MOLECULAR AND CELLULAR BIOLOGY, Sept. 2007, p. 6163–6176 Vol. 27, No. 170270-7306/07/$08.00�0 doi:10.1128/MCB.01735-06Copyright © 2007, American Society for Microbiology. All Rights Reserved.

ROCK2 and Its Alternatively Spliced Isoform ROCK2m PositivelyControl the Maturation of the Myogenic Program�

Michele Pelosi,1* Francesco Marampon,2 Bianca M. Zani,1,2 Sabrina Prudente,3 Emerald Perlas,1Viviana Caputo,3 Luciano Cianetti,4 Valeria Berno,1 Shuh Narumiya,5 Shin W. Kang,6

Antonio Musaro,7 and Nadia Rosenthal1*EMBL Mouse Biology Unit, Monterotondo (Roma), Italy1; Department of Experimental Medicine, University of L’Aquila, L’Aquila,

Italy2; CSS-Mendel Institute, Roma, Italy3; Department of Hematology, Oncology and Molecular Medicine, Istituto Superiore diSanita, Roma, Italy4; Department of Pharmacology, Faculty of Medicine, Kyoto University, Sakyo-ku, Kyoto, Japan5;

IRCCS Fondazione Santa Lucia, Roma, Italy6; and Department of Histology and Medical Embryology,University La Sapienza, Roma, Italy7

Received 14 September 2006/Returned for modification 20 October 2006/Accepted 10 June 2007

Signal transduction cascades involving Rho-associated kinases (ROCK), the serine/threonine kinases down-stream effectors of Rho, have been implicated in the regulation of diverse cellular functions including cytoskel-etal organization, cell size control, modulation of gene expression, differentiation, and transformation. Here weshow that ROCK2, the predominant ROCK isoform in skeletal muscle, is progressively up-regulated duringmouse myoblast differentiation and is highly expressed in the dermomyotome and muscle precursor cells ofmouse embryos. We identify a novel and evolutionarily conserved ROCK2 splicing variant, ROCK2m, that ispreferentially expressed in skeletal muscle and strongly up-regulated during in vivo and in vitro differentiationprocesses. The specific knockdown of ROCK2 or ROCK2m expression in C2C12 myogenic cells caused asignificant and selective impairment of the expression of desmin and of the myogenic regulatory factors Mrf4and MyoD. We demonstrate that in myogenic cells, ROCK2 and ROCK2m are positive regulators of the p42and p44 mitogen-activated protein kinase–p90 ribosomal S6 kinase–eucaryotic elongation factor 2 intracellularsignaling pathways and, thereby, positively regulate the hypertrophic effect elicited by insulin-like growthfactor 1 and insulin, linking the multifactorial functions of ROCK to an important control of the myogenicmaturation.

During mammalian embryogenesis, the development ofskeletal muscle is an ordered multistep process, initially requir-ing the myogenic determination of pluripotent mesodermalprecursor cells, followed by myoblast fusion and by the conse-quent progression of a program of muscle-specific gene expres-sion (4, 22, 31, 49). The in vivo orchestration of this processinvolves a complex interplay between intrinsic programs oflineage specification and extrinsic controls such as humoralfactors in the embryonic environment, cell-cell interactions,innervation, and the extracellular milieu (4, 31, 49).

The coordinated action of members of the muscle-specifichelix-loop-helix myogenic regulatory factors (MRFs) and ofthe MADS-box myocyte enhancer factors family (MEFs) reg-ulates the differentiation of pluripotent stem cells into multinu-cleated myotubes and the muscle-specific gene expression (31,44, 49). Although the temporal patterns of MRF and MEFgene expression controlling muscle differentiation have beenquite well characterized, the signaling events that dictate andregulate the progression and the maturation of the myogenicprogram have not been fully described.

Rho GTPases are a family of a molecular switches that

control a variety of cell functions (14). Signaling through theRho GTPases has been implicated in the differentiation ofsmooth, cardiac, and skeletal muscle cells (6, 18, 38, 40, 47, 48).It has been shown that Rho activation is required to promotemyoblast differentiation (6, 12, 39, 40, 48). However, somereports showed RhoA to be a negative regulator of myogenesis(7, 8, 29). Thus, the accumulated evidence describes contrast-ing roles for RhoA in skeletal myogenesis.

Rho-dependent kinases (ROCK) have been identified amongthe major downstream effectors of RhoA (19, 21, 32, 38). Themammalian ROCK family is composed of ROCK1 and ROCK2isoforms, encoded by two different genes (28). ROCK1 andROCK2 are serine/threonine kinases of about 160 kDa that con-tain a highly conserved amino-terminal kinase domain, with 92%amino acid identity between the two isoforms, and carboxyl-ter-minal domains which are substantially different between the twoisoforms (28). ROCK has been suggested to play a role in thecontrol of differentiation processes of some cell types includingskeletal and cardiac myocytes and keratinocytes (7, 12, 24, 29, 39,47). However, the specific functions of each ROCK isoform, invitro and in vivo, remain largely unknown and require furtherinvestigation. Loss of ROCK1 through gene targeting in miceresults in open eyes at birth and an omphalocele phenotype (37,42), while the loss of ROCK2 results in placental dysfunctions,intrauterine growth retardation, and fetal death (41, 42). How-ever, a small number of ROCK2 knockout mice are born aliveand apparently grow normally (41). Overall, the observation thatgene disruption of ROCK1 and ROCK2 displays obviously dif-

* Corresponding author. Mailing address: EMBL Mouse BiologyUnit, Campus Buzzati-Traverso, via Ramarini 32, 00016 Montero-tondo-Scalo, Roma, Italy. Phone: 39-06-90091241. Fax: 39-06-90091272. E-mail for Michele Pelosi: mpelosi@embl-monterotondo.it.E-mail for Nadia Rosenthal: rosenthal@embl-monterotondo.it.

� Published ahead of print on 2 July 2007.

6163

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/m

cb o

n 06

Feb

ruar

y 20

22 b

y 45

.5.1

31.1

72.

ferent phenotypes suggests that the two genes play specific indi-vidual roles and cannot fully compensate for each other’s loss.

The regulation of myogenic growth and differentiation isdependent on the activation of signal transduction cascadeswith the complex involvement of several kinases, including thep42 and p44 mitogen-activated protein kinases (p42/p44MAPK) (2, 15, 20, 35, 50). Transient p42/p44 MAPK activationin the presence of growth factors is required to sustain theproliferative state of the myoblast that is not permissive for theexpression of muscle-specific genes. However, during the onsetof terminal differentiation, a potent and progressive p42/p44MAPK activation is observed, which is independent of mito-gens and essential for regulating myogenic differentiation (2,15, 35).

It has been demonstrated that during terminal differentiation,insulin-like growth factor 1 (IGF-1) and insulin can promote thefull maturation of the myogenic program, contributing to theactivation of the components of the kinase cascade includingp42/p44 MAPK and p90 ribosomal S6 kinase (p90RSK) and in-ducing myotube hypertrophy (9, 26, 27, 46).

In this study, we addressed the isoform-specific role ofROCK2 during the in vivo and in vitro process of myogenicdifferentiation, using loss-of-function models of a ROCK2knockout mouse line and specific RNA interference experi-ments. We demonstrate that ROCK2, but not ROCK1, isstrongly up-regulated during myogenic differentiation andduring embryonic development. In addition, we identifiedROCK2m, a novel and evolutionarily conserved ROCK2 splic-ing variant, expressed predominantly in skeletal muscles. Se-lective knockdown of ROCK2 or ROCK2m in C2C12 myo-genic cells in vitro and analysis of ROCK2 knockout mice invivo revealed a muscle-specific signaling mechanism that linksROCK2/ROCK2m with the selective regulation of the myo-genic maturation. These findings reveal a new axis of regula-tion in the maturation of the myogenic program and suggestimportant implications for the coordination of intracellularsignaling events during myogenic progression.

MATERIALS AND METHODS

Cell culture and RNA interference. C2C12 mouse myoblasts were maintained inDulbecco’s modified Eagle’s medium (DMEM) with 20% fetal bovine serum(growth medium [GM]). To induce differentiation, cells were shifted to DMEM with2% horse serum (differentiation medium [DM]). All media, sera, and complementsfor cell cultures were from Gibco-Invitrogen. Small interfering RNA (siRNA) du-plex target sequences were designed by QIAGEN Gene Silencing. The siRNA targetsequences were 5�-CAGAAGCGTTGTCTTATGCAA-3� (exon 20) or 5�-TTGGATAAACATGGACATCTA-3� (exon 5) for mouse ROCK2 and 5�-CTCTCTCAGTATTGCCACTAA-3� (exon 27�) for mouse ROCK2m. These siRNAs againstROCK2 or ROCK2m produced consistent biological effects, thereby minimizing thepossibility of off-target effects (see Results).

Undifferentiated C2C12 cells in GM at 70% of confluence plated in six-welldishes were transfected with 200 pmol/well of ROCK2 siRNA, ROCK2m siRNA,or nonsilencing (nonspecific) control siRNA (QIAGEN), using 20 �l/well ofRNAi-Fect transfection reagent (QIAGEN). Eighteen hours after transfection,cells were shifted in DM or in DM containing 50 nM IGF-1 (R&D Systems) or130 nM insulin (Sigma). Cells were harvested at 18, 42, 66, or 90 h aftertransfection, and total proteins were extracted for Western blotting analysis, ortotal RNA was extracted for quantitative PCR analysis.

Protein extraction and immunoblots. C2C12 cells were washed with phos-phate-buffered saline and lysed with a buffer containing 150 mM NaCl, 20mM Tris-HCl (pH 7.4), 1% wt/vol sodium dodecyl sulfate (SDS), 50 mM NaF, 1mM 2-glycerophosphate, 1 mM sodium orthovanadate, 10 mM Na4P2O7, 2 mMEGTA, and a protease inhibitor cocktail (Complete Mini; Roche). Cells weresonicated and cleared by centrifugation (14,000 � g for 10 min at 4°C), and the

whole-cell lysate supernatant was collected. Proteins were resolved on SDS-polyacrylamide gel electrophoresis (PAGE), transferred onto Protran nitrocel-lulose membrane (Schleicher & Schuell), and immunoblotted under standardconditions with primary antibodies, as follows: ROCK2 and ROCK1 (BD Bio-sciences); desmin, cyclin D1, cyclin D3, and Cdkn1a (Santa Cruz); MyoD1 clone5.8A (DakoCytomation); myogenin (a kind gift from Giulio Cossu); �-tubulin(Sigma); phospho-p42/p44 MAPK, total p42/p44 MAPK, phospho-p90RSK Thr-359/Ser-363, phospho-p90RSK Ser-380, total p90RSK, and eEF2 Thr56 (CellSignaling). Anti-mouse or anti-rabbit immunoglobulin G horseradish peroxidaselinked F(ab�)2 secondary antibodies and an enhanced chemiluminescence detec-tion kit were from Amersham.

RNA preparation from mouse tissues and Northern blotting analysis. TotalRNA was extracted from tissues of 15-week-old wild-type C57BL/6J mice and fromwhole embryos (at embryonic day 11.5 [E11.5], E15.5, and E17.5) by RNA-TRIzolextraction (Invitrogen). Total RNA (15 �g) was size fractionated by electrophoresison 1% agarose-formaldehyde gels, subjected to Northern blotting, and hybridized aspreviously described (26). cDNA probes for Northern blotting analysis were pre-pared by reverse transcription-PCR (Access RT-PCR system; Promega) usingmouse liver total RNA as the template for ROCK1 and ROCK2 cDNA. Theprimers used in the RT-PCR for ROCK1 and for ROCK2 Northern blotting probeswere the same as those described by Nakagawa et al. (28). The PCR fragments weresubcloned into a pGEM-T Easy Vector (Promega), and the inserted cDNA wasconfirmed by sequencing. The excised probes (QIAGEN gel extraction kit) werepurified and radioactively labeled by random priming (Amersham).

PCR amplifications and sequencing of ROCK2 isoforms. (i) Mouse isoforms.Total RNA (1 �g) from each mouse tissue or cell line was subjected to reversetranscription-PCR by the Access RT-PCR system (Promega). The primers weredesigned according to a mouse ROCK2 genomic sequence (GenBank referencesequence NM_009072) as follows: “a” (sense), 5�-AGCTGGAGATCAAAGAGATGATGG-3� (in the exon 23 of ROCK2); “b” (antisense), 5�-AGTGTTGTTGCGCACAGGCAAT-3� (in the exon 28 of ROCK2); “c” (sense), 5�-CCATATCACTCAGTCACACAC-3� (in the intron 27 of ROCK2); and “d” (antisense),5�-CCCACTGGTTCCACTGGAAAC-3� (in the exon 31 of ROCK2).

(ii) Human and rat ROCK2m isoforms. A panel of first-strand cDNA fromdifferent human (Caucasian) or rat (Sprague-Dawley) tissues (BD Biosciences)was used as the template for PCR amplification. The primers were designedaccording to the human ROCK2 genomic sequence (GenBank reference se-quence NM_004850), as follows: “c” (sense), as described above; and “e” (anti-sense), 5�-CCAACTGGCTCCACTGGAAAT-3� (in exon 31 of the ROCK2human sequence).

(iii) PCR amplification of rat ROCK2m isoforms. The primers were designedaccording to a rat ROCK2 genomic sequence (reference sequence NM_013022),as follows: “c” (sense), as described above; and “f” (antisense), 5�-AGTGTTGTTCCGCACAGGCAAC-3� (in the exon 28 of the ROCK2 rat sequence). PCRfragments were sequenced in forward and reverse directions using BigDye Ter-minator chemistry and an ABI 3100 DNA sequencer (Applied Biosystems).

Real-time quantitative PCR. Total RNA was extracted from C2C12 cells usingRNeasy mini columns (QIAGEN) and treated with RNase-free DNase. First-strand cDNA was generated from 1 �g of total RNA by using a T-primedfirst-strand kit Ready-To-Go (Amersham). Gene expression by real-time quan-titative PCR was performed on an SDS-ABI Prism 7700 (Applied Biosystems)using premade 6-carboxyfluorescein (FAM)-labeled TaqMan assays for �-tubu-lin, ROCK1, ROCK2, MRF4 (gene alias, Myf6), Mef2c, myogenin, MyoD1,Cdkn1a, cyclin D1, cyclin D3, and desmin (Applied Biosystems). For the house-keeping gene, ubiquitin-B, and for the ROCK2m isoform, we designed twospecific TaqMan assays (Applied Biosystems) consisting of sense primer 5�-CTCTGAAGAAGATTCCTTGCCTTACT-3�, antisense primer 5�-ACAGGCAATGACAACCATCCTT-3�, and FAM-labeled probe 5�-CCTATTAGTACAGAATCAAGATTAG-3� for ROCK2m; and sense primer 5�-ACCTGGTCCTCCGTCTGA-3�, antisense primer 5�-TGGCTAGAGTGCAGAGTAATGC-3�,and FAM-labeled probe 5�-CCAGTGGGCAGTGATG-3� for ubiquitin-B.Changes in gene expression (n-fold) were calculated by the 2���CT method,using the lower expressor as the calibrator (expression � 1).

Bioinformatics analysis. Exon-intron prediction analysis was performed byGenscan software (http://genes.mit.edu/GENSCAN.html) (5). Conservationscores analysis for ROCK2 gene exonic and intronic structures from 17 differentvertebrate species was performed by Vertebrate Multiz Alignment/Conservationat UCSC (http://genome.ucsc.edu). Prediction of potential phosphorylation siteswas performed by NetPhos server (http://www.cbs.dtu.dk/services/NetPhos/) soft-ware (3).

Generation of the ROCK2m-expressing vector. The expression plasmid pCAG-Myc-ROCK2 (generated by S. Narumiya, Kyoto University [28]) was cut withEcoRV restriction enzyme, and the resulting 2,800-bp fragment, containing the

6164 PELOSI ET AL. MOL. CELL. BIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/m

cb o

n 06

Feb

ruar

y 20

22 b

y 45

.5.1

31.1

72.

ROCK2 cDNA sequence from nucleotide 1665 to the end of the 3� untranslatedregion of the insert, was ligated into the EcoRV site of vector pGEM-5Zf(�)(Promega) to generate the plasmid pGEM5-EcoRV-ROCK2. Moreover, a cDNAlibrary from skeletal muscle was subjected to PCR amplification using the primersforward-25 (5�-GCTGTGAATAAGTTGGCGGAGATC-3�), located in the exon25, and reverse-32 (5�-CCGACTAACCCATTTCTGCTG-3�), located in the exon32 of ROCK2, and high-fidelity DNA polymerase AccuPrime-Pfx (Invitrogen). TwoPCR products were obtained, corresponding to the cDNA fragments from exon 25to exon 32 of ROCK2 alternative spliced forms, a 930-bp fragment from ROCK2and an 1,100-bp fragment from ROCK2m cDNA. The blunt-ended 1,100-bp PCRfragment was purified, subjected to 3�-dATP tailing with Taq polymerase (Promega),and cloned into pGEM-Teasy vector (Promega). The resulting plasmid pGEM-T-(25-32)ROCK2m was cut with AflII and XbaI restriction enzymes, and an 826-bpAflII-XbaI fragment was purified and ligated in the plasmid pGEM5-EcoRV-ROCK2, previously cut with AflII and XbaI, to generate a pGEM5-EcoRV-ROCK2m construct. The plasmid pGEM5-EcoRV-ROCK2m was cut with ScaI tofragment the scaffold plasmid and with EcoRV to generate a 3,000-bp fragment thatwas purified and ligated into the final destination vector pCAG-Myc-ROCK2, pre-viously cut with EcoRV, to finally obtain the pCAG-Myc-ROCK2m expressionplasmid. All the constructs and PCR fragments were verified by restriction analysisand direct sequencing.

Transfections, immunoprecipitation, and kinase assay. C2C12 and HEK293Tcells were transiently transfected with pCAG-Myc-ROCK2 and pCAG-Myc-ROCK2m or with the empty vector by Lipofectamine 2000 (Invitrogen). Thirty-six hours after transfection, cells were lysed in 1% Nonidet P-40-containingbuffer and immunoprecipitated by protein G-Sepharose with a monoclonal anti-Myc tag antibody, 9E10 (BD Pharmingen). ROCK2 and ROCK2m catalyticactivity levels were assessed in vitro by incubating anti-Myc immunoprecipitatesfor 5 min at 37°C in 25 mM Tris-HCl (pH 7.5), 50 mM NaCl, 10 mM MgCl2, 0.1mM Na3VO4, 500 �M dithiothreitol, 33 ng/ml myelin basic protein (Sigma) as asubstrate, and 5 �M [-P32]ATP (specific activity 10 mCi/�mol). Samples wereresolved in SDS-PAGE and transferred onto nitrocellulose membranes. Radio-labeled bands were visualized by autoradiography and quantitated by usingImageQuant software after scanning in a PhosphorImager (Amersham).

Confocal immunofluorescence. Transfected C2C12 or HEK293T cells werefixed in formaldehyde and permeabilized by 0.5% Triton X-100. Anti-Myc tag(71D10) rabbit monoclonal antibody (MAb; Cell Signaling) and RhoA (26C4)mouse MAb (Santa Cruz) were incubated overnight at 4°C. Alexa-488 or Alexa-633-conjugated secondary antibodies (Molecular Probes) and 4�,6-diamidino-2-phenylindole (DAPI; 1 �g/ml) were used prior to mounting in Vectashield(Vector Laboratories). Omitting any of the primary antibodies resulted in a lossof the immunostaining, demonstrating the lack of the cross-reactivity in thedouble labeling. Cells were imaged using a confocal laser microscope (LeicaTCS-SP5 system) with a Leica �63 HCX plan-apochromat 1.4-numerical-aper-ture objective. Whole-cell volumes were collected at 1-�m Z steps, and imageswere projected.

ROCK2 knockout mouse embryo generation. Previously generated and de-scribed ROCK2�/� heterozygous mice from the C57BL/6J genetic background(41) were intercrossed to produce ROCK2�/� offspring. Genotyping was per-formed by PCR, according to conditions previously described (41), usinggenomic DNA prepared from tails of mice or visceral yolk sacs from embryos.

�-Galactosidase staining and immunohistochemistry. Dissected embryoswere fixed in 2% formaldehyde for 30 min and stained for -galactosidase asdescribed previously (41). After color development, embryos were dehydratedand embedded in paraffin. Transverse sections (8 �m thick) were cut for histo-logical observation. Antibody staining of sections was performed using antigenretrieval with 10 mM Tris buffer (pH 8.0). The antibodies antidesmin (SantaCruz), antilaminin (Sigma), anti-phospho-p42/p44 MAPK (Cell Signaling), andanti-p42/p44 MAPK (Santa Cruz) were used. Anti-ROCK2m antiserum produc-tion was performed by immunization of New Zealand rabbits, with the peptideSHTLLDFDSEEDSLP conjugated to keyhole limpet hemocyanin. Immunohis-tochemical analysis of the myosin heavy chain, as shown in Fig. 8, was performedas previously described (27).

Nucleotide sequence accession number. DNA sequences were deposited in theGenBank database under accession number DQ864977.

RESULTS

ROCK2 is highly expressed in skeletal muscle. Rho hasbeen implicated in the control of skeletal muscle differentia-tion, and ROCK1 and ROCK2 are among the downstreameffectors of Rho. However, in vivo and in vitro, the specific

functions of each ROCK isoform in muscle differentiationhave not been elucidated. Accordingly, we performed a de-tailed analysis of expression and regulation of ROCK isoformsduring muscle differentiation. Quantitative PCR analysis of avariety of adult mouse tissues revealed the highest expressionof ROCK2 in heart, brain, and lung and in most skeletalmuscles (Fig. 1B). ROCK1 and ROCK2 were expressed athigh levels in heart and lung, whereas ROCK2 predominatedin the brain and skeletal muscles (Fig. 1A and B). Northernblotting analysis of ROCK1 and ROCK2 expression was con-sistent with the results of the quantitative PCR (see Fig. 1Cand reference 28). Consistent with the report by Nakagawa etal. (28), Northern blotting experiments showed one transcriptof approximately 6.6 kb for ROCK1, whereas ROCK2 gener-ated three major transcripts of approximately 7.6, 6.6, and 5.6kb. The difference in tissue distribution of ROCK1 andROCK2 raises the possibility that these kinases play distinctfunctions and that ROCK2 is likely to be more functionallyrelevant during muscle differentiation.

ROCK2m is a novel splicing isoform of ROCK2, evolution-arily conserved and predominantly expressed in skeletal mus-cle. We performed a bioinformatic analysis of the mouseROCK2 gene sequence, using a predictive program to detectputative exonic and intronic regions in genomic sequences (5).This analysis predicted the presence of a potentially alterna-tively spliced exon located in an evolutionarily conserved re-gion of intron 27 of the mouse ROCK2 gene (Fig. 2A). Toconfirm this in silico result, we performed a reverse-transcrip-tion PCR analysis using RNA from different mouse tissues andtwo primers located in exon 23 and exon 28 of the mouseROCK2 gene (Fig. 2A and B, left panel). The PCR productsfrom brain, lung, liver, spleen, and intestine showed the am-plification of a single band of the expected size (655 bp),whereas the PCR products from four skeletal muscles (quad-riceps, diaphragm, soleus, and gastrocnemius) showed ampli-fication of the expected band of 655 bp plus an additional bandof about 850 bp (Fig. 2B, left panel). This latter band wassequenced, and it revealed the inclusion of an extra exon be-tween exon 27 and exon 28 (Fig. 2A). This extra exon, whichwas called exon 27�, encodes an open reading frame (ORF) of171 bp (Fig. 2A). This ORF is in frame with the coding se-quence of ROCK2 and generates a novel splicing variant (Fig.2A), which we named ROCK2m (GenBank accession numberDQ864977).

Expression of the ROCK2m isoform was tested using twoprimers located in exon 27� and in exon 30 of the mouseROCK2 gene (Fig. 2B, right panel). ROCK2m was expressedin all the adult skeletal muscles and in both the proliferatingmyoblast and the differentiated myotubes of the mouse myo-genic cell line C2C12 (Fig. 2B, right panel). ROCK2m tran-scripts were also detected in heart and skin (Fig. 2B, rightpanel). However, quantitative analysis revealed that ROCK2mis only minimally expressed in these two latter tissues (Fig. 2C).Notably, ROCK2m was not present in brain and lung (Fig. 2B),although ROCK2 expression was found to be very high in thesetissues (Fig. 1B). Moreover, ROCK2m was also undetectablein tissues rich in smooth muscle (i.e., stomach and intestine,see Fig. 2B) where ROCK plays an important physiologicalrole in the process of Ca2� sensitization (38).

Evaluation of relative ROCK2m expression levels was per-

VOL. 27, 2007 ROCK2 ISOFORMS IN MYOGENIC DIFFERENTIATION 6165

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/m

cb o

n 06

Feb

ruar

y 20

22 b

y 45

.5.1

31.1

72.

formed by quantitative real-time PCR assay, developing a spe-cific TaqMan probe (Fig. 2C and D). Quantitative PCR mea-surements of ROCK2m transcripts showed a high expressionof this isoform in skeletal muscles (with the exception of thelatissimus dorsi), whereas very low levels of ROCK2m weredetected in heart and skin (Fig. 2C). ROCK2m mRNA wasundetectable by quantitative PCR in brain, liver, intestine,stomach, lung, testis, kidney, and spleen (Fig. 2C and data notshown). The ratio between the ROCK2m versus total ROCK2expression level was obtained by the real-time PCR compara-tive CT method and showed that ROCK2m accounts for 20 to50% of the total ROCK2 in all the adult mouse muscles tested,whereas in heart and skin, ROCK2m accounts for only 3% and9% of the total ROCK2, respectively (Fig. 2D).

To evaluate the evolutionary conservation of ROCK2m, weperformed PCR experiments with cDNA panels from differenthuman and rat tissues, using ortholog DNA primers. ROCK2mwas detected in heart and skeletal muscles, whereas it wasabsent from the other tissues tested (Fig. 2F). Thus, the mech-anism and pattern of alternative splicing of mouse ROCK2/ROCK2m are conserved in other mammalian species.

An anti-peptide polyclonal antibody specifically recognizingthe ROCK2m isoform was unable to recognize ROCK2m inWestern blotting experiments but was effective in the immu-nohistochemical detection of ROCK2m in muscle (Fig. 2E).

The amino acidic sequence encoded by the exon 27� ORFrevealed an extremely high content of Ser/Thr (one-third ofthe total amino acid residues), half of which was predicted in

silico as potential phosphorylation sites (3). The amino acidicsequence of the exon 27� is located in the carboxyl-terminalregion of ROCK2, flanking the pleckstrin homology domain(Fig. 2A). The carboxyl-terminal region of ROCK has beendemonstrated to function as an autoregulatory inhibitor do-main of the amino-terminal kinase region (1). Therefore, wecan speculate that the insertion of an extra domain in suchregulatory position could have potential effects on the confor-mation and the activity of the kinase.

In order to address the structure/function of ROCK2m, wecloned the full-length cDNA of ROCK2m (see Materials andMethods), and we evaluated ROCK2m kinase activity, cellularlocalization, and interaction with RhoA. HEK293T and C2C12cells were transfected with plasmids expressing N-terminallyMyc-tagged ROCK2 or ROCK2m. Immunofluorescence mi-croscopy revealed that about 80% of the HEK293T cells andabout 10% of the C2C12 cells were efficiently transfected bythe ROCK2- or the ROCK2m-expressing plasmid (data notshown).

As shown in Fig. 2G, when anti-Myc immunoprecipitatesfrom transiently transfected HEK293T cells were subjected toan in vitro kinase assay using myelin basic protein (MBP) asthe exogenous substrate, ROCK2m displayed a kinase activityon MBP not significantly different than that of ROCK2 (Fig.2G). Indeed, after normalization for the amount of immuno-precipitated proteins (Fig. 2G, right panel), the ratio betweenthe catalytic activity of ROCK2m and the catalytic activity ofROCK2 on MBP was found to be 0.84 � 0.31 (mean � stan-

FIG. 1. ROCK2 is the predominant isoform in mouse skeletal muscles. (A and B) Quantitative PCR analysis of ROCK1 and ROCK2expression in adult mouse tissues. First-strand cDNA was generated from 1 �g of total RNA. The expression data were normalized for theexpression of ubiquitin-B. Results are expressed as means � standard deviations. (C) Northern blotting analysis of isoforms ROCK1 and ROCK2in adult mouse tissues. Northern blots of total RNA (15 �g/lane) were hybridized with 32P-labeled cDNA probes specific for ROCK1 or forROCK2; ethidium bromide (EtBr) staining of the 18S and 28S rRNA was used to verify equal amounts of loading and integrity of the RNAsamples.

6166 PELOSI ET AL. MOL. CELL. BIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/m

cb o

n 06

Feb

ruar

y 20

22 b

y 45

.5.1

31.1

72.

FIG. 2. ROCK2m is a novel splicing isoform of ROCK2, expressed predominantly in skeletal muscle. (A, upper panel) Exon-intron predictionanalysis of part of the ROCK2 mouse gene sequence, performed by Genscan software (http://genes.mit.edu/GENSCAN.html). Arrows represent primers,as described in Materials and Methods. (Middle panel) The alternatively spliced exon 27� is located in a highly evolutionarily conserved part of intron27 of the ROCK2 gene. The histogram shows evolutionary conservation scores for ROCK2 genes from 17 different vertebrate species obtained byVertebrate Multiz Alignment/Conservation UCSC (http://genome.ucsc.edu). (Lower panel) Schematic representation of the predicted functionaldomains of ROCK2/ROCK2m proteins (28). Amino acids are numbered. (B) PCR analysis of mouse ROCK2/ROCK2m. (Left panel) RT-PCR analysisof total RNA from mouse tissues, using primers “a” in exon 23 and “b” in exon 28. (Right panel) RT-PCR analysis of ROCK2m expression, using primers“c” in exon 27� and “d” in exon 30 of the ROCK2 gene. (C) Quantitative PCR analysis of ROCK2m in mouse tissues. Results are expressed as means �standard deviations. (D) Relative expression of ROCK2m in mouse tissues. Total ROCK2 and ROCK2m levels of expression were detected byquantitative PCR analysis. Numbers and white bars represent the percentages of ROCK2m expression with respect to those of the total ROCK2.(E) Immunohistochemical detection of ROCK2m in back muscles of an E18.5 mouse embryo. A specific anti-ROCK2m polyclonal antibody was raised,using an antigenic peptide corresponding to part of the mouse amino acid sequence of exon 27�. (F) PCR analysis of ROCK2m expression in human andrat tissues. cDNA panels from human and rat tissues were subjected to PCR amplification using primer “c” located in exon 27� and primer “e” (locatedin the human exon 31) or primer “f” (located in the rat exon 28). (G) ROCK2 and ROCK2m kinase activity. HEK293T cells were transfected with theempty vector pCAG or with the same plasmid encoding Myc-ROCK2 or Myc-ROCK2m. The immunoprecipitated kinases were subjected to an in vitrokinase assay in the presence of [-32P]ATP and in the presence (lanes 1, 2, and 3) or in the absence (lanes 4, 5, and 6) of the exogenous substrate MBPand developed by autoradiography (left panel) and PhosphorImager analysis. The amount of protein kinase in each immunoprecipitated sample wasdetermined by immunoblotting (right panel, lanes 7, 8, and 9) and densitometry. The experiment shown is representative of three independentdeterminations. (H) Cellular localization of ROCK2 and ROCK2m. Immunofluorescent cells were imaged using a confocal laser microscope. Panels Aand B show C2C12 cells transfected with ROCK2 or ROCK2m and stained with an anti-Myc antibody (red); DAPI was used to stain the DNA (blue).Panels C and D show HEK293T cells transfected with ROCK2 or ROCK2m and double stained with an anti-Myc antibody (red) and with an anti-RhoAantibody (green); DAPI was used to stain the DNA (blue). White bar � 25 �m in panels A and B and 10 �m in panels C and D.

6167

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/m

cb o

n 06

Feb

ruar

y 20

22 b

y 45

.5.1

31.1

72.

dard deviation [SD], n � 3). In the presence of MBP in thekinase assay, both of the immunoprecipitated ROCK2 iso-forms underwent autophosphorylation, whereas in the absenceof the exogenous substrate, we could not observe any signifi-cant autophosphorylation (Fig. 2G).

Notably, ROCK2m displayed an autophosphorylation levelsignificantly lower than that of ROCK2 (Fig. 2G). After nor-malization for the amount of immunoprecipitated proteins, theratio between the autophosphorylation level of ROCK2m andthe autophosphorylation level of ROCK2 was found to be0.42 � 0.20 (n � 3). Under the conditions of immunoprecipi-tation employed, we did not see any differential phosphory-lated band coprecipitating with ROCK2 or with ROCK2m(Fig. 2G).

Cellular localization of ROCK2 isoforms was analyzed withC2C12 myocytes and HEK293T cells. ROCK2 and ROCK2mshowed cellular distributions that were similar: both isoformswere mainly diffused, with some localization at the cell margins(Fig. 2H). This cell distribution was similar to that previouslyobserved for ROCK2 (19, 21).

It has been reported that ROCK2 could be partly recruitedto the plasma membrane by RhoA (19, 21). We addressed thefunctional interaction between RhoA and ROCK2m by a co-localization study. We were able to efficiently stain endogenousRhoA in HEK293T cells, but the same antibodies were notefficient in C2C12 cells (data not shown). Using confocal im-munofluorescence microscopy, we observed that ROCK2 andthe endogenous RhoA partly colocalize in HEK293T cells(Fig. 2H, panel C), consistent with previous observations (19).

However, in HEK293T cells, we could not observe the samecolocalization between the ROCK2m splicing isoform and theendogenous RhoA (Fig. 2H, panel D). These data suggest thatROCK2m may constitute a specialized pool of the cellularROCK2 not directly recruited at the plasma membrane byRhoA. Further biochemical studies are needed to uncover thedistinct properties of ROCK2m and to explore whether theinsertion of the extra domain can modulate its binding capacityto RhoA and its autophosphorylation properties.

ROCK2 and ROCK2m are progressively expressed duringmuscle differentiation. To study the dynamics of ROCK1,ROCK2, and ROCK2m expression during muscle differentia-tion, we utilized C2C12 cells as an in vitro myogenic differen-tiation model system which allows a detailed characterizationof molecular events, such as expression of cell cycle regulatorsand myogenic markers (36). Using quantitative real-time PCR,we observed that in C2C12 cells, ROCK1 mRNA was lessabundant compared to that of ROCK2 (Fig. 3A). Consistently,ROCK1 expression remained relatively constant during thecourse of the differentiation, whereas ROCK2 expression wasup-regulated more than threefold during the process of myo-tube formation and maturation (Fig. 3A). ROCK2m ac-counted for about 50% of the total ROCK2 in every stage ofdifferentiation of C2C12 cells, with kinetics of expression sim-ilar to those of ROCK2 (Fig. 3B). The dynamics of ROCKisoform expression levels during C2C12 myogenic differentia-tion were compared to markers of myogenic differentiation(Fig. 3C to F). Expression levels of MyoD, myogenin, MEF-2C, desmin, cyclin D1, cyclin D3, and Cdkn1a were consistent

FIG. 3. ROCK2 and ROCK2m are progressively expressed during myogenic differentiation. (A through F) C2C12 undifferentiated myoblastswere maintained in GM; differentiation was induced by switching subconfluent cells to DM for 1 to 3 days. RNA was extracted, and first-strandcDNA was generated from 1 �g of total RNA. The expression data from quantitative real-time PCR analysis were normalized for the expressionof ubiquitin-B. Results are expressed as means � standard deviations.

6168 PELOSI ET AL. MOL. CELL. BIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/m

cb o

n 06

Feb

ruar

y 20

22 b

y 45

.5.1

31.1

72.

with the progressive transition from the proliferative stage tothe cell cycle withdrawal and myogenic differentiation (36).While MyoD, myogenin, Cdkn1a, and desmin expression in-creased early after cell shifting in DM and decreased at laterstages of differentiation, MEF-2C, cyclin D3, and Mrf4 showeda more sustained expression level, and cyclin D1 was progres-sively and dramatically down-regulated (Fig. 3C to F). Theseanalyses established the relative dynamics of ROCK expres-sion and provided a model for detailed analysis of ROCKisoform functions during muscle differentiation.

ROCK2 expression during mouse embryonic skeletal mus-cle development. Analysis of ROCK isoform expression wasperformed with RNA isolated from total mouse embryos atdifferent developmental stages. As seen in Fig. 4, ROCK1expression remained relatively constant from E11.5 to E17.5,whereas ROCK2 expression was up-regulated approximatelytwofold in the same developmental window (Fig. 4A and B).ROCK1 mRNA was consistently less abundant than ROCK2mRNA (data not shown). ROCK2m expression was progres-sively up-regulated by about 70-fold from E11.5 to E17.5 (Fig.4C), and at E17.5, ROCK2m accounted for about 20% of thetotal ROCK2 expressed by the whole embryo (Fig. 4D).

To monitor the expression pattern of ROCK2 during em-bryonic development, we took advantage of a mouse line pre-viously generated and described by Narumiya and collabora-tors, in which the ROCK2 gene was disrupted by the insertionof the lacZ reporter cassette into exon 3, in frame with the startcodon (41). In this previous analysis, the majority of theROCK2�/� embryos died in utero in the late stage of preg-nancy, and growth retardation was observed in the small num-ber of surviving ROCK2�/� embryos and pups (41). Statisticalanalysis of genotype distribution in utero revealed that untilE15.5, ROCK2�/� embryos were viable (41). Moreover, untilE15.5, no growth retardation and no gross anatomical abnor-

malities were observed (Fig. 4E and reference 41). We there-fore restricted our analysis to stages earlier than E15.5. Sincethe secondary wave of muscle differentiation begins approxi-mately at E15.5 (17), our muscle phenotypic analysis was re-stricted to initial myofiber formation.

LacZ reporter expression in ROCK2�/� heterozygous mice(Fig. 4E to I) recapitulated the spatiotemporal expression ofROCK2 during mouse embryogenesis (41). Migrating muscleprecursor cells delaminate from the lateral dermomyotome, aderivative of the somite, to generate the muscles of limbs (30).At E11.5, LacZ expression was high in the somites (Fig. 4I), inthe dermomyotome and in the limb muscle precursors (Fig. 4Gand H) as well as in multiple neuronal precursor populations(particularly in the anterior horn of the spinal cord, see Fig.4H). Similar patterns of LacZ expression in the dermomyo-tome and in migrating muscle precursors cells were also ob-served in E11.5 ROCK2�/� embryos (not shown), implyingthat the total absence of ROCK2 protein did not impair its owntranscription in these embryonic structures, nor did it impairthe process of delamination/migration of the muscle precursorcells. Analysis of skeletal muscles (distal forelimb and hindlimb muscles, hypaxial musculature at the trunk level, dia-phragm, back muscles, and intrinsic tongue muscles) at E13.5and E15.5 in ROCK2�/� did not show any gross muscle alter-ation (not shown), suggesting that ROCK2 is dispensable forthe proper execution of the primary wave of myogenesis.

ROCK2 and ROCK2m positively regulate myogenic differ-entiation. In previous studies, the synthetic inhibitor Y-27632or the forced expression of the dominant active form ofROCK1 has been used to investigate the possible functions ofROCK isoforms during myogenic differentiation (7, 29). How-ever, Y-27632 cannot distinguish isoform-specific effects andcan inhibit kinases other than ROCK isoforms (32, 41). More-over, the overexpression of a dominant active truncated form

FIG. 4. ROCK2 expression during mouse embryonic development. (A to D) Expression dynamics of ROCK1, ROCK2, and ROCK2m during mouseembryonic development. Total RNA was isolated from total mouse embryos at E11.5, E15.5, and E17.5, and first-strand cDNA was generated. Theexpression data from quantitative PCR analysis were equalized for the expression of ubiquitin-B. Results are expressed as means � standard deviations.(E) Phenotype of the E11.5 offspring obtained by crossing ROCK2�/� with ROCK2�/�; results are shown for E11.5 ROCK2�/� (left bars), E11.5ROCK2�/� (center bars), and E11.5 ROCK2�/� (right bars) littermates. ROCK2 expression was revealed by X-Gal (5-bromo-4-chloro-3-indolyl--D-galactopyranoside) staining of the LacZ reporter gene. Genotyping was performed by PCR with embryonic visceral yolk sacs. (F to I) Transverse sectionsof E11.5 ROCK2�/� embryo at the trunk level. Dashed lines indicated by g, h, and i in panel F refer, respectively, to the sections shown in panels G, H,and I. dm, dermomyotome; s, somite; mp, limb muscle precursors; sc, anterior horn of the spinal cord.

VOL. 27, 2007 ROCK2 ISOFORMS IN MYOGENIC DIFFERENTIATION 6169

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/m

cb o

n 06

Feb

ruar

y 20

22 b

y 45

.5.1

31.1

72.

of ROCK could have generalized effects on actin cytoskeleton,thus complicating the interpretation of muscle-specific effectsof ROCK isoforms (18, 32).

We addressed the isoform-specific role of ROCK2 andROCK2m in muscle differentiation. We selectively knockeddown the endogenous ROCK2 or ROCK2m, using siRNA.

Specific knockdown of ROCK2 in C2C12 cells by siRNAreduced the total endogenous ROCK2 and ROCK2m tran-scripts (Fig. 5A). Transfection of C2C12 cultures with siRNAspecific for the ROCK2m isoform resulted in a dramatic de-crease of ROCK2m transcripts (Fig. 5B). ROCK2m transcriptsaccount for about 50% of the total ROCK2 in C2C12 cells(Fig. 3). Indeed, knockdown of ROCK2m also resulted in asignificant reduction in the levels of total ROCK2 (Fig. 5B).

Microscopy examination of the general morphology of thetransfected C2C12 cells showed that siRNA-mediated knock-down of ROCK2 or of ROCK2m did not cause any obviousimpairment in the fusion process or any significant alterationof myotube formation and maintenance during the culturetime examined (up to 4 days in DM; results not shown).

To test whether the knockdown of ROCK2/ROCK2m couldinduce general changes in C2C12 differentiation, we analyzedthe expression profile of MyoD, myogenin, MEF-2C, Mrf4,

and desmin in cells transfected with siRNA for ROCK2 or forROCK2m (Fig. 5A and B). Inhibition of the expression ofROCK2 or ROCK2m caused a reduction of desmin, Mrf4, andMyoD expression, while no changes were induced in levels ofmyogenin and MEF-2C transcripts (Fig. 5A and B). Notably,the knockdown of ROCK2m resulted in an even greater down-modulation of myogenic markers compared to that of totalROCK2. Considering that ROCK2m comprises up to 50% ofthe total ROCK2 in C2C12 cells, the interference of ROCK2mexpression had a more significant effect in myogenic differen-tiation compared to the interference of total ROCK2.

To test whether effects of the siRNA-mediated knockdownof ROCK2/ROCK2m were also reproducible at the proteinlevel, the total cell lysates from C2C12 cells transfected withsiRNA for ROCK2 or for ROCK2m were analyzed by Westernblotting (Fig. 5C). The knockdown of ROCK2 or ROCK2mcaused a significant decrease in the protein levels of totalROCK2, MyoD, and desmin, whereas myogenin, �-tubulin,and ROCK1 (Fig. 5C and data not shown) were not signifi-cantly affected. By using anti-ROCK2 antibody, we detected aproportional decrease of the protein levels of total ROCK2after ROCK2m knockdown (Fig. 5C).

We investigated whether ROCK2/ROCK2m may regulate

FIG. 5. ROCK2 and ROCK2m positively regulate myogenic differentiation. Undifferentiated C2C12 cells in GM were transfected with ROCK2siRNA, ROCK2m siRNA, or control (nonsilencing) siRNA and after 18 h were switched into DM. (A and B) Quantitative PCR analysis ofmyogenic differentiation factors in siRNA-transfected C2C12 cells. siRNA-transfected cells were harvested after 2 days in DM; total RNA wasextracted, and first-strand cDNA was generated from 1 �g of total RNA. Expression data were normalized for the expression of ubiquitin-B.Results are expressed as means � standard deviations. (C) Western blotting analysis of siRNA-transfected C2C12 cells. siRNA-transfected cellswere harvested in GM or after 1 or 2 days in DM, and total protein was extracted. Total cell lysate (70 �g) was subjected to Western blottinganalysis with the indicated specific antibodies.

6170 PELOSI ET AL. MOL. CELL. BIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/m

cb o

n 06

Feb

ruar

y 20

22 b

y 45

.5.1

31.1

72.

cell cycle-related markers. We monitored the protein levels ofcyclin D1 and cyclin D3, two well-established mitogenic sen-sors that couple cell cycle withdrawal to muscle differentiation(36, 44, 45). Cyclin D1 is down-regulated, whereas expressionof cyclin D3 is up-regulated during the process of normaldifferentiation of C2C12 cells (Fig. 3). Notably, no majorchanges were induced in the protein levels of both cyclins byROCK2 siRNA or by ROCK2m siRNA during myogenic dif-ferentiation (Fig. 5C).

The acquisition of apoptosis resistance by myogenic precur-sor cells is also a critical event during differentiation (44).Cdkn1a, which promotes cell cycle exit, also confers apoptosisresistance to the differentiating myoblasts (44, 45). The knock-down of ROCK2 or ROCK2m did not perturb the expressionof Cdkn1a (Fig. 5C).

ROCK2 and ROCK2m positively control the activation ofp42/p44 MAPK and of p90RSK. p42/p44 MAPK have a keyrole in myogenesis in that sustained p42/p44 MAPK activationis required at the onset of differentiation in myogenic cells (2,15, 35). To determine whether signals arising from ROCK2/ROCK2m can influence p42/p44 MAPK activity, we per-formed Western blotting analysis with differentiating C2C12cells in the presence of ROCK2 or ROCK2m siRNA. Normal(control) C2C12 differentiating myocytes strongly induced

phospho-active p42/p44 MAPK compared to proliferating cellsin growth medium (Fig. 6A and B). In the differentiation win-dow explored, p44 MAPK activation was largely predominantover that of p42 MAPK (Fig. 6A and B). In C2C12 cellstransfected with siRNA targeting ROCK2 or ROCK2m, sig-nificant decreases in the activation of both p42 and p44 MAPKwere observed (Fig. 6A and B). The total amounts of p42 andp44 MAPK proteins during differentiation were not affected byROCK2 or ROCK2m siRNA (Fig. 6A). Analogously, the ac-tivation and the total amount of p38 and AKT, two kinasesimportant for myogenic growth and differentiation (20, 43, 50),were not affected by ROCK2 or by ROCK2m siRNA (notshown).

The p90RSK is activated by p42/p44 MAPK in vivo and invitro via phosphorylation of specific sites (10, 51). Once acti-vated, p90RSK phosphorylates elongation factor 2 kinase(eEF2K), thus inhibiting eEF2K activity (Fig. 8C and reference46). Inhibition of eEF2K activity results in a decrease in theinhibitory phosphorylation at the Thr56 residue of eEF2, thusactivating eEF2 (Fig. 8C and 46). Indeed, only the dephosphor-ylated form of eEF2 can mediate the translocation step ofelongation and promote protein synthesis (46). Using antibod-ies specific for activated forms of p90RSK, we observed thatp90RSK undergoes full activation during the normal progres-

FIG. 6. ROCK2 and ROCK2m positively control the activation of p42/p44 MAPK and of p90RSK during myogenic differentiation. Undiffer-entiated C2C12 cells in GM were transfected with ROCK2 siRNA, ROCK2m siRNA, or control (nonsilencing) siRNA and after 18 h wereswitched into DM. (A) Western blotting analysis of siRNA-transfected C2C12 cells. siRNA-transfected cells were harvested in GM or after 1 or2 days in DM, and total protein was extracted. Total cell lysates (70 �g) were subjected to Western blotting analysis with the indicated specificantibodies. (B) Quantitation by densitometry of the phospho-p42/p44MAPK immunoblot bands shown in panel A.

VOL. 27, 2007 ROCK2 ISOFORMS IN MYOGENIC DIFFERENTIATION 6171

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/m

cb o

n 06

Feb

ruar

y 20

22 b

y 45

.5.1

31.1

72.

sion of C2C12 myogenic differentiation, concomitantly to p42/p44 MAPK activation (Fig. 6A). Specific down-modulation ofROCK2 or ROCK2m caused a dramatic decrease in the levelsof phospho-active p90RSK during C2C12 differentiation, thusincreasing the level of phospho-Thr56 eEF2 (Fig. 6A). Theseresults suggest that in myogenic cells, ROCK2/ROCK2m canpositively regulate differentiation-induced protein synthesis.

To explore the relevance of these findings in vivo, we com-pared equivalent histological cross-sections of E13.5 and E15.5distal forelimbs from ROCK2�/� and ROCK2�/� (control)embryos (Fig. 7). At these stages of limb development, skeletalmuscle cells already express muscle-specific proteins such asdesmin and laminin, which were visualized by immunohisto-chemistry. In the limb cross-sections from ROCK2�/� em-bryos, both extensor and flexor groups of muscles were nor-mally developed, and compared with sections from controlheterozygous embryos, there were no significant reductions inthe size of the muscle fibers and in the number of laminin-positive cells. However, at E13.5 and E15.5, there was a sig-nificant reduction in the number of desmin-positive cells and,notably, in the level of activated phospho-p42/p44 MAPK (Fig.7). These in vivo results mirror the effects of ROCK2 knock-down in C2C12 myogenic cells (Fig. 5 and Fig. 6) and indicatethat although the absence of ROCK2 or ROCK2m protein didnot alter the number of muscle fibers and the gross anatomy oflimb muscles, it reduced the steady-state levels of desmin pro-tein and perturbed the phosphorylation/activation of p42/p44MAPK.

ROCK2 and ROCK2m positively control the hypertrophiceffect elicited by insulin and IGF-1. We demonstrated that inmyogenic cells, ROCK2 and ROCK2m positively control theactivation of p42/p44 MAPK, p90RSK, and eEF2. There isample evidence showing that in muscle, insulin and IGF-1stimulation activate the components of the kinase cascade in-cluding p42/p44 MAPK and p90RSK, leading to eEF2 activa-tion and sustaining the machinery of terminal differentiation(9, 33, 46). Moreover, we and others demonstrated that IGF-1

and insulin promote the maturation of the myogenic programthrough induction of myotube hypertrophy (9, 26, 27). Collec-tively, these data suggest the hypothesis that in muscle,ROCK2/ROCK2m may be involved in the regulation the hy-pertrophic pathways activated by IGF-1 and/or insulin. To testthis hypothesis, we explored the ability of the myotubes tosustain the biological response to IGF-1/insulin during myogenicdifferentiation, under the knockdown conditions of ROCK2/ROCK2m.

C2C12 myoblasts were transfected with siRNA for ROCK2or ROCK2m. Eighteen hours after transfection, the cells wereswitched to DM in order to permanently withdraw them fromtheir cell cycles and to commit them to myogenic differentia-tion (day zero). The day after, the cells were switched to DMcontaining IGF-1 or insulin and allowed to differentiate for 2days. Cells were subjected to morphological and biochemicalanalysis (Fig. 8).

It has been previously reported that during in vitro myogenicdifferentiation, IGF-1-induced hypertrophy is morphologicallycharacterized not only by the increased size of the single myo-tubes but also by the progressive fusion of adjacent myotubes(26, 27). Indeed, as previously reported (26, 27), myogenicdifferentiation in the presence of IGF-1 or insulin inducedsignificant morphological hypertrophy of myotubes in the con-trol (Fig. 8A). On the contrary, microscopy examination of thegeneral morphology of C2C12 myotubes derived from myo-blasts transfected with siRNA against ROCK2 or ROCK2mrevealed a significant reduction in the hypertrophic responseelicited by IGF-1 and insulin (Fig. 8A). It is notable that themyosin expression was not significantly affected (Fig. 8A anddata not shown), indicating that down-regulation of ROCK2/ROCK2m does not dramatically affect the early phase of mus-cle differentiation. It is more probable that ROCK2 andROCK2m have a greater effect on the control of the latterinduction of hypertrophy (Fig. 8A) (26, 27).

Western blotting analysis revealed that differentiation ofC2C12 cells in the presence of IGF-1 or insulin caused a

FIG. 7. ROCK2 affects the levels of desmin and the activation of p42/p44 MAPK in mouse forelimbs. Muscle-specific proteins were examinedby immunohistochemistry in equivalent histological cross-sections at the level of lower forelimbs of E13.5 and E15.5 ROCK2�/� and ROCK2�/�

(control) mouse embryos. Panels a, b, a�, and b� are higher magnifications (�100) of the muscle groups indicated by arrows in the lowermagnification panel (�20). Note the radius (R) and ulna (U). ROCK2�/� or ROCK2�/� E13.5 and E15.5 embryos were obtained by crossingROCK2�/� with ROCK2�/� mice, and genotyping of the offspring was performed by PCR using embryonic visceral yolk sacs.

6172 PELOSI ET AL. MOL. CELL. BIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/m

cb o

n 06

Feb

ruar

y 20

22 b

y 45

.5.1

31.1

72.

significant increase in the levels of MyoD, desmin, phospho-active p42/p44 MAPK, and phospho-active p90RSK and thedephosphorylation of eEF2, activating its role in promotingprotein synthesis (Fig. 8B). On the contrary, myotubes derivedfrom myoblasts transfected with siRNA against ROCK2 orROCK2m revealed a dramatic reduction in the insulin- orIGF-1-induced up-regulation of MyoD, desmin, phospho-ac-tive p42/p44 MAPK, phospho-active p90RSK, and active eEF2(Fig. 8B). These data strongly suggest that in myogenic cells,ROCK2 and ROCK2m are positive regulators of the p42/p44MAPK-p90RSK-eEF2 intracellular signaling pathway andthereby positively regulate the hypertrophic effect elicited byinsulin and IGF-1 (Fig. 8C).

DISCUSSION

ROCK consists of two isoforms, ROCK1 and ROCK2, en-coded by two different genes (28). Recent in vivo work dem-onstrated that gene disruption of ROCK1 or ROCK2 in micedisplays obviously different phenotypes, strongly suggesting

that the two genes play specific individual roles and cannotcompensate for each other’s loss (37, 41, 42). However thespecific functions of each ROCK isoform remain largely un-known.

In this study, we elucidated the isoform-specific regulationand function of ROCK2 during the in vivo and in vitro pro-cesses of myogenic differentiation, using loss-of-function mod-els with a ROCK2 knockout mouse line and specific RNAinterference experiments.

We show that ROCK2 is the highly predominant isoform inskeletal muscle, indicating that ROCK2 protein might have aspecialized role in this tissue. A detailed analysis of ROCKisoform expression during muscle differentiation in C2C12myogenic cells revealed that ROCK2 expression was stronglyup-regulated at the onset of differentiation and progressivelyincreased during the formation and maturation of myotubes.Conversely, ROCK1 mRNA was consistently less abundantcompared to that of ROCK2, and its expression remainedrelatively constant during the course of the differentiation. Thiswas consistent with a previous report which showed that

FIG. 8. ROCK2 and ROCK2m positively control the hypertrophic effect elicited by IGF-1 and insulin. C2C12 cells in GM were transfected withROCK2 siRNA, ROCK2m siRNA, or control (nonsilencing) siRNA and after 18 h were switched to DM in order to withdraw them from the cellcycle and to commit them to myogenic differentiation. After 24 h in DM, the cells were switched to DM containing 50 nM IGF-1 or 130 nM insulin.After 48 h, cells were analyzed by morphologic examination and immunoblotting. (A) Transfected and differentiated cell cultures, after 2 days inDM plus IGF-1 or DM plus insulin, were immunostained with an antibody against the embryonic isoform of myosin heavy chain. Panels 1, 2, 5and 6 show cells transfected with control (nonsilencing) siRNA. Panels 3 and 7 show cells transfected with ROCK2 siRNA. Panels 4 and 8 showcells transfected with ROCK2m siRNA. (B) Transfected and differentiated cell cultures, after 2 days in DM plus IGF-1 or in DM plus insulin, weresubjected to Western blotting analysis. Cells were harvested, and total protein was extracted; total cell lysate (70 �g) was subjected to Westernblotting analysis with the indicated specific antibodies. (C) A schematic model of the signaling connections identified in this study, showing thecontrol of ROCK2/ROCK2m on the IGF-1/insulin–p42/p44MAPK–p90RSK–eEF2 axis.

VOL. 27, 2007 ROCK2 ISOFORMS IN MYOGENIC DIFFERENTIATION 6173

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/m

cb o

n 06

Feb

ruar

y 20

22 b

y 45

.5.1

31.1

72.

ROCK1 activity is progressively and strongly down-regulatedduring C2C12 myogenesis (29). Overall, these data suggestthat ROCK2 rather than ROCK1 is likely to be the function-ally relevant isoform in muscle differentiation.

In this study we describe ROCK2m, a novel splicing isoformof ROCK2, characterized by the inclusion of an extra exon atthe boundary between exon 27 and exon 28. The expressionpattern of ROCK2m further underscores the functional rele-vance of ROCK2 in the skeletal muscle tissues. ROCK2mexpression in human, mouse, and rat was prevalently restrictedto striated muscle tissues, up-regulated during the differentia-tion of C2C12, and accounted for 20 to 50% of the totalROCK2 in all the murine skeletal muscles and myogenic celllines tested.

Low levels of ROCK2m expression were also consistentlydetected in heart and skin, where Rho-associated kinases havebeen implicated as potential regulators of the differentiationprocesses (24, 47). Notably, ROCK2m was undetectable inother tissues characterized by high expression of ROCK2,namely brain and lungs, and in tissues rich in smooth muscle(e.g., stomach and intestine), where ROCK has been impli-cated in the physiological phenomenon of contraction at aconstant intracellular Ca2� concentration (38). The ROCK2misoform, therefore, is most likely to be involved in modulatingspecific functions in skeletal muscle.

We selectively knocked down the endogenous expression of theROCK2 or ROCK2m isoform in the myogenic C2C12 cells, usingRNA interference, to investigate whether these kinases play aspecific and relevant role in the progression of the muscle differ-entiation. siRNA-mediated knockdown of ROCK2 isoforms didnot cause any obvious impairment in the fusion process or anysignificant alteration of the morphology of the myotubes. AftersiRNA-mediated ROCK2 or ROCK2m knockdown, MyoD,Mrf4, and desmin were significantly decreased, whereas cyclinD1, cyclin D3, Cdkn1a, myogenin. and MEF2C were not affected.The similarity in the expression profiles of myogenic-specific fac-tors in the expression of cell cycle regulator proteins and in theactivation of the p42/p44 MAPK-p90RSK-eEF2 pathway aftersiRNA-mediated ROCK2 or ROCK2m knockdown suggests thatin skeletal muscles ROCK2m may constitute a specialized pool ofthe total ROCK2 protein, which controls specific events duringmyogenic differentiation.

The link between the activation of the p42/44 MAPK path-way and the expression of MyoD has been described byGredinger et al. (15). Inhibition of p42/44 MAPK by the in-hibitor PD098059 resulted in inhibition of MyoD expressionand a decrease in MyoD transcriptional activity, while theexpression of myogenin and other muscle-specific markers wasnot affected (15). Because desmin is a transcriptional target ofMyoD, it is expected that the expression levels of desmin RNAare also decreased when p42/44 MAPK is inhibited, as we showin our study.

Our experimental observation demonstrates that ROCK2and ROCK2m are strongly and progressively up-regulated dur-ing the progression of myogenesis and suggests that they arenot implicated in the cell cycle withdrawal, the myogenic fusionprocess, and the myotube formation. ROCK2 and ROCK2mappear to play a selective role downstream of early myogenicevents, in later stages of the myogenic program, where they arenormally expressed (Fig. 3 and Fig. 5). These results are con-

sistent with previous in vivo genetic experiments showing thatmuscles developed relatively normally in MyoD knockout micebut were defective in later mechanisms of differentiation andregeneration (25, 31, 34). Our results are also consistent withprevious observations that the myoblast cell line L6E9 fails toexpress MyoD and desmin but undergoes myogenic fusion andmuscle differentiation (26, 27).

Consistent with these observations, skeletal muscle defectsin ROCK2-deficient embryos were not immediately evidentand were revealed only through a detailed analysis of theROCK2�/� embryonic muscular phenotype. Due to the lethal-ity of ROCK2�/� embryos (41), our phenotype analysis wasrestricted to the first wave of myofiber formation (i.e., up toE15.5). Nevertheless, even at these developmental stages, theabsence of ROCK2 protein caused a reduction in steady-statelevels of active p42/p44 MAPK and desmin, without signifi-cantly altering muscle mass or the number of laminin-positivecells. There may be some functional defects in ROCK2�/�

muscle that cannot be definitively revealed at these develop-mental stages. We speculate, based on our in vivo and in vitrodata, that ROCK2 likely plays a more significant function dur-ing the later stages of myogenesis, such as during the matura-tion of the muscle phenotype.

Our study suggests that ROCK2 isoforms play a positive rolein the later stages of the differentiation process. However, ourdata directly contradict the previous in vitro studies thatshowed the synthetic total ROCK inhibitor Y-27632 can en-hance myoblast fusion and eventually facilitate skeletal muscledifferentiation, suggesting an inhibitory role for ROCKs (7, 8,29). In order to reconcile the apparent discrepancy, we offerthe following explanations. (i) There is a possibility that thedosage of ROCK activity may be important for determiningwhether ROCK plays a positive or negative role in muscledifferentiation. While we have shown that specific gene silenc-ing of ROCK2/ROCK2m has negative effects on muscle dif-ferentiation, we cannot exclude the possibility that when bothROCK1 and ROCK2 are inhibited by Y-27632, the muscledifferentiation may be positively affected. Unfortunately, inthis study we could not address this question using siRNA-mediated knockdown because, under our experimental condi-tions, the combinations of different siRNA for ROCK1 andROCK2 caused a dramatic loss of the silencing effect (resultsnot shown). This phenomenon of loss-of-silencing effects forpools of heterogeneous siRNAs has been described previously(23). (ii) There is a possibility that the inhibitor Y-27632 ismore effective in inhibiting ROCK1 than ROCK2 (16). There-fore, the previous studies using Y-27632 may be indicative ofROCK1 function. However, when we performed a specificknockdown of ROCK1 in C2C12 cells by using two differentspecific siRNA, we observed no significant effect (positive ornegative) on muscle differentiation markers and in the differ-entiation-induced p42/p44 MAPK and p90RSK activation(data not shown). (iii) There may be an off-target effect otherthan ROCKs due to the nonspecific effect of Y-27632. Forexample, it is well established that Y-27632 inhibits kinasesother than ROCKs, such as protein kinase C-related kinases(11) and citron kinase (16), whose respective roles in myogenicdifferentiation are not known (32, 41).

At the onset of terminal differentiation, a potent and pro-longed activation of p42/p44 MAPK is required (2, 15, 35).

6174 PELOSI ET AL. MOL. CELL. BIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/m

cb o

n 06

Feb

ruar

y 20

22 b

y 45

.5.1

31.1

72.

There is ample evidence showing that in skeletal muscle, insu-lin and IGF-1 stimulation can activate the components of thekinase cascade including p42/p44 MAPK and p90RSK, leadingto eEF2 activation and sustaining the machinery of terminaldifferentiation (13, 33, 46). It is also well established that IGF-1or insulin plays a physiological role in promoting the matura-tion of the myogenic program through the induction of myo-tube hypertrophy (9, 26, 27). Moreover, activation of p90RSKrequires phosphorylation by p42/p44 MAPK (10, 46). p90RSKis highly expressed in skeletal muscle and is potently activatedby insulin, IGF-1, and exercise, resulting in increased proteinsynthesis (13). The decrease in IGF-1/insulin-induced p90RSKactivation resulting from the ROCK2/ROCK2m knockdownestablishes a link between p90RSK-dependent control of theprotein synthesis through increased levels of inactive eEF2 thatpresumably down-modulates the peptide-chain elongation pro-cess (Fig. 8C and 46).

Collectively, our findings establish a critical link betweenROCK2 isoforms and the IGF-1/insulin-p42/p44MAPK-p90RSK-eEF2 signaling axis, providing a greater insight intoour understanding of the regulation of hypertrophy and mat-uration of the myogenic program. Further in vivo and in vitrostudies are required to address whether this is a general sig-naling mechanism regulating protein synthesis, hypertrophy,and differentiation processes in other cell types where ROCK2isoforms are present.

ACKNOWLEDGMENTS

We thank T. Nastasi, O. Bereshchenko, G. Dobrowolny, and C.Cupo for scientific discussions, help, and encouragement. We thank C.Serguera, E. Slonimsky, N. Winn, O. Mirabeau, M. P. Santini, B.Kablar, P. Segnalini, M. Hede, L. Tatangelo, P. Kratios, L. Barberi, G.Falcone, and G. Romeo, who contributed to this work.

This work was supported by grants to N.R. from the MuscularDystrophy Association, the Leducq Foundation for Transatlantic Re-search, the MYORES European Muscle Development Network, byEU grant QLG2-CT-2002-00930 (EUMORPHIA), and by grants toA.M. from AIRC and ASI.

REFERENCES

1. Amano, M., K. Chihara, N. Nakamura, T. Kaneko, Y. Matsuura, and K.Kaibuchi. 1999. The COOH terminus of Rho-kinase negatively regulatesrho-kinase activity. J. Biol. Chem. 274:32418–32424.

2. Bennett, A. M., and N. K. Tonks. 1997. Regulation of distinct stages ofskeletal muscle differentiation by mitogen-activated protein kinases. Science278:1288–1291.

3. Blom, N., S. Gammeltoft, and S. Brunak. 1999. Sequence and structure-based prediction of eukaryotic protein phosphorylation sites. J. Mol. Biol.294:1351–1362.

4. Buckingham, M., L. Bajard, T. Chang, P. Daubas, J. Hadchouel, S. Meilhac,D. Montarras, D. Rocancourt, and F. Relaix. 2003. The formation of skeletalmuscle: from somite to limb. J. Anat. 202:59–68.

5. Burge, C., and S. Karlin. 1997. Prediction of complete gene structures inhuman genomic DNA. J. Mol. Biol. 268:78–94.

6. Carnac, G., M. Primig, M. Kitzmann, P. Chafey, D. Tuil, N. Lamb, and A.Fernandez. 1998. RhoA GTPase and serum response factor control selec-tively the expression of MyoD without affecting Myf5 in mouse myoblasts.Mol. Biol. Cell 9:1891–1902.

7. Castellani, L., E. Salvati, S. Alema, and G. Falcone. 2006. Fine regulation ofRhoA and Rock is required for skeletal muscle differentiation. J. Biol.Chem. 281:15249–15257.

8. Charrasse, S., F. Comunale, Y. Grumbach, F. Poulat, A. Blangy, and C.Gauthier-Rouviere. 2006. RhoA GTPase regulates M-cadherin activity andmyoblast fusion. Mol. Biol. Cell 17:749–759.

9. Coolican, S. A., D. S. Samuel, D. Z. Ewton, F. J. McWade, and J. R. Florini.1997. The mitogenic and myogenic actions of insulin-like growth factorsutilize distinct signaling pathways. J. Biol. Chem. 272:6653–6662.

10. Dalby, K. N., N. Morrice, F. B. Caudwell, J. Avruch, and P. Cohen. 1998.Identification of regulatory phosphorylation sites in mitogen-activated pro-

tein kinase (MAPK)-activated protein kinase-1a/p90rsk that are inducible byMAPK. J. Biol. Chem. 273:1496–1505.

11. Davies, S. P., H. Reddy, M. Caivano, and P. Cohen. 2000. Specificity andmechanism of action of some commonly used protein kinase inhibitors.Biochem. J. 351:95–105.

12. Dhawan, J., and D. M. Helfman. 2004. Modulation of acto-myosin contrac-tility in skeletal muscle myoblasts uncouples growth arrest from differenti-ation. J. Cell Sci. 117:3735–3748.

13. Dufresne, S. D., C. Bjorbaek, K. El-Haschimi, Y. Zhao, W. G. Aschenbach,D. E. Moller, and L. J. Goodyear. 2001. Altered extracellular signal-regu-lated kinase signaling and glycogen metabolism in skeletal muscle from p90ribosomal S6 kinase 2 knockout mice. Mol. Cell. Biol. 21:81–87.

14. Etienne-Manneville, S., and A. Hall. 2002. Rho GTPases in cell biology.Nature 420:629–635.

15. Gredinger, E., A. N. Gerber, Y. Tamir, S. J. Tapscott, and E. Bengal. 1998.Mitogen-activated protein kinase pathway is involved in the differentiation ofmuscle cells. J. Biol. Chem. 273:10436–10444.

16. Ishizaki, T., M. Uehata, I. Tamechika, J. Keel, K. Nonomura, M. Maekawa,and S. Narumiya. 2000. Pharmacological properties of Y-27632, a specificinhibitor of rho-associated kinases. Mol. Pharmacol. 57:976–983.

17. Kelly, A. M., and S. I. Zacks. 1968. Neuromuscular development in inter-costal muscle of the fetal and newborn rat. J. Neuropathol. Exp. Neurol.27:109.

18. Kuwahara, K., T. Barrientos, G. C. Pipes, S. Li, and E. N. Olson. 2005.Muscle-specific signaling mechanism that links actin dynamics to serumresponse factor. Mol. Cell. Biol. 25:3173–3181.

19. Leung, T., X.-Q. Chen, E. Manser, and L. Lim. 1996. The p160 RhoA-binding kinase ROK� is a member of a kinase family and is involved in thereorganization of the cytoskeleton. Mol. Cell. Biol. 16:5313–5327.

20. Lluis, F., E. Perdiguero, A. R. Nebreda, and P. Munoz-Canoves. 2006. Reg-ulation of skeletal muscle gene expression by p38 MAP kinases. Trends CellBiol. 16:36–44.

21. Matsui, T., M. Amano, T. Yamamoto, K. Chihara, M. Nakafuku, M. Ito, T.Nakano, K. Okawa, A. Iwamatsu, and K. Kaibuchi. 1996. Rho-associatedkinase, a novel serine/threonine kinase, as a putative target for small GTPbinding protein Rho. EMBO J. 15:2208–2216.

22. McKinsey, T. A., C. L. Zhang, and E. N. Olson. 2002. Signaling chromatin tomake muscle. Curr. Opin. Cell Biol. 14:763–772.

23. McManus, M. T., B. B. Haines, C. P. Dillon, C. E. Whitehurst, L. van Parijs,J. Chen, and P. A. Sharp. 2002. Small interfering RNA-mediated genesilencing in T lymphocytes. J. Immunol. 169:5754–5760.

24. McMullan, R., S. Lax, V. H. Robertson, D. J. Radford, S. Broad, F. M. Watt,A. Rowles, D. R. Croft, M. F. Olson, and N. A. Hotchin. 2003. Keratinocytedifferentiation is regulated by the Rho and ROCK signaling pathway. Curr.Biol. 13:2185–2189.

25. Megeney, L. A., B. Kablar, K. Garrett, J. E. Anderson, and M. A. Rudnicki.1996. MyoD is required for myogenic stem cell function in adult skeletalmuscle. Genes Dev. 10:1173–1183.

26. Musaro, A., K. J. McCullagh, F. J. Naya, E. N. Olson, and N. Rosenthal.1999. IGF-1 induces skeletal myocyte hypertrophy through calcineurin inassociation with GATA-2 and NF-ATc1. Nature 400:581–585.

27. Musaro, A., and N. Rosenthal. 1999. Maturation of the myogenic program isinduced by postmitotic expression of insulin-like growth factor I. Mol. Cell.Biol. 19:3115–3124.

28. Nakagawa, O., K. Fujisawa, T. Ishizaki, Y. Saito, K. Nakao, and S. Narumiya.1996. ROCK-I and ROCK-II, two isoforms of Rho-associated coiled-coil form-ing protein serine/threonine kinase in mice. FEBS Lett. 392:189–193.

29. Nishiyama, T., I. Kii, and A. Kudo. 2004. Inactivation of Rho/ROCK sig-naling is crucial for the nuclear accumulation of FKHR and myoblast fusion.J. Biol. Chem. 279:47311–47319.

30. Ordahl, C. P., and N. M. Le Douarin. 1992. Two myogenic lineages withinthe developing somite. Development 114:339–353.

31. Parker, M. H., P. Seale, and M. A. Rudnicki. 2003. Looking back to theembryo: defining transcriptional networks in adult myogenesis. Nat. Rev.Genet. 4:497–507.

32. Riento, K., and A. J. Ridley. 2003. Rocks: multifunctional kinases in cellbehaviour. Nat. Rev. Mol. Cell. Biol. 4:446–456.

33. Roux, P. P., S. A. Richards, and J. Blenis. 2003. Phosphorylation of p90ribosomal S6 kinase (RSK) regulates extracellular signal-regulated kinasedocking and RSK activity. Mol. Cell. Biol. 23:4796–4804.

34. Rudnicki, M. A., T. Braun, S. Hinuma, and R. Jaenisch. 1992. Inactivationof MyoD in mice leads to up-regulation of the myogenic HLH gene Myf-5and results in apparently normal muscle development. Cell 71:383–390.

35. Sarbassov, D. D., L. G. Jones, and C. A. Peterson. 1997. Extracellularsignal-regulated kinase-1 and -2 respond differently to mitogenic and differ-entiative signaling pathways in myoblasts. Mol. Endocrinol. 11:2038–2047.

36. Shen, X., J. M. Collier, M. Hlaing, L. Zhang, E. H. Delshad, J. Bristow, andH. S. Bernstein. 2003. Genome-wide examination of myoblast cell cyclewithdrawal during differentiation. Dev. Dyn. 226:128–138.

37. Shimizu, Y., D. Thumkeo, J. Keel, T. Ishizaki, H. Oshima, M. Oshima, Y.Noda, F. Matsumura, M. M. Taketo, and S. Narumiya. 2005. ROCK-I

VOL. 27, 2007 ROCK2 ISOFORMS IN MYOGENIC DIFFERENTIATION 6175

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/m

cb o

n 06

Feb

ruar

y 20

22 b

y 45

.5.1

31.1

72.

regulates closure of the eyelids and ventral body wall by inducing assemblyof actomyosin bundles. J. Cell Biol. 168:941–953.

38. Somlyo, A. P., and A. V. Somlyo. 2003. Ca2� sensitivity of smooth muscleand nonmuscle myosin II: modulated by G proteins, kinases, and myosinphosphatase. Physiol. Rev. 83:1325–1358.

39. Sordella, R., W. Jiang, G. C. Chen, M. Curto, and J. Settleman. 2003.Modulation of Rho GTPase signaling regulates a switch between adipogen-esis and myogenesis. Cell 113:147–158.

40. Takano, H., I. Komuro, T. Oka, I. Shiojima, Y. Hiroi, T. Mizuno, and Y.Yazaki. 1998. The Rho family G proteins play a critical role in muscledifferentiation. Mol. Cell. Biol. 18:1580–1589.

41. Thumkeo, D., J. Keel, T. Ishizaki, M. Hirose, K. Nonomura, H. Oshima, M.Oshima, M. M. Taketo, and S. Narumiya. 2003. Targeted disruption of themouse Rho-associated kinase 2 gene results in intrauterine growth retarda-tion and fetal death. Mol. Cell. Biol. 23:5043–5055.

42. Thumkeo, D., Y. Shimizu, S. Sakamoto, S. Yamada, and S. Narumiya. 2005.ROCK-I and ROCK-II cooperatively regulate closure of eyelid and ventralbody wall in mouse embryo. Genes Cells 10:825–834.

43. Tureckova, J., E. M. Wilson, J. L. Cappalonga, and P. Rotwein. 2001. Insu-lin-like growth factor-mediated muscle differentiation: collaboration be-tween phosphatidylinositol 3-kinase-Akt-signaling pathways and myogenin.J. Biol. Chem. 276:39264–39270.

44. Walsh, K., and H. Perlman. 1997. Cell cycle exit upon myogenic differenti-ation. Curr. Opin. Genet. Dev. 7:597–602.

45. Wang, J., and K. Walsh. 1996. Resistance to apoptosis conferred by Cdkinhibitors during myocyte differentiation. Science 273:359–361.

46. Wang, X., W. Li, M. Williams, N. Terada, D. R. Alessi, and C. G. Proud.2001. Regulation of elongation factor 2 kinase by p90(RSK1) and p70 S6kinase. EMBO J. 20:4370–4379.

47. Wei, L., W. Roberts, L. Wang, M. Yamada, S. Zhang, Z. Zhao, S. A. Rivkees,R. J. Schwartz, and K. Imanaka-Yoshida. 2001. Rho kinases play an oblig-atory role in vertebrate embryonic organogenesis. Development 128:2953–2962.

48. Wei, L., W. Zhou, J. D. Croissant, F. E. Johansen, R. Prywes, A. Balasubra-manyam, and R. J. Schwartz. 1998. RhoA signaling via serum responsefactor plays an obligatory role in myogenic differentiation. J. Biol. Chem.273:30287–30294.

49. Weintraub, H. 1993. The MyoD family and myogenesis: redundancy, net-works, and thresholds. Cell 75:1241–1244.

50. Zetser, A., E. Gredinger, and E. Bengal. 1999. p38 mitogen-activated proteinkinase pathway promotes skeletal muscle differentiation. Participation of theMef2c transcription factor. J. Biol. Chem. 274:5193–5200.

51. Zhao, Y., C. Bjorbaek, and D. E. Moller. 1996. Regulation and interaction ofpp90(rsk) isoforms with mitogen-activated protein kinases. J. Biol. Chem.271:29773–29779.

6176 PELOSI ET AL. MOL. CELL. BIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/m

cb o

n 06

Feb

ruar

y 20

22 b

y 45

.5.1

31.1

72.