SPECIAL ISSUE RESEARCH ARTICLE

In Vivo Analyzes of Dystroglycan FunctionDuring Somitogenesis in Xenopus laevisMagdalena Hidalgo,1,2 Cathy Sirour,1 Valerie Bello,1 Nicole Moreau,1 Michele Beaudry,2 andThierry Darribere1*

Dystroglycan (Dg) is a cell adhesion receptor for laminin that has been reported to play a role in skeletalmuscle cell stability, cytoskeletal organization, cell polarity, and signaling. Here we show that Dg isexpressed at both the notochord/somite and the intersomitic boundaries, where laminin and fibronectin areaccumulated during somitogenesis. Inhibition of Dg function with morpholino antisense oligonucleotidesor a dominant negative mutant results in the normal segmentation of the presomitic mesoderm but affectsthe number, the size, and the integrity of somites. Depletion of Dg disrupts proliferation and alignment ofmyoblasts without affecting XMyoD and XMRF4 expression. It also leads to defects in laminin deposition atthe intersomitic junctions, whereas expression of integrin �1 subunits and fibronectin assembly occurnormally. Our results show that Dg is critical for both proliferation and elongation of somitic cells and thatthe Dg-cytoplasmic domain is required for the laminin assembly at the intersomitic boundaries.Developmental Dynamics 238:1332–1345, 2009. © 2009 Wiley-Liss, Inc.

Key words: somitogenesis; muscular differentiation; dystroglycan; laminin; Xenopus laevis

Accepted 20 October 2008

INTRODUCTION

Dystroglycan (Dg) is a membrane-span-ning cell adhesion receptor encoded by asingle gene whose product is cleavedposttranslationally to yield the maturetwo-chain form, �-Dg and �-Dg(Ibraghimov-Beskrovnaya et al., 1992).The �-Dg modulates signal transduc-tion, acting as a scaffold for the extra-cellular-signal-related kinase–mitogen-activated protein (ERK-MAP) kinasecascade (Spence et al., 2004). It hasalso been shown that �-Dg sequestersproteins in separate cellular locationsto regulate their adhesion-dependentactivation (Spence et al., 2004), and

modulates actin reorganization bymeans of Cdc42 (Batchelor et al.,2007). The �-Dg contains a large mu-cin-like domain with potential sitesfor O-glycosylation and binds extra-cellular matrix (ECM) molecules, suchas laminins, perlecan, agrin, andneurexin (Michele and Campbell,2003).

In adult tissue, mutations in the O-glycosylation pathway decrease li-gand-binding activity, which leads todystroglycanopathies (Barresi andCampbell, 2006). Dg was found to beaberrantly expressed in a variety ofhuman cancers including oral squa-

mous cell carcinoma, breast, colon,and prostate cancers (Sgambato andBrancaccio, 2005). During develop-ment, Dg appears to be essential forthe formation of the Reichert’s mem-brane that separates the rodent em-bryo from the maternal circulation(Williamson et al., 1997). In mice, tar-geted disruption of the Dg gene in pe-ripheral nerve revealed decreasednerve conduction velocity, reduced so-dium channel density, and abnormalmyelin sheath folding, suggesting aunique role of Dg for both myelinationand nodal architecture (Colognato etal., 2007). Dg has been implicated in

1Universite Pierre et Marie Curie Paris 6 UMR CNRS 7622, Laboratoire de Biologie du Developpement, equipe Matrice Extracellulaire etDeveloppement, Paris, France2Universite Paris 13, EA 2363, UFR-SMBH 93017-Bobigny Cedex, FranceGrant sponsor: P13 University; Grant number: 26818-2007; Grant sponsor: the “Association pour la Recherche contre le Cancer”; Grantnumber: 3513; Grant number: 7867.*Correspondence to: Thierry Darribere, Universite Pierre et Marie Curie Paris 6 UMR CNRS 7622, Laboratoire de Biologie duDeveloppement, equipe Matrice Extracellulaire et Developpement, 9 quai Saint-Bernard, 75005 Paris, France.E-mail: thierry.darribere@upmc.fr

DOI 10.1002/dvdy.21814Published online 10 December 2008 in Wiley InterScience (www.interscience.wiley.com).

DEVELOPMENTAL DYNAMICS 238:1332–1345, 2009

© 2009 Wiley-Liss, Inc.

branching morphogenesis of lung andsalivary glands (Durbeej et al., 2001).In vivo, depletion of Dg affects kidneymorphogenesis and can lead to renalagenesis (Bello et al., 2008). In Dro-sophila, Dg has been reported to playa fundamental role for polarizing theepithelial cells and the oocytes in theovaries (Deng et al., 2003). In larvalbody wall muscles, errors in muscleattachment, muscle contraction, andmuscle membrane resistance havebeen associated with Dg mutant al-leles and RNAi mediated reduction ofDg (Haines et al., 2007). Originally,Dg was isolated from skeletal musclemembranes and was found to be anessential component of the dystro-phin–glycoprotein complex, whichlinks the ECM surrounding myofibersto the actin cytoskeleton. Disruptionof Dg–dystrophin interaction hasbeen described in Duchenne-type orlimb-girdle-type muscular dystrophy(Deconinck and Dan, 2007). It hasbeen proposed that Dg forms a contin-uous link from the ECM to the actincytoskeleton, providing structural in-tegrity and perhaps transduction sig-nals (Winder, 2001). Whereas its roleis well established in muscles, muchless is known about its implication inearly skeletal myogenesis, in particu-lar during somitogenesis in verte-brates.

In vertebrate embryos, somites areregular transient structures repeatedalong the anterior/posterior axis of theembryo, which then differentiate intoa part of the dermis, bone, cartilage,tendon-cell lineages, and skeletalmuscles. Somitogenesis in Xenopuslaevis displays several unique fea-tures characterized by the orches-trated rotation of blocks of cells. Dur-ing gastrulation, presomitic cells ofthe paraxial mesoderm intercalate ra-dially and mediolaterally, separatefrom the rest of the mesoderm, estab-lish the notochord/somite boundaryand finally form blocks of 7–9 cellswidth (Danker et al., 1992). In thoseblocks, cells change shape, lengthenalong their mediolateral axis and nar-row along their anteroposterior axis(Wilson et al., 1989; Afonin et al.,2006). At the onset of somitogenesis,the presomitic cells bend anteriorlyand undergo a 90° rotation relative tothe anteroposterior axis of the embryoto form mono-nucleate muscle cells

that are aligned parallel to the noto-chord (Keller, 2000). These observa-tions suggest a complex process de-pending on a series of coordinatedchanges in cell shape, cell–cell, andcell–matrix adhesion. Yet little isknown about the molecular pathwaysthat coordinate these changes and ad-hesion during somitogenesis. Type Icadherins are required for cell–celladhesion during rotation and pro-tocadherin for somite boundary for-mation (Kim et al., 2000; Giacomelloet al., 2002). The �5�1 integrin is ex-pressed during somitogenesis and adominant negative form of �1 integrinsubunit alters somite formation(Marsden and DeSimone, 2003).Moreover, a laminin- and fibronectin-containing ECM is localized at noto-chord/somite boundaries and inter-somitic junctions (Wedlich et al., 1989;Fey and Hausen, 1990). We have pre-viously reported that Dg transcriptswere present in Xenopus presomiticmesoderm (PSM) and persisted insomites upon their formation (Moreauet al., 2003), suggesting that Dg pro-teins might play a role in somitogen-esis.

Our results show that Dg is local-ized at notochord/somite boundariesand enriched at the intersomiticjunctions. Using targeted depletionof the protein, we demonstrate thatDg is not required for the segmenta-tion of the PSM and that Dg does notinterfere with the myogenic signal-ing pathway but may have a centralrole in somitic cell proliferation andin the setting of contractile proteinsthat characterize muscular cells. Wedemonstrate that the presence of Dgis crucial for the laminin depositionin the extracellular matrix, for an-choring cells to laminin, and conse-quently for cell alignment and elon-gation within somites. Our datashow that Dg depletion does not af-fect integrin �1 subunit expressionand fibronectin fibrillogenesis atintersomitic junctions. Finally, weestablish by using dominant nega-tive mutant that the cytoplasmic do-main of Dg is required for the lami-nin assembly at the intersomiticboundaries. Together, these dataprovide new in vivo insights into Dgfunction in the developing verte-brate somites.

RESULTS

Dystroglycan Is Localized tothe Intersomitic and theNotochord/Somite Boundaries

We have previously cloned the full-length cDNA of Dg from Xenopus laevisand described the mRNA expressionpattern during early development (Mo-reau et al., 2003). To better characterizeits expression pattern, whole-mount insitu hybridizations were performed atdifferent developmental stages usingdigoxigenin (DIG) -labeled �Dg RNAantisense probes. Dg mRNAs were de-tected in the dorsal region, in the parax-ial mesoderm in a series of stripes andposteriorly in the PSM where newsomites are forming (Fig. 1A–C). Tran-scripts were also detected at the brainlevel, the visceral arches, the otic vesi-cle, the pronephros, and the duct (Fig.1A–C). At stage 45, when all the somitepairs are formed, the transcripts re-mained present in somites. A sense �Dgcontrol probe showed no staining pat-tern in embryos at any stage (data notshown). To carry this study further, weexamined the Dg protein expression intime and space using the monoclonalantibody 43DAG/8D5 raised againstthe human �-Dg carboxy-terminus. Bywhole-mount immunolocalization theprotein was progressively detected fromthe onset of somitogenesis (stage 17). Itwas localized at both the notochord/somite and intersomitic boundariesduring somite formation (data notshown). As development proceeds, Dgwas found to be enriched at intersomiticjunctions, and later at intermyotomaljunctions in tadpoles (Fig. 1D).

Taken together these data show thatDg transcripts are present in the PSMand both the Dg transcripts and theproteins persist in somites upon theirsegmentation and maturation. An obvi-ous question is whether Dg is involvedduring somitogenesis. To address thisissue, we used loss-of-function experi-ments using Dg antisense morpholinosand mRNA encoding a mutated Dgmissing the �-C-terminus tail.

Dystroglycan DepletionCauses a Disorganization ofthe Paraxial Mesoderm

Because Xenopus laevis is allotet-raploid, several pseudoalleles of Dg

DYSTROGLYCAN FUNCTION DURING SOMITOGENESIS 1333

Fig. 1. Dystroglycan transcript and protein expressions. A–C: Lateralview, whole-mount in situ hybridization at stage 21 (A), 23 (B), and 27 (C).Dg mRNAs are detected in the dorsal region, posteriorly in the pre-somitic mesoderm and in the mesoderm in a series of stripes. D: Stage40, lateral view, in toto immunodetection. The staining of the Dg proteinis associated with the intersomitic junctions and the notochord (whitearrow). �m; b, brain; ov, otic vesicle; psm, presomitic mesoderm; s,somites; va, visceral arches; p, pronephros and duct. Scale bar � 400�m.

Fig. 2. Immunodetection of dystroglycan in the dorsal region of mor-pholino (MO) treated embryos. A: Protein extracts of dorsal explantsdissected at stage 24 embryos. (1) Noninjected embryos, a 43-kDa bandcorresponding to �-Dg is detected. (2–4) Embryos injected with 12 ng(2), 20 ng (3) and 28 ng (4) of Dg MO per blastomere. A dose dependentreduction of the band corresponding to �-Dg is highlighted. B: Stage 40,frontal section. Embryo was unilaterally injected at the two-cell stagewith 28 ng of Dg MO. The immunodetection shows that the Dg proteinis present in the control side whereas no staining is observed in theMO-injected side. WT, wild-type; Mo, injected side. Scale bar � 100 �m.

Fig. 3.

1334 HIDALGO ET AL.

exist. The nucleotides sequence of theallele that we have cloned (Moreau etal., 2003) was too divergent aroundthe translational start site with othersalleles identified in the expressed se-quence tag (EST) databases to be tar-geted by one morpholino. Therefore,we have used two Dg antisense mor-pholinos (Dg-MO) to examine the Dgfunction in Xenopus laevis somitogen-esis. Their efficiency has been demon-

strated previously (Bello et al., 2008).However, to verify the depletion of theDg protein in paraxial mesoderm andsomites with the Dg-MO mixture, in-creasing concentrations were injectedbilaterally in two-cell stage embryos.At stage 24, dorsal explants contain-ing somites were dissected and pro-teins were analyzed by Western blot-ting to establish the presence of Dg. Incontrol embryos, a band at 43 kDarepresenting �-Dg was identifiedwhile in Dg-MO injected embryos, the43-kDa band was greatly reduced orabsent in a dose-dependent manner(Fig. 2A). To further control the deple-tion of Dg at the somite level, 28 ng ofthe Dg-MO mixture were unilaterallyinjected at the two-cell stage and thelocalization of Dg was analyzed by in-direct immunofluorescence. On fron-tal sections, the Dg protein waspresent at the intersomitic junctionson the control side, whereas no stain-ing was observed on the MO-injectedside (Fig. 2B). Because both Westernblotting and immunostaining showeddrastically reduced levels of Dg insomites, subsequent knockdown ex-periments were made with a 28-ngmixture of MO.

To determine the requirement of Dgprotein during somitogenesis, unilat-eral injections of Dg-MO were per-formed at the two-cell stage. Injectedembryos developed normally throughcleavage, gastrulation, and early neu-rulation. At the beginning of somito-genesis (stage 17), the morphology ofembryos appears normal except that abend appears along the anterior/pos-terior axis of injected embryos sug-gesting a shortening of the axis (Fig.3A–D). Ninety-eight percent of theDg-MO injected embryos (n � 180)show this phenotype. The bend wasmaintained as the development pro-gressed, and was more pronounced atthe tail bud stages, with the concaveside corresponding to the injectedside. Such curved embryos had neverbeen observed when injected with thestandard MO (data not shown), a five-nucleotide mismatch Dg-MO or a mix-ture of Dg mRNA and Dg-MO (Fig.5Q,R). At older stages, the embryosexhibited jerky movements and a dis-turbed stroke characterized by a cir-cular trajectory suggesting defectiveand/or less muscles in the injectedsides. Somites were counted at differ-

ent stages in both the injected andcontrol sides (n � 35). Between stages24 and 41, there was a reduction of 2to 3 somites on the injected side com-pared with the control side (Fig. 3E).These phenotypes could be the conse-quence of a developmental delay pro-voked by either the MO injection itselfor the Dg depletion. The first sugges-tion seems unlikely because the injec-tion of a five-nucleotide mismatchDg-MO or a mixture of Dg mRNA andDg-MO had no effects on somite num-ber and development (not shown, seeFig. 5Q,R).

To investigate the consequence ofthe Dg depletion, patterning and mor-phology of somites were analyzed bywhole-mount immunostaining withthe 12/101 antibodies, a myotome andskeletal muscular tissue-specificmonoclonal antibody (Kintner andBrockes, 1984). When the dorsal re-gion of the embryos was observed, anextensive disorganization of thesomitic tissue and a weaker intensityof the labeling on the injected sidewere noticed (Fig. 3F–M). On the Dgdepleted side, the somite region wassignificantly reduced when comparedwith the noninjected side (Fig. 3F–I).On the injected side, repetitive unitswere observed (Fig. 3A–D). Theseunits appeared reduced in number.For example, at stage 28, thirteensomite-like structures were observedon the injected side whereas sixteenwere detected in the uninjected side(Fig. 3B). Moreover in the Dg-depletedside, the repetitive units were ob-served but their characteristic chev-ron pattern was often affected (com-pare Fig. 3F,G and Fig. 3H,I). Tofurther analyze these phenotypes, se-lected affected embryos were fixed,sectioned and stained with Hoechstand the 12/101 antibodies (Fig. 3J–L).On transversal sections of the truncalregion of the embryos, somites of theinjected side appeared smaller alongthe dorsoventral and mediolateralaxes than somites of the control side(Fig. 3J–L). On frontal sectionsthroughout the embryo, we observedthat the 12/101 antibody staining wasweaker on the Dg depleted side (Fig.3M). Blocks of somitic cells were dis-cernable in the Dg-MO injected side,but they appeared less compact andless cohesive with each other (Fig.3M). Moreover, the cell organization

Fig. 4. Dystroglycan depletion reduces thecell proliferation in somites. Stage 24 embryosinjected unilaterally were sectioned (n � 8).Then sections were stained with the phospho-Histone H3 antibodies (mitosis marker). Posi-tive cells were counted in somite areas in boththe injected and control sides. The averagenumber of positive cells per somite on the con-trol and the injected sides is compared on thegraph. Bars indicate SE. The difference (38%)between the control and the injected side issignificant (t-test: P � 0.045). WT, wild-type;MO, injected side.

Fig. 3. Somites are disorganized and reducedwhen dystroglycan is depleted. A–E: Embryoswere injected with 28 ng of Dg morpholino (MO)into the left side at the two-cell stage andstained with the 12/101 antibodies. A–D: Dorsalviews of stage 24 (A), 28 (B), 32 (C), and 45 (D).E: Statistical analysis of the somite number atdifferent developmental stages. Bars indicateSE (t-test: Pstage24 � 0.0004; Pstage28 � 0.0014;Pstage32 � 0.005; Pstage41 � 0.047). F–I: An em-bryo at stage 24. F,G: Lateral view of the controlside. The segmentation of the paraxial meso-derm is clearly visible. H,I: Lateral view of theinjected side. The repetitive units often lacktheir characteristic chevron pattern. J–L: Stage24, transversal section (injected side on theright). J: Hoechst (blue). K: Immunodetectionwith the 12/101 antibodies (red). L: Merge. Theinjected sides show a reduction of the somitesize. M: Stage 24, frontal section, immunode-tection with the 12/101 antibodies (injected sideat the right). The segmentation is not affected;the somites are less compact and less cohe-sive. WT, wild-type; Mo, injected side. Scalebar � 200 �m.

DYSTROGLYCAN FUNCTION DURING SOMITOGENESIS 1335

Fig. 5.

1336 HIDALGO ET AL.

within each somite appeared affected(Fig. 3M, see also Fig. 5D–F). Alto-gether, these observations indicatethat depletion of Dg does not affect thesegmentation of the PSM but reducesthe number, the size and the organi-zation of the somites.

Dystroglycan DepletionDecreases Cell Proliferationin Somites

Several steps characterize somito-genesis. After the muscular determi-nation of somitic cells, the premyo-blasts proliferate before aligning

themselves. Embryos injected withDg-MO exhibited a reduction of thenumber and the size of somites sug-gesting that Dg depletion could pro-mote apoptosis or alter cell prolifera-tion. To identify the apoptosis patternin Dg depleted somites, embryos (n �10) were fixed at stage 24, sectionedby a cryostat at a 14�m thickness andterminal deoxynucleotidyl transferase–mediated deoxyuridinetriphosphatenick end-labeling (TUNEL) stained.TUNEL-positive cells were counted on10 frontal sections of 11 somitic blocksper embryo in both the control and theinjected sides. On all sections, no statis-tically significant differences in thenumber of apoptotic cells were foundbetween the two sides of the embryos.

To compare the rate of prolifera-tion of myoblast cell progenitors incontrol and Dg-MO injected sides,the phosphorylated-Histone H3 (P-H3) was used as a mitosis marker.Immunodetections were performedwith the anti–phospho-Histone H3antibodies on cryostat sections ofstage 24 embryos (n � 8). The num-ber of P-H3 positive cells was deter-mined in both the control and theinjected sides as described above.The average of positive cells persomite is presented on Figure 4. Thedata indicate that cell proliferationin somites was reduced by 38% in theDg depleted side compared with thecontrol side (Fig. 4). These resultssuggest that the reduction of thenumber and the size of somites ob-served in Dg-MO injected embryoscould be correlated to a decrease inmyoblast cell proliferation.

Dystroglycan DepletionDisrupts Myotome CellsAlignment and IntersomiticCohesion

During somite formation, somitic cellsare stacked in blocks and undergo a90° rotation relative to the anteropos-terior axis of the embryo to form myo-tome fibers that are aligned parallel tothe notochord (Keller, 2000). On fron-tal section, we have observed that dur-ing somite rotation, the morphology ofnuclei changed. From a round shapebefore the rotation (Fig. 5A,C), theyprogressively adopted an oval/elon-gated shape (Fig. 5A,C,G,J) that couldbe correlated to the elongation of cells.

Moreover, nuclei curved and alignedalong the dorsoventral and proximo-distal axes within each block (Fig.5A,C). Thus at the end of rotation,nuclei are arranged in regularly or-dered stripes so that one stripe corre-sponds to one somite (Fig. 5A,C). Toanalyze somite formation, selected Dgdepleted embryos were fixed, sec-tioned and double stained with the 12/101 antibodies and Hoechst. ThemRNA encoding a membrane-taggedgreen fluorescent protein (GFP) wasalso used to identify cell plasma mem-brane.

On frontal sections, nuclei werealigned within each somite on the con-trol side, whereas they were randomlydistributed on the injected side (Fig.5D–F). Once again we noticed that the12/101 antibodies staining appearedweaker on the injected side of embryos(Fig. 5E). On para-sagittal sections(Fig. 5G–L), we observed a dramaticdisorganization of cells within the Dgdepleted somites (Fig. 5J–L) com-pared with the control side (Fig. 5G–I). Nuclei were not arranged in regu-larly spaced stripes (compare Figs. 5Gand 5J). Moreover, nuclei adopted around shape instead of a typical oval/elongated shape (compare Figs. 5Gand 5J), suggesting that cells failed toelongate and to align parallel to thenotochord. We also observed thatsomites were less cohesive to eachother (Fig. 5K) and that the somiteboundaries were less reactive to the12/101 antibodies than those detectedin the control side (compare Figs. 5Hand 5K). On frontal and para-sagittalsections of embryos expressing themembrane-tagged GFP, somites werecorrectly arranged in the control side(Fig. 5M,O). Cells that had rotatedwere aligned parallel to the notochordand showed the characteristic chevronpattern (Fig. 5M,O). In the Dg de-pleted side, somites were disorganizedand cells were randomly arranged(compare Fig. 5M and 5N and Fig. 5Oand 5P). They were not able to antero-posteriorly extend and to align withineach somite (compare Fig. 5M and 5Nand Fig. 5O and 5P). Moreover, theintersomitic junctions failed to formcorrectly (Fig. 5N,P). These pheno-types suggest that myoblasts have ini-tiated their rotation but fail to align,to extend and to establish a correctanchoring at intersomitic junctions.

Fig. 5. Dystroglycan depletion disrupts myo-tome cells alignment and intersomitic cohesion.A,B–C: Stage 24. Control embryo sectionedand stained with Hoechst (blue, A), and immu-nostained with the 12/101 antibodies (red, B);merged image (C). Arrows show cells undergo-ing their rotation. D–L: Embryos were unilater-ally injected at the two-cell stage with 28 ng ofDg morpholino (MO). At stage 24, embryoswere fixed, sectioned and double stained withthe 12/101 antibodies (red) and Hoechst (blue).D–F: Frontal section. The nuclei are randomlydistributed on the injected side (lower part ofthe panel). G–I: Para-sagittal section, controlside of the embryo. G: The nuclei exhibit atypical oval/elongated shape and are aligned ineach somite. H: The somites are strongly cohe-sive the ones with the others. The 12/101 stain-ing is concentrated at the intersomitic junctions(white arrows). I: Merge. J–L: Para-sagittal sec-tion of the injected side. J: The nuclei are ran-domly distributed. K: The somites are less com-pact and less cohesive; the 12/101 staining isabsent at the intersomitic junctions (white ar-rows); within somites, the cell alignment ap-pears affected. L: Merge. M–P: Embryos wereunilaterally injected at the two-cell stage withDg MO and mRNAs encoding a membrane-tagged green fluorescent protein (GFP). M:Stage 24, frontal section, control side. Cells arealigned parallel to the notochord and somiteboundaries are well delimited (white arrows). N:Stage 24, frontal section, injected side. Thesomites are disorganized. Cells fail to extendand align within each somite. The intersomiticjunctions fail to form correctly (white arrows). O:Stage 24, para-sagittal section, control side.Cells are aligned. Somite boundaries exhibittheir characteristic chevron pattern (white ar-rows). P: Stage 24, para-sagittal section, in-jected side. Cells fail to align and boundariesare not well delimited (white arrows). Q: Frontalsection of a rescued embryo stained withHoechst (blue) and the 12/101 antibodies (red).R: Frontal section of an embryo injected with afive-nucleotide mismatch Dg-MO and stainedwith Hoechst (blue) and the 12/101 antibodies(red). The somite morphology and their numberare not affected. Cells are aligned parallel to thenotochord. WT, wild-type; Mo, injected side.Scale bar � 70 �m.

DYSTROGLYCAN FUNCTION DURING SOMITOGENESIS 1337

To check the specificity of observedphenotypes, co-injections of 400 pg ofXenopus full-length Dg mRNA withthe Dg-MO were performed. Embryoswere fixed, sectioned and doublestained with the 12/101 antibodiesand Hoechst (Fig. 5Q). Experimentswere also done with the standard MO(data not shown) and a five-nucleotidemismatch Dg-MO (Fig. 5R). In thesethree cases, no obvious morphologicaland cell arrangements abnormalitieswere observed in somites. Theseknockdown, control and rescue exper-iments show that Dg depletion resultsin somite defects including misorien-tation of cells and impairment ofintersomitic boundary formation.

Dystroglycan Depletion DoesNot Disrupt the MyogenicSignaling Pathway inSomites

The above observations (Fig. 3K,M,5E) show that the 12/101 antibodieslabeling is reduced in myotome cells ofthe Dg-MO injected side. To verifythese observations, we have used thein vitro system to induce muscle tis-sue in explants derived from the ani-mal region of the blastula, so-calledanimal caps (Okabayashi andAsashima, 2003). Dg-MO was injectedat the two-cell stage into animal poleof both cells and animal caps wereisolated from the blastula (stage 9).After incubation with activin, animalcaps were grown until stage 24 equiv-alent. Then, explants were fixed, sec-tioned and double stained with the 12/101 antibodies and Hoechst. Incontrols (nWT � 45), explants were or-ganized in elongated structures (datanot shown), and contractions of ex-plants were observed, suggesting thepresence of contractile proteins, char-acteristic of muscular fibers. By con-trast, Dg-depleted explants (nMo � 60)were spherical and not able to con-tract. Sections of control explantsshowed a large block of muscle cellspresents in the center and the immu-nostaining revealed that these cellswere reactive to the 12/101 antibodies(Fig. 6A). The Hoechst staining indi-cated that some nuclei were alignedwithin the structure but they werenever arranged in regular spacedstripes like in somites (Fig. 6B). Incontrast, the fluorescence was absent

on sections of MO-treated explants(Fig. 6D–F), and nuclei were notaligned within the structure (Fig.6E,F). These observations suggestthat muscle cell differentiation did notoccur in Dg depleted explants (Fig.6D–F).

Because the expression of the 12/101 muscle marker was weaker or ab-sent in MO-injected embryos or ex-plants, it is possible that Dg depletioncould affect the somitic fate by inhib-iting the expression of genes involvedin the muscle cell differentiation.Therefore, we investigated whetherthe Dg interfered with the myogenicsignaling pathway by analyzing theexpression pattern of XMyoD andXMRF4, respectively the earliest anda late marker of muscle cell differen-tiation (Hopwood et al., 1992; DellaGaspera et al., 2006). The expressionof these specific genes was analyzed inDg-MO injected embryos and ex-plants, by whole-mount in situ hybrid-ization and immunostaining of sec-tions.

In embryos, the comparison be-tween the MO-injected side and thecontrol side did not reveal significantdifference in the expression of XMyoDat any stage. At stage 28, blocks ofsomites were apparent but boundarieswere disorganized (Fig. 6H,H�). Onfrontal section, indirect immunofluo-rescence with an anti-XMyoD anti-body revealed a nuclear staining inboth the control and the injected sides(Fig. 6I). In control and Dg-MO ex-plants, the XMyoD transcript (Fig.6J,K) and protein (Fig. 6L,M) were de-tected, suggesting that the muscularinduction took place, notably in MOinjected tissues that were not reactiveto the 12/101 antibodies (Fig. 6D–F).

To analyze the late cell differentia-tion, we investigated the expressionpattern of XMRF4, a late myogenictranscription factor. Whole-mount insitu hybridization revealed that tran-scripts were expressed in both the con-trol and the Dg depleted sides al-though the somites structure wasdisrupted (Fig. 6N,N�). On frontal sec-tion, immunostaining with the anti-XMRF4 antibodies revealed a nuclearlocalization of this factor in both thecontrol and the injected sides of em-bryos (Fig. 6O). This result was con-firmed on sections of control andDg-MO explants (Fig. 6P–P�).

Together, these data show that Dgdepletion does not disturb the XMyoDand XMRF4 expression pattern andsuggest that the Dg does not interferewith the myogenic signaling pathwayin somites but with somite morpho-genesis.

Dystroglycan DepletionPrevents the LamininDeposition at the SomiteBoundaries

Dg is one of the known laminin recep-tors. Previous data in mice suggestthat it plays a critical role in the for-mation of basement membrane duringearly development by anchoring thismatrix component to the cell surface(Williamson et al., 1997). Becauselaminin is present in Xenopus aroundthe notochord and at the intersomiticjunctions (Fig. 7A), we investigatedthe possibility that the alignment de-fect of myoblasts within somites wasdue to a disrupted laminin organiza-tion at the intersomitic boundaries.Dg-MO were injected at the two-cellstage into one or both blastomeres (seethe Experimental Procedures section),embryos and explants were fixed, sec-tioned and stained with the anti-lami-nin antibodies. Both sides of the em-bryo displayed a laminin distributionaround the notochord; however, lami-nin was absent at the intersomiticjunctions where Dg was depleted (Fig.7B,C). In MO-treated explants, nolaminin deposition was detected (com-pared Fig. 7D with 7E). Together,these findings demonstrate that theDg is required to deposit the lamininin the extracellular matrix, at theintersomitic boundaries. They supportthe possibility that the cell disorgani-zation within somites, in Dg depletedembryos, could be the consequence ofthe absence of laminin at the inter-somitic junctions.

Integrins are also known to functionas receptors for laminin (Henry et al.,2001). To verify the organization inthe somitic tissue of the beta1 inte-grins, we used the 8C8 monoclonal an-tibody. On frontal sections, the indi-rect immunofluorescence indicatedthat integrin �1 subunits were con-centrated at both the intersomitic andthe notochord/somite boundaries inboth the injected and the control sides(Fig. 7F). Integrins can also bind fi-

1338 HIDALGO ET AL.

bronectin and have been shown to beessential for the formation of somites(Rifes et al., 2007). To analyze the or-ganization of this other matrix compo-nent in the absence of Dg, we used ananti-fibronectin antibody. On frontalsections, this protein is distributedaround the notochord and at the inter-somitic boundaries in both the controland injected sides (Fig. 7G). The invitro experiments support these re-sults because both integrin �1 sub-units and fibronectin were detected inthe Dg depleted structures (Fig. 7H,I).

The above results show that the Dgdepletion in somites affects interac-tion between cells and the extracellu-lar matrix containing laminin atintersomitic boundaries. Dg is knownto interact with adaptors and signal-ing molecules such as Grb2, ERK, andMAPK (Spence et al., 2004). To deter-mine whether the cytoplasmic domainof Dg was required during somitogen-esis, a construct coding for a GFP-tagged Dg missing the �-C-terminustail (Dg-�Cyto) was used. At the eight-cell stage, embryos were unilaterallyinjected into the presumptive somiteregion with Dg-�Cyto mRNA. Simi-larly to the MO-injected embryos, theDg-�Cyto mRNA injected embryosdeveloped normally until tail budstage when a bend appeared along theanterior/posterior axis. Immunohisto-chemistry using the 12/101 antibodiesand Hoechst revealed a similar pheno-type to the Dg-MO injected embryos: aweaker 12/101 staining, a defective co-hesion between somites, a dramaticdisorganization of cells and nuclei(Fig. 7J–L). As expected, immunohis-tochemistry using the anti-lamininantibodies showed a positive stainingat both the somite/notochord andintersomitic boundaries in the controlside of embryos (Fig. 7J,K). In the Dg-�Cyto mRNA-injected side, the lami-nin was present at the somite/noto-chord boundaries, but its deposition atthe intersomitic junctions was se-verely disrupted (Fig. 7J,L). Interest-ingly, the fluorescence appearedsparse and discontinuous within theputative somites (Fig. 7L). Using anantibody to GFP to detect mutant Dg-�Cyto proteins, we showed thatsomitic cells expressing the Dg-�Cytoproteins were also stained by the anti-laminin antibodies (Fig. 7M,N). Thisresult indicated that cells expressing

Fig. 6. Dystroglycan depletion does not disrupt the myogenic signaling pathway. A–C: Sectionsof control explant at stage 24 equivalent. A: Immunostaining with the 12/101 antibodies. B:Hoechst. Some nuclei are aligned. C: Merge. D–F: Sections of morpholino (MO) -injected explantat stage 24 equivalent. D: Immunodetection with the 12/101 antibodies. No staining is observed. E:Hoechst. F: Merge. G–H�: Whole-mount in situ hybridization of XMyoD. G: Stage 24, dorsal view ofunilaterally injected embryo (injected side on the right). H: Stage 28, lateral view of the control side.H�: Stage 28, lateral view of the injected side. The intersomitic junctions are disorganized (whitearrows). I: Stage 24, frontal section, immunodetection of the XMyoD protein (injected side at theright). J,K: Stage 24 equivalent, whole-mount in situ hybridization of XMyoD. J: Control explant. K:MO-injected explant. L,M: Stage 24 equivalent, immunodetection of the MyoD protein. L: Controlexplant. M: MO-injected explant. N–N�: Stage 28, whole-mount in situ hybridization of XMRF4. N:Lateral view of the control side. N�: Lateral view of the injected side. O–P�: Stage 28, immunode-tection of the MRF4 protein. O: Frontal section (injected side on the right). P: Control explant. P�:MO-injected explant. WT, wild-type; Mo, injected side. Scale bar � 70 �m.

DYSTROGLYCAN FUNCTION DURING SOMITOGENESIS 1339

the Dg-�Cyto were able to bind thelaminin on their surface but the poly-merization of this extracellular matrixcomponent was disrupted when thecytoplasmic domain of Dg was absent.

Together, these findings demon-strate that Dg depletion severely dis-rupts the extracellular laminin depo-sition at the intersomitic boundariesbut it does not alter the localization offibronectin and integrins of the �1family. They also show that �1 familyintegrins cannot compensate for thelack of Dg and cannot accumulate thelaminin at the intersomitic bound-aries. Finally, they show that somiticcells expressing a cytoplasmic mutantof Dg are able to bind laminin but thecorrect assembly of this extracellularmatrix component at the intersomiticjunctions is severely compromised.

DISCUSSION

In this in vivo and in vitro study, weprovide several lines of evidence for afundamental role of Dg in Xenopuslaevis somitogenesis. First, inhibitionof Dg synthesis or overexpression ofthe Dg construct lacking the cytoplas-mic domain, affected the size, the in-tegrity and the cellular organizationwithin somites, independently of thePSM segmentation. Second, Dg deple-tion decreased cell proliferation insomites, affected myotome cell elonga-tion, did not disrupt the somitic fatebut interfered with the muscular dif-ferentiation. Third, somitic cells ex-pressing the cytoplasmic mutant of Dgwere able to bind laminin on their sur-face but the correct laminin assemblyat the intersomitic junctions was se-verely compromised. Finally, the de-pletion of Dg prevented the laminindeposition without disturbing integrin�1 subunit distribution and fibronec-tin organization in the extracellularmatrix, at the intersomitic bound-aries. These data demonstrate thatDg/laminin interactions ensure thematrix laminin deposition at theintersomitic junctions and the cellelongation within somites, two succes-sive steps required for cell orientation,somite organization, and muscle celldifferentiation. They also demon-strate that the cytoplasmic domain ofDg is required for the laminin assem-bly at the intersomitic boundaries.

Dg transcripts are expressed in the Fig. 7.

1340 HIDALGO ET AL.

Xenopus PSM and persist in somitesupon their formation (this study andMoreau et al., 2003). Here, we showthat the protein is detected from theonset of somitogenesis and is enrichedat intersomitic junctions and at thenotochord/somite boundaries. Somito-genesis requires the mediolateralelongation of presomitic cells, the for-mation of filopodia protrusions at theonset of rotation, the break of adhe-sive interactions with the ECM at thenotochord/somite boundary, the bend-ing of cells around the dorsoventralaxis and finally the elongation andalignment of cells parallel to the noto-chord (Afonin et al., 2006). Dg ispresent during these events suggest-ing that Dg/ECM interactions mightgovern adhesive properties of cellsduring somitogenesis and muscle for-mation. To address this question, wecarried out loss-of-function experi-ments with Dg morpholino antisenseoligonucleotides and overexpression ofa Dg construct that lacks the cytoplas-mic domain.

Two oligonucleotides known for tar-geting the translation of Dg were used(Bello et al., 2008). We tested theirspecificity in MO-injected embryos byWestern blotting of proteins extractedfrom somites and by indirect immuno-fluorescence. These two methods

showed a specific depletion of Dg with28 ng of MO. This MO amount wasinjected into one side of the embryosin the region that will become somites.During the development, embryos ex-hibited a bend along the anterior/pos-terior axis with the concave side cor-responding to the Dg depleted side.Older embryos had difficulties to swimand presented a circular trajectory.These observations reflect a right/leftasymmetry of the dorsal region of theembryos, suggesting that either thesegmentation of the PSM or the somi-togenesis were affected. We found thatthe depletion of Dg leads to a normalsegmentation of the PSM, because re-petitive units have been observed. Asdevelopment occurs, the repetitiveunits give rise to abnormal somites,reduced in number and smaller insizes. Abnormal somite boundary for-mation was also observed with a laterloss of the canonical chevron-like con-formation. Because the proliferationof premyoblasts is known to be anearly step of the myogenesis, one hy-pothesis was that the depletion of Dgcould increase apoptosis during thesomites development. We did not ob-serve fragmented nuclei or pycnoticcells after MO injections making thispossibility rather unlikely. However,we cannot exclude that apoptosis oc-

curs before the stages that we haveanalyzed, for instance in younger em-bryos, during the segregation betweenthe notochord and the paraxial meso-derm at neurulation. Another possi-bility would be that the depletion ofDg affects cell proliferation in devel-oping somites. In agreement with thishypothesis, we showed a correlationbetween Dg depletion and a reductionof 38% of cell proliferation, leading toa reduction of the somitic tissue andthe muscle mass among older embryos.This suggests that Dg function influ-ences intracellular signaling processesinvolved in the cell cycle regulation.Previous studies have provided evi-dence that Dg influences the function ofthe intracellular phosphatase PTEN, akey molecule regulating several signal-ing cascades such as the PI3-K/AKT cellsurvival pathway (Muschler et al.,2002; Sgambato and Brancaccio, 2005).Of interest, in cell cultures, Dg has beenshown to modulate the cell cycle by af-fecting extracellular signal-regulatedkinase levels (Higginson et al., 2008).Although further studies are required,our results are consistent with those de-scribing the role of Dg in Xenopus kid-ney development (Bello et al., 2008) andwith the hypothesis that Dg is a compo-nent of the signaling cascade regulatingcell proliferation during early develop-ment.

We found that treatment with MOinduced changes in somite morphol-ogy. One possibility was that it couldbe the consequence of a disruption ofsomite specification. We, therefore, in-vestigated whether the PSM was ap-propriately specified. The data ob-tained in vivo and in vitro show that,in absence of Dg, the transcript andprotein expression patterns of bothMyoD and MRF4, the earliest and latemarkers of muscle cell fate, were notdisturbed. Both markers showed ap-propriate temporal and spatial ex-pression patterns all along somito-genesis suggesting that somiticspecifications do not require Dg/ECMinteractions. This is in agreementwith similar phenotypes and conclu-sions that have been previously re-ported in inhibition experiments ofthe integrin �5 and �1 subunits and ofthe fibronectin in Xenopus as well asin mice and Zebrafish, in which MyoDexpression is not affected (Drake etal., 1992; George et al., 1993; Giaco-

Fig. 7. Dg depletion and Dg without the cytoplasmic domain prevents laminin deposition at thesomite boundaries. A–E: Immunodetection of the laminin. A: Frontal section of control embryo,stage 24. The laminin is present around the notochord, the neural tube, and at the intersomiticjunctions. B: Frontal section of the ventral region of a unilaterally morpholino (MO) -injected embryo,stage 24. Laminin is detected at the intersomitic boundaries in the uninjected side (left) and absentin the injected side (right). C: Frontal section of the dorsal region of a unilaterally Mo-injectedembryo, stage 28. The laminin is absent in the injected side (right) but present at the intersomiticjunctions of the control side. D,E: Immunodetection of the laminin in explants. D: Laminin isdeposited in control explant. E: MO-treated explant. Laminin was not detected. F: Frontal section,stage 28. Immunodetection of the integrins (injected side on the right). The integrins are localizedat the intersomitic junctions in both the control and injected sides. G: Frontal section, stage 28.Immunodetection of the fibronectin (injected side on the right). The fibronectin is detected at thesomite boundaries and around the notochord. H,I: MO-injected explants. H: Immunodetection ofthe integrins. I: Immunodetection of the fibronectin. J–N: Dg-�Cyto mRNA was delivered unilater-ally and effects of the overexpression of the Dg-cytoplasmic mutant was analyzed at stage 24 onfrontal section. J–L: Immunochemistry with Hoechst (blue) and the anti-laminin antibodies (red). J:Laminin is detected around the notochord and at the intersomitic boundaries in the control side(left). Laminin is absent at the intersomitic junctions in the injected side (right). K: Magnification ofthe control side. The nuclei exhibit a typical oval/elongated shape and are aligned in each somite.The laminin is present at both the somite/notochord and intersomitic boundaries. Somites arecohesives. L: Magnification of the Dg-�Cyto mRNA-injected side. The nuclei are randomly distrib-uted. The laminin is present at the somite/notochord boundaries, but its deposition at the inter-somitic junctions is severely disrupted. The fluorescence appears sparse within the putativesomites (white arrows). M,N: Immunodetection of the green fluorescent protein (GFP) -taggedDg-�Cyto (green) and the laminin (red). M: The Dg-�Cyto proteins are detected at the surface ofsomitic cells (white arrows). N: Magnification of one cell expressing the Dg-�Cyto protein. Somiticcells expressing the Dg-�cyto proteins (white arrows) were also stained by the anti-lamininantibodies (blue arrows) and both proteins are co-localized at the level of the plasma membrane.Ch, notochord; TN, neural tube; WT, wild-type; Mo, injected side. Scale bar � 100 �m.

DYSTROGLYCAN FUNCTION DURING SOMITOGENESIS 1341

mello et al., 2002; Marsden and DeSi-mone, 2003; Kragtorp and Miller,2007). These data contrast with phe-notypes obtained with inhibition oftype I cadherins where MyoD is down-regulated (Giacomello et al., 2002).One potential explanation would bethat cell–cell interactions have a cen-tral role in muscle cell fate, whereascell–ECM interactions are required insomite formation and maintenance.This would suggest that the observedphenotypes are due to an alteration inthe morphological processes duringsomitogenesis, which may be attrib-uted to an incorrect modulation of in-teractions between somitic cells andthe extracellular matrix.

We have observed that the 12/101monoclonal antibodies that specifi-cally recognize a not yet identifiedepitope of the differentiated musclecells (Kintner and Brockes, 1984), ap-peared weaker on the injected side ofembryos and was absent in MO-in-jected explants, although MyoD andMRF4 are expressed. This apparentdiscrepancy between in vivo and invitro data might be due to artifacts ofthe immunfluorescence analysis.However, we observed the sameweaker staining in dominant negativeexperiments and we did not observe

differences in 12/101 staining in boththe rescue and the five-nucleotide mis-match MO experiments (see Figs. 5Qand 5R, respectively), making thispossibility rather unlikely. It appearsmore likely that, in Dg depleted cells,differences in 12/101 staining in bothin vivo and in vitro conditions mightresult from changes in tissue environ-ment. Indeed the patterning and thefate determination of somitic cells oc-cur in response to extrinsic signals,produced and secreted from the adja-cent tissues in particular the neuraltube, the notochord, the surface ecto-derm, and the somitic compartmentthemselves. Candidate molecules forthese signaling activities includeSonic Hedgehog, the Wnt proteins,Noggin, and BMP-4 (Yusuf andBrand-Saberi, 2006). It is possiblethat in MO-treated explants, one ormore of these proteins are absent ormisexpressed leading to an incom-plete muscle cell differentiation insynergy with the depletion of Dg andexplaining the lack of 12/101 stainingthat is specific of differentiated musclecells. This possibility is supported bythe fact that control explants are ableto contract whereas MO-treated ex-plants do not. On the contrary, in MOtreated embryos, the adjacent tissues

are present and may provide sufficientsignaling molecules to activate the 12/101 epitope expression. However, thedepletion of Dg could induce a deple-tion of a scaffold to which the proteinrecognized by the 12/101 antibodiescould bind and thus localized at theintersomitic junctions. This hypothe-sis is consistent with reports showingthat Dg is required to target proteinsto membrane ruffles in fibroblast cellsor to the subsarcolemmal compart-ment of muscle fibers (Spence et al.,2004). Nevertheless, it remains to bedetermined if this is a direct or anindirect consequence of the Dg loss offunction, caused for example by theloss of intersomitic junctions and thesubsequent lack of specialization ofmyoblast cell extremities. The ob-served difficulties of MO treated em-bryos to swim may be due to the loss ofthis somite structural integrity.

Dg depletion and overexpression ofthe Dg deleted of the cytoplasmic do-main results in a dramatic disorgani-zation of the myotome and a misorien-tation of cells within somites. Nucleiadopted a round shape instead of atypical oval/elongated shape and nu-clei were not arranged in regularlyspaced stripes suggesting that cellswere unable to elongate and to extendparallel to the notochord. Tracing MO-depleted cells with a membrane tar-geted GFP confirm these results. It isestablished that the Dg is required inthe Drosophila to organize the cy-toskeleton and to ensure the cellularpolarity of epithelial cells and oocytesin the ovary (Deng et al., 2003). More-over, the cytoplasmic domain of �-DGinteracts with ezrin, a protein able tomodulate the actin reorganization andto induce the formation of filopodia atthe periphery of cells (Batchelor et al.,2007). Therefore, we propose that cellsdepleted in Dg rotate but exhibit dif-ficulties to orient within each blockbecause they are unable to anchor tothe intersomitic ECM. So, they cannotprovide the traction required to ex-tend during their alignment leading toa random distribution of cells in themyotome. In agreement with this pos-sibility, we have observed that the de-pletion of Dg affects the intersomiticcohesion and the organization ofsomitic boundaries.

Dg binds by means of its �-subunit tolaminin-1 (Ervasti and Campbell,

Fig. 8. Model of dystroglycan function during Xenopus somitogenesis. Diagram illustrating how Dgmay contribute to somitogenesis. During gastrulation, presomitic cells of the paraxial mesodermintercalate radially and mediolaterally and separate from the rest of the mesoderm (Keller, 2000). Theyestablish interactions with the extracellular matrix (ECM) at the notochord/somite boundaries. Interac-tions between �5 integrins and fibronectin allow cells to bend anteriorly and to undergo a 90° rotationrelative to the anteroposterior axis (Kragtorp and Miller, 2007). The model predicts that after theirrotation the somitic cells expressing Dg assemble laminin in the ECM at the intersomitic boundaries.Then, cells anchor to the fibronectin/laminin matrix, elongate and align parallel to the notochord. Finally,both integrin/fibronectin and Dg/laminin systems function cooperatively in maintaining somite bound-aries and later somite cohesion and integrity. (Adapted from Alfonin et al., 2006.)

1342 HIDALGO ET AL.

1993), laminin-2 (Talts et al., 1999), andlaminin-10/11 (Yu and Talts, 2003). Inlaminin-1, laminin-2, and laminin-10/11, the association with Dg has beenmapped within the LG4, LG1-3 plusLG4-5, and LG4-5 regions of their�-chains, respectively (Talts et al.,1999; Yu and Talts, 2003). The poly-clonal antibody L9393 recognized thelaminin �1 or �1 chains in addition tothe isoform specific laminin �1 chain.Laminin isoforms prevalent during Xe-nopus somitogenesis have not yet beenelucidated. During mouse somitogen-esis, laminin �1 has been detected indermomyotome (Tiger and Gullberg,1997) and it has been shown to be ex-pressed in somite/notochord area in Ze-brafish (Pollard et al., 2006). Therefore,one possibility is that Dg might bindlaminin-1 during Xenopus somitogen-esis. Because other laminin chains likethe �2, �4, and �5 have overlappingfunctions with laminin �1 chain (Pol-lard et al., 2006), we cannot excludethat these chains participate also to Dgbinding. Although further studies areneeded, our results demonstrate thatDg expressed by myoblasts has a piv-otal role in laminin deposition at somiteboundaries but not at somite/notochordjunction.

In this study, we have shown thatlaminin is present around the noto-chord in both the Dg-depleted embryosand Dg-cytoplasmic mutants but it isabsent at the intersomitic junctions. Weobserved that laminin is absent in bothMO-treated embryos and explants butsomitic cells expressing the Dg-cyto-plasmic mutant bind laminin to theircell surfaces. Moreover, our data showthat fibronectin is deposited in the ECMat both the somite/notochord and theintersomitic junctions, in MO treatedembryos. They also show that the inte-grin �1 subunit is expressed on Dg-de-pleted cells but that they do not accu-mulate laminin. Integrins are requiredin mice, chick, and Zebrafish for somiteboundary formation (Yang et al., 1999,Zagris et al., 2004; Koshida et al., 2005;Bajanca et al., 2006). In mouse embry-onic stem cells, Dg-laminin interactionhave been shown to be essential for theinitial binding of laminin to cell surface,whereas �1-integrins were required forsubsequent laminin–matrix assembly(Henry et al., 2001). In Xenopus laevis, anECM containing fibronectin and lamininis deposited at the somitic boundaries

throughout somitogenesis (Wedlich et al.,1989; Fey and Hausen, 1990; Danker etal., 1992; this work). Fibronectin appearsin the first steps of somitogenesis, duringrotation, whereas laminin occurs aftersomites have already been formed(Wedlich et al., 1989). A dominant nega-tive form of integrin �1 subunit alterssomite formation (Marsden and DeSi-mone, 2003). Knockdown of the �5 inte-grin prevents fibronectin deposition inthe matrix and cell rotation withinsomites (Kragtorp and Miller, 2007). To-gether these findings suggest that Dg and�1-integrin functions are nonoverlappingduring somitogenesis and support the fol-lowing model for integrins and Dg func-tions during Xenopus laevis somitogen-esis (Fig. 8). In this model, integrin/fibronectin interactions allow fibronectinfibrillogenesis at the intersomitic junc-tions and rotation of somitic cells(Wedlich et al., 1989; Danker et al., 1992;Kragtorp and Miller, 2007), whereas Dg/laminin interactions ensure the deposi-tion of laminin in the intersomitic matrixand the elongation of cells within somites.These two steps are required for cell ori-entation, intersomitic cohesion and orga-nization, and then muscle cell differenti-ation. Later in development, both Dgand integrin/ECM interactions are in-volved in the maintenance of somiteboundaries and then in somite cohe-sion and integrity.

EXPERIMENTALPROCEDURES

Xenopus laevis Embryos andAnimal Cap Assays

Xenopus laevis embryos were obtainedas previously described (Bello et al.,2008). At stage 9 (Nieuwkoop andFaber, 1967), the presumptive ecto-derm explants (animal cap) were iso-lated in 1� MBS (Modified Barth’s Sa-line) and induction of muscle wasobtained by incubation for 1 hr in 10ng/ml of activin (A4941 Sigma). Thenexplants were cultured in 1� MBS un-til the appropriate stage.

Morpholino andMicroinjections

Two specific antisense morpholino oli-gomers (Dg-MO1 and Dg-MO2) wereused in this study, and the sequenceswere 5� CAGCACACCTAATATCCA-

TTTTGGC 3� (Dg-MO1) and 5�TGTTA-CAGCGTAGGAGGCA 3� (Dg-MO2).GenBank searches failed to detect sig-nificant homologies of the two morpho-linos elsewhere in the Xenopus genome.Gene Tools MO-standard (5� CCTCT-TACCTCAGTTACAATTTATA 3�) anda five-nucleotide mismatch Dg-MO (5�TcTTAgAGCcTAGcAGGgA 3�) wereused for control morpholino injections.The morpholinos were suspended insterile water to a concentration of 1mM. Morpholino oligos were injected atthe two- or four-cell stage into one orboth blastomeres depending on the ex-periment.

Plasmid Constructs andmRNA Injection Experiments

Two constructs were engineered fromDg cDNA cloned into the EcoRI andXhoI sites of the polylinker of the pBlue-script II SK () vector (Moreau et al.,2003): the wild-type dystroglycan (Dg-wt), and a construct, which the �-cyto-plasmic domain was deleted (Dg-�cyto).The constructs were amplified by poly-merase chain reaction (PCR), subclonedin the pcR II-TOPO vector (Invitrogen),and sequenced. The dystroglycan con-structs were then cloned into the pCS2mycGFP vector as described previously(Bello et al., 2008). A GFP-expressionplasmid in the pCS2 vector was alsoused. Synthetic capped mRNAs weremade by in vitro transcription as de-scribed in Djiane et al. (2000). Embryoswere injected at the two, four or eight-cell stage with Dg-wt, Dg-�cyto mRNAsor membrane-tagged GFP mRNA intothe mediolateral zone of blastomeres totarget the region that will give rise tothe paraxial mesoderm.

Western Blot Analysis

At stage 24, dorsal regions were re-moved in 1X MMR and explants werehomogenized in lysis buffer (1% TritonX100, 150 mM NaCl, pH 7.5, 10 mMTris, 1 mM ethylenediaminetetraace-tic acid [EDTA], 1 mM ethyleneglycol-tetraacetic acid [EGTA], 0.5% NP40,0.2 mM phenylmethyl sulfonyl fluoride[PMSF]) supplemented with protease in-hibitors. Protein samples, separated by12% sodium dodecyl sulfate-polyacrylam-ide gel electrophoresis, were transferredonto nitrocellulose (Hybond) as describedby Towbin (Towbin et al., 1979). The

DYSTROGLYCAN FUNCTION DURING SOMITOGENESIS 1343

membrane blocked in 1% bovine serumalbumin was incubated with the mouseanti-� dystroglycan antibody (43DAG1/8D5, Novacastra, Newcastle-upon-Tyne,UK, 1:25), and then with the anti-mouseantibody conjugated with biotin (JacksonImmunoResarch 715-065-150, 1:20,000)and streptavidin coupled to peroxidase(Immunotech 016-030-084, 1:10,000).The revelation was made by chemilumi-nescence (kit super signal West Pico Che-mioluminescent substrate, Pierce).

Synthesis of Probes andWhole-Mount In SituHybridization

Plasmids were incubated with the ap-propriate restriction enzyme, BamHI orApaI. Transcription was realized usingSp6 RNA polymerase and probes weresynthesized with a DIG labeling kit(Roche). Embryos were fixed inMEMFA (0.5 M MOPS, pH 7.4, 100 mMEGTA, 1 mM MgSO4, 4% formalde-hyde) for 1 hr and stored in methanoluntil used. Whole-mount in situ hybrid-izations were carried out as describedby Harland (Harland, 1991). Probeswere visualized with anti-DIG antibodycoupled to alkaline phosphatase (Roche11093274910, 1:2,000) and 5-bromo-4-chloro-3-indolyl phosphate/nitrobluetetrazolium (BCIP/NBT; Sigma) for thecolor reaction.

Immunochemistry

Whole-mount immunostaining.

Embryos fixed in MEMFA were incu-bated with a muscle specific 12/101antibody (DSHB I9393, 1:1,000) andwith the anti-mouse secondary anti-body alkaline phosphatase-conjugated(Jackson Immunoresearch 115-056-062, 1:40,000). BCIP/NBT was usedfor the color reaction.

Immunostaining of frozen sections.

Embryos were embedded with 15%cold-water fish gelatin (FLUKA, bio-chemika) and 15% sucrose. Tissueswere sectioned at 14-�m thickness bya cryostat (Leica CM 3050S). Sectionswere blocked in 20% goat serum andthe following primary antibodies wereused: mouse anti-� dystroglycan(43DAG1/8D5,Novacastra,Newcastle-upon-Tyne, UK, 1:50), mouse mono-clonal 12/101 (DSHB I9393, 1:2,000),

rabbit anti-laminin (Sigma L9393,1:25), mouse anti-MyoD (D7F2, DSHB),guinea pig anti-MRF4 (kindly providedby Dr. B. Della Gaspera, Universite Paris5, France, 1:50), mouse anti-�1 integrin(8C8, kindly provided by Dr. D. Alfan-dari, University of Massachusetts, Low-ell, MA; 1:400), mouse anti-fibronectin(4H2, kindly provided by Dr. D. DeSi-mone, University of Virginia, Charlottes-ville, VA; 1:100). After washing they wereincubated with the appropriate second-ary antibodies: anti-mouse CY3 conju-gated (Sigma C2181, 1:100), anti-rabbitfluorescein isothiocyanate (FITC) con-jugated (Jackson Immunoresearch 111-095-144, 1:40), anti-guinea pig Alexa 488conjugated (Invitrogen A11073, 1:500).Nuclei were stained with HoechstH33258 (Sigma, 1:1,000). Sections werewashed and mounted in Immunomount(Thermo electron corporation).

All imaging was done at room tem-perature. Light micrographs weretaken using a stereoscopic microscopeNikon SM2 1500 equipped with a digi-tal camera Nikon DMX1200 and theimage acquisition software LUCIA. Im-munofluorescent staining was imagedusing a Nikon Eclipse E800 microscopeequipped with a QEi Evolution camera(Media Cybernetics; ARC N° 7867). A�4 (Plan 0.1 NA), �10 (Plan-apochro-mat 0.45 NA), �20 (Plan-apochromat0.75 NA), or �40 (Plan-apochromat0.95 NA) were used and the image ac-quisition software was Image-Pro Plus.The figures were created using Photo-shop CS2 software (Adobe) and whenthe brightness and contrast of the wholeimage needed adjustment, the bright-ness/contrast adjustment function wasused.

TUNEL Staining andProliferation Assays

Embryos (n � 10) were fixed and sec-tioned. TUNEL staining was carriedout following the protocol as previ-ously described (Bello et al., 2008).Sections were mounted in Immuno-mount (Thermo electron corporation),and positive cells were counted insomite areas in both the control andthe injected sides.

Proliferation assays were performedin myotome sections of embryos (n � 8)stained with the rabbit anti-humanphospho-Histone H3 antibody (Ser 10,mitosis marker, Euromedex H5110-

14B, 1:500) and the anti-rabbit alkalinephosphatase-conjugated (Jackson Im-munoresearch 111-055-144, 1:5,000).Positive cells were counted in both theinjected and the uninjected sidesthrough the microscope at a magnifica-tion of 100 and in nonoverlapping fields.

ACKNOWLEDGMENTSWe thank Dr. D. Alfandari (fundingNIH DE016289, Department of Veteri-nary and Animal Sciences, Universityof Massachusetts, Lowell, MA) for pro-viding reagents (the anti-�1 integrinsubunit antibody) and helpful com-ments during this project and for criti-cal comments on the manuscript. Wealso thank M. Brockop and Dr. A.Gaultier (UCSD, Department of Pathol-ogy, La Jolla, CA) for reading and com-ments on the manuscript. We alsothank Dr. B. Della Gaspera (UniversiteParis 5, France) for providing theXMRF4 plasmid and antibody. Wethank Dr. D. DeSimone (University ofVirginia, Charlottesville, VA) for pro-viding the anti-fibronectin antibody. Wethank Dr. F. Broders (Institut Curie,CNRS-UMR 144, Paris, France) for theXMyoD probe. The Anti-MyoD antibodydeveloped by J. Gurdon and H. J.Standley, and the 12/101 antibody de-veloped by J. P. Brockes were obtainedfrom the Developmental Studies Hy-bridoma Bank developed under the aus-pices of the NICHD and maintained byThe University of Iowa, Department ofBiological Sciences, Iowa City, IA52242. M.H. received a predoctoral fel-lowship from P13 University.

REFERENCES

Afonin B, Ho M, Gustin JK, Meloty-Kapella C, Domingo CR. 2006. Cell be-haviors associated with somite segmen-tation and rotation in Xenopus laevis.Dev Dyn 235:3268–3279.

Bajanca F, Luz M, Raymond K, Martins G,Sonnenberg A, Tajbakhsh S, BuckinghamM, Thorsteinsdottir S. 2006. Integrinalpha6beta1-laminin interactions regulateearly myotome formation in the mouseembryo. Development 133:1635–1644.

Barresi R, Campbell KP. 2006. Dystrogly-can: from biosynthesis to pathogenesis ofhuman disease. J Cell Sci 119:199–207.

Batchelor CL, Higginson JR, Chen YJ,Vanni C, Eva A, Winder SJ. 2007. Re-cruitment of Dbl by ezrin and dystrogly-can drives membrane proximal Cdc42activation and filopodia formation. CellCycle 6:353–363.

1344 HIDALGO ET AL.

Bello V, Sirour C, Moreau N, Denker E,Darribere T. 2008. A function for dystro-glycan in pronephros development in Xe-nopus laevis. Dev Biol 317:106–120.

Colognato H, Galvin J, Wang Z, Relucio J,Nguyen T, Harrison D, Yurchenco PD,Ffrench-Constant C. 2007. Identificationof dystroglycan as a second laminin recep-tor in oligodendrocytes, with a role in my-elination. Development 134:1723–1736.

Danker K, Hacke H, Wedlich D. 1992. Effectsof heat shock on the pattern of fibronectinand laminin during somitogenesis in Xeno-pus laevis. Dev Dyn 193:136–144.

Deconinck N, Dan B. 2007. Pathophysiologyof Duchenne muscular dystrophy: currenthypotheses. Pediatr Neurol 36:1–7.

Della Gaspera B, Sequeira I, Charbonnier F,Becker C, Shi DL, Chanoine C. 2006. Spa-tio-temporal expression of MRF4 tran-scripts and protein during Xenopus laevisembryogenesis. Dev Dyn 235:524–529.

Deng WM, Schneider M, Frock R,Castillejo-Lopez C, Gaman EA, Baum-gartner S, Ruohola-Baker H. 2003. Dys-troglycan is required for polarizing theepithelial cells and the oocyte in Dro-sophila. Development 130:173–184.

Djiane A, Riou J, Umbhauer M, Boucaut J, ShiD. 2000. Role of frizzled 7 in the regulation ofconvergent extension movements duringgastrulation in Xenopus laevis. Develop-ment 127:3091–3100.

Drake CJ, Davis LA, Hungerford JE, LittleCD. 1992. Perturbation of beta 1 inte-grin-mediated adhesions results in al-tered somite cell shape and behavior.Dev Biol 149:327–338.

Durbeej M, Talts JF, Henry MD, YurchencoPD, Campbell KP, Ekblom P. 2001. Dys-troglycan binding to laminin alpha1LG4module influences epithelial morphogene-sis of salivary gland and lung in vitro. Dif-ferentiation 69:121–34.

Ervasti JM, Campbell KP. 1993. A role forthe dystrophin-glycoprotein complex as atransmembrane linker between lamininand actin. J Cell Biol 122:809–823.

Fey J, Hausen P. 1990. Appearance and distri-bution of laminin during development of Xe-nopus laevis. Differentiation 42:144–152.

George EL, Georges-Labouesse EN, Patel-King RS, Rayburn H, Hynes RO. 1993.Defects in mesoderm, neural tubeand vascular development in mouse em-bryos lacking fibronectin. Development119:1079–1091.

Giacomello E, Vallin J, Morali O, CoulterIS, Boulekbache H, Thiery JP, BrodersF. 2002. Type I cadherins are requiredfor differentiation and coordinated rota-tion in Xenopus laevis somitogenesis. IntJ Dev Biol 46:785–92.

Harland RM. 1991. In situ hybridization: animproved whole-mount method for Xenopusembryos. Methods Cell Biol 36:685–695.

Haines N, Seabrooke S, Stewart BA. 2007.Dystroglycan and protein O-Mannosyl-transferases 1 and 2 are required to

maintain integrity of Drosophila larvalmuscles. Mol Biol Cell 18:4721–4730.

Henry MD, Satz JS, Brakebusch C, Costell M,Gustafsson E, Fassler R, Campbell KP.2001. Distinct roles for dystroglycan, beta1integrin and perlecan in cell surface lamininorganization. J Cell Sci 114:1137–1144.

Higginson JR, Thompson O, Winder SJ.2008. Targeting of dystroglycan to thecleavage furrow and midbody in cytokine-sis. Int J Biochem Cell Biol. 40:892–900.

Hopwood ND, Pluck A, Gurdon JB, Dil-worth SM. 1992. Expression of XMyoDprotein in early Xenopus laevis embryos.Development 114:31–38.

Ibraghimov-Beskrovnaya O, Ervasti JM,Leveille CJ, Slaughter CA, Sernett SW,Campbell KP. 1992. Primary structure ofdystrophin-associated glycoproteins link-ing dystrophin to the extracellular matrix.Nature 355:696–702.

Keller R. 2000. The origin and morphogen-esis of amphibian somites. Curr Top DevBiol 47:183–246.

Kim SH, Jen WC, De Robertis EM, KinterC. 2000. The protocadherin PAPC estab-lishes segmental boundaries duringsomitogenesis in xenopus embryos. CurrBiol 10:821–830.

Kintner C, Brockes JP. 1984. Monoclonalantibodies identify blastemal cells de-rived from dedifferentiating limb regen-eration. Nature 308:67–69.

Koshida S, Kishimoto Y, Ustumi H,Shimizu T, Furutani-Seiki M, Kondoh H,Takada S. 2005. Integrinalpha5-depen-dent fibronectin accumulation for main-tenance of somite boundaries in ze-brafish embryos. Dev Cell 8:587–598.

Kragtorp KA, Miller JR. 2007. Integrin al-pha5 is required for somite rotation andboundary formation in Xenopus. DevDyn 236:2713–2720.

Marsden M, DeSimone DW. 2003. Inte-grin-ECM interactions regulate cad-herin-dependent cell adhesion and arerequired for convergent extension in Xe-nopus. Curr Biol 13:1182–1191.

Michele DE, Campbell KP. 2003. Dystro-phin-glycoprotein complex: post-transla-tional processing and dystroglycan func-tion. J Biol Chem 278:15457–15460.

Moreau N, Alfandari D, Gaultier A, CousinH, Darribere T. 2003. Cloning and ex-pression patterns of dystroglycan duringthe early development of Xenopus laevis.Dev Genes Evol 213:355–359.

Muschler J, Levy D, Boudreau R, Henry M,Campbell K, Bissell MJ. 2002. A role fordystroglycan in epithelial polarization:loss of function in breast tumor cells.Cancer Res 62:7102–7109.

Nieuwkoop PD, Faber J. 1967. Normal ta-ble of Xenopus laevis, New York: Gar-land Publishing.

Okabayashi K, Asashima M. 2003. Tissuegeneration from amphibian animal caps.Curr Opin Genet Dev 13:502–507.

Pollard SM, Parsons MJ, Kamei M, Kettle-borough RN, Thomas KA, Pham VN, Bae

MK, Scott A, Weinstein BM, StempleDL. 2006. Essential and overlappingroles for laminin alpha chains in noto-chord and blood vessel formation. DevBiol 289:64–76.

Rifes P, Carvalho L, Lopes C, Andrade RP,Rodrigues G, Palmeirim I, ThorsteinsdottirS. 2007. Redefining the role of ectoderm insomitogenesis: a player in the formation ofthe fibronectin matrix of presomitic meso-derm. Development 134:3155–3165.

Sgambato A, Brancaccio A. 2005. The dys-troglycan complex: from biology to can-cer. J Cell Physiol 205:163–169.

Spence HJ, Dhillon AS, James M, WinderSJ. 2004. Dystroglycan, a scaffold for theERK-MAP kinase cascade. EMBO Rep5:484–489.

Talts JF, Andac Z, Gohring W, Brancaccio A,Timpl R. 1999. Binding of the G domains oflaminin alpha1 and alpha2 chains and per-lecan to heparin, sulfatides, alpha-dystrogly-can and several extracellular matrix pro-teins. EMBO J 18:863–870.

Tiger CF, Gullberg D. 1997. Absence oflaminin alpha1 chain in the skeletalmuscle of dystrophic dy/dy mice. MuscleNerve 20:1515–24.

Towbin H, Staehelin T, Gordon J. 1979. Elec-trophoretic transfer of proteins from poly-acrylamide gels to nitrocellulose sheets:procedure and some applications. Proc.Natl. Acad. Sci. USA 76:4350–4354.

Wedlich D, Hacke H, Klein G. 1989. Thedistribution of fibronectin and laminin inthe somitogenesis of Xenopus laevis. Dif-ferentiation 40:77–83.

Williamson RA, Henry MD, Daniels KJ,Hrstka RF, Lee JC, Sunada Y, Ibraghi-mov-Beskrovnaya O, Campbell KP.1997. Dystroglycan is essential for earlyembryonic development: disruption ofReichert’s membrane in Dag1-null mice.Hum Mol Genet. 6:831–841.

Wilson PA, Oster G, Keller R. 1989. Cellrearrangement and segmentation in Xe-nopus: direct observation of cultured ex-plants. Development 105:155–166.

Winder SJ. 2001. The complexities of dystro-glycan. Trends Biochem Sci. 26:118–124.

Yang JT, Bader BL, Kreidberg JA, Ullman-Cullere M, Trevithick JE, Hynes RO.1999. Overlapping and independentfunctions of fibronectin receptor inte-grins in early mesodermal development.Dev Biol 215:264–277.

Yu H, Talts JF. 2003. Beta1 integrin and al-pha-dystroglycan binding sites are localizedto different laminin-G-domain-like (LG)modules within the laminin alpha5 chain Gdomain. Biochem J 371:289–299.

Yusuf F, Brand-Saberi B. 2006. The event-ful somite: patterning, fate determina-tion and cell division in the somite. AnatEmbryol (Berl) 211:21–30.

Zagris N, Christopoulos M, Giakoumaki A.2004. Developmentally regulated expres-sion and functional role of alpha 7 inte-grin in the chick embryo. Dev GrowthDiffer 46:299–307.

DYSTROGLYCAN FUNCTION DURING SOMITOGENESIS 1345