Mechanisms underlying intranuclear rod formationAna Domazetovska,1,3 Biljana Ilkovski,1 SandraT. Cooper,1,3 Majid Ghoddusi,4 Edna C. Hardeman,4

Laurie S. Minamide,5 Peter W.Gunning,2,3 James R. Bamburg5 and Kathryn N. North1,3

1Institute for Neuromuscular Research, 2Oncology Research Unit, Children’s Hospital at Westmead, 3Discipline of Paediatricsand Child Health, University of Sydney, 4Muscle Development Unit, Children’s Medical Research Institute, Sydney, NSW 2145,Australia and 5Department of Biochemistry and Molecular Biology, Colorado State University, Fort Collins, Colorado 80523-1870, USA

Correspondence to: Prof. Kathryn N. North, Children’s Hospital at Westmead, Locked Bag 4001,Westmead, NSW 2145,AustraliaE-mail: kathryn@chw.edu.au

Specific mutations within the a-skeletal actin gene (ACTA1) result in intranuclear rod myopathy (IRM), charac-terized by rod-like aggregates containing actin and a-actinin-2 inside the nucleus ofmuscle cells.Themechanismleading to formation of intranuclear aggregates containing sarcomeric proteins and their impact on cell functionand contribution to disease pathogenesis is unknown. In this study, we transfected muscle and non-muscle cellswith mutants of a-skeletal actin (Val163Leu,Val163Met) associated with intranuclear rod myopathy. By live-cellimaging we demonstrate that nuclear aggregates of actin form within the nuclear compartment, rather thanentering the nucleus after formation in the cytoplasm, and are highly motile and dynamic structures.Thus, thenuclear environment supports the polymerization of actin and the movement and coalescence of the polymer-ized actin into larger structures.We show that the organization of actin within these aggregates is influenced bythe binding of a-actinin, and that a-actinin is normally present in the nucleus of muscle and non-muscle cells.Furthermore, we demonstrate that, under conditions of cell stress (cytoskeletal disruption and ATP depletion),WTskeletal actin forms aggregates within the nucleus that are similar in morphology to those formed by themutant actin, suggesting a common pathogenic mechanism for aggregate formation. Finally, we show thatthe presence of intranuclear actin aggregates significantly decreases the mitotic index and hence impacts onthe function of the cell. Intranuclear aggregates thus likely contribute to the pathogenesis of muscle weaknessin intranuclear rod myopathy.

Keywords: intranuclear rod myopathy; nuclear aggregates; a-skeletal actin; a-actinin

Abbreviations: ACTA1=(a-skeletal actin gene; CFTD=congenital fibre-type disproportion; IRM= intranuclear rod myopa-thy; LMB= leptomycin B; NES=nuclear export sequences; NLS=nuclear localization sequence.

Received May 24, 2007. Revised August 14, 2007. Accepted September 17, 2007. Advance Access publication October 10, 2007

IntroductionMutations in the gene encoding a-skeletal actin (ACTA1) areresponsible for a number of congenital myopathy subtypesincluding nemaline myopathy (Nowak et al., 1999; Ilkovskiet al., 2001), intranuclear rod myopathy (IRM) (Sparrowet al., 2003; Hutchinson et al., 2006), myopathy associatedwith accumulation of actin (actin myopathy) (Nowak et al.,1999; Sparrow et al., 2003), core myopathy (Kaindl et al.,2004) and congenital fibre-type disproportion (CFTD)(Laing et al., 2004). The variety of changes in muscle histologylikely result from fundamental differences in the way thatACTA1 mutations disrupt muscle function (Sparrow et al.,2003; Ilkovski et al., 2004; Clarke et al., 2007). IRM associatedwith ACTA1 mutations is characterized by congenital onsetmuscle weakness and pathologically by the presence of nuclear

aggregates in skeletal muscle, that are highly enriched forboth filamentous actin and the actin-binding protein at theZ-line, a-actinin-2 (Goebel and Warlo, 1997; Hutchinsonet al., 2006).

Mutations of amino acid Val163 in ACTA1 have beenidentified in a three generation family (Val163Met)(Hutchinson et al., 2006) and in two unrelated patients(Val163Leu) (Goebel et al., 1997a; Nowak et al., 1999) withIRM. In each of the affected individuals, rods occurredalmost exclusively within the nuclei of their muscle cells,with a lack of other myopathic features at the lightmicroscopy level, suggesting that this particular mutationpredisposes to intranuclear rod formation.

Many questions remain to be answered concerning themechanisms underlying the formation of nuclear aggregates

doi:10.1093/brain/awm247 Brain (2007), 130, 3275^3284

� The Author (2007). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org

containing actin and a-actinin, and their contribution tomuscle weakness in IRM. Finding accumulations ofsarcomeric proteins within the nucleus is an unusualobservation, as both, a-skeletal actin and actinin-2 havenot been shown to localize to the nucleus. It is thus notknown whether the nuclear aggregates form within thenucleus or whether they enter the nucleus after formingwithin the cytoplasm. We have utilized IRM mutants ofa-skeletal actin (Val163Leu, Val163Met) in order to providefurther insight into the mechanisms of intranuclear rodformation and their effect on cell function.

Material and MethodsUnless otherwise specified, all cell culture reagents were purchasedfrom Invitrogen and all molecular biology reagents were obtainedfrom Roche.

Cell cultureMouse C2C12 myoblasts were cultured in 40% high glucoseDulbecco’s Modified Eagle Medium (DMEM) and 40% F-12Nutrient Mixture (HAM) with L-glutamine, supplemented withfoetal bovine serum (FBS) and horse serum (HS), each to 10%.Mouse NIH3T3 fibroblasts were cultured in high glucose DMEM,supplemented with FBS to 10%. For transfections and drugtreatments, cells were cultured on collagen/matrigel coatedthermanox coverslips (Nunc) in 24-well plates.

ConstructsGeneration of the WT-actinEGFP and V163L/M-actinEGFP con-structs has been previously described (Ilkovski et al., 2004). Allactin constructs were generated in the pEGFP-N1 backbone(Clontech). To generate the V163L/M-actinuntagged constructs, asegment of the actin cDNA containing the V163M/L mutationswas excised from V163L/M-actinEGFP with PpuMI and XhoI andsubcloned into WT-actinuntagged that had also been digested withthe same enzymes.

TransfectionsC2C12 and NIH3T3 cells were trypsinized and plated onthermanox coverslips (Nunc) 1 day prior to transfection ingrowth medium. The cells were washed twice with phosphate-buffered saline (PBS) and medium without antibiotics was added30min prior to transfection. The cells were transfected at 70–90%confluence using Lipofectamine2000TM, according to the manu-facturer’s instructions. Briefly, per 2 cm2 culture area, 1.2mg ofDNA and 3.6ml lipid were prepared in 0.1ml of OptimemTM andincubated in 0.6ml of growth medium without antibiotics. Cellswere transfected for 6 h, washed twice in PBS and replenished withgrowth medium.

Cell treatmentsLMB (Sigma) was diluted at 2.5, 5 or 10 ng/ml of culture mediumand added to the culture wells overnight. LMB activity was testedusing the Rev-NES-GFP expression system. The pRev(NES)-GFPplasmid has been described previously (Henderson andEleftheriou, 2000). In untreated cells expressing the construct,Rev-NES-GFP localized to the cytoplasm, whereas in cells treated

with LMB for 3 h or overnight, it localized exclusively to thenucleus.Cytochalasin D (final concentration 1 mM; Sigma) and latrun-

culin A (final concentration 5 mM; Sigma) were diluted in culturemedium and added to the cells for 45min–1 h.For ATP depletion, cells were washed with PBS and incubated

in ATP-depletion medium (10mM sodium azide, 6mM2-deoxyglucose in PBS) for 30min at 37�C without CO2 supply.Following cell treatments, the coverslips were briefly washed inPBS and fixed.

Fixation and immunostaining of cultured cellsC2C12 and NIH3T3 cells grown on thermanox coverslips (Nunc)were fixed in PBS containing 3% paraformaldehyde and permea-bilized in 0.1% triton-X 100 for 20min at room temperature (RT).Samples were washed three times in PBS, then incubated inblocking buffer (PBS plus 2% bovine serum albumin) for 10min atRT before immunostaining as described previously (Ilkovski et al.,2004). After immunostaining, samples were washed three times inPBS and mounted on 22� 50mm2 glass coverslips usingFluorsaveTM mounting reagent (Calbiochem).

Antibodies and fluorophoresPrimary antibodies: (mAb) a-actinin (1:300) (Sigma), (mAb)emerin (1:100) (Novacastra Laboratories Ltd), (pAb) a-actinin-2(4B2, 1:4000) and a-actinin-4 (6A2, 1:600) (Dr Alan Beggs,Harvard Medical School), (mAb) actin (C4, 1:200) (BDBiosciences Pharmigen). Secondary antibodies: Cy3-conjugatedgoat anti-mouse or anti-rabbit IgG (1:250), Cy5-conjugateddonkey anti-mouse IgG (1:200) (Jackson ImmunoResearchLaboratories. Inc.), Alexa Fluor 488 goat anti-mouse or anti-rabbit IgG (1:200) (Molecular Probes). For some experiments,TRITC-phalloidin (1:500) (Sigma) and/or ToPro3 iodide (1:200)(Molecular Probes) were added with the secondary antibody.

Extraction of pelleted and solubleprotein poolsTransfected cells were rinsed twice in PBS and then scraped inextraction buffer [50mM MES pH 6.8, 1mM EGTA pH 8.0, 50mMKCl, 1mM MgCl2, 0.5% Triton X-100, protease inhibitor (PI)cocktail from Sigma was added immediately prior to use (1:500)],followed by 1 h ultracentrifugation at 100 000g at 4�C to get poolsenriched for filamentous actin, which likely contains the intra-nuclear aggregates (the pellet, P), and globular actin (the solublefraction, S). The pelleted and soluble fractions were then separatedand 200 ml of the soluble fraction was transferred to a microfuge tubecontaining 50 ml of 5� SDS sample buffer [312.5mM Tris pH 6.8,10% SDS, 50% glycerol, 250mM DTT, PI cocktail (1:500) andbromophenol blue (BPB)]. The pellet was re-suspended in 1� SDSsample buffer [62.5mM Tris pH 6.8, 6% SDS, 10% glycerol, 50mMDTT, PI cocktail (1:500), BPB] and briefly sonicated. All sampleswere heat-inactivated for 4min at 94�C and stored at –20�C.

Western blotSamples were thawed and heated to 94�C for 1min immediatelyprior to loading on 5% stacking, 9% resolving SDS–PAGE gels.Western blot was performed as described previously (Cooperet al., 2003).

3276 Brain (2007), 130, 3275^3284 A. Domazetovska et al.

ImagingConfocal microscopy was performed using a Leica TCS SP2Scanning Confocal Microscope equipped with HCX Plan Apo(PH3) 40�/1.25 and 63�/1.32 oil immersion objective lenses.EGFP or alexa, Cy3 or TRITC and Cy5 or ToPro3 were excited at488, 543 and 633 nm, respectively. For live-cell imaging experi-ments, cells transfected on 42mm collagen/matrigel-coatedcircular glass coverslips were transferred at 6 h after transfectionto microscope incubator and scanning stage equipped withtemperature regulator 37-2 digital and CTI controller 3700 CO2IR sensor. The temperature was maintained at 37�C and CO2 at5%. Images were merged using Leica LCS softwareand figures were assembled using Adobe Photoshop.The mitotic index was determined by counting the number of

transfected cells in mitosis as a proportion of the total number oftransfected cells in randomly chosen fields at 40�magnification.A total of 6919 C2C12 cells were counted from 20 coverslips and2809 NIH3T3 cells from 12 coverslips. Results are expressed asmean� SEM; statistical significance was determined by non-parametric 2-tailed Mann–Whitney U-test.

Electron microscopyCultured cells were fixed in situ with modified Karnovski’s fixative(2.5% glutaraldehyde, 4% paraformaldehyde solution in 0.1Mcacodylate buffer, pH 7.4) for 1 h. Cells were post-fixed with 2%osmium tetroxide, dehydrated through an ascending series ofethanol and embedded in Spurr’s epoxy resin. Sections were cutwith DIATOME diamond knife on Leica ULTRACUT Sultramicrotome at 70 nm thickness, double contrasted withuranyl acetate and lead citrate. The sections were viewed andphotographed with Philips CM120 BioTwin transmission electronmicroscope.

Statistical analysisThe statistical significance was assessed by non-parametrictwo-tailed Mann–Whitney U-test.

ResultsMutant a-skeletal actin forms filamentousaggregates with a-actinin inside the nucleusof different cell typesPatients bearing mutations within ACTA1 at positionVal163 exhibit an IRM, with electron dense aggregatespositive for filamentous actin (phalloidin) and a-actinin-2within the nuclei of their skeletal muscle fibres (Goebeland Warlo, 1997; Hutchinson et al., 2006). Transfectionstudies using a-skeletal actin constructs bearing thesemutations (V163L, V163M) and tagged with a greenfluorescent protein (V163L/M-actinEGFP) or untagged(V163L/M-actinuntagged), revealed striking intranuclearaccumulations of actin, both needle-like and star-like inC2C12 myoblasts (Ilkovski et al., 2004; Fig. 1) and inNIH3T3 fibroblasts (Fig. 1), that closely resembled theintranuclear rod phenotype observed in the patient muscle.WT-actinEGFP, however, localizes to the cytoplasmicmicrofilaments and does not form intranuclear aggregates

(See Supplementary materiall). We performed electronmicroscopy on C2C12 and NIH3T3 cells transientlyexpressing V163L-actinEGFP or V163L-actinuntagged anddemonstrated that the mutant a-skeletal actin aggregatesare surrounded by a double nuclear membrane (Fig. 1A)and are thus located inside, rather than above or below theplane of the nucleus. Higher magnification showed that theaggregates have a filamentous appearance (Fig. 1A), whichin combination with the phalloidin labelling demonstratesthat a-skeletal actin can exist in a filamentous form insidethe nucleus.

Muscle and non muscle-specific isoforms of a-actininlabelled a subset of intranuclear aggregates of actin formedwhen V163L/M-actinuntagged was transfected in muscle(C2C12 myoblasts) and non-muscle (NIH3T3 fibroblasts)cells (Fig. 1B–D). Staining of C2C12 cells expressing V163L/M-actinuntagged with phalloidin and an antibody againsta-actinin-2 and a-actinin-4 showed that both a-actininisoforms accumulate with filamentous actin inside thenucleus (Fig. 1B and data not shown). a-Actinin-4 alsoaccumulated with filamentous actin in the nuclei ofNIH3T3 cells expressing V163L/M-actinuntagged (Fig. 1Cand D). The formation of intranuclear aggregates of mutant(V163L/M) a-skeletal actin and a-actinin inside the nucleusof a variety of cell types, suggests that the intracellularenvironment necessary for intranuclear aggregate formationis not cell-type specific.

The aggregates co-labelled by a-actinin have a distinctmorphology to those not labelled by a-actinin. Thea-actinins were only present within aggregates witha ‘star-like’ shape, localizing to an intensely fluorescentfocus from which the phalloidin-labelled actin filamentsradiate (Fig. 1B–D, insets). These star-like aggregates wereoften accompanied by numerous smaller focal aggregatescontaining both a-actinin and filamentous actin (Fig. 1D,inset 1). The needle-like aggregates did not label withantibodies against the a-actinins (Fig. 1D, inset 2). Thus,the a-actinins likely cross-link the intranuclear aggregatesand may influence their morphology.

Intranuclear mutant actin aggregatesform inside the nucleusTo address the question whether rods are producedwithin the nuclei or enter the nuclei after cytoplasmicformation, we performed live-cell imaging of NIH3T3fibroblasts expressing V163L-actinEGFP and demonstratethat the intranuclear aggregates form inside the nucleus.Diffuse V163L-actinEGFP fluorescence first appeared inthe cytoplasm at �6 h after transfection with the firstdetectable nuclear fluorescence appearing as bright punctaat �15–16 h after transfection, concurrent with a rapidincrease in V163L-actinEGFP expression in the cytoplasm(Fig. 2A; see Video 1 Supplementary materiall). With timethe bright puncta increased in number and size inside thenucleus, they joined upon contact and their morphology

Nuclear actin aggregates in myopathy Brain (2007), 130, 3275^3284 3277

increasingly resembled needle-shaped aggregates (Fig. 2A).The live-cell imaging results were replicated in C2C12myoblasts (data not shown). Analysis of pelletable (P) andsoluble (S) actin pools in C2C12 myoblasts transfected withWT-actinEGFP or V163L-actinEGFP revealed significantly

higher levels of pelletable V163L-actinEGFP (49% totalactin) compared to WT-actinEGFP (26% total actin) at15 h after transfection (Fig. 2B), consistent with thetiming of appearance of first fluorescent puncta inside thenucleus (Fig. 2A).

Fig. 1 Mutant a-skeletal actin forms filamentous aggregates with a-actinin inside the nucleus. (A) NIH3T3 cells transfected withV163L-actinEGFP result in intranuclear aggregates (left panel, arrow) that have a filamentous structure at the electron microscopy level (centre andright panel, arrows). EM performed on cells expressing V163L-actinUnT showed consistent results (data not shown). (B) C2C12 myoblaststransfected withV163L-actinUnT shows that intranuclear aggregates stain with phalloidin (red) and a-actinin-2 (green). Cells are co-stainedwith emerin to demarcate the nuclear envelope (blue). (C) NIH3T3 fibroblasts transfected withV163L-actinUnT shows labelling of intranuc-lear aggregates with phalloidin (red) and a-actinin-4 (green). (D) a-Actinin-4 localizes to focal aggregates but not to needle-like aggregates(green, overlay and insets).

3278 Brain (2007), 130, 3275^3284 A. Domazetovska et al.

Fig. 2 Intranuclear aggregates of mutant actin are dynamic structures that form inside the nucleus. (A) Intranuclear aggregate formationwas assessed using live-cell imaging of an NIH3T3 fibroblast expressing V163L-actinEGFP 15h after transfection. A single confocal sectionis shown for each selected time point (time in hours and minutes; for full video seeVideo 1 Supplementary materiall). The intranuclearaggregates first appear as small puncta inside the nucleus (2:18 inset; arrow). The intranuclear aggregates fuse upon contact (3:47^4:05;arrows). (B) Levels of pelleted (P) and soluble (S) actin separated by ultracentrifugation were analysed by immunoblotting with anti-GFPantibody at 12.5, 15 and 24h after transfection of C2C12 myoblasts withWT-actinEGFP and V163L-actinEGFP.V163L-actinEGFP exhibitssignificantly higher levels of pelleted actin compared to WT-actinEGFP at 15h after transfection. Data are presented as mean� S.E.M.Statistical significance by non-parametric 2-tailed Mann^Whitney U-test is shown (�P=0.021, n=4). (C) Live-cell imaging of an NIH3T3fibroblast expressing V163L-actinEGFP shows that the intranuclear aggregates are dynamic structures. They move and can bend inside thenucleus (11:09; arrows, for full video seeVideo 2 Supplementary materiall).

Nuclear actin aggregates in myopathy Brain (2007), 130, 3275^3284 3279

Intranuclear mutant actin aggregatesare dynamic structuresLive-cell imaging also revealed that the intranuclearaggregates containing V163L-actinEGFP are highly dynamicstructures. Besides being able to fuse, the needle-shapedaggregates can move rapidly and bend inside the nucleus(Fig. 2C; see Video 2 Supplementary material).Interestingly, the bending of aggregates on opposite sidesof the nucleus was synchronous and their movement wasrestricted to one area of the nucleus and appeared to occuraround the point of bending. This suggests that theaggregates may be tethered to a nucleoskeleton.

a-Actinin-2 normally resides withinthe nucleusa-Actinin has not previously been shown to reside withinthe nucleus—yet it is a major constituent of intranuclearrods in patient muscle and in our cell culture model ofintranuclear rod formation. To determine whethera-actinin normally resides within the nucleus, we examinedthe effect of blocking nuclear export on nuclear aggregateformation. We demonstrate that endogenous a-actinin-2can accumulate within the nucleus and its export issensitive to leptomycin B (LMB), which blocks theCRM1/exportin-mediated pathway. In response to LMBtreatment of C2C12 cells, endogenous a-actinin-2 formedpredominantly focal aggregates that did not label withphalloidin in the area of the nucleus of �20% of cells

(Fig. 3A). In a subset of these cells, a-actinin-2 formedneedle-like aggregates, reminiscent of rods seen in associa-tion with the ACTA1 mutants (Fig. 3B).

The actin proteins contain two nuclear export sequences(NES) and b-actin has been shown to accumulate withinthe nucleus and form paracrystalline structures in responseto LMB (Wada et al., 1998) or depletion of the nuclearexporter, exportin 6 (Exp6) (Stuven et al., 2003; Bohnsacket al., 2006). We hypothesized that if a-skeletal actin alsoresides within the nucleus and its export is mediatedthrough CRM1, then LMB treatment would lead to itsaccumulation within the nucleus. However, immunostain-ing of LMB-treated C2C12 cells with phalloidin and the C4antibody, which recognizes all isoforms of actin, failed toshow accumulation of actin within the nucleus (data notshown). In addition, LMB treatment of C2C12 and NIH3T3cells expressing WT-actinEGFP did not result in observableintranuclear aggregates (Fig. 3C). Therefore, if intranuclearaggregates of a-skeletal actin result from blocked nuclearexport, then this is not mediated through CRM1 in the celllines used in this study.

Wild-type a-skeletal actin formsintranuclear aggregates in response to actinmicrofilament disruption and ATP depletionVarious types of cell stress lead to the formation of actin-containing aggregates in the cytoplasm or in the nucleus(Nishida et al., 1987; Minamide et al., 2000). Since previous

Fig. 3 a-Actinin-2 can be found within the nucleus in the absence of mutant actin. (A) C2C12 cells treated with leptomycin B (LMB) formfocal aggregates within the nucleus that label with a-actinin-2 (green) but not with phalloidin (red).ToPro-3 was used to label DNA (blue).(B) Some of the a ^actinin-2 aggregates had needle-like shape (inset). (C) C2C12 cells expressing WT-actinEGFP were treated with LMB.WT-actinEGFP did not form aggregates in the nucleus. Nuclear envelope was demarcated with emerin (blue).

3280 Brain (2007), 130, 3275^3284 A. Domazetovska et al.

studies have focused on non-muscle actins (Nishida et al.,1987; Minamide et al., 2000), we aimed to determinewhether WT a-skeletal actin can form intranuclearaggregates in response to cytoskeletal disruption andischaemia. C2C12 myoblasts were transfected with WT-actinEGFP and treated with 1 mM cytochalasin D (CD) or5mMLatrunculin A (LatA) (model of cytoskeletal disruption)or incubated in ATP-depletion medium for 30min(ischemia). Needle-like aggregates of WT a-skeletal actinformed inside the nucleus in response to all treatments,resembling those formed by expression of mutant (V163L/M)actin (Fig. 4). The experiment was also performedin NIH3T3 fibroblasts and similar results were obtained(data not shown). Thus, actin microfilament disruption,ischaemia and the presence of mutations at V163 lead toa common effect of a-skeletal actin accumulation insidethe nucleus.

Intranuclear aggregates affect mitotic indexAccumulation of actin in the nucleus leads to decreased cellproliferation and may be toxic at the cellular level(Perrimon et al., 1989; Wada et al., 1998; Stuven et al.,2003). Here we demonstrate that the intranuclear aggregatesdue to mutations at position V163 in a-skeletal actin leadto reduced mitotic index. By counting the number of cellsin mitosis for cells transfected with WT-actinEGFP orV163L-actinEGFP at different time-points after transfection,we found that an increase in the number of transfected cellswith intranuclear rods is associated with reduced propor-tion of cells in mitosis (the mitotic index). At 24 h and 48 hafter V163L-actinEGFP transfection, �37% and 81% of thetransfected C2C12 myoblasts respectively (�31% and 74%in NIH3T3 fibroblasts) contained intranuclear rods and themitotic index was reduced by �50% and 97%, respectively(�55% and 85% in NIH3T3 fibroblasts) (Fig. 5 and datanot shown). Replication of the results in NIH3T3fibroblasts demonstrates that the reduced mitotic index is

not a result of myoblast differentiation. In addition, theaggregate-containing cells were not apoptotic, as judged bytheir chromatin arrangement on electron microscopy(Fig. 1A). Furthermore, there was no change in thenumber of nuclei per cell in the aggregate-containingcells. A small proportion of mitotic cells expressing V163L-actinEGFP contained aggregates. These aggregates were smallin size and were pushed to the side of the condensed DNAduring mitosis (Fig. 5C), suggesting that the size of theaggregates could be a factor in determining the extent towhich normal cellular processes are affected. Figure 5Dshows that the large and ‘star-shaped’ intranuclear aggre-gates displace the DNA.

DiscussionIn one of the first descriptions of IRM, Goebel et al.(1997b) commented that

‘finding rods within the muscle fibre nuclei is a surprising andspectacular observation. Normal muscle fibre contents ofa defined structure are not an intranuclear feature. Thus,encountering rods within nuclei arouses considerablespeculation . . .Were the rods produced within the nuclei ordid they enter the nuclei from the sarcoplasm after cytoplasmicformation?’

In this study, we demonstrate that intranuclear aggregatesof mutant a-skeletal actin form within the nuclearcompartment. On live-cell imaging, the mutant actin firstaccumulated into small aggregates inside the nucleus thatgradually increased in number and size and coalesced intolarger aggregates that appear filamentous at the EM leveland label with phalloidin (Figs 1 and 2). Thus, the nuclearenvironment supports the polymerization of actin and themovement and coalescence of the polymerized actin intolarger structures. Interestingly, the intranuclear aggregatesof mutant actin can form in both muscle and non-musclecells (Fig. 1). This suggests that the intranuclear aggregates

Fig. 4 WT a-skeletal actin forms intranuclear aggregates following disruption of the actin microfilament system and ATP-depletion.C2C12myoblasts expressing WT-actinEGFP were treated with (A) cytochalasin D (CD) (B) latrunculin A (LatA) and (C) ATP-depletion medium for30min.WT-actinEGFP formed needle-like intranuclear aggregates in response to all three conditions. DNAwas stained withToPro-3 (blue).

Nuclear actin aggregates in myopathy Brain (2007), 130, 3275^3284 3281

form through a process that does not require musclelineage- specific proteins and that the ability to formintranuclear aggregates is intrinsic to the mutant protein. Inaddition, our data demonstrate that intranuclear aggregatesmay form independently from cytoplasmic aggregates, as nosuch aggregates formed within the cytoplasm of myoblasts.

The a-actinins are a family of actin-binding proteinsinvolved in establishing and maintaining cell structure(Lazarides, 1976; Rajfur et al., 2002; Nakagawa et al.,2004). The actins and a-actinins form a variety of dynamicand highly ordered networks in the cytoplasm involvedin processes such as cell motility, cell adhesion and

Fig. 5 Intranuclear aggregates disrupt mitotic index. (A) Representative images of C2C12 myoblasts expressing WT-actinEGFP and V163L-actinEGFP at 24h post-transfection. DNA is labelled withToPro3 (blue; left panel). Mitotic nuclei appear intense blue in the pseudo-colouredimage (right panel, arrow).Nuclei containing rods are notmitotic (right panel, arrowhead). (B) Mitotic index of C2C12 myoblasts expressingWT-actinEGFP (black bars) or V163L-actinEGFP (grey bars) at 24h and 48h after transfection. There is a significant reduction in the mitoticindex inV163L-actinEGFP transfected cells compared to those transfected withWT-actinEGFP, which is enhanced after 48h of transfection.Data are expressed as the mean� S.E.M. Statistical significance by non-parametric 2-tailed Mann^Whitney U-test is shown. (C) Small rodswere observed occasionally <1% of mitotic cells. The inset shows a rod-containing cell in metaphase. (D) Large intranuclear rods displacethe DNA (arrows). Cells expressing V163L-actinUnTwere stained withToPro-3 to visualize DNA (blue) and phalloidin (red) to visualizefilamentous actin and intranuclear rods.

3282 Brain (2007), 130, 3275^3284 A. Domazetovska et al.

cell division. Our results suggest that in the presenceof mutations at Val163 in ACTA1, a-skeletal actin is ableto enter the nucleus and accumulate into aggregates witha-actinin (Fig. 1). Unexpectedly, we also demonstrate thata-actinin-2 can normally reside within the nucleus(Fig. 3). The a-actinins belong to the spectrin proteinsuperfamily; in recent years, a number of other membersof this protein superfamily have been shown to localizewithin the nucleus and are suggested to play a structuralrole in anchoring other proteins to the nuclear membrane,or perform functions in nuclear processes such as DNArepair (Young and Kothary, 2005). Our finding suggeststhat, similar to the formation of cytoplasmic rods innemaline myopathy, a-actinin-2 cross-links bundles ofmutant actin filaments inside the nucleus into largeraggregates and is thus likely involved in the organizationof mutant actin aggregates into a more complex three-dimensional structure, although additional factors may berequired in vivo for the formation of nemaline bodies inmature skeletal muscle.The process that drives the accumulation of mutant actin

inside the nucleus remains to be determined. In contrast toa-actinin-2, a-skeletal actin did not accumulate inside thenucleus in response to inhibition of CRM1-dependentnuclear export. It is still possible, however, that intranuclearaggregates of mutant actin result from blocked exportthrough another exporter, such as exportin-6, which isknown to export b-actin from the nucleus (Stuven et al.,2003; Bohnsack et al., 2006).Alternately the Val163 mutation may result in increased

trafficking of actin into the nucleus. Amino acid 163 lies inthe hinge region of the actin monomer and, based onmolecular modelling, is likely to affect nucleotide exchangeand the folding of the actin monomer (Sparrow et al., 2003;Costa et al., 2004). The mutant actins preferentially localizeto the nucleus and their incorporation into cytoplasmicmicrofilaments is reduced (Fig. 1). Thus the predispositiontoward intranuclear rod formation secondary to thismutation may be due to the increased availability of actinmonomers in the cytoplasm that cannot properly incorpo-rate into thin filaments, triggering increased trafficking ofactin into the nucleus. This possibility is supported by theobservation on live cell imaging that the appearance of thefirst intranuclear aggregates is concurrent with a rapidincrease in V163L-actinEGFP expression in the cytoplasm(Fig. 2A; Video 1 Supplementary materiall).Actin is not known to contain a nuclear localization

sequence (NLS) and the mechanism leading to its entryinto the nucleus is controversial. Whereas monomeric actinis small enough to enter the nucleus by diffusion (Wadaet al., 1998; Bohnsack et al., 2006), cells have very little freeactin monomer that is not associated with monomersequestering proteins; thus, others propose that actin istranslocated into the nucleus by a transporter moleculesuch as cofilin (Nishida et al., 1987; Matsuzaki et al., 1988;Gonsior et al., 1999; Pendleton et al., 2003; Chhabra and

dos Remedios, 2005). Previous studies in non-muscle cellssuggest that treatment of cells with latrunculin or ATPdepletion, leads to cofilin-mediated translocation of b-actininto the nucleus (Pendleton et al., 2003). In the currentstudy, treatment of both myoblast and fibroblast lines withcytochalasin D, latrunculin A and ATP depletion inducedthe formation of aggregates inside the nucleus composed ofWT a-skeletal actin that closely resembled the aggregatesformed by the V163L/M actin. This raises the possibilitythat both the expression of mutant actin in IRM and cellstress (such as disruption of the actin microfilament systemand ATP depletion), result in intranuclear rod formationdue to a common pathogenic mechanism, such as anincrease in ADP-actin monomers in the cytoplasm. Sincecofilin has increased binding affinity for ADP-actin (Carlieret al., 1997), this would likely result in increased import ofmonomers of a-skeletal actin into the nucleus, where theysubsequently polymerise to form aggregates, that are thencross-linked into more complex structures by a-actinin.Further studies are necessary to dissect the molecular basisof accumulation of mutant actin inside the nucleus, inparticular to determine whether the aggregates are due todecreased export or increased import of the mutant actinand, in the case of increased import, whether this is due topassive diffusion or active transport by a transportermolecule such as cofilin.

It is difficult to imagine that large aggregates inside thenucleus would not affect nuclear function and structure.One possible explanation for the synchronic bending andmovement of the intranuclear aggregates (Fig. 2C) is thatthey are tethered to a nucleoskeleton. Figure 5 shows thatthe intranuclear aggregates displace the DNA and theirappearance correlates with a reduced mitotic index. Lenartet al. (2005) showed that nuclear actin networks areinvolved in the congression of chromosomes in oocytes andthat this process can fail when the actin cytoskeleton isdisrupted, however, the existence of nuclear actin networksremains to be demonstrated in other cell types.Alternatively, the effect of intranuclear aggregates onmitotic index could be the result of a general affect oncell metabolism, disruption of transcription or a toxic effectdue to the sequestration of normal sarcomeric proteinssuch as a-actinin-2, so that they are no longer available toperform their normal functions. It is likely that thesealterations in muscle cell function (rather than structure)contribute to muscle hypotrophy and weakness in IRM,since patients with pure IRM have relatively normalsarcomeric structure and demonstrate very little of themyofibrillar disorganization usually attributed as the causeof weakness in patients with nemaline myopathy andcytoplasmic rods (Hutchinson et al., 2006).

Finally, why do cells accumulate actin inside the nucleus?High levels of nuclear actin are known to interfere withdevelopmental programs and are toxic at the cellular level(Perrimon et al., 1989; Stuven et al., 2003). The sequestra-tion of actin into intranuclear aggregates may be part of

Nuclear actin aggregates in myopathy Brain (2007), 130, 3275^3284 3283

a protective response by the cell which minimizes theimpact of unincorporated mutant actin in the cytoplasm.Thus, packing of the mutant actin into intranuclearaggregates and their cross-linking by the a-actinins maybe an effective way for the cell to minimize the impact ofthe mutation on sarcomeric assembly. However, onceformed, the presence of large aggregates within the nucleuscan disrupt nuclear morphology and impair normal cellfunction.A limitation of our tissue culture model is that our

observations are in mononuclear myoblasts whereas maturemuscle fibres are multinucleate. This limits extrapolation ofour results to determine the effect on whole skeletal musclein vivo. Nevertheless, our study provides valuable informa-tion concerning the time-course and mechanisms under-lying intranuclear rod formation and provides indirectevidence that the presence of these structures disrupts thefunction of the muscle cell, which is likely to impact onmuscle bulk and muscle strength in humans with IRM.

AcknowledgementsWe thank Dr Beric Henderson (University of Sydney,Sydney, Australia) for his gift of the pRev(NES)-GFPconstruct. This work was funded by the National Healthand Medical Research Council (NHMRC #139039 and#403941) Australia. A.D. is supported by AustralianPostgraduate Award (APA) scholarship.

ReferencesBohnsack MT, Stuven T, Kuhn C, Cordes VC, Gorlich D. A selective block

of nuclear actin export stabilizes the giant nuclei of Xenopus oocytes.

Nat Cell Biol 2006; 8: 257–63.

Carlier MF, Laurent V, Santolini J, Melki R, Didry D, Xia GX, et al. Actin

depolymerizing factor (ADF/cofilin) enhances the rate of filament

turnover: implication in actin-based motility. J Cell Biol 1997; 136:

1307–22.

Chhabra D, dos Remedios CG. Cofilin, actin and their complex observed

in vivo using fluorescence resonance energy transfer. Biophys J 2005; 89:

1902–8.

Clarke NF, Ilkovski B, Cooper S, Valova VA, Robinson PJ, Nonaka I, et al.

The pathogenesis of ACTA1-related congenital fiber type disproportion.

Ann Neurology 2007; 61: 552–61.

Cooper ST, Lo HP, North KN. Single section Western blot: improving the

molecular diagnosis of the muscular dystrophies. Neurology 2003; 61:

93–7.

Costa CF, Rommelaere H, Waterschoot D, Sethi KK, Nowak KJ,

Laing NG, et al. Myopathy mutations in alpha-skeletal-muscle

actin cause a range of molecular defects. J Cell Sci 2004; 117: 3367–77.

Goebel HH, Anderson JR, Hubner C, Oexle K, Warlo I. Congenital

myopathy with excess of thin myofilaments. Neuromuscul Disord 1997a;

7: 160–8.

Goebel HH, Piirsoo A, Warlo I, Schofer O, Kehr S, Gaude M. Infantile

intranuclear rod myopathy. J Child Neurol 1997b; 12: 22–30.

Goebel HH, Warlo I. Nemaline myopathy with intranuclear rods–

intranuclear rod myopathy. Neuromuscul Disord 1997; 7: 13–9.

Gonsior SM, Platz S, Buchmeier S, Scheer U, Jockusch BM, Hinssen H.

Conformational difference between nuclear and cytoplasmic actin as

detected by a monoclonal antibody. J Cell Sci 1999; 112: 797–809.

Henderson BR, Eleftheriou A. A comparison of the activity, sequence

specificity, and CRM1-dependence of different nuclear export signals.

Exp Cell Res 2000; 256: 213–24.

Hutchinson DO, Charlton A, Laing NG, Ilkovski B, North KN.

Autosomal dominant nemaline myopathy with intranuclear rods due

to mutation of the skeletal muscle ACTA1 gene: clinical and

pathological variability within a kindred. Neuromuscul Disord 2006;

16: 113–21.

Ilkovski B, Cooper ST, Nowak K, Ryan MM, Yang N, Schnell C, et al.

Nemaline myopathy caused by mutations in the muscle alpha-skeletal-

actin gene. Am J Hum Genet 2001; 68: 1333–43.

Ilkovski B, Nowak KJ, Domazetovska A, Maxwell AL, Clement S,

Davies KE, et al. Evidence for a dominant-negative effect in ACTA1

nemaline myopathy caused by abnormal folding, aggregation and altered

polymerization of mutant actin isoforms. Hum Mol Genet 2004; 13:

1727–43.

Kaindl AM, Ruschendorf F, Krause S, Goebel HH, Koehler K, Becker C,

et al. Missense mutations of ACTA1 cause dominant congenital

myopathy with cores. J Med Genet 2004; 41: 842–8.

Laing NG, Clarke NF, Dye DE, Liyanage K, Walker KR, Kobayashi Y, et al.

Actin mutations are one cause of congenital fibre type disproportion.

Ann Neurol 2004; 56: 689–94.

Lazarides E. Actin, alpha-actinin, and tropomyosin interaction in the

structural organization of actin filaments in nonmuscle cells. J Cell Biol

1976; 68: 202–19.

Lenart P, Bacher CP, Daigle N, Hand AR, Elis R, Terasaki M, et al. A

contractile nuclear actin network drives chromosome congression in

oocytes. Nature 2005; 436: 812–8.

Matsuzaki F, Matsumoto S, Yahara I, Yonezawa N, Nishida E, Sakai H.

Cloning and characterization of porcine brain cofilin cDNA. Cofilin

contains the nuclear transport signal sequence. J Biol Chem 1988; 263:

11564–8.

Minamide LS, Striegl AM, Boyle JA, Meberg PJ, Bamburg JR.

Neurodegenerative stimuli induce persistent ADF/cofilin-actin rods

that disrupt distal neurite function. Nat Cell Biol 2000; 2: 628–36.

Nakagawa T, Engler JA, Sheng M. The dynamic turnover and functional

roles of alpha-actinin in dendritic spines. Neuropharmacology 2004; 47:

734–45.

Nishida E, Iida K, Yonezawa N, Koyasu S, Yahara I, Sakai H. Cofilin is a

component of intranuclear and cytoplasmic actin rods induced in

cultured cells. Proc Natl Acad Sci USA 1987; 84: 5262–6.

Nowak KJ, Wattanasirichaigoon D, Goebel HH, Wilce M, Pelin K,

Donner K, et al. Mutations in the skeletal muscle alpha-actin gene in

patients with actin myopathy and nemaline myopathy. Nat Genet 1999;

23: 208–12.

Pendleton A, Pope B, Weeds A, Koffer A. Latrunculin B or ATP depletion

induces cofilin-dependent translocation of actin into nuclei of mast

cells. J Biol Chem 2003; 278: 14394–400.

Perrimon N, Engstrom L, Mahowald AP. Zygotic lethals with specific

maternal effect phenotypes in Drosophila melanogaster. I. Loci on the X

chromosome. Genetics 1989; 121: 333–52.

Rajfur Z, Roy P, Otey C, Romer L, Jacobson K. Dissecting the link

between stress fibres and focal adhesions by CALI with EGFP fusion

proteins. Nat Cell Biol 2002; 4: 286–93.

Sparrow JC, Nowak KJ, Durling HJ, Beggs AH, Wallgren-Pettersson C,

Romero N, et al. Muscle disease caused by mutations in the skeletal

muscle alpha-actin gene (ACTA1). Neuromuscul Disord 2003; 13:

519–31.

Stuven T, Hartmann E, Gorlich D. Exportin 6: a novel nuclear export

receptor that is specific for profilin.actin complexes. EMBO J 2003; 22:

5928–40.

Wada A, Fukuda M, Mishima M, Nishida E. Nuclear export of actin: a

novel mechanism regulating the subcellular localization of a major

cytoskeletal protein. EMBO J 1998; 17: 1635–41.

Young KG, Kothary R. Spectrin repeat proteins in the nucleus. Bioessays

2005; 27: 144–52.

3284 Brain (2007), 130, 3275^3284 A. Domazetovska et al.