, 2018 2007
00.5
1
1.5
Angptl4 Lpl Gpihbp1
0
0.5
1
1.5
Fatp1 Cd36 Got2
0
0.5
1
1.5
Atgl Hsl Perilipin
WT Chow
KO Chow
WT 1d HFD
KO 1d HFD
WT 1m HFD
KO 1m HFD
pHSL S660
tHSL  
pP44 MAPK
pP42 MAPK
tP44 MAPK
tP42 MAPK
pAKT s473
tAKT
B-actin
pA
KT
/tA
KT
pP
44
/tP
44
pP
42
/tP
42
pH
SL
/tH
SL
tHS
L/b
act
in
tAK
T/B
act
in
W
T 1
dH
FD
KO
 1d
HF
D
W
T C
ho
w
KO
 C
ho
w
A B C
D E
R
el
at
iv
e 
ex
pr
es
si
on
R
el
at
iv
e 
ex
pr
es
si
on
R
el
at
iv
e 
ex
pr
es
si
on
A
rb
itr
ar
y 
U
ni
ts
*
* *
*
*
*
*
* * * * * * * *
*
0
40
80
120
160
200
240
280
D
pm
/u
l o
f s
er
um
F
I
0
2
4
6
8
10
12
sc
W
AT
 N
E
/E
 F
A
20
40
60
80
100
120
140
D
pm
/m
g 
of
 s
cW
AT
*
* *
J
0
Fas Elovl6 Scd1 Srebp1c
M
P
0.5
1
1.5
0
R
el
at
iv
e 
E
xp
re
ss
io
n
R
el
at
iv
e 
E
xp
re
ss
io
n
*
*
0
0.5
1
1.5
N
R
el
at
iv
e 
E
xp
re
ss
io
n
Irs1 Irs2 F4/80
0
0.2
0.4
0.6
0.8
1
1.2
1.4
pAKT s473
tAKT
pP42 MAPK
tP42 MAPK
B-actin A
rb
itr
ar
y 
U
ni
ts
pA
KT
/tA
KT
pP
42
/tP
42
W
T 1
dH
FD
KO
 1d
HF
D
W
T C
ho
w
KO
 C
ho
w
WT
KO
Chow 1d HFD 1m HFD
*
G H
K L
0
0.5
1
1.5
2
2.5
3
3.5
0
500
1000
1500
2000
2500
3000
 A
di
po
cy
te
 A
re
a 
(μ
m
²)
3500
*
*
*
*
*
*
0
0.5
1
1.5
2
2.5
* *
* *
3
Angptl4 Lpl Fatp1 Dgat1 Dgat2
0
0.5
1
1.5
2
O
R
el
at
iv
e 
E
xp
re
ss
io
n 2.5
Aox Cpt1a Cpt1b Lcad Pgc1a Pgc1b
*
*
Figure 3. Mice Lacking PPARg2 Cannot Appropriately Store Lipid in Adipose Tissue
(A–C) Gene expression in WAT for (A) lipoprotein lipase and regulatory molecules and (B) FA uptake genes. (C) Markers of lipolysis (n = 8 mice per group except
1 month HFD KO [n = 7]).
(D and E) Western blots of WAT: (D) representative western blots and (E) quantification (n = 8 mice per group).
(legend continued on next page)
2008 Cell Reports 24, 2005–2012, August 21, 2018
and S1K) than PPARg2-KOmice after 1dHFD (Figure 2E). Finally,
we conducted a fasting-refeeding experiment and demon-
strated that mice lacking PPARg2 had greatly increased insulin
in response to refeeding after an overnight fast (Figures S2A
and S2B). Overall, our data suggested a lack of PPARg2
impaired adipose tissue lipid storage rate and a failure in adipose
tissue storage rate caused insulin resistance in response to
overnutrition.
We next sought to confirm that PPARg2-KO adipose tissue
was unable to buffer lipid appropriately. We first analyzed gene
expression in scWAT. In terms of genes controlling hydrolysis
of TG to FFAs, lipoprotein lipase (Lpl) tended to be upregulated,
whereas the LPL inhibitory protein angiopoietin-like 4 (Angptl4)
was downregulated in PPARg2-KO scWAT after 1dHFD (Fig-
ure 3A). The FFA transporter-encoding genes FA transport pro-
tein (Fatp1) and Cd36 were downregulated in PPARg2-KO
mice by 1dHFD (Figure 3B). The trend to upregulation of Lpl
and the downregulation of Angptl4, Cd36, and Fatp1 were
consistent with the normal TG levels and increased FFAs in
PPARg2-KO mice following 1dHFD. To exclude a role for lipol-
ysis in regulating circulating FFA levels, we analyzed expression
of adipose TG lipase (Atgl), which was downregulated under
high-fat feeding conditions and hormone sensitive lipase (Hsl),
which was downregulated under all nutritional conditions (Fig-
ure 3C) in PPARg2-KO mice. HSL phosphorylation at the activa-
tory S660 site was decreased in the scWAT of PPARg2-KOmice
following 1dHFD (Figures 3D and 3E), while insulin signaling, a
known suppressor of lipolysis, was either similar (ratio of phos-
pho-AKT to total AKT) or upregulated (ratio of phospho-P44/42
ERK to total P44/42 ERK), in line with the increased circulating
insulin levels. Overall, and in line with our previous work (Rodri-
guez-Cuenca et al., 2012), the elevated FFAs in PPARg2-KO
mice were unlikely to be a result of FFA release by adipose tissue
but a result of a defect in FFA uptake.
To confirm a defect in FFA transport into adipose tissue, we
used the fact our HFDhas an increased nonessential (NE)/essen-
tial (E) FA ratio compared with chow. Feeding mice an HFD leads
to a progressive increase in NE/E FA ratio in white adipose tissue
(WAT) (Yew Tan et al., 2015). Wild-type (WT) and PPARg2-KO
mice had similar NE/E FA ratios under chow conditions, but
1 day HFD increased NE/E FA ratio only in WT mice (Figures
3F and S2C), a difference that was normalized between geno-
types after 1month of HFD. Importantly, this analysis detects ex-
isting fat as well as newly stored fat. The 15% increase in NE/E
ratio in WT mice was in good agreement with the 20% increase
in fat mass observed in response to 1dHFD (Figure S1G), sug-
gesting that most of the fat accumulating in the scWAT of WT(F) The ratio of non-essential to essential FAs in adipose tissue fromWT and PPAR
WT chow, 1 month HFD WT and KO).
(G andH)Morphometric analysis of sections ofWAT fromWT and PPARg2-KOmi
adipocyte cross-sectional area (n = 8 per group KO chow, 1 day HFD KO and 1
(I and J) Radioactive palmitate levels in (I) blood and (J) scWAT from WT and PPA
group).
(K and L) Western blots from muscle: (K) representative western blots and (L) qu
(M–P) Gene expression in muscle for (M) FA oxidative markers, (N) insulin sensitiv
(n = 8 mice per group except 1 month HFD KO [n = 7]).
For multiple time points, two-way ANOVA was performed, and if significant, pairw
All mice were 4–5 months old. All data are represented as mean ± SEM. See alsmice was dietary derived. Further support for a lack of lipid up-
take to adipocytes was provided by histological analysis of WT
and PPARg2-KO mouse adipocyte sizes. Although WT mice
increased their adipocyte size following 1dHFD, PPARg2-KO
mice did not (Figures 3G and 3H). Finally, we traced palmitate
uptake to WAT under euglycemic-hyperinsulinemic clamp con-
ditions. PPARg2-KO mice exhibited increased 14C-palmitate in
blood (Figure 3I) and decreased incorporation into adipose tis-
sue (Figures 3J and S2D–S2F). Overall, these results suggested
that elevated circulating FFA levels in PPARg2-KO mice were
caused by a failure to uptake and esterify FFAs into WAT, a
concept further supported by the elevated FFAs during a lipid
tolerance test (LTT) (Figure S1B).
The failure of adipose tissue to clear FFAs (Figure 3J) suggested
apotential increasedfluxofFFAs tootherorgans (Virtue andVidal-
Puig, 2010), which we next sought to investigate. In liver, mice
lacking PPARg2 exhibited a more lipogenic response to acute
overfeeding than WTs (Figures S2G–S2K), potentially reflecting
the increased insulin levels (Figure 2G), but no differences in FA
oxidation markers or changes in the phosphorylation status of in-
sulin signalingmolecules (Figures S2L and S2M). Similarly to liver,
no defects in brown adipose tissue (BAT) insulin signaling (Figures
S3A and S3B), or glucose disposal into BAT or WAT (Figure S3C)
were detected. Conversely, AKT phosphorylationwas impaired in
skeletal muscle after 1dHFD in PPARg2-KOmice (Figures 3K and
3L). Muscle gene expression showed that several oxidativemeta-
bolism genes in PPARg2-KO mice (Figure 3M) reached levels of
expression after 1dHFD that were similar to or greater than those
observed for either WT or PPARg2-KOmice after 1 month of high
feeding. Additionally, lipid uptake andTGstoragegenes tended to
be increased in the muscle of PPARg2-KO mice compared with
controls (Figure 3O). HFD did not modifymRNAmarkers of insulin
sensitivity in PPARg2-KO muscles (Figure 3N) or markers of lipo-
genesis (Figure 3P) between genotypes.
To validate gene expression changes, we analyzed muscle
lipid composition. On a chow diet, WT and PPARg2-KO mice
had largely similar muscle lipidomes (Figure S3D) that could
not be separated using multivariate analyses (Figures 4A and
4B), but following 1dHFD, a significant separation in the lipid pro-
files was detectable (Figures 4A and 4B). PPARy2-KO mice
accumulatedmore longer chain FA-containing TGs than controls
(Figure 4C). Finally, we measured the acylcarnitine profile of the
muscles. Under chow conditions, and consistent with their
increased carbohydrate oxidation in the light phase, mice lack-
ing PPARg2 had less short-chain acyl carnitines, but this pheno-
type was reversed by 1dHFD (Figures 4D and S3E). In summary,
our data supported the idea that PPARg2-KO mice exhibit ag2-KOmice (n = 8 per group KO chow, 1 day HFDWT and KO, n = 7 per group,
ce: (G) representative images (scale bar shows 100 mM) and (H) quantification of
month HFD WT, n = 7 per group, KO chow, 1dHFD WT 1 month HFD KO).
Rg2-KO mice under hyperinsulinemic-euglycemic clamp conditions (n = 6 per
antification (n = 8 per group).
ity markers, (O) lipid uptake and storage genes, and (P) lipid biosynthetic genes
ise comparisons were performed using two-tailed Student’s t test (*p < 0.05).
o Figures S2 and S3.
Cell Reports 24, 2005–2012, August 21, 2018 2009
00.6
1.2
1.8
F480
0
0.2
0.4
0.6
0.8
1
1.2
Angptl4 Fatp1 Lpl
0
0.2
0.4
0.6
0.8
1
1.2
1.4
Aox Cpt1a Lcad Pgc1a Pgc1b
0
0.2
0.4
0.6
0.8
1
1.2
Dgat1 Dgat2 Elovl6 Fas Scd1 Srebp1c
0
0.2
0.4
0.6
0.8
1
1.2
Irs1 Irs2
0
1.5
3
4.5
RD GIR HGP
0
20
60
70
50
40
30
10uM
ol
/m
ou
se
/m
in
S
up
re
ss
io
n 
of
 H
G
P
%
0
20
50
40
30
10
*
B
od
y 
W
ei
gh
t (
g)
R
el
at
iv
e 
E
xp
re
ss
io
n
R
el
at
iv
e 
E
xp
re
ss
io
n
R
el
at
iv
e 
E
xp
re
ss
io
n
R
el
at
iv
e 
E
xp
re
ss
io
n
*
*
E F
H I
J K L
WT 1 year old
KO 1 year old
*
*
*
*** * *
G
R
el
at
iv
e 
E
xp
re
ss
io
n
R² = 0.217
P=0.001
R² = 0.130
P=0.016
R² = 0.094
P=0.04
0 2 4 6 8 10 12 14 16
0
2
4
6
8
10
12
M
-V
al
ue
0
0.5
1
1.5
2
2.5
3
IR
S
1
0
0.2
0.4
0.6
0.8
1
1.2
1.4
G
lu
t4
PPARγ2
0 2 4 6 8 10 12 14 16
PPARγ2
0 2 4 6 8 10 12 14 16
PPARγ2
M N O
-4
-2
0
2
4
-8 -6 -4 -2 0 2 4 6 8
t[2
]
t[1]
KO1
KO2
KO3
KO4
O 5
KO6
KO7 KO8
Chow WT1
WT2
WT3 WT4
WT5
WT6
WT7
Chow WT8
-5
0
5
-12-10-8-6 -4-2 0 2 4 6 8 1012
t[2
]
t[1]
KO2
KO3KO4
KO5KO6
KO7
KO8
WT1
WT2
WT3WT4T5WT6
WT7WT8
-4
-3
-2
-1
0
1
2
3
4
t[1] t[2]
0
50
100
150
200
250
0
5
10
15
20
25
30 WT Chow
KO Chow
WT 1d HFD
KO 1d HFD
Lo
ad
in
g 
C
oe
ffi
ci
en
ts
Chow 1d HFD
Tr
ig
ly
ce
rid
es
ng
/m
g 
pr
ot
ei
n
A
cy
l C
ar
na
tin
es
ng
/m
g 
pr
ot
ei
n
>50 Carbons <50 Carbons Short Chain Mid/long Chain
A B
C D
*
*
0.07
0.06*
Figure 4. Loss of Adipose Tissue Lipid
Buffering Leads to Aging-Related Insulin
Resistance
(A) Principal-component analysis of lipid species
frommuscleof chow (left) or 1dHFD (right) fedmice.
(B) Quantification of the first two principal com-
ponents from (A).
(C) Muscle TG species.
(D) Muscle acylcarnitines species (n = 6–8 per
group).
(E–I) Gene expression in muscle of 1-year-old
mice: (E) F4/80, (F) FA oxidative markers, (G) lipid
uptake and storage genes, (H) lipid biosynthetic
genes, and (I) Insulin sensitivity markers (n = 10
WT, n = 7 KO, male mice chow fed).
(J–L) Clamps from 1-year-old mice: (J) rate of
glucose disposal (RD), glucose infusion rate (GIR),
and hepatic glucose production (HGP) under
hyperinsulinemic condition; (K) suppression of
HGP between the basal and hyperinsulinemic
state; and (L) body weights of mice used for
clamps (n = 7 WT and 6 KO mice, chow fed).
(M–O) Correlations between PPARg2 expression
in human scWAT and (M) M-value during a clamp,
(N) IRS1 expression, and (O) Glut 4 expression
(n = 45 subjects).
For multiple time points, two-way ANOVA was
performed, and if significant, pairwise compari-
sons were performed using two-tailed Student’s
t test (*p < 0.05). Correlations are Pearson’s, and
exact p values are reported. All mice were
4–5 months old. All data are represented as
mean ± SEM. See also Figures S3 and S4.primary defect in adipose tissue storage rate, which in the
context of acute overfeeding drives FFA flux to muscle.
Increased FFA flux to muscle leads to lipid accumulation, lipo-
toxicity-induced insulin resistance, and compensatory upregula-
tion of the FA oxidation program (Randle et al., 1963).
Although 1dHFDwas able to unmask the underlyingmetabolic
defects of PPARg2 ablation, even under chow conditions, mice
lacking PPARg2 had impaired metabolic flexibility (Figure 1E)
and increased expression of the lipid oxidation genes Aox and2010 Cell Reports 24, 2005–2012, August 21, 2018Cpt1b (Figure 3M) in skeletal muscle. To
determine if a loss of metabolic flexibility
due to adipose tissue dysfunction led to
longer term metabolic complications, we
aged PPARg2-KO mice and controls on
a chow diet for 1 year. Aged mice lacking
PPARg2 exhibited decreased expression
ofPgc1a (Figure 4F) and insulin sensitivity
markers including the FA synthesis pro-
gram and Irs1 (Figures 4H and 4I) and
increased expression of the macrophage
marker F4/80 (Figure 4E) in muscle.
Notably, other than a tendency toward
reducedScd1 expression, these changes
were absent in themuscles of youngmice
on a chow diet (Figures 3M–3P). These
molecular changes manifested in skeletalmuscle insulin resistance, with aged PPARg2-KO mice exhibit-
ing lower RD, reduced GIR, and normal HGP (Figures 4J–4L)
as well as increased fed insulin levels (Figures S3F–S3J)
compared with WTs. Adipose tissue showed relatively little
further impairment with aging compared with young animals
(Figures S4A–S4E), and liver exhibited almost no changes be-
tween PPARg2-KO and WTs (Figures S4F–S4J).
The human relevance of our observations was confirmed in
adipose tissue from 45 obese subjects who had undergone a
euglycemic-hyperinsulinemic clamp. PPARg2, but not PPARg1,
expression in scWAT correlated with M-value as well as IRS1
and GLUT4 (Figures 4M–4O and S4K–S4M).
Our study identifies PPARg2 as a crucial regulator of adipose
tissue lipid storage rate, in addition to its role in adipose tissue
development and total storage capacity. A reduction in adipose
tissue storage rate limits the ability of adipose tissue to respond
acutely to large nutrient influxes and causes metabolic inflexi-
bility that chronically leads to insulin resistance. Our data there-
fore suggest a possible evolutionary rationale for the presence of
PPARg2 in mature adipocytes, as a nutritional integrator of lipid
uptake and release that controls the availability of FFAs (Rodri-
guez-Cuenca et al., 2012) and gluconeogenic substrates (Perry
et al., 2015; Titchenell et al., 2016) in the fasted state and permits
appropriate lipid storage in the fed state. Because the mouse
model of PPARg2 ablation we have used is a total KO, we cannot
exclude effects of PPARg2 in other organs, but PPARy2 is
almost exclusively expressed in adipose tissue (Escher et al.,
2001). Furthermore, although it is possible that effects on lipid
uptake and release could be secondary to alterations in adipo-
genesis, we observed these defects in chow-fed mice at
an age when PPARg2-KO animals possess normal adipocyte
numbers (Figures 3G and 3H).
Under laboratory conditions of ad libitum access to food, mice
do not eat like humans, consumingmany small meals throughout
the day (Ellacott et al., 2010). Conversely, most humans
consume a small number of large meals a day and occasionally
tend to over-eat. As such, the nutrient load that humans face is
more infrequent, with a larger and more variable amplitude
than that faced by mice. Although not perfectly analogous to
the human breakfast-lunch-dinner feeding pattern, the 1dHFD
model in rodents increases the amplitude of their nutrient intake
and importantly keeps them in circadian phase. The 1dHFD
modelmay therefore provide amore translatablemetabolic chal-
lenge than traditional chronic HFDs for the study of adipose lipid
storage function, and it is likely that other mouse models with
similar defects maybe uncovered using a 1dHFD protocol.
Our results also raise questions regarding how metabolic dis-
ease is diagnosed and studied in humans. Currently there is a
bias toward studying human and mouse metabolism in the
fasted state. Equally, diagnostic tests for human metabolic dis-
ease are conducted almost exclusively in the fasted state to con-
trol for human meal choices. By extension, our data suggest that
there are humans with primary adipose tissue defects who
exhibit normal GTT results but become metabolically dysfunc-
tional after eating. It is possible that a hypercaloric mixed meal
tolerance test would be the most suitable way to detect such in-
dividuals. Finally, our data suggest that ‘‘murinizing’’ human
feeding patterns may be a way to treat individuals with adipose
dysfunction and storage defects.STAR+METHODS
Detailed methods are provided in the online version of this paper
and include the following:
d KEY RESOURCES TABLE
d CONTACT FOR REAGENT AND RESOURCE SHARINGd EXPERIMENTAL MODEL AND SUBJECT DETAILS
B Animals
B Rodent Diets
B Feeding regimes
d METHOD DETAILS
B Glucose tolerance test
B Inlet conditions
B Temperature Program:
B MSD conditions:
B Mouse Euglycaemic-hyperinsulinaemic clamp
B Mouse 2-Deoxyglucose uptake
B Calorimetry and assessment of metabolic flexibility
B Lipid tolerance test
B Liver fat %
B Serum Biochemistry
B RNA extraction and real time PCR
B Western Blotting
B Histological analysis of white adipose tissue
B GC-MS FAME analysis
B Lipid uptake to adipose tissue
B LC-MS analyasis
d QUANTIFICATION AND STATISTICAL ANALYSIS
d DATA AND SOFTWARE AVAILABILLITYSUPPLEMENTAL INFORMATION
Supplemental Information includes four figures and one table and can be
found with this article online at https://doi.org/10.1016/j.celrep.2018.07.063.ACKNOWLEDGMENTS
We thank Helen Westby, Sarah Grocott, Charley Beresford, Jade Bacon,
Laura McKinven, and Agnes Lukasik for their excellent technical assistance.
All animal work was carried out in the Disease Model Core (MRC Metabolic
Diseases Unit [MRC_MC_UU_12012/5]; Wellcome Trust Strategic Award
[100574/Z/12/Z]). All serum biochemistry was conducted by the Biochemistry
Assay Lab (MRCMetabolic Diseases Unit [MRC_MC_UU_12012/5]). We thank
the BHF (RG/12/13/29853), FIS (PI15/01934), and MRC (MC_UU_12012/2) for
funding this work.AUTHOR CONTRIBUTIONS
S.V. designed and conducted experiments and wrote the manuscript. K.P.,
D.H., B.J., and M.D. conducted experiments. A.K. designed and conducted
experiments. J.M.M.-N. and J.M.F.-R. provided and analyzed human
adipose gene expression data. A.V.-P. designed experiments and wrote the
manuscript.DECLARATION OF INTERESTS
The authors declare no competing interests.
Received: November 27, 2017
Revised: March 15, 2018
Accepted: July 18, 2018
Published: August 21, 2018
REFERENCES
Asterholm, I.W., and Scherer, P.E. (2010). Enhancedmetabolic flexibility asso-
ciated with elevated adiponectin levels. Am. J. Pathol. 176, 1364–1376.Cell Reports 24, 2005–2012, August 21, 2018 2011
Asterholm, I.W., Mundy, D.I., Weng, J., Anderson, R.G., and Scherer, P.E.
(2012). Altered mitochondrial function and metabolic inflexibility associated
with loss of caveolin-1. Cell Metab. 15, 171–185.
Danforth, E., Jr. (2000). Failure of adipocyte differentiation causes type II dia-
betes mellitus? Nat. Genet. 26, 13.
Ellacott, K.L., Morton, G.J., Woods, S.C., Tso, P., and Schwartz, M.W. (2010).
Assessment of feeding behavior in laboratory mice. Cell Metab. 12, 10–17.
Escher, P., Braissant, O., Basu-Modak, S., Michalik, L., Wahli, W., and Des-
vergne, B. (2001). Rat PPARs: quantitative analysis in adult rat tissues and
regulation in fasting and refeeding. Endocrinology 142, 4195–4202.
Folch, J., Lees, M., and Sloane Stanley, G.H. (1957). A simple method for the
isolation and purification of total lipides from animal tissues. J. Biol. Chem.
226, 497–509.
Frayn, K.N. (2002). Adipose tissue as a buffer for daily lipid flux. Diabetologia
45, 1201–1210.
Gao, S., McMillan, R.P., Zhu, Q., Lopaschuk, G.D., Hulver, M.W., and Butler,
A.A. (2015). Therapeutic effects of adropin on glucose tolerance and substrate
utilization in diet-induced obese mice with insulin resistance. Mol. Metab. 4,
310–324.
Goodpaster, B.H., and Sparks, L.M. (2017). Metabolic flexibility in health and
disease. Cell Metab. 25, 1027–1036.
Isken, F., Klaus, S., Petzke, K.J., Loddenkemper, C., Pfeiffer, A.F., and
Weickert, M.O. (2010). Impairment of fat oxidation under high- vs. low-glyce-
mic index diet occurs before the development of an obese phenotype. Am. J.
Physiol. Endocrinol. Metab. 298, E287–E295.
Jackowski, S., and Leonardi, R. (2014). Deregulated coenzyme A, loss of
metabolic flexibility and diabetes. Biochem. Soc. Trans. 42, 1118–1122.
Kim, J.Y., van de Wall, E., Laplante, M., Azzara, A., Trujillo, M.E., Hofmann,
S.M., Schraw, T., Durand, J.L., Li, H., Li, G., et al. (2007). Obesity-associated
improvements in metabolic profile through expansion of adipose tissue.
J. Clin. Invest. 117, 2621–2637.
Longo, K.A., Charoenthongtrakul, S., Giuliana, D.J., Govek, E.K., McDonagh,
T., Qi, Y., DiStefano, P.S., andGeddes, B.J. (2008). Improved insulin sensitivity
and metabolic flexibility in ghrelin receptor knockout mice. Regul. Pept. 150,
55–61.
McQuaid, S.E., Hodson, L., Neville, M.J., Dennis, A.L., Cheeseman, J., Hum-
phreys, S.M., Ruge, T., Gilbert, M., Fielding, B.A., Frayn, K.N., and Karpe, F.
(2011). Downregulation of adipose tissue fatty acid trafficking in obesity: a
driver for ectopic fat deposition? Diabetes 60, 47–55.
Medina-Gomez, G., Virtue, S., Lelliott, C., Boiani, R., Campbell, M., Christo-
doulides, C., Perrin, C., Jimenez-Linan, M., Blount, M., Dixon, J., et al.
(2005). The link between nutritional status and insulin sensitivity is dependent
on the adipocyte-specific peroxisome proliferator-activated receptor-
gamma2 isoform. Diabetes 54, 1706–1716.
Medina-Gomez, G., Gray, S.L., Yetukuri, L., Shimomura, K., Virtue, S., Camp-
bell, M., Curtis, R.K., Jimenez-Linan, M., Blount, M., Yeo, G.S., et al. (2007).
PPAR gamma 2 prevents lipotoxicity by controlling adipose tissue expand-
ability and peripheral lipid metabolism. PLoS Genet. 3, e64.
Muoio, D.M., Noland, R.C., Kovalik, J.P., Seiler, S.E., Davies, M.N., DeBalsi,
K.L., Ilkayeva, O.R., Stevens, R.D., Kheterpal, I., Zhang, J., et al. (2012). Mus-2012 Cell Reports 24, 2005–2012, August 21, 2018cle-specific deletion of carnitine acetyltransferase compromises glucose
tolerance and metabolic flexibility. Cell Metab. 15, 764–777.
Perry, R.J., Camporez, J.G., Kursawe, R., Titchenell, P.M., Zhang, D., Perry,
C.J., Jurczak, M.J., Abudukadier, A., Han, M.S., Zhang, X.M., et al. (2015). He-
patic acetyl CoA links adipose tissue inflammation to hepatic insulin resistance
and type 2 diabetes. Cell 160, 745–758.
Randle, P.J., Garland, P.B., Hales, C.N., and Newsholme, E.A. (1963). The
glucose fatty-acid cycle. Its role in insulin sensitivity and the metabolic distur-
bances of diabetes mellitus. Lancet 1, 785–789.
Riachi, M., Himms-Hagen, J., and Harper, M.E. (2004). Percent relative cumu-
lative frequency analysis in indirect calorimetry: application to studies of trans-
genic mice. Can. J. Physiol. Pharmacol. 82, 1075–1083.
Rodriguez-Cuenca, S., Carobbio, S., Velagapudi, V.R., Barbarroja, N., Mor-
eno-Navarrete, J.M., Tinahones, F.J., Fernandez-Real, J.M., Oresic, M., and
Vidal-Puig, A. (2012). Peroxisome proliferator-activated receptor g-dependent
regulation of lipolytic nodes and metabolic flexibility. Mol. Cell. Biol. 32, 1555–
1565.
Storlien, L., Oakes, N.D., and Kelley, D.E. (2004). Metabolic flexibility. Proc.
Nutr. Soc. 63, 363–368.
Titchenell, P.M., Quinn, W.J., Lu, M., Chu, Q., Lu, W., Li, C., Chen, H., Monks,
B.R., Chen, J., Rabinowitz, J.D., and Birnbaum, M.J. (2016). Direct hepatocyte
insulin signaling is required for lipogenesis but is dispensable for the suppres-
sion of glucose production. Cell Metab. 23, 1154–1166.
Ukropcova, B., Sereda, O., de Jonge, L., Bogacka, I., Nguyen, T., Xie, H., Bray,
G.A., and Smith, S.R. (2007). Family history of diabetes links impaired sub-
strate switching and reduced mitochondrial content in skeletal muscle. Dia-
betes 56, 720–727.
Vacanti, N.M., Divakaruni, A.S., Green, C.R., Parker, S.J., Henry, R.R., Ciaraldi,
T.P., Murphy, A.N., andMetallo, C.M. (2014). Regulation of substrate utilization
by the mitochondrial pyruvate carrier. Mol. Cell 56, 425–435.
Vaitheesvaran, B., LeRoith, D., and Kurland, I.J. (2010). MKR mice have
increased dynamic glucose disposal despite metabolic inflexibility, and hepat-
ic and peripheral insulin insensitivity. Diabetologia 53, 2224–2232.
Vidal-Puig, A.J., Considine, R.V., Jimenez-Lin˜an,M.,Werman, A., Pories,W.J.,
Caro, J.F., and Flier, J.S. (1997). Peroxisome proliferator-activated receptor
gene expression in human tissues. Effects of obesity, weight loss, and regula-
tion by insulin and glucocorticoids. J. Clin. Invest. 99, 2416–2422.
Virtue, S., and Vidal-Puig, A. (2010). Adipose tissue expandability, lipotoxicity
and the Metabolic Syndrome–an allostatic perspective. Biochim. Biophys.
Acta 1801, 338–349.
Vroegrijk, I.O., van Diepen, J.A., van den Berg, S.A., Romijn, J.A., Havekes,
L.M., van Dijk, K.W., Darland, G., Konda, V., Tripp, M.L., Bland, J.S., and
Voshol, P.J. (2013). META060 protects against diet-induced obesity and insu-
lin resistance in a high-fat-diet fed mouse. Nutrition 29, 276–283.
Yew Tan, C., Virtue, S., Murfitt, S., Roberts, L.D., Phua, Y.H., Dale, M., Griffin,
J.L., Tinahones, F., Scherer, P.E., and Vidal-Puig, A. (2015). Adipose tissue
fatty acid chain length and mono-unsaturation increases with obesity and in-
sulin resistance. Sci. Rep. 5, 18366.
STAR+METHODSKEY RESOURCES TABLEREAGENT or RESOURCE SOURCE IDENTIFIER
Antibodies
pHSL S660 Cell Signaling Cat 4126; RRID: AB_490997
Total HSL Cell Signaling Cat 4107 ; RRID: AB_2296900
Phospho P44/42 MAPK Cell Signaling Cat 4376; RRID: AB_331772
Total P44/42 MAPK Cell Signaling Cat 4695; RRID: AB_390779
pAKT S473 Cell Signaling Cat 4060; RRID: AB_2315049
Total AKT Cell Signaling Cat 2920; RRID: AB_1147620
B-actin Cell Signaling Cat 8226; RRID: AB_306371
Biological Samples
Human adipose tissue biopsies Endocrinology Service of the
Hospital of Girona ‘Dr Josep
Trueta’
N/A
Chemicals, Peptides, and Recombinant Proteins
D-Glucose (6,6-D2 99%) Goss Scientific DLM-349-1
D-Glucose (2-D, 98%) Goss Scientific DLM-1271-1
Pyridine Sigma 270970-100ml
Hydroxyamine hydrochloride Sigma 431362-50 g
Acetic anhydride Sigma 320102
Ethyl Acetate Sigma 34972
Fentanyl Martindale Pharma PL 00156/0038
Midazolam (Hypnovel) Roche PL00031/0126
Acepromazine Novartis GTIN5037694004606
3-3H glucose Perkin Elmer NET331A250UC
Deoxy-D-glucose, 2-[1-14C] Perkin Elmer NEC495A250UC
Insulin Novo Nordisk EU/1/02/230/003
Olive oil Heritage
STAT 60 AMS Biotech CS-502
Reverse transcriptase kit N/A N/A
Reverse Transcriptase M-MLV Promega M170b
M-MLV RT 5x buffer Promega M351A
MgCl2 25mM Promega A351B
Random Primers Promega C118A
dNTP Mix 10 mM Promega U151B
RNeasy Lipid Tissue Mini Kit QIAGEN 74804
Chlorofom Sigma 34854
Methanol Sigma 34860
BF3 Methanol Sigma B1127
Hexane Sigma 34859
1-14C Palmitate Perkin Elmer NEC075H050UC
BondElut NH2 columns Agilent 12102014
Isolute Strong anion exchange columns, 100 mg
bed, 6 ml reservoir.
Biotage 500-0010-C
Formic acid Sigma F0507-500ML
Ammonium acetate Sigma A1542-250G
(Continued on next page)
Cell Reports 24, 2005–2012.e1–e7, August 21, 2018 e1
Continued
REAGENT or RESOURCE SOURCE IDENTIFIER
Critical Commercial Assays
Alpha Trak 2 glucose meter Zoetis N/A
Insulin Elisa (Mouse Clamp studies) Crista Chem 90080
NEFA assay (Mouse clamp studies) WAKO R1 436-91995
R2 434-91795
S: 270-77000
Triglyceride assays Siemens Healthcare Diagnostics N/A
Roche Free Fatty Acid Kit (half-micro test) Roche 1138175001
Insulin ECL Meso Scale Discovery K152BZC
Deposited Data
Raw data This paper https://data.mendeley.com/datasets/
whn4s7nyww/draft?a=22693287-b601-
457c-85a9-9903c6d88639; Reserved:
https://doi.org/10.17632/whn4s7nyww.2
Experimental Models: Organisms/Strains
Mouse: C57BL/6J The Jackson Laboratory JAX: 000664
Mouse: PPARy2 KO C57BL/6J.sv129 (Medina-Gomez et al., 2005) N/A
Oligonucleotides
PPARg2 F CCAACCAATCTTTTGCAAGACATAGAC (Medina-Gomez et al., 2005) N/A
PPARg2 R ACATGCAATTTCACCCACACATGAGTG (Medina-Gomez et al., 2005) N/A
PPARg2 Asc AATGGCCGCTTTTCTGGATTCATCGAC (Medina-Gomez et al., 2005) N/A
Sybr Green Primer and TaqMan primer and probe
sequences see Table S1
N/A N/A
GLUT4 TaqMan primer and probe set (Human) ThermoFisher Hs00168966_m1
Human PPIA (Cyclophilin A) Endogenous Control
(FAM/MGB probe, non-primer limited)
ThermoFisher 4333763F
IRS1 TaqMan primer and probe set (Human) ThermoFisher Hs00178563_m1
Software and Algorithms
IBM SPSS Statistics 22 IBM corporation N/A
MassHunter Workstation Software Quantitative
Analysis (Version B.07.00)
Agilent Technologies Inc N/ACONTACT FOR REAGENT AND RESOURCE SHARING
Further information and requests for resources and reagents should be directed and will be fulfilled by Dr. Samuel Virtue
(sv234@medschl.cam.ac.uk).
EXPERIMENTAL MODEL AND SUBJECT DETAILS
Animals
PPARg2 KOmicewere generated as described previously (Medina-Gomez et al., 2005).Micewere phenotyped on amixedC57BL/6/
sv129 genetic background. Mice homozygous for a deletion in PPARg2 (PPARg2
/
) and their wild-type littermates were generated
by mating heterozygous mice. All animals used were male mice. Ages are provided in figure legends. C57BL/6 wild-type mice were
purchased fromCharles River UK. This research has been regulated under the Animals (Scientific Procedures) Act 1986 Amendment
Regulations 2012 following ethical review by the University of Cambridge AnimalWelfare and Ethical ReviewBody (AWERB). Animals
were housed in a specific pathogen free facility with 12 hour light and 12 hour dark cycles. Unless otherwise stated all animals were
studied under fed conditions and at 24C.
Rodent Diets
Mice were fed either standard breeders chow (Safe Diets, DS-105) or a 45% fat diet (HFD) (Research Diets D12451) all diets were
provided ab libitum.e2 Cell Reports 24, 2005–2012.e1–e7, August 21, 2018
Feeding regimes
For the one day high-fat diet challengemice were single housed 1 week prior to the start of the diet to permit food intake assessment.
The diet commenced in a staggered manner between 10 am and midday. Mice were alternated in genotype for both feeding of the
diet and also for culling. Mice were killed in the same sequence that they were fed the diet in to keep the mice as close to receiving
high-fat diet for 24 hours as possible. No significant correlation between the order the mice were killed in and the levels of FFAs, in-
sulin or TG in blood was detected for the time course of the experiment. 1 month and 1 week HFD fed animals mice were culled be-
tween 10 am and 12pm in order to maintain the animals in the same stage of the circadian cycle as the chow controls and the 1 day
high-fat fed animals
Patient recruitment
Forty five subcutaneous adipose tissue (SAT) samples from morbidly obese (BMI > 35 kg/m2) participants with different degrees of
insulin action (measured using hyperinsulinaemic–euglycaemic clamp) were analyzed. All these participants were of Caucasian
origin, reported that their bodyweight had been stable for at least threemonths before the study andwere studied in the post-absorp-
tive state. BMI was calculated as weight (in kg) divided by height (in m) squared. Patients had no systemic disease other than obesity
and all were free of any infections in the previous month before the study. Liver diseases (specifically tumoral disease and hepatitis
C virus infection) and thyroid dysfunction were specifically excluded by biochemical work-up. All these participants were recruited at
the Endocrinology Service of the Hospital of Girona ‘Dr Josep Trueta’. All participants gave written informed consent, validated and
approved by the Ethics Committee of the Hospital of Girona ‘Dr Josep Trueta’, after the purpose of the study was explained to them.
METHOD DETAILS
Glucose tolerance test
For fasted glucose tolerance testsmicewere fasted overnight from 4pmuntil 9 am the next day and then allowed to acclimatise to the
procedural room for 1 hour. For fed glucose tolerance tests mice were allowed to acclimatise to the procedural room for one hour
between 9 and 10 am. Glucose was administered at 1 g/kg IP. A fixed dose of glucose was given to all mice in the study group based
on the average weight of the group. Fed glucose tolerance tests were performed at 10 am in the morning. Blood samples were
collected at 0, 10, 20, 30, 60, 90 and 120 minutes.
For stable isotope GTTsmice were injected with a 50:50mix of D2-6,6 Glucose and D2-2 Glucose (Goss Scientific, UK). Bloodwas
collected on blood spot cards. A 6 mm punch was removed from each blood spot card and extracted by wetting with 40 ul of water
followed by addition of 400 ul of ethanol. Samples were shaken for 45minutes at room temperature before undergoing centrifugation
at 16,000 g. EtOH extracts were dried under vacuum at ambient temperature using a Speed Vac. Extracts underwent aldonitrile pen-
taacetate derivitisation as follows: Dried extracts were resuspended in 50 ul of pyridine containing 2% w/v hydroxylamine. Samples
were heated for 45 minutes at 90C then allowed to cool. After cooling 100 ul of acetic anhydride was added and samples were heat-
ed for 30 minutes at 60C for 30 minutes. Samples were allowed to cool and then dried under nitrogen at 60C. Dried samples were
dissolved in 200 ul of ethyl acetate, centrifuged at 16,000 g for 1 minute and the top 150 ul of supernatant was transferred to an au-
tosampler vial containing a 300 ul glass insert. Gas-Chromatography Mass spectrometry analysis was performed using an Aglient
7890B GC and an Agilent 5977A MSD using a 0.25 uM x 30 M DB-5 (Cat: 122-5031, Agilent) column. The GC-Conditions were as
follows:
Inlet conditions
Inlet temperature: 250C
Split 20:1
Inlet liner Liner, ultra inert, splitless, single taper, glasswool (cat#6322616) (Agilent Technologies)
Column Flow 1 ml/min
Temperature Program:
80C Hold for 1 minute
20C/min until 280C
280C Hold for 3 minutes
MSD transfer line 290C
MSD conditions:
Scan 45-850, 4 Hz
MS Source temperature 230C
MS Quad temperature 150CCell Reports 24, 2005–2012.e1–e7, August 21, 2018 e3
Analysis was performed usingMSQuantitative analysis. Ions analyzed were 217 (M0) 218 (M1), 219 (M2) 220 (M3) and 221 (M4). To
determine the relative ratio of endogenous to exogenous glucose blood samples were compared to the glucose composition of
mouse blood from animals uninjected with labeled glucose and a sample of D2-6,6 glucose (10 ul, 10mM) that was dried onto a blood
spot card and extracted as described above). The ion pattern of each blood sample was compared to the tracer and endogenous
blood glucose. To determine the relative contribution of endogenous and tracer glucose we defined 5 equations for each ion as
follows:
Equation System 1 :
M0calc= x M0Endo+ ð1­xÞ M0Exo
M1calc= x M1Endo+ ð1­xÞ M1Exo
M2calc= x M2Endo+ ð1­xÞ M2Exo
M3calc= x M3Endo+ ð1­xÞ M3Exo
M4calc= x M4Endo+ ð1­xÞ M4Exo
Where x represents the Fractional concentration of the endogenous glucose in the blood and 1-x the Fractional concentration of
exogenous glucose in the blood.
We defined a further 5 equations:
Equation System 2 :
M0obs­M0calc
M1obs­M1calc
M2obs­M2calc
M3obs­M3calc
M4obs­M4calc
We then used the Solver function of Excel to minimize the sum of the squares of the result of the 5 equations by changing the value of
x in equation system 1 using the GRG Non-Linear Engine.
The solution for x defined the fractional concentration of endogenous glucose and 1-x the fractional concentration of D2-6,6
glucose. As D2-6,6 glucose represented 50% of the glucose injected in the GTT we multiplied the fractional concentration of
D2-6,6 glucose by 2 to get the fractional concentration of the exogenous glucose and multiplied it by the observed blood concen-
tration of glucose to get the concentration of exogenous glucose in mMol. By subtraction the exogenous glucose from the mouse’s
total blood glucose concentration we obtained the endogenous glucose concentration.
Mouse Euglycaemic-hyperinsulinaemic clamp
Animals were fasted from 4 pm until 8 am the next morning. Animals were anaesthetised with 11 ml per gram of body weight using a
combination of the following drugs at stated concentrations fentanyl (0.03125 mg/ml), midazolam (0.625 mg/ml), and acepromazine
(0.625 mg/ml). Once unconscious mice were placed on temperature controlled heating pads (Harvard Apparatus) and core body
temperature was maintained at 37C. Mice underwent a basal infusion period. The basal infusate comprised of 3-3H glucose
(0.23MBq/ml) and 3mg/ml sodium citrate in 0.9% saline. The basal infusion rate was 100 ul/hour and was conducted for 90minutes.
For the clamp phase the basal infusate was stopped and the hyperinsulinaemic infusion was started. The hyperinsulinaemic infusate
comprised of 0.11 mU per ml of insulin, 0.231 MBQ/ml of 3-3H glucose, 3 mg/ml sodium citrate, 2% FFA-free low-endotoxin BSA in
0.9% saline. The hyperinsulinaemic infusion was initiated with a bolus of 30 ul of infusate followed by 100 ul/hour. A variable rate
infusion of glucose was then started using 12.5% glucose in 0.9% saline with 3 mg/ml sodium citrate. Blood glucose was measured
every 5 minutes and glucose infusion rate was adjusted to clamp mice to their basal glucose levels. The hyperinsulinaemic infusion
was continued for at least 90 minutes, or if mice had not reached steady state until steady state was achieved. Steady state was
defined as mice exhibiting blood glucoses that were ± 0.5 mMol of their starting blood glucose and having a stable glucose infusion
rate for three consecutive measurements. Blood samples were collected for the determination of FFA, Insulin and serum specific
activity at the end of the basal and clamped states.
Mouse 2-Deoxyglucose uptake
Mice were fed a high-fat diet for 1 day, with mice placed on the diet between 8 and 9 am. The following day mice were anaesthetised
in the sequence they were anaesthetised between 8 and 9 am. The anesthetic was 11 ml per gram of body weight using a combination
of the following drugs at stated concentrations fentanyl (0.03125mg/ml), midazolam (0.625mg/ml), and acepromazine (0.625mg/ml).
Micewere injected IVwith 0.046MBqof 14C 2-Deoxyglucose and blood glucose levels weremeasured and serum samples collected
at 10,20 and 30 minutes post injection in order to determine the specific activity of glucose in the blood. Mice were killed 30 minutes
after injection with 14C-2-Deoxyglucose and tissues were harvested.
Phospho deoxyglucose was separated from deoxyglucose using solid-phase extraction with strong anion exchange columns. Tis-
sues were weighed (100 mg of WAT and 50 mg of BAT) and homogenized in 1 mL of distilled water in glass tubes. The homog-
enates were heated to 95C for 10 minutes before cooling to room temperature. Homogenates were then transferred to eppendorf
tubes and centrifuged at 16,000 G and 66 ul of homogenate was taken for scintillation counting. SAX-SPE columns were conditionede4 Cell Reports 24, 2005–2012.e1–e7, August 21, 2018
with 5ml of distilled water and then collection vials placed below the columns before transferring 660 ul of homogenate to the column.
Columns were washed 3 times with 2 mL of distilled water. The wash fractions were vortexed and 500 ul of the wash fraction was
added to 20 mL of scintillant for counting. New collection columns were placed under the SPE columns and the phosphodeoxyglu-
cosewas eluted three times using 2mL of 0.2M formic acid and 0.5Mammonium acetate pH 5.0. The elution fractions were vortexed
to mix them and 500 ul was added to 20 mL of scintillant and counted immediately.
Calorimetry and assessment of metabolic flexibility
Mice underwent calorimetry in a custom built calorimetry system (Creative Scientific UK) for up to 48 hours. Carbon dioxide and
oxygen concentrations were determined every 11 minutes for each chamber and the incurrent air supply. Flow rates were
400 ml/minute. Energy expenditure was calculated from the VO2 and CO2 according to the modified Weir equation (EE J/min =
15.818xVO2 (ml/min) + 5.176*VCO2 (ml/min)). Metabolic flexibility was assessed bymeasuring the amplitude of respiratory exchange
ratio (RER) frommice in free living calorimetry chambers. Amplitude of RERwas defined as the difference between the average of the
lowest 10% of RER values subtracted from the average of the highest 10% of RER values (dRER). Amplitude of RER has been es-
tablished as a read out of metabolic flexibility by multiple groups (Asterholm et al., 2012; Asterholm and Scherer, 2010; Gao et al.,
2015; Isken et al., 2010; Jackowski and Leonardi, 2014; Longo et al., 2008; Muoio et al., 2012; Riachi et al., 2004; Vaitheesvaran
et al., 2010; Vroegrijk et al., 2013).
Lipid tolerance test
Mice were fasted overnight from 4 pm until 9 am the next day. Mice were gavaged with 200 ml olive oil (Heritage, NISA, UK). Blood
samples were collected at 0,1,2,4 and 8 hours post gavage for determination of TG and FFA levels. See Serum biochemistry for de-
tails of TG and FFA measurement.
Liver fat %
100-200mg of liver was weighed and lipids extracted according to the Folch extraction procedure (Folch et al., 1957). Briefly, 100mg
of liver was homogenized in 1mL volumes (w/v) of 2:1 chloroformmethanol using aMPBiomedical Fast Prep. Homogenized samples
were centrifuged (16,000 G for 10minutes) and the supernatant transferred to a clean Eppendorf tube. 200 ul of water was added and
samples vortexed for 2 minutes. Samples were centrifuged (16,000 G for 10 minutes) and the lower organic phase was collected into
a pre-weighed Eppendorf tube. A second round of extraction was carried out by adding 700 ul of chloroform to the sample and vor-
texing (2 minutes) and centrifuging (16,000 G for 10 minutes). The lower organic phase was collected again and combined with the
previous extract. The extract was then dried under nitrogen. The weight of the tube and extracted lipid wasmeasured and the weight
of the empty tube subtracted to give the quantity of lipid in the sample. The liver fat extract was expressed as a% of wet liver weight.
Serum Biochemistry
Triglycerides weremeasured on the Dimension RXL analyzer (Siemens Healthcare). Free Fatty Acidsweremeasured using the Roche
Free Fatty Acid Kit (half-micro test) (kit code 11383175001). Insulin, was measured using electrochemical luminescence immuno-
assay on the MesoScale Discovery immunoassay platform.
RNA extraction and real time PCR
RNA was extracted using STAT-60 (AMS biotech) according to manufacturer’s procedures. Reverse transcription (500 ng of total
RNA) was performed using random hexamers using the Reverse Transcriptase System (Promega) according to manufacturer’s in-
structions. Real-time PCR was carried out in 13ul reactions using TaqMan or Sybr Green reagents and performed in a 384 well plate
on an ABI7900 real-time PCR machine using default thermal cycler conditions. For PCR primer sequences see Table S1.
Western Blotting
Protein was extracted from tissues using a modified RIPA buffer (50 mM Tris HCL pH 8, 150 mM NaCl, 1% NP-40, 0.5% Sodium
Deoxycholate, 0.1% SDS) and quantified by the BioRad DC protein assay. Protein was heated with loading dye containing DTT
and heated to 95oC for 5 minutes to denature proteins before loading. 10 ug per well of protein was and subjected to SDS-PAGE
in a 4%–12% gradient gel using the Novex NuPage midi system (Life Technologies) and transferred using the iBlot transfer system
and reagents (Life Technologies) for 7 minutes. Membranes were probed for: pHSL S660 (Cat: 4126 Cell Signaling, USA) total HSL
(Cat: 4107Cell Signaling, USA), Phospho-p44/42MAPK (Cat: 4376Cell Signaling, USA), total p44/42MAPK (Cat: 4695 Cell Signaling,
USA), Phospho AKT (Cat: 4060 Cell Signaling, USA), total-AKT (Cat: 2920 Cell Signaling, USA) and B-actin (Cat: 8226 Abcam, UK).
Histological analysis of white adipose tissue
Adipose tissue was fixed in formalin. Tissues were embedded in paraffin and sections were cut at a thickness of 100 mM. Slides were
imaged using an automated slide scanning microscope (Axioscan Z1 using a Hamamatsu orca flash 4.0 V3 camera) using a 20x
objective with a numerical aperture of 0.8. Adipocytes size was quantified using Halo (Indica Labs inc) utilizing the Vacuole module
v.1.4. The cut off point for minimum vacuole diameter was set at 15 and the maximum at 150 mm. Adipocyte, blood vessels and slide
background were determined using the tissue classifier method, only adipocytes were quantified.Cell Reports 24, 2005–2012.e1–e7, August 21, 2018 e5
GC-MS FAME analysis
Lipid extraction and analysis were described in detail previously (Yew Tan et al., 2015). Briefly, lipids were extracted by the Folch
extraction procedure. Lipids were esterified to form methyl esters using BF3 methanol. Methyl esters were analyzed by Gas-chro-
matography Mass-Spectrometry. The identity of species was determine by retention time andmass spectra and compared to a food
industry fame standard (Thames Restek, UK). Lipid species were quantified based on integrated peak area.
Lipid uptake to adipose tissue
Mice were clamped as described above. 30 minutes from the end of the clamp a bolus of 0.2 MBQ of 1-14C palmitate conjugated to
2% BSA in PBS was injected IV. After 30 minutes serum and adipose tissue was collected. Serum DPMs were measured directly.
Adipose tissue was weighed and lipids extracted using the Folch method (see ‘‘Liver fat%’’ above). Lipids were separated into
Neutral Lipid, Phospholipid and FFA fractions by solid phase extraction. Solid phase extraction was performed by resuspending
the adipose tissue extract in 1mL of dry chloroform. Of this 1 mL 100 ul was reserved for scintillation counting and 900 ul was applied
to the SPE column and the flow through collected. Two further washes of 1 mL of dry chloroform were performed and combined with
the initial flow through; this represented the neutral lipid fraction. Phospholipids were then eluted from the columns into a fresh tube
using 2 mL of Chloroform:Methanol 60:40). The NEFA fraction was eluted in 2.0ml chloroform/methanol/glacial acetic acid (100:2:2).
All SPE stepswere performed using a vacuummanifold. Unlikemany SPE lipid extraction protocols, no activation of the columnswas
required and between collection of fractions the bed was dried under vacuum for 2-3 s to minimize carry over between fractions.
LC-MS analyasis
Sample extraction
The pre-weighedmuscle tissue samples (10-20mg) were transferred to a 2mL Eppendorf screw cap tubes. A single 5mm stainless
steel ball bearing and 400 mL of the chloroform:methanol mix solution (2:1, respectively) were added to the tissue samples, followed
by a 3 s cycle of vigorous vortexing. The tissues were then homogenized in the extraction solvent using a Bioprep-24-1004 homog-
enizer (Allsheng, Hangzhou City, China) run at speed; 6 m/s, time; 30 s for 2 cycles. Following the homogenization, 150 mL of the sta-
ble isotope labeled lipid internal standard mix was added along with 10 mL the stable isotope 5-hydroxytryptamine-d4 acylcarnitine
internal marker. The mixtures was then thoroughly vortexed before the addition of 600 mL of the chloroform:methanol mix solution
(2:1, respectively). The samples were homogenized again (speed; 6 m/s, time; 30 s for 2 cycles). To the homogenates, 400 mL of
HPLC-water was added with an additional cycle of vortexing. The samples were then centrifuged (5 mins at 21,000 g) to produce
aqueous and organic biphasic extracts. The two phases were then separated by pipetting each into separate 2 mL Eppendorf screw
cap vials making sure not to break up the undissolved protein pellet. To perform a double extraction, 1mL of the chloroform:methanol
mix solution (2:1, respectively) was added to each sample, then vortexed thoroughly. The samples were then homogenized on the
tissuelyser (speed; 6m/s, time; 30 s for 2 cycles). Then 400 mL of HPLC-water was added to each sample, vortexed and homogenized
on the tissuelyser (speed; 6 m/s, time; 30 s for 2 cycles). The samples were then centrifuged (5 mins at 21,000 g) to produce the
biphasic extracts, followed by phase separation into the corresponding 2 mL Eppendorf screw cap vials containing the first extracts.
Acylcarnitine sample preparation
Half of the corresponding organic fractions and half the aqueous fraction were mixed with each other into separate 2 mL Eppendorf
screw cap vials. Themixed extracts were dried completely under a gentle stream of oxygen free nitrogen heated to 45C. To the dried
extracts, 200 mL of water:acetonitrile (4:1, respectively) was added to reconstitute the acylcarnitine species. The samples were then
vortexed thoroughly. The samples were then transferred into 2 mL amber glass vials containing a 300 mL vial-insert which were then
analyzed by liquid chromatography with mass spectrometry detection.
Acylcarnitine LC-MS
Chromatographic separation was achieved using an ACE Excel 2 C18-PFP (150mm * 2.1mm, 2 mm) LC-column with a Shimadzu
UPLC system (Shimadzu UK Limited, Wolverton, Milton Keynes). The column was maintained at 55Cwith a flow rate of 0.5 mL/min.
A binary mobile phase system was used with mobile phase A; water (with 0.1% formic acid), and mobile phase B; acetonitrile (with
0.1% formic acid). The gradient profile was as follows; at 0 minutes_0% mobile phase B, at 0.5 minutes_100% mobile phase B, at
5.5 minutes_100% mobile phase B, at 5.51 minutes_0% mobiles phase B, at 7 minutes_0% mobile phase B. Mass spectrometry
detection was performed on a Thermo Exactive orbitrap mass spectrometer (Thermo Scientific, Hemel Hempstead, UK) operating
in positive ion mode. A heated electrospray source was used; the sheath gas was set to 40 (arbitrary units), the aux gas set to 15
(arbitrary units) and the capillary temperature set to 250C. The instrument was operated in full scan mode from m/z 75–1000 Da.
Acylcarnitine species were identified by detecting a signal peak for the corresponding accurate mass at the correct retention
time. Signals were normalized to the dried tissue mass (dried protein pellet left after the extraction), their final semiquantitative con-
centrations were determined by comparing the acylcarnitine species intensity to the internal marker intensity.
Lipid sample preparation, 50 mL of the organic fractions were transferred into 2mL amber glass vials containing a 300 mL vial-insert.
The samples were then dried completely under a gentle stream of oxygen free nitrogen at 45C. Then, 90 mL of isopropanol:acetoni-
trile (4:1, respectively) was added to reconstitute the lipid species. The samples were then vortexed thoroughly before being analyzed
by liquid chromatography with mass spectrometry detection.
Lipid LC-MS, Chromatographic separation was achieved using aWaters Acquity UPLC CSHC18 (50mm * 2.1mm, 1.7 mm) LC-col-
umn with a Shimadzu UPLC system (Shimadzu UK Limited, Wolverton, Milton Keynes). The column was maintained at 55C with ae6 Cell Reports 24, 2005–2012.e1–e7, August 21, 2018
flow rate of 0.5 mL/min. A binary mobile phase system was used with mobile phase A; acetonitrile:water mix (6:4, respectively, with
10 mM ammonium formate), and mobile phase B; isopropanol:acetonitrile mix (9:1, respectively, with 10 mM ammonium formate).
The gradient profile was as follows; at 0 minutes_40% mobile phase B, at 0.4 minutes_43% mobile phase B, at 0.45 minutes_50%
mobile phase B, at 2.4 minutes_54% mobile phase B, at 2.45 minutes_70% mobile phase B, at 7 minutes_99% mobile phase B, at
8 minutes_99% mobile phase B, at 8.3 minutes_40% mobile phase B, at 10 minutes_40% mobile phase B. Mass spectrometry
detection was performed on a Thermo Exactive orbitrap mass spectrometer (Thermo Scientific, Hemel Hempstead, UK) operating
in positive ion and negative ion continuous switchingmode. Heated electrospray sourcewas used; the sheath gaswas set to 40 (arbi-
trary units), the aux gas set to 15 (arbitrary units) and the capillary temperature set to 300C. The instrument was operated in full scan
mode from m/z 150–1200 Da. Lipid species were identified by detecting a signal peak for the corresponding accurate mass at the
correct retention time. Signals were normalized to the dried tissue mass (dried protein pellet left after the extraction), their final semi-
quantitative concentrations were determined by comparing the lipid species intensity to the appropriate stable isotope labeled lipid
internal standard.
Human Adipose tissue handling
Adipose tissue samples were obtained from SAT depots during elective surgical procedures (cholecystectomy, abdominal hernia
surgery and gastric bypass surgery). Samples of adipose tissue were immediately transported to the laboratory (5-10 min). The
handling of tissue was carried out under strictly aseptic conditions. Adipose tissue samples were washed in PBS, cut with forceps
and scalpel into small pieces (100 mg) and immediately flash-frozen in liquid nitrogen before being stored at 
80C.
Hyperinsulinemic-euglycemic clamp in humans
After an overnight fast, two catheters were inserted into an antecubital vein, one for each arm, used to administer constant infusions
of glucose and insulin, and to obtain arterialized venous blood samples. A 2-h hyperinsulinaemic-euglycaemic clampwas initiated by
a two-step primed infusion of insulin (80mU/m2/min for 5 min, 60mU/m2/min for 5 min) immediately followed by a continuous infusion
of insulin at a rate of 40mU/m2/min (regular insulin (Actrapid, NovoNordisk, NJ)). Glucose infusion began atminute 4 at an initial perfu-
sion rate of 0.011 mmol/kg/min being then adjusted to maintain plasma glucose concentration at 4.9 – 5.5 mmol/l. Blood samples
were collected every 5minutes for determination of plasma glucose and insulin. Insulin sensitivity was assessed as themean glucose
infusion rate during the last 40 min. In the stationary equilibrium, the amount of glucose administered (M) equals the glucose taken by
the body tissues and is a measure of overall insulin sensitivity.
RNA expression for human SAT
RNA purification was performed using an RNeasy Lipid Tissue Mini Kit (QIAgen, Izasa, Barcelona, Spain) and the integrity was
checked using an Agilent Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA). Gene expression was assessed by real-time
PCR using a LightCycler 480 Real-Time PCR System (Roche Diagnostics, Barcelona, Spain), using TaqMan and SYBRgreen tech-
nology suitable for relative genetic expression quantification. The RT-PCR reaction was performed in a final volume of 12 mL. The
cycle program consisted of an initial denaturing of 10 min at 95C then 40 cycles of 15 s denaturizing phase at 95C and 1 min an-
nealing and extension phase at 60C. A threshold cycle (Ct value) was obtained for each amplification curve and a DCt value was first
calculated by subtracting theCt value for human cyclophilin A (PPIA) RNA from theCt value for each sample. Fold changes compared
with the endogenous control were then determined by calculating 2-DCt, so gene expression results are expressed as expression ratio
relative to PPIA gene expression according to manufacturers’ guidelines. The commercially available and pre-validated TaqMan
primer/probe sets used were as follows: Peptidylprolyl isomerase A (cyclophilin A) (4333763, PPIA as endogenous control), insulin
receptor substrate 1 (IRS1, Hs00178563_m1), solute carrier family 2 member 4 (SLC2A4 or GLUT4. Hs00168966_m1). Human
PPARg1 [forward: 50-GGCCGCAGATTTGAAAGAAG-30 and reverse: 50-GGAGAGATCCACGGAGCTGAT-30] and PPARg2 [forward:
50-AAACCCCTATTCCATGCTGTTATG-30 and reverse: 50- TGTCAACCATGGTCATTTCTTGTG-30] weremeasured using SYBRGreen
technology.
QUANTIFICATION AND STATISTICAL ANALYSIS
All statistics were performed using SPSS 25. Data points were excluded if they exhibit a value of more than two standard deviations
from the mean. For all metabolic tests animals were randomly ordered into metabolic chambers (calorimetry) or to order in which
experiments were conducted (GTT, Euglycaemic-hyperinsulinaemic clamps). Statistical significant was set at a p value of < 0.05.
Specific tests are detailed in the figure legends.
DATA AND SOFTWARE AVAILABILLITY
https://data.mendeley.com/datasets/whn4s7nyww/draft?a=22693287-b601-457c-85a9-9903c6d88639Cell Reports 24, 2005–2012.e1–e7, August 21, 2018 e7