Biochimica et Biophysica Acta 1849 (2015) 1081–1094

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta journal homepage: www.elsevier.com/locate/bbagrm

Jmjd2C increases MyoD transcriptional activity through inhibiting G9a-dependent MyoD degradation Eun-Shil Jung a,1, Ye-Ji Sim a,1, Hoe-Su Jeong a, Su-Jin Kim a, Ye-Jin Yun a, Joo-Hoon Song b, Su-Hee Jeon c, Chungyoul Choe d, Kyung-Tae Park e, Chang-Hoon Kim a,⁎, Kye-Seong Kim a,⁎ a

Department of Biomedical Science, Graduate School of Biomedical Science and Engineering, Hanyang University, Seoul 133-791, Republic of Korea Bio Focus Co., Ltd., Gyeonggi-do 437-753, Republic of Korea Department of Biological & Environmental Science, Dongguk University, Seoul 100-175, Republic of Korea d Samsung Biomedical Research Institute, School of Medicine, Sungkyunkwan University, Seoul 135-710, Republic of Korea e Center for Cancer Research, Department of Pathology and Laboratory Medicine, University of Tennessee Health Science Center, Memphis, TN, USA b c

a r t i c l e

i n f o

Article history: Received 13 February 2015 Received in revised form 17 June 2015 Accepted 2 July 2015 Available online 4 July 2015 Keywords: Skeletal muscle differentiation Methylation-dependent MyoD degradation Jmjd2C G9a

a b s t r a c t Skeletal muscle cell differentiation requires a family of proteins called myogenic regulatory factors (MRFs) to which MyoD belongs. The activity of MyoD is under epigenetic regulation, however, the molecular mechanism by which histone KMTs and KDMs regulate MyoD transcriptional activity through methylation remains to be determined. Here we provide evidence for a unique regulatory mechanism of MyoD transcriptional activity through demethylation by Jmjd2C demethylase whose level increases during muscle differentiation. G9a decreases MyoD stability via methylation-dependent MyoD ubiquitination. Jmjd2C directly associates with MyoD in vitro and in vivo to demethylate and stabilize MyoD. The hypo-methylated MyoD due to Jmjd2C is significantly more stable than hyper-methylated MyoD by G9a. Cul4/Ddb1/Dcaf1 pathway is essential for the G9a-mediated MyoD degradation in myoblasts. By the stabilization of MyoD, Jmjd2C increases myogenic conversion of mouse embryonic fibroblasts and MyoD transcriptional activity with erasing repressive H3K9me3 level at the promoter of MyoD target genes. Collectively, Jmjd2C increases MyoD transcriptional activity to facilitate skeletal muscle differentiation by increasing MyoD stability through inhibiting G9a-dependent MyoD degradation. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Myogenic stem cell differentiation is a lineage decision process by which the embryonic and adult skeletal muscle development occurs. Lineage-specific differentiation of myogenic stem cells is governed by genetic networks of myogenic regulatory factors (MRFs), a family of basic helix-loop-helix transcription factors. It is well-known that epigenetic modifications regulate muscle specific gene expression [1]. Therefore, epigenetic modifiers must control the accessibility of MRFs to muscle gene promoters that is required for muscle cell differentiation. Among MRFs, the muscle specific transcription factor MyoD lies at the core of MRF networks [2,3]. Extensive investigations revealed that MyoD-dependent transcription is regulated primarily through MyoD stability [4–6] and epigenetic modification via G9a methyltransferase [7,8]. However, the causal connection between methylation and stability of MyoD and its transcriptional activity remains largely unstudied.

⁎ Corresponding authors. E-mail addresses: [email protected] (C.-H. Kim), [email protected] (K.-S. Kim). 1 These authors contributed equally.

http://dx.doi.org/10.1016/j.bbagrm.2015.07.001 1874-9399/© 2015 Elsevier B.V. All rights reserved.

Epigenetic modifications of proteins influence a diverse range of biological processes including muscle cell differentiation. One such modification, methylation, occurs on both lysine and arginine residues by protein lysine (KMTs) and arginine (RMTs) methyltransferases. KMTs illustrated below have a broad range of substrate specificity in histone proteins as well as non-histone proteins. Indeed, a number of studies have shown that the histone methyltransferase SET7 co-localizes and directly interacts with DNMT1 [9]. Thus, the level of methylated DNMT1 peaks during the S and G2 phases of the cell cycle, and it is prone to proteasome-mediated degradation. Likewise, Set7 methylates Sox2 and activates methylation-dependent Sox2 ubiquitination [10]. Consistent with the above, the methylation of FoxO3 by the histone methyltransferase SET9 decreases FoxO3 protein stability [11]. These findings clearly suggest that the methylation of non-histone proteins is coupled to protein degradation pathway [12]. Recently, Ling and colleagues demonstrated that MyoD is methylated by G9a methyltransferase, resulting in inhibition of muscle differentiation [7,8]. However, how methylated MyoD suppresses muscle differentiation remains unexplored. Studies on histone methylation reveal that the steady-state-level of histone lysine methylation is controlled by KTMs and lysine demethylases (KDMs) [13]. KDMs can be categorized into two families

1082

E.-S. Jung et al. / Biochimica et Biophysica Acta 1849 (2015) 1081–1094

A

C

B D0

D1

D2

D0

D3

D1

D2

MW 250 150

D0

D1

D3

D5

MW 250 150 100

Jmjd2C Jmjd2C

100

D

D3

D0

D1

D2

D3

JMJD2C Jmjd2A IB: JMJD2C

IB: Jmjd2C 250

250

MyoD

IB: MHC

Jmjd2B

150 100

150 100

IB: MHC

MyoG

Jmjd2C

50

50

37

37

Gapdh

IB: MYOD

Myf5

IB: MyoD

37 37

Myl6

IB: MYF5

IB: Myf5 37

37

Gapdh

IB: GAPDH

IB: Gapdh

E

F

MB I

JmjdD2C

MyoD

Jmjd2C+MyoD

DAPI

Merge

C

MT3 N

I

C

N

MW 250

Myoblasts

150 100 IB: Jmjd2C 250 150 IB: G9a MyoD

Jmjd2C+MyoD

DAPI

Myocytes

JmjdD2C

Merge

20 15 IB: Histone3 37 IB: Gapdh

Fig. 1. Skeletal muscle differentiation is accompanied by the induction and nuclear localization of Jmjd2C in C2C12 cells. (A) The protein expression patterns of Jmjd2C, MHC, MyoD, and Myf5 during C2C12 muscle cell differentiation. Differentiation was induced by transferring the cells to DMEM supplemented with 4% horse serum. Gapdh was used as a protein loading control. (B) The mRNA expression patterns of Jmjd2C, MyoD, MyoG, Myf5, and Myl6 during C2C12 muscle cell differentiation. Gapdh was used an internal mRNA control. (C) The expression patterns of JMJD2C, MHC, MYOD, and MYF5 during human skeletal muscle differentiation. GAPDH was served as a protein loading control. The samples were manipulated as in (A). (D) The mRNA levels of Jmjd2A, Jmjd2B, and Jmjd2C increases during C2C12 muscle differentiation. Gapdh was served as an internal mRNA control. (E) The immunostaining analysis shows the differential distribution patterns of Jmjd2C and MyoD in C2C12 myoblasts and myotubes. Distribution patterns were visualized after staining with primary antibodies of anti-Jmjd2C and anti-MyoD, and secondary antibodies of AlexaFluor 594- and 488-conjugated goat anti-mouse or rabbit IgG along with DAPI, and the images were captured with a LSM 510 META confocal laser scanning microscope. (F) A subcellular fractionation assay shows that Jmjd2C distributes differently in C2C12 myoblasts and myotubes, but G9a localizes mainly to the nuclear fraction. Histone H3 and Gapdh were used as nuclear and cytoplasmic markers, respectively. I: input; C: cytoplasmic fraction; N; nuclear fraction.

based on the type of cofactor, FAD and iron; FAD-dependent monoamine oxidases are such as LSD1/KDM1A and LSD2/KDM1B, while the list of iron-dependent dioxygenases includes the JmjC domaincontaining demethylase (Jmjd) family. The Jmjd2 family is composed of six members: Jmjd2A, Jmjd2B, Jmjd2C, Jmjd2D, Jmjd2E, and Jmjd2F. Recent studies have shown that the Jmjd2 family demethylates histone H3K9me3, H3K9me2, and H3K36me3 [14,15]. Especially, demethylation of H3K9me3 by Jmjd2C is reported to relieve chromatin compaction by recruiting epigenetic writers and their readers such as HP1α and KAP-1 [16]. For instance, Jmjd2C-mediated reduction of H3K9me3 level and thereby chromatin structure alteration is very important in maintaining mouse embryonic stem cells (mESCs), since its deficiency

disrupts mESC pluripotency and induces differentiation [17–19]. These findings suggest that Jmjd2C plays a key role in the maintenance of self-renewal potential of stem cells by lowering H3K9m3 level. Emerging evidence indicates that non-histone proteins are also substrates of KMTs [9–12]. Although histone modifications have been intensively studied, the epigenetic regulation of non-histone proteins in muscle cells remains to be investigated. Thus, in this study we demonstrate that Jmjd2C demethylates MyoD that is methylated by G9a. Once methylated, MyoD is subject to Cul4/Ddb1/Dcaf1-mediated degradation. On the other hand, demethylated MyoD by Jmjd2C is resistant to Cul4/Ddb1/Dcaf1-mediated ubiquitinatoin and degradation. Therefore, Jmjd2C promotes MyoD transcription activity by increasing MyoD

Fig. 2. Jmjd2C associates with MyoD. (A) GST-MyoD binds to Jmjd2C in a pull-down assay. GST-MyoD with or without Jmjd2C-3xHA were transfected to 293T cells. A pull-down assay was performed as described in the Materials and methods section. (B) The JmjN, JmjC, and 2xPHD domains of Jmjd2C associates with MyoD. GST-MyoD with C-terminal 3xHA tagged Jmjd2C mutants were transfected to 293T cells. The samples were manipulated as in (A). (C) A schematic representation of MyoD binding sites and relative binding strength in Jmjd2C. (D) GSTJmjd2C associates with MyoD in a pull-down assay. The samples were manipulated as in (A). (E) Jmjd2C binds to the basic, HLH, and C-terminal domains as well as full-length MyoD. The samples were manipulated as in (A). (F) A schematic representation of Jmjd2C binding sites and relative binding strength in MyoD. (G) Immunoprecipitation experiments show the differential association of MyoD with Jmjd2C and G9a in C2C12 myoblasts and myotubes. But, GSK3β as a negative control does not associate with MyoD. An immunoprecipitation assay was performed as described in the Materials and methods section.

E.-S. Jung et al. / Biochimica et Biophysica Acta 1849 (2015) 1081–1094

A

1083

B 1% Input

PD GST-Con GST-MyoD Jmjd2C-3xHA

+ +

WCL

+ +

PD/Glutathione

+ +

+

+ + -

+ + -

+ +

+ -

+ -

+ -

+ -

+ -

+ -

+

-

-

-

-

Low-3xHA 2xPHD-3xHA 2xTUD-3xHA Jmjd2C-3xHA

-

-

-

+ -

+ -

+ -

+

-

-

-

+ -

+ -

+ -

+

+

MW 250 150 100 75 50

WCL

GST-MyoD 3xHA JmjN-3xHA JmjC-3xHA

37

+

+

+

+

+

+

+

+

+

MW 250 150 100 75

25 20 IB: HA

IB: HA

IB: GST

50

37

37 IB: GAPDH

C

1

140

310

670

Jmjd2C JmjN JmjC

Low

25

868

1056

Binding to MyoD

2xPHD 2xTUD

IB: HA

IB: HA

IB: GST

+ + +

JmjN

+ +

JmjC

+ +

Low

20

37 IB: GAPDH

-

2xPHD

+ + +

2xTUD

-

D

E

GST-Con GST-Jmjd2C MyoD-3xHA

1% Input

PD + +

WCL + +

PD/Glutathione + -

+

GST-Jmjd2C 3xHA MyoD1-115-3xHA MyoD116-166-3xHA MyoD167-318-3xHA MyoD-3xHA

+

+

MW 250 150 100 75

WCL

+ +

+ -

+ -

+ -

+ -

+

-

-

-

-

+

-

+ -

+ -

+ -

+

-

+ -

+ -

+ -

+

+

+

MW 250 150 100 75

50 37 25

50

20

37 IB: HA

IB: HA

IB: GST 25

37

20 IB: HA

IB: GAPDH

IB: HA

IB: GST 37 IB: GAPDH

G

F 116

1 MyoD MyoD1-115

Basic

IP

Binding to 318 Jmjd2C + + + C-terminal

166

HLH

WCL MB

+ + +

MT3

α rabbit IgG MB

MT3

α rMyoD MB

MT3

MW 250 MyoD116-166

+ + +

MyoD167-318

++

150 100

Jmjd2C IB: Jmjd2C

250

G9a

150 100 IB: G9a 50

GSK3β IB: GSK3β

50 37 IB: mMyoD

1084

E.-S. Jung et al. / Biochimica et Biophysica Acta 1849 (2015) 1081–1094

A

B In vitro methylation 3 H-SAM

GST-MyoD GST-Con GST-G9a SET

+ + -

GST-MyoD GST-Con GST-wt Jmjd2C

+ + -

+ +

+ -

GST-190/192 Jmjd2C

-

-

+

1.00

0.25

1.09

+ +

75

5-Carboxy-8HQ

+

-

-

-

GST-MyoD 3xHA Jmjd2C-3xHA

+ + -

+ -

+ +

+ ++

75 3

H-SAM autoradiography MW 250 150 100

MW 250 150 100

50

+

IB: Mono/di-methyl lysine

+

+ +

-

++

250 150 100 75

Jmjd2C MyoD

50

Jmjd2C

75

37 IB: GST

G9a SET MyoD

75

+

-

MW 75

3 H-SAM autoradiography

G9a SET MyoD

75

WCL

PD: Glutathione

In vitro demethylase assay

25 20

37

50

25

IB: GST IB: GST

GST

IB: HA

37

IB: GST

IB: GAPDH

D

E

F

PD: Glutathione

-

-

-

+

+

+

+

+

190/192 Jmjd2C-3xHA

-

+

-

-

wt Jmjd2C-3xHA

-

-

+

-

+ + -

GST-MyoD 3xHA Jmjd2C-3xHA

+ +

MW 250 150 100

MW 75 IB: Mono/di-methyl lysine

pSingle-Luc

+

-

GST-MyoD

pSingle-Jmjd2C

-

+

H1166K G9a-GFP wt G9a-GFP

+ -

+ -

+ + -

+ +

HA-Ub

-

+

+

+

MW 250 150 100

Ub-MyoD

UNC0638 GST-MyoD

PD: Glutathione

IP: rMyoD

Ub-MyoD

PD: Glutathione

75

75

MW 250 150 100

50

75

50

IB: Ub

IB: Ub

IB: GST

Ub-MyoD

C

75 IB: HA

50 75

WCL GST-MyoD

+

+

+

+

190/192 Jmjd2C-3xHA

-

+

-

-

wt Jmjd2C-3xHA

-

-

+

-

37

IB: GST WCL

WCL GST-MyoD Jmjd2C-3xHA

MW 250 150 100

+ -

IB: Jmjd2C

IB: GST

IB : MyoD

+ + -

+ +

IB: GFP 75 IB: Gapdh IB: HA

IB: GAPDH

+ -

37

150 100

37

+ -

MW 250 150 100

37

MW 250

IB: GST

GST-MyoD H1166K G9a-GFP wt G9a-GFP

50

50

75

WCL

MW 250 150 100

+ +

MW 100 75

IB: HA

75

IB: mMyoD

IB: GST

IB: GST

37

37 IB: GAPDH IB: GAPDH

IP: HA

H

WCL

GST-MyoD

+

+

+

+

+

+

+

5xMyc-wt Jmjd2C 5xMyc-190/192 Jmjd2C wt G9a-GFP H1166K G9a-GFP MG132

-

-

+

+ -

+ -

+ -

+ -

HA-Ub

-

+

+

+

+

+

+

+ -

+ -

+ -

+ + -

+ + -

+ + -

+ +

IP: HA MyoD-3xHA BIX01249 HA-Ub

+ -

+ +

MW 250 150 100

Ub-MyoD

100

250 150 100

75 IB: GFP / Myc

75 IB: HA

75

IB: GST

37

IB: GST

75

IB: HA

G9a Jmjd2C

IB: GAPDH

+

+

+

75 IB: GST 37

75 MW 250 150

WCL + + + Ub-MyoD

G

IB: GST

IB: GAPDH

E.-S. Jung et al. / Biochimica et Biophysica Acta 1849 (2015) 1081–1094

stability and myogenic conversion of mouse embryonic fibroblasts. Additionally, Jmjd2C promotes MyoD transcriptional activity by removing repressive H3K9me3 level. In summary, Jmjd2C promotes muscle differentiation by stabilizing the transcription factor MyoD through inhibiting G9a-dependent MyoD degradation. 2. Materials and methods 2.1. Cell culture, plasmids, shRNAs, and transfection G9a−/− and G9a−/− + G9a(L) mESCs were obtained from Dr. Makoto Tachibana (Kyoto University, Japan) and cultured in DMEM containing 15% FBS, LIF, and 2i inhibitors [20]. C2C12 and 293T cells were cultured in DMEM media supplemented with 10% FBS. Human primary myoblasts (cat # A12555) were obtained from Invitrogen and cultured with human skeletal muscle cell growth media (cat # SKM-M from Zenbio). After isolating mRNA from mouse C2C12 cells, cDNA was made with Superscript II and oligo dT20 primer according to the manufacturer's protocol. The coding sequences (CDS) of MyoD, Jmjd2C, G9a, and Ub were amplified with gene-specific PCR primers (Table S1). The amplicons of MyoD, Jmjd2C, and G9a were digested with SalI and NotI, whereas Ub was digested with XhoI and NotI. The digested amplicons were cloned into SalI/NotI-digested N-terminal GST-, N-terminal 5xMyc-, C-terminal 3xHA-, or N-terminal 3xFLAGtagged vectors. SalI/BamHI-digested pEGFP-N1 was used to clone G9a. For knocking down endogenous Jmjd2C and Dcaf1 in mouse C2C12 cells, we utilized the pSingle-tTS-shRNA vector system (Clontech) with gene-specific oligonucleotide sets (Table S1) according to the manufacturer's instructions. We transfected the plasmids into 293T cells using polyethylenimine (1 μg/μl PEI, 25-kD linear form from Polysciences, Cat# 24765). A 2.5:1 ratio of PEI (μg): total DNA (μg) in Opti-MEM was used. Lipofectamine 2000 was used to transfect the C2C12 and human myoblast cells. The growth medium in each well of a 12-well dish was replaced with 400 μl of the Opti-MEM mixture that contains 1:3 DNA:Lipofectamine 2000, and the cells were incubated for 1 h. One hour after transfection, the Opti-MEM mixture was removed, and the cells were washed once with growth medium and then cultured to reach ~ 80% confluence. This modified transfection method produced an 80% transfection efficiency based on GFP fluorescence using pEGFP-N1. Myoblasts were transiently cotransfected with the 4RE-luciferase reporter and pCMVβ (as a control for transfection efficiency and sample handling) in a 10:1 ratio using Lipofectamine 2000. Luciferase assays were performed using a firefly-luciferase kit (Promega). Firefly luciferase activity was normalized to βgalactosidase activity. To induce differentiation, differentiation medium (DMEM supplemented with 4% horse serum) of C2C12 and human primary myoblast cells was changed everyday for the indicated periods of time. Unless otherwise indicated, all of the experiments were repeated at least three times. 2.2. Reverse transcription-polymerase chain reaction (RT-PCR) and Western blot Total RNA was extracted with TRIzol reagent (Invitrogen) according to the manufacturer's protocol. RNA pellets were dissolved in 0.1% diethyl pyrocarbonate-treated water. The RNA concentration of each sample was determined with a NanoDrop. A reverse-transcription

1085

reaction was performed with 1 μg of RNA and SuperScript II Reverse Transcriptase (Invitrogen) after incubating the samples with DNase I to eliminate any contaminating genomic DNA. The cDNA was amplified by the Platinum PCR Master Mix (Invitrogen) using the appropriate primer sets (Table S2). The amplification conditions were as follows: 5 min at 95 °C, denaturing (94 °C, 30 s), annealing (55 °C, 30 s), and extension (72 °C, 30 s) cycles with a final extension at 72 °C for 5 min. The PCR products were visualized with ethidium bromide after electrophoresis on a 2% agarose gel. For Western blot analysis, the plasmids were transfected into 293T, C2C12, or human myoblast cells using PEI or Lipofectamine 2000. Two days after transfection, the cells were washed with PBS and lysed with RIPA buffer. The cell lysates were used for Western blotting or further experiments. Twenty-microgram protein samples from each lysate were subjected to SDS-polyacrylamide gel electrophoresis, transferred onto a PVDF membrane, and probed with appropriate antibodies. Proteins were detected using HRP-conjugated secondary antibodies and ECL reagents. The bands were visualized using blue X-ray film or an ImageQuant LAS 4000 (GE Healthcare). 2.3. GST pull-down and immunoprecipitation, immunocytochemistry, antibodies, and Inhibitors Molecular cloning was performed as described [21]. For the pulldown analysis, we generated a mammalian GST expression system. GST-transgenes were transfected into 293T cells using PEI transfection. Two days after transfection, the cells were lysed with RIPA buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 1% sodium deoxycholate, 1% Triton X-100, and 0.1% SDS). After the cell lysates were clarified using the centrifuge, Glutathione Sepharose 4B was added to the lysates, then incubated overnight at 4 °C. After the beads were washed with RIPA buffer three times, 1×SDS sample buffer was added, and Western blot analysis was performed. To identify endogenous protein interactions, the C2C12 lysates were treated with DNase I to remove genomic DNA contamination and incubated with 1 μg of anti-MyoD antibody. One microgram of mouse IgG was added to myoblast lysates as a control. For the immunocytochemistry assays, the cells were fixed with 4% paraformaldehyde in PBS for 15 min, permeabilized with 0.2% Triton X-100, and blocked using 1% BSA in PBS containing 0.1% Triton X-100. The permeabilized cells were incubated with appropriate antibodies overnight at 4 °C. Samples were washed three times with 0.1% Triton X-100 in PBS, then incubated with secondary antibodies (AlexaFluor 488- or 594conjugated goat anti-mouse IgG (488: A-11001, 594: A-11005, Invitrogen) or AlexaFluor 488- or 594-conjugated goat anti-rabbit IgG (488: A-11008, 594: A-11012)) at room temperature for 1 h. Cell nuclei were stained with DAPI for 10 min. Images were obtained using a Nikon fluorescence microscope (model ECLIPSE Ti) and a LSM 510 META confocal laser scanning microscope (Carl Zeiss). The following inhibitors were used in this study: 5 μM of BIX01249 (Stemgent, 04-0002), 1 μM of UNC0638 (Sigma, U4885), 2 μM of 5-carboxy-8HQ (Merck, 420201), 50 μg/ml of cycloheximide (Sigma, C4859), and 2.5 μM of MG132 (Merck, 474790). List of the antibodies used in this study is as follows: anti-FLAG (Sigma, F3165); anti-Myc, anti-MyoD, and antiMHC (Developmental Studies Hybridoma Bank, 9E10, D7F2, and MF20, respectively); anti-MyoD, anti-Myf5, anti-Ub, anti-β-Actin, antiGFP, and anti-HA (Santa Cruz Biotechnology, sc-760, sc-302, sc-9133, sc-47778, sc-9996, and sc-7392, respectively); anti-Jmjd2C (Bethyl, A300-885A; Santa Cruz Biotechnology, sc-98678); anti-G9a (Millipore,

Fig. 3. Jmjd2C decreases MyoD methylation and ubiquitination. (A) Jmjd2C demethylates MyoD directly in vitro demethylation assay. An in vitro methylation and demethylation assays were performed as described in Materials and Methods section. (B) Jmjd2C expression leads to the reduction of MyoD methylation. GST-MyoD with or without Jmjd2C-3xHA were transfected to 293T cells. One day before harvesting cells, the Jmjd2C inhibitor 5-carboxy-8HQ was added to transfected 293T cells. A pull-down assay was performed as described in the Materials and methods section. (C) Wt Jmjd2C and UNC0638, a G9a specific inhibitor, reduce the methylation levels of MyoD compared to H190G/E192A Jmjd2C. The samples were manipulated as in (B). (D) Jmjd2C reduces MyoD ubiquitination using an ubiquitination assay. GST-MyoD with or without Jmjd2C-3xHA was transfected to 293T cells. The samples were prepared as in (B). (E) Silencing of endogenous Jmjd2C by pSingle-Jmjd2C,a doxycycline-inducible and Jmjd2C-specific shRNA, increases MyoD ubiquitination in C2C12 myoblast cells. (F) Wt G9a but not H1166K G9a, a catalytically inactive mutant, increases MyoD ubiquitination determined by an ubiquitination assay. The samples were manipulated as in (B). (G) Wt G9a, H1166K G9a, wt Jmjd2C, and H190G/E192A Jmjd2C show the different effects on MyoD ubiquitination. (F) The G9a/GLP inhibitor BIX01249 suppresses MyoD ubiquitination. The samples were prepared as in (B).

1086

E.-S. Jung et al. / Biochimica et Biophysica Acta 1849 (2015) 1081–1094

A MyoD-3xHA 3xHA G9a-GFP

+ + -

+ + -

+ + -

+ + -

+ + -

+ +

+ +

+ +

+ +

+ +

+ -

+ -

+ -

+ -

+ -

5xMyc-Jmjd2C

-

-

-

-

-

-

-

-

-

-

+

+

+

+

+

0 1/2 1

2

4

0 1/2 1

2

4

0 1/2 1

2

4

50 μg/ml CHX

-

-

-

-

-

+

+

+

+

+

-

-

-

-

-

0 1/2 1

2

4

0 1/2 1

2

4

0 1/2 1

2

4

Con G9a-GFP 5xMyc-Jmjd2C

hrs

MW 250 150 100 75

0.8

50

0.6

37

0.4

1

0.2 25

0

20

0

1/2

1

2

4

hrs

IB: GFP

IB: HA / Myc 37

IB: GAPDH

B BIX01249

Con BIX01249 5-Carboxy-8HQ

5-Carboxy-8HQ

MyoD-3xHA

+

+

+

+

+

+

+

+

+

+

+

+

+

+

+

50 μg/ml CHX

0

1

2

3

4

0

1

2

3

4

0

1

2

3

4

hrs

1 0.8

MW 50

0.6 IB: HA

0.4 0.2

37

0

IB: GAPDH

C

0

1

2

3

4

hrs

E MyoD-3xHA

+

+

+

+

+ +

+

+

+

+

+ +

wt G9a-GFP H1166K G9a-GFP

+ -

+ -

+ -

+ -

+ + - -

+

+

+

+

- + +

0 1/2 1

2

4 6

0 1/2 1

2

4

50 μg/ml CHX

6

mESCs

G9a-/-

hrs

MW 250 150 100

MyoD-3xHA

+

+

+

pEGFP-N1

+

-

-

Jmjd2C-3xHA

-

+

-

G9a-GFP

-

-

+

wt G9a-GFP H1166K G9a-GFP IB: GFP

75 50

1

MW 50

0.8

MyoD/Gapdh ratio

IB: HA 1.00

1.98

0.53

0.6

37

WCL

0.4 25

0.2

20

0

IB: HA

250 0

1/2

1

2

IB: HA/GFP

150

6 hrs

4

37

37 IB: GAPDH

D

20

MyoD-3xHA 5xMyc-wt Jmjd2C 5xMyc-190/192 Jmjd2C

+ + -

+ + -

+ + -

+ + -

+ + -

+ +

+ +

+ +

+ +

+ +

50 μg/ml CHX

0

1

2

4

6

0

1

2

4

6

MW 250 150 100

IB: Gapdh

IB: H3K9me2

15

wt Jmjd2C-3xHA 190/192 Jmjd2C-3xHA

hrs

F G9a-/-

1 IB: Myc

75

0.8

MyoD-3xHA DMSO UNC0638

0.6

50

0.4

37

0 IB: HA

37

MyoD/Gapdh ratio 0

1

2

4

6

hrs

IB: HA 1.00

1.69

WCL 37

IB: GAPDH

mESCs

+ +

MW 50

0.2 25 20

+ + -

IB: Gapdh

E.-S. Jung et al. / Biochimica et Biophysica Acta 1849 (2015) 1081–1094

09-071); anti-Dcaf1/VPRBP (Cell Signaling, 14966); anti-H3K9me3 and anti-histone H3 (Abcam, ab6001 and ab1791, respectively); antimono-/di-methyl lysine (Abcam, ab23366); and anti-Gapdh (Cell Signaling, 2118). 2.4. In vitro methylation, demethylation, ubiquitination, and subcellular fractionation In vitro methylation reaction was conducted as follows: GST-MyoD and GST-G9a(SET) were transfected into 293T cells, individually isolated by glutathione beads, mixed together, and then incubated in methylation buffer (50 mM Tris-HCl pH 8.8, 5 mM MgCl2, and 4 mM DTT) with S-adenosyl-L-[methyl-3H] methionine (85 Ci/mmol from a 0.5 mCi/ml stock solution, Perkin-Elmer) for 6 h at room temperature. After methylation reaction, the isolated GST, GST-wt Jmjd2C, or GST-190/192 Jmjd2C was added to the methylated GST-MyoD, and then incubated in demethylation buffer [20 mM Tris-HCl pH 7.3, 150 mM NaCl, 50 μM (NH4)2Fe(SO4)2°6H2O, 1 mM α-ketoglutarate, 2 mM ascorbic acid, and 10% glycerol] for 5 h at room temperature. The reaction was stopped by addition of 2×SDS sample buffer, and Western blot analysis was performed. For ubiquitination assays, GST-MyoD and MyoD-3xHA with HA-Ub and 3xFLAG-Ub were transfected into 293T cells, respectively. MyoD ubiquitination was detected using anti-HA or anti-FLAG antibodies after the pull-down or immunoprecipitation experiments. After the beads were washed with RIPA buffer three times, 1×SDS sample buffer was added, and Western blot analysis was performed. The subcellular fractionation was performed as described in the Abcam protocol provided by Dr. Richard Pattern. Briefly, the cells were harvested with fractionation buffer (250 mM sucrose, 20 mM HEPES pH 7.5, 10 mM KCl, 1.5 mM MgCl2, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, and protease inhibitor cocktail) and passed through a 25-G needle, and put on ice for 20 min. Lysates were centrifuged at 1,300 ×g for 5 min to obtain cytosolic and nuclear fraction. The cytosolic supernatant was further centrifuged at 15,000 ×g for 20 min and the supernatant, cytosolic fraction, was collected. The nuclear pellet was resuspended with nuclear lysis buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, and 10% glycerol) and incubated on ice for 30 min for nuclei lysis. Thereafter, nuclear fractions were isolated using centrifugation at 10,000 ×g for 10 min. The fractionation efficiency was examined by measuring quantitative enrichment of Gapdh for the cytoplasmic fraction and histone H3 for the nuclear fraction. 2.5. Myogenic conversion assay Embryonic STO fibroblast cells were used for myogenic conversion assays. Briefly, 2 μg of indicated the DNAs were transfected into to STO fibroblast cells cultured in a 12 well dish using Lipofectamine 2000 according to the manufacturer's protocol. Forty-eight hours after transfection, the cells were switched to DMEM containing 4% horse serum for 72 h, then processed for firefly-luciferase assays or the immunocytochemistry with anti-MF20 antibody to examine the myogenic conversion of fibroblasts. 2.6. ChIP assay Chromatin immunoprecipitation (ChIP) assays were carried out using the Abcam X-ChIP protocol. C2C12 cells were incubated with 1%

1087

formaldehyde in phosphate-buffered saline (PBS) at room temperature for 10 min. The cross-linking reaction was stopped by adding of glycine to a final concentration of 0.125 M. After sonicating samples, 1 μg of anti-MyoD, anti-H3K9me3, and rabbit pre-Immune IgG were added in incubation step. Eluted DNA was used as template for real-time PCR with the following primers encompassing the E-box element as in our earlier report [22,23]: Chrna1/AChRα (237 bp), sense, 5′ GAC AAG CCT CTG ACT CAT GAT CTA TGT, antisense, 5′ GCT GCC GGT CCT ACT CCA CCC TGG CT; Myh1 (306 bp), sense, 5′ TGA GTA GGG ACC TGG CTT TG, antisense, 5′ GCA CCC CAG CTT CAC TTT TA; Myl6 (133 bp), sense, 5′ CCT CAG TTA CCC GAA GGT CA, antisense, 5′ CTT GCC TCT CAA GCG GAT AC; and Tnnc1 (129 bp), sense, 5′ GGA ATG TAG CAG GAG GTG GA, antisense, 5′ TCA CCG CAG CTT TGT AGA TG. 2.7. Statistics Data represent mean ± SEM. Groups were compared using the Student's t-test for parametric data. A value of P b 0.01 was considered statistically significant. 3. Results 3.1. Jmjd2C is a myotube-enriched protein during muscle differentiation G9a methyltransferase mono- and di-methylates a master transcription factor MyoD, and suppresses skeletal muscle differentiation by an unknown mechanism [7]. Since emerging evidence suggests that nonhistone proteins are also substrates of Jmjd2C [24–28], we hypothesized that the Jmjd2C demethylase might reverse the methylation state of MyoD. To test this hypothesis, we investigated the role of Jmjd2C/ Kdm4C in muscle differentiation that MyoD promotes. Mouse skeletal muscle differentiation led to the increased expression of the muscle differentiation marker MHC (Fig. 1A, second panel). Coincident with this, both Jmd2C and MyoD levels increased during mouse muscle differentiation (Fig. 1A). In contrast, Myf5 that acts as a myoblast commitment factor was decreased as muscle differentiation progressed (Fig. 1A). We also examined changes of the mRNA expression levels of Jmjd2C, MyoD, MyoG, Myf5, and Myh6 using RT-PCR analysis (Fig. 1B). Overall, the protein expression patterns of JMJD2C, MYOD, and MYF5 in human primary skeletal muscle cells were similar to those of the three proteins in mouse muscle (Fig. 1C). A previous study indicated that JMJD2A protein decreases during human muscle cell differentiation, contrasting to mouse C2C12 cells in which Jmjd2A protein increases during muscle differentiation [29]. To identify the expression pattern of the mouse Jmjd2 family, we examined mRNA expression patterns using RT-PCR analysis. Consistent with the increased protein level of Jmjd2C (Fig. 1A, first panel), the mRNA levels of the Jmjd2 family members including Jmjd2A, Jmjd2B, and Jmjd2C gradually increased during C2C12 skeletal muscle differentiation (Fig. 1D). These findings suggest that the Jmjd2 family may be implicated in skeletal muscle differentiation. Most nuclear proteins act as shuttle proteins that localize to both the cytoplasm and the nucleus [30]. To test this notion, we performed an immunostaining to look into the endogenous localization of Jmjd2C and MyoD in skeletal muscle cells. Mouse Jmjd2C appeared to be distributed throughout the entire myoblasts but was mostly localized in the nuclei of myotubes (Fig. 1E). On the other hand, MyoD was exclusively localized in the nuclei. Thus, MyoD exclusively co-localized with

Fig. 4. Hypo-methylated MyoD exhibits a long half-life. (A) Jmjd2C increases the half-life of MyoD, whereas G9a shortens the half-life of MyoD. Two days after transfection, 50 μg/ml of CHX was administered for the indicated times before collection of protein samples. The graph shows relative quantification of MyoD levels analyzed with the ImageJ program, MyoD level at time 0 h is calculated as 1. Error bars represent standard deviation. Experiments were repeated three times, *P b 0.01. (B) The G9a inhibitor BIX01249 increases the half-life of MyoD more than does the Jmjd2C inhibitor 5-carboxy-8HQ. The samples were manipulated as in (A). (C) The catalytically inactive G9a mutant H1166K G9a increases the half-life of MyoD more than wt G9a does. The samples were manipulated as in (A). (D) The catalytically inactive Jmjd2C mutant H190G/E192A Jmjd2C does not significantly decrease the half-life of MyoD. The samples were manipulated as in (A). (E) MyoD expression is dependent on Jmjd2 or G9a expression in G9a−/− mESCs. Exogenous MyoD-3xHA, Jmjd2C-3xHA, or G9a-GFP was transfected into G9a−/− mESCs using Lipofectamine 2000. (F) GLP may be involved in MyoD methylation in the same way as G9a. The G9a/GLP inhibitor UNC0638 increases MyoD expression in G9a−/− mESCs.

1088

E.-S. Jung et al. / Biochimica et Biophysica Acta 1849 (2015) 1081–1094

A

E

C2C12 MB DMSO

+

-

UNC0638

-

+

D1 D2 D3

IB: MyoD 37 MyoD/Gapdh ratio

1.0

+ -

5xMyc 5xMyc-Jmjd2A 5xMyc-Jmjd2B 5xMyc-Jmjd2C

MW 50

2.6

H

C2C12

+ -

+ -

D2 D3

D2 D3

+ -

+ -

+ -

UNC0638

DMSO

5-Carboxy-8HQ

D2 D3

+ -

+

+

MW 250

37

150 100

IB: Gapdh

IB: MHC

B

250

DMSO UNC0638 5-Carboxy-8HQ

+ -

+ -

60 Myogenic Index (%)

C2C12 MT3

150 100

+

IB: Myc 37

MW 50 IB: MyoD

37

IB: Gapdh

MyoD/Gapdh ratio

1.00

2.35

0.72

37

IB: Gapdh

F pEGFP-N1

C

5xMyc-G9a

40

20

0 DMSO

+

-

-

UNC0638 5-Carboxy-8HQ

-

+ -

+

5xMyc-Jmjd2C

hSK

I

C2C12

D0 D1 D3 D5 D1 D3 D5 3xHA JMJD2C-3xHA

+ -

+ -

+ -

+ -

+

+

D0 D1 D2 D3

+

pSingle-Luc pSingle-Jmjd2C

MW 250

+ -

+ -

+ -

D1 D2 D3

+ -

+

+

+

MW 250 Myogenic Index (%)

150 100 IB: MHC 250 150 100 IB: HA 37 IB: GAPDH

D

150 100

30

IB: MHC 250

20

150 100

10

Jmjd2C IB: Jmjd2C

0 pEGFP-N1

+

-

-

5xMyc-G9a 5xMyc-Jmjd2C

-

+ -

+

37 IB: Gapdh

hSK Con

G

5-Carboxy-8HQ

C2C12

D0 D1 D3 D5 D1 D3 D5

D0 D1 D2 D3 D1 D2 D3 D1 D2 D3

MW 250 150 100

IB: MHC

pEGFP-N1 5xMyc-Jmjd2C 5xMyc-G9a

+ -

+ -

+ -

+ -

+ -

+ -

+ -

+

+

J

C2C12 Con

+

5-Carboxy-8HQ

D0 D1 D2 D3 D1 D2 D3 MW 250

37

IB: GAPDH

MW 50

150 100

37

IB: MHC

IB: MyoD 250 150 100

37 IB: Myc

IB: Gapdh

37 IB: Gapdh 20 15 IB: H3K9me3 20 15 IB: Ponceau S (H3)

Fig. 5. Jmjd2C induces muscle differentiation. (A) The G9a inhibitor UNC0638 increases endogenous MyoD in C2C12 myoblasts. The numbers indicate the relative quantitation of ratio MyoD normalized by Gapdh analyzed by the ImageJ program. MyoD/Gapdh ratio of the control is calculated as 1. Experiments were repeated three times, *P b 0.01. (B) The G9a inhibitor UNC0638 but not the Jmjd2 inhibitor 5-carboxy-8HQ increases endogenous MyoD in C2C12 myotubes. The samples were manipulated as in (A). (C) JMJD2C expression activates human muscle differentiation. JMJD2C-3xHA was transfected to human myoblast cells using Lipofectamine 2000. Differentiation was induced by transferring the cells to DMEM supplemented with 4% horse serum. The samples were manipulated as in (A). (D) The Jmjd2C inhibitor 5-carboxy-8HQ decreases human skeletal muscle differentiation. The samples were manipulated as in (C). (E) Expression of Jmjd2 family members activates C2C12 muscle differentiation. The samples were manipulated as in (C). (F) Jmjd2C but not G9a increases muscle differentiation determined by the myogenic index with the representative figures of C2C12 myotubes expressing pEGFP-N1, 5xMyc-G9a, and 5xMyc-Jmjd2C. (G) Jmjd2C expression increases MyoD protein level and decreases H3K9me3 level during C2C12 differentiation, but G9a does not. (H) The G9a/GLP inhibitor UNC0368 increases muscle differentiation, whereas the Jmjd2 inhibitor 5-carboxy-8HQ decreases it. The samples were manipulated as in (F). (I) Jmjd2C silencing by pSingle-Jmjd2C expression compromises C2C12 muscle differentiation. (J) Treatment with 5-carboxy-8HQ reduces C2C12 muscle differentiation.

E.-S. Jung et al. / Biochimica et Biophysica Acta 1849 (2015) 1081–1094

Jmjd2C in the nuclei of myotubes (Fig. 1E, lower panel). The localization patterns of mouse Jmjd2C we observed were unexpected because Jmjd2C has been reported to localize exclusively in the nucleus [14, 15]. This finding implies that Jmjd2C could be a shuttle protein in muscle cells. When the protein sequence of Jmjd2C was analyzed by a web-based program, NucPred (www.sbc.su.se/~maccallr/nucpred/), NLS (nuclear localization sequence) within the low complexity domain was found (Supplementary Fig. 1A). To prove the prediction, we investigated localization of various Jmjd2C truncated mutants. Jmjd2C consists of five protein domains: the jumonji (Jmj) C, JmjN, low complexity (Low), 2xplant homeodomain (2xPHD), and 2xTUDOR domain. The mutants of C terminal GFP-tagged JmjN and JmjN + JmjC were distributed throughout the cells, while the JmjN + JmjC + Low mutant was localized to the nucleus (Supplementary Fig. 1B), indicating that the low complexity domain has the nuclear localization signal (NLS) indeed. As expected, the low complexity domain containing mutants, JmjN + JmjC + Low + 2xPHD and full-length Jmjd2C, localized primarily to the nucleus (Supplementary Fig. 1B). We also examined the distribution patterns of full length 5xMyc tagged Jmjd2A, Jmjd2B, and Jmjd2C in 293T cells. Most Jmjd2A, Jmjd2B, and Jmjd2C localized to the nucleus, but some proteins seemed to localize in the cytoplasm (Supplementary Fig. 2A). To determine conclusively the localization patterns of these proteins, we used Jmjd2CGFP. Most GFP signal was found in the nuclei while only weak GFP signal was detected in the cytoplasm of cells expressing Jmjd2C-GFP (Supplementary Fig. 2B). This result was confirmed by a subcellular fractionation assay, demonstrating that Jmjd2C-GFP was in the nuclear fraction as well as in the cytoplasmic fraction (Supplementary Fig. 2C). When leptomycin B (LMB), a specific nuclear export inhibitor, was added to the cells, the cytoplasmic fraction of Jmjd2C-GFP was diminished. To examine this phenomenon further, we conducted a subcellular fractionation assay using C2C12 extracts. G9a is highly expressed at myoblast stage, but its expression is decreased with differentiation [7]. G9a appeared exclusively in the nuclear fraction of myoblasts (Fig. 1F, second panel). However, Jmjd2C showed different patterns when compared with G9a. Consistent with our Jmjd2C staining data (Fig. 1E), Jmjd2C existed in both the cytoplasm and the nucleus in myoblasts, but occurred mainly in the nucleus in myotubes (Fig. 1F, first panel), indicating that Jmjd2C is a shuttle protein localized to both the cytoplasm and the nucleus in accordance with differentiation status. Taken together, MyoD and Jmjd2C protein levels increase during skeletal muscle differentiation, and Jmjd2C is a shuttle protein co-localizing with MyoD mainly in the nucleus in differentiated skeletal myotubes. 3.2. Jmjd2C associates with MyoD Considering that G9a methylates MyoD to suppress muscle differentiation and that MyoD co-localizes with Jmjd2C, we wanted to look into whether MyoD methylation might be reversed by Jmjd2C. Thus, we first examined the protein-protein interactions using the pull-down and immunoprecipitation assays. GST-MyoD bound to full length Jmjd2C3xHA in a pull-down assay (Fig. 2A). To map the MyoD-binding domains, Jmjd2C truncation mutants were constructed and expressed as C-terminal 3xHA-tagged proteins. GST-MyoD was found to be associated with three truncation mutants containing JmjN, JmjC, or 2xPHD domain of Jmjd2C (Fig. 2B and C). In a complementary experiment, GSTJmjd2C interacted with full-length MyoD (Fig. 2D). Next, we examined the domains of MyoD bound by Jmjd2C. The basic (amino acids 1-115 of MyoD), HLH (helix-loop-helix, amino acids 116-166), and Cterminal (amino acids 167-318) MyoD mutants bound to Jmjd2C in the same way as does full-length MyoD (Fig. 2E and F). In addition, we conducted an immunoprecipitation experiment to examine the endogenous interaction between Jmjd2C and MyoD in C2C12 cells. As reported before [7], we were able to detect the interaction of MyoD with G9a in myoblasts (Fig. 2G, second panel). We also

1089

observed a strong interaction of MyoD with Jmjd2C in myotubes, but GSK3β as a negative control didn't associate with MyoD. Collectively, these data indicate that Jmjd2C associates with MyoD in vitro and in vivo. 3.3. Jmjd2C decreases the methylation and ubiquitination levels of MyoD Since Jmjd2C interacts with MyoD, we next decided to conduct an in vitro MyoD demethylation assay to examine whether Jmjd2C demethylates MyoD directly. To this end, GST-MyoD and GSTG9a(SET) were purified from glutathione beads for the in vitro MyoD methylation reaction as described in the Materials and methods section. An in vitro demethylation assay was performed by mixing the purified GST, GST-wt Jmjd2C, and GST-H190G/E192A Jmjd2C, which is a catalytically inactive mutant, with the (afore-described) methylated MyoD. Our results clearly showed that JmjdD2C directly demethylated MyoD in vitro compared to the control and H190G/E192A Jmjd2C (Fig. 3A). Based on the finding above, we hypothesized that Jmjd2C antagonizes G9a activity during skeletal muscle differentiation. To test this idea, we examined whether Jmjd2C can reverse the methylation of MyoD catalyzed by G9a in vitro. Exogenous MyoD was highly methylated (Fig. 3B, second lane in left panel) by endogenous G9a in 293T cells (Supplementary Fig. 3A), however, Jmjd2C expression reversed MyoD methylation in a dose-dependent fashion (Fig. 3B, compare lane 2 with 3 and 4 in left panel). The MyoD demethylation by Jmjd2C was prevented by treatment with 5-carboxy-8HQ, a competitive inhibitor of the Jmjd2 family proteins (Fig. 3B, see lanes 1 to 2). Moreover, H190G/E192A Jmjd2C increased the methylation level of MyoD, whereas UNC0638, a G9 specific inhibitor, produced an opposite effect (Fig. 3C). Change of the methylation status in some proteins may be required for protein degradation [9,11,12]. To test this idea, we conducted an ubiquitination assay to examine whether Jmjd2C regulates MyoD degradation. Jmjd2C expression highly reduced MyoD ubiquitination (Fig. 3D). In line with the findings, knockdown of Jmjd2C expression using shRNA Jmjd2C, pSingle-Jmjd2C, increased MyoD ubiquitination in C2C12 myoblast cells (Fig. 3E). On the contrary, wt G9a increased MyoD ubiquitination compared to H1166K G9a, a catalytically inactive G9a mutant (Fig. 3F), suggesting that MyoD methylation status controlled by Jmjd2C and G9a determines MyoD stability in vivo. Next, we would like to explore the relationship between MyoD methylation and ubiquitination-dependent degradation under various conditions described below. The proteasome inhibitor MG132 highly increased MyoD ubiquitination level, indicating that MyoD is degraded mainly through the ubiquitin-mediated proteasome pathway (Fig. 3G, lane 3). Wt G9a increased MyoD ubiquitination to a level comparable with that induced by MG132 (Fig. 3G, lane 6). However, H1166K G9a suppressed this effect (compare lanes 6 and 7). As anticipated, wt Jmjd2C expression decreased MyoD ubiquitination relative to the control (lanes 2 and 4), whereas H190G/E192A Jmjd2C a little augmented MyoD ubiquitination compared to wt Jmjd2C (lanes 4 and 5). Consistent with the above, a G9a/GLP inhibitor, BIX01249, decreased MyoD ubiquitination (Fig. 3H). Taken together, Jmjd2C demethylates G9amethylated MyoD, thereby antagonizing G9a activity. Thus, MyoD stability is regulated by the activity of Jmjd2C and G9a that determines the net methylation status of MyoD. 3.4. Hypo-methylated MyoD exhibits a longer half-life than methylated MyoD Our data indicate that MyoD methylation is vital for ubiquitinmediated MyoD degradation. Therefore, hyper-methylated MyoD may be more degradable than hypo-methylated MyoD. To test this possibility, we measured the half-life of MyoD in the presence of G9a or Jmjd2C. G9a expression shortened the half-life of MyoD (Fig. 4A). In contrast, Jmjd2C delayed MyoD degradation (Fig. 4A). We also used the Jmjd2 inhibitor 5-carboxy-8HQ or the G9a inhibitor BIX01249 to change the

1090

E.-S. Jung et al. / Biochimica et Biophysica Acta 1849 (2015) 1081–1094

A

B

C2C12 D0

D1

D2

PD: Glutathione

150

IP: HA

+

-

wt MyoD-3xHA

+

-

GST-MyoD 5xMyc-Dcaf1

+

+ +

K104R MyoD-3xHA 5xMyc-Dcaf1

+

+ +

100

MW 250

MW 250

150 100

150 100

IB: Dcaf1 37 IB: Gapdh RT-PCR D1

5xMyc-Dcaf1

D2

D3

+

IB: rMyoD Dcaf1

GST-Con

+

-

MyoD

GST-MyoD

-

+

Gapdh

MW 250 150 100 75

IB: GST

WCL

IB: Myc

WCL

wt MyoD-3xHA

+

-

K104R MyoD-3xHA 5xMyc-Dcaf1

+

+ +

GST-MyoD 5xMyc-Dcaf1 G9a-GFP Jmjd2C-GFP

MW 250

+ -

+ + -

150 100

75

250

+ + + Dcaf1

IB: Myc 150 100

37 MyoD

50 25

+ + + -

MW 250

Dcaf1

150 100 MyoD

50

+ + +

75

+

150 100

+ + + -

IB: Myc

50

MW 250

Dcaf1

+ + -

+ -

150 100

IB: Myc

WCL

PD: Glutathione

GST-MyoD 5xMyc-Dcaf1 G9a-GFP Jmjd2C-GFP MW 250

IB: Myc

D0

D

GST-Con

D3

MW 250

C

IB: GFP

GST

IB: Myc / HA

Ponceau S

75

MyoD

37

37

IB: GST IB: GAPDH

IB: GAPDH

37 IB: GAPDH

F

WCL + + -

+ + + +

+ +

+ +

MW 250 150 100 75

pSingle-Luc -

-

+

G9a

50

Dcaf1

+

-

+

-

-

+

-

-

+

-

-

+

-

pSingle-Dcaf1-A

-

pSingle-Dcaf1-B

-

75

25

50

20 50

37

37 IB: rMyoD

IP: rMyoD

+ -

+

Ub-MyoD

MW 250 150 100 75

5xMyc 5xMyc-Dcaf1(f) Dcaf1 5xMyc-Dcaf1401-700 5xMyc-Dcaf11001-1492

+ -

+ -

+ -

IB: Myc 50

37

IB: mMyoD WCL + -

+ -

+ -

MW 250 150 100

IB: MyoD

IB: Ub

IB: Gapdh

IB: Myc

37 IB: MyoD

IB: mMyoD

IB: MyoD 37

Dcaf1401-700

50

37

37

37

Dcaf11001-1492

50

50

50 + Dcaf1(f)

75

37 IB: Gapdh

G9a

IB: GFP

WCL + -

+

+

250 150 100

50

37 IB: mMyoD

37

C2C12 MB

75

50 50

IB: Ub

IB: Gapdh

IP: rMyoD

MW 250 150 100

IB: Ub

50

IB: mMyoD

WCL

+

37

37

Ub-MyoD

+ -

50

I

C2C12 MB

5xMyc 5xMyc-Dcaf1

IB: Dcaf1 50

37

IB: Gapdh

+

75

150 100

IB : MyoD

50

+ -

MW 250 150 100

IB: Ub

IB: Myc

IB: HA / GFP

-

250

150 100

MyoD

IP: rMyoD pEGFP-N1 G9a-GFP

-

+

37

H

C2C12 MB WCL

MW 250

IB: FLAG

G

C2C12 MB IP: rMyoD

Ub-MyoD

+ + +

Ub-MyoD

IP: HA + + -

Ub-MyoD

E MyoD-3xHA pEGFP-N1 G9a-GFP 5xMyc-Dcaf1 3xFLAG-Ub

IB: Gapdh

E.-S. Jung et al. / Biochimica et Biophysica Acta 1849 (2015) 1081–1094

methylation status of MyoD. BIX01249 increased the MyoD half-life compared with the control, whereas 5-carboxy-8HQ decreased it (Fig. 4B), supporting our idea that methylated MyoD is degraded by the UPS pathway. We also evaluated the alteration of half-life of MyoD using wt G9a or a catalytically inactive H1166K G9a. The expression of H1166K G9a increased the half-life of MyoD probably by outcompeting endogenous G9a for MyoD (Fig. 4C). On the contrary, H190G/E192A Jmjd2C deceased the half-life of MyoD compared to wt Jmjd2C (Fig. 4D). This suggests that the catalytically inactive Jmjd2C mutant prevents endogenous Jmjd2C from demethylating MyoD, thereby promoting G9amediated degradation. A previous report demonstrated that the K104 residue of MyoD is mono- and di-methylated by G9a [7]. Thus, the K104R MyoD mutant may be refractory to G9a-mediated degradation, causing this mutant to have a long half-life. To test this idea, we examined the half-life of K104R MyoD in the presence of G9a or Jmjd2C. In contrast to wt MyoD, the half-life of K104R MyoD is prolonged by G9a but is also increased by Jmjd2C (Supplementary Fig. 4), suggesting that K104R MyoD is regulated differently from wt MyoD due to removal of the G9a methylation site. Taken together, MyoD stability is primarily governed by its methylation status. To further confirm our findings above, we utilized G9a−/− mouse embryonic stem cells (mESCs) (Supplementary Fig. 3B) [20]. Jmjd2C expression increased MyoD protein level compared to control in G9a−/− mESCs. However, exogenous G9a expression decreased MyoD protein level (Fig. 4E). G9a/Ehmt2 belongs to the same family as GLP/Ehmt1. GLP shows a similar function with G9a in vivo [31]. To examine the role of GLP on MyoD stability, we added UNC0638, an inhibitor of GLP, to the G9a−/− mESCs. Treatment of UNC0638 increased MyoD protein level (Fig. 4F), suggesting that GLP also regulates MyoD stability through methylation. Collectively, Jmid2C extends the half-life of MyoD by reversing the methylation of MyoD catalyzed by G9a methyltransferase. 3.5. Jmjd2C increases skeletal muscle differentiation Our results demonstrate that MyoD methylation is important for MyoD degradation through methylation-dependent ubiquitination. As shown in Fig. 1, endogenous MyoD and Jmjd2C levels are low in C2C12 myoblasts that highly express G9a relative to myotubes. Therefore we decided to investigate the possibility that G9a-dependent methylation leads to MyoD degradation in myoblasts. When we treated C2C12 myoblasts with UNC0638, a G9a/GLP inhibitor, as anticipated, MyoD protein level increased over two-fold (Fig. 5A). We also examined the effect of UNC0638 and 5-carboxy-8HQ in myotubes. As predicted, UNC0638 increased MyoD protein level, probably because of suppression of G9a and GLP in myotubes, whereas 5-carboxy-8HQ significantly reduced MyoD protein level as compared to the control (Fig. 5B). Since Jmjd2C expression increases concomitant with myoblast differentiation (Fig. 1), we investigated whether Jmjd2C might regulate muscle differentiation. As expected, JMJD2C expression enhanced human skeletal muscle differentiation (Fig. 5C). On the contrary, suppressing JMJD2 activity with 5-carboxy-8HQ inhibited human and mouse muscle differentiation (Fig. 5D and J). The Jmjd2 family, which consists of Jmjd2A, Jmjd2B, Jmjd2C, and Jmjd2D, may activate muscle differentiation utilizing the mechanism

1091

described above. To test this hypothesis, we introduced the plasmids of Jmjd2A, Jmjd2B, or Jmjd2C into C2C12 cells. Actually all three of them increased C2C12 muscle differentiation (Fig. 5E). Consistent with our results, Jmjd2C increased myotube formation based on the examination of myogenic index, whereas G9a suppressed it (Fig. 5F). Jmjd2C is a demethylase which decreases H3K9me3 level [14,15]. Therefore, we examined changes in H3K9me3 level during muscle differentiation. In fact, Jmjd2C reduced H3K9me3 level but increased MyoD level (Fig. 5G). However, G9a led to opposite effects compared with Jmjd2C expression (Fig. 5F and G). Consistent with the above results, UNC0638 increased myogenic index, but 5-carboxy-8HQ reduced it (Fig. 5H). In addition, pSingle-Jmjd2C, doxycycline-inducible shRNA to silence endogenous Jmjd2C, decreased muscle differentiation compared to pSingle-Luc (Fig. 5I and Supplementary Fig. 5). This result is consistent with the observation that knockdown of Jmjd2C decreases myogenin and muscle creatine kinase gene expression in C2C12 cells [29]. Collectively, our results demonstrate that Jmjd2C potentiates muscle differentiation by increasing MyoD stability. 3.6. Methylated MyoD is degraded through the Dcaf1/Ddb1/Cul4 ubiquitination pathway MyoD is known to be degraded by the Cul1/Skp/Fbxo32-mediated ubiquitin pathway [5,6]. However, a recent study showed that Dcaf1/ Vprbp adaptor and DDB1/Cul4 E3 ligase mediate the degradation of methylated proteins [12]. Thus, we wanted to investigate whether methylated MyoD is degraded by Dcaf1/Cul4 E3 ligase-dependent UPS pathway. We first examined the mRNA expression patterns of Dcaf1/ Vprbp, Fbxo32/Atrogin1, MyoD, and Myh6 in C2C12 cells (Supplementary Fig. 6). The mRNA levels of both Decaf1 and Fbxo32 were higher in myoblasts than in myotubes, but MyoD mRNA levels were similar in both myoblasts and myotubes. However, MyoD protein level was lower in myoblasts than in myotubes (Fig. 1A and C). Therefore, it is plausible that Dcaf1 functions as an E3 ligase for MyoD in myoblasts. To test the possibility, we examined the expression levels of Decaf1 protein and mRNA during muscle differentiation. As expected, the expressions of Decaf1 protein and mRNA were decreased with differentiation (Fig. 6A), indicating that Dcaf1 is a myoblast-enriched protein. We hypothesized that G9a-methylated MyoD is actively targeted by Dcaf1 in myoblasts for the UPS-dependent degradation. If Dcaf1 acts as an E3 ligase for MyoD, we may be able to detect a protein-protein interaction between MyoD and Dcaf1. To test this possibility, we conducted a pull-down experiment. As anticipated, GST-MyoD associated with Dcaf1 (Fig. 6B). To examine whether MyoD methylation is necessary to associate with Dcaf1, the association of Dcaf1 with wt MyoD and K104R MyoD was examined. Wt MyoD, but not K104R MyoD, is bound to Dcaf1 (Fig. 6C), indicating that the association between MyoD and Dcaf1 is dependent on MyoD methylation. Moreover, G9a increased the association between MyoD and Dcaf1, but Jmjd2C weakened it (Fig. 6D), indicating that the Dcaf1-mediated MyoD association is dependent on the methylation status of MyoD which is determined by G9a and Jmjd2C. To examine whether Dcaf1-mediated MyoD interaction is required for MyoD degradation, we conducted an ubiquitination assay. Dcaf1 expression enhanced ubiquitination of G9a-methylated MyoD (Fig. 6E), suggesting that Dcaf1 functions as an E3 ligase for methylated MyoD

Fig. 6. Methylated MyoD is degraded by the Cul4/Ddb1/Dcaf1 pathway. (A) The expression levels of Dcaf1/Vprbp protein and mRNA are decreased with muscle differentiation. Differentiation was induced by transferring the cells to DMEM supplemented with 4% horse serum. Gapdh was used as a protein and mRNA control. (B) MyoD associates with Dcaf1. GST-MyoD with or without 5xMyc-Dcaf1 were transfected to 293T cells. A pull-down assay was performed as described in the Methods. (C) The association of Dcaf1 with MyoD is dependent on MyoD methylation. 5xMyc-Dcaf1 with wt MyoD or K104R MyoD were transfected to 293T cells. An immunoprecipitation experiment was performed as described in the Methods. (D) The methylation status of MyoD by G9a and Jmjd2C determines the association of MyoD with Dcaf1. The samples were manipulated as in (B). (E) Dcaf1/Vprbp activates G9a-methylated MyoD ubiquitination in 293T cells. The samples were manipulated as in (C). (F) Knockdown of endogenous Dcaf1 by pSingle-Dcaf1,a doxycycline -inducible and Dcaf1-specific shRNA, decreases MyoD ubiquitination in C2C12 myoblast cells. pSingle-Luc, -Dcaf1-A, or -Dcaf1-B was transfected to C2C12 cells using Lipofectamine 2000. The samples were manipulated as in (C). (G) G9a expression in myoblasts increases MyoD ubiquitination. The samples were manipulated as in (C). (H) Dcaf1 expression increases MyoD ubiquitination in C2C12 myoblasts, which express G9a and methylated MyoD at high levels. (I) Wt Dcaf1 expression increases MyoD ubiquitination, but Dcaf1401-700 and Dcaf11001–1492 decreased MyoD ubiquitination in C2C12 myoblasts. The samples were manipulated as in (C).

1092

E.-S. Jung et al. / Biochimica et Biophysica Acta 1849 (2015) 1081–1094

by G9a. On the contrary, silencing of endogenous Dcaf1 by Dcaf1 specific shRNAs, two pSingle-Dcaf1-A and -B, decreased MyoD ubiquitination in C2C12 myoblast cells that resulted in the increase of MyoD protein level in the input (Fig. 6F). Hence, we were tempted to speculate that high Dcaf1 in myoblasts leads to low MyoD protein level in myoblasts. In fact, high G9a expression increased MyoD ubiquitination in C2C12 myoblasts presumably via Dcaf1-mediated ubiquitination (Fig. 6G). Indeed, Dcaf1 expression significantly increased MyoD ubiquitination in C2C12 myoblasts (Fig. 6H). To confirm this finding, we investigated the effects of Dcaf1 mutants in C2C12 myoblasts. Both the Dcaf1401–700 mutant, which has only a chromo domain to recognize methylated proteins, and the Dcaf11001–1492 mutant, which has a WD40 domain responsible for physical binding of Dcaf proteins to Ddb1 [12], suppressed MyoD ubiquitination at myoblasts when compared to full-length Dcaf1 (Fig. 6I), suggesting that methylated MyoD is degraded by the Cul4/ Ddb1/Dcaf1 ubiquitin pathway in myoblast muscle cells. Taken together, Cul4/Ddb1/Dcaf1 pathway is involved in methylation-dependent MyoD degradation in myoblasts. 3.7. Jmjd2C increases MyoD transcriptional activity G9a and Jmjd2C are members of histone modifiers, which regulate gene expression through changes in histone H3 methylation [17,31]. Change of H3K9me3 levels by KMT and KDM regulates gene expression through modulation of chromatin compaction by recruiting epigenetic readers such as HP1α and KAP-1 [16]. Jmjd2C and G9a might regulate MyoD transcriptional activity through chromatin structure alteration in the promoter of MyoD target genes. We tested the idea using 4RELuc reporter system that contains four repeated MyoD-binding E-box sites [23]. Wt Jmjd2C enhanced MyoD transcriptional activity in 293T and C2C12 cells, whereas wt G9a suppressed the MyoD activity (Fig. 7A and B). The results suggest that Jmjd2C might stimulate MyoD transcriptional activity by increasing MyoD protein level. To further investigate physiological relevance of MyoD regulation by G9a and Jmjd2C during muscle differentiation process, we wanted to test the idea that Jmjd2C and G9a control MyoD-mediated myogenic conversion of embryonic fibroblasts using embryonic STO fibroblast cell line [32]. In line with the above results, wt Jmjd2C increased wt MyoD transcriptional activity in STO cells compared to while wt G9a suppressed it (Fig. 7C). Moreover, wt Jmjd2C also increased K104R MyoD transcriptional activity compared to other transgenes in STO cells (Fig. 7D), suggesting that Jmjd2C enhances K104R MyoD stability as shown in Supplementary Fig. 4. Consistent with the reporter assays, Jmjd2C promoted both wt MyoD- and MyoD-K104R-mediated myogenic conversion (Fig. 7E and F), suggesting that Jmjd2C facilitates the myogenic conversion of fibroblasts by increasing MyoD transcriptional activity and also relaxing the chromatin compaction of MyoD target genes. Jmjd2C is a demethylase which decreases H3K9me3 level, a transcription repressor marker [14,15]. To test the hypothesis that Jmjd2C promotes MyoD-mediated myogenic gene expression via decreasing repressive H3K9me3 level, we performed ChIP assays to examine the levels of H3K9me3 and MyoD in the promoter of myogenic genes. Silencing of Jmjd2C decreased the bound level of MyoD (Fig. 7G), but increased the repressive H3K9me3 (Fig. 7H) in the E-box of muscle specific target genes, suggesting that Jmjd2C enhances transcription of myogenic genes through increasing MyoD level and decreasing H3K9me3 level. Collectively, our data unequivocally demonstrate that Jmjd2C activates MyoD transcriptional activity and muscle differentiation by stabilizing MyoD and lowering H3K9me3 (Fig. 7I). 4. Discussion Elucidating the mechanisms by which muscle stem cells differentiate into muscle fibers during development is essential for understanding the potential of muscle stem cells. Lineage decisions in muscle

stem cells are influenced by genetic networks of regulatory factors and epigenetic modifiers. Epigenetic modulation is an elegant regulatory mechanism that functions by covalently modifying histone proteins and DNA without causing genetic code changes. Epigenetic regulation by histone methylation has gained considerable attention due to its variety of biological implications. However, the epigenetic regulation of non-histone proteins such as MyoD in muscle cells remains poorly understood. In this study, we found that Jmjd2C associates with MyoD in vitro and in vivo to demethylate MyoD that has been methylated by G9a. Hyper-methylated MyoD by G9a is more readily ubiquitinated than hypo-methylated MyoD by Jmjd2C, indicating that the methylation status of MyoD determines whether MyoD is degraded by the UPS or not. Another important function of Jmjd2C is to increase MyoD transcriptional activity. Our observation that Jmjd2C expression is increased during muscle differentiation, along with Ling et al.'s report that G9a is down-regulated with muscle differentiation [7], suggests that the epigenetic modification of histones is important for muscle cell differentiation. The mRNA levels of the Jmjd2 family increase during differentiation. Likewise, Jmjd2C protein levels also increase throughout differentiation. After all, the expression pattern of Jmjd2C is in line with our finding that ectopic introduction of the Jmjd2C into embryonic fibroblasts induces muscle differentiation (Fig. 7). The different localization patterns of Jmjd2C protein during mouse muscle differentiation are unexpected. In contrast with the previous reports that Jmjd2C solely localizes to the nucleus [14,15], mouse skeletal Jmjd2C showed differential distribution patterns during muscle differentiation. This indicates that Jmjd2C shuttles between the cytoplasm and the nucleus in muscle cells (Fig. 1). Our data demonstrate that the low complexity domain of Jmjd2C contains the NLS required for nuclear translocation (Supplementary Fig. 1). Indeed, LMB treatment prevents Jmjd2C from localizing to the cytoplasm (Supplementary Fig. 2). Posttranslational modifications, such as methylation and phosphorylation, may be responsible for translocation of Jmjd2C between the nucleus and the cytoplasm in mouse muscle cells. G9a/Ehmt2 exhibits a specificity to a range of substrates that include histone proteins and non-histone proteins [7,33]. The Jmjd2 family demethylates the methylated lysine residues in WIZ, CDYL1, CSB, and G9a/Ehmt2 proteins [34]. In line with this observation, G9a and Jmjd2 exhibit similar substrate specificities, despite their opposing functions [35]. Therefore, the methylation status of MyoD is controlled by the differential expression of G9a and Jmjd2C (Fig. 3). MyoD protein level is inversely proportional to MyoD methylation level, because G9a activates MyoD ubiquitination while Jmjd2C has the opposing effect (Figs. 3 and 4). Thus, Jmjd2C plays a key role in MyoD stabilization through the reduction of methylation-dependent degradation. Consistent with the above conclusion, muscle differentiation is facilitated by overexpression of Jmjd2C in muscle cells (Fig. 5). Knocking down Jmjd2C in C2C12 cells does not completely suppress muscle differentiation due to the presence of other Jmjd2 family members (Fig. 1D). The increased stability of MyoD by other Jmjd2 family members should also be taken into consideration. Our results demonstrated that Jmjd2C stimulates MyoD-dependent myogenic conversion and muscle differentiation via two distinct and synergistic mechanisms. First, Jmjd2C increases MyoD stability by demethylating the G9a-methylated MyoD. Second, Jmjd2C reduces the repressive H3K9me3 level in the myogenic gene promoters and results in the enhanced MyoD target gene transcription. All of our data consistently support our model summarized in Fig. 7I. Protein modification is intricately coupled to the ubiquitindependent degradation pathway. As seen with the widespread phosphorylation-mediated degradation [36], protein methylation also promotes ubiquitin-mediated degradation through the recruitment of the substrate-specific ubiquitin E3 ligase. The histone methyltransferase SET7 specifically methylates K142 of DNMT1 to activate proteasomemediated degradation [9]. In addition, the histone methyltransferase

E.-S. Jung et al. / Biochimica et Biophysica Acta 1849 (2015) 1081–1094

B Luciferase/β-Galactosidase

15 10 5

0 4RE-Luc MyoD-3xHA wt G9a-GFP H1166K G9a-GFP wt Jmjd2C-3xHA 190/192 Jmjd2C-3xHA

+ -

+ + -

D

+ + + -

+ + + -

+ + + -

+ -

+ + -

+ + + -

+ + + -

+ + + -

25

+ -

+ + -

+ + + -

G MyoD ChIP

+ + + -

+ + + -

30 20 10 + -

+ + -

+ + + -

+ + + -

+ + + -

+ + -

+ + + -

+ + + -

+ + + -

+ + +

STO myogenic coversion

40 30 20 10

0 4RE-Luc K104R MyoD-3xHA wt G9a-GFP H1166K G9a-GFP wt Jmjd2C-3xHA 190/192 Jmjd2C-3xHA

+ + +

+ -

+ + -

+ + + -

+ + + -

+ + + -

+ + +

MT3 IgG control MT3/pSingle-Luc MT3/pSingle-Jmjd2C

H3K9me3 ChIP Relative Bound Level

2 Relative Bound Level

+ -

H

MT3 IgG control MT3/pSingle-Luc MT3/pSingle-Jmjd2C

1

0.5

0

25

50

40

0 4RE-Luc wt MyoD-3xHA wt G9a-GFP H1166K G9a-GFP wt Jmjd2C-3xHA 190/192 Jmjd2C-3xHA

+ + +

50

F

50

50

75

0 4RE-Luc wt MyoD-3xHA wt G9a-GFP H1166K G9a-GFP wt Jmjd2C-3xHA 190/192 Jmjd2C-3xHA

+ + +

STO myogenic coversion MHC positive cell #

Luciferase/β-Galactosidase

5

E

STO

0 4RE-Luc K104R MyoD-3xHA wt G9a-GFP H1166K G9a-GFP wt Jmjd2C-3xHA 190/192 Jmjd2C-3xHA

10

0 4RE-Luc MyoD-3xHA wt G9a-GFP H1166K G9a-GFP wt Jmjd2C-3xHA 190/192 Jmjd2C-3xHA

+ + +

STO

15

MHC positive cell #

Luciferase/β-Galactosidase

20

C

C2C12

293T

Luciferase/β-Galactosidase

A

1093

1.5 1 0.5 0

Myh1

Myh6 Tnnc1 Chrna1/ MyoG AchRα

I

Myh1

Myh6 Tnnc1 Chrna1/ MyoG AchRα

Myoblast stage

MyoD

G9a K9me3

Me

K9me3

K9me3

Svu39h1

K104

K104

Myotube stage

Me

Cul4/ Ddb1/ Dcaf1

K104

MyoD

Jmjd2C K9

K9

K9

E-box

E-box MyoD-mediated E box Transcriptional activation Fig. 7. Jmjd2C activates MyoD transcriptional activity through increasing MyoD protein level (A-B), Wt Jmjd2C increases MyoD-mediated transcriptional activity of 4RE-Luc compared with wt G9a, H1166K G9a, or H190G/E192A Jmjd2C in 293T (A) and C2C12 cells (B). These experiments were repeated three times in duplicate, *P b 0.01. (C) Wt Jmjd2C increases wt MyoD-mediated transcriptional activity of 4RE-Luc compared with wt G9a, H1166K G9a, and H190G/E192A Jmjd2C in STO cells. The DNA constructs were transfected into STO fibroblast cells using Lipofectamine 2000. Forty-eight hours after transfection, the cells were switched to DMEM with 4% horse serum for 72 h, then performed a firefly luciferase assay. The samples were manipulated as in (A). (D) Wt Jmjd2C increases K104R MyoD-mediated transcriptional activity of 4RE-Luc compared with wt G9a, H1166K G9a, and H190G/E192A Jmjd2C in STO cells. The samples were manipulated as in (C). (E) STO myogenic conversion assays show that wt Jmjd2C increases wt MyoD-mediated myogenic conversion. The samples were manipulated as in (C). Forty-eight hours after transfection the cells were switched to DMEM with 4% horse serum for 72 h, then conducted an immunostaining analysis with anti-MF20 antibody. These experiments were repeated three times, *P b 0.01. (F) Wt Jmjd2C also increases K104R MyoD-mediated myogenic conversion. The samples were manipulated as in (E). (G–H) Silencing of Jmjd2C reduces the bound MyoD level (G), but increases the H3K9me3 level (H) in muscle specific genes compared with pSingle-Luc. The relative levels of MyoD and H3K9me3 under Jmjd2C deficiency were normalized with the amount of MyoD and H3K9me3 in pSingle-Luc. A ChIP assay was conducted as described in the Methods. These experiments were repeated at least three times, *P b 0.01. (I) A model depicting Jmjd2Cmediated MyoD activation by preventing methylation-dependent MyoD degradation and by enhancing MyoD transcriptional activity.

1094

E.-S. Jung et al. / Biochimica et Biophysica Acta 1849 (2015) 1081–1094

SET9 decreases FoxO3 protein stability [11]. The above findings are supported by a recent study that methylation-dependent ubiquitination is catalyzed by the DDB1/CUL4 E3 ubiquitin ligase complex and a DCAF1 adaptor [12]. Consistent with this, our study shows that Dcaf1 promotes G9a-methylated MyoD ubiquitination-dependent degradation in myoblasts (Fig. 6). Collectively, we speculate that MyoD is degraded through a novel ubiquitination pathway: methylation-dependent ubiquitination via Cul4/Ddb1/Dcaf1 in myoblasts (Fig. 6). In this regard, we strongly suggest that methylation at K104 acts as a methyl degron for MyoD methylationdependent ubiquitination and subsequent proteosomal degradation. In summary, we have demonstrated that Jmjd2C demethylates MyoD and promotes muscle differentiation by protecting MyoD from G9a-mediated methylation-dependent degradation and activating MyoD transcriptional activity. Our study provides important insight into epigenetic regulations of muscle differentiation. In addition, this work will contribute to a better understanding of muscle stem cell biology and its potential applications. Our future direction is to elucidate how Jmjd2C expression is regulated during muscle development. Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.bbagrm.2015.07.001. Funding This research was supported by the grant from the Korea Health Technology R&D Project, Ministry of Health & Welfare (A120262), Republic of Korea. Conflict of interest The authors have no conflict of interest to declare. Transparency document The Transparency document associated with this article can be found, in the online version. Acknowledgments We are grateful to Ms. Patty Mantel for editing the manuscript. We thank Dr. Lin Mei (Medical College of Georgia, USA) for providing the mouse C2C12 cell line and Dr. Makoto Tachibana (Kyoto University, Japan) for providing mouse G9a−/− and G9a−/− + G9a(L) mESCs. References [1] P. Asp, R. Blum, V. Vethantham, F. Parisi, M. Micsinai, J. Cheng, C. Bowman, Y. Kluger, B.D. Dynlacht, Genome-wide remodeling of the epigenetic landscape during myogenic differentiation, Proc. Natl. Acad. Sci. U. S. A. 108 (2011) E149–E158. [2] M.A. Rudnicki, R. Jaenisch, The MyoD family of transcription factors and skeletal myogenesis, BioEssays 17 (1995) 203–209. [3] S. Yokoyama, Y. Ito, H. Ueno-Kudoh, H. Shimizu, K. Uchibe, S. Albini, K. Mitsuoka, S. Miyaki, M. Kiso, A. Nagai, T. Hikata, T. Osada, N. Fukuda, S. Yamashita, D. Harada, V. Mezzano, M. Kasai, P.L. Puri, Y. Hayashizaki, H. Okado, M. Hashimoto, H. Asahara, A systems approach reveals that the myogenesis genome network is regulated by the transcriptional repressor RP58, Dev. Cell 17 (2009) 836–848. [4] V. Saccone, P.L. Puri, Epigenetic regulation of skeletal myogenesis, Organogenesis 6 (2010) 48–53. [5] L.A. Tintignac, J. Lagirand, S. Batonnet, V. Sirri, M.P. Leibovitch, S.A. Leibovitch, Degradation of MyoD mediated by the SCF (MAFbx) ubiquitin ligase, J. Biol. Chem. 280 (2005) 2847–2856. [6] S. Batonnet, M.P. Leibovitch, L. Tintignac, S.A. Leibovitch, Critical role for lysine 133 in the nuclear ubiquitin-mediated degradation of MyoD, J. Biol. Chem. 279 (2004) 5413–5420. [7] B.M. Ling, N. Bharathy, T.K. Chung, W.K. Kok, S. Li, Y.H. Tan, V.K. Rao, S. Gopinadhan, V. Sartorelli, M.J. Walsh, R. Taneja, Lysine methyltransferase G9a methylates the transcription factor MyoD and regulates skeletal muscle differentiation, Proc. Natl. Acad. Sci. U. S. A. 109 (2012) 841–846. [8] B.M. Ling, S. Gopinadhan, W.K. Kok, S.R. Shankar, P. Gopal, N. Bharathy, Y. Wang, R. Taneja, G9a mediates Sharp-1-dependent inhibition of skeletal muscle differentiation, Mol. Biol. Cell 23 (2012) 4778–4785.

[9] P.O. Esteve, H.G. Chin, J. Benner, G.R. Feehery, M. Samaranayake, G.A. Horwitz, S.E. Jacobsen, S. Pradhan, Regulation of DNMT1 stability through SET7-mediated lysine methylation in mammalian cells, Proc. Natl. Acad. Sci. U. S. A. 106 (2009) 5076–5081. [10] L. Fang, L. Zhang, W. Wei, X. Jin, P. Wang, Y. Tong, J. Li, J.X. Du, J. Wong, A methylation-phosphorylation switch determines Sox2 stability and function in ESC maintenance or differentiation, Mol. Cell 55 (2014) 537–551. [11] D.R. Calnan, A.E. Webb, J.L. White, T.R. Stowe, T. Goswami, X. Shi, A. Espejo, M.T. Bedford, O. Gozani, S.P. Gygi, A. Brunet, Methylation by Set9 modulates FoxO3 stability and transcriptional activity, Aging (Albany NY) 4 (2012) 462–479. [12] J.M. Lee, J.S. Lee, H. Kim, K. Kim, H. Park, J.Y. Kim, S.H. Lee, I.S. Kim, J. Kim, M. Lee, C.H. Chung, S.B. Seo, J.B. Yoon, E. Ko, D.Y. Noh, K.I. Kim, K.K. Kim, S.H. Baek, EZH2 generates a methyl degron that is recognized by the DCAF1/DDB1/CUL4 E3 ubiquitin ligase complex, Mol. Cell 48 (2012) 572–586. [13] C. Van Rechem, J.R. Whetstine, Examining the impact of gene variants on histone lysine methylation, Biochim. Biophys. Acta 1839 (2014) 1463–1476. [14] P.A. Cloos, J. Christensen, K. Agger, A. Maiolica, J. Rappsilber, T. Antal, K.H. Hansen, K. Helin, The putative oncogene GASC1 demethylates tri- and dimethylated lysine 9 on histone H3, Nature 442 (2006) 307–311. [15] R.J. Klose, K. Yamane, Y. Bae, D. Zhang, H. Erdjument-Bromage, P. Tempst, J. Wong, Y. Zhang, The transcriptional repressor JHDM3A demethylates trimethyl histone H3 lysine 9 and lysine 36, Nature 442 (2006) 312–316. [16] M.K. Ayrapetov, O. Gursoy-Yuzugullu, C. Xu, Y. Xu, B.D. Price, DNA double-strand breaks promote methylation of histone H3 on lysine 9 and transient formation of repressive chromatin, Proc Natl Acad Sci U S A, 111 9169-9174. [17] P.P. Das, Z. Shao, S. Beyaz, E. Apostolou, L. Pinello, L. Angeles Ade, K. O'Brien, J.M. Atsma, Y. Fujiwara, M. Nguyen, D. Ljuboja, G. Guo, A. Woo, G.C. Yuan, T. Onder, G. Daley, K. Hochedlinger, J. Kim, S.H. Orkin, Distinct and combinatorial functions of Jmjd2b/Kdm4b and Jmjd2c/Kdm4c in mouse embryonic stem cell identity, Mol. Cell 53 (2014) 32–48. [18] Y.H. Loh, W. Zhang, X. Chen, J. George, H.H. Ng, Jmjd1a and Jmjd2c histone H3 Lys 9 demethylases regulate self-renewal in embryonic stem cells, Genes Dev. 21 (2007) 2545–2557. [19] M.T. Pedersen, K. Agger, A. Laugesen, J.V. Johansen, P.A. Cloos, J. Christensen, K. Helin, The demethylase JMJD2C localizes to H3K4me3-positive transcription start sites and is dispensable for embryonic development, Mol. Cell. Biol. 34 (2014) 1031–1045. [20] M. Tachibana, K. Sugimoto, M. Nozaki, J. Ueda, T. Ohta, M. Ohki, M. Fukuda, N. Takeda, H. Niida, H. Kato, Y. Shinkai, G9a histone methyltransferase plays a dominant role in euchromatic histone H3 lysine 9 methylation and is essential for early embryogenesis, Genes Dev. 16 (2002) 1779–1791. [21] J.W. Jang, W.Y. Lee, J.H. Lee, S.H. Moon, C.H. Kim, H.M. Chung, A novel Fbxo25 acts as an E3 ligase for destructing cardiac specific transcription factors, Biochem. Biophys. Res. Commun. 410 (2011) 183–188. [22] J. Choi, H. Jang, H. Kim, S.T. Kim, E.J. Cho, H.D. Youn, Histone demethylase LSD1 is required to induce skeletal muscle differentiation by regulating myogenic factors, Biochem. Biophys. Res. Commun. 401 (2010) 327–332. [23] C.H. Kim, H. Neiswender, E.J. Baik, W.C. Xiong, L. Mei, Beta-catenin interacts with MyoD and regulates its transcription activity, Mol. Cell. Biol. 28 (2008) 2941–2951. [24] A. Burton, C. Azevedo, C. Andreassi, A. Riccio, A. Saiardi, Inositol pyrophosphates regulate JMJD2C-dependent histone demethylation, Proc Natl Acad Sci U S A, 110 18970-18975. [25] A. Ishimura, M. Terashima, H. Kimura, K. Akagi, Y. Suzuki, S. Sugano, T. Suzuki, Jmjd2c histone demethylase enhances the expression of Mdm2 oncogene, Biochem. Biophys. Res. Commun. 389 (2009) 366–371. [26] X. Li, S. Dong, Histone demethylase JMJD2B and JMJD2C induce fibroblast growth factor 2: mediated tumorigenesis of osteosarcoma, Med Oncol, 32 53. [27] W. Luo, R. Chang, J. Zhong, A. Pandey, G.L. Semenza, Histone demethylase JMJD2C is a coactivator for hypoxia-inducible factor 1 that is required for breast cancer progression, Proc Natl Acad Sci U S A, 109 E3367-3376. [28] L. Rui, N.C. Emre, M.J. Kruhlak, H.J. Chung, C. Steidl, G. Slack, G.W. Wright, G. Lenz, V.N. Ngo, A.L. Shaffer, W. Xu, H. Zhao, Y. Yang, L. Lamy, R.E. Davis, W. Xiao, J. Powell, D. Maloney, C.J. Thomas, P. Moller, A. Rosenwald, G. Ott, H.K. Muller-Hermelink, K. Savage, J.M. Connors, L.M. Rimsza, E. Campo, E.S. Jaffe, J. Delabie, E.B. Smeland, D.D. Weisenburger, W.C. Chan, R.D. Gascoyne, D. Levens, L.M. Staudt, Cooperative epigenetic modulation by cancer amplicon genes, Cancer Cell, 18 590-605. [29] L. Verrier, F. Escaffit, C. Chailleux, D. Trouche, M. Vandromme, A new isoform of the histone demethylase JMJD2A/KDM4A is required for skeletal muscle differentiation, PLoS Genet. 7 (2011) e1001390. [30] D. Gorlich, U. Kutay, Transport between the cell nucleus and the cytoplasm, Annu. Rev. Cell Dev. Biol. 15 (1999) 607–660. [31] Y. Shinkai, M. Tachibana, H3K9 methyltransferase G9a and the related molecule GLP, Genes Dev. 25 (2011) 781–788. [32] L. Lattanzi, G. Salvatori, M. Coletta, C. Sonnino, M.G. Cusella De Angelis, L. Gioglio, C.E. Murry, R. Kelly, G. Ferrari, M. Molinaro, M. Crescenzi, F. Mavilio, G. Cossu, High efficiency myogenic conversion of human fibroblasts by adenoviral vector-mediated MyoD gene transfer. An alternative strategy for ex vivo gene therapy of primary myopathies, J. Clin. Invest. 101 (1998) 2119–2128. [33] P. Rathert, A. Dhayalan, M. Murakami, X. Zhang, R. Tamas, R. Jurkowska, Y. Komatsu, Y. Shinkai, X. Cheng, A. Jeltsch, Protein lysine methyltransferase G9a acts on nonhistone targets, Nat. Chem. Biol. 4 (2008) 344–346. [34] V.K. Ponnaluri, D.T. Vavilala, S. Putty, W.G. Gutheil, M. Mukherji, Identification of non-histone substrates for JMJD2A-C histone demethylases, Biochem. Biophys. Res. Commun. 390 (2009) 280–284. [35] V.K. Ponnaluri, D.T. Vavilala, M. Mukherji, Studies on substrate specificity of Jmjd2ac histone demethylases, Biochem. Biophys. Res. Commun. 405 (2011) 588–592. [36] L.K. Nguyen, W. Kolch, B.N. Kholodenko, When ubiquitination meets phosphorylation: a systems biology perspective of EGFR/MAPK signalling, Cell Commun. Signal. 11 (2013) 52.