Neural Crest-Specific and General Expression of Distinct Metalloprotease–Disintegrins in EarlyXenopus laevisDevelopment

Neural Crest-Specific and General Expression ofDistinct Metalloprotease–Disintegrins in EarlyXenopus laevis Development

Hui Cai,* Jorn Kratzschmar,*,1 Dominique Alfandari,†Gary Hunnicutt,‡ and Carl P. Blobel*,2

*Cellular Biochemistry and Biophysics Program, Sloan-Kettering Institute, Memorial Sloan-Kettering Cancer Center, New York, New York 10021; †Laboratoire de Biologie Cellulaire etMoleculaire du Developpement, Universite Paris VI, CNRS, Paris UMR 7622, France; and‡Population Council, Center for Biomedical Research, 1230 York Avenue,New York, New York 10021

Metalloprotease–disintegrins are a family of membrane-anchored glycoproteins that have been implicated in diversecellular processes, including fertilization and myoblast fusion, release of TNFa from the plasma membrane, andneurogenesis. Here we report the cloning of cDNAs encoding three full-length (xMDC9, xMDC11b, and xMDC13), and onepartial (xMDC11a) metalloprotease–disintegrin from the amphibian Xenopus laevis, and the analysis of their expressionduring early X. laevis development and in adult tissues. The most notable finding was the highly localized and specificexpression pattern of xmdc11a at the tailbud stage in the cranial neural crest and in a subset of neural tube cells in the trunkregion. In contrast, expression of the closely related xmdc11b was not detectable during the early stages of X. laevisdevelopment, and remained low in the adult tissues examined here. Distinct expression patterns were also observed for twoother highly related X. laevis genes, xmdc13 and adam13 (Alfandari et al., 1997). While adam13 is expressed in the somiticmesoderm and in neural crest cells, but not in adult testis, xmdc13 expression is low and ubiquitous in the developingembryo, but is clearly present in adult testis. Finally, xmdc9, the putative orthologue of human and mouse mdc9, was foundat all stages of development, and in all tissues examined, suggesting a function that may be utilized by most or all cells. Thenoteworthy features of these four xmdc genes and the implications of their distinct spatial and temporal expression patternsare discussed. © 1998 Academic Press

INTRODUCTION

The metalloprotease–disintegrin protein family (also re-ferred to as MDC proteins, or ADAMs) is thought tofunction in cell–cell interactions and in the proteolysis ofluminal or extracellular protein domains (Blobel, 1997;Wolfsberg and White, 1996). In general, proteins with a rolein cell–cell interactions, and proteases that cleave extracel-lular or luminal proteins, are known to play important roles

in development and in the maintenance of an adult organ-ism. Cell adhesion proteins of the cadherin family areinvolved in morphogenesis and the establishment of epithe-lial cell polarity (Gumbiner, 1996), and integrins are essen-tial for distinct steps in development as well as in adulttissues (Alfandari et al., 1995; Brower et al., 1995; Fassler etal., 1996; Hynes, 1996; Lallier et al., 1996; Ramos andDeSimone, 1996; Ramos et al., 1996). Drosophila Tolloid isan example of a metalloprotease that has a role duringembryogenesis (Finelli et al., 1994; Shimell et al., 1991).Tolloid mediates dorsal/ventral pattern formation presum-ably by cleaving SOG (short gastrulation), which is aninhibitor of the ventralizing activity of DPP (decapentaple-gic) (Marques et al., 1997). The Xenopus laevis metallopro-tease Xolloid, which is related to Tolloid, has been shownto cleave Chordin, the functional ortholog of SOG, which

1 Present address: Institute of Cellular and Molecular Biology,Research Laboratories of Schering AG, D-13342 Berlin, Germany.

2 To whom correspondence should be addressed at CellularBiochemistry and Biophysics Program, Sloan–Kettering Institute,Memorial Sloan–Kettering Cancer Center, Box 368, 1275 YorkAve., New York, NY 10021. Fax: (212) 717-3047. E-mail: [email protected].

DEVELOPMENTAL BIOLOGY 204, 508–524 (1998)ARTICLE NO. DB989017

0012-1606/98 $25.00Copyright © 1998 by Academic Press

All rights of reproduction in any form reserved.508

leads to the release of active BMP from a Chordin/BMPcomplex (Piccolo et al., 1997). Because the released BMPinduces ventral structures, these results suggests that Xol-loid metalloprotease activity is also critical for the induc-tion of dorsal/ventral body axis.

Based on the known roles of cell adhesion molecules andproteases in development, it is tempting to speculate thatmembers of the metalloprotease–disintegrin protein familymight also be involved in specific aspects of development.Eleven of the currently known 21 MDC proteins have apredicted catalytic site sequence (HEXXH) in their metal-loprotease domain, and therefore are predicted to possesscatalytic activity. The presence of both a metalloproteasedomain and a putative cell adhesion domain suggests thatthe adhesive and proteolytic function might be somehowcombined, although alternatively these functions could beapplied and regulated independently from one another (Blo-bel, 1997). Metalloprotease–disintegrins that do not containa catalytic site in their metalloprotease domain are notthought to function as metalloproteases, but instead may beinvolved mainly in cell–cell or cell–matrix interactions.

To date, most of the insight into the function ofmetalloprotease–disintegrins has come from studies of ei-ther cell–cell interactions such as sperm–egg binding andfusion, and muscle cell fusion, or from studying the role ofthe metalloprotease domain. Fertilin, which is a het-erodimeric complex of two metalloprotease–disintegrins(Blobel et al., 1990, 1992; Primakoff et al., 1987; Wolfsberget al., 1993) and cyritestin/ADAM3 (Heinlein et al., 1994;Linder and Heinlein, 1997; Yuan et al., 1997), have impor-tant roles in fertilization, most likely by binding to anintegrin on the egg (Almeida et al., 1995; Evans et al.,1997a,b; 1995; Myles et al., 1994). Other cell–cell fusionevents include the fusion of myoblasts into myotubes andthe fusion of cells from the macrophage/monocyte lineageinto bone-resorbing osteoclasts (Huovila et al., 1996; Rood-man, 1996). The metalloprotease–disintegrin meltrin a hasbeen implicated in the process of cell–cell binding andfusion in the muscle cell line C2C12 (Yagami-Hiromasa etal., 1995), indicating that MDC proteins may also play arole during myogenesis in vivo. However, in bone cells,expression of meltrin a and the related meltrin b was foundin osteoblasts, but not in osteoclasts (Harris et al., 1997;Inoue et al., 1998) arguing against a direct role for these twoproteins in osteoclast fusion.

With respect to the role of the metalloprotease domainof metalloprotease– disintegrins during development, it isknown that Drosophila KUZ is essential both during veryearly and during later stages of development, includingneurogenesis and axon extension (Fambrough et al., 1996;Pan and Rubin, 1997; Rooke et al., 1996; Sotillos et al.,1997). The function of KUZ seems to be conserved indifferent species, since expression of a dominant negativemouse homologue of KUZ (MKUZ, also referred to asADAM10 or MADM (Howard et al., 1996)) in X. laevisembryos leads to an increase in the number of neuralcells. Further evidence for the evolutionary conservation

of KUZ function comes from studies with Caenorhabdi-tis elegans, where SUP-17, a KUZ homologue, has beenshown to function in LIN-12/Notch signaling (Wen et al.,1997). Finally it should be noted that two separatemetalloprotease– disintegrins, the TNFa convertase(TACE, or ADAM 17) (Black et al., 1997; Moss et al.,1997) and ADAM 10 (MADM) (Howard and Glynn, 1995;Howard et al., 1996), have been implicated in the cleav-age of the membrane-anchored cytokine TNFa (Lunn etal., 1997; Rosendahl et al., 1997). This finding raises thepossibility that metalloprotease– disintegrins may alsofunction in the cleavage of other proteins, includinggrowth factors, cytokines, cytokine receptors, and adhe-sion proteins that are known to be released from the cellsurface by membrane-anchored metalloproteases (Blobel,1997; Hooper et al., 1997).

Additional clues about functions of metalloprotease–disintegrins in development have emerged from the analy-sis of the expression pattern of X. laevis ADAM 13 and C.elegans adm-1 (ADAM 14). During early X. laevis develop-ment, ADAM13 is expressed in somitic mesoderm andneural crest cells, suggesting that ADAM13 might be in-volved in neural crest cell adhesion and migration as well asmyoblast differentiation (Alfandari et al., 1997). In C. el-egans, the adm-1 (ADAM 14) gene is expressed in syncytialorgans, sperm, and sheath cells of sensory organs in C.elegans embryos, consistent with a potential function incell adhesion and/or cell–cell fusion (Podbilewicz, 1996).

The main goal of this study was to identifymetalloprotease– disintegrins that are expressed duringearly development in X. laevis, and to delineate thedevelopmental stages, embryonic structures, and adulttissues in which these proteins may function. We reportthe cDNA cloning of four metalloprotease– disintegringenes, and the analysis of their expression patterns indeveloping and adult X.laevis. Putative mammalian or-thologues are known for two of these four genes, andtherefore the observed expression patterns in X. laevismay also have implications for the expression and func-tion of the corresponding mammalian proteins. Theresults are discussed in the context of what is currentlyknown about the function of metalloprotease– disintegrinproteins.

MATERIALS AND METHODS

Animals and reagents. Adult male or female X. laevis wereobtained from Nasco (Fort Atkinson, MI). The A6 X. laevis kidneycell line was kindly supplied by Dr. B. Gumbiner (MemorialSloan–Kettering Cancer Center). All reagents were purchased fromBoehringer Mannheim (Indianapolis, IN), unless indicated other-wise. Radiolabeled nucleotides were supplied by NEN (Boston,MA).

cDNA cloning of xmdc9, xmdc11a, xmdc11b, and xmdc13.PCR generated cDNA tags of xmdc9, xmdc11a, xmdc11b, andxmdc13 (Shilling et al., 1997) were separately labeled with[a-32P]dCTP and used to probe a X. laevis testis cDNA library

509Metalloprotease–Disintegrins in Early Xenopus Development

Copyright © 1998 by Academic Press. All rights of reproduction in any form reserved.

constructed in the Lambda ZAPII bacteriophage vector (Stratagene,La Jolla, CA) (Shilling et al., 1997). These screens yielded 5 cloneswith xmdc9 cDNA, 2 with xmdc11a cDNA, 14 with xmdc11bcDNA, and 6 with xmdc13 cDNA. All clones were subjected to invivo excision following the manufacturer’s protocol. cDNA cloneswith the longest 59 sequence were identified by PCR using a vectorspecific primer (T3) and an antisense PCR primer designed tohybridize with the disintegrin domain cDNA sequence tag of eachgene. The results were confirmed by restriction analysis of thedifferent clones. Clones with the longest cDNA insert were se-quenced on both strands using a primer walk approach (Sequenase,USB, Cleveland, OH). The xmdc9, xmdc11b, and xmdc13 clonescontained apparently full-length open reading frames (xmdc9cDNA insert, 2993 bp; xmdc11b cDNA insert, 3447 bp; xmdc13cDNA insert, 3083 bp), whereas the xmdc11a clone only containeda partial cDNA insert (1588 bp). Sequence analysis was carried outusing the MacVector program (Kodak IBI), and the alignment ofeach of the protein sequences with the most related currentlyknown MDC proteins was generated with the Megalign module ofthe DNAstar (Madison, WI) software program.

Northern blot analysis. Total RNA was isolated from unfertil-ized eggs, from developing embryos (10 embryos per stage ana-lyzed), from adult male X. laevis testes, heart, muscle, and liver,and from the A6 X. laevis kidney cells as described (Chomczynskiand Sacchi, 1987). Fifteen micrograms of total RNA was loaded perlane for each tissue. The blots were probed sequentially withrandom primed [a-32P]dCTP-labeled cDNA fragments of the fourxmdc cDNAs (xmdc9: 1.8-kb NcoI fragment between nucleotides179 and 2000; xmdc11a: 3-kb ClaI/XhoI fragment between nucle-otide 933 and the vector multiple cloning site; xmdc11b: 900-bpEcoRV fragment between nucleotides 1677 and 2573; xmdc13:0.7-kb XhoI fragment between nucleotide 2348 and the vectormultiple cloning site) under high stringency conditions. It shouldbe noted that the xmdc13 probes for Northern blot analysis and insitu RNA hybridization were derived from noncoding 39 sequencesthat are not present in the adam13 cDNA. As a control, theNorthern blot was stripped and reprobed with an [a-32P]dCTP-labeled X. laevis fibronectin cDNA probe (corresponding to a 1-kbEcoRI/SalI fragment, see DeSimone et al., 1992).

In situ RNA hybridization. Whole-mount in situ hybridiza-tions were performed on albino X. laevis embryos according to themethod of Harland (1991). Digoxigenin-rUTP-labeled antisenseprobes for xmdc9, xmdc11a, xmdc11b, and xmdc13 were synthe-sized using T7 RNA polymerase from identical cDNA fragments asdescribed above for Northern blot analysis. As negative controls,sense probes were synthesized using T3 RNA polymerase. Hybrid-ized transcripts were visualized in situ with anti-digoxigeninantibodies and BM purple and photographed using a Leica dissect-ing microscope with an attached Nikon camera. To evaluateexpression of xmdc11a in a section of the truncal crest, an embryostained as described above was dehydrated in MeOH and embeddedin polyethylene glycol 400-distearate (Aldrich). After cutting 10mM serial sections, the embedding medium was dissolved withmethanol. Sections were washed once in xylene and mounted inBaume du Canada.

Generation of antibodies against xMDC9. A construct forexpression of a GST fusion protein with the cytoplasmic tail ofxMDC9 was generated by ligating a PCR synthesized cDNAfragment encoding for the cytoplasmic amino acid residues 716 to803 of xmdc9 into the pGEX 4T3 vector (Pharmacia LKB, Upsala,Sweden) via restriction sites added to the PCR primers. GST–xMDC9–cytotail fusion proteins were purified from bacterial BL21

cells using glutathione beads (Pharmacia LKB) following protocolssupplied by the manufacturer, and subsequently used for antibodyproduction following standard protocols (Harlow and Lane, 1988).Protein A–Sepharose was used to purifiy IgG from the polyclonalrabbit antiserum raised against the GST–xMDC9–cytotail fusionprotein. The resulting IgG were depleted of antibodies reactivewith GST alone by incubation with GST coupled to CNBr-activated CL4B beads (Pharmacia LKB) following the manufactur-er’s instructions. The IgG thus obtained reacted with xMDC9 andare referred to as xMDC9 IgG. The xMDC9 IgG were then incu-bated with the GST–xMDC9–cytotail fusion protein coupled toCNBr-activated CL4B beads to deplete antibodies reacting with thexMDC9 cytoplasmic tail. These antibodies are referred to ascontrol IgG. Affinity-purified xMDC9 antibodies were obtained byeluting the xMDC9-specific antibodies bound to the GST–xMDC9–cyto fusion protein with 0.1 M glycine, pH 3, followed bydialysis with PBS. The successful depletion of antibodies againstthe GST–xMDC9–cytotail, or of GST-reactive antibodies, or theaffinity purification of xMDC9-specific antibodies was confirmedby probing Western blots of the GST–xMDC9–cytotail fusionprotein and of GST alone (not shown).

Western blot analysis. Different tissues were removed frommale X. laevis and homogenized with a Polytron homogenizer(Kinematica, Littau, Switzerland) in cell lysis buffer (1% NP-40 inPBS, supplemented with a protease inhibitor cocktail (Blobel et al.,1990)). Ten milliliters of cell lysis buffer was used per gram oftissue. X. laevis A6 cell lysates were prepared by lysing a confluent15-cm culture dish in 1.5 ml of cell lysis buffer. All lysates werecentrifuged at 27,000g for 30 min. The supernatants were incubatedovernight at 4°C with a 500-ml bed volume of Con A–Sepharose(Pharmacia LKB) per 10 ml lysate. Con A-bound glycoproteins wereeluted in 500 ml sample loading buffer per 500 ml Con A by heatingat 95°C for 5 min in the presence or absence of 5 mM DTT, asindicated. Twenty microliters of eluted glycoproteins was run perlane and processed for Western blot analysis as described(Weskamp et al., 1996). For Western blot analysis of differentdevelopmental stages, 50 embryos from each stage were collectedand lysed in 1% NP-40, followed by Con A enrichment forglycoproteins as outlined above.

Expression of xMDC9 in COS-7 cells. The full length openreading frame of xmdc9 was subcloned into the pcDNA3 expres-sion vector (Invitrogen, San Diego, CA) using a NotI/ApaI frag-ment, and 2 mg of the resulting construct was transfected intoCOS-7 cells using Lipofectamine (Gibco/BRL) as a delivery vehicle.Cells were lysed 2 days after transient transfection in cell lysisbuffer and processed as described above for Western blot analysis.

Cell surface biotinylation and immunoprecipitation. A 90%confluent A6 cell culture was incubated in 1 mg/ml NHS–LC–biotin (Pierce, Rockford, IL) in PBS for 1 h at 4°C. Labeling wasquenched by washing cells three times in PBS buffer containing 50mM glycine. After being labeled and quenched, the cells were lysedin cell lysis buffer, and after removal of nuclei by centrifugation,the extract was subjected to immunoprecipitation with the anti-xMDC9 IgG or with control IgG. After addition of sample loadingbuffer, the immunoprecipitated proteins were heated to 95°C withor without 10 mM DTT, separated on 10% SDS–PAGE, andtransferred to nitrocellulose. Biotinylated proteins were detectedby probing the blot with horseradish peroxidase-labeled streptavi-din (Pierce, Rockford, IL) followed by treatment with a chemilu-minescent detection kit (Amersham, Arlington Heights, IL) andexposure to Kodak XAR autoradiography film.

510 Cai et al.


RESULTS

Cloning and cDNA Sequencing of xmdc9,xmdc11a, xmdc11b, and xmdc13

In a previous study, short cDNA sequences encodingpartial disintegrin domains of xmdc9, xmdc11a, xmdc11b,and xmdc13 were generated by PCR from X. laevis testiscDNA (Shilling et al., 1997). For the present study, each ofthese short cDNA sequences was used separately to probe acDNA library constructed from X. laevis testis mRNA toisolate corresponding cDNA clones. Of the positive clonesthus identified, the clone with the longest cDNA insert ineach case was detected by restriction digest and PCRanalysis, and subsequently sequenced completely on bothstrands (see Materials and Methods for details). The proteinsequences deduced from the individual cDNAs are shownaligned with the most closely related metalloprotease–disintegrin proteins that are presently known in Figs 1A–1C. xMDC9 was aligned with human and mouse MDC9(Fig. 1A) (Weskamp et al., 1996), xMDC11a and 11b werealigned with human MDC/ADAM11 (Fig. 1B) (Emi et al.,1993; Katagiri et al., 1995), and xMDC13 was aligned withADAM13 (Fig. 1C) (Alfandari et al., 1997). The deducedprotein sequences for xMDC9, 11b, and 13 contain a pre-dicted signal sequence, a prodomain, metalloprotease do-main, disintegrin domain, cysteine-rich domain, EGF re-peat, transmembrane domain, and a cytoplasmic tail. ForxMDC11a, only a partial cDNA sequence is available, withthe deduced protein sequence beginning in the middle ofthe metalloprotease domain. xMDC9 and xMDC13 have acatalytic site consensus sequence and are thus predicted tobe catalytically active, whereas xMDC11a and xMDC11bdo not contain this sequence, and therefore are not pre-dicted to have metalloprotease activity. Table 1 presents asummary of the features that can be deduced from thetranslated protein sequences of these four xMDC pro-teins, including predicted MW, and percentage sequencesimilarity with the most related presently knownmetalloprotease–disintegrin proteins.

Northern Blot Analysis

Northern blot analysis under high stringency conditionswas used to establish the expression pattern of xmdc9,

xmdc11a, xmdc11b, and xmdc13 in unfertilized eggs, ingastrula (stage 13), neurula (stages 18), tailbud (stage 23),and tadpole stage embryos (stage 40, Fig. 2A), and in adulttestis, heart, muscle, and liver, and in the kidney epithelialA6 cell line (Fig. 2B).

xmdc9 was found to be expressed in all developmentalstages and adult tissues examined, and in the X. laevis A6kidney cell line. The highest expression levels for xmdc9were found in testis, heart, and in A6 cells, whereasexpression levels were lower in muscle and liver (Fig. 2B). Inall samples the most abundant transcript was 6 kb long,although a minor transcript of 9 kb could also be detected.Testis was the only tissue containing several closely comi-grating transcripts around 3 kb. Northern blot analysis ofdifferent developmental stages revealed that transcripts ofxmdc9 are stored maternally and that the gene is expressedthroughout development, including the swimming tadpolestage (Fig. 2A). In the Northern blot of developmental stageembryos, both the more abundant 6.5-kb band and theweaker 9-kb band were detected, which correspond to themost abundant transcript in the adult tissue Northern blot.The different xmdc9 transcript sizes could derive from theuse of alternative polyadenylation signals, or from alterna-tive splice products of a single gene, or may representclosely related mdc genes that cross-hybridize with thexmdc9 cDNA probe.

When Northern blots of identical samples were probedwith labeled xmdc11a cDNA, expression of a 6-kb mRNAwas detected beginning with neurula stage (stage 18) em-bryos, and was clearly visible in tailbud embryos (stage 30)and tadpoles (stage 40) (Fig. 2A). In adult tissues, expressionof xmdc11a can be detected in testis, but is barely visible inheart and muscle, and not detectable in liver and A6 cells(Fig. 2B). In contrast to xmdc11a, there was no apparentexpression of the related xmdc11b in developing embryos atdifferent stages (data not shown). In adult testis, an 8-kbband and one or more bands between 3 and 4.5 kb arevisible. The 8-kb band can also be seen in heart and muscle,but not in liver or A6 cells. For xmdc13, expression of a 5-kbmRNA band in embryos is detectable starting at stage 13,and can be seen up until stage 40 (Fig. 2A). In adult tissuesthe strongest expression appears to be in testis, followed byheart and muscle, but no expression could be seen in liveror A6 cells. Testis and heart also contained a weaker band of

FIG. 1. Alignment of the deduced amino acid sequences of Xenopus laevis metalloprotease–disintegrins with the most highly relatedmammalian protein sequences. xMDC9 is aligned with human and mouse MDC9 in A, xMDC11a and xMDC11b are aligned with humanMDC/ADAM11 in B, and xMDC13 is aligned with ADAM13 in C. Residues present in at least two of the aligned sequences are boxed.Predicted signal sequence cleavage sites (von Heijne, 1986) are marked with an arrowhead. Predicted domain boundaries are marked by avertical bar with an arrowhead, and the metalloprotease catalytic site consensus sequence HEXXH is surrounded by a hatched box in A andC. The putative integrin binding sequence which is found in lieu of the RGD sequence in snake venom disintegrins is underlined (Wolfsbergand White, 1996). Examples of cytoplasmic proline-rich regions which may function as SH3 ligand domains (Alexandropoulos et al., 1995)are boxed, although it should be noted that other potential signaling motifs and sequences predicting cytoplasmic protein–proteininteractions can be found. The cDNA and protein sequences have been deposited with GenBank under the following Accession Ns.:xMDC9, AF032382; xMDC11a, AF032384; xMDC11b, AF032383; xMDC13, AF032385.



FIGURE 1

512 Cai et al.

FIG. 1—Continued



FIG. 1—Continued

514 Cai et al.

approximately 3 kb, which is not detectable in muscle andliver. As a control for equal loading, both Northern blotswere probed with a fibronectin probe. It should be notedthat expression of the fibronectin gene cannot be detectedin A6 kidney epithelial cells, but since xmdc9 and xmdc13are expressed in these cells, these two probes served toverify the presence of intact mRNA in the A6 cell sample.

In Situ mRNA Hybridization with xmdc9,xmdc11a, xmdc11b, and xmdc13 Probes

An important goal of this study was to elucidate theexpression pattern of xmdc9, xmdc11a, xmdc11b, andxmdc13 mRNA in early X. laevis embryos. To this end, thelocalization of these transcripts was examined by whole-mount in situ mRNA hybridization in gastrula, neurula andtailbud stage embryos. At the gastrula stage (stage 13),xmdc9 mRNA is localized uniformly in the whole embryo(Fig. 3C). In later stages, xmdc9 transcripts continue to bewidely expressed, although increased levels of expressionare visible in the somites, notochord, and the head region ofthe tailbud stage embryos (Figs. 3D–3F). In contrast,xmdc11a expression could not be detected in embryos untilneurulation (stage 18/20). At this stage (Fig. 3I) staining ispresent in a symmetrical pattern on either side of the neuraltube. In the anterior part of the embryo, the xmdc11amRNA is present in three groups of cells (Fig. 3G). Thislocalization is similar to the expression of xap2 (Schuh etal., 1993), xslug (Mancilla and Mayor, 1996), and adam13(Alfandari et al., 1997), all of which are expressed in thecranial neural crest. In the trunk, xmdc11a differs from

xslug and adam13 in that it extends into more dorsal partsof the embryo. On a section of the trunk of an early tailbudstage embryo (Fig. 3K, stage 22) the staining appears local-ized to a subset of neuronal cells within the neural tube.This mRNA staining pattern is very similar, but slightlymore dorsal compared to that described for the a6 integrin(Lallier et al., 1996). During later development of thetailbud, expression of the xmdc11a mRNA is found in thespaces surrounding the brain and optic vesicle as well as inthe mandibular, hyoid and branchial arches (Fig. 3L). Theseareas are known to be colonized by neural crest cells.xmdc11a mRNA is low or absent in the optic vesicle, aswell as between the epidermis (Figs. 3H, 3J, and 3L). At thislater stage, xmdc11a expression in the trunk is still re-stricted to two rows of cells in the dorsal part of the neuraltube. The appearance of the strong neural staining patternat stage 30 correlates with the presence of the xmdc11aband on stage-specific Northern blots (Fig. 2A). Expressionof the closely related xmdc11b could not be detected by insitu hybridization or by Northern blot analysis in any of theembryonic stages examined here (data not shown). In con-trast to adam13, which is expressed in somitic mesodermand neural crest cells (Alfandari et al., 1997), expression ofthe highly related xmdc13 was relatively weak and ubiqui-tous (data not shown).

Several control experiments were undertaken to confirmthe specificity of the xmdc9, xmdc11a, and xmdc13 anti-sense probes. These included sense transcripts as negativecontrol probes and adam13 antisense transcripts as a posi-tive control probe. Staining with the control sense tran-script of xmdc9 is shown in Figs. 3A and 3B.

TABLE 1Predicted Features of xMDC9, xMDC11a, xMDC11b, and xMDC13

Xenopus laevis

MDCs

xMDC9

xMDC11axMDC11b 938 104.135 µ (QSLAH) VNDC 1 hMDC11, 51.3%

ADAM 13, 90%4AGSC+ (HEIGH)99.084910xMDC13

n/a n/a µ (QTLGQ) LNEC 0 hMDC11, 75%

873 95.255 + (HELGH) ANEC 3 hMDC9, 64.1%; mMDC9, 59.6%

Amino acid

residues

Predicted

MW

Metalloprotease

catalytic site

Predicted integrin binding sequence

Potential SH-3 binding

sequences

% Sequence similarity to:

Signal sequence

Pro- domain

Metalloprotease domain

Disintegrin domain

Cysteine-rich region

EGF repeat

TM

Cytoplasmic domain

HEXXH*

��

��

Note. The domain organization of a typical metalloprotease–disintegrin is shown above. A catalytic site consensus sequence (HEXXH)is present in xMDC9 and xMDC13, but not in xMDC11a and xMDC11b. The position of the predicted integrin binding sequence in thedisintegrin domain is marked by an asterisk. The cytoplasmic domains of xMDC9, xMDC11b, and xMDC13 contain proline-rich sequencesthat may function as potential SH-3 ligand domains.



Analysis of xMDC9 Protein Expression andProcessing

Polyclonal antibodies were raised against GST–fusionproteins with the cytoplasmic tail of xMDC9, xMDC11a,xMDC11b to analyze the expression and processing of theseMDC proteins in X. laevis tissues and in A6 cells. Prelimi-nary screening for reactivity of the different antisera withthe corresponding protein by Western blot analysis wasperformed using Con A-purified glycoproteins from adult X.laevis heart. Although all antisera reacted well with the

fusion proteins that had been used as antigens, the onlyMDC protein that could reliably be detected on a tissueWestern blot was xMDC9. The failure to detect xMDC11aand 11b may be due to relatively low expression of theseproteins and/or an insufficient titer of the antiserum.

To further characterize the xMDC9 antiserum, threedifferent antibody samples were generated (see Materialsand Methods for details). Protein A-purified IgG were firstdepleted of antibodies reacting with GST alone, and arereferred to as xMDC9 IgG. After depletion of GST-

FIG. 2. Northern blot analysis of xMDC9, 11a, 11b, and 13 expression during X. laevis development and in adult tissues and A6 cells. (A)A Northern blot containing RNA extracted from 10 embryos at different stages of development (from the unfertilized egg through theswimming tadpole stage), was probed with [32P]dCTP-labeled xmdc9, xmdc11a, and xmdc13 cDNA, as indicated. (B) Northern blotscontaining 15 mg of total RNA per lane from adult X. laevis testis (lane 1), heart (lane 2), muscle (lane 3), liver (liver 4), and A6 cells (lane5) were probed sequentially with [32P]dCTP-labeled cDNA fragments of xmdc9, xmdc11a, xmdc11b, and xmdc13, and with X. laevisfibronectin cDNA as control.

516 Cai et al.


reactive antibodies, the protein A-purified IgG was fur-ther depleted of all antibodies reacting with GST–xMDC9 – cytotail fusion protein, and this sample isreferred to as control IgG. Finally, antibodies binding tothe GST–xMDC9 – cytotail were eluted with 0.1 M gly-cine, pH 3.0, and these antibodies are referred to asaffinity-purified xMDC9 IgG. Blots of Con A-purifiedglycoproteins from a X. laevis heart lysate were probedwith xMDC9 IgG (Fig. 4, lanes 1 and 2), with affinity-purified xMDC9 IgG (lanes 3 and 4), with control IgG(lanes 5 and 6), or with the secondary antibody alone (lane7). Both the xMDC9 IgG and affinity-purified xMDC9IgG, but not the control IgG or the secondary antibodyalone, recognized a band of 75 kDa under nonreducingconditions, and a band of 95 kDa in samples separatedunder reducing conditions. A significant increase in theapparent MW after reduction is consistent with therelatively high cysteine content of the extracellular se-quence of xMDC9, and is a general feature ofmetalloprotease– disintegrin proteins.

To further confirm the specificity of the xMDC9 antibod-ies, Western blots of nonreduced extracts of COS-7 cellstransfected with xMDC9, or with the expression vectoralone, were probed with either anti-xMDC9 IgG or controlIgG. In the extract of xMDC9-transfected COS-7 cells, theanti-xMDC9 IgG recognized a 75-kDa band (Fig. 4B, lane 1),which was not present in the vector transfected cells (lane2), and was also not recognized by the control IgG, (Fig. 4B,lane 3). In Western blots of different X. laevis tissues and A6cells, the xMDC9 protein was visible as a 75-kDa proteinunder nonreducing condition (Fig. 4C). The expression levelcorrelated well with the mRNA expression level seen in theNorthern blot analysis, with the highest expression intestis, heart, and A6 cells, and relatively low expression inmuscle and liver. Immunoblot analysis of various embry-onic stages under nonreducing conditions showed that thexMDC9 protein is expressed as a 75-kDa glycoprotein at allstages examined including unfertilized eggs (Fig. 4E), inagreement with the temporal expression pattern of xMDC9RNA.

To determine whether xMDC9 is present on the surfaceof A6 cells, these cells were cell-surface biotinylated with anon-membrane-permeable biotinylation reagent, andxMDC9 was subsequently immunoprecipitated usingxMDC9 IgG. Figure 4D (lane 1) shows that the biotinylatedmaterial which can be immunoprecipitated with thexMDC9 IgG consists mainly of a 95-kDa and a weaker65-kDa band under reducing conditions, and of a 75- and a48-kDa band under nonreducing conditions. These bandsappear to be specific for xMDC9 since they are not immu-noprecipitated by the control antibodies. This result indi-cates that xMDC9 is present on the cell surface of A6 cells,and is consistent with the observation that mouse MDC9can be cell-surface biotinylated in NIH 3T3 fibroblasts(Weskamp et al., 1996).

DISCUSSION

The main goal of this study was to gain a better under-standing of the potential functions of four metalloprotease–disintegrins (xmdc9, xmdc11a, xmdc11b, and xmdc13)through an analysis of their expression patterns in develop-ing and adult X. laevis. Our results indicate that expressionof xmdc11a is restricted to neural crest derivatives such ascranial and truncal neural crest cells during early X. laevisdevelopment, suggesting a role for xmdc11a in the migra-tion or differentiation of cells or tissues derived from theneural crest. xmdc9 is expressed at all stages of develop-ment and in all adult tissues analyzed. These results indi-cate that xmdc9 may have a more general function that isutilized by most or all cells. xmdc11b expression was notobserved in developing embryos and was relatively weak inthe adult tissues examined, whereas xmdc13 expressionwas low yet ubiquitous in early embryos and thus differedfrom the expression pattern of the highly related adam13(Alfandari et al., 1997) in somitic mesoderm and in cranialneural crest cells.

The first step of this study was to attempt to isolate andsequence full-length cDNAs for four metalloprotease–disintegrin PCR sequence tags which had previously beenidentified in X. laevis testis (Shilling et al., 1997). Theprotein sequences deduced from the longer cDNA se-quences presented here essentially confirm the initial com-parison between the PCR sequence tags and other knownmetalloprotease–disintegrins (Shilling et al., 1997). Thepresent and previous analyses both suggest that two of thecDNA fragments (xMDC9, 11a) have a known putativemammalian orthologue, that a third one is closely related tothe human MDC11 protein (xMDC11b), and that one ishighly related to X. laevis ADAM13 (Alfandari et al., 1997).Following is a discussion of the noteworthy features andexpression pattern of each of these four metalloprotease-disintegrins.

xmdc9

Like human and mouse MDC9, the putative X. laevisxMDC9 contains a metalloprotease domain with a catalyticsite consensus sequence HEXXH, and cytoplasmic signal-ing motifs, including proline-rich putative SH3–ligand do-mains. In the case of mouse MDC9, these proline-richsequences have been shown to bind the SH3 domain of srcin a blot overlay assay, suggesting that in principle thesedomains may be able to bind to SH3–ligand domains(Weskamp et al., 1996). The strong conservation of theseproline-rich sequences between mouse, human, and X.laevis MDC9 relative to less conserved adjacent cytoplas-mic sequences lends further support to the idea that theproline-rich sequences might be important for the functionof MDC9.

Northern blot analysis revealed expression of xmdc9 atall stages examined, including the unfertilized egg. In situhybridization further established that xmdc9 mRNA is


518 Cai et al.

uniformly expressed in the whole embryo at the gastrulaand neurula stages. In later tailbud stage embryos, xmdc9mRNA is detectable throughout the embryo, although anelevated level of expression is seen in the somite, head, andnotochord region. The pattern of xmdc9 expression in X.laevis embryos is similar to that reported for themetalloprotease–disintegrin kuz, a maternal gene which isfirst expressed widely in gastrula embryos, whereas at laterstages an increased level of expression is detected in neuraltissues (Pan and Rubin, 1997). It should be noted thatalthough kuz expression is not restricted to developingneural tissues, it nevertheless has a specific role in neuro-genesis. The expression pattern of xmdc9 in all stages ofdevelopment, including the unfertilized egg, suggests thatxMDC9 plays a general role that is utitilized or required byall cells and tissues, although such a general expressiondoes not rule out specific roles in different tissues ordevelopmental stages. Since mouse MDC9 has been shownto have catalytic activity (Roghani et al., manuscript sub-mitted), it is possible that xMDC9 may have a role inprotein ectodomain processing, in cell–cell interactions, orboth (Blobel, 1997; Weskamp et al., 1996).

Antibodies raised against xMDC9 were used to extendthe analysis of xMDC9 expression to the protein level andessentially confirmed the results of the Northern blotanalysis. Expression of xMDC9 in the unfertilized eggsuggests a maternal mRNA contribution, which must bereplenished by endogenous expression at later stages ofdevelopment. In a previous study, a peptide correspondingto the predicted integrin binding sequence of xMDC9,which differs from the sequence of mouse and humanMDC9, has been shown to block X. laevis fertilization(Shilling et al., 1997). Since xMDC9 is apparently presenton both sperm and egg, this finding raises that possibilitythat a metalloprotease–disintegrin on the egg might also beinvolved in binding an integrin-type receptor on the sperm,

in addition to the predicted role of sperm metalloprotease–disintegrins in binding integrins on the egg. In Northernblots, different xmdc9 mRNA transcripts are found insomitic tissues and in testis, indicating that xmdc9 mRNAmight be alternatively spliced or have divergent 39 untrans-lated regions. Yet in all tissues and developmental stagesexamined here, there was no evidence for different polypep-tide species. The observed MW of xMDC9 is similar to thatof mouse and human MDC9, and suggests that the pre-dominant form of xMDC9 contains a membrane anchoredmetalloprotease and disintegrin domain, but lacks a pro-domain (Weskamp et al., 1996). In the case of mouseMDC9, furin or a related pro-protein convertase appears tobe responsible for removal of the pro-domain in the trans-Golgi network (Roghani et al., manuscript submitted).

xmdc11a and xmdc11b

The high degree of sequence identity between xMDC11aand human MDC11 (Emi et al., 1993; Katagiri et al., 1995),including short and nearly identical cytoplasmic sequences,suggests that these two proteins are orthologues. xMDC11bis clearly more related to xMDC11a and human MDC11than to other presently known MDC proteins, but has alonger cytoplasmic domain with at least one potential SH3ligand domain. While the xMDC11a cDNA clone isolatedin this study does not contain a full-length open readingframe (it lacks the signal sequence, pro-domain, and parts ofthe metalloprotease domain), the deduced xMDC11b se-quence appears to be full-length because a hydrophobicsignal sequence and a pro-domain are present in the openreading frame. The alignment shown in Fig. 1B furthersuggests that the full-length human MDC/ADAM11 se-quence has not yet been determined, since the predictedinitial methionine in the reported human protein sequence(Emi et al., 1993; Katagiri et al., 1995) is not followed by a

FIG. 3. Whole mount in situ mRNA hybridization of embryos at different developmental stages with xmdc9 and xmdc11a cRNA, andhybridization of a tailbud stage embryo section with xmdc11a cRNA. Albino X. laevis embryos at various developmental stages werehybridized with an xmdc9 sense probe (A, B), an xmdc9 antisense probe (C–F), or an xmdc11a antisense probe (G–L). All embryos areoriented with their anterior toward the left of the figure. A shows a lateral view of a late neurula embryo (stage 20), and B shows a tailbudembryo (stage 30). Both embryos were hybridized with the sense xmdc9 control probe. C shows that xmdc9 expression can be detected inthe entire gastrula stage embryo. In a neurula embryo (E, stage 20) xmdc9 mRNA is present along the entire anteroposterior axis with amore pronounced staining on the dorsal side. At stage 25 (D), an essentially similar localization is observed in the head, the dorsal trunkstructures and the tail bud. In addition, a more pronounced staining of the somites is visible. In a dorsal view of a tailbud embryo (F, stage25), expression of xmdc9 appears strongest in dorsal structures. G and I show a lateral and dorsal view, respectively, of a late neurula embryo(stage 20) hybridized with an xmdc11a antisense probe. The most anterior signal is divided into three populations of cells (arrows) extendingfrom the neural tube. In the trunk, a thin line of cells on either side of the neural tube express xmdc11a (black arrowheads). K shows atransverse section through the trunk of a tailbud embryo (stage 22). Two rows of xmdc11a-expressing cells (small black arrowhead) extendalong the sides of the neural tube (nt). The somites (s), gut (g), and notochord structures (nc) are also indicated. In a lateral view of a tailbudstage embryo (H, stage 25), xmdc11a staining is present in four groups of cells (arrows) within the head and branchial arches. J demonstratesthat expression in the same anterior structures as in H can be seen in a dorsal view of a tailbud embryo. No xmdc11a expression is detectedin the optic vesicle (J, red arrowhead). In the trunk, the staining clearly appears along two stripes on either side of the midline. L depictsa magnification of the anterior region of a tailbud stage embryo. xmdc11a mRNA is localized to the four segments derived from the cranialneural crest. From posterior to anterior, these structures correspond to the posterior branchial crest, the anterior branchial crest, the hyoidcrest, and the mandibular crest. The optic vesicle is marked by a red arrowhead, and the brain is indicated (b).



hydrophobic signal sequence, and since parts of the pro-domain are missing compared to xMDC11b (see Fig. 1B), orother known MDC proteins (data not shown).

By Northern blot analysis, xmdc11a expression couldfirst be detected in stage 18 embryos. It is also present inadult testis, heart, and muscle, but not in liver and A6 cells.In contrast, xmdc11b expression was not observed duringearly development, in A6 cells or adult liver, and washighest in adult testis, followed by heart and muscle. By insitu hybridization, xmdc11a RNA is first detected duringneurulation in cells derived from the neuroectoderm. In thehead of the tailbud stage embryo, xmdc11a is only detectedin the cranial neural crest. The cranial neural crest struc-tures are thought to emerge from the anterior neural tubeand migrate ventrally to fill spaces between the neural tubeand laterally under the epidermis (Bronner-Fraser, 1993,1994; Chang and Hemmati-Brivanlou, 1998; Krotoski et al.,1988; Mayor et al., 1995; Sadaghiani and Thiebaud, 1987).These cells give rise to facial structures including cartilage,facial muscles, the sclera of the eye, and the sensoryganglia. Interestingly, the cranial neural crest staining pat-tern of xmdc11a resembles that of adam13 (Alfandariet al., 1997). This raises the possibility that the twometalloprotease–disintegrins xMDC11a and ADAM13 mayinteract, as has been described for the a and b subunit of theheterodimeric sperm protein fertilin (Blobel et al., 1990,1992). In the trunk, transversal sections show that xmdc11ais expressed in the dorsal half of the neural tube on eitherside of the midline. While we are not aware of a detaileddescription of the cell types in this region in X. laevis, itseems unlikely that these cells represent neural crest cellprecursors because at this stage the truncal neural crestcells have already emerged and begun migrating ventrally.

Since xMDC11a does not have a catalytic site in itsmetalloprotease domain, any role in the neural crest shouldbe quite different from that of catalytically active proteinssuch as KUZ. xMDC11a may therefore function in cell–cellinteractions, perhaps as an integrin ligand, or it couldengage other proteins or receptors on the cell surface or

extracellular matrix. It is thus conceivable that xMDC11ahas a role in certain aspects of neural crest cell migration,which is thought to require dynamic changes in cell–celland cell–matrix interactions (Bronner-Fraser, 1993, 1994;Erickson and Perris, 1993; Mayor et al., 1995; Monier-Gavelle and Duband, 1997).

xmdc13

The fourth cDNA presented here is highly related to X.laevis ADAM13. Because X. laevis is tetraploid, it is notunusual to isolate highly related, but distinct cDNA clonesfrom X. laevis cDNA libraries (DeSimone and Hynes, 1988).While there is a high degree of sequence conservationbetween ADAM13 and xMDC13, the respective mRNAshave different sizes on a Northern blot, and there are cleardifferences between the expression of these two genes inembryos and adult tissues. Both observations suggest thatadam13 and xmdc13 are not pseudoalleles (DeSimone andHynes, 1988). During early development, adam13 expres-sion begins at the midblastula transition and is localized tothe cranial neural crest, while xmdc13 expression firstappears at stage 13, and is weak but ubiquitous in embryos.Furthermore, xmdc13 is expressed in adult testis, whereasadam13 is not. One interpretation of these observations isthat xMDC13 and ADAM13 may have distinct functionsdespite their strong sequence similarity. While ADAM13most likely is important in early development (Alfandari etal., 1997), peptides corresponding to the predicted integrin-binding site of xMDC13 block X. laevis fertilization in aconcentration-dependent manner, suggesting a potentialrole of xMDC13 in fertilization (Shilling et al., 1997). Asdescribed previously (Alfandari et al., 1997), ADAM 13 andxMDC13 belong to a subfamily of metalloprotease–disintegrins that also includes Meltrin a (Yagami-Hiromasaet al., 1995) and Meltrin b (Inoue et al., 1998), but do notappear to be orthologues of either of these two mammalianproteins.

In summary, the identification of cDNA clones for

FIG. 4. Western blot analysis of xMDC9 in different tissues, and at different developmental stages. A and B show the characterization ofthe xMDC9 cytotail antibodies. All lanes in A contain glycoproteins isolated from adult X. laevis heart as described under Materials andMethods, which were treated with or without DTT prior to electrophoresis as indicated. Lanes 1 and 2 were probed with xMDC9–cyto IgG,lanes 3 and 4 with affinity-purified xMDC9–cyto IgG, lanes 5 and 6 with control IgG depleted of xMDC9 reactive antibodies, and lane 7with the secondary antibody alone. The generation of the antibody samples used here is described in more detail under Materials andMethods. In B, a Western blot of glycoproteins from an extract of COS-7 cells expressing xMDC9 (lanes 1 and 3) or of COS-7 cells transfectedwith the pcDNA3 vector alone (lanes 2 and 4) was probed with xMDC9–cyto IgG (lanes 1 and 2) or control IgG (lanes 2 and 4). In C,glycoproteins purified from extracts of Xenopus testis (lane 1), heart (lane 2), muscle (lane 3), and liver (lane 4) and from extracts of A6 cells(lane 5) were probed with xMDC9–cytotail IgG (top) or control IgG (bottom). The 75-kDa band appars to be the only band that is specificallyrecognized by the xMDC9–IgG, but not by the control IgG. In D, extracts of cell-surface-biotinylated A6 cells were immunoprecipitatedwith xMDC9–cyto IgG (lanes 1 and 2), or with control IgG (lanes 3 and 4), and treated with 10 mM DTT (lanes 1 and 3) or left nonreduced(lanes 2 and 4) prior to SDS–PAGE and then transferred to nitrocellulose. The immunoprecipitated biotinylated material was detected withhorseradish peroxidase-coupled streptavidin (see Materials and Methods). E shows a Western blot of glycoprotein extracts from eggs andfrom embryos at different stages of development. Only the 75-kDa protein was specifically recognized by the xMDC9–IgG (see arrow), butnot by the control antibody (not shown).

520 Cai et al.



metalloprotease–disintegrins in X. laevis presented herehas unveiled apparent orthologues to mammalian membersof this protein family, and has provided new insights intothe spatial and temporal expression of these genes in X.laevis. The expression of xmdc11a is highly localized todeveloping neural tissues, suggesting a role for this gene inthe migration or differentiation of tissues derived from theneural crest. The expression of xmdc9 in developing X.laevis is consistent with a general role which may berequired in all cells and tissues, but may neverthelessmanifest itself differently in diverse tissues. Since X. laeviscan be used as a system to evaluate protein function in earlydevelopment, this study represents an important first steptoward analyzing the role of xMDC9, xMDC11a, andxMDC13 in vertebrate development.

ACKNOWLEDGMENTS

We thank Ms. Selina Noramly for isolating the xMDC11a and bclones from a cDNA library, H. Cousin and C. Montmory for theirexcellent assistance in sectioning embryos, Drs. F. Fagotto, L. Jaffe,B. Gumbiner, D. DeSimone, J. L Duband, T. Wolfsberg, L. Howard,L. Lum and J. Schlondorff for helpful discussions and advice, andDr. W.-D. Schleuning for his continued interest and encourage-ment. This work was supported by NIH Grant R55GM51988 toC.P.B., and the Cancer Center Support Grant NCI-P30-CA-08748.

REFERENCES

Alexandropoulos, K., Cheng, G., and Baltimore, D. (1995). Proline-rich sequences that bind to Src homology 3 domains withindividual specificities. Proc. Natl. Acad. Sci. USA 92, 3110–3114.

Alfandari, D., Whittaker, C. A., DeSimone, D. W., and Darribere, T.(1995). Integrin alpha v subunit is expressed on mesodermal cellsurfaces during amphibian gastrulation. Dev Biol 170, 249–61.

Alfandari, D., Wolfsberg, T. G., White, J. M., and DeSimone, D. W.(1997). ADAM13: A novel ADAM expressed in somitic meso-derm and neural crest cells during Xenopus laevis development.Dev. Biol. 182, 314–330.

Almeida, E. A. C., Huovila, A.-P. J., Sutherland, A. E., Stephens,L. E., Calarco, P. G., Shaw, L. M., Mercurio, A. M., Sonnenberg,A., Primakoff, P., Myles, D. G., and White, J. M. (1995). Mouseegg integrin a6b1 functions as a sperm receptor. Cell 81, 1095–1104.

Black, R., Rauch, C. T., Kozlosky, C. J., Peschon, J. J., Slack, J. L.,Wolfson, M. F., Castner, B. J., Stocking, K. L., Reddy, P., Srini-vasan, S., Nelson, N., Boiani, N., Schooley, K. A., Gerhart, M.,Davis, R., Fitzner, J. N., Johnson, R. S., Paxton, R. J., March, C. J.,and Cerretti, D. P. (1997). A metalloprotease disintegrin thatreleases tumour-necrosis factor-a from cells. Nature 385, 729–733.

Blobel, C. P. (1997). Metalloprotease-disintegrins: Links to celladhesion and cleavage of TNFa and Notch. Cell 90, 589–592.

Blobel, C. P., Myles, D. G., Primakoff, P., and White, J. W. (1990).Proteolytic processing of a protein involved in sperm-egg fusioncorrelates with acquisition of fertilization competence. J. CellBiol. 111, 69–78.

Blobel, C. P., Wolfsberg, T. G., Turck, C. W., Myles, D. G.,Primakoff, P., and White, J. M. (1992). A potential fusion peptideand an integrin ligand domain in a protein active in sperm–eggfusion. Nature 356, 248–252.

Bronner-Fraser, M. (1993). Mechanisms of neural crest cell migra-tion. Bioessays 15, 221–30.

Bronner-Fraser, M. (1994). Neural crest cell formation and migra-tion in the developing embryo. FASEB J. 8, 699–706.

Brower, D. L., Brabant, M. C., and Bunch, T. A. (1995). Role of thePS integrins in Drosophila development. Immunol. Cell Biol. 73,558–564.

Chang, C., and Hemmati-Brivanlou, A. (1998). Neural crest induc-tion by Xwnt7B in Xenopus. Dev Biol 194, 129–34.

Chomczynski, P., and Sacchi, N. (1987). Single-step method ofRNA isolation by acid guanidium thiocyanate–phenol–chloroform extraction. Anal. Biochem. 162, 156–159.

DeSimone, D. W., and Hynes, R. O. (1988). Xenopus laevis inte-grins. Structural conservation and evolutionary divergence ofintegrin beta subunits. J. Biol. Chem. 263, 5333–40.

DeSimone, D. W., Norton, P. A., and Hynes, R. O. (1992). Identi-fication and characterization of alternatively spliced fibronectinmRNAs expressed in early Xenopus embryos. Dev Biol 149,357–69.

Emi, M., Katagiri, T., Harada, Y., Saito, H., Inazawa, J., Ito, I.,Kasumi, F., and Nakamura, Y. (1993). A novel metalloprotease/disintegrin-like gene at 17q21.3 is somatically rearranged in twoprimary breast cancers. Nature genetics 5, 151–157.

Erickson, C. A., and Perris, R. (1993). The role of cell–cell andcell–matrix interactions in the morphogenesis of the neuralcrest. Dev Biol 159, 60–74.

Evans, J. P., Kopf, G. S., and Schultz, R. M. (1997a). Characteriza-tion of the binding of recombinant mouse sperm fertilin betasubunit to mouse eggs: Evidence for adhesive activity via an eggb1 integrin-mediated interaction. Dev. Biol. 187, 79–93.

Evans, J. P., Schultz, R. M., and Kopf, G. S. (1995). Mouse sperm-eggplasma membrane interactions: Analysis of roles of egg integrinsand the mouse homologue of PH-30 (fertilin) b. J. Cell Sci. 108,3267–3278.

Evans, J. P., Schultz, R. M., and Kopf, G. S. (1997b). Characteriza-tion of the binding of recombinant mouse sperm fertilin alphasubunit to mouse eggs: evidence for function as a cell adhesionmolecule in sperm–egg binding. Dev. Biol. 187, 94–106.

Fambrough, D., Pan, D., Rubin, G. M., and Goodman, C. S. (1996).The cell surface metalloprotease/disintegrin kuzbanian is re-quired for axonal extension in Drosophila. Proc. Natl. Acad. Sci.USA 93, 13233–13238.

Fassler, R., Georges-Labouesse, E., and Hirsch, E. (1996). Geneticanalysis of integrin function in mice. Curr. Opin. Cell Biol. 8,641–646.

Finelli, A. L., Bossie, C. A., Xie, T., and Padgett, R. W. (1994).Mutational analysis of the Drosophila tolloid gene, a humanBMP-1 homolog. Development 120, 861–870.

Gumbiner, B. M. (1996). Cell adhesion: The molecular basis oftissue architecture and morphogenesis. Cell 84, 345–357.

Harland, R. M. (1991). In situ hybridization: An improved whole-mount method for Xenopus embryos. Methods Cell Biol. 36,685–95.

Harlow, E., and Lane, D. (1988). “Antibodies: A LaboratoryManual.” Cold Spring Harbor Laboratories, Cold Spring Harbor,NY.

522 Cai et al.


Harris, H. A., Murrills, R. J., and Komm, B. S. (1997). Expression ofmeltrin-alpha mRNA is not restricted to fusagenic cells. J. CellBiochem. 67, 136–42.

Heinlein, U. A. O., Wallat, S., Senftleben, A., and Lemaire, L.(1994). Male germ cell-expressed mouse gene TAZ83 encodes aputative, cysteine rich transmembrane protein (cyritestin) shar-ing homologies with snake venom toxins and sperm egg fusionproteins. Dev. Growth Differ. 36, 49–58.

Hooper, N. M., Karran, E. H., and Turner, A. J. (1997). Membraneprotein secretases. Biochem. J. 321, 265–279.

Howard, L., and Glynn, P. (1995). Membrane-associated metallo-proteinase recognized by characterisitic cleavage of myelin basicprotein: assay and isolation. In “Proteolytic Enzymes: Asparticand Metalloproteases” (A. J. Barrett, Ed.), Vol. 248, pp. 388–395.Academic Press, San Diego, CA.

Howard, L., Lu, X., Mitchell, S., Griffiths, S., and Glynn, P. (1996).Molecular cloning of MADM: A catalytically active disintegrin-metalloprotease expressed in various cell types. Biochem. J. 317,45–50.

Huovila, A. P. J., Almeida, E. A., and White, J. M. (1996). ADAMsand cell fusion. Curr. Opin. Cell Biol. 8, 692–9.

Hynes, R. O. (1996). Targeted mutations in cell adhesion genes:What have we learned from them? Dev. Biol. 180, 402–412.

Inoue, D., Reid, M., Lum, L., Kratzschmar, J., Weskamp, G.,Myung, Y. M., Baron, R., and Blobel, C. P. (1998). Cloning andinitial characterization of mouse meltrin beta and analysis of theexpression of four metalloprotease–disintegrins in bone cells.J. Biol. Chem. 273, 4180–4187.

Katagiri, T., Harada, Y., Emi, M., and Nakamura, Y. (1995). Humanmetalloprotease/disintegrin-like (MDC) gene: Exon–intron orga-nization and alternative splicing. Cytogenet. Cell Genet. 68,39–44.

Krotoski, D. M., Fraser, S. E., and Bronner-Fraser, M. (1988).Mapping of neural crest pathways in Xenopus laevis using inter-and intraspecific cell markers. Dev Biol 127, 119–32.

Lallier, T. E., Whittaker, C. A., and DeSimone, D. W. (1996).Integrin a6 expression is required for early nervous systemdevelopment in Xenopus laevis. Development 122, 2539–2554.

Linder, B., and Heinlein, U. A. (1997). Decreased in vitro fertiliza-tion efficiencies in the presence of specific cyritestin peptides.Dev. Growth Differ. 39, 243–7.

Lunn, C. A., Fan, X., Dalie, B., Miller, K., Zavodny, P. J., Narula,S. K., and Lundell, D. (1997). Purification of ADAM10 frombovine spleen as TNFalpha convertase. FEBS Lett. 400, 333–335.

Mancilla, A., and Mayor, R. (1996). Neural crest formation inXenopus laevis: Mechanisms of Xslug induction. Dev Biol 177,580–9.

Marques, G., Musacchio, M., Shimell, M. J., Wunnenberg-Stapleton, K., Cho, K. W., and O’Connor, M. B. (1997). Produc-tion of a DPP activity gradient in the early Drosophila embryothrough the opposing actions of the SOG and TLD proteins. Cell91, 417–26.

Mayor, R., Morgan, R., and Sargent, M. G. (1995). Induction of theprospective neural crest of Xenopus. Development 121, 767–77.

Monier-Gavelle, F., and Duband, J. L. (1997). Cross talk betweenadhesion molecules: control of N-cadherin activity by intracel-lular signals elicited by beta1 and beta3 integrins in migratingneural crest cells. J. Cell Biol. 137, 1663–81.

Moss, M. L., Jin, S.-L. C., Milla, M. E., Burkhart, W., Cartner, H. L.,Chen, W.-J., Clay, W. C., Didsbury, J. R., Hassler, D., Hoffman,C. R., Kost, T. A., Lambert, M. H., Lessnitzer, M. A., McCauley,P., McGeehan, G., Mitchell, J., Moyer, M., Pahel, G., Rocque, W.,

Overton, L. K., Schoenen, F., Seaton, T., Su, J.-L., Warner, J.,Willard, D., and Becherer, J. D. (1997). Cloning of a disintegrinmetalloproteinase that processes precursor tumour-necrosisfactor-a. Nature 385, 733–736.

Myles, D. G., Kimmel, L. H., Blobel, C. P., White, J. M., andPrimakoff, P. (1994). Identification of a binding site in thedisintegrin domain of fertilin required for sperm–egg fusion.Proc. Natl. Acad. Sci. USA 91, 4195–4198.

Pan, D., and Rubin, J. (1997). KUZBANIAN controls proteolyticprocessing of NOTCH and mediates lateral inhibition duringDrosophila and vertebrate neurogenesis. Cell 90, 271–280.

Piccolo, S., Agius, E., Lu, B., Goodman, S., Dale, L., and DeRobertis, E. M. (1997). Cleavage of Chordin by Xolloid metallo-protease suggests a role for proteolytic processing in the regula-tion of Spemann organizer activity. Cell 91, 407–16.

Podbilewicz, B. (1996). ADM-1, a protein with metalloprotease-and disintegrin-like domains, is expressed in syncitial organs,sperm and sheath cells of sensory organs in Caenorhabditiselegans. Mol. Biol. Cell 7, 1877–1893.

Primakoff, P., Hyatt, H., and Tredick-Kline, J. (1987). Identificationand purification of a sperm surface protein with a potential rolein sperm–egg membrane fusion. J. Cell Biol. 104, 141–149.

Ramos, J. W., and DeSimone, D. W. (1996). Xenopus embryonic celladhesion to fibronectin: position-specific activation of RGD/synergy site-dependent migratory behavior at gastrulation. J CellBiol 134, 227–40.

Ramos, J. W., Whittaker, C. A., and DeSimone, D. W. (1996).Integrin-dependent adhesive activity is spatially controlled byinductive signals at gastrulation. Development 122, 2873–83.

Roodman, G. D. (1996). Advances in bone biology: the osteoclast.Endocrin. Rev. 17, 308–332.

Rooke, J., Pan, D., Xu, T., and Rubin, G. M. (1996). KUZ, aconserved metalloprotease–disintegrin protein with two roles inDrosophila neurogenesis. Science 273, 1227–1230.

Rosendahl, M. S., Ko, S. C., Long, D. L., Brewer, M. T., Rosenzweig,B., Hedl, E., Anderson, L., Pyle, S. M., Moreland, J., Meyers,M. A., Kohno, T., Lyons, D., and Lichenstein, H. S. (1997).Identification and characterization of a pro-tumor necrosisfactor- alpha-processing enzyme from the ADAM family of zincmetalloproteases. J. Biol. Chem. 272, 24588–93.

Sadaghiani, B., and Thiebaud, C. H. (1987). Neural crest developmentin the Xenopus laevis embryo, studied by interspecific transplanta-tion and scanning electron microscopy. Dev. Biol. 124, 91–110.

Schuh, T. J., Hall, B. L., Kraft, J. C., Privalsky, M. L., and Kimelman,D. (1993). v-erbA and citral reduce the teratogenic effects ofall-trans retinoic acid and retinol, respectively, in Xenopusembryogenesis. Development 119, 785–98.

Shilling, F. M., Kratzschmar, J., Cai, H., Weskamp, G., Gayko, U.,Leibow, J., Myles, D. G., Nuccitelli, R., and Blobel, C. P. (1997).Identification of metalloprotease/disintegrins in Xenopus laevistestis with a potential role in fertilization. Dev. Biol. 186, 155–164.

Shimell, M. J., Ferguson, E. L., Childs, S. R., and O’Connor, M. B.(1991). The Drosophila dorsal–ventral patterning gene tolloid isrelated to human bone morphogenic protein 1. Cell 67, 469–481.

Sotillos, S., Roch, F., and Campuzano, S. (1997). Themetalloprotease-disintegrin Kuzbanian participates in Notch ac-tivation during growth and patterning of Drosophila imaginaldiscs. Development 124, 4769–4779.

von Heijne, G. (1986). A new method for predicting signal sequencecleavage sites. Nucleic Acids Res. 14, 4683–4690.

Wen, C., Metzstein, M. M., and Greenwald, I. (1997). SUP-17, aCaenorhabditis elegans ADAM protein related to Drosophila



KUZBANIAN, and its role in LIN-12/NOTCH signaling. Devel-opment 124, 4759–4767.

Weskamp, G., Kratzschmar, J. R., Reid, M., and Blobel, C. P. (1996).MDC9, a widely expressed cellular disintegrin containing cyto-plasmic SH3 ligand domains. J. Cell Biol. 132, 717–726.

Wolfsberg, T. G., Bazan, J. F., Blobel, C. P., Myles, D. G., Primakoff,P., and White, J. M. (1993). The precursor region of a proteinactive in sperm–egg fusion contains a metalloprotease and adisintegrin domain: Structural, functional and evolutionary im-plications. Proc. Natl. Acad. Sci. USA 90, 10783–10787.

Wolfsberg, T. G., and White, J. M. (1996). ADAMs in fertilizationand development. Dev. Biol. 180, 389–401.

Yagami-Hiromasa, T., Sato, T., Kurisaki, T., Kamijo, K., Na-beshima, Y., and Fujisawa-Sehara, A. (1995). A metalloprotease–disintegrin participating in myoblast fusion. Nature 377, 652–656.

Yuan, R., Primakoff, P., and Myles, D. G. (1997). A role for thedisintegrin domain of cyritestin, a sperm surface protein belong-ing to the ADAM family, in mouse sperm–egg plasma membraneadhesion and fusion. J. Cell Biol. 137, 105–112.

Received for publication May 7, 1998Revised July 13, 1998

Accepted July 14, 1998

524 Cai et al.


Documents

Neural Crest-Specific and General Expression of Distinct Metalloprotease–Disintegrins in EarlyXenopus laevisDevelopment