1338 VOLUME 19 | NUMBER 10 | OCTOBER 2013 nature medicine
t e c H n i c a L r e P O r t s
White adipose tissue displays high plasticity. We developed
a system for the inducible, permanent labeling of mature
adipocytes that we called the AdipoChaser mouse. We
monitored adipogenesis during development, high-fat diet
(HFD) feeding and cold exposure. During cold-induced
‘browning’ of subcutaneous fat, most ‘beige’ adipocytes
stem from de novo–differentiated adipocytes. During
HFD feeding, epididymal fat initiates adipogenesis
after 4 weeks, whereas subcutaneous fat undergoes
hypertrophy for a period of up to 12 weeks. Gonadal fat
develops postnatally, whereas subcutaneous fat develops
between embryonic days 14 and 18. Our results highlight
the extensive differences in adipogenic potential in
various fat depots.
The ongoing obesity epidemic in both the western and developing
worlds has raised awareness of the complex physiology of adipose
tissue. It is widely appreciated that anatomically distinct adipose tis-
sues differ substantially in their contributions to energy balance and
nutrient homeostasis. Adipose tissue distribution is a strong predictor
of the occurrence of the metabolic syndrome in the context of obesity.
Obese individuals who preferentially expand visceral adipose tissue
are at a greater risk for diabetes and cardiovascular disease than are
equally obese individuals who store excess energy in subcutaneous
adipose tissue1–3. In fact, the expansion of subcutaneous adipose tis-
sue can be potently protective against metabolic complications of
HFD feeding4,5.
The mechanism by which individual adipose depots expand may
also be a critical determinant of the metabolic syndrome in obesity.
In principle, adipose tissue expansion can occur through an enlarge-
ment in adipocyte size (hypertrophy) or an increase in adipocyte
numbers (hyperplasia). Differentiated adipocytes are post-mitotic;
therefore, hyperplasia represents an increase in de novo adipocyte
formation (adipogenesis). Adipocyte hypertrophy is closely linked to
adipose dysfunction: this pathological expansion of white adipose tis-
sue (WAT) is a major component of the metabolic syndrome in obese
individuals6. Both adipocyte hypertrophy and hyperplasia contrib-
ute to adipose tissue expansion during HFD challenge6,7. However,
it remains under debate whether adipogenesis is induced immedi-
ately after HFD exposure and whether there are depot differences in
response to HFD with respect to adipogenesis8,9. Recent studies have
highlighted the critical role of in vivo adipogenesis as it relates to key
adipogenic precursor cells10–15.
WAT is known to have high physiological plasticity. Exposure to
cold or pharmacological treatment with β-adrenergic receptor (β3)
agonists that enhance lipolysis in adipocytes triggers the appearance
of a subset of potentially beneficial adipocytes within the WAT that
are positive for uncoupling protein 1 (UCP1) and share additional
characteristics with brown adipocytes16–19. Unlike classic brown
adipocytes, these ‘brown-like’ fat cells, termed beige adipocytes, do
not derive from precursors that are positive for myogenic factor 5
(Myf5)20. Recent studies have suggested that in subcutaneous adi-
pose tissue, newly induced brown adipocytes are not associated with
precursor proliferation21 and may arise from the transdifferentiation
of existing white adipocytes22–24. Conversely, others have found that
there is a CD137+ precursor cell population that is at rest in subcuta-
neous adipose tissue and can be activated to differentiate into brown
adipocytes after appropriate stimulation15. However, to date, there
has been no direct evidence found for the unambiguous lineage of
these beige adipocytes. Leptin reporter mice have been valuable in
identifying early waves of adipogenesis occurring developmentally25;
however, inducible lineage-tracing mouse models to examine adipo-
genesis are still lacking.
To better understand the dynamics of adipocytes in different fat
depots, we developed a doxycycline-inducible, mature adipocyte–
specific tracing system that we refer to as the AdipoChaser mouse.
We are thus in a unique position to answer the questions outlined
above in vivo with a high degree of temporal resolution. By using a
pulse-chase system that allows us to label all pre-existing mature adi-
pocytes, we uncovered the differential adipogenic capacity of epidi-
dymal and subcutaneous adipose tissue during HFD-induced adipose
tissue expansion. Our studies also offer insights into the unresolved
issue of whether brown-like fat cells induced by cold exposure or β3
agonist treatment in subcutaneous adipose tissue arise from existing
mature adipocytes or derive from de novo differentiation. Most nota-
bly, we found that cold exposure or β3 agonist stimulation induces
massive white adipogenesis in epididymal adipose tissue. In addition,
our labeling system allowed us to determine the exact developmental
time frame of adipocyte differentiation in gonadal and subcutaneous
adipose tissue.
Tracking adipogenesis during white adipose tissue
development, expansion and regeneration
Qiong A Wang1, Caroline Tao1, Rana K Gupta1 & Philipp E Scherer1,2
1Touchstone Diabetes Center, Department of Internal Medicine, The University of Texas Southwestern Medical Center, Dallas, Texas, USA. 2Department of Cell Biology,
The University of Texas Southwestern Medical Center, Dallas, Texas, USA. Correspondence should be addressed to P.E.S. (philipp.scherer@utsouthwestern.edu).
Received 7 May; accepted 9 July; published online 1 September 2013; doi:10.1038/nm.3324
np
g
©
20
13
N
at
ur
e A
m
er
ic
a,
In
c.
A
ll
rig
ht
s
re
se
rv
ed
.
t e c H n i c a L r e P O r t s
nature medicine VOLUME 19 | NUMBER 10 | OCTOBER 2013 1339
RESULTS
The contribution of hyperplasia to adipose tissue expansion
In order to study the fate of mature adipocytes under different meta-
bolic challenges, we developed the AdipoChaser mouse, which is an
inducible adipocyte-tagging system. This model is a combination of
three transgenic lines that we and other labs have generated: the adi-
ponectin promoter–driven tetracycline-on (Tet-on) transcription fac-
tor rtTA (adiponectinP-rtTA)26, a Tet-responsive Cre (TRE-cre) line
that can be activated by rtTA in the presence of doxycycline27 and a
transgenic line carrying Rosa26 promoter–driven loxP-stop-loxP–β-
galactosidase (Rosa26-loxP-stop-loxP-lacZ)28 (Fig. 1a). In the absence
of doxycycline, there was no LacZ expression in mature adipocytes.
After treatment with doxycycline, rtTA activates the TRE promoter to
induce cre expression. Cre protein subsequently eliminates the floxed
transcriptional stop cassette and permanently turns on LacZ expres-
sion in all mature adipocytes present during doxycycline exposure.
Cells can be stained blue with an appropriate β-galactosidase (β-gal)
substrate. After 2 weeks of doxycycline treatment, all adipocytes in
both the epididymal and subcutaneous white adipose tissues (eWAT
and sWAT, respectively) of the triple transgenic (AdipoChaser) mice
were uniformly labeled blue (Fig. 1b,c), reflecting the presence of
β-gal activity. Other cell types, such as endothelial cells in the subcu-
taneous adipose tissue, were not labeled (Fig. 1c). We also used β-gal
substrate to stain adipocytes from control (containing only TRE-cre
and Rosa26-loxP-stop-loxP-lacZ) mice treated with doxycycline and
adipocytes from AdipoChaser mice kept on chow diet; none of these
stainings showed a β-gal signal (Fig. 1b,c). We next tested the washout
period of doxycycline in mice. We kept AdipoChaser mice on doxycy-
cline diet for 7 d and then switched them to chow diet without doxy-
cycline for time periods ranging from overnight to up to 3 d before
analyzing them. Real-time quantitative PCR (qPCR) results showed
that each group of AdipoChaser mice had a similar level of rtTA
expression in subcutaneous adipose tissue (Supplementary Fig. 1).
Cre expression levels in AdipoChaser mice after 3, 2 or 1 d or over-
night doxycycline withdrawal were comparable to the expression
levels in AdipoChaser mice kept on chow diet without doxycycline
and were much lower than those in AdipoChaser mice treated con-
tinuously in the presence of doxycycline (Supplementary Fig. 1).
b
eWAT
Control with dox AdipoChaser without dox AdipoChaser with dox
c
sWAT
Control with dox AdipoChaser without dox AdipoChaser with dox
a adnP rtTA
Doxycycline Rosa26 lacZcreTRE
Rosa26 Stop lacZ
loxP loxP
Figure 1 Inducible labeling of mature adipocytes. (a) The inducible
labeling system of mature adipocytes, produced by crossing adiponectinP-
rtTA (adnP-rtTA) transgenic mice with TRE-cre and Rosa26-loxP-stop-
loxP-lacZ transgenic mice. The triple transgenic mouse, called the
AdipoChaser mouse, expresses rtTA in mature adipocytes but does not
express LacZ in any cell type while maintained on food not containing
doxycycline (dox). When doxycycline is included in the food, adipocytes
that express rtTA will have the TRE promoter activated so that cre
expression is induced. The Cre protein will specifically cut out the floxed
transcriptional stop cassette and then turn on LacZ expression. Even
after withdrawal of doxycycline from the food, these adipocytes will
permanently express LacZ, whereas any new adipocytes that develop after
doxycycline exposure will not express LacZ. (b,c) Representative β-gal
(blue) staining of eWAT (b) and sWAT (c) in male control (mice with only
TRE-cre and Rosa26-loxP-stop-loxP-lacZ) or AdipoChaser mice. Solid
arrows (b,c), LacZ-positive cells; open arrows (b,c), LacZ-negative cells.
Scale bar (black, shown in b, applies to b and c), 200 µm; (blue, shown
in b, applies to the insets in b and c), 50 µm. Throughout the figure,
n = 2 male mice per group.
eWAT
Chow 3 d Chow 59 da
Chow 3 d + HFD 7 d Chow 3 d + HFD 35 d Chow 3 d + HFD 56 d Chow 3 d + HFD 89 d
eWAT
b
sWAT
Chow 3 d + HFD 7 d Chow 3 d + HFD 35 d Chow 3 d + HFD 56 d Chow 3 d + HFD 89 dd
sWAT
Chow 3 d Chow 59 dc
Normal
Hypertrophy Hypertrophy and hyperplasia
Hypertrophy
Prolonged
HFD
Prolonged
HFD
HFDeWAT
Normal
Hypertrophy
HFDsWAT
e
Figure 2 HFD-induced adipose tissue hypertrophy and hyperplasia.
(a–d) Representative β-gal staining of eWAT (a,b) and sWAT (c,d) from
9- to 10-week-old male AdipoChaser mice that were kept on doxycycline
diet for 7 d followed by chow diet for 3 or 59 d (a,c) or chow diet for 3 d
and HFD for 7, 35, 56 or 89 d (b,d). Solid arrows (b), LacZ-positive cells;
open arrows (b), LacZ-negative cells. Scale bar (shown in a, applies to
a–d), 200 µm. For a–d, n = 3 male mice per group. (e) Schematic model
of the depot-dependent contribution of hyperplasia to adipose tissue
expansion after HFD feeding. HFD-induced adipose tissue expansion
is contributed mainly by hypertrophy in both eWAT and sWAT at the
early stages. After prolonged HFD exposure (i.e., longer than 1 month),
a wave of adipogenesis is preferentially initiated in eWAT (hyperplasia),
but adipogenesis does not occur at measurable levels in sWAT.
Adipocytes surrounded by blue circles represent old LacZ-positive
cells, and adipocytes surrounded by white circles represent new
LacZ-negative cells.
np
g
©
20
13
N
at
ur
e A
m
er
ic
a,
In
c.
A
ll
rig
ht
s
re
se
rv
ed
.
t e c H n i c a L r e P O r t s
1340 VOLUME 19 | NUMBER 10 | OCTOBER 2013 nature medicine
During the same time course, AdipoChaser mice had similar levels of
LacZ expression throughout the doxycycline withdrawal periods that
were much higher than those in mice never treated with doxycycline
(Supplementary Fig. 1). These results show that an overnight with-
drawal is sufficient to completely wash out doxycycline in mice.
Using this system, we initially focused on the rate of appearance
of de novo–differentiated adipocytes during HFD challenge. We first
gave AdipoChaser mice doxycycline diet for 7 d to ensure uniform
and permanent labeling of mature adipocytes with LacZ, which we
followed with 3 d of chow diet to ensure that the doxycycline was
fully washed out. Thereafter, we fed the mice either chow or HFD
(60% of calories from fat) for various lengths of time. HFD feed-
ing for 12 weeks increased the body weight of AdipoChaser mice by
51% (P < 0.001) as compared to the basal body weight of the mice
before beginning the HFD, whereas AdipoChaser mice kept on chow
had only a 15% increase in body weight (P < 0.001) (Supplementary
Fig. 2a). The weights of epididymal adipose tissue and subcutaneous
adipose tissue of 12-week HFD-fed AdipoChaser mice were increased
by 84% (P < 0.001) and 163% (P < 0.001), respectively, compared to
age-matched AdipoChaser mice kept on chow diet (Supplementary
Fig. 2b,c). AdipoChaser mice fed with chow diet showed no new
adipogenesis, as determined by the fact that all fat depots displayed
nearly 100% positive β-gal staining in adipocytes, even at up to 59 d
after doxycycline treatment (Fig. 2). Thus, extremely low levels of
adipogenesis were present in both epididymal and subcutaneous
adipose tissues (Fig. 2a,c). After HFD feeding for 7 d, adipocytes in
both the epididymal and subcutaneous adipose tissues still showed
nearly 100% LacZ labeling with no obvious morphological changes
(Fig. 2b,d). When we kept mice on HFD for 35 d, the average size of
the adipocytes in both depots increased markedly, reflecting a high
capacity for cell hypertrophy (Fig. 2b,d). After 56 or 89 d of HFD
feeding, epididymal adipose tissue showed a high adipogenesis rate,
as determined by a large number of β-gal–negative cells (Fig. 2b);
in contrast, the subcutaneous adipose depot maintained a relatively
low rate of adipogenesis, as evidenced by nearly 100% LacZ labeling
(Fig. 2d). These observations indicate that HFD-induced adipose
tissue expansion is contributed mainly by hypertrophy during the
first month of HFD. After prolonged HFD exposure (i.e., longer than
1 month), a wave of adipogenesis is preferentially initiated in epidi-
dymal adipose tissue, whereas only negligible levels of adipogenesis
occur in subcutaneous adipose tissue depots (Fig. 2e). These experi-
ments reveal the surprising property of subcutaneous adipose tissue
to use de novo adipogenesis only minimally as a mechanism to cope
with chronic caloric excess.
Beige adipocytes arise by de novo adipogenesis
It is not clear whether the beige cells that are induced by cold exposure
or β3 agonist treatment arise through transdifferentiation of exist-
ing white adipocytes or by de novo adipogenesis from a subgroup
of precursor cells. We therefore studied newly developed beige adi-
pocytes within subcutaneous adipose tissue under those conditions.
We pretreated AdipoChaser mice with doxycycline diet for 7 d to
ensure uniform, permanent expression of LacZ in mature adipocytes,
which we followed with 3 d of chow diet to wash out the doxycy-
cline. Overnight cold exposure in doxycycline-pretreated mice on a
chow diet showed clusters of cells within the subcutaneous adipose
tissue that were smaller than normal white adipocytes and that dis-
played negative β-gal staining (Fig. 3a). Three days of cold exposure
induced massive browning of subcutaneous adipose tissue, with the
AdipoChaser mice showing large areas of beige fat cells with multi-
ple small lipid droplets, and most of these tissues had negative β-gal
a
b
d
e
f
g
c
CE overnight CE 3 d CE 3 d + RT 7 d
CE 3 d with dox
Beige
Beige Beige Beige
BeigeBeigeBeige
Beige Beige
White
White
White
White White
White
White
White
WhiteWhite
White White
Merge
Merge
Merge
Brightfield color
Brightfield color
Brightfield color
DAPI
DAPI
DAPI
Perilipin
Ucp1
Cited1
RT
6 °C
Mature white adipocyte
A subgroup of precursors
Beige
adipocytes
De n
ovo
differ
entia
tion
Dedifferentiation
Brightfield B&W
Brightfield B&W
Brightfield B&W
β3 7 d
Figure 3 Lineage of the brown-like adipocytes in subcutaneous adipose
tissue after cold exposure. (a) Representative β-gal staining of sWAT
from 10-week-old male AdipoChaser mice that were kept on doxycycline
diet for 7 d followed by chow diet for 3 d and then exposed to cold (CE)
overnight (left) or for 3 d (middle) or exposed to cold for 3 d
followed by 7 d in room temperature (RT; right). (b) Representative
β-gal staining of sWAT from 10-week-old male AdipoChaser mice that were
on doxycycline diet before and during 3 d of cold exposure as a positive
control group. (c) Representative β-gal staining of sWAT from 10-week-old
male AdipoChaser mice on doxycycline diet for 7 d followed by chow diet
for 3 d and then given 7 d of daily β3 agonist treatment. Solid arrows (a–c),
LacZ-positive cells; open arrows (a,c), LacZ-negative cells. Scale bar
(shown in a, applies to a–c), 100 µm. For a–c, n = 3 male mice per
group. (d–f) Immunofluorescence staining for perilipin (green) (d),
Ucp1 protein (green) (e) or Cited1 protein (green) (f) on slides prestained
with β-gal. The male AdipoChaser mice in d–f were pretreated with
doxycycline diet and exposed to cold for 3 d on a chow diet. Blue
indicates pre-existing white adipocytes. Scale bar (shown in d, applies to
d–f), 200 µm. The yellow dashed outlines indicate the borders between
white and beige adipocytes. For d–f, n = 2 male mice per group.
B&W, black and white. (g) Schematic model showing that the beige cell
population arises predominantly from de novo adipogenesis rather than
transdifferentiation. After cold exposure or β3 agonist treatment, most
beige adipocytes are induced by differentiating from cell populations other
than existing mature adipocytes (beige precursors) rather than through
dedifferentiation of mature white adipocytes.
np
g
©
20
13
N
at
ur
e A
m
er
ic
a,
In
c.
A
ll
rig
ht
s
re
se
rv
ed
.
t e c H n i c a L r e P O r t s
nature medicine VOLUME 19 | NUMBER 10 | OCTOBER 2013 1341
staining (Fig. 3a). These new adipocytes were preserved for at least 7 d
in the subcutaneous adipose tissue after we switched the mice back to
room temperature (Fig. 3a). In mice that we kept on doxycycline diet
throughout the cold exposure, ~100% of the beige adipocytes showed
LacZ-positive signals, demonstrating that these newly emerging cells
can indeed be labeled if the mouse remains exposed to doxycycline
during cold exposure (Fig. 3b). We observed similar results in the β3
agonist–treated mice: most of the newly generated beige adipocytes
in these mice were negative for β-gal staining (Fig. 3c).
Immunofluorescence staining demonstrated that LacZ-negative
cells that formed in the subcutaneous adipose tissue of AdipoChaser
mice were positive for the lipid droplet–specific marker perilipin, Ucp1
and the beige cell marker Cbp/p300-interacting transactivator with
Glu/Asp-rich C-terminal domain, 1 (Cited1)29, whereas the LacZ-
positive blue adipocytes were positive for perilipin only (Fig. 3d–f).
These results indicate that after cold exposure or β3 agonist treat-
ment, most beige adipocytes are induced by de novo differentiation
from cell populations other than existing mature adipocytes (i.e.,
Cited1+Cd137+ beige precursors) rather than by a transdifferentiating
mature white adipocyte (Fig. 3g). Beige cells can also revert to a white
adipocyte phenotype, as has been recently reported30 (Fig. 3g).
Cold-induced adipogenesis in epididymal adipose tissue
Previous studies have shown that the appearance of brown-like cells
is observed mainly in subcutaneous adipose tissue. Very few of these
newly emerging cells can be detected in the visceral fat20. Notably, after
cold exposure during as short a time period as overnight, epididymal
adipose tissue of mice pretreated with doxycycline on a chow diet
showed num