peroxisomeproliferator-activatedreceptor coactivator-1 ... · alternative promoter usage in the...

Peroxisome Proliferator-activated Receptor � Coactivator-1 �Isoforms Selectively Regulate Multiple Splicing Events onTarget Genes*

Received for publication, November 20, 2015, and in revised form, May 19, 2016 Published, JBC Papers in Press, May 26, 2016, DOI 10.1074/jbc.M115.705822

Vicente Martínez-Redondo‡1, Paulo R. Jannig‡§2,3, Jorge C. Correia‡2, Duarte M. S. Ferreira‡1, Igor Cervenka‡,Jessica M. Lindvall¶, Indranil Sinha¶, Manizheh Izadi‡, Amanda T. Pettersson-Klein‡4, Leandro Z. Agudelo‡,Alfredo Gimenez-Cassina�, Patricia C. Brum§, Karin Dahlman-Wright¶, and Jorge L. Ruas‡5

From the ‡Department of Physiology and Pharmacology, Molecular and Cellular Exercise Physiology Unit and �Department ofMedical Biochemistry and Biophysics, Karolinska Institutet, SE-171 77 Stockholm, Sweden, §School of Physical Education andSport, University of São Paulo, 05508-030 São Paulo, Brazil, and ¶Department of Biosciences and Nutrition, Novum, KarolinskaInstitutet, SE-141 83 Huddinge, Sweden

Endurance and resistance exercise training induces specificand profound changes in the skeletal muscle transcriptome.Peroxisome proliferator-activated receptor � coactivator-1 �(PGC-1�) coactivators are not only among the genes differen-tially induced by distinct training methods, but they also partic-ipate in the ensuing signaling cascades that allow skeletal mus-cle to adapt to each type of exercise. Although endurancetraining preferentially induces PGC-1�1 expression, resistanceexercise activates the expression of PGC-1�2, -�3, and -�4.These three alternative PGC-1� isoforms lack the arginine/ser-ine-rich (RS) and RNA recognition motifs characteristic ofPGC-1�1. Discrete functions for PGC-1�1 and -�4 have beendescribed, but the biological role of PGC-1�2 and -�3 remainselusive. Here we show that different PGC-1� variants can affecttarget gene splicing through diverse mechanisms, includingalternative promoter usage. By analyzing the exon structure ofthe target transcripts for each PGC-1� isoform, we were able toidentify a large number of previously unknown PGC-1�2 and-�3 target genes and pathways in skeletal muscle. In particular,PGC-1�2 seems to mediate a decrease in the levels of choles-terol synthesis genes. Our results suggest that the conservationof the N-terminal activation and repression domains (and notthe RS/RNA recognition motif) is what determines the gene pro-grams and splicing options modulated by each PGC-1� isoform.By using skeletal muscle-specific transgenic mice for PGC-1�1and -�4, we could validate, in vivo, splicing events observed in invitro studies. These results show that alternative PGC-1� vari-ants can affect target gene expression both quantitatively andqualitatively and identify novel biological pathways under thecontrol of this system of coactivators.

PGC-1�6 coactivators are transcriptional regulators withimportant roles in skeletal muscle adaptation to physical exer-cise (1). Upon specific physiological signals, a single PGC-1�gene expresses different variants with discrete biological roles(2– 4). For example, in skeletal muscle, resistance exercise and�2-adrenergic stimulation activate a distal PGC-1� promoterthat controls the expression of alternative isoforms PGC-1�2,PGC-1�3, and PGC-1�4 (3, 5, 6), which undergo additionalalternative splicing of their primary transcript (Fig. 1A) (3).Induction of the canonical PGC-1�1 isoform in skeletal musclepromotes mitochondrial biogenesis and an oxidative pheno-type that mediates adaptive responses to endurance training(7). Although widely different in its structure, PGC-1�4 main-tains transcriptional activity and regulates distinct target genenetworks involved in resistance exercise-driven hypertrophy inskeletal muscle (3, 6). To date, the biological roles of theremaining alternative isoforms of PGC-1� are still unknown,and conventional 3� expression arrays have failed to identifyspecific gene networks targeted by PGC-1�2 and -�3 (3).

PGC-1�1 exhibits a dual role with regard to transcriptionalregulation of its target genes as it can coactivate transcriptionfactor-mediated gene expression but also modulate alternativesplicing of the nascent transcript (8). Co-transcriptional splic-ing of pre-mRNA is a common phenomenon (9) where mRNAediting factors such as serine/arginine (SR) proteins can dock tothe hyperphosphorylated C-terminal domain (CTD) of theRNA polymerase II to mature the nascent mRNA (10).Although this mechanism is not yet fully understood, it hasbeen shown that transcriptional coactivators can interactdirectly with the spliceosome and condition the splicingoptions that follow (11–13). In this sense, PGC-1�1 acts muchlike an SR protein due to two characteristic domains located inits own CTD: an arginine/serine-rich (RS) domain that medi-ates protein-protein interactions (with other SR proteins) and

* This work was supported in part by grants from the Swedish Research Coun-cil, the Strategic Research Programme in Diabetes at Karolinska Institutet,the Novo Nordisk Foundation (Denmark), the Swedish Diabetes Founda-tion, and the Malin and Lennart Phiipson Foundation (to J. L. R). Theauthors declare that they have no conflicts of interest with the contents ofthis article.

1 Supported in part by a postdoctoral fellowship from the Wenner-GrenFoundations (Sweden).

2 Both authors contributed equally to this work.3 Recipient of a Fundação de Amparo à Pesquisa do Estado de São Paulo

fellowship (Brazil).4 Supported in part by Svenska Sällskapet för Medicinsk Forskning (Sweden).5 To whom correspondence should be addressed. E-mail: [email protected].

6 The abbreviations used are: PGC-1�, peroxisome proliferator-activatedreceptor � coactivator 1-�; RRM, RNA recognition motif; CTD, C-terminaldomain; qRT-PCR, quantitative RT-PCR; Ddx27, DEAD box polypeptide 27;Osbpl1a, oxysterol-binding protein-like 1A; Ndrg4, n-myc downstreamregulated gene 4; IPA, Ingenuity Pathway Analysis; Mcm, minichromo-some maintenance; EAA, Exon Array Analyzer; Myo-PGC-1�4, myogeninpromoter-PGC-1�4; Mck-PGC-1�1, muscle creatine kinase promoter-PGC-1�1; PPAR, peroxisome proliferator-activated receptor.

crossmarkTHE JOURNAL OF BIOLOGICAL CHEMISTRY VOL. 291, NO. 29, pp. 15169 –15184, July 15, 2016

© 2016 by The American Society for Biochemistry and Molecular Biology, Inc. Published in the U.S.A.

JULY 15, 2016 • VOLUME 291 • NUMBER 29 JOURNAL OF BIOLOGICAL CHEMISTRY 15169

by guest on August 21, 2020

http://ww

w.jbc.org/

Dow

nloaded from

http://crossmark.crossref.org/dialog/?doi=10.1074/jbc.M115.705822&domain=pdf&date_stamp=2016-5-26

http://www.jbc.org/

an RNA recognition motif (RRM) that allows interaction withthe nascent mRNA (14). Deletion of any of those CTD regions issufficient to alter the co-transcriptional splicing activity ofPGC-1�1 on the expression of both an artificial minigenereporter (8) and endogenous genes (15).

One of the most striking structural features of non-canonicalPGC-1� isoforms (PGC-1�2, -�3, and -�4) is that all three iso-forms lack the CTD, including the RS and RRM domains (3)(Fig. 1A). This suggests that different PGC-1� variants couldhave selective effects on target gene splicing. Hence, in thisstudy, we have investigated whether the biological functions ofthe novel PGC-1� variants are mediated by quantitative and/orqualitative changes in target gene transcription produced bydistinct isoform expression. To this end, we have determinedthe occurrence of alternative splicing events under the controlof different PGC-1� coactivators in mouse primary myotubes.With this approach, we have identified several gene pathwaysregulated by PGC-1�2 and -�3. Our results show that the dif-ferent PGC-1� proteins promote both quantitative and quali-tative changes in gene expression. By using several complemen-tary bioinformatics tools, we identified a broad number of genesthat display differential splicing regulated by distinct PGC-1�isoforms. By designing targeted qRT-PCR strategies, we veri-fied some of the different alternative splicing events underPGC-1� control. These include the PGC-1�2- and -�3-depen-dent regulation of alternative 3�-UTR DEAD box polypeptide27 (Ddx27) isoform expression and PGC-�1- and -�4-medi-ated alternative promoter usage of oxysterol-binding protein-like 1A (Osbpl1a) and n-myc downstream regulated gene 4(Ndrg4). This study identifies a mechanism of action forPGC-1� isoforms and provides a framework for the analysis ofisoform-specific effects on gene expression.

Results

Alternative PGC-1� Isoforms Display Distinct Transcrip-tional Regulation and Protein Stability—The importance ofalternative promoter usage in the expression of PGC-1� iso-forms has been clearly demonstrated (3, 6, 16). In addition, wehave found that, in murine skeletal muscle, �2-adrenergic stim-ulation with clenbuterol results in differential PGC-1� pro-moter regulation as it increases the levels of alternative iso-forms but represses PGC-1�1 expression (Fig. 1B).

The specific transcriptional regulation of PGC-1� isoformsis accompanied by significant changes in their mRNA struc-ture. We hypothesized that differences in the C terminus of thealternative PGC-1� isoforms that include binding sites forsplicing factors might affect target gene splicing. To investigatethis hypothesis, we have performed global analysis of geneexpression under the control of each PGC-1� variant by usingan exon-based array designed for whole-transcriptome analysisat the gene and exon level. Each PGC-1� variant (or GFP ascontrol) was expressed in mouse primary myotubes (Fig. 1C)through the use of recombinant adenoviral vectors (3). Asreported previously (3), we observed that, even when normal-ized by mRNA expression, the different PGC-1� proteins accu-mulate at different levels (Fig. 1D). These differences inPGC-1� isoform protein accumulation are also observed whenassessing their basal levels in vivo in adult skeletal muscle (3).

Although PGC-1�1 has been previously reported to be a shortlived protein due to ubiquitin-proteasome degradation (17, 18),not much is known about the protein stability of the otherPGC-1� isoforms. To determine whether all PGC-1� variantsare targeted to the same degradation pathway, myotubesexpressing each of the PGC-1� isoforms were treated for 4 hwith the proteasome inhibitor MG132. All PGC-1� proteinsaccumulate (to different degrees) upon MG132 treatment (Fig.1E), indicating that they are all degraded through the ubiquitin-proteasome pathway. To better compare the protein degrada-tion dynamics between the different isoforms, we performed acycloheximide chase. Upon inhibition of protein synthesis, thedegradation patterns were strikingly different (Fig. 1F).Although PGC-1�2 and PGC-1�3 displayed a half-life similarto PGC-1�1 (�30 min; Fig. 1F and Ref. 19), PGC-1�4 showedgreater stability with a remarkably long half-life, similar to whatis described for yet another PGC-1� isoform, NT-PGC-1� (17).

We used the fact that the ectopically expressed PGC-1� iso-forms (Fig. 1G) are themselves the result of alternative splicing(3) to validate the ability of our system to detect alternativesplicing events. The probe set signal for exon 2, the only regioncommon to all the PGC-1� variants, was increased in all exper-imental conditions when compared with the GFP control (Fig.1, G and H, arrowhead “b”). Only PGC-1�1 expressionincreased the probe set signal for all the exons along thePGC-1� gene (Fig. 1H), including unique exon 1a (Fig. 1, G andH, arrowhead “a”), whereas exons specific to the alternativePGC-1� isoforms (namely exon 1b in PGC-1�2 and PGC-1�4and exon 1b� in PGC-1�3) were not detected by the array (Fig.1H). In the Geneview representation, we also observed theexpression pattern characteristic to the specific C terminus ofeach alternative PGC-1� isoform (Fig. 1, G and H, arrowheads“c” and “d”). We validated the expression of each PGC-1� iso-form by qRT-PCR using primer pairs specific to the differentexon combinations (Fig. 1I). By probing exons 3–5, we detectedPGC-1� transcripts with a conserved N terminus, namelyPGC-1�1 and PGC-1�4 (Fig. 1I), and complementarily, whenprobing exons 3– 8 we detected exclusively isoforms with dis-rupted activation and repression domains, PGC-1�2 and PGC-1�3 (Fig. 1I). Exogenous expression of PGC-1� isoformsresulted in mild changes in their endogenous levels. For exam-ple, PGC-1�2 expression decreases its own endogenous levelsbut increases PGC-1�3 (Fig. 1I). Additionally, endogenousPGC-1�2 was induced (50%) by PGC-1�1.

Analysis of Gene Networks under the Control of PGC-1�Isoforms—The workflow and thresholds established for the dis-crimination of differentially expressed genes and exons aredescribed in detail under “Experimental Procedures” and sum-marized in Fig. 2A. For clarity, we use the term “gene” whenreferring to the identity of a particular gene and “transcript”when referring to a specific variant of a given gene. Each arrayprobe set generates a signal specific for the 5�-UTR, exons, andthe 3�-UTR of a single gene. Principal component analysis ofthe variance distribution indicated good clustering for all thedata sets generated for the different PGC-1� isoforms (Fig. 2B).We detected significant changes in gene expression modulatedby each of the different PGC-1� variants (cutoff of 1.2-foldchange versus control GFP, p � 0.05) (Fig. 2C). Using Partek

PGC-1� Isoforms Selectively Regulate Target Gene Splicing

15170 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 291 • NUMBER 29 • JULY 15, 2016


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

AP

1801 403 565 632 676 752 797

Activation Repression

Activation

RS RRM

13 107 274 403 543

4 107 274 403 543

13 180 263

Activati ression

Activati ression

Repr

PGC-1α1

PGC-1α2

PGC-1α3

PGC-1α4

P PLLXXL

LLXXL

LLXXL

LLXXL

LLXXL

LLXXL

LXXLL

LXXLL

(379 AA)

(370 AA)

(266 AA)

Splicingfactors

Med 1

APP

1 2 3 45 6 7 8 109 11 12 13*PGC-1α1

1’ 2 3 7 8

*3’ UTR

PGC-1α2

1’’ 2 3 87

*3’ UTR

PGC-1α3

2 3 45 6 7 8 109 11 12 131’ PGC-1α4

a

b

*

dc2

5

8

Exons

11

5' 3'14

α2 α3 α4GFP α1a

bc

d

Transcript structure Transduced PGC-1α variants - Geneview (Partek)

exon 2 - common to all isoforms no probes(exon 1b, 1b’)

exon 1aG

prob

e se

t sig

nal i

nten

sity

H

0

5x105

1x106

1.5x106

2x106

0.5

1

1.5

2

0

1x104

2x104

3x104

4x104

5x104

0

200

400

600

800

1000

0

5x104

1x105

1.5x105

GFP PGC-1α1PGC-1α1 Exon 3 to 5 Exon 3 to 8 PGC-1α2 (endogenous) PGC-1α3 (endogenous)

* *

*

*

* *

**

PGC-1α2 PGC-1α3 PGC-1α4

rela

tive

mR

NA

expr

essi

on

(arb

itrat

y un

its)

rela

tive

mR

NA

expr

essi

on

(arb

itrat

y un

its)

rela

tive

mR

NA

expr

essi

on

(arb

itrat

y un

its)

rela

tive

mR

NA

expr

essi

on

(arb

itrat

y un

its)

rela

tive

mR

NA

expr

essi

on

(arb

itrat

y un

its)

I

0

B

0

1

2

3

4

0

1

2

3

4

0

10

20

30

40

0.0

0.5

1.0

1.5

0

10

20

30

rela

tive

mR

NA

expr

essi

on

fold

cha

nge

over

con

trol

Total PGC-1α PGC-1α1 PGC-1α2 PGC-1α3 PGC-1α4

PBS Clenbuterol

*

*

* * *

0

1000

2000

3000Total PGC-1α

GFPPGC-1α1

PGC-1α2

PGC-1α3

PGC-1α4

100 kDa75 kDa

50 kDa

37 kDa

C

*

*

**re

lativ

e m

RN

A ex

pres

sion

(a

rbitr

aty

units

)

GFP

PGC-1α1

PGC-1α2

PGC-1α3

PGC-1α4

D

DMSO MG-132

PGC-1α2

PGC-1α3

PGC-1α4

0' 20' 240' 40' 60' 120'

PGC-1α1

Cycloheximide E Cycloheximide chase

0 20 40 60 120 2400.00

0.25

0.50

0.75

1.00

1.25

min

arbi

trary

uni

ts

PGC-1α1:PGC-1α2:PGC-1α3:

PGC-1α4:

Half-life

240 min*

27 min29 min28 min

F

FIGURE 1. PGC-1� isoform protein dynamics. A, Modular structure of PGC-1� isoforms (amino acid numbers are relative to PGC-1�1). Main domains depicted:activation, repression, RS, and RRM. Leucine-rich motifs (LLXXL and LXXLL) indicate interaction sites for transcription factors and other coactivators. “P”indicates phosphorylation sites for p38 MAPK. C-terminal RS and RRM domains interact with splicing factors and the mediator complex via Med1. AA, aminoacids; PP, proximal promoter; AP, alternative promoter. B, �-adrenergic stimulation regulates PGC-1� isoform expression in skeletal muscle. Gene expressionin the gastrocnemius muscle was analyzed by qRT-PCR. PBS control was compared with clenbuterol (i.p. 2 mg/kg for 16 h). Error bars represent S.E. *, p � 0.05versus PBS. C, expression of PGC-1� isoforms in mouse primary myotubes. Differentiated myotubes were transduced with recombinant adenovirus to expressGFP alone (control) or GFP and each PGC-1� isoform from independent promoters. Error bars represent S.D. D, protein immunoblot for myotubes transducedwith PGC-1� isoforms. E, PGC-1� protein dynamics. 4 h of 10 �M MG132 treatment stabilizes all the PGC-1� isoforms. F, time course for cycloheximide chase andhalf-life determination for all transduced PGC-1� isoforms. *, at least 240 min as this was the end point of the chase. G, exon/intron structure of PGC-1� isoforms.H, probe signal intensity for transduced PGC-1� isoforms obtained in exon arrays. Arrows and letters indicate regions of interest: a, exon 1a specific for PGC-1�1;b, exon 2 common to all isoforms; c, exon 7-8 region present in PGC-1�1/�2/�3; d, 3� regions only present in PGC-1�1. * indicates stop codon for each isoform,crossed exons indicate alternative splicing-incomplete exon. I, qRT-PCR validation of the expression of PGC-1� isoforms ectopically expressed in myotubes.Error bars represent S.D. *, p � 0.05 versus GFP.




http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

software, we identified 1279 annotated genes (from a total of1592 regulated genes) modulated by PGC-1�1. Remarkably, weidentified a similar number of genes targeted by PGC-1�2(1638 annotated from a total of 1980) and PGC-1�3 (1279 genesannotated from a total of 1572). PGC-1�4 modulated the larg-est number of genes (2262 annotated from a total of 2606) (Fig.2C). The discrepancy between these results and the previouslypublished 3� expression array data is likely due to the fact thatexon-based arrays achieve higher sensitivity by providing awider coverage of the transcriptome (1.2 million probe sets inAffymetrix Mouse Exon 1.0 ST array versus 45,000 probes inAffymetrix GeneChip Mouse Genome 430 2.0 array), probablyreflecting the detection of a higher number of transcript vari-ants of the targeted genes.

Critical N-terminal motifs that fine-tune PGC-1�1 tran-scriptional activity (1) are spliced out in PGC-1�2, -�3, and -�4.To investigate how the inclusion or exclusion of these domainsimpacts the transcriptional activity of PGC-1�, we comparedthe gene sets regulated by each PGC-1� isoform. This analysistook into account whether genes regulated by different PGC-1�isoforms were activated or repressed by both variants (co-regu-lated) or activated by one but repressed by the other (contra-regulated). The highest number of co-regulated genes wasfound between isoforms PGC-1�1 and -�4 (650) and PGC-1�2and -�3 (421) (Fig. 2D). Contrarily, PGC-1�2 and -�4 proved tocontra-regulate the largest gene set (486) (Fig. 2D). However,

even excluding co-/contra-regulated genes, we could still iden-tify a large number of target genes specific to each PGC-1�isoform (i.e. PGC-1�1, 426; PGC-1�2, 729; PGC-1�3, 566;PGC-1�4, 1063 genes, respectively; data not shown).

PGC-1�2 and PGC-1�3, which lack part of the activationand repression domains (amino acids 107–274 of the canonicalisoform), produce a distinct cluster from PGC-1�4, which inturn showed higher similarities with the genes targeted byPGC-1�1. This indicates that the transcriptional activity of thePGC-1� isoforms is dictated by the conservation of the N-ter-minal activation domain rather than the presence or absence ofthe RS/RRM motifs.

Identification of Gene Pathways Specifically Regulated byEach PGC-1� Isoform—To better understand which biologicalfunctions are associated with the transcriptional programs con-trolled by each PGC-1� isoform, we performed a comprehen-sive pathway and network analysis of each target gene set usingthe Ingenuity Pathway Analysis platform (Fig. 3A). We thenvalidated by qRT-PCR selected genes from each of the identi-fied pathways (Fig. 3, B–H).

Analysis of pathways regulated by PGC-1�1 in myotubesconfirmed its well known involvement in transcriptional pro-grams associated with mitochondrial biology (mitochondrialdysfunction, tricarboxylic acid cycle, acetyl-CoA biosynthesis,and branched-chain �-keto acid dehydrogenase complex) aswell as the characteristic increase in oxidative metabolism

2.01.81.61.41.21.00.80.6

0.0

0.40.2

-0.2-0.4-0.6-0.8-1.0-1.2-1.4-1.6-1.8-2.0

GFPPGC-1α1

PGC-1α2

PGC-1α3

PGC-1α4

A

C

Gene Level Analysis

Venn Diagram

Exon Level Analysis

ExperimentalSkeletal muscle

Ingenuity Pathway Analysis

Event cathegorization

AltAnalyze

Alt.Spl.Attribute≤0.0001Partek

qRT-PCRHit validation

Exon Array AnalyzerSI > 1 & -1 > SI

Mouse 1.0 ST Array

α1

1032 58 1032

α3

189

967 1326

117

195

α1 α2

619 160311

α1 α4

650

α2

1201

421

16 842

α3

1038 1661

115

486

α2 α4

883 1866

140

256

α4α3

co-regulated contra-regulated

D

PGC-1 1 PGC-1 2 PGC-1 3 PGC-1 40

500

1000

1500

2000

2500RepressedInduced

n° g

enes

GFPPGC-1α1PGC-1α2PGC-1α3PGC-1α4

PCAB

FIGURE 2. Global gene expression analysis by exon arrays in myotubes expressing each PGC-1� isoform. A, workflow for the exon array gene/isoformexpression, molecular pathway analysis, and alternative splicing determination. Splicing hits were double filtered in Partek (alt.splice attribute �0.0001) andEAA (�1 � splicing index � 1). B, principal component analysis (PCA) of the exon array data sets. C, heat map for PGC-1� isoform target gene expression inmyotubes transduced plus number of differentially expressed genes (1.2-fold change, p � 0.05) by each PGC-1� isoform versus GFP. Red, induced; green,repressed. D, Venn diagram for gene regulation comparison between each PGC-1� isoform. The union area is divided into two subareas to indicate co-regu-lated versus contra-regulated genes.




http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

(mitochondrial L-carnitine shuttle) (Fig. 3A). Among the genesthat defined the pathways targeted by PGC-1�1, we validatedby qRT-PCR previously known targets such as estrogen-relatedreceptor � (Esrra), carnitine palmitoyltransferase 1b (Cpt1b),and cytochrome c (Cyc) (Fig. 3B). Some of these targets are alsoinduced by PGC-1�4, although cytochrome c remains a uniquetarget for PGC-1�1.

Among the PGC-1� isoforms, PGC-1�2 and PGC-1�3 sharethe highest sequence similarity, only differing in the first aminoacids. Accordingly, PGC-1�2 and PGC-1�3 co-regulated�30% of their individual target genes. However, because of thedifference in the magnitude of effects on each gene set, therewas not much overlap between the pathways regulated by bothisoforms.

Cyc

0

1

2

3

4Cpt1b

0

2

4

6

8

10 Esrra

fold

cha

nge

over

con

trol Hmgcs1

0.0

0.5

1.0

1.5

2.0

2.5 Nsdhl Sc5d

0.0

0.5

1.0

1.5

2.0 Msmo1

Baiap2

0

1

2

3

4 Myl2Myh7

0.0

0.5

1.0

1.5

2.0 Egfr Mcm3

0

1

2

3 Mcm5Mcm4 Mcm6

0

1

2

3

4

5

Mmp15

0.0

0.5

1.0

1.5Src1

0.0

0.5

1.0

1.5

2.0

2.5Vegfa

0

1

2

3

4 Vegfd Phd2

Map2k6

0.00.51.01.52.02.53.03.54.04.5 Smad7

Ptgs2

0.0

0.5

1.0

1.5

2.0 Pik3r1Mapkapk2

B Bioenergetics D TGFβ-2 signalingC Cholesterol metabolism

E Epithelial Adherens Junctions F Minichromosomal maintenance

G HIF signaling H IL-17 signaling

*

** *

*

* * ** * *

*

* *

** *

*

*

*

* **

*

*

*

* *

*

p=0,14

* * **

p=0,07

p=0,14

p=0,07

* * **

* ** *

p=0,06

p=0,06

*

*

*

n.s.

*


Top 10 IPA Pathways-log(p-value)

Mitochondrial DysfunctionTCA Cycle II

Valine Degradation I

Acetyl-Coa Biosynthesis I Stellate Cell Activation

Mitochondrial L-Carnitine Shuttle

2-ketoglutarate Dehydrogenase Complex

2-oxobutanoate Degradation Ascorbate Recycling

Branched-chain α-keto acid Dehydrogenase Complex

Prostanoid Biosynthesis

0 2 4 6 8 10 12 14 16 18 0.0 0.5 1.0 1.5 2.0 2.5 3.0

Cholesterol Biosynthesis ICholesterol Biosynthesis II Cholesterol Biosynthesis III

Superpathway of Cholesterol BiosynthesisMacrophages/Endothelial Cells in RA

Methylmalonyl Pathway

Oleate Biosynthesis I Wnt/ β-catening Signaling

IL-10 SignalingTGF-β Signaling

0.0 0.5 1.0 1.5 2.0

Epithelial Adherens Junction SignalingAssembly of RNA Polymerase III Complex

Regulation of Cellular Mechanics by Calpain ProteaseArsenate Detoxification I (Glutaredoxin)

Cardiomyocite Differentiation via BMP ReceptorsOvarian Cancer Signaling

Estrogen-mediated S-phase EntryRole of p14/p19ARF in Tumor Supression

RhoA SignalingAssembly of RNA Polymerase I Complex

PGC-1α1

IL-17 SignalingColorectal Cancer Signaling

HIF-1α SignalingPancreatic Adenocarcinoma Signaling

NRF2 Oxidative Stress ResponseInhibition of Angiogenesis

AMPK SignalingIL-22 Signaling

IL-17A in Arthritisp53 Signaling

PGC-1α2

PGC-1α3 PGC-1α40.0 0.5 1.0 1.5 2.0 2.5

PGC-1α isoform Exon Array Same PGC-1α isoform on Gene Array

A

rela

tive

mR

NA

expr

essi

on

fold

cha

nge

over

con

trol

rela

tive

mR

NA

expr

essi

on

fold

cha

nge

over

con

trol

rela

tive

mR

NA

expr

essi

on

fold

cha

nge

over

con

trol

rela

tive

mR

NA

expr

essi

on

fold

cha

nge

over

con

trol

rela

tive

mR

NA

expr

essi

on

fold

cha

nge

over

con

trol

rela

tive

mR

NA

expr

essi

on

fold

cha

nge

over

con

trol

rela

tive

mR

NA

expr

essi

on

fold

cha

nge

over

con

trol

rela

tive

mR

NA

expr

essi

on

fold

cha

nge

over

con

trol

rela

tive

mR

NA

expr

essi

on

fold

cha

nge

over

con

trol

rela

tive

mR

NA

expr

essi

on

fold

cha

nge

over

con

trol

rela

tive

mR

NA

expr

essi

on

fold

cha

nge

over

con

trol

FIGURE 3. Target gene regulation by PGC-1� isoforms in myotubes. A, top pathways from the Ingenuity Pathway Analysis platform. Gray bars indicatescores for the same pathways from a previous data set (3). B–H, qRT-PCR validation for genes associated with the molecular pathways identified by IPA. Errorbars represent S.D. *, p � 0.05 versus GFP, n.s., non-significant statistical differences versus GFP. TCA, tricarboxylic acid; BMP, bone morphogenetic protein; RA,rheumatoid arthritis; ARF, alternative reading frame; AMPK, AMP kinase; HIF-1�, hypoxia-inducible factor-1�.




http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

Interestingly, genes regulated by PGC-1�2 in myotubesgrouped largely under the cholesterol biosynthesis pathway(Fig. 3A). Based on this prediction, we next validated by qRT-PCR that PGC-1�2 expression leads to a significant reductionin the mRNA levels of several upstream genes of the superpath-way of cholesterol biosynthesis such as methylsterol monooxy-genase 1 (Msmo1) and NAD(P)-dependent steroid dehydroge-nase-like (Nsdhl). Some other members of the cholesterolsynthesis pathway showed a non-significant trend for reducedexpression as is the case for 3-hydroxy-3-methylglutaryl-CoAsynthase 1 (Hmgcs1) and the sterol-C5-desaturase (Sc5d) (Fig.3C). Several genes involved in the regulation of inflammatoryprocesses were also among the top scoring PGC-1�2-regulatedpathways, including transforming growth factor � (TGF-�) sig-naling. In this case, a consistent up-regulation of the expressionof Smad7, the intracellular antagonist of TGF�2 in musclegrowth and maintenance, was validated by qRT-PCR (Fig. 3D).

The pathways identified for PGC-1�3 were suggestive ofdevelopment and cell cycle control. We validated the expres-sion of targets involved in the epithelial adherens junction sig-naling pathway (Fig. 3A) and confirmed the reductions inmRNA levels of myosin heavy chain 7 (Myh7), myosin lightchain (Myl2), and epidermal growth factor receptor (Egfr) plusa clear trend in the induction of brain-specific angiogenesisinhibitor 1-associated protein 2 (Baiap2) (Fig. 3E). Althoughthe Ingenuity Pathway Analysis (IPA) nomenclature of severalpathways tends to be very general, epithelial adherens junctionsignaling is also one of the top canonical pathways differentiallyregulated in both an integrative study of Duchenne muscle dys-trophy in humans (20) and a comprehensive multilevel omicsanalysis in dystrophy-deficient mice (21).

By manual curation of each gene list, we found that theminichromosome maintenance family of genes (Mcm3– 6) wasconsistently induced by both PGC-1�2 and PGC-1�3 (Fig. 3F).Proteins of the Mcm family configure the prereplication com-plex necessary for eukaryotic DNA replication (22). Contrarily,PGC-1�1 or PGC-1�4 expression reduced Mcm3– 6 levels(statistically significant only in the case of PGC-1�1 but with aclear trend for PGC-1�4), an effect that has also been observedupon expression of PGC-1� (23). As shown in Fig. 2D, there area significant number of genes contra-regulated by PGC-1�1versus PGC-1�2 and -�3. In agreement, many PGC-1�1 targetgenes were either unchanged or slightly repressed by PGC-1�2(Esrra, tricarboxylic acid cycle, and ATP synthesis genes). Thisanalysis did not predict any major PGC-1�2 effects on genesrelated to energy metabolism or mitochondrial biogenesis/function (Fig. 3A).

PGC-1�4 regulates muscle mass and hypoxia-driven angio-genesis in skeletal muscle (3, 24). PGC-1�4 and the structurallyrelated NT-PGC-1� (2) are able to induce Vegfa expression andan angiogenic program independently of hypoxia-induciblefactor-1� (24). Interestingly, in addition to observing that PGC-1�4 was able to induce Vegfa expression under the tested con-ditions (Fig. 3G), our results suggested PGC-1�4 as a negativeregulator of hypoxia-inducible factor-1� signaling. Indeed, sev-eral hypoxia-inducible factor-1� targets such as matrix metal-lopeptidase 15 (Mmp15) or prolyl hydroxylase domain-con-taining protein 2 (Phd2) showed reduced levels upon PGC-1�4

expression (Fig. 3G). Our results further identify new genepathways under PGC-1�4 control in myotubes, includinginterleukin 17 (IL17) and p53 signaling. Of the suggested IL-17signaling-associated genes targeted by PGC-1�4, we validatedby qRT-PCR the increases in the expression of Mapk26 and thep85 member phosphoinositide 3-kinase regulatory subunit 1(Pik3r1) and a reduction in the levels of prostaglandin-en-doperoxidase synthase 2 (Ptgs2) (Fig. 3H).

Because PGC-1� isoforms seem to be differentially regulatednot only at the transcriptional level (Fig. 1B) but also in terms ofprotein accumulation (Fig. 1D), we decided to validate targetgene specificity upon increasing expression of each PGC-1�variant (Fig. 4A). This strategy resulted in greater accumulationof PGC-1�1, -�2, and -�3 but not of PGC-1�4 protein (Fig. 4B).Subsequently, we could observe that the expression levels ofPGC-1�2 and PGC-1�3 target genes (e.g. Mcm3, Mcm4, andMcm6) remained unaffected even in the presence of higherPGC-1�1 levels (Fig. 4C). This target gene specificity was alsoobserved for genes that show reduced expression in the pres-ence of PGC-1�2 and PGC-1�3 (e.g. Msmo1, Nsdhl, and Myl2)(Fig. 4C). The opposite is also true, and specific targets for PGC-1�1 and PGC-1�4 (e.g. Cyc and Vegfa) remained exclusive uponincreased levels of PGC-1�2 and -�3 (Fig. 4C).

Global Analysis of Target Transcript Structure Reveals NovelGene Networks Regulated by PGC-1� Isoforms—To understandwhether our results are fundamentally different from the pre-viously published PGC-1� variant target gene data (3), we com-pared both gene sets. To this end, we reanalyzed our data sideby side with the previously published 3� gene array (3) to com-pare the top 10 IPA gene pathways identified for each PGC-1�isoform (Fig. 3A). Both gene expression analysis strategies wereequally efficient in the identification of PGC-1�1-regulatedgenes, which were also the most similar when compared withthe other isoforms. Accordingly, the top 10 canonical pathwaysand functions for PGC-1�1 obtained with the exon arraymatched strongly with those obtained for the previously pub-lished 3� array (3) (Fig. 3A). However, this was not true for theother PGC-1� isoforms, in particular for PGC-1�2 and -�3. Ofnote, canonical pathways with stronger statistical significance(i.e. higher �log(p value)) were more reproducible through dif-ferent arrays as is the case of isoforms of PGC-1�1. Remarkably,although the previous Affymetrix gene array (3) reported asmall number of genes targeted by PGC-1�2 and PGC-1�3(under 100), our current strategy identified well over 1000genes regulated by either coactivator (Fig. 2C). The most prom-inent functions associated with PGC-1�4 in the previous studywere energy production, lipid metabolism, and skeletal muscledevelopment. However, the top canonical pathways we foundin the present study are represented by several moleculesrelated to cell proliferation and survival (p53, cyclin D1, andseveral MAPKs) and the inflammatory response (IL-17 recep-tors, prostaglandin-endoperoxide synthase 2, and interleukin6) that were annotated under the names IL-17 signaling, colo-rectal cancer signaling, hypoxia-inducible factor-1� signaling,pancreatic adenocarcinoma signaling, and p53 signaling. In thiscase and despite the significant number of PGC-1�4 targetgenes identified with both gene expression profiling strategies,we did not observe a significant overlap between them. Our




http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

GFPPGC-1α1 +

PGC-1α1 ++

PGC-1α2 +

PGC-1α2 ++

PGC-1α3 +

PGC-1α3 ++

PGC-1α4 +

PGC-1α4 ++

0

1000

2000

3000

4000

0.0

Total PGC-1αfo

ld c

hang

e ov

er c

ontro

l

fold

cha

nge

over

Tite

r x1

0.0

0.5

1.0

1.5

2.0

fold

cha

nge

over

Tite

r x1

0.0

0.5

1.0

1.5

2.0

fold

cha

nge

over

Tite

r x1

0.0

0.5

1.0

1.5

2.0

2.5

fold

cha

nge

over

Tite

r x1

0.0

0.5

1.0

1.5PGC-1α1

PGC-1α2

PGC-1α4

PGC-1α3

Titer x1 Titer x2

0

1

2

3 Cyc

fold

cha

nge

over

con

trol * *

*

0

1

2

3

4

5 Vegfa

fold

cha

nge

over

con

trol

**

0

1

2

3

4

5 Mcm3

fold

cha

nge

over

con

trol

* *

** *

*

0.5

1.0

1.5

2.0

2.5 Mcm4

fold

cha

nge

over

con

trol

* *

** *

*

0

1

2

3 Mcm6

fold

cha

nge

over

con

trol

*

*

**

*

0.0

0.5

1.0

1.5 Msmo1

fold

cha

nge

over

con

trol

*

0.0

0.5

1.0

1.5

2.0 Nsdhl

fold

cha

nge

over

con

trol

* *

0

1

2

3 Myl2fo

ld c

hang

e ov

er c

ontro

l

* * * *

PGC-1α1 PGC-1α3 PGC-1α4PGC-1α2

* * *

A B

C

GFPPGC-1α1 +

PGC-1α1 ++

PGC-1α2 +

PGC-1α2 ++

PGC-1α3 +

PGC-1α3 ++

PGC-1α4 +

PGC-1α4 ++

20 40 60 80

1000

800

600

400

0

200

Glucose 5 mM

OC

R (p

mol

/min

)/60.

000

cell

minutes

Oligomycin FCCP Antimycin


minutes20 40 60 80

2000

1500

1000

500

0

Pyruvate 1 mM

OC

R (p

mol

/min

)/60.

000

cell

Oligomycin FCCP Antimycin


**

*

F

G

α1 vs G

FP

α2 vs G

FP

α3 vs G

FP

α4 vs G

FP0

50

100

150

200

250

α1 vs α

2

α1 vs α

3

α1 vs α

4

α2 vs α

3

α2 vs α

4

α3 vs α

40

200

400

600

800

1000

0

50

100

150

200

α1 vs G

FP

α2 vs G

FP

α3 vs G

FP

α4 vs G

FP0

200

400

600

800

α1 vs α

2

α1 vs α

3

α1 vs α

4

α2 vs α

3

α2 vs α

4

α3 vs α

4

n° of alternative spliced genes

n° of differentially expressed exons

Alt N-ter

Alt prom

Bleeding Exon

Cassette Exon

Exon Region Exclusion

Intron Retention

Mutual Exon Exclusion

Alt3'

Alt5'

Alt C-ter

α3 vs α4α2 vs α4α2 vs α3

3%

8%

17%1%1%18%27%

17%8%

1% 4%

8%

18%

3%1%16%

30%

10%9%

1% 4%

7%

18%

3%1%16%

34%

9%7%

1%

Alternative splice events categorization (AltAnalyze Software)

α1 vs α4α1 vs α3α1 vs α2

5%

10%

15%2%1%15%31%

9%11%

1% 2%

9%14%

1%2%

14%36%

8%13%

6%

8%

16%2%0,4%17%30%

11%9%

1%

H

D

E

FIGURE 4. Global analysis of alternative splicing in myotubes expressing PGC-1� isoforms. A, qRT-PCR analysis of total PGC-1� (exon 2) on myotubestransduced with increasing virus titer (�, single titer; ��, double titer). Error bars represent S.D. B, protein levels of distinct PGC-1� variants upon increasingvirus titer plus quantification (ImageLab) normalized by the lower titer in each experiment. Error bars represent S.E. *, p � 0.05 versus titer �1. C, qRT-PCRanalysis of target gene specificity for the different PGC-1� isoforms upon increasing protein expression (�, single titer; ��, double titer). Error bars representS.D. *, p � 0.05 versus GFP. D, extracellular flux analysis (Seahorse) of myotubes expressing each PGC-1� isoform in medium supplemented with 1 mM pyruvate.Error bars represent S.D. *, p � 0.05 versus GFP control. E, extracellular flux analysis of myotubes expressing each PGC-1� isoform on medium supplementedwith 5 mM glucose. Error bars represent S.D. F, left, number of differentially spliced genes between GFP control and each PGC-1� isoform. Right, number ofdifferentially spliced genes between PGC-1� isoforms. G, left, number of differentially expressed exons between GFP control and each PGC-1� isoform. Right,number of differentially expressed exons between PGC-1� isoforms. H, splicing events categorization by AltAnalyze: Alt3�, alternative 3� splicing; Alt5�,alternative 5� splicing; Alt C-ter, alternative C-terminal splicing; Alt N-ter, alternative N-terminal splicing; Alt prom, alternative promoter. FCCP, carbonyl cyanide-4-(trifluoromethoxy)phenylhydrazone; OCR, oxygen consumption rate.




http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

results indicate that the use of gene expression profiling withexon level resolution was necessary to uncover several newpathways under PGC-1� control. This was particularly impor-tant to identify targets for the non-canonical PGC-1� variants.

Expression of PGC-1� Isoforms Modulates Oxygen Consump-tion Rates in Primary Muscle Cells—Because our analysis ofPGC-1� isoform target genes indicated effects on severalimportant biochemical pathways (Fig. 3A), we next evaluatedtheir impact on cellular respiration by performing extracellularflux analysis. This allowed us to measure the oxygen consump-tion rate in real time on intact myotubes transduced with eachPGC-1� variant. We observed that when using 1 mM pyruvateas energy substrate all PGC-1� isoforms significantly increased(40 – 60%) the basal oxygen consumption rate of cultured myo-tubes (Fig. 4D). Treatment with the ATP synthase inhibitoroligomycin revealed that all PGC-1�-expressing cultures dis-played a higher fraction of respiration coupled to ATP produc-tion compared with controls (Fig. 4D). However, upon additionof the uncoupling ionophore carbonyl cyanide-4-(trifluorome-thoxy)phenylhydrazone, myotubes expressing PGC-1�1, PGC-1�2, and PGC-1�3 showed higher maximal respiratory capac-ity compared with PGC-1�4 and GFP, although only PGC-1�2and PGC-1�3 showed significant statistical differences (Fig.4D). When the medium was supplemented with 5 mM glucose,we observed no significant changes in the oxygen consumptionrate of the cultured myotubes (Fig. 4E). Together with the geneexpression data, these results indicate that the differentPGC-1� variants can affect cellular metabolism through dis-tinct mechanisms.

Expression of Each PGC-1� Isoform in Myotubes SelectivelyChanges the Splicing Landscape of the Cell—Our results clearlyshow that several PGC-1� target genes are only revealed whentaking into consideration the exon composition of the targettranscript. This indicates that coactivation by different PGC-1�proteins affects target gene splicing. Furthermore, PGC-1�2,-�3, and -�4 lack the C-terminal RS/RRM domains present inPGC-1�1, which could help explain why their target gene setsare the most different when comparing our results with thepreviously published array (3). Our next aim was therefore toinvestigate how these domains affect the expression of specifictarget gene variants. To identify and visualize alternative splic-ing events under the control of the different PGC-1� variants,we analyzed our exon array data using two complementary soft-ware platforms as indicated in Fig. 2A. Only splicing hits recog-nized by both platforms were further considered. Probably dueto the elevated mRNA levels of ectopically expressed PGC-1�variants, Ppargc1a was identified as the gene with the highestsplicing score (Partek; alt.splice p value � 0). In addition, whenwe compared the PGC-1� mRNA profiles in PGC-1�1- and-�4-expressing myotubes, Exon Array Analyzer (EAA) indi-cated high probability for 5� splicing (splicing index � 2). Whencomparing the transcript structure of PGC-1�2 or PGC-1�3with PGC-1�4, we could identify several splicing events classi-fied as uncategorized. This is most likely due to the incompleteannotation of these transcripts in the available databases.

Each of the 10 possible comparisons between PGC-1� iso-forms or controls yielded a substantial number of differentiallyspliced genes and expressed exons (Fig. 4, F and G). In fact,

expressing either PGC-1� variant in myotubes changes thesplicing pattern of over 100 genes when compared with con-trols. This number does not change much when comparingsplicing events under PGC-1�1 control versus those regulatedby PGC-1�4 (177) and PGC-1�2 versus PGC-1�3 (180). Inaddition, it is interesting that the number of “alternativelyspliced genes” (Fig. 4F) and “differentially expressed exons”(Fig. 4G) between PGC-1�1 versus PGC-1�4 and PGC-1�2 ver-sus PGC-1�3 was almost the same. This suggests that foralmost all the alternatively spliced genes within those specificcomparisons the resulting isoforms will mainly diverge in onedifferentially expressed exon. However, for all the other com-parisons, especially for PGC-1�2 versus PGC-1�4, the ratio wasclose to 1.5, indicating larger differences in target transcriptstructure, which in some cases will diverge in more than oneexon. Similar to what we observed on the gene level analysis, wefound larger differences in target gene splicing events whencomparing the exon expression profile controlled by PGC-1�2or PGC-1�3 versus PGC-1�4 (967 and 550, respectively).Because several of these events can map to the same gene, thenumber of alternatively spliced genes was lower than that ofdifferentially expressed exons, although both have the sameoverall distribution profile (Fig. 4, F and G).

We next systematically categorized the splicing events underthe control of each PGC-1� variant (please see “Isoform LevelAnalysis” under “Experimental Procedures”). Fig. 4C summa-rizes the eight most representative events found for each com-parison. Despite the large differences in differentially expressedexons observed between groups, the relative frequency of thecategorized events did not change in any of the situations com-pared. The most common splicing events observed in all groupswere those of “cassette exon” (27–36%), “alternative C-termi-nal,” and “bleeding exon” (both 14 –18%) (Fig. 4H).

PGC-1�2 and -�3 Promote Expression of a Truncated Ddx27Form with an Alternative 3�-UTR—Ddx27, a member of theDEAD box family of putative RNA helicases related to ribo-some and spliceosome assembly, showed a highly significantsplicing probability in both Partek and EAA platforms. Welocated the existence of a discrete splicing event on Ddx27 for aunique probe set when PGC-1�2 or PGC-1�3 was expressed(Fig. 5A). Analysis of the Ddx27 transcripts annotated onEnsembl indicated that the differentially expressed exonmaps to an alternative 3�-UTR found in isoform Ddx27-002(ENSMUST00000150571, 292 amino acids) (Fig. 5B). To con-firm these splicing events, we performed qRT-PCR validationusing oligonucleotides specific for constitutive and alternativeexons (see Table 1).

To analyze Ddx27 isoform expression, we targeted constitu-tive exons 8 and 9 for the detection of canonical full-lengthisoform Ddx27-001 (ENSMUST00000018143, 760 aminoacids) and alternative exon 8b for the detection of isoformDdx27-002. As expected, the relative levels of Ddx27-001proved to be independent of PGC-1� variant expression (Fig.5C). However, myotubes expressing PGC-1�2 or PGC-1�3 hadhigher levels of the Ddx27-002 isoform, which contains thealternative 3�-UTR (Fig. 5C) and a premature stop codon. Forthis reason, Ddx27-002 lacks the C-terminal helicase domain,which likely impacts protein functionality.




http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

PGC-1�1 and -�4 Induce a Shift toward Alternative Pro-moter Usage and 5�-UTR Isoforms—Ndrg4, a cytoplasmic pro-tein from the �/�-hydrolase superfamily that has been shownto regulate cancer progression (25) and cardiac development inzebrafish (26), showed one of the highest probabilities for alter-native splicing on Partek (Fig. 5D). Interestingly, the expressionof PGC-1�1 or PGC-1�4 resulted in the clear induction ofNdrg4 variants expressed from an internal gene promoter (27).The specific promoter usage was confirmed by qRT-PCRamplification of the 5�-UTR and coding sequence of exon 1,

unique to isoforms Ndrg4-001 and Ndrg4-002 (Fig. 5F). In thebrain, activation of an upstream promoter produces the longerisoform Ndrg4-003 (27) (404 amino acids) that contains a dis-tinct N terminus (Fig. 5E). In contrast to the internal gene pro-moter, the Ndrg4 upstream promoter was not responsive to theexpression of any PGC-1� variants in myotubes. In line withthis, we observed no differences between conditions when tar-geting regions specific for Ndrg4-003 (constitutive exons 1 and2) (Fig. 5F). Ndrg4-002 (339 amino acids) differs from thecanonical Ndrg4-001 (352 amino acids) in that the terminal

DEAD box polypetide 27 (Ddx27)

* differentially expressed exon 1

3.25

5.5

7.75

105' 3'

A

Ddx27-001: ENSMUST00000018143

Ddx27-002

e8 e9

e8b Alt 3´UTR

isoform-specific exoncommon exon

probe 1

probe 2

B

probe 1

C

probe 2

GFP α1 α2 α3 α4

Exo

n si

gnal

*

GFP α1 α2 α3 α4

**

0.0

0.5

1.0

1.5

0

1

2

3

4

Ddx27e8 e9e7 e10

Ddx27-002: ENSMUST00000150571

Ddx27-001

Exo

n si

gnal

* alternative promoter 5

6.75

8.5

10.25

12

D

5' 3'GFP α1 α2 α3 α4

probe 1 probe 2

E

isoform-specific exonconstitutive / common exon

e1 e2 e4e3

Ndrg4-003

Alt 5´e1b

Ndrg4-001, Ndrg4-002

e2

Ndrg4e1 e2 e3 e4

Ndrg4-003: ENSMUST00000041318

Ndrg4-001: ENSMUST00000073139

Ndrg4-002: ENSMUST00000080666

F

probe 1 probe 2

GFP α1 α2 α3 α4

0.0

0.5

1.0

1.5

0

5

10

15* *

N-myc downstream regulated gene (Ndrg4)

rela

tive

mR

NA

expr

essi

on

fold

cha

nge

over

con

trol

rela

tive

mR

NA

expr

essi

on

fold

cha

nge

over

con

trol

FIGURE 5. Alternative splicing of Ddx27 and Ndrg4 induced by distinct PGC-1� isoforms. A, Ddx27 exon array probe set signal intensity. *, differentiallyexpressed exon. B, graphic representation of the mRNA structure of Ddx27 isoforms indicating exons of interest. e8, constitutive exon 8; e8b, alternative exon8b, Alt 3�-UTR, alternative 3�-UTR); e9, exon9. C, qRT-PCR analysis of specific Ddx27 isoform expression in myotubes expressing each PGC-1� isoform. Error barsrepresent S.D. *, p � 0.05 versus GFP. D, Ndrg4 exon array probe set signal intensity. *, alternative promoter. E, graphic representation of Ndrg4 isoform mRNAstructure indicating exons of interest: e1, constitutive exon 1; e1b, alternative exon 1b; Alt 5�, alternative 5�-UTR; e2– e4, exons 2– 4. F, specific Ndrg4 isoformexpression in myotubes expressing each PGC-1� isoform was analyzed by qRT-PCR. Error bars represent S.D. *, p � 0.05 versus GFP.




http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

exon 14 is spliced out, producing an isoform with a shorter Cterminus. Because exon 14 is present in both Ndrg4-001 andNdrg4-003, we cannot determine by transcript analysis whichspecific isoform (Ndrg4-001 or Ndrg4-002) is induced uponPGC-1�1 and -�4 expression.

Osbpl1a is a member of the oxysterol-binding protein familythat acts as intracellular lipid receptor (28). This gene showedone of the highest scores for alternative splicing on Partek anal-ysis (Fig. 6A). We mapped two specific alternative splicing siteson Obspl1a transcript that matched with alternative 3�-UTR(Fig. 6B, probe 2) and alternative promoter usage (Fig. 6B, probe4), respectively. Accordingly, by qRT-PCR analysis, we con-firmed that PGC-1�1 and -�4 induced the expression of anOsbpl1a isoform derived from a previously described internalalternative promoter between exons 15 and 16 of the canonicalvariant (28). Besides the distinct exon 1b, isoforms spawningfrom this internal promoter share the rest of the sequence withthe canonical Osbpl1a-001 and produce smaller C-terminalprotein variants (Ensembl Osbpl1a-003, -004, -005, -007, and-017). Thus, two separate probes, one for the internal 5�-UTR(Fig. 6C, probe 4) and another for the constitutive exon 28 fromthe 3�-end of the Osbpl1a transcripts (Fig. 6C, probe 5), showeda 3– 4-fold induction compared with the GFP control.

Analysis of the exon array indicated with high probabilitythat PGC-1�2 and -�3 could induce the expression of an alter-native Osbpl1a isoform with a distinct 3�-UTR due to a prema-ture stop codon in exon 12 that matched with features fromisoform Osbpl1a-009. However, qRT-PCR analysis did notshow statistical significance for that specific splicing event (Fig.6C, probe 2). We observed no additional differences betweenconditions when detecting a constitutive exon common to allisoforms containing the canonical N terminus (Fig. 6C, probe 1)or when probing for exons specific to the canonical full-lengthOsbpl1a-001 isoform (Fig. 6C, probe 3).

PGC-1�1 and PGC-1�4 Induce Specific Promoter Usage ofTarget Genes in Vivo—Next, we wanted to validate in vivo someof the splicing events we saw in vitro. For that we have used twoavailable muscle-specific transgenic mouse lines, one forcanonical PGC-1�1 (7) and another for alternative PGC-1�4(3), which was backcrossed to a C57BL/6 background. Uponanalysis by qRT-PCR of Osbpl1a isoform expression in the dif-ferent skeletal muscle samples (gastrocnemius), we observedthat both PGC-1� variants induce the expression of the shortOsbpl1a isoforms transcribed from the internal promoter (Fig.7B) when compared with wild type littermates. This is shown bya 3– 4-fold increase in the mRNA levels of Osbpl1a transcripts

TABLE 1Primers used for detection of gene expression by qRT-PCR

Forward Reverse

PGC-1� isoformsTotal PGC-1� (exon 2) TGATGTGAATGACTTGGATACAGACA GCTCATTGTTGTACTGGTTGGATATGPGC-1�1 GGACATGTGCAGCCAAGACTCT CACTTCAATCCACCCAGAAAGCTPGC-1�2 CCACCAGAATGAGTGACATGGA GTTCAGCAAGATCTGGGCAAAPGC-1�3 AAGTGAGTAACCGGAGGCATTC TTCAGGAAGATCTGGGCAAAGAPGC-1�4 TCACACCAAACCCACAGAAA CTGGAAGATATGGCACATExons 3–5 AGCCGTGACCACTGACAACGAG GCTGCATGGTTCTGAGTGCTAAGExons 3–8 AAGACGGATTGCCCTCATTTG GCTTTGGCGAAGCCTTGAA

Gene validationBaiap2 AGAACTCTGCTGCTTACCATTC TTGCTTTGAGACCCCACAAATCpt1b GCACACCAGGCAGTAGCTTT CAGGAGTTGATTCCAGACAGGTACyc ACAAGAAGACTCAAATGTGTTTCAGTTT TGCACTGTCAAGAATAGACAGTTGCEgfr GCCATCTGGGCCAAAGATACC GTCTTCGCATGAATAGGCCAATEsrra GGGGAGCATCGAGTACAGC AGACGCACACCCTCCTTGAHmgcs1 AACTGGTGCAGAAATCTCTAGC GGTTGAATAGCTCAGAACTAGCCMap2k6 ATGTCTCAGTCGAAAGGCAAG TTGGAGTCTAAATCCCGAGGCMapkapk2 TTCCCCCAGTTCCACGTCA GCAGCACCTTCCCGTTGATMcm3 AGCGCAGAGAGACTACTTGGA CAGCCGATACTGGTTGTCACTMcm4 GAGGAAAGCAGGTCGTCACC AGGGCTGGAAAACAAGGCATTMcm5 CAGAGGCGATTCAAGGAGTTC CGATCCAGTATTCACCCAGGTMcm6 ACCAACCCAAGGTTTGGAGG TAATGCTCTCAGCGGTCTGTTMmp15 GGCAACTTCGACACAGTGG AGCCAGTAGCGGTTACCTTTGMsmo1 AAACAAAAGTGTTGGCGTGTTC AAGCATTCTTAAAGGGCTCCTGMyh7 ACTGTCAACACTAAGAGGGTCA TTGGATGATTTGATCTTCCAGGGMyl2 ATCGACAAGAATGACCTAAGGGA ATTTTTCACGTTCACTCGTCCTNsdhl TCATGGTGAATCAAAGCGAGG CCGGGGGTTATCAAAGCCTTGPhd2 TTGTTACCCAGGCAACGGAAC CCTTGGCGTCCCAGTCTTTPik3r1 AAACTCCGAGACACTGCTGA AGACTCATTCCGGTAGTGGTPtgs2 TGAGCAACTATTCCAAACCAGC GCACGTAGTCTTCGATCACTATCSc5d TCTCAGTGCCGCCGATTACTA CTTGACAGTGAACACGATCTCASmad7 GGCCGGATCTCAGGCATTC TTGGGTATCTGGAGTAAGGAGGSrc-1 GGACCCAGGTACTGAGAACCA GGAGGCCGAGGTAATCGTCVegfa GCACATAGGAGAGATGAGCTTCC CTCCGCTCTGAACAAGGCTVegfd TTGAGCGATCATCCCGGTC GCGTGAGTCCATACTGGCAAG

Splicing validationDdx27, probe 1 CGTCTTGGAGCGTCTGATCT GAAGAGCGGCTTCCTGAGCATTDdx27, probe 2 TGCAGGCCCTTAATTCCATCACT TACATAATACAGCTCCTCTGCCGTGNdrg4, probe 1 ATGAAGGTGCTGGGACACA AGCAGGGGCTTCTCCTCAGNdrg4, probe 2 TACCTCCAGACCGAGTGACTCAG GGCGTCTCGATGTCATGTTCOsbpl1a, probe 1 CACAAACAAGTGGTCGAGGA GCTCCTTTCGTCCTGTGAAGOsbpl1a, probe 2 TGCTTTGATGACACTGTTCACT AGTTCTGGGAAAGACAGAAGTCGOsbpl1a, probe 3 GAGTGTGATGTGGCCAAAGA GACAGTCATTCAACGCCACAOsbpl1a, probe 4 ACTTAGGGAGACCTTGCGGTA GCACGCTTCAGCACAGATGOsbpl1a, probe 5 CAGAGAAGTCAAGGTGTTTGC GTACCAGATGCCTTGCCCTA




http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

containing the alternative 5�-UTR or exons that code for theC-terminal end of the protein. No changes were observed whenprobing for sequences exclusive to the full-length Osbpl1a-001isoform (Fig. 7, A and B). Changes in gene expression wereconfirmed at the protein level in both transgenic models with a

clear induction of the 50-kDa short C-terminal isoform ofOsbpl1a (Fig. 7, A and B).

A similar approach was used to validate Ndrg4 isoformexpression in skeletal muscle from muscle creatine kinase pro-moter-PGC-1�1 (Mck-PGC-1�1) and myogenin promoter-PGC-1�4 (Myo-PGC-1�4) mice. The specific expression ofPGC-1�1 in skeletal muscle prompted a 60% increase on themRNA levels of Ndrg4 isoforms derived from the internal genepromoter, whereas no changes were observed for distal pro-moter-derived Ndrg4-003 (Fig. 7C). However, in PGC-1�4skeletal muscle-specific transgenic mice, we observed an induc-tion of similar amplitude (3– 4-fold) for all the Ndrg4 isoformsregardless of the promoter of origin.

When we assessed Ndrg4 protein levels, we observed that,contrarily to what occurs for transcript levels, the protein con-tent of Ndrg4 short isoforms in Mck-PGC-1�1 and Myo-PGC-1�4 muscles is reduced. Because the transcript levels of Ndrg4-003 are very low compared with those of Ndrg4-001 and -002and only a unique band appeared in the immunoblot thatmatches the expected size, we assumed it corresponded to oneof the short Ndrg4 isoforms. Although all three Ndrg4 isoformscan be detected by immunoblotting in vitro (data not shown),only one variant could be detected in vivo in agreement withprevious observations (29).

Discussion

To date, more than eight alternative PGC-1� isoforms havebeen described in skeletal muscle (4). Of these, several are tran-scribed from an alternative gene promoter, including PGC-1�2, PGC-1�3, and PGC-1�4. PGC-1� transcription istightly regulated and dependent on specific stimuli. Here, weshow that administration of clenbuterol to mice induces notonly PGC-1�4 expression in skeletal muscle (3) but alsoPGC-1�2 and PGC-1�3 while reducing the levels of canon-ical promoter-derived PGC-1�1 (Fig. 1B). In addition totheir distinct transcriptional regulation, PGC-1� proteinsaccumulate at different levels (3, 30) (Fig. 1D). We haveobserved that changes in the sequence of PGC-1� isoformsalter protein degradation dynamics and confer increased sta-bility to some of the alternative PGC-1� variants. Finally, wedemonstrate that alternative PGC-1� isoforms regulate theexpression of large gene networks and can, in addition, shiftthe splicing pattern of target genes. These findings add addi-tional nodes to the already complex regulatory network con-trolled by the PGC-1� system.

Several biological functions have been already attributed tonon-canonical PGC-1� isoforms in skeletal muscle (24, 31).However, until now, not enough data were available to identifyand characterize the gene networks under the control of PGC-1�2 and -�3. By using a high resolution splicing array, we couldidentify several target genes for these isoforms.

Interestingly, among the PGC-1� variants analyzed, PGC-1�1 proved to regulate gene networks in the most coordinatedfashion, i.e. with the highest number of target genes within eachpathway. The conservation of nuclear receptor interactionand/or regulatory sites located in the N-terminal part of theprotein might explain the differences in transcriptional activityof the non-canonical PGC-1� variants as seen also in

Oxysterol binding protein-like 1A (Osbpl1a)Ex

on s

igna

l

e3 e14

Alt5´ e28

e3 e12

e28

probe 1 probe 4probe 3probe 2 probe 5

probe 5

probe 2

probe 4

probe 1

A

B

C

GFP α1 α2 α3 α4

constitutive / common exonENSMUST00000074352ENSMUST00000121774ENSMUST00000122175

GFPα1α2α3α4

probe 3

1

3.25

5.5

7.75

10

* alternative promoter

5' 3'

*

e4

e4

e12

Alt3´

e15

* ***

0.0

0.5

1.0

1.5

0.0

0.5

1.0

1.5

0

2

4

6

0

1

2

3

4

0

1

2

3

4

e12 e13e3 e4

Osbpl1a

e14 e28

Osbpl1a-001

Osbpl1a-009

Osbpl1a-003,-004,-005, -006, -007

rela

tive

mR

NA

expr

essi

on

fold

cha

nge

over

con

trol

rela

tive

mR

NA

expr

essi

on

fold

cha

nge

over

con

trol

FIGURE 6. Regulation of Osbpl1a alternative promoter usage by distinctPGC-1� isoforms. A, Osbpl1a exon array probe set signal intensity. *, alter-native promoter. B, graphic representation of the mRNA structure of Osbpl1aisoforms: e, exon; Alt3�, alternative 3�-UTR; Alt5�, alternative 5�-UTR. C, qRT-PCR analysis of specific Osbpl1a isoform expression in myotubes expressingeach PGC-1� isoform. Error bars represent S.D. *, p � 0.05 versus GFP.




http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

NT-PGC-1� (2), a PGC-1� isoform structurally related toPGC-1�4 (4). Like PGC-1�1, NT-PGC-1� can physically inter-act with PPAR� and PPAR�. However, its coactivator functionis greatly dependent on PPAR ligands because NT-PGC-1�lacks the ligand-independent nuclear receptor activationdomain (338 – 403 amino acids) functional in PGC-1�1 (2).

Of note, a significant number of PGC-1�1 target genes arecontra-regulated by PGC-1�2 and PGC-1�3 (i.e. repressedinstead of induced or vice versa). We could only predict cleareffects on energy metabolism pathways when performing net-

work analysis on classical PGC-1�1 (Fig. 3A). However, whenmeasuring oxygen consumption rates in cell culture, weobserved that myotubes expressing PGC-1�2 and PGC-1�3behave similarly to those expressing PGC-1�1 by displayingenhanced pyruvate mitochondrial respiration. From our geneand network analysis, it is unlikely that PGC-1�2 or -�3 regu-lates mitochondrial biogenesis. We therefore speculate that themechanism by which those isoforms elicit these effects mightinclude coordinated splicing of genes involved in pyruvatemetabolism. Effects on oxidative metabolism have been

A

0.0 0

0.5

1.0

1.5

Osbpl1a isoform 1

0

1

2

3

4

Osbpl1a Alt5’

*

1

2

3

4

5

Osbpl1a exon 28

*

Osbpl1a isoform 1 Osbpl1a exon 28Osbpl1a Alt5’

0.0

0.5

1.0

1.5

2.0

*

0.0

0.5

1.0

1.5

2.0

2.5

*

0.0

0.5

1.0

1.5

0.0

0.5

1.0

1.5

2.0

0.0

0.5

1.0

1.5

2.0 *

0

2

4

6

*

0

1

2

3

4 *

Ndrg4-1&2 Ndrg4-3 Ndrg4-1&2 Ndrg4-35

BMck-PGC-1α1 transgenic mouse Myo-PGC-1α4 transgenic mouse

Mck-PGC-1α1 transgenic mouse Myo-PGC-1α4 transgenic mouseC D

WT Mck-PGC-1α1

Osbpl1a55

45Ponceau 45

WT Myo-PGC-1α4

Osbpl1a

Ponceau

0

1

2

3

0.0

0.5

1.0

1.5

0.00.51.01.52.02.5

0.0

0.5

1.0

1.5

* *

45

WT Mck-PGC-1α1

Ndrg4 35Ponceau 45

WT Myo-PGC-1α4

Ndrg4

Ponceau

Basal conditions

gene 1

promoter A promoter B

RNA Pol II

Full-length isoform

C-terminal isoform

RNA Pol IIgene 1

promoter A promoter B

RNA Pol II RNA Pol II

PGC-1α1/α4

PGC-1α1/α4 induction

C-terminal isoform

E

Full-length isoform

TF

WT Mck-PGC-1α1 WT Myo-PGC-1α4

WT Mck-PGC-1α1 WT Myo-PGC-1α4

*n.s.

rela

tive

mR

NA

expr

essi

on

fold

cha

nge

over

con

trol

rela

tive

mR

NA

expr

essi

on

fold

cha

nge

over

con

trol

rela

tive

mR

NA

expr

essi

on

fold

cha

nge

over

con

trol

rela

tive

mR

NA

expr

essi

on

fold

cha

nge

over

con

trol

Pro

tein

qua

ntifi

catio

nfo

ld c

hang

e ov

er W

T

Pro

tein

qua

ntifi

catio

nfo

ld c

hang

e ov

er W

T

Pro

tein

qua

ntifi

catio

nfo

ld c

hang

e ov

er W

T

Pro

tein

qua

ntifi

catio

nfo

ld c

hang

e ov

er W

T

kDa

70100

kDa

55

70100

55kDa

35

55kDa

FIGURE 7. Analysis of Ndrg4 and Osbpl1a alternative promoter usage in skeletal muscle of PGC-1�1 or PGC-1�4 transgenic mice. A and B, qRT-PCR andprotein immunoblotting analysis of distinct Osbpl1a isoform expression in gastrocnemius from PGC-1�1 (A) or PGC-1�4 (B) skeletal muscle-specific transgenicmice. C and D, qRT-PCR and protein immunoblotting analysis of distinct Ndrg4 isoform expression in gastrocnemius from PGC-1�1 (C) or PGC-1�4 (D) skeletalmuscle-specific transgenic mice (Mck-PGC-1�1 (6) and Myo-PGC-1�4 (4)). E, proposed model for alternative promoter usage upon induction of specific PGC-1�isoforms. In all panels, error bars represent S.E. *, p � 0.05 versus WT; n.s., nonsignificant. Pol, polymerase; Alt5�, alternative 5�-UTR.




http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

described for other PGC-1� alternative isoforms such asNT-PGC-1� (32) and PGC-1�-b (31, 33).

Although originating from the same promoter and uponidentical stimuli, the alternative PGC-1� isoforms regulate dif-ferent transcriptional programs. This is not surprising consid-ering the differences in protein sequence and structure amongthese variants. For example, PGC-1�2 and -�3 lack amino acids120 –284 of PGC-1�1 that seem to be necessary for PGC-1� tointeract with PPAR� (34). This might explain in part the rela-tive clustering in target gene regulation we observed for PGC-1�1 and PGC-1�4 versus PGC-1�2 and PGC-1�3. Indeed, theability to bind specific transcriptional partners can probablydefine the biological activity of the different PGC-1� coactiva-tors. This particular functional clustering of the PGC-1� iso-forms was also found when assessing their potential role astranscriptional repressors. Although PGC-1�1 has always beenconsidered as a direct positive regulator of gene expression, ithas been shown recently that its partnership with heat shockfactor 1 can result in repression of target gene expression (35).When analyzing our exon array data set to determine whetherother PGC-1� variants can reproduce the effects of PGC-1�1on heat shock factor 1 target gene regulation in skeletal muscle(35), we found that the expression of either PGC-1�1 or PGC-1�4 in primary myotubes produced equivalent reductions inheat shock factor 1 target genes expression levels, whereasPGC-1�2 and -�3 had only minimal effects.

Indeed, some similarities can also be seen when comparingthe gene programs triggered by PGC-1�2 and -�3. These vari-ants are the closest related PGC-1� isoforms, their sequencebeing completely homologous save for the first amino acid, andinterestingly the inclusion of exon 1b� (PGC-1�3) instead ofexon 1b (PGC-1�2) seems to affect the basal protein accumu-lation (Fig. 1D). Other groups have shown that alternativePGC-1� isoforms that only differ in exon 1 (namely PGC-1�-aand PGC-1�-b) can elicit similar biological responses (33).However, the induction of endogenous PGC-1�3 seen whenectopic PGC-1�2 is expressed (Fig. 1I) might influence theiroverall transcriptional activity and bring their respective geneprograms closer. Both PGC-1�2 and PGC-1�3 induce theexpression of Mcm3– 6, known markers of proliferation (36).Although our results were obtained in fully differentiated myo-tubes, this could indicate a potential positive role on muscleregeneration and growth. Interestingly, PGC-1� homologPGC-1� represses Mcm expression and prevents proliferationof vascular smooth cells (23). Of note and contrary to PGC-1�2and -�3, PGC-1�1 and -�4 repress the expression of severalmembers of the Mcm family (Fig. 3F).

Despite their similarities, PGC-1�2 still has unique targetgenes that are not shared with PGC-1�3. The overall regulationof genes targeted by PGC-1�2 suggests a consistent decrease incholesterol synthesis that might indicate a negative effect onskeletal muscle differentiation (37, 38). Other signals that advo-cate for a role in myogenesis and muscle recovery include thespecific PGC-1�2 induction of the intracellular antagonist ofTGF-�, Smad7 (39) (Fig. 3D), an event that might cooperatewith the effects of PGC-1�4 on skeletal muscle growth andmaintenance after resistance training. PGC-1�3 seems to alsoexert functions in cell cycle control, proliferation, and tissue

remodeling as seen in the regulation of the epithelial adherensjunction pathway.

In any case, because the distinct PGC-1� isoforms accumu-late to distinct levels under different conditions, concern abouttarget gene specificity could be raised. However, the patterns oftarget gene expression observed upon incremental dosage ofeach PGC-1� variant argue against it (Fig. 4C). In addition, if weconsider, for example, isoforms PGC-1�2 and PGC-1�3 (themost similar proteins that only differ in the first few aminoacids), if the target gene differences were due to the differentprotein levels the prediction would be that the PGC-1�2 geneset would be included in the PGC-1�3 gene set. However, wesee that of the 1622 genes regulated by PGC-1�2 only 437 areshared with the more abundant PGC-1�3. The same canbe observed for any other comparison.

It has been previously shown that PGC-1�4 expression inskeletal muscle promotes muscle hypertrophy (3) and angio-genesis (24). In addition, here we show that PGC-1�4 can acti-vate several gene programs related to cancer signaling.

Our initial working hypothesis was that differences inPGC-1� isoform sequences might have an impact on splicingregulation and that conservation of the RS/RRM C-terminaldomains would determine the splicing patterns of target genes(8). However, our results indicate that most distinct splicingpatterns are produced by PGC-1� isoforms that differ in their Ntermini. PGC-1�4 expression in myotubes produces splicingpatterns that are remarkably similar to those generated byPGC-1�1. In addition, these contrast strongly with the splicingpatterns seen upon PGC-1�2 and PGC-1�3 expression (Fig.4F). Thus, the conservation of the activation domain mighthave a greater impact on defining both the gene programs andthe splicing options made during transcript maturation.

The fact that different PGC-1� variants have specific effectson target gene expression and splicing greatly enhances ourunderstanding of how these coactivators promote different bio-logical functions. For example, although changes in brainNdrg4 isoform expression have been reported previously (27),no mechanism for this shift has been proposed. Our resultsshow that PGC-1�1 and -�4 have the specific ability to induce ashift in Ndrg4 expression that favors isoform 001 and/or 002instead of 003. This expression pattern seems to mimic what isobserved during brain development where Ndrg4 isoformsfrom the upstream promoter (e.g. Ndrg4-003) peak duringembryonic development but disappear at 6 weeks of age (29).Although the regulation of Ndrg4 isoform expression was sim-ilar in vitro and in vivo, additional regulation might affect Ndrg4protein stability and explain the differences between transcriptand protein regulation observed in the transgenic models.

PGC-1�1 and PGC-1�4 induce the expression of a shortC-terminal isoform of Osbpl1a from an internal gene promoterboth in vitro and in vivo (Fig. 7E). This isoform, coined ORP1S,retains the oxysterol binding domain and promotes traffickingof oxysterols from the cytoplasm to the nucleus where the cho-lesterol metabolites can act as ligands for the liver X receptor(40) and activate gene expression. Conversely, ORP1L, the full-length canonical isoform of Osbpl1a that is not induced byPGC-1�1 or -�4, works as a cholesterol sensor that determineslate endosome positioning (41) and might have a lesser effect on




http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

gene regulation. This suggests that PGC-1� isoforms mightplay a role in liver X receptor transcriptional activity fine-tun-ing by altering ligand delivery.

The precise mechanisms by which coactivators regulate co-transcriptional splicing are still not completely understood.From the two co-existing models, studies on PGC-1�1 andMed23 (12) support the recruitment model in which splicingfactors are recruited to the preinitiation complex to assist in thematuration of the transcript. The recruitment model relies on acentral role of RNA polymerase II CTD in recruiting splicingfactors while still loaded on a distinct promoter (43). The iden-tity of the factors docked on the RNA polymerase II CTD willdetermine the nature of the resulting splicing. It has beenshown that the RS/RRM domains are not needed for PGC-1� tointeract with the preinitiation complex (44). Accordingly, ourresults also support the recruitment model (40) in whichnuclear receptors and/or cofactors will dictate its associationwith certain promoters and help in the recruitment of splicingfactors to the preinitiation complex. This association would aswell be independent of the CTD and rely on conservation of theactivation domain. The specific association of PGC-1�1 withother members of the family might ultimately define the geneprograms initiated.

In light of our results, we conclude that regulation of tran-scriptional networks by the PGC-1� system is more complexthan initially thought. We show that the identification ofPGC-1� isoform target genes is greatly facilitated by taking intoaccount the exon structure of the corresponding transcripts.This work provides a new framework for our understanding ofhow PGC-1� coactivators mediate skeletal muscle adaptationsto exercise.

Experimental Procedures

Cell Culture and PGC-1� Isoform Expression—Mouse pri-mary myoblasts were isolated as described previously (45). Dur-ing the growth phase, myoblasts were plated in collagen-coatedplates and maintained in Ham’s F-10 nutrient mixture (Gibco),20% FBS (Sigma), 1% penicillin/streptomycin (Gibco), and 2.5ng/ml basic fibroblast growth factor (Gibco). After achieving90% confluence, cells were changed to differentiation mediumconsisting of DMEM high glucose with pyruvate (Gibco), 5%horse serum (Gibco), and 1% penicillin/streptomycin. Uponmyotube formation (48 h after induction of differentiation), weexpressed GFP alone or GFP and one of the four PGC-1� iso-forms via adenoviral vectors at a multiplicity of infection of 100.GFP and PGC-1� isoforms are expressed from independentpromoters, and the adenoviruses expressing the differentmouse PGC1� isoforms or GFP have been described previously(3). Twelve hours after transduction, medium was replacedwith fresh differentiation medium. Approximately 60 h post-transduction, cells were harvested in Isol-RNA lysis reagent(5PRIME) and purified with Nucleospin� RNA II columns(Macherey-Nagel) according to the manufacturers’ instruc-tions to achieve optimal purity levels. Experiments were per-formed at least in triplicate using three replicates per condition.

Microarray Procedures—For global analysis of gene expres-sion, we used the Affymetrix GeneChip Mouse Exon 1.0 STarray (46). Gene expression profiling experiments were per-

formed at the Bioinformatics and Gene Expression Analysiscore facility of the Karolinska Institutet. All data sets were fil-tered through quality control steps and normalized using therobust multianalysis method for background correction.

Gene Level Analysis—Differential gene expression was deter-mined by separately comparing the summarized probe setintensity for genes regulated by each PGC-1� isoform versusthe control condition (GFP). Differences in gene expressionwere calculated with a two-sided unpaired t test assuming equalvariance and applying a cutoff of 1.2-fold change and a p value�0.05. Qlucore Omics Explorer was used to perform hierarchi-cal clustering analysis and to generate gene regulation heatmaps. We used the Venn diagram generator (VENNY) (42)for the graphical representation of different sets of datacomparison.

Pathway Analysis—We performed comprehensive pathwayand network analysis for gene expression by using the IPA plat-form. IPA output was filtered to include only functions andpathways experimentally validated in skeletal muscle. The top10 significant pathways (p � 0.05) were considered for subse-quent qRT-PCR validation of target genes.

Isoform Level Analysis—To identify and characterize splicingevents promoted by the different PGC-1� variants, data setswere paired for all the possible comparisons. Splicing hits werethen double filtered according to the workflow depicted in Fig.2A. Partek Genomic Suite (Partek) assigned a unique alterna-tive splicing statistic value to each gene (we set the cutoff atalt.splice attribute �0.0001). Additionally, EAA (47) generateda unique splicing index for each probe set and paired compari-son by relating single probe set intensities to average gene probeset intensity (equivalent to gene expression). Only probe setswith a �1 � splicing index � 1 were considered. By performingthis double filter, we aimed to increase the robustness of thealternative splicing candidates and profited from the intrinsicdifferences of the analysis performed by both programs. Forglobal splicing categorization, we used AltAnalyze (48). Asin the microarray procedures, for all the bioinformaticsapproaches described above we used robust multianalysis back-ground correction.

Gene/Exon Expression Validation by Quantitative RT-PCR—To design primers targeting characteristic exons/regions ofannotated isoforms from candidate genes, we used the “probe-to-transcript” mapping feature from EAA combined with theEnsembl database. Hits for gene expression and for alternativesplicing/alternative promoter usage were validated using SYBRGreen-based qRT-PCR. Gene/exon expression was normalizedusing hypoxanthine-guanine phosphoribosyltransferase ashousekeeping gene. The relative expression of different targetgene isoforms was confirmed by specific probe set-primerdesign. All oligonucleotide sequences used in the study aresummarized in Table 1.

Western Blotting Analysis—Protein extracts were prepared ina modified FLAG lysis buffer (50 mM Tris-HCl, pH 7.4, 180 mM

NaCl, 1% Triton X-100, 15% glycerol, and 1 mM EDTA) supple-mented with 1 mM DTT and 0.5 mM PMSF. Protein extractswere resolved by SDS-polyacrylamide gel electrophoresis andtransferred onto polyvinylidene difluoride membranes. Immu-noblotting was performed with antibodies that recognize all




http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

PGC-1� isoforms analyzed in this study (4C1.3, Merck Milli-pore), C-terminal residues of Osbpl1a (anti-ORP1, ab131165,Abcam), and the C-terminal end of Ndrg4 (anti-NDRG4, 9039,Cell Signaling Technology).

Protein Stability and Half-life Determination—Myoblastswere differentiated and transduced as described above in “CellCulture and PGC-1� Isoform Expression” under “ExperimentalProcedures.” On the 4th day of differentiation, cells weretreated with 10 �M proteasome inhibitor MG132 or DMSO ascontrol for 4 h. Alternatively and to determine PGC-1� proteinhalf-lives, myotubes were treated for different times (20 min, 40min, 1 h, 2 h, and 4 h) with 100 �g/ml cycloheximide. After thedifferent incubation times, cells were harvested for cell lysatesas described above in “Western Blotting Analysis” under“Experimental Procedures.”

Extracellular Flux Analysis (Seahorse)—Myoblasts were dif-ferentiated and transduced as described before. On the 4th dayof differentiation, mitochondrial oxidative phosphorylationfrom myotubes transduced with adenovirus expressing PGC-1�1, -�2, -�3, -�4, or GFP control were analyzed using extra-cellular flux analysis (XF24, Seahorse Biosciences) in DMEM,pH 7.4 (Sigma-Aldrich) by using either 1 mM pyruvate or 5 mM

glucose as substrate (Sigma-Aldrich). Baseline oxygen con-sumption rate was measured. Following baseline measure-ments, oligomycin (1 �M), carbonyl cyanide-4-(trifluorome-thoxy)phenylhydrazone (0.3 �M), and antimycin A (2 �M) weresequentially injected to measure oxygen consumption ratesupon different challenges.

Animal Experimentation—Mck-PGC-1�1 skeletal muscletransgenic mice have been described previously (3, 7). Myo-PGC-1�4 skeletal muscle transgenic mice are a C57BL/6 10thgeneration backcrossing of the previously described Myo-Myo-PGC-1�4 transgenic line (3). All experiments and protocolswere approved by the regional animal ethics committee ofNorthern Stockholm, Sweden.

Array Data—All the array data in this study are deposited atGene Expression Omnibus (GEO) (www.ncbi.nlm.nih.gov/geo/) under accession number GSE75448. Complementaryanalysis has been done with the previously published microar-ray data with accession number GSE42473 (3).

Author Contributions—V. M.-R. and J. L. R. conceived, coordinated,and designed the study and wrote the paper. V. M.-R., J. M. L., I. S.,and K. D.-W. performed the bioinformatics analysis. I. C. providedtechnical assistance and contributed to the preparation of figures.V. M.-R., P. R. J., J. C. C., I. C., D. M. S. F., M. I., A. T. P-K., L. Z. A.,and A. G.-C. performed and analyzed experiments. P. C. B. analyzedexperiments. All authors reviewed the results and approved the finalversion of the manuscript.

References1. Correia, J. C., Ferreira, D. M., and Ruas, J. L. (2015) Intercellular: local and

systemic actions of skeletal muscle PGC-1s. Trends Endocrinol. Metab.26, 305–314

2. Zhang, Y., Huypens, P., Adamson, A. W., Chang, J. S., Henagan, T. M.,Boudreau, A., Lenard, N. R., Burk, D., Klein, J., Perwitz, N., Shin, J.,Fasshauer, M., Kralli, A., and Gettys, T. W. (2009) Alternative mRNAsplicing produces a novel biologically active short isoform of PGC-1�.J. Biol. Chem. 284, 32813–32826

3. Ruas, J. L., White, J. P., Rao, R. R., Kleiner, S., Brannan, K. T., Harrison,B. C., Greene, N. P., Wu, J., Estall, J. L., Irving, B. A., Lanza, I. R., Rasbach,K. A., Okutsu, M., Nair, K. S., Yan, Z., Leinwand, L. A., and Spiegelman,B. M. (2012) A PGC-1� isoform induced by resistance training regulatesskeletal muscle hypertrophy. Cell 151, 1319 –1331

4. Martínez-Redondo, V., Pettersson, A. T., and Ruas, J. L. (2015) The hitch-hiker’s guide to PGC-1� isoform structure and biological functions. Dia-betologia 58, 1969 –1977

5. Chinsomboon, J., Ruas, J., Gupta, R. K., Thom, R., Shoag, J., Rowe, G. C.,Sawada, N., Raghuram, S., and Arany, Z. (2009) The transcriptional co-activator PGC-1� mediates exercise-induced angiogenesis in skeletalmuscle. Proc. Natl. Acad. Sci. U.S.A. 106, 21401–21406

6. Nader, G. A., von Walden, F., Liu, C., Lindvall, J., Gutmann, L., Pistilli, E. E.,and Gordon, P. M. (2014) Resistance exercise training modulates acutegene expression during human skeletal muscle hypertrophy. J. Appl.Physiol. 116, 693–702

7. Lin, J., Wu, H., Tarr, P. T., Zhang, C. Y., Wu, Z., Boss, O., Michael, L. F.,Puigserver, P., Isotani, E., Olson, E. N., Lowell, B. B., Bassel-Duby, R., andSpiegelman, B. M. (2002) Transcriptional co-activator PGC-1� drives theformation of slow-twitch muscle fibres. Nature 418, 797– 801

8. Monsalve, M., Wu, Z., Adelmant, G., Puigserver, P., Fan, M., and Spiegel-man, B. M. (2000) Direct coupling of transcription and mRNA processingthrough the thermogenic coactivator PGC-1. Mol. Cell 6, 307–316

9. Kornblihtt, A. R., Schor, I. E., Alló, M., Dujardin, G., Petrillo, E., and Mu-ñoz, M. J. (2013) Alternative splicing: a pivotal step between eukaryotictranscription and translation. Nat. Rev. Mol. Cell Biol. 14, 153–165

10. Du, L., and Warren, S. L. (1997) A functional interaction between thecarboxy-terminal domain of RNA polymerase II and pre-mRNA splicing.J. Cell Biol. 136, 5–18

11. Auboeuf, D., Dowhan, D. H., Kang, Y. K., Larkin, K., Lee, J. W., Berget,S. M., and O’Malley, B. W. (2004) Differential recruitment of nuclearreceptor coactivators may determine alternative RNA splice site choice intarget genes. Proc. Natl. Acad. Sci. U.S.A. 101, 2270 –2274

12. Huang, Y., Li, W., Yao, X., Lin, Q. J., Yin, J. W., Liang, Y., Heiner, M., Tian,B., Hui, J., and Wang, G. (2012) Mediator complex regulates alternativemRNA processing via the MED23 subunit. Mol. Cell 45, 459 – 469

13. Du, C., Ma, X., Meruvu, S., Hugendubler, L., and Mueller, E. (2014) Theadipogenic transcriptional cofactor ZNF638 interacts with splicing regu-lators and influences alternative splicing. J. Lipid Res. 55, 1886 –1896

14. Puigserver, P., Wu, Z., Park, C. W., Graves, R., Wright, M., and Spiegel-man, B. M. (1998) A cold-inducible coactivator of nuclear receptors linkedto adaptive thermogenesis. Cell 92, 829 – 839

15. Thijssen-Timmer, D. C., Schiphorst, M. P., Kwakkel, J., Emter, R., Kralli,A., Wiersinga, W. M., and Bakker, O. (2006) PGC-1� regulates the isoformmRNA ratio of the alternatively spliced thyroid hormone receptor � tran-script. J. Mol. Endocrinol. 37, 251–257

16. Silvennoinen, M., Ahtiainen, J. P., Hulmi, J. J., Pekkala, S., Taipale, R. S.,Nindl, B. C., Laine, T., Häkkinen, K., Selänne, H., Kyröläinen, H., andKainulainen, H. (2015) PGC-1 isoforms and their target genes are ex-pressed differently in human skeletal muscle following resistance and en-durance exercise. Physiol. Rep. 3, e12563

17. Trausch-Azar, J. S., Abed, M., Orian, A., and Schwartz, A. L. (2015) Iso-form-specific SCF(Fbw7) ubiquitination mediates differential regulationof PGC-1�. J. Cell. Physiol. 230, 842– 852

18. Sano, M., Tokudome, S., Shimizu, N., Yoshikawa, N., Ogawa, C., Shi-rakawa, K., Endo, J., Katayama, T., Yuasa, S., Ieda, M., Makino, S., Hattori,F., Tanaka, H., and Fukuda, K. (2007) Intramolecular control of proteinstability, subnuclear compartmentalization, and coactivator function ofperoxisome proliferator-activated receptor � coactivator 1�. J. Biol. Chem.282, 25970 –25980

19. Olson, B. L., Hock, M. B., Ekholm-Reed, S., Wohlschlegel, J. A., Dev, K. K.,Kralli, A., and Reed, S. I. (2008) SCFCdc4 acts antagonistically to thePGC-1� transcriptional coactivator by targeting it for ubiquitin-mediatedproteolysis. Genes Dev. 22, 252–264

20. Tian, L., Cao, J., Deng, X., Zhang, C., Qian, T., Song, X., and Huang, B.(2014) Unveiling transcription factor regulation and differential co-ex-pression genes in Duchenne muscular dystrophy. Diagn. Pathol. 9, 210

21. Roberts, T. C., Johansson, H. J., McClorey, G., Godfrey, C., Blomberg,




http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

K. E., Coursindel, T., Gait, M. J., Smith, C. I., Lehtiö, J., El Andaloussi, S.,and Wood, M. J. (2015) Multi-level omics analysis in a murine model ofdystrophin loss and therapeutic restoration. Hum. Mol. Genet. 24,6756 – 6768

22. Camarda, G., Siepi, F., Pajalunga, D., Bernardini, C., Rossi, R., Monte-cucco, A., Meccia, E., and Crescenzi, M. (2004) A pRb-independent mech-anism preserves the postmitotic state in terminally differentiated skeletalmuscle cells. J. Cell Biol. 167, 417– 423

23. Guo, Y., Fan, Y., Zhang, J., Chang, L., Lin, J. D., and Chen, Y. E. (2013)Peroxisome proliferator-activated receptor � coactivator 1� (PGC-1�)protein attenuates vascular lesion formation by inhibition of chromatinloading of minichromosome maintenance complex in smooth musclecells. J. Biol. Chem. 288, 4625– 4636

24. Thom, R., Rowe, G. C., Jang, C., Safdar, A., and Arany, Z. (2014) Hypoxicinduction of vascular endothelial growth factor (VEGF) and angiogenesisin muscle by truncated peroxisome proliferator-activated receptor � co-activator (PGC)-1�. J. Biol. Chem. 289, 8810 – 8817

25. Kotipatruni, R. P., Ferraro, D. J., Ren, X., Vanderwaal, R. P., Thotala, D. K.,Hallahan, D. E., and Jaboin, J. J. (2012) NDRG4, the N-Myc downstreamregulated gene, is important for cell survival, tumor invasion and angio-genesis in meningiomas. Integr. Biol. 4, 1185–1197

26. Qu, X., Jia, H., Garrity, D. M., Tompkins, K., Batts, L., Appel, B., Zhong,T. P., and Baldwin, H. S. (2008) Ndrg4 is required for normal myocyteproliferation during early cardiac development in zebrafish. Dev. Biol. 317,486 – 496

27. Qu, X., Zhai, Y., Wei, H., Zhang, C., Xing, G., Yu, Y., and He, F. (2002)Characterization and expression of three novel differentiation-relatedgenes belong to the human NDRG gene family. Mol. Cell. Biochem. 229,35– 44

28. Johansson, M., Bocher, V., Lehto, M., Chinetti, G., Kuismanen, E., Ehn-holm, C., Staels, B., and Olkkonen, V. M. (2003) The two variants of ox-ysterol binding protein-related protein-1 display different tissue expres-sion patterns, have different intracellular localization, and are functionallydistinct. Mol. Biol. Cell 14, 903–915

29. Nakada, N., Hongo, S., Ohki, T., Maeda, A., and Takeda, M. (2002) Mo-lecular characterization of NDRG4/Bdm1 protein isoforms that are dif-ferentially regulated during rat brain development. Brain Res. Dev. BrainRes. 135, 45–53

30. Trausch-Azar, J., Leone, T. C., Kelly, D. P., and Schwartz, A. L. (2010)Ubiquitin proteasome-dependent degradation of the transcriptional co-activator PGC-1� via the N-terminal pathway. J. Biol. Chem. 285,40192– 40200

31. Wen, X., Wu, J., Chang, J. S., Zhang, P., Wang, J., Zhang, Y., Gettys, T. W.,and Zhang, Y. (2014) Effect of exercise intensity on isoform-specific ex-pressions of NT-PGC-1� mRNA in mouse skeletal muscle. Biomed. Res.Int. 2014, 402175

32. Chang, J. S., Fernand, V., Zhang, Y., Shin, J., Jun, H. J., Joshi, Y., and Gettys,T. W. (2012) NT-PGC-1� protein is sufficient to link �3-adrenergic re-ceptor activation to transcriptional and physiological components ofadaptive thermogenesis. J. Biol. Chem. 287, 9100 –9111

33. Tadaishi, M., Miura, S., Kai, Y., Kano, Y., Oishi, Y., and Ezaki, O. (2011)

Skeletal muscle-specific expression of PGC-1�-b, an exercise-responsiveisoform, increases exercise capacity and peak oxygen uptake. PLoS One 6,e28290

34. Vega, R. B., Huss, J. M., and Kelly, D. P. (2000) The coactivator PGC-1cooperates with peroxisome proliferator-activated receptor � in tran-scriptional control of nuclear genes encoding mitochondrial fatty acidoxidation enzymes. Mol. Cell. Biol. 20, 1868 –1876

35. Minsky, N., and Roeder, R. G. (2015) Direct link between metabolic regu-lation and the heat-shock response through the transcriptional regulatorPGC-1�. Proc. Natl. Acad. Sci. U.S.A. 112, E5669 –E5678

36. Ryu, S., and Driever, W. (2006) Minichromosome maintenance proteinsas markers for proliferation zones during embryogenesis. Cell Cycle 5,1140 –1142

37. Muscat, G. E., Wagner, B. L., Hou, J., Tangirala, R. K., Bischoff, E. D.,Rohde, P., Petrowski, M., Li, J., Shao, G., Macondray, G., and Schulman,I. G. (2002) Regulation of cholesterol homeostasis and lipid metabolism inskeletal muscle by liver X receptors. J. Biol. Chem. 277, 40722– 40728

38. Trapani, L., Segatto, M., La Rosa, P., Fanelli, F., Moreno, S., Marino, M.,and Pallottini, V. (2012) 3-Hydroxy 3-methylglutaryl coenzyme A reduc-tase inhibition impairs muscle regeneration. J. Cell. Biochem. 113,2057–2063

39. Cohen, T. V., Kollias, H. D., Liu, N., Ward, C. W., and Wagner, K. R. (2015)Genetic disruption of Smad7 impairs skeletal muscle growth and regen-eration. J. Physiol. 593, 2479 –2497

40. Lee, S., Wang, P. Y., Jeong, Y., Mangelsdorf, D. J., Anderson, R. G., andMichaely, P. (2012) Sterol-dependent nuclear import of ORP1S promotesLXR regulated trans-activation of apoE. Exp. Cell Res. 318, 2128 –2142

41. Vihervaara, T., Uronen, R. L., Wohlfahrt, G., Björkhem, I., Ikonen, E., andOlkkonen, V. M. (2011) Sterol binding by OSBP-related protein 1L regu-lates late endosome motility and function. Cell. Mol. Life Sci. 68, 537–551

42. Oliveros, J. C. (2015) VENNY, BioinfoGP, Centro Nacional de Biotec-nología, Madrid, Spain

43. Das, R., Yu, J., Zhang, Z., Gygi, M. P., Krainer, A. R., Gygi, S. P., and Reed,R. (2007) SR proteins function in coupling RNAP II transcription topre-mRNA splicing. Mol. Cell 26, 867– 881

44. Wallberg, A. E., Yamamura, S., Malik, S., Spiegelman, B. M., and Roeder,R. G. (2003) Coordination of p300-mediated chromatin remodeling andTRAP/mediator function through coactivator PGC-1�. Mol. Cell 12,1137–1149

45. 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

46. Lockstone, H. E. (2011) Exon array data analysis using Affymetrix powertools and R statistical software. Brief. Bioinform. 12, 634 – 644

47. Gellert, P., Uchida, S., and Braun, T. (2009) Exon Array Analyzer: a webinterface for Affymetrix exon array analysis. Bioinformatics 25,3323–3324

48. Emig, D., Salomonis, N., Baumbach, J., Lengauer, T., Conklin, B. R., andAlbrecht, M. (2010) AltAnalyze and DomainGraph: analyzing and visual-izing exon expression data. Nucleic Acids Res. 38, W755–W762




http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

Karin Dahlman-Wright and Jorge L. RuasPettersson-Klein, Leandro Z. Agudelo, Alfredo Gimenez-Cassina, Patricia C. Brum,

Igor Cervenka, Jessica M. Lindvall, Indranil Sinha, Manizheh Izadi, Amanda T. Vicente Martínez-Redondo, Paulo R. Jannig, Jorge C. Correia, Duarte M. S. Ferreira,

Regulate Multiple Splicing Events on Target Genes Isoforms Selectivelyα Coactivator-1 γPeroxisome Proliferator-activated Receptor

doi: 10.1074/jbc.M115.705822 originally published online May 26, 20162016, 291:15169-15184.J. Biol. Chem.

10.1074/jbc.M115.705822Access the most updated version of this article at doi:

Alerts:

When a correction for this article is posted•

When this article is cited•

to choose from all of JBC's e-mail alertsClick here

http://www.jbc.org/content/291/29/15169.full.html#ref-list-1

This article cites 47 references, 18 of which can be accessed free at


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/lookup/doi/10.1074/jbc.M115.705822

http://www.jbc.org/cgi/alerts?alertType=citedby&addAlert=cited_by&cited_by_criteria_resid=jbc;291/29/15169&saveAlert=no&return-type=article&return_url=http://www.jbc.org/content/291/29/15169

http://www.jbc.org/cgi/alerts?alertType=correction&addAlert=correction&correction_criteria_value=291/29/15169&saveAlert=no&return-type=article&return_url=http://www.jbc.org/content/291/29/15169

http://www.jbc.org/cgi/alerts/etoc

http://www.jbc.org/content/291/29/15169.full.html#ref-list-1

http://www.jbc.org/

peroxisomeproliferator-activatedreceptor coactivator-1 ... · alternative promoter usage in the...

Documents