A PGC-1a Isoform Induced

by Resistance Training Regulates

Skeletal Muscle Hypertrophy
Jorge L. Ruas,1,5,7,* James P. White,1,5 Rajesh R. Rao,1 Sandra Kleiner,1,6 Kevin T. Brannan,1,6 Brooke C. Harrison,2,6
Nicholas P. Greene,3 Jun Wu,1 Jennifer L. Estall,1 Brian A. Irving,4 Ian R. Lanza,4 Kyle A. Rasbach,1 Mitsuharu Okutsu,3
K. Sreekumaran Nair,4 Zhen Yan,3 Leslie A. Leinwand,2 and Bruce M. Spiegelman1,*
1Department of Cell Biology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA 02115, USA
2Department of Molecular, Cellular, and Developmental Biology, University of Colorado at Boulder, Boulder, CO 80309, USA
3Cardiovascular Medicine, Department of Medicine, University of Virginia, Charlottesville, VA 22908, USA
4Endocrine Research Unit, Division of Endocrinology, The Mayo Clinic, 200 1st Street SW, Joseph 5-194, Rochester, MN 55905, USA
5These authors contributed equally to this work
6These authors contributed equally to this work
7Present address: Department of Physiology and Pharmacology, Karolinska Institutet, S-171 77 Stockholm, Sweden

*Correspondence: jorge.ruas@ki.se (J.L.R.), bruce_spiegelman@dfci.harvard.edu (B.M.S.)

http://dx.doi.org/10.1016/j.cell.2012.10.050

SUMMARY endurance training, including mitochondrial biogenesis, fiber-

PGC-1a is a transcriptional coactivator induced
by exercise that gives muscle many of the best
known adaptations to endurance-type exercise but
has no effects on muscle strength or hypertrophy.
We have identified a form of PGC-1a (PGC-1a4)
that results from alternative promoter usage and
splicing of the primary transcript. PGC-1a4 is highly
expressed in exercised muscle but does not regulate
most known PGC-1a targets such as the mitochon-
drial OXPHOS genes. Rather, it specifically induces
IGF1 and represses myostatin, and expression of
PGC-1a4 in vitro and in vivo induces robust skeletal
muscle hypertrophy. Importantly, mice with skeletal
muscle-specific transgenic expression of PGC-1a4
show increased muscle mass and strength and
dramatic resistance to the muscle wasting of cancer
cachexia. Expression of PGC-1a4 is preferentially
induced in mouse and human muscle during resis-
tance exercise. These studies identify a PGC-1a pro-
tein that regulates and coordinates factors involved
in skeletal muscle hypertrophy.
INTRODUCTION
PGC-1a is a transcriptional coactivator that controls the expres-
sion of genes involved in oxidative metabolism. PGC-1a was
originally identified as a coactivator of PPARg in brown adipose
tissue, but it is enriched in many tissues that are active in oxida-
tive metabolism, such as heart, skeletal muscle, and the fasted
liver. Muscle PGC-1a is induced by exercise in both mice and
humans (Short et al., 2003). When expressed in skeletal muscle
in vivo, PGC-1a causes many of the changes associated with
type switching, stimulation of fatty acid oxidation, angiogenesis,
and resistance to muscle atrophy (Arany, 2008). This reprogram-
ming of muscle results in increased muscle endurance. Genetic
loss of PGC-1a in murine muscle causes glucose intolerance,
especially on a high-fat diet (Choi et al., 2008; Handschin et al.,
2007). Elevated PGC-1a in muscle does not protect against
insulin resistance stimulated by a high-fat diet in young mice
(Choi et al., 2008) but dramatically protects against the sarcope-
nia, obesity, and diabetes that accompany aging (Wenz et al.,
2009). Although PGC-1a is induced in exercise and has remark-
able effects on muscle endurance, it has no clear-cut effects
on either muscle size or strength. Here, we identify a transcript
from the Pgc-1a gene that is expressed abundantly in skeletal
muscle, particularly in the setting of resistance training in mice
and humans. This protein, termed PGC-1a4, does not regulate
the same set of genes induced by PGC-1a but rather regulates
the IGF1 and myostatin pathways, both of which are known
regulators of muscle size and strength (Florini, 1987; McPherron
et al., 1997). This results in muscle hypertrophy and increased
strength. Remarkably, mice with skeletal-muscle-specific trans-
genic expression of PGC-1a4 show a dramatic resistance to the
muscle wasting of cancer cachexia. Taken together, these data
suggest that PGC-1a4 integrates resistance training with a gene
program of muscle hypertrophy, which results in several impor-
tant health benefits.
RESULTS
Identification, Characterization, and Expression of
Alternative Isoforms of PGC-1a
We and others have shown that transcription of the Pgc-1a gene
can be driven by two distinct promoter regions (Chinsomboon
et al., 2009; Yoshioka et al., 2009). One is located immediately
50 of the known exon 1 (proximal promoter), and another (alterna-
tive promoter) is located 13 kb upstream (Figure 1A). By using

Cell 151, 1319–1331, December 7, 2012 ª2012 Elsevier Inc. 1319

 A

C D

Figure 1. Cloning and Characterization of PGC-1a Isoforms

(A) Schematic representation of the conservation between human and mouse PGC-1a gene (http://www.dcode.org). Two promoters that can drive expres-
sion of the PGC-1a gene. Ex indicates exons seen in the depicted region. Structure of the different PGC-1a isoform mRNA is shown. / indicates partial
conservation. * indicates stop codon.
(B) PGC-1a protein domain conservation. Amino acid numbers refer to mouse PGC-1a (hereafter PGC-1a1). Numbers in brackets indicate the number of amino
acids for each isoform. Red boxes indicate new N- and C-terminal amino acid sequences.
(C) Three different exon1 coding sequences result in different N-terminal amino acid sequences. PGC-1a2 and a4 share the same alternative exon1 and,
therefore, share the same first 12 amino acids. All isoforms share exon 2.
(D) Differential promoter usage and splicing options result in proteins with different molecular weights. The different PGC-1a isoforms were expressed in HEK293
cells. Whole-cell extracts were resolved by SDS PAGE followed by immunoblotting by using an anti-PGC-1a antibody (Zhang et al., 2009) that we have found to
recognize all isoforms described here.
See also Figure S1.

1320 Cell 151, 1319–1331, December 7, 2012 ª2012 Elsevier Inc.

 Figure 2. Gene Expression Profiling of
Quantitative mRNA Expression

A Total
18 PGC-1a Isoforms and Their Target Genes
PGC-1α1
16 (A) Tissue-specific PGC-1a isoform expression
PGC-1α2
14 patterns. Absolute quantification of gene expres-
PGC-1α3
12 sion in mouse tissues (n = 6) by quantitative real-
10 PGC-1α4
time PCR using isoform-specific primers.
8
(B) Heatmap summary of relative changes in gene
6
4 expression by each PGC-1a isoform. Gene ex-
2 pression was analyzed (Affymetrix) in myotubes
0 expressing GFP alone (control) or together with
Sk. Musc Heart Kidney Liver Brain BAT each PGC-1a isoform. Experiments were per-
formed in triplicate, and results were analyzed with
B C dChip software.
(C) Venn diagram represents the number of genes
PGC-1α1
PG 1α1

PG 1α2

PG 1α3

regulated by PGC-1a1, PGC-1a4, and in common

1α

PGC-1α4
C-

C-

C-

C-

between both isoforms.

FP
PG
G

(D–F) From the Affymetrix results, gene sets were

validated by quantitative real-time PCR using
specific primers. RNA was prepared as described
in (B). Bars depict mean values, and error bars
represent SD. *p < 0.05 between indicated group
D and control. #p < 0.05 between all groups.
Relative mRNA Expression

10 GFP
* See also Figure S2.
PGC-1 1
8
PGC-1 4
6 *

4 *
*,# * * * * 41.9 and 41.0 kDa, respectively. PGC-1a4
2 * *,# * *,#
shares the same alternative exon1 with
0 PGC-1a2 (and therefore the same N
terminus, Figure 1C), but its mRNA struc-
a

C 4

D
Vb

PD F
1b
C

Fb
t4
R

G
A
yt

lu

PT
ER

PD

G
ox

VE
C
C

ture is different in that it contains a 31

C

nucleotide insertion between exons 6

E F
GFP and 7, which generates a premature stop
Relative mRNA Expression

Relative mRNA Expression

6 * GFP 2.0 PGC-1 1 codon (Figures 1A, 1B, and S1C).

PGC-1 1 PGC-1 4
5 * PGC1a4 is predicted to encode 266
PGC-1 4 1.5 *
4 amino acids, a protein of 29.1 kDa. The
3 * 1.0 different alternative first exons generate
* *,#
* different N termini that seem to signifi-
2 *,#
* 0.5 cantly affect protein accumulation (Fig-
1 *
** ures 1C, 1D, and S1B). With the use of
0 0.0
specific qPCR probes, we determined
P1

P2

P3

P4

P5

P6
F1

F2

1B

2A

2B
N

TN
IG

IG

FB

FB

FB

FB

FB

FB

ST

total PGC-1a mRNA content (targeting

VR

VR

VR

FS
IG

IG

IG

IG

IG

IG

AC

AC

AC

exon 2, which is present in all forms) or

the expression of each individual isoform
in an array of mouse tissues (Figure 2A).
a targeted PCR strategy (Figure S1A available online), we cloned In fed liver, PGC-1a1 levels approach total PGC-1a content,
and characterized the transcripts that result from the use of and no significant alternative isoform expression is observed
either promoter in mouse skeletal muscle. We cloned four (Figure 2A). However, in skeletal muscle, heart, and brown
full-length messenger RNAs (mRNAs); one of these corresponds adipose tissue (BAT), all isoforms are expressed at comparable
to the previously described PGC-1a (hereafter referred to levels and represent a significant part of total PGC-1a mRNA.
as PGC-1a1) and originates exclusively from the proximal Figures S2A and S2B show PGC-1a isoform protein levels in
promoter. The other three transcripts (hereafter referred to as superficial white quadriceps (glycolytic, low PGC-1a1 levels)
PGC-1a2, a3, and a4) result from alternative promoter usage and tibialis anterior (oxidative, high PGC-1a1 levels) muscle, as
and alternative splicing of the Pgc-1a gene (Figures 1A and well as liver, kidney, and BAT. Primary myotubes treated with
S1C). The new isoforms encode significantly different proteins the known PGC-1a inducer forskolin show high levels of all iso-
(Figure 1B). PGC-1a2 and a3 have distinct first exons but have forms (Figure S2C). Similarly, cold exposure induces expression
the same remaining exon/intron structure (Figure S1C); this of all PGC-1a variants in BAT (Figures S2D and S2E). These
results in two new proteins with overall similar domain structure results show that total PGC-1a mRNA content includes signifi-
but discrete N termini (Figure 1C). PGC-1a2 and a3 are 379 and cant levels of each isoform, which can be increased in vitro
370 amino acids long and have a predicted molecular weight of and in vivo by known inducers of the Pgc-1a gene.

Cell 151, 1319–1331, December 7, 2012 ª2012 Elsevier Inc. 1321

 A B

C D

Figure 3. Myotubes Expressing PGC-1a4 Show Cellular Hypertrophy

(A) Fluorescence microscopy analysis of myotubes expressing GFP alone or together with PGC-1a1 or PGC-1a4. Fully differentiated myotubes were transduced
with the different adenovirus and were observed under a fluorescence microscope 36 hr later with a 103 objective.
(B) Protein accumulation normalized by genomic DNA content in myotubes expressing GFP control, PGC-1a1, or PGC-1a4. Experiments were performed as
described in (A), but cells were processed for either total protein or DNA quantification. Graphs show the total protein/genomic DNA ratio.

1322 Cell 151, 1319–1331, December 7, 2012 ª2012 Elsevier Inc.

PGC-1a4 Regulates a Discrete Gene Program in Primary
Myotubes
Differentiated primary myotubes were transduced with adeno-
virus expressing different PGC1a isoforms. Figure 2B shows
a heat map generated by comparing the gene expression profile
of cells receiving each PGC-1a isoform, compared to green fluo-
rescent protein (GFP) alone. Interestingly, PGC-1a1 and PGC-
1a4 drive many changes in gene expression that are distinct
from each other; only 98 genes were coregulated by both
PGC-1a1 and PGC-1a4 (Figure 2C). PGC-1a2 and 3 seem to
affect the expression of only a very small set of genes (110 and
69 gene IDs, respectively). The functions of PGC-1a2 and a3
remain under investigation. Importantly, expression of PGC-
1a4 in myotubes did not affect the regulation of many classic
PGC-1a1 targets, including CytC (cytochrome C), CoxVb (cyto-
chrome c oxidase subunit Vb), Glut4 (glucose transporter type 4),
CPT1 (carnitine palmitoyltransferase-I), MCAD (medium chain
acyl CoA dehydrogenase), and PDGFb (platelet-derived growth
factor B) (Figure 2D). Several other known PGC-1a target genes
were induced by PGC-1a4 expression, though to a much
lesser extent than upon expression of PGC-1a1 (Figure 2D),
including ERRa, PDK4 (pyruvate dehydrogenase kinase, iso-
enzyme 4), and VEGFa (vascular endothelial growth factor A).
These results strongly suggest distinct functions for PGC-1a1
and PGC1a4.
Expression of PGC-1a4 Specifically Induces IGF1 and
Represses Myostatin Gene Expression
Pathway analysis of the microarray data identified cell mor-
phology, growth, and proliferation and IGF1 signaling as the
top pathways predicted to be under PGC-1a4 regulation (data
not shown). From quantitative real-time PCR, we confirmed
that PGC-1a4 (but not a1) specifically induces expression of
IGF1 (3.7-fold) while minimally affecting IGF2 (1.5-fold) levels
(Figure 2E). The expression levels of some members of the
IGF-binding protein (IGFBP) family were also selectively affected
by PGC-1a4 expression. IGF1 is among the best-known activa-
tors of skeletal muscle hypertrophy (Adams, 2002). PGC-1a4
expression also reduced mRNA levels of myostatin, a powerful,
negative regulator of muscle size in rodents and humans (Fig-
ure 2F) (Lee, 2004; McPherron et al., 1997), as well as the tran-
script levels of its receptors ACVRIIa and ACVRIIb (40% and
30%, respectively). The levels of ACVRIb remained unaffected
by expression of either PGC-1a1 or PGC1a4, whereas both iso-
forms repress follistatin expression (Figure 2F). Taken together,
these results indicate that PGC-1a4 controls the expression
of genes in two important pathways for regulating skeletal
muscle size.
PGC-1a4 Expression Leads to Powerful Myotube
Hypertrophy
Myotubes expressing PGC-1a4 appear significantly larger than
those expressing GFP control or PGC-1a1 (Figure 3A), with
a 2-fold elevation in the ratio of total protein to genomic DNA
(Figure 3B). We observed no significant differences in fusion of
myoblasts expressing GFP or the different PGC-1a isoforms,
as assessed by the number of nuclei per myotube (Figure S3A).
Importantly, the PGC-1a4-dependent increase in myotube size
and protein accumulation could be inhibited by an IGF1 receptor
(IGF1R) inhibitor (BMS-754807; Dinchuk et al., 2010) (Figures 3C
and S3B). Under the same conditions, no significant changes in
total protein accumulation were observed in cells expressing
GFP or PGC-1a1. Although we observed an increase in expres-
sion of the myogenic transcription factors Myf5 and 6, the levels
of MyoD and myogenin were only minimally affected (Figure 3D).
Small effects were also observed in the expression of the
atrophy-related genes MuRF-1 and atrogin-1/MAFbx (Figure 3D)
(Bodine et al., 2001; Gomes et al., 2001). Both PGC-1a isoforms
show a predominantly nuclear localization in skeletal myotubes
(Figure S3C). Taken together, these results clearly suggest a
role for PGC-1a4 in the regulation of myotube size and protein
content, mediated (at least in part) by IGF1.
PGC-1a4 Loss of Function Blunts Skeletal Muscle Cell
Hypertrophy in Culture
To evaluate the requirement for PGC-1a4 in a cellular model of
skeletal muscle hypertrophy, we developed an isoform-specific
short hairpin RNA (shRNA). As shown in Figure S3D, PGC-1a4
shRNA efficiently reduces PGC-1a4 mRNA levels in trans-
duced myotubes while not affecting PGC-1a1. Myotube hyper-
trophy was induced by treatment with the b-adrenergic agonist
clenbuterol (McMillan et al., 1992); this resulted in robust hyper-
trophy (Figure 3E) accompanied by a 5-fold increase in endoge-
nous PGC-1a4 levels and also resulted in a 1.9-fold increase in
protein/DNA ratio (Figures S3D and 3E). No changes were ob-
served in PGC-1a1 levels with clenbuterol treatment. However,
shRNA-mediated reduction in PGC-1a4 levels blunted clen-
buterol-induced myotube hypertrophy (Figure 3E) and protein

(C) Protein accumulation normalized by genomic DNA content in myotubes expressing GFP control, PGC-1a1, or PGC-1a4 and treated with the IGF1R inhibitor
BMS-754807. Myotubes were transduced as described in (A) and treated with 5 nM BMS-754807. At the end of 36 hr, cells were processed for total protein or
DNA quantification.
(D) Analysis of gene expression for markers of myogenic differentiation. Experiments were performed as described in (A), and gene expression was analyzed by
quantitative real-time PCR by using primers specific to the indicated genes.
(E) PGC-1a4 loss of function blunts clenbuterol-induced myotube hypertrophy. Fully differentiated primary myotubes were treated with 500 nM Clenbuterol
(or vehicle) and transduced with adenovirus expressing PGC-1a4-specific or scrambled control shRNAs. 48 hr later, cells were processed for fluorescence
microscopy or analysis of protein/DNA content.
(F) ChIP of DNA regions associated with Acetyl-H3K9. Myotubes expressing GFP alone or with PGC-1a4 were processed for ChIP. Purified DNA
fragments were identified and quantified by PCR using primers targeting the IGF1 gene at 1 kb intervals. Graph shows enrichment relative to input after
normalized by IgG.
(G) ChIP of regions associated with di/trimethyl-H3K9. Myotubes were processed as described above. Purified DNA fragments were identified and quantified by
PCR using primers targeting the Myostatin gene at 1 kb intervals. Graph shows enrichment relative to input normalized by IgG.
Bars depict mean values, and error bars represent SD. *p < 0.05 between indicated group and control. #p < 0.05 between all groups. See also Figure S3.

Cell 151, 1319–1331, December 7, 2012 ª2012 Elsevier Inc. 1323

accumulation, compared to a scrambled control (Figure S3E).
These results show that PGC-1a4 is required for myotube hyper-
trophy in this cellular model.
Loss of Estrogen-Related Receptors a or g Does Not
Affect PGC-1a4-Mediated Myotube Hypertrophy
The estrogen-related receptors are master drivers of OXPHOS
gene expression under PGC-1a1 coactivation (Giguère, 2008).
We investigated whether myotubes genetically deficient in
ERRa or ERRg could still support PGC-1a4 function. Differenti-
ated myotubes were transduced with adenovirus expressing
GFP control or PGC-1a4, and even in the absence of ERRa,
PGC-1a4 was still able to induce robust myotube hypertrophy
(Figure S4F); both IGF1 and myostatin genes were regulated
as previously observed in wild-type cells (Figure S3G). Similar
results were obtained with ERRg-deficient myotubes (data not
shown). In agreement with these results, and in contrast to
PGC-1a1, PGC-1a4 did not coactivate ERRa function in lucif-
erase assays using two different reporters for ERR-mediated
transactivation (Figure S3H). These results indicate that PGC-
1a4 does not coactivate ERR function, which likely contributes
to the lack of effect on OXPHOS gene expression.
PGC-1a4 Expression Induces Changes in Histone
Modifications near the IGF1 and Myostatin Genes
The PGC-1s have been shown to associate with other coregula-
tory complexes to promote chromatin remodeling and enhance
transcription of target genes. One such complex is anchored
by the histone acetyltransferase CBP/p300, which mediates
histone tail acetylation as a mark for positive gene regulation
(Ogryzko et al., 1996). Because IGF1 gene transcription is en-
hanced by PGC-1a4, we performed chromatin immunoprecipi-
tation (ChIP) experiments by using an anti-acetyl-H3K9 antibody
to identify potential IGF1 gene enhancer regions. We thus
identified three regions of DNA where association with acetyl-
H3K9 was significantly enriched upon expression of PGC-1a4
(Figure 3F). Those included the gene promoter (2.7-fold) and
30 untranslated region (UTR) (3.3-fold) and a region located
4 kb upstream of the transcription initiation site (5.9-fold). To
identify regions of the myostatin gene that could mediate the
repressive effect of PGC1a4, we performed ChIP experiments
targeting DNA regions associated with di/trimethyl-H3K9, a
mark for negative gene regulation (Heard et al., 2001). Of the
identified regions, a sequence located 5 kb upstream of the
myostatin gene showed the highest PGC-1a4-induced associa-
tion with di/trimethyl-H3K9 (74.3-fold) (Figure 3G), suggest-
ing the potential location of a regulatory region. Upon expres-
sion of PGC-1a4, the 30 UTR of the myostatin gene showed
a 3.2-fold higher association with di/tri-methyl-H3K9, whereas
the gene promoter showed a decrease of 80%. These results
identify putative regulatory regions near the IGF1 and myosta-
tin genes, which could be the target of PGC-1a4-mediated
regulation.
Delivery of PGC-1a4 to Skeletal Muscle Causes Marked
Hypertrophy In Vivo
To evaluate the effect of PGC-a4 expression in vivo, we per-
formed intramuscular injections of the adenoviral vectors. Mice
1324 Cell 151, 1319–1331, December 7, 2012 ª2012 Elsevier Inc.
with severe combined immunodeficiency (SCID) were used to
maximize adenovirus uptake by muscle and to reduce any
immune response to the adenoviruses (Chakkalakal et al.,
2010). Each mouse received an injection of GFP control adeno-
virus into the gastrocnemius of one limb (Figure S4A) and
received an equivalent injection of a PGC-1a4-expressing
adenovirus into the contralateral limb. Fibers expressing PGC-
1a4 showed a 59.5% increase in average cross-sectional area
(CSA) 7 days after injection, compared to controls (Figures 4A
and S4B). In particular, PGC1a4 drove the appearance of very
large fibers (in the >1,200 mm2 interval) that were rarely seen
with the control (Figure 4B). PGC-1a4 expression in vivo resulted
in a 2.3-fold increase in IGF1 gene expression and in a 50%
reduction in myostatin gene expression (Figure S4C). Further-
more, PGC-1a4 expression in muscle resulted in increased S6
ribosomal protein phosphorylation levels (Figure S4D), which is
indicative of S6K activation (Pende et al., 2004).
We also performed electroporation of plasmids encoding
PGC-1a4 (or a control) into the tibialis anterior of C57Bl/6
mice. As before, each mouse received the control plasmid in
one limb and received the PGC-1a4-expression plasmid in the
contralateral limb. Recently, another PGC-1a splicing variant
has cloned from a mouse BAT complementary DNA (cDNA)
library and was named NT-PGC-1a (Zhang et al., 2009).
Although we have been unable to detect expression of endoge-
nous NT-PGC-1a transcript in skeletal muscle, we tested
whether electroporation of a plasmid encoding this PGC-1a
variant would induce changes in the CSA of targeted fibers, as
observed for PGC-1a4. Although PGC-1a4 and NT-PGC-1a
mRNAs were expressed at similar levels (Figure 4C), we could
not detect NT-PGC-1a protein accumulation by immunoblotting
(Figure 4D). We observed an average increase of 28% in CSA of
PGC-1a4-expressing fibers when compared to controls (Fig-
ure S4E). Results are also shown as CSA frequency distribution
in Figure 4E and show a 34.7% reduction in the smallest CSA
interval (<1,000 mm2) and show increases of 45.5% and 99.4%
in the 2,000 and 2,500 mm2 intervals (respectively). As before,
we observed fibers of CSA in the 3,000 and >3,000 mm2 inter-
vals, which were not significantly observed in controls (Fig-
ure 4E). This increase in average CSA was accompanied by
a 10% increase in muscle weight, compared to control limb (Fig-
ure S4F). PGC-1a4 expression resulted in higher phospho-S6K
levels (Figure S4G). Expression of NT-PGC-1a resulted in a
decrease of 46.7% and 66.9% in the number of larger fibers
with CSA in the 2,000 and 2,500 mm2 intervals (respectively)
and resulted in an increase of 33.8% in smaller fibers with CSA
in the 1,500 mm2 interval (Figure 4F). These results indicate
that in vivo delivery of PGC-1a4 results in a marked increase in
the CSA of targeted fibers.
Regulation of PGC-1a4 Expression during Hindlimb
Suspension/Reloading
We next investigated whether PGC-1a4 levels are regulated by
a hindlimb suspension/reloading protocol, a model of skeletal
muscle hypertrophy in rodents (Hanson et al., 2010). As shown
in Figure 5A, hindlimb suspension resulted in a small but signif-
icant decrease in total PGC-1a content in the soleus muscle
(0.85-fold when compared to control), and reloading resulted
 GFP Figure 4. In Vivo Expression of PGC-1a4
Induces Skeletal Muscle Hypertrophy
A B PGC-1α4 (A) Adenovirus-mediated expression of PGC-1a4
60

Distribution (%)
CSA Frequency
GFP PGC-1α4 in mouse skeletal muscle. Cross-section of the
* gastrocnemius muscle 7 days after intramuscular
40 injection of adenovirus expressing GFP alone or
with PGC-1a4.
*
(B) Cross-sectional area frequency distribution
20 (sections from six mice per group).
* (C) PGC-1a4 and NT-PGC-1a mRNA expression
0 levels in electroporated tibialis anterior analyzed

00

00

0
by quantitative real-time PCR using primers tar-

20

20
<4

-8
geting exon 2 (present in both isoforms).

-1

>1
Fiber Size (μm2) (D) PGC-1a4 and NT-PGC-1a protein expression
levels in electroporated muscle.
C D (E and F) Electroporation-mediated delivery of
Relative mRNA Expression

α
-1
100 plasmids into the tibialis anterior (TA). Each mouse

α4

C
PG
-1
* (n = 6 per group) received a control plasmid in one
C

T-
80
PG
limb (pCI-neo) and received in the contralateral
60
L R L R N limb the plasmid encoding PGC-1a4 or NT-PGC-
50 kDa
* 1a. Bars depict mean values, and error bars
40 α-PGC-1α
37 kDa represent SE. *p < 0.05 between indicated group
20 and control.
37 kDa α-GAPDH See also Figure S4.
0
eo

α4

α
-1
I-n

-1

C
C
pC

PG
PG

T-
N

E F Different Exercise Protocols Result

in Differential PGC-1a Isoform
pCIneo
* Expression in Human Muscle
Distribution (%)

Distribution (%)
CSA Frequency

CSA Frequency

pCIneo
60 60 NT-PGC-1α
PGC-1α4 We determined the expression of PGC-
40 40 1a4 in human skeletal muscle with
* training. Percutaneous muscle biopsies
*
20 * * * 20 of the vastus lateralis muscle were ob-
* tained at baseline and 48 hr after the
0 0 *
last training session. Figure 5C shows
-1 0
-1 0

-2 0
-2 0
-3 0
>3 0
0
-2 0
-2 0
-3 0
>3 0
0

00
00

50
00
50
00
00
50
00
50
00
00

the change in total PGC-1a expression

<1
<1

2
at the end of the indicated exercise
Fiber size (μm2) Fiber size (μm ) protocol (Table S1). All exercise programs
elevated total PGC-1a levels in muscle,
although to different degrees. Notably,
in a 1.6-fold increase. Changes in muscle wet weights are the combined endurance/resistance protocol induced the most
shown in Figure S5A. No significant changes in PGC-1a1 significant increase in PGC-1a expression. We observed no
mRNA levels were observed upon suspension, but a marked changes in PGC-1a4 expression in the endurance protocol
reduction was seen after reloading (50% versus control), when group (Figure 5D). However, both the resistance and the
hypertrophic signaling cascades are initiated. Accordingly, combined exercise programs led to a 1.5- and 3-fold increase
during the suspension and reloading phases, we observed a in the expression of PGC-1a4 (respectively). We observed a
reduction in the expression of transcripts driven by the proxi- reduction in myostatin gene expression only in the resistance
mal promoter of the PGC-1a gene, as assessed by quantitative and combined groups (Figure 5E) but observed no statistically
real-time PCR using primers directed at that 50 UTR region significant changes in IGF1 expression (Figure 5F). Interestingly,
(Figure S5B). Interestingly, PGC-1a4 expression levels drop a very significant correlation across all groups (R = 0.64) was
to 22% of starting levels after suspension and increase by observed between the changes in PGC-1a4 expression and
18.7-fold with reloading (Figure 5A). The increase in PGC-1a4 performance in the leg press exercise (Figure 5G).
mRNA expression during the reloading phase was accom-
panied by significant protein accumulation (Figure S5C). This Transgenic Muscle-Specific Expression of PGC-1a4
expression pattern matches the suspension/atrophy and re- Drives Increased Muscle Mass, Strength, and Reduced
loading/hypertrophy phases and coincided with an increase Adiposity
in IGF1 and a decrease in myostatin gene expression (1.8- To evaluate the effects of chronically elevated PGC-1a4 levels
fold and 40%, respectively; Figure 5B). Similar results were ob- in muscle, we generated the transgenic mouse model Myo-
tained with the gastrocnemius/plantaris muscles (Figures S5A PGC-1a4. This mouse (Figure 5H) expresses PGC-1a4 under
and S5D). the control of a MEF2C enhancer/myogenin promoter (Li et al.,

Cell 151, 1319–1331, December 7, 2012 ª2012 Elsevier Inc. 1325

A Figure 5. PGC-1a4 Expression Increases
25 B
Control Control during Muscle Hypertrophy and Resistance
Relative mRNA Expression

*,# 2.5

Relative mRNA Expression

20 Suspension Suspension Training
Reloading Reloading (A and B) PGC-1a4 expression increases during
15 2.0 * the hypertrophy phase of a suspension/reloading
10 protocol. C57Bl/6 mice were divided into groups
1.5 (n = 4 each): control, 10 days hindlimb suspension
2.0 *,# (Suspension) or 10 days suspension plus 24 hr
1.5 1.0 of reloading (Reloading). The soleus muscles were
1.0 * * harvested and processed for gene expression
* 0.5 analysis by quantitative real-time PCR using
0.5
* primers specific for the indicated genes.
0.0 0.0 (C–F) Analysis of gene expression in skeletal
muscle biopsies from human volunteers. Percu-
α1

α4
l
ta

F1

N
ST
To

-1

-1

IG
taneous vastuls lateralis biopsies were obtained at
C

M
PG

PG

baseline and 8 weeks later, 48 hr after the last

training session. Gene expression was analyzed
C D E F by quantitative real-time PCR using primers
Total hPGC-1α hPGC-1α4 hMSTN hIGF1 specific to the indicated genes.

Relative mRNA Expression

Relative mRNA Expression
Relative mRNA Expression

Relative mRNA Expression

4 4 *,# 2.0 5 (G) Increase in PGC-1a4 expression correlates

4
with improvement in leg press exercise perfor-
3
*,# 3 1.5 mance. Graph shows the percent change in the
3
1.0 number of leg press repetitions observed between
2 2
* *
* * 2 baseline and after 8 weeks of training.
*
1 1 0.5 1 (H) Morphology of hindlimbs from wild-type and
PGC-1a4 transgenic mice (Myo-PGC1a4). Immu-
0 0 0.0 0
noblot using an anti-PGC-1a antibody (a-PGC-1a)
du ry

is ce

d
om ce
du ry

is ce

d
om ce
d u ary

ce

d
om ce
y

d
e

ne
ne
ne
ar

nc

nc

ne

a
En nta

shows PGC-1a4 protein levels in the gastrocne-

R ran

C an
R ran

C an
R ran

C an

En nt
nt

bi
nt

bi
bi
ra

ta

bi

t
t
t

de
de
de

is
de

is
du

om

mius muscle.
es
es
es

Se
Se
Se
es
Se

En
En

C
R

(I) Total PGC-1a and PGC-1a4 mRNA levels in the

Myo-PGC-1a4 mouse line were determined by
quantitative real-time PCR and compared to the
G H I
Leg Press Wild-type Myo-PGC-1α4 wild-type littermate controls. Bars depict mean
35 Wild-type values, and error bars represent SD. *p < 0.05
% Change over 8 weeks

Relative mRNA Expression

60
*,# *,# Myo-PGC-1α4
30 between indicated group and control. #p < 0.05
40 *
between all groups.
20 * 25 See also Figure S5.
0 20
5 *
-20 4
3
-40 2
α-PGC-1α

1
S6B). Analysis of myosin heavy chain
y

d
e
ar

nc

nc

ne

0
nt

ra

ta

bi

37 kDa
Total PGC-1α4
de

expression in transgenic muscle revealed

is
du

om
es
Se

En

C
R

a higher representation of type IIa and IIx

fibers at the expense of IIb (Figures 6C
and S6C). Interestingly, these animals
2005). Transgenic muscle expresses PGC-1a4 mRNA 30-fold also show reduced adiposity. Skeletal-muscle-specific PGC-
above endogenous levels (Figure 5I) and shows significant 1a4 transgenics show a reduction of 20.7% and 40% in the
PGC-1a4 protein accumulation (Figure 5H). Importantly, these mass of the epididymal and retroperitoneal fat depots (Fig-
levels of expression bring PGC-1a4 levels up to those observed ure 6D). Among the tissues analyzed, this observation proved
during the reloading phase of the suspension/reloading experi- to be selective to fat (Figure S6D). When compared to the wild-
ments (i.e., 19-fold, Figure 5A). Analysis of the Myo-PGC-1a4 type controls, the PGC1a4 transgenics show a 20% increase in
mouse shows redder muscle when compared to the wild-type maximum force generated (Figure 6E). Interestingly, transgenic
littermate controls (Figure 5H). The pattern of gene expression muscle also proved to be more fatigue-resistant than wild-type
observed in the muscle of Myo-PGC-1a4 mice (Figure S5E) (Figure 6F). These results show that elevation of PGC-1a4 levels
was similar to the one observed upon expression of PGC-1a4 in skeletal muscle results in a hypertrophy phenotype accompa-
in myotubes (Figures 2D–2F) with the exception of IGF1. In these nied by a significant increase in strength.
animals, and consistent with our previous results, we observed
a 64.4% reduction in myostatin expression. However, there Myo-PGC-1a4 Mice Show Less Muscle Loss upon
was no increase in IGF1 expression (Figure S5E). PGC-1a4 Hindlimb Suspension and Improved Exercise
transgenic mice showed an increase in muscle wet weight as Performance
observed in Figure 6A. The increase in muscle mass was accom- To evaluate the physiological alterations in the Myo-PGC-1a4
panied by an average increase of 12% in fiber CSA (Figure 6B), mouse, we performed hindlimb suspension/reloading experi-
with a significant increase in larger fibers (Figures S6A and ments. Myo-PGC-1a4 mice lost significantly less muscle mass

1326 Cell 151, 1319–1331, December 7, 2012 ª2012 Elsevier Inc.

 Myo- Figure 6. Myo-PGC-1a4 Skeletal Muscle
A B C WT
PGC-1α4 Transgenics Have Increased Muscle Mass
and Strength

MyHC
(Normalized by tibia length)

Wild-type (A) Determination of muscle wet weight from

I
15 Myo-PGC-1α4 PGC-1a4 transgenics and wild-type controls
0.10
* (n = 6). Muscle weights are normalized by tibia

(Arbitrary Units)
Muscle Mass

MyHC
0.08 length.
10

IIa
*
(B) Fiber cross-sectional area in the gastrocne-

CSA
0.06
mius muscle of wild-type and PGC-1a4 trans-
5 0.04
genics.

MyHC
*

IIx
0.02 (C) Immunohistochemical analysis of gastrocne-
0 0.00 mius muscle from wild-type (WT) and Myo-PGC-
Wild- Myo- 1a4 animals by using antibodies against different
TA

us

ris
c

MyHC
tro

ua

-type PGC-1α4
le

ta

myosin heavy chain (MyHC) types.

IIb
Q
as

So

an
G

(D) Determination of Epididimal (EAT) and Peri-

Pl

renal (PRAT) adipose tissue wet weight from

PGC-1a4 transgenics and wild-type controls
D E F (n = 6).
(E) Maximal force measurements of the gastroc-
Adipose Tissue Weight (mg)

Wild-type Myo-PGC-1α4
nemius muscle of wild-type and PGC1a4 trans-
Myo-PGC-1α4 2.5 Wild-type
Maximum force (N)

100
Force (% maximum)

genics (Myo-PGC-1a4).
400 *
2.0 80 (F) Muscle fatigue test performed under the same
300 experimental settings as in (E).
* 1.5 60 (G) Changes in normalized muscle mass with
*
200 1.0 40 * hindlimb suspension/reloading. n = 4 per group.
* (H) Exercise tolerance test. Muscle-specific PGC-
100
* 20 *
0.5 1a1 (MCK-PGC-1a) and PGC-1a4 (Myo-PGC-
Stimulation Recovery
0 1a4) transgenics ran to exhaustion on a treadmill.
0 0.0 0 1 2 3 4 5 6 7 8 9
EAT PRAT Wild- Myo- Data are normalized by values obtained with
-type PGC-1α4 Time (mins) wild-type animals (n = 6 per group). Bars depict
mean values, and error bars represent SE. *p <
G 30 H 0.05 between indicated group and control. #p <
Wild-type 0.05 between all groups.
20
MCK-PGC-1α1 See also Figure S6 and Table S1.
% Change Normalized

10 Myo-PGC-1α4
2.0
Muscle Mass

0 *
Wild-type *
Suspension *,#
Fold Change

-10 1.5 *,#

Wild-type
-20 Reloading
Myo-PGC-1α4
1.0
-30 in food consumption (Figure S6F) or
Suspension
* movement (data not shown) were ob-
Myo-PGC-1α4 0.5
-40
served. These results show that the
Reloading
-50 *
0.0 increase in muscle mass and strength
Time Distance
observed in the Myo-PGC-1a4 mice
does not result in an overt metabolic
phenotype but contributes to better exer-
during the suspension phase (Figure 6G) and trended toward cise performance and protects against hindlimb unloading-
accumulating more muscle mass during the reloading process. induced atrophy.
Myo-PGC-1a4 mice were evaluated for their performance on
an exercise tolerance test and were compared to littermate Myo-PGC-1a4 Mice Show Resistance to Cancer-
controls and to the previously reported mice with muscle- Induced Cachexia with Improved Glucose Tolerance and
specific PGC-1a1 expression (MCK-PGC-1a). As expected Preserved Muscle Mass
(Lin et al., 2002), MCK-PGC-1a mice remained on the tread- Cancer cachexia is a very important complication of many malig-
mill 60% longer than controls (Figure 6H). The Myo-PGC-1a4 nancies, resulting in severe muscle wasting and negatively
mice showed improved performance, with a 28% increase in affecting patient outcome (Fearon et al., 2012). Wild-type and
time to exhaustion compared to controls (Figure 6H). No differ- Myo-PGC-1a4 mice were inoculated with the Lewis lung carci-
ences in glucose tolerance were observed between these noma (LLC) cells, and the cachectic phenotype was evaluated
mouse models and controls (Figure S6E). Metabolic profiling 24–28 days after inoculation. Importantly, this model of cachexia
of the Myo-PGC-1a4 mice by using Comprehensive Laboratory is characterized by loss of muscle mass and strength and by
Animal Monitoring System (CLAMS) revealed that, although development of glucose intolerance (Busquets et al., 2012;
these animals have higher maximum oxygen consumption Das et al., 2011). Upon necropsy analysis of tumor-bearing
(VO2) and carbon dioxide elimination (VCO2), their respiratory animals, muscles of Myo-PGC-1a4 mice looked less pale and
exchange ratio (RER) is unchanged (Figure S6F). No changes healthier when compared to cachectic muscle from wild-type

Cell 151, 1319–1331, December 7, 2012 ª2012 Elsevier Inc. 1327

A Myo-PGC-1 4 B Figure 7. PGC-1a4 Transgenic Mice Show
Wild-type

% change in Gastroc Weight

(Normalized to NTB control)
Wild-type Resistance to Muscle Wasting during
+LLC +LLC 0
Experimental Cancer Cachexia
-10 (A) Representative images of hindlimb muscles
from wild-type and tumor-bearing (+LLC) wild-
*,#
-20 type or Myo-PGC-1a4 mice.
(B) Gastrocnemius muscle mass in wild-type and
-30
Myo-PGC-1a4 tumor-bearing mice normalized to
*
-40 their own genotype non-tumor-bearing control.
Wild-type Myo-PGC-1 4 (C) Muscle force production in wild-type and Myo-
PGC-1a4 tumor-bearing mice normalized to their
Wild-type own genotype non-tumor-bearing control.
C D 3 Wild-type+LLC (D) Myostatin and IGF-1 mRNA expression in wild-
Relative mRNA Expression
(Normalized to NTB control)

* Myo-PGC-1α4 type and Myo-PGC-1a4 with or without LLC

% change in Force (Po)

-10 Myo-PGC-1α4+LLC tumor.

2 (E) Physical activity throughout the progression of
-20
*,# *,# tumor load.
-30 (F) Glucose tolerance test.
-40 1 # *,# (G) Quantification of glucose clearance. Bars
* depict mean values, and error bars represent SE.
-50 *
Wild-type Myo-PGC-1 4
*p < 0.05 between indicated group and its geno-
0 type control. #p < 0.05 between indicated group
MSTN IGF1 and group receiving same treatment across
genotypes. *,#p < 0.05 between all groups. yp <
0.05 between indicated group and wild-type mice.
E Wild-type F G zp < 0.05 between indicated group and wild-type +
Wild-type+LLC
GTT (area under curve)

Myo-PGC-1 4+LLC 40000 LLC mice.

Wt *
See also Figure S7.
Physical Activity (Counts)

Wt+LLC
2500 500 30000
Glucose (mg/dl)

† † † PGC-1a4+LLC *,#
2000 400 † 20000
1500
†,‡
300 †,‡ †,‡
10000
1000 200 †,‡
100 0 the development of the cachexia pheno-
500 †
C
LC
pe

† type for physical activity, glucose toler-

LL

0
-ty

+L

4+

0
ild

pe

t 0 20 40 60 90 120
W

-ty

-1

e
on Pr te er
e
ance, and food consumption. No
C

ra
ild

ev
G

C e Time (min)
-P
W

od S
changes in food intake were observed
yo

M
M

Tumor Progression between wild-type and Myo-PGC-1a4

mice with or without LLC cell inoculation
(Figure S7E). Although physical activity
mice (Figure 7A). Strikingly, tumor-bearing Myo-PGC-1a4 mice was similar between wild-type and Myo-PGC-1a4 mice before
lost only 10% of gastrocnemius muscle mass (Figure 7B), the inoculation of the cancer cells (Figure 7E), it declined more
compared to wild-type tumor-bearing animals (29%). This was strongly with tumor load in wild-type than in Myo-PGC-1a4
accompanied by improvements in both muscle force produc- mice (Figure 7E). Tumor-bearing Myo-PGC-1a4 mice displayed
tion and myofiber cross-sectional area (Figures 7C and S7A). improved glucose tolerance compared to cachectic wild-type
Cancer-induced cachexia has been shown to occur concomi- mice, as shown by their ability to clear a bolus of glucose (Figures
tantly with dysregulation of myostatin and IGF1 expression in 7F and 7G). Thus, transgenic expression of PGC-1a4 in muscle
muscle (Busquets et al., 2012; Fearon et al., 2012; White et al., dramatically ameliorates cancer-induced cachexia through re-
2011). Analysis of muscle gene expression in the different groups duced loss of muscle mass and strength and improved glucose
revealed that cachectic wild-type animals indeed have a 2.2-fold homeostasis during cancer progression.
increase in myostatin gene expression, but only a 1.2-fold in-
crease was seen in tumor-bearing Myo-PGC-1a4 animals (Fig- DISCUSSION
ure 7D). Conversely, IGF1 gene expression was 59% lower in
cachectic wild-type muscle, whereas the corresponding Myo- Exercise is usually considered in two broad categories: endur-
PGC-1a4 mice showed only a 27% reduction (Figure 7D). ance training, which involves low-resistance work of longer dura-
SMAD phosphorylation, a marker of myostatin signal transduc- tion, and resistance training, which requires more powerful
tion, was clearly suppressed in the Myo-PGC-1a4 mice (Fig- movements of shorter duration. Health studies suggest that
ure S7B). The expression of other muscle wasting markers most humans should participate in both types of exercise, espe-
such as MuRF1 and atrogin1/MAFbx1 (Figure S7C) was also cially in aging populations (Nair, 2005). Molecular mechanisms
strongly reduced in cachectic Myo-PGC-1a4 mice (2.3- and underlying the different exercise modalities are not well under-
1.9-fold, respectively) when compared to wild-type tumor- stood, but the importance of maintaining muscle mass and
bearing controls (5.6- and 7.1-fold, respectively). Tumors in strength as humans age has motivated considerable research
both genotypes of mice grew at the same rates (Figure S7D). into the pathways of muscle growth; these have highlighted
Wild-type and Myo-PGC-1a4 mice were studied throughout the importance of the IGF1 and myostatin systems (Schiaffino

1328 Cell 151, 1319–1331, December 7, 2012 ª2012 Elsevier Inc.

and Mammucari, 2011; Lee, 2004; McPherron et al., 1997). PGC-
1a, now termed PGC-1a1, is induced by exercise and regulates
many of the effects of endurance training: mitochondrial bio-
genesis, fiber-type switching, angiogenesis, and resistance to
muscle atrophy. However, gain-of-function studies with PGC-
1a have not shown any increase in either muscle mass or
strength (Lin et al., 2002; Sandri et al., 2006; Wende et al.,
2007). Similarly, other PGC-1a variants have been shown to spe-
cifically affect energy metabolism in BAT and skeletal muscle,
but no effects on cell size have been reported (Tadaishi et al.,
2011; Yoshioka et al., 2009; Zhang et al., 2009).
Here, we show that the Pgc-1a gene encodes a protein (PGC-
1a4) that is expressed robustly in muscle and is particularly
induced by resistance training. The Pgc-1a gene has been
shown to be subject to alternative splicing, but the PGC-1a4 iso-
form is undescribed. It is somewhat similar to the NT-PGC-1a
isoform (Zhang et al., 2009), but PGC-1a4 comes from a different
gene promoter and has a completely distinct N terminus derived
from the upstream exon 1. Our data strongly suggest that the
distinct N-terminal sequence of the PGC-1a4 is important in
allowing this protein to accumulate in skeletal muscle and
perhaps in other tissues.
PGC-1a4 specifically activates the expression of IGF1 and
suppresses myostatin in skeletal muscle. That the net result of
PGC-1a4 expression in cultured myotubes represents bona
fide hypertrophy, as opposed to hyperplasia, can be deduced
from the fact that protein content per unit DNA doubles. In a
classical cellular system of muscle cell hypertrophy, stimula-
tion with clenbuterol, PGC1a4 is required for this increase in
myotube size. PGC-1a4 also induces muscle fiber hypertrophy
in vivo in three independent expression systems used here:
transduction with adenoviral vectors, delivery of naked DNA,
and transgenic expression with a muscle-selective promoter.
All of these systems give very comparable results, with muscle
fiber hypertrophy, modulation of myostatin expression, and in
some cases, increases in IGF1 expression. Importantly, we see
no changes in mitochondrial gene expression, which is a virtual
hallmark of PGC-1a1 expression in skeletal muscle. This could
be partly explained by the fact that PGC-1a4 does not robustly
coactivate ERRs.
Transgenic mice with an increase in PGC-1a4 expression
within or close to the physiological range results in a 60%
decrease in myostatin mRNA expression. This is comparable to
the 50% decrease observed in mice heterozygous for a null muta-
tion in the myostatin gene. Muscle mass changes in myostatin
heterozygous mice range from 7% to 35%, depending on the
muscle type and study (Gentry et al., 2011; Lee, 2007; Mendias
et al., 2006). The changes observed in the Myo-PGC-1a4 mouse
are comparable to those reported in two of those studies (Gentry
et al., 2011; Mendias et al., 2006). Although effects on other
molecular systems are entirely plausible, the effect of PGC-1a4
on myostatin gene expression is likely to be a major contributor
to the phenotype seen in the Myo-PGC-1a4 animals.
Importantly, PGC-1a4 transgenic mice have an increase in
muscle force production proportional to the increase in muscle
mass, indicating that forced PGC-1a4 expression in muscle
results in functional hypertrophy. The increase in muscle mass
and strength results in other significant functional conse-
quences, including exercise tolerance and resistance to hin-
dlimb suspension-induced muscle atrophy. The transgenic ex-
pression of PGC-1a4 also had a strikingly protective effect in
the pathogenesis of cancer-induced cachexia. Although tumors
grew at an equal rate in mice of both genotypes, tumor-bearing
Myo-PGC-1a4 mice lose less muscle mass and strength than
controls and do not develop the glucose intolerance observed
in the cachectic control animals. This protective effect is clearly
reflected in the spontaneous physical activity of the tumor-
bearing transgenic animals, which approaches that of non-
tumor-bearing wild-type controls, even in the most severe
phases of tumor development.
Because PGC-1a4 contains the complete activation domain
also present in PGC-1a1, it seems highly likely that PGC-1a4
also functions as a coactivator. However, because PGC-1a1
does not stimulate IGF1 and myostatin gene expression, PGC-
1a4 may be interacting with a very different set of DNA-binding
proteins. Interestingly, histone deacetylase (HDAC) inhibition
can induce muscle hypertrophy (Iezzi et al., 2004) through a
mechanism that involves induction of follistatin expression and
increased myoblast recruitment and fusion. Follistatin is a
negative regulator of myostatin action and therefore acts posi-
tively on muscle mass independent of the IGF1 pathway (Iezzi
et al., 2004). We have observed a reduction in follistatin mRNA
levels upon PGC-1a4 expression (Figure 2F), which would sug-
gest that it does not contribute to PGC-1a4-induced muscle
hypertrophy. In agreement with this result, PGC-1a4 can pro-
mote myotube hypertrophy without an increase in myoblast
fusion (Figure S3A).
Finally, the potential of PGC-1a4 in therapeutics seems
obvious. Loss of muscle mass and strength in aging, wasting
diseases and in certain muscular dystrophies affects both quality
of life and longevity. While protein therapeutics for both the IGF1
and the myostatin pathways are being developed, PGC-1a4 can,
in principle, modulate both of these systems in a highly coordi-
nated way. Because PGC-1a4 comes from a distinct promoter
with some degree of distinct regulation in vivo, it will be important
to find chemical matter that can modulate PGC-1a4 gene
expression. In addition, PGC-1a4 gene expression might be
used as a helpful readout for optimizing resistance training that
focuses on increasing muscle strength.
EXPERIMENTAL PROCEDURES
PGC-1a Isoform Cloning and Detection
PGC-1a isoforms were cloned from a mouse soleus cDNA library. Primer
sequences used for cloning and for quantitative real-time PCR detection of
the different isoforms in mouse and human tissues can be found in Extended
Experimental Procedures.
Cell Culture, Western Blotting, and Immunocytochemistry
Primary satellite cells (myoblasts) were isolated, maintained, and differentiated
as described previously (Megeney et al., 1996). To detect all PGC-1a variants
by immunoblotting, anti-PGC-1a antibodies were obtained from Calbiochem
(ST1202). For PGC-1a isoform immunoprecipitation experiments, antibodies
were obtained from Santa Cruz Biotechnologies (sc-13067).
Adenovirus-Mediated Expression
Adenovirus expressing the different mouse PGC1a isoforms or GFP was
generated by using the pAdTrack/pAdEasy system (Stratagene). For cell
Cell 151, 1319–1331, December 7, 2012 ª2012 Elsevier Inc. 1329
culture experiments, differentiated myotubes were transduced with an adeno-
virus at an MOI of 100 overnight. For intramuscular delivery of adenovirus,
PGC1a4 or GFP viral stock (10 ml, 2 3 1010 infectious particles) was injected
into the lower limbs of young SCID mice. Each mouse received the control
adenovirus in one limb and the PGC-1a4 in the contralateral limb. Tissues
were harvested 7 days after injection.
Clenbuterol-Induced Myotube Hypertrophy and PGC-1a4
Knockdown
Differentiated primary myotubes were treated with 500 nM clenbuterol (Sigma)
and PGC-1a4 shRNA (ATA AAT GTG CCA TAT CTT CCA) or scrambled shRNA
for 48 hr. Cells were then divided in aliquots for genomic DNA and total protein
quantification.
Chromatin Immunoprecipitation
ChIP was carried out according to a protocol from Upstate Biotechnology.
Input DNA and immunoprecipitated DNA were analyzed by quantitative PCR.
Primer sequences used to detect each DNA region are available upon request.
Electric-Pulse-Mediated Gene Transfer
Electric-pulse-mediated gene transfer was modified from previous studies (Mir
et al., 1999). A detailed description can be found in Extended Experimental
Procedures.
Fiber Type and Cross-Sectional Area Determination
Fiber type determination was done as previously described (Arany et al., 2007).
CSA was determined with ImageJ software.
Hindlimb Suspension and Reloading
HS was performed as described previously (Hanson et al., 2010). All protocols
were approved by the University of Colorado Institutional Animal Care and Use
Committee. A detailed description is included in Extended Experimental
Procedures.
Human Exercise Training
Table S1 provides a complete description of the exercise programs, and
a detailed protocol can be found in Extended Experimental Procedures.
Generation of Transgenic Mice and Animal Experimentation
Generation of Myo-PGC-1a4 mice was done by cloning the PGC-1a4 cDNA in
front of a MEF2C enhancer/Myogenin promoter DNA sequences (Li et al., 2005),
followed by the human growth hormone polyA. Mice were maintained under
standard conditions. All experiments and protocols were performed in accor-
dance with the Dana-Farber Cancer Institute or Beth Israel Deaconess Medical
Center Animal Facility Institutional Animal Care and Use Committee regulations.
Muscle Force Measurements
Maximal muscle force was measured as previously described (Axell et al.,
2006) with an isometric transducer (ADInstruments, Colorado Springs, CO).
Specific maximal force was quantified by correcting for muscle mass. A fatigue
test was performed following the maximal contractions.
Exercise Tolerance Test
Exercise tolerance was performed as previously described (Arany et al., 2007).
Cachexia Studies
Wild-type and Myo-PGC-1a4 mice were given a subcutaneous injection in the
left flank of either 107 LLC cells in PBS or PBS alone (control). Stages of tumor
severity were defined as follows: Pre (before development of tumor), 1–5 days
after inoculation; Moderate (moderate tumor size), 10–15 days after inocula-
tion; and Severe (large tumor size), 23–28 days after inoculation. Mice with
tumor development were sacrificed 28 days after injection.
Statistical Analysis
All results are expressed as means ±SD for cell experiments and ±SEM for
animal experiments. Two-tailed Student’s t test was used to determine
p values. Statistical significance was defined as p < 0.05.
1330 Cell 151, 1319–1331, December 7, 2012 ª2012 Elsevier Inc.
ACCESSION NUMBERS
The GenBank accession numbers for mRNA sequences are JX866946,
JX866947, and JX866948. Microarray data are deposited at GEO with the
accession number GSE42473.
SUPPLEMENTAL INFORMATION
Supplemental Information includes Extended Experimental Procedures, seven
figures, and one table and can be found with this article online at http://dx.doi.
org/10.1016/j.cell.2012.10.050.
ACKNOWLEDGMENTS
We thank Drs. Srikripa Devarakonda and Sibylle Jäger for valuable discus-
sions. ERRa and ERRg KO myoblasts were a kind gift from Dr. Zhidan Wu
(Novartis Institutes for Biomedical Research). The MEF2C/Myogenin promoter
cassette was kindly provided by Dr. Eric Olson (University of Texas
Southwestern Medical Center). LLC cells were kindly donated by Dr. Jose
M. Garcia (Baylor College of Medicine). This project was supported by grants
(DK061562) from the NIH and from Novartis to B.M.S. J.L.R. was supported in
part by a grant from the Wenner-Gren Foundations, Sweden. This research
was supported in part by grants to B.C.H. (NIH, 5K01AR55676-2), N.P.G.
(NIH, T32HL07284), J.W. (AHA, 09POST2010078 and 12SDG8070003),
B.A.I. (RR024151 and AG09531), Z.Y. (NIH, AR050429), and L.A.L. (NIH,
GM29090). B.M.S. is a shareholder and consultant to Ember Therapeutics
and has received funding in the form of sponsored research from Novartis, Inc.
Received: February 1, 2012
Revised: June 30, 2012
Accepted: October 26, 2012
Published: December 6, 2012
REFERENCES
Adams, G.R. (2002). Invited review: autocrine/paracrine IGF-I and skeletal
muscle adaptation. J. Appl. Physiol. 93, 1159–1167.
Arany, Z. (2008). PGC-1 coactivators and skeletal muscle adaptations in health
and disease. Curr. Opin. Genet. Dev. 18, 426–434.
Arany, Z., Lebrasseur, N., Morris, C., Smith, E., Yang, W., Ma, Y., Chin, S., and
Spiegelman, B.M. (2007). The transcriptional coactivator PGC-1beta drives
the formation of oxidative type IIX fibers in skeletal muscle. Cell Metab. 5,
35–46.
Axell, A.M., MacLean, H.E., Plant, D.R., Harcourt, L.J., Davis, J.A., Jimenez,
M., Handelsman, D.J., Lynch, G.S., and Zajac, J.D. (2006). Continuous testos-
terone administration prevents skeletal muscle atrophy and enhances resis-
tance to fatigue in orchidectomized male mice. Am. J. Physiol. Endocrinol.
Metab. 291, E506–E516.
Bodine, S.C., Latres, E., Baumhueter, S., Lai, V.K., Nunez, L., Clarke, B.A.,
Poueymirou, W.T., Panaro, F.J., Na, E., Dharmarajan, K., et al. (2001). Identifi-
cation of ubiquitin ligases required for skeletal muscle atrophy. Science 294,
1704–1708.
Busquets, S., Toledo, M., Orpı́, M., Massa, D., Porta, M., Capdevila, E., Padilla,
N., Frailis, V., López-Soriano, F.J., Han, H.Q., and Argilés, J.M. (2012). Myosta-
tin blockage using actRIIB antagonism in mice bearing the Lewis lung carci-
noma results in the improvement of muscle wasting and physical performance.
J Cachexia Sarcopenia Muscle 3, 37–43.
Chakkalakal, J.V., Nishimune, H., Ruas, J.L., Spiegelman, B.M., and Sanes,
J.R. (2010). Retrograde influence of muscle fibers on their innervation revealed
by a novel marker for slow motoneurons. Development 137, 3489–3499.
Chinsomboon, J., Ruas, J., Gupta, R.K., Thom, R., Shoag, J., Rowe, G.C., Sa-
wada, N., Raghuram, S., and Arany, Z. (2009). The transcriptional coactivator
PGC-1alpha mediates exercise-induced angiogenesis in skeletal muscle.
Proc. Natl. Acad. Sci. USA 106, 21401–21406.
Choi, C.S., Befroy, D.E., Codella, R., Kim, S., Reznick, R.M., Hwang, Y.J., Liu,
Z.X., Lee, H.Y., Distefano, A., Samuel, V.T., et al. (2008). Paradoxical effects of
increased expression of PGC-1alpha on muscle mitochondrial function and
insulin-stimulated muscle glucose metabolism. Proc. Natl. Acad. Sci. USA
105, 19926–19931.
Das, S.K., Eder, S., Schauer, S., Diwoky, C., Temmel, H., Guertl, B., Gorkie-
wicz, G., Tamilarasan, K.P., Kumari, P., Trauner, M., et al. (2011). Adipose
triglyceride lipase contributes to cancer-associated cachexia. Science 333,
233–238.
Dinchuk, J.E., Cao, C., Huang, F., Reeves, K.A., Wang, J., Myers, F., Cantor,
G.H., Zhou, X., Attar, R.M., Gottardis, M., and Carboni, J.M. (2010). Insulin
receptor (IR) pathway hyperactivity in IGF-IR null cells and suppression of
downstream growth signaling using the dual IGF-IR/IR inhibitor, BMS-
754807. Endocrinology 151, 4123–4132.
Fearon, K.C., Glass, D.J., and Guttridge, D.C. (2012). Cancer cachexia: medi-
ators, signaling, and metabolic pathways. Cell Metab. 16, 153–166.
Florini, J.R. (1987). Hormonal control of muscle growth. Muscle Nerve 10,
577–598.
Gentry, B.A., Ferreira, J.A., Phillips, C.L., and Brown, M. (2011). Hindlimb skel-
etal muscle function in myostatin-deficient mice. Muscle Nerve 43, 49–57.
Giguère, V. (2008). Transcriptional control of energy homeostasis by the
estrogen-related receptors. Endocr. Rev. 29, 677–696.
Gomes, M.D., Lecker, S.H., Jagoe, R.T., Navon, A., and Goldberg, A.L. (2001).
Atrogin-1, a muscle-specific F-box protein highly expressed during muscle
atrophy. Proc. Natl. Acad. Sci. USA 98, 14440–14445.
Handschin, C., Choi, C.S., Chin, S., Kim, S., Kawamori, D., Kurpad, A.J., Neu-
bauer, N., Hu, J., Mootha, V.K., Kim, Y.B., et al. (2007). Abnormal glucose
homeostasis in skeletal muscle-specific PGC-1alpha knockout mice reveals
skeletal muscle-pancreatic beta cell crosstalk. J. Clin. Invest. 117, 3463–3474.
Hanson, A.M., Stodieck, L.S., Cannon, C.M., Simske, S.J., and Ferguson, V.L.
(2010). Seven days of muscle re-loading and voluntary wheel running following
hindlimb suspension in mice restores running performance, muscle mor-
phology and metrics of fatigue but not muscle strength. J. Muscle Res. Cell
Motil. 31, 141–153.
Heard, E., Rougeulle, C., Arnaud, D., Avner, P., Allis, C.D., and Spector, D.L.
(2001). Methylation of histone H3 at Lys-9 is an early mark on the X chromo-
some during X inactivation. Cell 107, 727–738.
Iezzi, S., Di Padova, M., Serra, C., Caretti, G., Simone, C., Maklan, E., Minetti,
G., Zhao, P., Hoffman, E.P., Puri, P.L., and Sartorelli, V. (2004). Deacetylase
inhibitors increase muscle cell size by promoting myoblast recruitment and
fusion through induction of follistatin. Dev. Cell 6, 673–684.
Lee, S.J. (2004). Regulation of muscle mass by myostatin. Annu. Rev. Cell Dev.
Biol. 20, 61–86.
Lee, S.J. (2007). Quadrupling muscle mass in mice by targeting TGF-beta
signaling pathways. PLoS ONE 2, e789.
Li, S., Czubryt, M.P., McAnally, J., Bassel-Duby, R., Richardson, J.A., Wiebel,
F.F., Nordheim, A., and Olson, E.N. (2005). Requirement for serum response
factor for skeletal muscle growth and maturation revealed by tissue-specific
gene deletion in mice. Proc. Natl. Acad. Sci. USA 102, 1082–1087.
Lin, J., Wu, H., Tarr, P.T., Zhang, C.Y., Wu, Z., Boss, O., Michael, L.F., Puig-
server, P., Isotani, E., Olson, E.N., et al. (2002). Transcriptional co-activator
PGC-1 alpha drives the formation of slow-twitch muscle fibres. Nature 418,
797–801.
McMillan, D.N., Noble, B.S., and Maltin, C.A. (1992). The effect of the beta-
adrenergic agonist clenbuterol on growth and protein metabolism in rat
muscle cell cultures. J. Anim. Sci. 70, 3014–3023.
McPherron, A.C., Lawler, A.M., and Lee, S.J. (1997). Regulation of skeletal
muscle mass in mice by a new TGF-beta superfamily member. Nature 387,
83–90.
Megeney, L.A., Perry, R.L., LeCouter, J.E., and Rudnicki, M.A. (1996). bFGF
and LIF signaling activates STAT3 in proliferating myoblasts. Dev. Genet. 19,
139–145.
Mendias, C.L., Marcin, J.E., Calerdon, D.R., and Faulkner, J.A. (2006).
Contractile properties of EDL and soleus muscles of myostatin-deficient
mice. J. Appl. Physiol. 101, 898–905.
Mir, L.M., Bureau, M.F., Gehl, J., Rangara, R., Rouy, D., Caillaud, J.M., De-
laere, P., Branellec, D., Schwartz, B., and Scherman, D. (1999). High-efficiency
gene transfer into skeletal muscle mediated by electric pulses. Proc. Natl.
Acad. Sci. USA 96, 4262–4267.
Nair, K.S. (2005). Aging muscle. Am. J. Clin. Nutr. 81, 953–963.
Ogryzko, V.V., Schiltz, R.L., Russanova, V., Howard, B.H., and Nakatani, Y.
(1996). The transcriptional coactivators p300 and CBP are histone acetyltrans-
ferases. Cell 87, 953–959.
Pende, M., Um, S.H., Mieulet, V., Sticker, M., Goss, V.L., Mestan, J., Mueller,
M., Fumagalli, S., Kozma, S.C., and Thomas, G. (2004). S6K1(-/-)/S6K2(-/-)
mice exhibit perinatal lethality and rapamycin-sensitive 50 -terminal oligopyri-
midine mRNA translation and reveal a mitogen-activated protein kinase-
dependent S6 kinase pathway. Mol. Cell. Biol. 24, 3112–3124.
Sandri, M., Lin, J., Handschin, C., Yang, W., Arany, Z.P., Lecker, S.H., Gold-
berg, A.L., and Spiegelman, B.M. (2006). PGC-1alpha protects skeletal muscle
from atrophy by suppressing FoxO3 action and atrophy-specific gene tran-
scription. Proc. Natl. Acad. Sci. USA 103, 16260–16265.
Schiaffino, S., and Mammucari, C. (2011). Regulation of skeletal muscle
growth by the IGF1-Akt/PKB pathway: insights from genetic models. Skelet.
Muscle 1, 4.
Short, K.R., Vittone, J.L., Bigelow, M.L., Proctor, D.N., Rizza, R.A., Coenen-
Schimke, J.M., and Nair, K.S. (2003). Impact of aerobic exercise training on
age-related changes in insulin sensitivity and muscle oxidative capacity. Dia-
betes 52, 1888–1896.
Tadaishi, M., Miura, S., Kai, Y., Kano, Y., Oishi, Y., and Ezaki, O. (2011). Skel-
etal muscle-specific expression of PGC-1a-b, an exercise-responsive iso-
form, increases exercise capacity and peak oxygen uptake. PLoS ONE 6,
e28290.
Wende, A.R., Schaeffer, P.J., Parker, G.J., Zechner, C., Han, D.H., Chen,
M.M., Hancock, C.R., Lehman, J.J., Huss, J.M., McClain, D.A., et al. (2007).
A role for the transcriptional coactivator PGC-1alpha in muscle refueling. J.
Biol. Chem. 282, 36642–36651.
Wenz, T., Rossi, S.G., Rotundo, R.L., Spiegelman, B.M., and Moraes, C.T.
(2009). Increased muscle PGC-1alpha expression protects from sarcopenia
and metabolic disease during aging. Proc. Natl. Acad. Sci. USA 106, 20405–
20410.
White, J.P., Baynes, J.W., Welle, S.L., Kostek, M.C., Matesic, L.E., Sato, S.,
and Carson, J.A. (2011). The regulation of skeletal muscle protein turnover
during the progression of cancer cachexia in the Apc(Min/+) mouse. PLoS
ONE 6, e24650.
Yoshioka, T., Inagaki, K., Noguchi, T., Sakai, M., Ogawa, W., Hosooka, T., Igu-
chi, H., Watanabe, E., Matsuki, Y., Hiramatsu, R., and Kasuga, M. (2009).
Identification and characterization of an alternative promoter of the human
PGC-1alpha gene. Biochem. Biophys. Res. Commun. 381, 537–543.
Zhang, Y., Huypens, P., Adamson, A.W., Chang, J.S., Henagan, T.M., Bou-
dreau, A., Lenard, N.R., Burk, D., Klein, J., Perwitz, N., et al. (2009). Alternative
mRNA splicing produces a novel biologically active short isoform of PGC-
1alpha. J. Biol. Chem. 284, 32813–32826.
Cell 151, 1319–1331, December 7, 2012 ª2012 Elsevier Inc. 1331
Supplemental Information

EXTENDED EXPERIMENTAL PROCEDURES

PGC-1a Isoform Cloning and Detection

PGC-1a isoforms were cloned using different combinations of the following forward and reverse primers: F1: CTG TCT GGA CTG
TCC ATA GG, F2: CCA CCA GAA TGA GTG ACA TGG A, and R1: TGC CTG CAG TAT CCT CTC C. Primer sequences used to detect
the different PGC-1a isoforms in mouse and humans samples are: mTotal_F: TGA TGT GAA TGA CTT GGA TAC AGA CA, mTotal_R:
GCT CAT TGT TGT ACT GGT TGG ATA TG, ma1_F: GGA CAT GTG CAG CCA AGA CTC T, ma1_R: CAC TTC AAT CCA CCC AGA
AAG CT, ma2_F: CCA CCA GAA TGA GTG ACA TGG A, ma2_R: GTT CAG CAA GAT CTG GGC AAA, ma3_F: AAG TGA GTA ACC
GGA GGC ATT C, ma3_R: TTC AGG AAG ATC TGG GCA AAG A, ma4_F: TCA CAC CAA ACC CAC AGA AA, ma4_R: CTG GAA GAT
ATG GCA CAT, hTotal_F: CAG CCT CTT TGC CCA GAT CTT, hTotal_R: TCA CTG CAC CAC TTG AGT CCA C, ha1_F: ATG GAG TGA
CAT CGA GTG TGC T, ha1_R: GAG TCC ACC CAG AAA GCT GT, ha2_F: AGT CCA CCC AGA AAG CTG TCT, ha2_R: ATG AAT GAC
ACA CAT GTT GGG, ha3_F: CTG CAC CTA GGA GGC TTT ATG C, ha3_R: CAA TCC ACC CAG AAA GCT GTC T, ha4_F: TCA CAC
CAA ACC CAC AGA GA, ha4_R: CTG GAA GAT ATG GCA CAT.

Animal Experimentation
MCK-PGC-1a, fat-specific PGC-1a knockout mice (FKO), and floxed and Adiponectin-Cre controls have been previously described
(Kleiner et al., 2012; Lin et al., 2002).
Electric Pulse-Mediated Gene Transfer
Under isoflurane anesthesia, both tibialis anterior (TA) muscles were injected with a mixture of DNA (50 mg of plasmid DNA encoding
PGC-1a4 or NT-PGC-1a with pGFP3 into one TA muscle with pCI-neo with pGFP3 to the contralateral muscle by use of a 0.5-ml
syringe with a 28-gauge needle at a rate of < 0.015 ml/min. Eight electric pulses (100 ms, 1 Hz, 100 V) were delivered immediately
to the injected TA muscle using a square-pulse stimulator (model S88K, Grass Telefactor) through a two-needle array (model 533,
BTX) placed on the medial and lateral sides of the TA muscle, so that the electrical field was perpendicular to the long axis of the
myofibers. Mice were placed under isoflurane anesthesia and TA muscles harvested 20 days following injection.

Hindlimb Suspension and Reloading

C57BL/6j male mice were divided into control (n = 4), 10 days of hindlimb suspension (HS; n = 4), or 10 days of hindlimb suspension
plus 1 day of reloading (HS+RL; n = 4). For the experiment with the Myo-PGC-1a4 mice, the suspension/reloading group was eval-
uated upon 10 days of reloading to achieve full recovery from the suspension process. Mice were housed individually in 19.1 cm x
29.2 cm x 12.7 cm, clear polycarbonate cages with ad libitum access to food and water. HS+RL animals were returned to normal
cage activity for 24 hr following the 10-day HS period.

Human Exercise Training

Participants were divided into the following groups. Control: The control participants were asked to maintain their current level of
physical activity during the no exercise control period. Following the no exercise control period, the control participants complete
8 weeks of combined endurance and resistance training. Endurance Training (ET): During week 1, participants completed 30 min
of stationary cycling at 65% VO2 peak 3 days per week. During week 2, participants completed 45 min of stationary cycling at
65% VO2 peak 3 days per week. During week 3, participants completed 45 min of stationary cycling at 65% VO2 peak 5 days per
week. During weeks 4-8, participants completed 60 min of stationary cycling at 65% VO2 peak 5 days per week. Resistance Training
(RT): During week 1, participants were familiarized with resistance training program and practiced the movements with light weight
during each of the four training sessions. During week 2, participants completed 2 sets of 8-10 repetitions to failure 4 days per week.
During week 3, participants completed 3 sets of 8-10 repetitions to failure 4 days per week. During weeks 4-8, participants completed
4 sets of 8-10 repetitions to failure 4 days per week. Table S1 presents the full exercise program. Combined Training (CT): The
progression of the ET was the same as that described for the ET group, except that the durations were half as long as the ET group
(i.e., 30 min versus 60 min). The progression of the RT was the same as that described for the RT group, except that the number of lifts
was less the RT group. Biopsy: Percutaneous muscle biopsies of the vastus lateralis (
400 mg) were obtained under local anesthesia
(lidocaine, 2%) (Nair et al., 1995) at baseline and
48h after the last training session. Cardiorespiratory Fitness and Strength: Peak
pulmonary oxygen uptake (VO2 Peak) was measured on a cycle ergometer using an incremental protocol to volitional fatigue with
continuous monitoring of expired gasses, heart rate, and blood pressure (Proctor and Beck, 1996). Lower- and upper-body strength
was assessed by determining the 1-repitition maximums (1-RM) for the leg-press and chest-press, respectively.

Metabolic Phenotyping
For glucose tolerance tests, animals were fasted overnight. The next morning, glucose levels in tail blood were measured with a stan-
dard glucometer prior to and at timed intervals following intraperitoneal injection of 2 g/kg D-glucose. Whole-body energy metabo-
lism was evaluated using a Comprehensive Lab Animal Monitoring System (CLAMS, Columbia Instruments). Mice were acclimated in
the metabolic chambers for 2 days before starting the experiment. CO2 and O2 levels were collected every 32 min for each mouse
over a period of 4 days. Movement and food intake are measured at regular intervals. For the analysis of mice during the development

Cell 151, 1319–1331, December 7, 2012 ª2012 Elsevier Inc. S1

of cancer cachexia, all mice were placed into the CLAMS system for four days for each time point and then returned to their cages in
between recording periods.

SUPPLEMENTAL REFERENCES

Kleiner, S., Mepani, R.J., Laznik, D., Ye, L., Jurczak, M.J., Jornayvaz, F.R., Estall, J.L., Chatterjee Bhowmick, D., Shulman, G.I., and Spiegelman, B.M. (2012).
Development of insulin resistance in mice lacking PGC-1a in adipose tissues. Proc. Natl. Acad. Sci. USA 109, 9635–9640.
Nair, K.S., Ford, G.C., Ekberg, K., Fernqvist-Forbes, E., and Wahren, J. (1995). Protein dynamics in whole body and in splanchnic and leg tissues in type I diabetic
patients. J. Clin. Invest. 95, 2926–2937.
Proctor, D.N., and Beck, K.C. (1996). Delay time adjustments to minimize errors in breath-by-breath measurement of Vo2 during exercise. J. Appl. Physiol. 81,
2495–2499.

S2 Cell 151, 1319–1331, December 7, 2012 ª2012 Elsevier Inc.

Figure S1. Identification of Three PGC-1a Isoforms by Targeted PCR, Related to Figure 1
(A) Schematic representation of the localization of primer target sequences used in order to amplify transcripts originating from the proximal or the alternative
promoter. Template cDNA originated from a mouse soleus cDNA library.
(B) PGC-1a mRNA expression analysis for the experiments described in (Figure 1D). Aliquots of the cells expressing each PGC-1a isoform or GFP alone as control
(described in D), were harvested and processed for RNA followed by analysis of gene expression by quantitative real-time PCR using primers directed at PGC-1a
exon 2, which is present in all isoforms.
(C) PGC-1a isoform coding sequences. Two exon 1 sequences (Ex1’ and 1’’) can be observed in transcripts originating from the alternative promoter, which are
different from exon1. Exons are shown alternating bold and regular font and the corresponding PGC-1a1 exons are indicated to the right of the sequence. (c) and
(p) denote complete or partial conservation, respectively. Red bars indicate new splicing options. Stop codon is indicated by yellow font. Bars depict mean values
and error bars represent standard deviation. *p < 0.05 between indicated group and control.

Cell 151, 1319–1331, December 7, 2012 ª2012 Elsevier Inc. S3

Figure S2. PGC-1a Isoform-Specific Expression Analysis, Related to Figure 2
(A) PGC-1a isoform protein levels in white quadriceps and red tibialis anterior (TA) muscles. Muscles were collected from C57bl/6 mice and tissues extracts were
prepared followed. PGC-1a isoform levels were determined by immunoprecipitation followed by immunoblotting.
(B) PGC-1a isoform protein expression in indicated tissues. Arrowheads indicate the different PGC-1a isoforms.
(C) PGC-1a isoform expression in myotubes treated with 10 mM forskolin for 6 hr compared to untreated controls.
(D) PGC-1a isoform expression in brown adipose tissue of mice at room temperature (RT), or exposed to 4 C for 6 hr.
(E) PGC-1a isoform protein levels in brown adipose tissue of mice at room temperature or exposed to 4 C for 24 hr. Flox, floxed PGC-1a allele; FKO, adipose
tissue-specific PGC-1a deletion, Cre, Adiponectin-Cre. Arrowheads indicate PGC-1a isoforms. *, indicates nonspecific bands. Bars depict mean values and error
bars represent standard deviation. *p < 0.05 between indicated group and control.

S4 Cell 151, 1319–1331, December 7, 2012 ª2012 Elsevier Inc.

Figure S3. PGC-1a4 Loss of Function Blunts Clenbuterol-Induced Myotube Hypertrophy, Related to Figure 3
(A) Average number of nuclei per myotube (total 300 myotubes per condition were scored).
(B) Fluorescence microscopy analysis of myotubes expressing GFP control, PGC-1a1 or PGC-1a4 and treated with 5 nM IGF1R inhibitor BMS-754807.
(C) PGC-1a4 localizes to the cell nucleus. Primary myotubes were infected as described in (A). 36 hr after infection cells were fixed and processed for immu-
nocytochemistry using a mouse monoclonal anti-PGC-1a antibody followed by an anti-mouse Cy3-conjugated secondary antibody. Nuclei were stained with
DAPI.
(D) PGC-1a1 and a4 expression levels in myotubes treated with 500 nM clenbuterol in the presence of PGC-1a4-specific or scrambled control shRNAs.
(E) Protein accumulation normalized by genomic DNA content. Experiments were conducted as described in (D) and cells were harvested for analysis of protein
accumulation.
(F) PGC-1a4-mediated myotube hypertrophy is independent of the ERRs. ERRa knockout myoblasts were differentiated in vitro and transduced with re-
combinant adenovirus encoding a GFP control or PGC-1a4. Thirty six hours after, myotubes were observed under a fluorescence microscope with a 10X
objective.
(G) Analysis of IGF1 and myostatin gene expression in ERRa-deficient myotubes. Experiments were performed as described in (A) and cells were harvested and
processed for RNA followed by analysis of gene expression by quantitative real-time PCR using primers specific to the indicated genes.
(H) PGC-1a4 does not coactivate ERRa activity on estrogen-response element (ERE) or ERRa promoter constructs. HEK293 cells were transfected with lucif-
erase (Luc) reporter plasmids and plasmids encoding for ERRa and/or PGC-1a1 or PGC-1a4 as indicated. Luciferase activity was assayed 36 hr later. Bars depict
mean values and error bars represent standard deviation. *p < 0.05 between indicated group and control. *,#, p < 0.05 between all groups.

Cell 151, 1319–1331, December 7, 2012 ª2012 Elsevier Inc. S5

Figure S4. Delivery of PGC-1a4 In Vivo Results in Skeletal Muscle Hypertrophy, Related to Figure 4
(A) Adenovirus-mediated delivery of PGC-1a4 to the gastrocnemius of SCID mice. Image shows a low magnification fluorescence microscopy image of a cross
section of the lower limb 7 days after one single injection of 2x1010 infectious particles.
(B) Average fiber cross sectional area of gastrocnemius muscle transduced with a GFP control or PGC-1a4 adenovirus.
(C) Gene expression analysis of muscle transduced with control GFP or PGC-1a4 adenovirus. One week after infection, animals were euthanized and the
gastrocnemius muscles were collected and processed for RNA followed by analysis of gene expression by quantitative real-time PCR using primers specific to
the indicated genes.
(D) Immunoblotting analysis of PGC-1a4 and phospho-S6 ribosomal protein (P-S6) levels in muscle transduced with GFP or PGC-1a4 adenovirus.
(E) Average fiber cross sectional area of tibialis anterior muscle after electroporation-mediated delivery of plasmids encoding PGC-1a4, NT-PGC-1a or control.
Each mouse received a control plasmid in one limb (pCI-neo), and in the contralateral limb the plasmid encoding PGC-1a4 or NT-PGC-1a.
(F) Changes in normalized muscle weight.
(G) Immunoblotting analysis of phospho-p70-S6 Kinase (P-S6K) levels in tibialis anterior muscle electroporated with pCIneo control plasmid or plasmid encoding
PGC-1a4. Bars depict mean values and error bars represent standard deviation. *p < 0.05 between indicated group and control.

S6 Cell 151, 1319–1331, December 7, 2012 ª2012 Elsevier Inc.

Figure S5. Suspension/Reloading-Induced Muscle Hypertrophy Induces PGC-1a4 Expression, Related to Figure 5
(A) Changes in normalized muscle mass with hindlimb suspension/reloading. N = 4 per group. Muscle wet weights normalized to tibia length.
(B) Analysis of PGC-1a proximal promoter activation by quantitative real-time PCR using primers directed at the 50 UTR region.
(C) Immunoblot analysis of PGC-1a4 protein accumulation during the suspension/reloading protocol. Arrowheads indicate PGC-1a4.
(D) Gene expression analysis of gastrocnemius/plantaris muscles. C57Bl/6 mice were divided into control (n = 4), 10 days hindlimb suspension (Suspension, n =
4), or 10 days hindlimb suspension plus 24 hr of reloading (Reloading, n = 4). The soleus muscles were harvested and processed for RNA followed by gene
expression analysis by quantitative real-time PCR using primers specific for the indicated genes.
(E) Gene expression analysis in the gastrocnemius muscle of PGC-1a4 transgenic mice (Myo-PGC-1a4). Bars depict mean values and error bars represent
standard error. *p < 0.05 between indicated group and control. *,#, p < 0.05 between all groups.

Cell 151, 1319–1331, December 7, 2012 ª2012 Elsevier Inc. S7

 A B
Wild-type
16 Wild-type Myo-PGC-1α4
Myo-PGC-1α4
Percentage of fibers (%)
14
12
10 *
8 *
6
4
2
0
50

0
0
10 0
12 0
15 0
17 0
20 0
22 0
25 0
27 0
30 0
32 0
35 0
37 0
40 0
>4 00
00
50
75
0
5
0
5
0
5
0
5
0
5
0
5
<2

Fiber size (μm2)

C D E
Wild-type
80 Wild-type
500 MCK-PGC-1α
Myo-PGC-1α4
% Positive Fibers

Wild-type

Glucose (mg/dl)
1500 400 Myo-PGC-1α4
Tissue weights (mg)

60 1000 Myo-PGC-1α4
* 250 300
40 *
200
200
150
20 * 100 100
50
0 0
0 Heart SpleenTestes Liver 0 20 40 60 90 120
IIb
I

IIa

IIx
C

Time (min)
yH

C
yH

yH

yH
M

F Wild-type
MCK-PGC-1α
Myo-PGC-1α4
5000
g/day/g body weight

5000
1.0 0.25
VO2 (ml/kg/hr)

VCO2 (ml/kg/hr)

*
4000 4000 *
Food Intake

0.8 0.20
3000 3000 0.6 0.15
RER

2000 2000 0.4 0.10

1000 1000 0.2 0.05
0 0 0.0 0.00

Figure S6. Functional Analysis of the Myo-PGC-1a4 Mouse, Related to Figure 6

(A) Fiber cross-sectional area (CSA) frequency distribution determined with the ImageJ software on sections from 6 mice.
(B) Hematoxylin and eosin staining of gastrocnemius cross-sections from wild-type and Myo-PGC-1a4 mice.
(C) Quantification of fiber type immunohistochemistry using antibodies against different myosin heavy chain (MyHC) types in the gastrocnemius muscles of wild-
type and Myo-PGC-1a4 mice.
(D) Tissue wet weight determination (n = 6 per group).
(E) Intraperitoneal glucose tolerance test of control and PGC-1a1 (MCK-PGC-1a) or PGC-1a4 (Myo-PGC-1a4) transgenic mice (n = 6 per group).
(F) Energy expenditure and food intake was measured for 72 hr in individually housed wild-type, and PGC-1a1 (MCK-PGC-1a) or PGC-1a4 (Myo-PGC-1a4)
muscle-specific transgenic mice. Bars depict mean values and error bars represent standard deviation. *p < 0.05 between indicated group and control. *,#,
p < 0.05 between all groups.

S8 Cell 151, 1319–1331, December 7, 2012 ª2012 Elsevier Inc.

 A B +LLC
Myo-
Wt Wt PGC-1 4

(Normalized to NTB control)

p SMAD2/3
0
SMAD2/3

% change in CSA -10 tubulin

4
*

(Normalized % IOD)
-20 *,#
3

p SMAD2/3
-30
2
-40
*
Wild-type Myo-PGC-1 4 1

C
LC
pe

LL
C

-ty

+L

4+
ild

pe
Wild-type

-ty

-1
C
ild

G
Wild-type+LLC

-P
W

yo
Myo-PGC-1 4
Relative mRNA Expression

M
10 Myo-PGC-1 4+LLC
*
8
*
6

4
*,# *,#
2

0
MURF1 Atrogin

D E Wild-type
Wild-type+LLC
Food intake (g/day/body weight)

15
Myo-PGC-1 4+LLC
Tumor Weight (g)

0.25

10 0.20

0.15

5 0.10

0.05

0 0.00
e

re
t

Wild-type Myo-PGC-1α4
on

Pr

at

ve
er
C

Se
od
M

Tumor Progression

Figure S7. Preservation of Muscle Mass in Myo-PGC-1a4 during Cancer Cachexia, Related to Figure 7
(A) Myofiber cross sectional area in wild-type and Myo-PGC-1a4 tumor bearing mice normalized to their own genotype non-tumor-bearing control.
(B) Phosphorylation status of SMAD2/3 in wild-type, wild-type + LLC tumor and Myo-PGC-1a4 + LLC tumor.
(C) Atrogin and MuRF1 mRNA expression wild-type and Myo-PGC-1a4 with or without LLC tumor.
(D) Tumor size in wild-type and Myo-PGC-1a4 mice after 28 days of inoculation.
(E) Food intake in wild-type, wild-type + LLC tumor and Myo-PGC-1a4 + LLC tumor.
All gene and protein expression normalized to wild-type non-tumor-bearing mice. Bars depict mean values and error bars represent standard error. *p < 0.05
between indicated group and its genotype control. #, p < 0.05 between indicated group and group receiving same treatment across genotypes. y, p < 0.05
between indicated group and wild-type mice.

Cell 151, 1319–1331, December 7, 2012 ª2012 Elsevier Inc. S9