Mechanisms underlying the resistance to diet-induced
obesity in germ-free mice
Fredrik Ba ¨ckhed*†, Jill K. Manchester*, Clay F. Semenkovich‡, and Jeffrey I. Gordon*§
*Center for Genome Sciences and‡Department of Medicine, Washington University School of Medicine, St. Louis, MO 63108
Edited by John J. Mekalanos, Harvard Medical School, Boston, MA, and approved November 22, 2006 (received for review June 27, 2006)
The trillions of microbes that colonize our adult intestines function
collectively as a metabolic organ that communicates with, and
complements, our own human metabolic apparatus. Given the
worldwide epidemic in obesity, there is interest in how interac-
tions between human and microbial metabolomes may affect our
energy balance. Here we report that, in contrast to mice with a gut
microbiota, germ-free (GF) animals are protected against the obe-
sity that develops after consuming a Western-style, high-fat,
sugar-rich diet. Their persistently lean phenotype is associated
with increased skeletal muscle and liver levels of phosphorylated
AMP-activated protein kinase (AMPK) and its downstream targets
involved in fatty acid oxidation (acetylCoA carboxylase; carnitine-
fasting-induced adipose factor (Fiaf), a circulating lipoprotein
lipase inhibitor whose expression is normally selectively sup-
pressed in the gut epithelium by the microbiota, are not protected
from diet-induced obesity. Although GF Fiaf?/? animals exhibit
similar levels of phosphorylated AMPK as their wild-type litter-
mates in liver and gastrocnemius muscle, they have reduced
expression of genes encoding the peroxisomal proliferator-
activated receptor coactivator (Pgc-1?) and enzymes involved in
fatty acid oxidation. Thus, GF animals are protected from diet-
induced obesity by two complementary but independent mecha-
nisms that result in increased fatty acid metabolism: (i) elevated
levels of Fiaf, which induces Pgc-1?; and (ii) increased AMPK
activity. Together, these findings support the notion that the gut
microbiota can influence both sides of the energy balance equa-
tion, and underscore the importance of considering our metabo-
lome in a supraorganismal context.
GF knockoutmice lacking
AMP-activated protein kinase ? fasting-induced adipose factor ?
fatty acid metabolism ? gut microbiota ? symbiosis
intake over expenditure. The startling rise in the number of
people who are obese, together with the inability of most
individuals to comply with treatment regimens that require
sustained lifestyle changes, has stimulated efforts to identify new
therapeutic targets for the treatment and prevention of this
One potential target is our gut microbes. The distal human
intestine can be viewed as an anaerobic bioreactor containing
trillions of bacteria and archaea, programmed to perform met-
abolic functions that we have not been required to evolve on our
own, including the ability to harvest otherwise inaccessible
nutrients from our diet (1). By comparing germ-free (GF) and
conventionally raised (CONV-R) mice, we have shown that the
gut microbiota functions as an environmental factor that regu-
lates fat storage (2). Colonization of adult GF C57BL/6J mice
with a microbiota harvested from the distal intestine (cecum) of
CONV-R animals (a process known as conventionalization)
produces a significant increase in body-fat content, and relative
insulin resistance within 14 days despite reduced food intake (2).
This effect occurs in males and females belonging to several
inbred strains of mice (2). Mechanistic studies revealed that the
transplanted microbiota not only increases calorie harvest from
lthough obesity stems from the interactions of genetic and
environmental factors, its root cause is an excess of caloric
dietary plant polysaccharides with glycosidic linkages that the
host is ill-equipped to cleave with their own complement of
glycoside hydrolases, but also modulates host genes that affect
energy deposition in adipocytes. Colonization increases glucose
uptake in the small intestine (2) as well as fermentation of
carbohydrates to short-chain fatty acids (SCFAs) in the distal gut
(3). SCFAs are absorbed with subsequent stimulation of de novo
synthesis of triglycerides in the liver. In addition, the microbiota
suppresses expression of fasting-induced adipose factor (Fiaf,
also known as angiopoietin-like protein-4), a secreted lipopro-
tein lipase (LPL) inhibitor; this suppression is confined to the
intestinal epithelium and does not occur at other sites where Fiaf
is produced (liver and fat) (2). LPL functions in a number of cell
lineages as the rate-limiting step for uptake of triglyceride-
derived fatty acids (4, 5). By suppressing Fiaf, colonization
increases LPL activity in adipocytes and enhances storage of
liver-derived triglycerides (2). The physiologic importance of
Fiaf was established by studying GF Fiaf?/? and wild-type
littermates fed a standard low-fat polysaccharide-rich diet; GF
knockout mice have the same degree of adiposity as their
conventionalized counterparts, indicating that Fiaf is a key
modulator of the microbiota-induced increase in fat storage (2).
Although LPL is the rate-limiting enzyme for import and
subsequent storage of triglyceride-derived fatty acids in adipo-
cytes, genetically engineered mice that express LPL only in their
myocytes gain weight normally and have a normal body-mass
composition. Instead of importing triglycerides from the circu-
lation, they increase de novo fatty acid synthesis in adipose tissue
(6). This finding raises the question of whether the lean pheno-
type of GF mice involves mechanisms beyond a Fiaf-mediated
reduction in LPL activity.
AMP-activated protein kinase (AMPK) is a heterotrimeric
enzyme that is conserved from yeast to humans and functions as
a ‘‘fuel gauge’’ that monitors cellular energy status; it is activated
in response to metabolic stresses that result in an increased
intracellular ratio of AMP to ATP (e.g., exercise, hypoxia, and
glucose deprivation; ref. 7). Adipocyte-derived leptin (8) and
adiponectin (9), as well as an elevated NAD:NADH ratio (10),
also increase AMPK activity. Activation of AMPK occurs by
phosphorylation of Thr-172 in its catalytic ? subunit (11, 12),
F.B. contributed new reagents/analytic tools; F.B., J.K.M., C.F.S., and J.I.G. analyzed data;
and F.B. and J.I.G. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS direct submission.
Freely available online through the PNAS open access option.
Abbreviations: AMPK, AMP-activated protein kinase; GF, germ-free; CONV-R, convention-
ally raised; Fiaf, fasting-induced adipose factor; LPL, lipoprotein lipase; Acc, acetylCoA
carboxylase; Cpt1, carnitine:palmitoyl transferase-1; Pgc-1?, peroxisomal proliferator-
activated receptor coactivator 1?; qRT-PCR, quantitative RT-PCR.
§To whom correspondence should be addressed. E-mail: firstname.lastname@example.org.
This article contains supporting information online at www.pnas.org/cgi/content/full/
© 2007 by The National Academy of Sciences of the USA
www.pnas.org?cgi?doi?10.1073?pnas.0605374104 PNAS ?
January 16, 2007 ?
vol. 104 ?
no. 3 ?
leading to suppression of ATP-consuming anabolic pathways
and induction of ATP-generating catabolic pathways (7).
Here, we show that GF mice are protected against obesity
produced by consumption of a high-fat high-sugar Western diet.
The mechanism involves AMPK and Fiaf operating through
GF Mice Are Protected Against Diet-Induced Obesity. To determine
whether GF mice are protected against diet-induced obesity,
adult C57BL/6J males, maintained since weaning on an auto-
claved low-fat chow diet (5% lipids; caloric density, 4.1 kcal/d),
were conventionalized with an unfractionated cecal microbiota
from a CONV-R donor that had also been fed low-fat chow.
Three weeks later, half of the animals were switched to a
‘‘Western diet’’, where 41% of the calories are in the form of fat;
41% as readily digested carbohydrates including simple sugars
(sucrose); and 18% as protein (caloric density, 4.8 kcal/g). Chow
consumption and weight gain were recorded weekly. After 8
weeks, conventionalized animals on the Western diet had gained
significantly more weight than their GF counterparts (5.3 ? 0.8
vs. 2.1 ? 0.5 g; n ? 5 mice per group; P ? 0.05 according to
Student’s t test) (Fig. 1A). Weight gain in the GF group was not
significantly different from the weight gain observed in GF mice
that had been maintained on the standard low-fat polysaccha-
ride-rich diet (data not shown). Epididymal fat-pad weights were
also significantly greater in conventionalized mice fed the West-
ern diet (37 ? 5 vs. 22 ? 1 mg/g of body weight; P ? 0.05).
GF and conventionalized mice consumed similar amounts of
n ? 5 per group; P ? 0.24), and there were no statistically
significant differences in the energy content of their feces, as
defined by bomb calorimetry (3.76 ? 0.01 vs. 3.81 ? 0.09 kcal/g;
n ? 5 per group; P ? 0.53). Fatty acid absorption appeared to
be similar in the two groups; when GF and conventionalized
mice were given a single gavage of olive oil after a overnight fast,
serum triglycerides rose rapidly over a 2-hr period, reaching
equivalent levels in the two groups (Fig. 1B). However, although
triglycerides were subsequently cleared from the circulation in
conventionalized mice, they remained elevated in GF mice, a
phenomenon that can be attributed to their reduced LPL activity
(2). This decrease in LPL activity was also manifested by higher
fasting serum triglyceride levels in GF compared with conven-
tionalized mice on the Western diet (Table 1).
GF Mice Have Increased Levels of Phosphorylated AMPK in Muscle and
Liver. To investigate whether AMPK is involved in mediating the
resistance of GF mice to diet-induced obesity, we compared
levels of phosphorylated (active) AMPK in gastrocnemius mus-
cles harvested from GF and conventionalized animals on the
Western diet. Immunoblots disclosed that phospho-AMPK con-
centrations are 40% higher in GF animals (n ? 4 per group; P ?
0.05; Fig. 2 A and B). There were no significant differences in the
total level of immunoreactive AMPK ? subunit (Fig. 2 A and B).
Consistent with the elevations in phospho-AMPK, biochemical
assays disclosed 50% higher levels of AMP in the gastrocnemius
muscles of GF compared with conventionalized mice and no
differences in ADP or ATP concentrations (Table 2).
Phosphorylated AMPK stimulates fatty acid oxidation in
peripheral tissues by directly phosphorylating acetylCoA car-
boxylase (Acc; converts acetyl CoA to malonylCoA). Phosphor-
ylation of Acc inhibits its activity, leading to decreased malo-
nylCoA levels. Because malonylCoA inhibits carnitine:palmitoyl
transferase-1 (Cpt1), which catalyzes the rate-limiting step for
entry of long-chain fatty acylCoA into mitochondria, diminished
malonylCoA concentrations result in increased Cpt1 activity and
increased fatty acid oxidation (7).
We documented a 43% increase in the levels of phospho-Acc
in the gastrocnemius muscle of GF animals by using an immu-
noblot assay (P ? 0.01; Fig. 2 A and B) and a modest but
statistically significant 17% increase in Cpt1 activity, as defined
by a biochemical assay (Fig. 2C). In addition, we detected a 31 ?
4.5% increase in medium-chain acylCoA dehydrogenous (Mcad)
expression in GF gastrocnemius by quantitative RT-PCR (qRT-
PCR) (n ? 4 per group; P ? 0.05). Mcad is a mitochondrial
enzyme that catalyzes the initial step in ? oxidation of C8–C12
C57BL/6J mice were conventionalized 3 weeks before they were switched to a
Serum triglycerides levels were measured at the indicated time points (n ? 5 per
tionalized mice over a 3-day period and then again after they had been on a
0.05;**, P ? 0.01; and***, P ? 0.001 compared with GF.
GF mice are protected against diet-induced obesity. (A) Adult male
www.pnas.org?cgi?doi?10.1073?pnas.0605374104Ba ¨ckhed et al.
fatty acids. Together, these findings suggest that the presence of
a gut microbiota suppresses skeletal muscle fatty acid oxidation
through a metabolic pathway that may involve phosphorylation
We found a similar increase in phosphorylated AMPK in the
livers of these GF animals (Fig. 3 A and B). Foretz et al. (13) have
shown that short-term adenoviral-mediated overexpression of a
constitutively active form of AMPK in the livers of CONV-R
mice produces mild hypoglycemia and reduced hepatic glycogen
stores. GF mice fed a Western diet for 5 weeks also have
significantly reduced hepatic glycogen levels and decreased
glycogen synthase activity (Fig. 3 C and D). In addition, we
observed significantly reduced serum glucose and insulin levels
in GF compared with conventionalized mice on the Western diet
(Table 1). Glucose and insulin tolerance tests confirmed their
increased insulin sensitivity relative to their conventionalized
obese counterparts (data not shown).
Although phospho-AMPK is increased in the livers of GF
mice (see above), there were no significant differences in hepatic
AMP:ATP ratios between GF and conventionalized animals.
However, biochemical assays disclosed that GF mice had 72%
higher levels of NAD? (Table 2), which also activates AMPK
(10). Similar regulation of AMPK and its targets in both muscle
and liver is consistent with recent reports that metabolic cross-
talk exists among these distinct tissues (14, 15). Collectively,
these findings suggest that insulin-sensitive GF mice are pro-
tected against diet-induced obesity at least in part because of
increased AMPK activity and increased fatty acid oxidation in
their peripheral tissues.
GF Fiaf-Deficient Mice Have Lost Their Resistance to Diet-Induced
Obesity. The obesity-resistant phenotype of GF mice can also be
attributed to their increased intestinal expression of Fiaf (Fig.
4A; note that no differences in hepatic Fiaf expression were
observed between the groups). When GF wild-type and Fiaf?/?
mice were fed a Western diet by using the protocol described
above, Fiaf-deficient animals gained significantly more weight
than their wild-type littermates (6.2 ? 0.9 vs. 2.7 ? 1.0 g over a
5-week period; n ? 5 per group; P ? 0.05) (Fig. 4B) and had
significantly greater epididymal fat-pad weights (Fig. 4C).
the Western diet had higher serum levels of leptin and insulin
than their GF wild-type littermates (Table 3). Serum triglycer-
ides but not free fatty acids were significantly reduced in GF
Fiaf?/? mice on the Western diet (Table 3), consistent with the
fact they lack this circulating inhibitor of LPL.
Fiaf Regulates Pgc-1? Expression in Gastrocnemius Muscle. Fiaf
deficiency in GF animals fed a Western diet is associated with
statistically significant 24–46% decreases in the expression of
genes encoding key enzymes involved in fatty acid oxidation in
muscle (Cpt1 and medium-chain acylCoA dehydrogenase; see
no statistically significant differences in phospho-AMPK, total
AMPK, phospho-Acc, AMP, ADP, ATP, NAD?, or NADH
levels in the gastrocnemius muscles and livers of GF Fiaf
knockout compared with their wild-type littermates [n ? 4 mice
per group; supporting information (SI) Table 4; data not shown].
However, we did find a significant reduction in expression of the
peroxisomal proliferator activated receptor coactivator 1? (Pgc-
1?) in GF Fiaf?/? gastrocnemius muscle (24 ? 7% compared
with GF Fiaf?/? littermates; n ? 6 mice per group; P ? 0.05;
Fig. 4D). Pgc-1? is capable of coactivating nearly all known
nuclear receptors, as well as many other transcription factors; it
of mitochondrial fatty acid oxidation, including Cpt1 and me-
dium-chain acylCoA dehydrogenous (16).
Locomotor Activity. Using an implantable detector of locomotion
(see Materials and Methods), we found that the absence of a
Table 1. Biochemical and ELISA studies of sera obtained after a 4-hr fast from GF and conventionalized wild-type
C57Bl/6J mice fed low- vs. high-fat diets
Low-fat diet Western diet
(n ? 5)
(n ? 5)
(n ? 8)
(n ? 8)
Free fatty acids, mM
5.46 ? 0.16
0.39 ? 0.03
1.65 ? 0.14
49 ? 4
110 ? 3
1.09 ? 0.12
7.28 ? 0.12
0.85 ? 0.08
2.70 ? 0.02
39 ? 5
103 ? 18
1.01 ? 0.10
10.4 ? 1.6
0.43 ? 0.07
2.01 ? 0.32
68.5 ? 3.8
173.5 ? 5.2
0.87 ? 0.08
13.9 ? 1.2
0.75 ? 0.07
5.89 ? 0.60
50.4 ? 4.1
185.1 ? 7.9
1.07 ? 0.07
Mean values ? SE are shown.
muscle of mice consuming a Western diet. (A) Immunoblotting of protein
lysates from gastrocnemius muscle harvested from 15-week-old male GF or
conventionalized C57BL/6J mice fed a Western diet for 5 weeks before death.
(C) Effects of the gut microbiota on Cpt activity in freeze-clamped gastrocne-
0.05 compared with GF; and**, P ? 0.01.
The gut microbiota suppresses AMPK activity in the gastrocnemius
Ba ¨ckhed et al. PNAS ?
January 16, 2007 ?
vol. 104 ?
no. 3 ?
microbiota is associated with significantly increased movement
in wild-type mice, whether these age- and gender-matched
animals were on a standard chow or a Western diet (n ? 4 per
group; Fig. 1C). Moreover, qRT-PCR assays revealed no statis-
tically significant differences in the levels of expression of
uncoupling protein 1 in the gastrocnemius muscles of these
animals (n ? 4 mice surveyed per treatment group; data not
shown). Finally, a comparison of GF Fiaf?/? and Fiaf?/?
littermates on the standard chow diet revealed no statistically
significant differences in their locomotor activity, despite sig-
nificant differences in their adiposity. The mechanisms under-
lying the increased locomotor activity of wild-type GF vs.
conventionalized mice are unknown, may reflect a heretofore
unappreciated link between the metabolic activity of their
microbiota, and behaviors which could contribute to the ob-
served differences in their adiposity. Nonetheless, we cannot
attribute the increased adiposity of GF Fiaf?/? vs. wild-type
littermates to this phenomenon.
Collectively, our data indicate that the gut microbiota is able to
modulate energy balance through a number of intertwined
pathways. A shift in gut microbial ecology occurs in genetically
obese (ob/ob) mice consuming a standard chow diet: compared
with their lean ?/? and ob/? littermates, the representation of
the Bacteroidetes diminishes by ?50%, and the Firmicutes
increase to a corresponding degree. Remarkably, these changes
are division-wide and not due to a suppression of one or a few
Bacteriodetes lineages or to bloom in one or a few members of
the Firmicutes (17). A similar shift in the ratio of Bacteroidetes
to Firmicutes occurs in obese compared to lean humans; more-
over, as humans lose weight, there is a division-wide increase in
the proportion of Bacteroidetes and reduction in Firmicutes
(18). Comparative metagenomic studies of the distal gut micro-
bial communities of ob/ob mice and their lean littermates fed a
standard low-fat rodent chow diet indicate that the ob/ob com-
munity is enriched for genes that are able to harvest calories
from complex plant-derived polysaccharides (19). Moreover,
transplantation of the gut microbiota from ob/ob donors to adult
GF ?/? recipients consuming a standard chow diet low in fat
and rich in polysaccharides results in a greater increase in
adiposity in the recipients over a 2-week period than does a
transplantation of a microbiota from lean ?/? donors (19).
Metagenomic studies of the gut microbial community of mice
with obesity due to consumption of a high-fat simple-sugar diet,
and microbiota transplant experiments analogous to those de-
scribed above, are needed to identify the organismal and gene
lineages present in their gut community and to characterize its
Combined with the observations described here, these find-
ings support an emerging view that gut microbes can affect both
sides of the energy balance equation, as a factor that influences
the harvest of energy from components of the diet, and as a
factor that affects host genes that regulate how energy is
expended and stored. The microbiota provides glycoside hydro-
lases and polysaccharide lyases required to cleave glycosidic
linkages in plant glycans (1). The resulting monosaccharides are
absorbed or metabolized to short-chain fatty acids, which are
delivered to the liver and converted to triacylglycerols; these de
novo synthesized lipids are then deposited in adipocytes through
a process that involves, in part, microbial suppression of intes-
blotting of protein lysates from liver samples obtained from 15-week-old GF
or conventionalized male mice fed a Western diet for 5 weeks before they
were killed. Representative results from two mice per group are shown. (B)
relative to actin. Effects of the gut microbiota on glycogen levels (C) and
glycogen synthase activity (D) in freeze-clamped livers (n ? 5 per group).**,
P ? 0.01 compared with GF; and***, P ? 0.001.
The gut microbiota suppresses AMPK activity in liver. (A) Immuno-
Table 2. Biochemical assays of various metabolites in gastrocnemius muscle and liver harvested from GF and
conventionalized wild-type C57Bl/6J mice fed a Western diet
P value GF Conventionalized
6.14 ? 0.39
6.36 ? 0.89
26.78 ? 1.79
1.99 ? 0.06
0.06 ? 0.02
4.09 ? 0.42
7.19 ? 0.18
28.56 ? 3.29
1.88 ? 0.12
0.09 ? 0.03
20.08 ? 0.85
4.25 ? 0.46
3.41 ? 0.73
1.98 ? 0.18
0.29 ? 0.03
20.87 ? 1.60
5.19 ? 0.52
4.13 ? 1.04
1.15 ? 0.21
0.34 ? 0.06
Mean values ? SE are shown (n ? 5).
www.pnas.org?cgi?doi?10.1073?pnas.0605374104Ba ¨ckhed et al.
tinal epithelial production of the circulating LPL inhibitor, Fiaf.
Comparisons of GF and colonized mice on a fat- and sugar-rich
diet indicate that the microbiota can also affect adiposity by
producing a physiologic state where AMPK activity is reduced
in muscle, leading to reduced phosphorylation of its downstream
target, Acc; increases in malonylCoA production; greater inhi-
bition of Cpt1; and diminished mitochondrial fatty acid oxida-
tion. Reduced fatty acid oxidation is also mediated through
suppression of Pgc-1? in skeletal muscle (20). Interestingly, we
(phosphorylation) in muscle (and liver) is independent of Fiaf
signaling. The mechanism by which Fiaf induces Pgc-1? expres-
sion is still unclear. Unlike other angiopoietins, which are known
to be potent stimulators of the Tie 2 receptor (21), a receptor for
Fiaf (angiopoietin-like protein 4) that could initiate intracellular
signaling has yet to be identified. In addition, the effects of the
microbiota on bile acid metabolism and on the bioavailability of
dietary lipids need to be further evaluated using methods such
as mass spectrometry and NMR.
We have found that CONV-R C57BL/6J transgenic mice,
where overexpression of Fiaf is achieved by using enterocyte-
specific transcriptional regulatory elements that are not affected
fatty acid-binding protein gene; Fabpi; ref. 22), exhibit statisti-
cally significant reductions in their adiposity compared with
nontransgenic littermates [8.3 ? 0.3% vs. 10.6 ? 0.4% total body
fat as defined by dual energy x-ray absorptiometry; n ? 6–8
animals fed a standard chow diet; P ? 0.01 (F.B. and J.I.G.,
unpublished observations)]. In addition, CONV-R aP2-Fiaf
transgenic mice with engineered forced expression of Fiaf in
adipocytes exhibit significant reduction of adipose tissue weight
by stimulating fatty acid oxidation and uncoupling (23).
Our studies in gnotobiotic mice have also identified microbial
determinants of Fiaf expression. When GF mice are colonized
with Bacteroides thetaiotaomicron, a prominent saccharolytic
member of the normal human colonic microbiota, together with
the dominant human colonic methanogen, Methanobrevibacter
smithii, the efficiency of polysaccharide fermentation is mark-
edly improved (24). De novo lipogenesis is augmented, and host
adiposity is increased compared with animals colonized with
with significantly increased suppression of intestinal Fiaf expres-
sion (81 ? 3% in cocolonized compared with GF mice vs. 65 ?
1% and 48 ? 5% in mice colonized with either B. thetaiotaomi-
cron or M. smithii alone, respectively; n ? 5 per group; P ? 0.05;
B. S. Samuel and J.I.G, unpublished observation).
Collectively, these findings suggest that the gut microbiota
contributes to mammalian adiposity by regulating more than one
node within the metabolic network that controls bioenergetics.
Manipulating microbial characteristics in ways that impact cal-
control of Pgc-1?, may represent new strategies for modifying
host energy balance to promote health.
Materials and Methods
Animals. GF wild-type C57BL/6J animals were maintained in
gnotobiotic isolators under a strict 12-h light cycle (lights on at
0600 h) and fed an autoclaved low-fat polysaccharide-rich chow
diet (B & K Universal, East Yorkshire, U.K.) ad libitum Fiaf?/?
mice on a mixed C57BL/J:129/Sv background were backcrossed
have lower expression of Pgc-1? and genes involved in fatty acid oxidation in
small intestines and livers of GF and conventionalized wild-type male mice
maintained on a low-fat diet since weaning or given a high-fat Western diet
for 8 weeks before being killed. Mean values ? SE are plotted. n ? 5 mice per
group.**, P ? 0.01, compared with GF mice on the chow diet; ?, P ? 0.05
mice were switched to the Western diet and their body weights monitored
weekly for 5 weeks (n ? 5 per group). (C) Epididymal fat-pad weights of the
mice shown in B after 5 weeks on the Western diet. (D) qRT-PCR assays of
gastrocnemius muscle RNAs prepared from GF Fiaf?/? mice and wild-type
littermates on the Western diet (n ? 6 per group). Mean values ? SE are
plotted.*, P ? 0.05 compared with wild-type animals;**, P ? 0.01.
GF Fiaf?/? mice are not protected against diet-induced obesity and
Table 3. Biochemical and ELISA studies of sera, obtained after
a 4-hr fast, from 15-week-old GF and conventionalized Fiaf?/?
mice and their wild-type littermates maintained on a
(n ? 8)
(n ? 6)
Free fatty acids
6.4 ? 1.6
0.39 ? 0.05
2.90 ? 0.51
84.5 ? 16.3
184.4 ? 8.4
0.95 ? 0.04
6.8 ? 0.8
0.54 ? 0.04
4.93 ? 0.77
54.7 ? 3.6
137.8 ? 9.2
0.99 ? 0.07
Mean values ? SE are shown.
Ba ¨ckhed et al. PNAS ?
January 16, 2007 ?
vol. 104 ?
no. 3 ?
one generation to C57BL/6J animals and rederived as GF, as
described (2). Wild-type and Fiaf-deficient littermates were used
in these studies.
GF mice were colonized at 6–10 weeks of age with cecal
contents harvested from an adult CONV-R mouse and kept in
their gnotobiotic isolators. Mice were either switched to an
irradiated Western diet (TD96132; Harlan Teklad, Madison,
WI) 2–3 weeks after conventionalization or maintained on their
autoclaved low-fat chow diet. Body weight and food consump-
tion were monitored weekly. Only male animals were used in this
study, which was performed by using protocols approved by the
Washington University Animal Studies Committee.
Locomotor Activity Measurements. Mice were anesthetized before
a transmitter (minimitter PDT-4000; Mini Mitter, Bend, OR)
was implanted intraabdominally. Mice were allowed to recover
for 7 days after implantation, and locomotor activity data were
emitted by the transmitter was detected by receivers positioned
underneath the plastic gnotobiotic isolators. Data were then
converted into activity counts by VitalView software (Mini
Mitter). Mice were subsequently switched to a Western diet, and
Isolation and Initial Processing of Tissues.Afteranimalswerekilled,
their small intestine was removed and divided into 16 equal-sized
segments. Segments 13–14, liver and gastrocnemius muscle from
each animal were snap-frozen, and total RNA was isolated
[Qiagen (Valencia, CA) RNeasy kit] for real-time qRT-PCR
Additional gastrocnemius and liver-tissue samples (?50 mg
each) were directly placed in 1 ml of lysis buffer [20 mM Tris (pH
7.5)/150 mM NaCl/1 mM EDTA/1% Triton X-100] containing
Complete protease inhibitor mixture (Roche Diagnostics, Indi-
anapolis, IN) and phosphatase inhibitor mixture 1 (Sigma, St.
Louis, MO). After homogenization at 4°C, extracts were cen-
trifuged for 10 min at 4°C at 15,000 ? g to remove insoluble
debris, and the protein concentration in the resulting superna-
tant fraction was determined (DC protein assay; Bio-Rad,
Hercules, CA). Tissue samples used for assaying enzyme activ-
ities and metabolites were harvested after freeze clamping (2)
and processed as described below.
Immunoblotting. Soluble proteins from liver and gastrocnemius
muscle were separated on 10% Bis-Tris gels (Invitrogen, Carls-
bad, CA) and transferred to PVDF membranes. Membranes
were placed in 5% BSA/0.1% Tween-20/PBS for 60 min at room
temperature and then incubated overnight at 4°C in 1% BSA/
0.1% Tween/PBS together with one of the following antibodies:
rabbit anti-phospho-Acc, rabbit anti-phospho-AMPK, and rab-
Beverly, MA; final dilution, 1:1,000), or rabbit antiactin (Sigma;
qRT-PCR. RNA prepared from each tissue sample was reverse-
transcribed by using SuperScript II (Invitrogen) and a dT15
primer (Roche Diagnostics), as described in ref. 2. qRT-PCR
assays were performed in 25-?l reactions containing gene-
specific primers (900 nM; for a list, see SI Table 5) and SYBR
green (Abgene, Epsom, U.K.). Data were normalized to L32
mRNA (??CTanalysis) (see SI Text).
Assays of Sera. Standard biochemical methods were used to assay
sera for glucose, cholesterol, triglycerides, and nonesterified
fatty acids (2). Insulin and leptin levels were determined by
ELISA (Crystal Chemical, Downers Grove, IL). Glucose and
insulin tolerance tests were performed as described in ref. 2.
Statistical Analysis.Data were analyzed by using Student’s t test or
ANOVA with Tukey’s post hoc analysis.
We thank Maria Karlsson, David O’Donnell, and Sabrina Wagoner for
superb technical assistance; Tim Nagy for performing bomb calorimetry
(P30DK56336); and Peter Crawford for helpful suggestions. This work
was supported in part by National Institutes of Health Grants DK70977
and P30 DK56341. F.B. is the recipient of a postdoctoral fellowship from
the Wenner-Gren Foundation.
1. Ba ¨ckhed F, Ley RE, Sonnenburg JL, Peterson DA, Gordon JI (2005) Science
2. Ba ¨ckhed F, Ding H, Wang T, Hooper LV, Koh GY, Nagy A, Semenkovich CF,
Gordon JI (2004) Proc Natl Acad Sci USA 101:15718–15723.
3. Wolin M (1981) Science 213:1463–1468.
4. Merkel M, Eckel RH, Goldberg IJ (2002) J Lipid Res 43:1997–2006.
5. Preiss-Landl K, Zimmermann R, Hammerle G, Zechner R (2002) Curr Opin
R, Breslow JL (1997) Proc Natl Acad Sci USA 94:10261–10266.
7. Kahn BB, Alquier T, Carling D, Hardie DG (2005) Cell Metab 1:15–25.
8. Minokoshi Y, Kim YB, Peroni OD, Fryer LG, Muller C, Carling D, Kahn BB
(2002) Nature 415:339–343.
9. Yamauchi T, Kamon J, Minokoshi Y, Ito Y, Waki H, Uchida S, Yamashita S,
Noda M, Kita S, Ueki K, et al. (2002) Nat Med 8:1288–1295.
10. Rafaeloff-Phail R, Ding L, Conner L, Yeh WK, McClure D, Guo H, Emerson
K, Brooks H (2004) J Biol Chem 279:52934–52939.
11. Crute BE, Seefeld K, Gamble J, Kemp BE, Witters LA (1998) J Biol Chem
12. Stein SC, Woods A, Jones NA, Davison MD, Carling D (2000) Biochem J
13. Foretz M, Ancellin N, Andreelli F, Saintillan Y, Grondin P, Kahn A, Thorens
B, Vaulont S, Viollet B (2005) Diabetes 54:1331–1339.
14. Ranalletta M, Jiang H, Li J, Tsao TS, Stenbit AE, Yokoyama M, Katz EB,
Charron MJ (2005) Diabetes 54:935–943.
15. Lee CH, Olson P, Hevener A, Mehl I, Chong LW, Olefsky JM, Gonzalez FJ,
Ham J, Kang H, Peters JM, Evans RM (2006) Proc Natl Acad Sci USA
16. Vega RB, Huss JM, Kelly DP (2000) Mol Cell Biol 20:1868–1876.
17. Ley RE, Ba ¨ckhed F, Turnbaugh P, Lozupone CA, Knight RD, Gordon JI
(2005) Proc Natl Acad Sci USA 102:11070–11075.
18. Ley RE, Turnbaugh PJ, Klein S, Gordon JI (2006) Nature 444:1022–1023.
19. Turnbaugh PJ, Ley RE, Mahowald MA, Magrini V, Mardis ER, Gordon JI
(2006) Nature 444:1027–1031.
20. Puigserver P, Spiegelman BM (2003) Endocr Rev 24:78–90.
J, Radziejewski C, Maisonpierre PC, Yancopoulos GD (1996) Cell 87:1161–1169.
22. Sweetser DA, Hauft SM, Hoppe PC, Birkenmeier EH, Gordon JI (1988) Proc
Natl Acad Sci USA 85:9611–9615.
23. Mandard S, Zandbergen F, van Straten E, Wahli W, Kuipers F, Muller M,
Kersten S (2006) J Biol Chem 281:934–944.
24. Samuel BS, Gordon JI (2006) Proc Natl Acad Sci USA 103:10011–10016.
www.pnas.org?cgi?doi?10.1073?pnas.0605374104 Ba ¨ckhed et al.