BMP1-Related Metalloproteinases Promote the Development of Ventral Mesoderm in Early Xenopus Embryos

DEVELOPMENTAL BIOLOGY 195, 144–157 (1998) ARTICLE NO. DB978840

BMP1-Related Metalloproteinases Promote the Development of Ventral Mesoderm in Early Xenopus Embryos Shelley A. Goodman,* Rodolpho Albano,* Fiona C. Wardle,* Glenn Matthews,† David Tannahill,‡ and Leslie Dale*,1 *Department of Anatomy and Developmental Biology, University College London, Gower Street, London, WC1E 6BT, United Kingdom; †Department of Surgery, University of Birmingham, Queen Elizabeth Hospital, Edgbaston, Birmingham, B15 2TH, United Kingdom; and ‡Department of Anatomy, University of Cambridge, Downing Street, Cambridge, CB2 3DY, United Kingdom

Bone morphogenetic protein 1 (BMP1) is a metalloproteinase closely related to Drosophila Tolloid (Tld). Tld regulates dorsoventral patterning in early Drosophila embryos by enhancing the activity of Dpp, a member of the TGF-b family most closely related to BMP2 and BMP4. In Xenopus BMP4 appears to play an essential role in dorsoventral patterning, promoting the development of ventral fates during gastrula stages. To determine if BMP1 has a role in regulating the activity of BMP4, we have isolated cDNAs for Xenopus BMP1 and a novel closely related gene that we have called xolloid (xld). Whereas xbmp1 is uniformly expressed at all stages tested, the initial uniform expression of xld becomes localized to two posterior ectodermal patches flanking the neural plate and later to the inner ectoderm of the developing tailbud. xld is also expressed in dorsal regions of the brain during tailbud stages and is especially abundant in the ventricular layer of the dorsal hindbrain caudal to the otic vesicle. Overexpression of either gene inhibits the development of dorsoanterior structures in whole embryos and ventralizes activin-induced dorsal mesoderm in animal caps. Since ventralization of activin-induced animal caps can be blocked by coinjecting a dominant-inhibitory receptor for BMP2 and BMP4, we suggest a role for BMP1 and Xld in regulating the ventralizing activity of these molecules. q 1998 Academic Press Key Words: Xenopus; ventralization; BMP1; tolloid; metalloendoprotease.

INTRODUCTION Bone morphogenetic proteins (BMPs) were initially identified in bone extracts that induced ectopic bone formation in rodents (Wozney et al., 1988). Most of these molecules define a subgroup of closely related members of the TGF-b family, that act as extracellular signals in a diverse range of developmental processes (reviewed by Hogan, 1996). In early Xenopus embryos BMP4 plays an important role in patterning the dorsoventral axis, promoting the development of both ventral mesoderm and epidermis. Overexpression of BMP4 converts dorsal into ventral mesoderm during gastrula stages (Dale et al., 1992; Jones et al., 1992; 1996), while inhibition of BMP signaling converts ventral into dorsal mesoderm (Graff et al., 1994; Maeno et al., 1994; Suzuki et al., 1994). In early gastrulae bmp4 is expressed 1 To whom correspondence should be addressed. Fax: 0171 380 7349. E-mail: [email protected].

throughout the animal cap and in ventral and lateral sectors of the marginal zone, but is excluded from a dorsal region of the marginal zone known as the organizer (Fainsod et al., 1994; Schmidt et al., 1995; Hemmati-Brivanlou and Thomsen, 1995). The organizer releases signals that are responsible for both dorsalizing lateral mesoderm and neuralizing the animal cap, signals that can be mimicked by three secreted factors—chordin, noggin, and follistatin (reviewed by Hemmati-Brivanlou and Melton, 1997; Thomsen, 1997). All three proteins are synthesized by the organizer during gastrula stages and directly bind BMP4, thereby neutralizing its activity (Piccolo et al., 1996; Zimmerman et al., 1996; Fainsod et al., 1997). The antagonistic interaction between these organizer signals and BMP4 may establish a morphogenetic gradient of BMP4 activity across the dorsoventral axis, cells adopting different fates as a function of its concentration (Dosch et al., 1997). A similar system of extracellular signals is responsible for dorsoventral patterning in the early Drosophila embryo, although the axis has been inverted (reviewed by De Rob0012-1606/98 $25.00 Copyright q 1998 by Academic Press All rights of reproduction in any form reserved.

144

AID

DB 8840

/

6x39$$$$$1

02-25-98 12:09:21

dbal

145

BMPs in Xenopus Development

ertis and Sasai, 1996). Genetic studies have identified a small number of zygotic genes that regulate dorsoventral patterning in the early embryo, including decapentaplegic (dpp), tolloid (tld) and short gastrulation (sog) (Ferguson and Anderson, 1992). The critical element in this system is Dpp, a member of the TGF-b family that is structurally and functionally homologous to BMP2 and BMP4 (Sampath et al., 1993; Padgett et al., 1993; Holley et al., 1995). Dpp appears to function as a morphogen, specifying dorsal structures when activity is high and more ventral structures when activity is low (Ferguson and Anderson, 1992). Moreover, the activity of Dpp is antagonized by ventrally expressed Sog, a structural and functional homologue of the Xenopus organizer signal chordin (Holley et al., 1995). In contrast, Tld is a metalloproteinase that enhances the activity of Dpp, possibly by releasing Dpp from inactive complexes with inhibitory binding proteins such as Sog (Shimell et al., 1991; De Robertis and Sasai, 1996). In addition to the N-terminal metalloproteinase domain, Tld also contains five C-terminal CUB repeats and two EGF repeats that are thought to mediate protein-protein interactions (Shimell et al., 1991). A similar structure is also found in the mTld splicing variant of vertebrate BMP1, the only BMP that is not a member of the TGF-b superfamily, while a shorter splicing variant has only three CUB repeats and a single EGF repeat (Wozney et al., 1988; Fukagawa et al., 1994; Takahara et al., 1994). The fact that BMP1 was copurified with BMP2 and BMP3 (Wozney et al., 1988) suggests that it may physically associate with one or more of the TGF-b-related BMPs, perhaps regulating the release of active BMP by proteolytic cleavage. More recently, BMP1 was shown to be identical to procollagen C-proteinase (PCP), an enzyme responsible for cleaving the C-terminal peptide from types I, II and III procollagen (Kessler et al., 1996; Li et al., 1996). PCP is also capable of correctly processing prolysyl oxidase (Panchenko et al., 1996), an enzyme involved in intermolecular cross-linking in collagen fibres. This suggests that BMP1 plays an important role in the assembly of mature collagen fibres, and a loss-of-function mutation in the mouse bmp1 gene disrupts the normal processing and deposition of collagen (Suzuki et al., 1996). However, this does not preclude a wider role for BMP1, or related proteinases, which may have many target proteins. If BMP1-related metalloproteinases are responsible for regulating the activity of BMP4, we might expect them to be involved in patterning the mesoderm during early Xenopus development. Preliminary experiments with human BMP1 suggested that this may indeed be the case, injected embryos typically lack a notochord (Dale, 1997). To further explore the role of BMP1-related metalloproteinases in Xenopus development, we have isolated cDNAs for Xenopus BMP1 (xbmp1) and a closely related gene that we have called xolloid (xld). We find that both genes are uniformly expressed at blastula and gastrula stages, but that xld becomes localized to the tip of the developing tail during neurula stages. Injection of synthetic mRNA for either gene into early Xenopus embryos generates a similar, albeit weaker, phenotype to that previously described for injection of

bmp4 RNA. The resulting embryos lack dorsoanterior structures, including the notochord. XBMP1 and Xld will also ventralize dorsal mesoderm induced in animal caps by activin, an effect that can be blocked by coinjection of a dominant-inhibitory receptor for BMP2 and 4. These results implicate BMP1-related metalloproteinases as regulators of TGF-b-related BMPs during dorsoventral patterning in early amphibian embryogenesis.

MATERIALS AND METHODS Embryo Culture and Manipulation Xenopus laevis embryos were obtained by artificial fertilization, allowed to rotate and then dejellied with 2% cysteine hydrochloride. Injected embryos were scored using the Dorso-Anterior Index of Kao and Elinson (1988), in which a score of 5 Å normal embryos, 4 Å reduced eyes and forehead, 3 Å cyclopic, 2 Å microcephalic, 1 Å acephalic and 0 Å fully ventralized. Animal caps were isolated from mid-blastulae and incubated in full-strength MBS (88 mM NaCl, 1 mM KCl, 2.4 mM NaHCO3 , 15 mM HEPES, 0.3 mM CaNO3 , 0.4 mM CaCl2 , 0.8 mM MgSO4 , 25 mg/l gentamycin, adjusted to pH 7.6) containing 10 units of human activin A until controls had reached stage 26. Embryos required for histology were fixed in 4% paraformaldehyde and stained overnight in a saturated solution of borax carmine (in 35% ethanol). Paraffin sections (10 mm) were cut and counter stained with 0.1% naphthalene black in saturated aqueous picric acid. Whole mount in situ hybridizations were performed on albino embryos essentially as described by Harland (1991), with the exception that the RNase digestion step was omitted, CHAPS buffer was replaced by Maleic Acid buffer (100 mM Maleic Acid, pH 7.8; 150 mM NaCl; 0.1% Tween 20), and BM Purple (Boehringer) was used as a substrate. DIG labeled probes were made from the same DNA fragments as RNase protection probes (see below).

cDNA Library Screening A Xenopus oocyte cDNA library prepared in lambda ZAP (Shuttleworth et al., 1988) was plated on E. coli BB4 cells. Approximately 5 1 105 plaques were screened with a random primed full-length probe for human BMP1 (Wozney et al., 1988). Hybridization was performed in 5 1 SSPE, 20% formamide, 1 1 Denhardt’s solution and 100 mg/ml yeast tRNA at 427C for 18 hours. Filters were washed in 2 1 SSC and 0.1% SDS at 607C. Three positive clones were isolated and sequenced using standard dideoxynucleotide chaintermination techniques. Two of the positive clones contained 3.2 and 4.8 kb cDNAs encoding xld, the larger one encoding the entire protein. The third clone contained a 2.4 kb cDNA encoding xbmp1, but lacked the initiating ATG and signal sequence. An additional 80 nucleotides, encoding the missing 25 N-terminal amino acids, was obtained by PCR from a Xenopus gastrula cDNA library prepared in pcDNA1 (Invitrogen). PCR was performed with a 5* anchored primer - 5*-GGTACCTAATACGACTCACTATAGGG-3* and a 3* gene specific primer - 5*-ACATAAGGAATGACCCC-3*. The 480 bp PCR product was ligated to the Xho I site of the 2.4 kb xbmp1 cDNA, such that the entire open reading frame was generated.

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

AID

DB 8840

/

6x39$$$$$1

02-25-98 12:09:21

dbal

146

Goodman et al.

02-25-98 12:09:21

dbal

147

BMPs in Xenopus Development

at the C-terminus of the metalloproteinase domain, generating the construct Dxbmp1 which was subcloned into pRN3. The Eco RIPvu II fragment of xld, encoding the full protein sequence, was also subcloned into pRN3. Human interstitial collagenase (mmp1) in pSP64 was provided by Dr. Dylan Edwards (University of Calgary). Six myc-tags were added to the C-terminus of XBMP1 and Xld by using PCR to subclone these cDNAs into the Bam H1 site of pCS2/MT. Capped synthetic mRNAs were transcribed with either SP6 or T3 polymerase and taken up in sterile RNase free water. Embryos were injected with 10-15 nl of mRNA (0.2 mg/ml) at the 14 cell stage in 10% MBS plus 3% Ficoll (Sigma, Type 400). Fully grown Xenopus oocytes were injected with 20-30 nl of mRNA (1 mg/ml) and incubated for 48 hours in full-strength MBS plus 1% BSA.

RNA Analysis

FIG. 1. (A) Alignments of the deduced amino acid sequences of Xld and XBMP1 (only nonconserved amino acids are indicated for XBMP1) are presented beneath a schematic of the domain structure of Xld; the identity of each domain is indicated above the schematic. Note that relative to XBMP1, Xld contains one additional EGF repeat and two additional CUB repeats. Conserved sequences in the proregion are underlined as are the highly conserved metalloproteinase domain sequences responsible for coordinating a zinc atom in the active site of astacin. These sequences will appear in the EMBL, GenBank, and DDBJ Nucleotide Sequence Databases under Accession Nos. Y09660 and Y09661. (B) Alignment of the highly conserved 21-amino-acid sequence in the proregion of the BMP1-related metalloproteinases Xolloid (Xld), Xenopus BMP1 (XBMP1), human BMP1 (HBMP1), murine BMP1 (MBMP1), mammalian Tolloid-like (MTLL), sea urchin BMP1 (SBMP1), Drosophila Tolloid-related 1 (DTLR1), and Drosophila Tolloid (DTLD). The murine sequence has been modified to take account of a possible error in the published sequence; the first 12 amino acids (underlined) are in a different reading frame to the last 9 amino acids. Although DTLD also has a single cysteine in the proregion, the flanking amino acids share no homology to the other seven proteins. (C) Percentage identity between Xolloid and other BMP1-related proteins is presented beneath schematics illustrating the domain structure of XBMP1 and Xolloid. The percentage identity between each domain is presented separately; figures for the highly diverged proregion are not included. We also include the Xenopus and human Tld splice variants of BMP1 (XTld and HTld), murine Tldlike (MTll), and Drosophila Tld (DTld) and Tld-related 1 (DTlr1).

Poly (A)/ mRNA was isolated on oligo dT columns (GIBCO-BRL) according to the manufacturer’s instructions. For Northern blots, RNAs were separated on formaldehyde-agarose (1.5%) gels, transferred to nylon (Amersham), and probed with random primed DNA probes made from the full-length cDNAs. RNase protections for xbmp1 and xld were performed in 50% formamide using up to 10 mg of total RNA per hybridization. Probes were: 230 bp Eco RI-Xho I fragment of xbmp1, 250 bp Hind III fragment of xld, and a 90 bp PCR fragment for ornithine decarboxylase (odc) (Isaacs et al., 1992). Following hybridization single stranded RNA was digested with 700 units/ml of RNase T1 (GIBCO-BRL) and 30 mg/ml RNase A (Sigma). Control hybridizations with 10 mg of tRNA were always negative and are not included in the figures. RNase protections for a-actin (Mohun et al., 1988) and a-globin (Walmsley et al., 1994) were performed using the Ambion RPA II kit according to the manufacturer’s instructions, except that RNase digestion was carried out with 700 units/ml of RNase T1 alone.

Protein Analysis Capped synthetic mRNAs were translated in a message dependent Xenopus cell-free egg extract (Matthews and Colman, 1991) in the presence of 1 mCi/ml of 35S-methionine (Amersham). To test for glycosylation we included the tripeptide (acetyl)-Asn-TyrThr-(amide) at 5 mM which inhibits N-glycosylation in this cellfree system. mRNAs were also translated in a message dependent rabbit reticulocyte lysate (Promega) according to the manufacturer’s instructions. Radiolabeled proteins were separated by SDSPAGE on a minigel apparatus (Bio-Rad) according to the manufacturer’s instructions. After electrophoresis, gels were fixed and prepared for fluorography with Enhance (DuPont). Oocytes were homogenized in 100 mM Tris-HCl (pH 7.5), 1% Triton X100, 1 mM PMSF and microfuged for 15 minutes to remove yolk platelets. Oocyte extracts and culture media were separated by SDS-PAGE and transferred to Nylon membranes (Amersham). XBMP1 and Xld were detected by ECL (Amersham), according to the manufacturer’s instructions, using an antibody (9E10) against the myc-tag epitope.

RESULTS

Microinjections

Isolation and Characterization of Xenopus BMP1-Related cDNAs

xbmp1 was subcloned into a modified version of pSP64T containing multiple cloning sites. A stop codon was introduced by PCR

cDNAs for Xenopus bmp1 (xbmp1) and a novel gene we have called xolloid (xld), were isolated from a screen of a

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

AID

DB 8840

/

6x39$$$$$1

02-25-98 12:09:21

dbal

148

Goodman et al.

FIG. 2. XBMP1 and Xld are secreted glycoproteins. (A) In vitro translation of xbmp1 and xld. Translation products of 80 (track 6) and 117 kDa (track 3) were obtained for xbmp1 and xld in a rabbit reticulocyte lysate (RL), while translation products of 90 (track 4) and 135 kDa (track 1) were obtained for the same mRNAs in a Xenopus cell-free extract (E0). Since translation products of only 78 (track 5) and 117 kDa (track 2) were obtained in the Xenopus extract in the presence of a tripeptide inhibitor of glycosylation (E/), both proteins must enter the endoplasmic reticulum and be glycosylated. The lower Mr for XBMP1 in track 5 relative to that in track 6 is probably a consequence of signal peptide removal; this is not always evident by SDS–PAGE as can be seen for Xld under the same conditions. Translation products were separated on a 7% SDS–polyacrylamide gel. (B) Western blot analysis of myc-tagged XBMP1 (tracks 5 and 6) and Xld (tracks 3 and 4) in Xenopus oocytes (O) and culture media (M). The equivalent of one-half an oocyte of cell extract and the culture media for four oocytes were separated on a 10% polyacrylamide gel. Although most of the protein is retained by the oocyte (tracks 3 and 5), both proteins are also secreted into the culture media (tracks 4 and 6). No signal could be detected in uninjected oocytes and culture media (tracks 1 and 2).

splicing variant of human BMP1 (Wozney et al., 1988; Takahara et al., 1994). A longer splicing variant of xbmp1 has recently been described by Lin et al., (1997). A 4.8 kb cDNA for xld encodes a protein of 1019 amino acids (Fig. 1A) that is closely related to BMP1. The N-terminal signal sequence is followed by a 121 amino acid proregion with little homology to that of BMP1, except for a 21 amino acid sequence that is also found in the pro-regions of human BMP1, mouse Tolloid-like, sea urchin BMP1, Aplysia TBL-1 and Drosophila Tld-related 1 (Fig. 1B). Although the published sequence for murine BMP1 lacks the first 12 amino acids of this sequence (Fukagawa et al., 1994), they can be found in a different reading frame of the cDNA. These 12 amino acids are included in a 35 amino acid stretch that is 89% identical to the equivalent region in human BMP1, suggesting that an error during cloning may have placed it in a different frame to the rest of the protein. There is currently no information to indicate the role of this conserved sequence, but it appears to contain an unpaired cysteine which may form an interchain disulphide bond with an as yet unidentified protein. Alternatively, this cysteine may act as a ‘‘cysteine switch’’, as proposed for the unpaired cysteine in the proregion of the matrix metalloproteinases (Van Wart and Birkedal-Hansen, 1990). In these latter proteins, the unpaired cysteine inactivates the protease by interacting with the active site zinc atom, activation is achieved by breaking this interaction. The 201 amino acid metalloproteinase domain of Xld is preceded by a short basic sequence (RVRR) believed to be a cleavage site for intracellular processing enzymes (Barr, 1991). A similar site is found at the same location in most BMP1-related metalloproteinases, and cleavage at this site is thought to be a requirement for enzymatic activity (Bode et al., 1992). The metalloproteinase domain of Xld is most closely related to mouse Tolloid-like (87% identity) and both Xenopus and mammalian BMP1 (85% identity) (Fig. 1C). Like BMP1, the metalloproteinase domain of Xld contains two highly conserved sequences, HExxHxxGFxHExxRxDRD and YDxxS(I/V)MHY (Fig. 1A), that are responsible for coordinating a zinc atom in the active site of astacin (Bode et al., 1992). The protease domain of Xld is followed by two CUB repeats, an EGF repeat and a third CUB repeat which together share 77% identity with Xenopus BMP1. xld also encodes additional EGF (x1) and CUB (x2) repeats, a similar structure to Drosophila Tld (Shimell et al., 1991) and the Tld splicing variant of BMP1 (Fukagawa et al., 1994; Takahara et al., 1994; Lin et al., 1997).

XBMP1 and Xld Are Secreted Glycoproteins Xenopus oocyte cDNA library (see Materials and Methods). A 2.5 kb cDNA for xbmp1 encodes a protein of 735 amino acids (Fig. 1A) that is 93% identical to human bmp1 (Wozney et al., 1988), and 95.5% identical to a previously described partial cDNA for xbmp1 (Maeno et al., 1993). In addition to the N-terminal metalloproteinase domain, this cDNA encodes three C-terminal CUB repeats and a single EGF repeat, a domain structure most similar to the short-

Synthetic mRNA for xbmp1 gave a translation product of Mr 80 kDa in a rabbit reticulocyte cell-free translation system (Fig. 2A, track 6), quite close to the 83.6 kDa predicted from the amino acid sequence, while a product of Mr 90 kDa was obtained when the same mRNA was translated in a Xenopus cell-free system that possesses a functional endoplasmic reticulum (track 4) (Matthews and Colman, 1991). That this increase in Mr was due to glycosylation was demonstrated

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

AID

DB 8840

/

6x39$$$$$1

02-25-98 12:09:21

dbal

149

BMPs in Xenopus Development

by translating xbmp1 in the presence of an inhibitor of Nlinked glycosylation, the tripeptide (acetyl)-Asn-Tyr-Thr(amide), which resulted in a smaller product of Mr 77.5 kDa (track 5). Translation of synthetic mRNA for xld gave a product of Mr 117 kDa in a rabbit reticulocyte cell-free translation system (track 3), which is quite close to the 115 kDa predicted from the amino acid sequence. An almost identical product was obtained following translation of xld in the Xenopus cell-free system in the presence of the tripeptide inhibitor of glycosylation (track 2). A product of 135 kDa was obtained in the absence of the inhibitor (track 1) demonstrating that Xld is also glycosylated. To confirm that XBMP1 and Xld are secreted, we made constructs which added six myctags to the C-terminus of each protein and injected the mRNAs into Xenopus oocytes. Oocytes and culture media were collected after 48 hours, and proteins analyzed by Western blots using the 9E10 myc antibody (Fig. 2B). Although most of the detectable BMP1 and Xolloid was retained by the oocyte (tracks 3 and 5), both proteins could also be detected in the media (tracks 4 and 6).

Expression of BMP1 and Xld

FIG. 3. Temporal and spatial expression of xbmp1 and xolloid in Xenopus embryos. (A) Northern blot analysis of poly (A)/ mRNA (2 mg/track) from 2- to 8-cell (tracks 1 and 3) and stage 28 embryos (tracks 2 and 4). The blot was initially probed with xbmp1, stripped, and then reprobed with xolloid. Whereas three transcripts of approximately 5.6, 4.5, and 2.4 kb were detected with the xbmp1 probe at all stages tested (tracks 1 and 2), only a single transcript of approximately 5.0 kb was detected with the xolloid probe in early cleavage stages through to neurula stages (tracks 3 and 4). Two weaker bands of approximately 3.5 kb were also detected at stage 28 with the xolloid probe (track 4). (B) RNase protection analysis of total RNA (10 mg/protection) isolated from indicated stages of Xenopus development using probes for either xbmp1 or xld. Each protection also contained probe for the ubiquitously expressed ornithine decarboxylase (odc) gene. Stages shown are 16-cell (stage 5), early blastula (stage 7), late blastula (stage 9), early gastrula (stage 10), late gastrula

Three transcripts of 2.5, 4.5 and 5.6 kb were detected when xbmp1 was used to probe a Northern blot of poly (A)/ mRNA isolated from cleavage stages (2-8 cell) and an early tailbud stage (stage 28) (Fig. 3A, tracks 1 and 2). This is similar to the situation found in mammals where the two largest transcripts are detected by probes specific for the mtld splicing variant (Takahara et al., 1994). When the blot was reprobed with xld, a major transcript of approximately 5.0 kb was detected at both stages (tracks 3 and 4), while two minor transcripts of approximately 3.5 kb were present only at stage 28 (track 4). Since a full-length probe for xld does not detect any of the transcripts detected by xbmp1, and a nearly full-length probe for xbmp1 does not detect transcripts for xld, it is clear that xld is a novel gene rather than a splicing variant of xbmp1. The temporal and spatial expression patterns of xbmp1 and xld were analyzed in detail by RNase protections (Figs. 3B, C). Since the probe for xbmp1 encompasses the proregion, it does not discriminate between the long and shortsplicing variants. xbmp1 and xld are expressed at all developmental stages tested, although transcript levels drop during gastrula and neurula stages before rising again in early

(stage 12), neurula (stage 16), late neurula (stage 20), early tailbud (stage 25), tailbud (stage 30), and tadpole (stage 40). (C) RNase protection analysis of total RNA (10 mg/protection) isolated from embryo fragments at the indicated stages using probes for either xbmp1 or xolloid. Each protection also contained probe for the ubiquitously expressed ornithine decarboxylase (odc) gene. The fragments analyzed were AP, animal pole; VP, vegetal pole; DMZ, dorsal marginal zone; VMZ, ventral marginal zone; and LMZ, lateral marginal zone. Head and Tail are self explanatory; Dorsal, dorsal abdomen; Ventral, ventral abdomen.

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

AID

DB 8840

/

6x39$$$$$1

02-25-98 12:09:21

dbal

150

Goodman et al.

FIG. 4. Whole-mount in situ hybridization analysis of xbmp1 and xolloid in Xenopus embryos. (A) Early gastrulae (stage 10) hybridized with a probe for xbmp1; the left embryo is viewed from the animal pole (AP) and the right embryo from the vegetal pole (VP). Transcripts are uniformly expressed in the animal hemisphere but cannot be detected in the vegetal hemisphere; failure to detect vegetal transcripts can be quite common using this whole-mount in situ hybridization technique. (B) Tailbud stage embryo (stage 28) showing widespread expression of xbmp1; transcripts are most abundant in the head and dorsal axis. (C) Neurula stage embryo (stage 17) showing punctate expression of xld in two dorsal patches flanking the posterior neural plate (NP). Anterior (Ant) is to the left and posterior (Post) to the right. (D) Tailbud stage embryo (stage 28) showing expression of xld at the distal tip of the tail, flanking the posterior endoderm and in dorsal parts of the brain. The arrow indicates expression of xld in the caudal hindbrain. (E) Transverse section through the hindbrain showing expression of xld in the dorsal ventricular layer posterior to the otic vesicle (see arrow in D). NT, neural tube; No, notochord; So, somite. (F) Transverse section through the tip of the tail (stage 28) showing xld expression in the inner layer of the epidermis. D, dorsal; V, ventral.

(xld) and late (xbmp1) tailbud stages (Fig. 3B). A probe for the uniformly expressed ornithine decarboxylase (odc) gene confirms that roughly equivalent levels of RNA were added to each assay. The result for xbmp1 contrasts with those previously reported by Maeno et al. (1993), who failed to detect any maternal expression of xbmp1. In our hands, maternal xbmp1 expression has been confirmed using three independent assays: Northern blotting (Fig. 3A), RNase protection (Fig. 3B) and RT-PCR (data not shown). Moreover, our xbmp1 cDNA was isolated from an oocyte library. RNase protection of total RNA isolated from animal and vegetal hemispheres dissected from blastulae (stage 8) suggested that both xbmp1 and xld are expressed in both hemispheres (Fig. 3C). A similar result was obtained at stage 10 (early gastrula), when xbmp1 was also expressed at similar levels in dorsal, lateral and ventral marginal zones. In contrast, there appeared to be slightly more xld RNA in the dorsal marginal zone (track 4) than in lateral or ventral mar-

ginal zones (tracks 3 and 5). At stage 28, expression of xld, but not xbmp1, was largely restricted to the tail (track 9). Finally, the spatial expression patterns of xbmp1 and xld were analyzed by whole mount in situ hybridization. xbmp1 was detected in most cells of the embryo at all stages tested (stage 10-28), although it appeared to be particularly abundant in the head and dorsal axis of tailbud stages (Figs. 4A, B). The lack of signal in the endoderm is quite common using this technique. However, this probe also detects the more abundantly expressed long-splicing variants, which may mask differential expression of the short-splicing variant used in this study. In contrast, xld could not be detected until neurula stages, when expression was observed in two ectodermal patches of positive cells interspersed by negative cells that flank the posterior neural plate (Fig. 4C). This region of the embryo ultimately gives rise to the epidermis of the tail (Tucker and Slack, 1995), and at early tailbud stages xld is abundantly expressed in the inner layer of epi-

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

AID

DB 8840

/

6x39$$$$$1

02-25-98 12:09:21

dbal

151

BMPs in Xenopus Development

TABLE 1 Overexpression of BMP1 and Xolloid Reduces the Development of Anterior Structures RNA

AD (%)

DAI

TD (%)

EP (%)

GD (%)

N

None Dxbmp1 xbmp1 xld mmp1

0 10 70 69 3

5.00 4.90 3.84 3.68 4.97

0 0 10 0 0

0 0 40 70 0

1 25 20 8 5

267 101 103 118 41

tralized phenotype and mean DAI scores of 3.84 and 3.68 respectively (Table 1). Embryos were rarely scored as fully ventralized and only 10-20% were scored as 2 or lower; anterior defects rarely extended into the hindbrain as indi-

Note. Embryos were injected at the one- to two-cell stage with 3 ng of the indicated RNAs and scored at stage 30. AD, anterior defects; DAI, dorso anterior index; TD, tail duplications; EP, enlarged proctodeum. Embryos scored as TD and EP were also scored for anterior defects and are included in the DAI score. GD, gastrulation defects; in these embryos the blastopore failed to close resulting in a spina bifida-like phenotype; they were not scored for anterior defects and are not included in the DAI score. N, total number of embryos scored, including GD embryos.

dermis at the distal tip of the tail (Figs. 4D, F). A weaker signal was seen throughout the forebrain and in dorsal regions of the midbrain and hindbrain at tailbud stages (Fig. 4D), and was particularly prominent in the dorsal ventricular layer of the caudal hindbrain just posterior to the otic vesicle (Figs. 4D, E). The results are largely consistent with the RNase protection data presented in Fig. 3C.

Overexpression of BMP1 and Xld Disrupts Anterodorsal Development Preliminary experiments had shown that injection of human bmp1 mRNA resulted in embryos that lacked dorsoanterior structures, including the notochord (Dale, 1997). To further investigate this phenotype we injected xbmp1 or xld mRNA into each blastomere at the 2-cell stage. As controls we injected synthetic mRNA for Dxbmp1, from which the C-terminal CUB and EGF repeats have been deleted (see Materials and Methods), or human collagenase (mmp1). Injected embryos developed normally until late gastrula stages, when a slight delay in blastopore closure was observed in embryos injected with xbmp1 or xld. By tailbud stages it was clear that most of these embryos had anterior defects (AD embryos in Table 1, see Fig. 5B), but the proctodeum was also frequently enlarged (EP embryos in Table 1, see Fig. 5D) and in some cases an additional taillike structure was observed (TD embryos in Table 1, see Fig. 5C). Since these tail-like structures rarely contained neural tissue or muscle, never contained notochord, and did not express Xpo (data not shown), a tail marker, it is unlikely that they represent true tails. Embryos were scored using the dorsoanterior index (DAI in Table 1) of Kao and Elinson (1988) in which a score of 5 represents a normal embryo and a score of 0 a fully ventralized embryo (see Materials and Methods). xbmp1 and xld injections resulted in embryos with a weakly ven-

FIG. 5. Overexpression of xbmp1 or xolloid results in anterior defects. (A) Control embryo injected with 2 ng of Dxbmp1 RNA into each dorsal blastomere at the 4-cell stage. (B) Embryo injected with 2 ng of xbmp1 RNA into each dorsal blastomere at the 4-cell stage exhibiting reduced head structures and a shorter anteroposterior axis. Note the continued presence of blood in the ventral blood island (Bl). (C) Embryo injected with 2 ng of xbmp1 RNA into each ventral blastomere at the 4-cell stage exhibiting a tail-like structure (T2) emerging from the proctodeum. Note the presence of melanocytes in the tail-like structure, although histological analysis showed that it contained no dorsal axial tissues. (D) Embryo injected with 2 ng of xld RNA into each dorsal blastomere at the 4cell stage exhibiting reduced head structures and a shorter anteroposterior axis. Note the continued presence of blood in the ventral blood island (Bl) and the enlarged proctodeum (P). (E) Embryo injected with 2 ng of xld RNA into each ventral blastomere at the 4-cell stage. Note the normal morphology apart from an abnormal accumulation of blood (BL) in the tail.

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

AID

DB 8840

/

6x39$$$$$1

02-25-98 12:09:21

dbal

152

Goodman et al.

FIG. 6. Dorsal axial defects in xbmp1-injected embryos. (A) Transverse section through the trunk of a control embryo injected with 2 ng of Dxbmp1 RNA into the dorsal blastomeres at the 4-cell stage. Note the normal arrangement of the notochord (No), somites (So), and neural tube (NT). (B) Transverse section through the trunk of an embryo injected with 2 ng of xbmp1 RNA into dorsal blastomeres at the 4-cell stage. Note the absence of a notochord and fusion of the somites (So) across the midline.

cated by normal expression of krox20 in whole mount in situ hybridizations (data not shown). In contrast, anterior defects were rarely observed in embryos injected with control RNAs and the mean DAI scores were 4.90 for Dxbmp1 and 4.83 for mmp1. When analyzed histologically the most severely truncated embryos were found to lack a notochord and as a consequence the somitic muscles had fused across the midline (Fig. 6B). The phenotypes we have described following the injection of xbmp1 and xld RNAs are very similar to those seen following the injection of low concentrations of xbmp4 RNA (Dale et al., 1992), and are suggestive of a partial ventralization of the mesoderm. If xbmp1 and xld ventralize dorsal mesoderm then localized injection of RNA into dorsal blastomeres should result in a stronger phenotype than injection into ventral blastomeres. This was confirmed by injecting either xbmp1 or xld mRNAs into two dorsal or two ventral blastomeres at the 4-cell stage. Although injection of xbmp1 RNA was less effective than in previous experiments, anterior defects were seen in 30% of embryos (DAI Å 4.53) following dorsal injections (Fig. 5B) but only in 6% of embryos (DAI Å 4.94) following ventral injections. Similarly, whereas dorsal injections of xld caused anterior defects in 68% of cases (DAI Å 3.68), ventral injections caused anterior defects in only 13% of cases (DAI Å 4.59) (Figs. 5D, E). In contrast, an enlarged proctodeum and tail-like structures were more frequent following ventral injections (Fig. 5C). Injection of mRNA for Dxbmp1 had little effect on anterior development irrespective of the site of injection (Table 2 and Fig. 5A). The results suggest that both xbmp1 and xld can ventralize dorsal mesoderm, resulting in anterior defects.

BMP1 and Xld Inhibit Dorsal Mesoderm in Animal Caps Treated with Activin We have previously shown that bmp4 blocks the formation of dorsal mesoderm in animal caps treated with activin (Dale et al., 1992; see also Jones et al., 1992; 1996), a potent inducer of mesoderm in this system. Since both xbmp1 and xld have a similar, albeit weaker, effect to bmp4 on whole embryos, we wished to know if they could also inhibit dorsal mesoderm formation in animal caps. To test this animal

caps were isolated at midblastula stages from embryos injected with either xbmp1, xld or control mRNAs and then incubated in the presence of 10 units of activin. These caps did not elongate in the absence of activin (Fig. 7A), and no mesoderm was detected upon histological examination (data not shown). In contrast, activin caused control caps to elongate (Fig. 7B), and histological analysis showed that 80-100% of them differentiated muscle (data not shown). Elongation did not occur in animal caps injected with either xbmp1 or xld mRNAs (Figs. 7C, D), and muscle was replaced by ventral mesoderm in all xld injected caps and in 79% of xbmp1 injected caps (data not shown). In a separate experiment, injected caps treated with activin were incubated until control embryos had reached stage 25 and assayed by RNase protection for the expression of musclespecific a-actin, a dorsal marker, and a-globin, a ventral marker (Fig. 8A). Animal caps from embryos injected with either xbmp1 or xld (tracks 4 and 5) had greatly reduced levels of a-actin when compared to controls (tracks 2, 3, and 6), while animal caps injected with xld exhibited a corresponding increase in a-globin (track 5). The results are consistent with xbmp1 and xld converting dorsal mesoderm into ventral mesoderm, as previously observed for bmp4 (Dale et al., 1992; Jones et al., 1992; 1996). One explanation for these results is that XBMP1 and Xld are activating TGF-b-related BMPs such as BMP2 or BMP4, which in turn ventralize the mesoderm. If this were the case we would expect inhibitors of BMP signaling pathways to block the ventralizing activity of XBMP1 and Xld. To test this we have coinjected xbmp1 and xld mRNAs with mRNA encoding a dominant-inhibitory receptor (tBR) for BMP2 and BMP4. This receptor lacks the cytoplasmic serine-threonine kinase domain and blocks the response of animal caps to both BMP2 and BMP4, but not activin (Suzuki et al., 1994). Animal caps were isolated from injected embryos at stage 8 and incubated in the presence of 10 units of activin. In contrast to animal caps injected with either xbmp1 or xld alone (Fig. 8B, tracks 2 and 4), animal caps coinjected with tbr both elongated (data not shown) and expressed a-actin (tracks 3 and 5). This result indicates that the ventralizing activity of both XBMP1 and Xld requires a functional BMP signaling pathway, consistent with a role

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

AID

DB 8840

/

6x39$$$$$1

02-25-98 12:09:21

dbal

153

BMPs in Xenopus Development

TABLE 2 Dorsal Injections Give More Extreme Anterior Defects Than Ventral Injections RNA

Blastomeres

AD (%)

DAI

TD (%)

EP (%)

GD (%)

N

None Dxbmp1 Dxbmp1 xbmp1 xbmp1 xld xld

None Dorsal Ventral Dorsal Ventral Dorsal Ventral

3 5 0 30 6 68 13

4.92 4.95 5.00 4.53 4.94 3.68 4.59

0 0 0 0 13 0 0

0 0 0 62 37 90 82

1 18 6 8 7 22 33

284 26 33 185 153 96 51

Note. The indicated RNAs were injected into either dorsal or ventral blastomeres at the four-cell stage. Both blastomeres were injected with 2 ng of RNA per blastomere; the resulting embryos were scored at stage 38. Abbreviations are the same as in Table 1. Embryos scored as TD and EP were also scored for anterior defects and are included in the DAI score.

for these metalloproteinases in regulating the activity of BMP2 and/or BMP4.

DISCUSSION In this paper, we describe the isolation and characterization of two Xenopus cDNAs encoding astacin-like metalloproteinases closely related to mammalian BMP1. Whereas the first of these encodes the short-splicing variant of Xenopus BMP1, the second encodes a novel metalloproteinase

FIG. 7. Inhibition of dorsal morphogenetic movements in activintreated animal caps. (A) Uninjected animal caps do not elongate in the absence of human activin A. An identical result was obtained for animal caps injected with 3 ng of either Dxbmp1 or mmp1 RNA. (B) Uninjected animal caps elongate when incubated in the presence of human activin A. An identical result was obtained for animal caps injected with 3 ng of either Dxbmp1 or mmp1 RNA and incubated in the presence human activin A. (C) Animal caps injected with 3 ng of xbmp1 RNA fail to elongate when incubated in the presence of human activin A. (D) Animal caps injected with 3 ng of xld RNA fail to elongate when incubated in the presence of human activin A. Embryos were injected at the 2-cell stage with the indicated RNA, 1.5 ng per blastomere, and animal caps isolated from midblastulae (stage 8). Animal caps were incubated in the absence (A) or presence (B–D) of human activin A until sibling embryos had reached stage 17.

we have called Xolloid (Xld). The domain structure of Xld is identical to that of Drosophila Tld (Shimell et al., 1991), both proteinases containing additional CUB and EGF repeats when compared to the short-splicing variant of BMP1. This domain structure is also shared with the Tld splicing variant of both mammalian and Xenopus BMP1 (Fukagawa et al., 1994; Takahara et al., 1994; Lin et al., 1997), a novel mouse gene called mTld-like (Takahara et al., 1996), Aplysia Tld/BMP1-like (Liu et al., 1997) and Drosophila Tld-related 1 (Nguyen et al., 1994; Finelli et al., 1995). A different arrangement of CUB and EGF repeats are found in astacin-like metalloproteinases from the sea urchin (BP10 and SpAN) (Lepage et al., 1992; Reynolds et al., 1992) and C. elegans (HCH-1) (Hishida et al., 1996). In addition, the Xenopus hatching enzyme (XHE) is an astacin-like metalloproteinase with two C-terminal CUB, but no EGF, repeats (Katagiri et al., 1997). Since CUB domains are thought to mediate protein-protein interactions (Bork and Beckmann, 1993), differences in the amino acid sequence and/or arrangement of these domains may target these metalloproteinases to different substrates. Evidence in favor of different substrate specificities has been provided by studies in Drosophila, where Tld and Tld-related 1 are not interchangeable (Nguyen et al., 1994; Finelli et al., 1995). Genetic studies in Drosophila suggest that the BMP1related metalloproteinase Tld enhances the activity of Dpp, a structural and functional homologue of both BMP2 and BMP4 (Shimell et al., 1991; Ferguson and Anderson, 1992). In this paper, we provide evidence that BMP1 and Xld may play a similar role in early Xenopus embryos. First, overexpression of BMP1 and Xld results in embryos that lack the most anterior structures of the head, and in the trunk fail to differentiate a notochord. A similar phenotype is also seen following injection of low concentrations of BMP4 mRNA (Dale et al., 1992; Dosch et al., 1997). Second, BMP1 and Xld ventralize animal caps exposed to activin, a member of the TGF-b family that induces dorsal mesoderm. Once again, similar results have been obtained for BMP4 (Dale et al., 1992; Jones et al., 1996). Finally, we have shown that the ventralizing activity of BMP1 and Xld, can be blocked by coexpressing a dominant-negative receptor for

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

AID

DB 8840

/

6x39$$$$$1

02-25-98 12:09:21

dbal

154

Goodman et al.

FIG. 8. XBMP1 and Xld ventralize the dorsal mesoderm in a BMPdependent manner. (A) RNase protection analysis showing that XBMP1 and Xolloid ventralize the dorsal mesoderm induced in animal caps by activin. Total RNA isolated from the same batch of animal caps was analyzed with all of the indicated probes: MS-Actin, a-actin; C-Actin, cytoskeletal actin; a-globin, a-globin; ODC, ornithine decarboxylase. Embryos were injected at the 2-cell stage with 3 ng of the indicated RNAs, 1.5 ng/blastomere. Animal caps were isolated from midblastulae (stage 8) and incubated until sibling embryos reached stage 25. (B) RNase protection analysis of activin-treated animal caps showing that a dominant-negative receptor for BMP2 and BMP4 (tBR) blocks the ventralizing activity of both XBMP1 and Xolloid. Total RNA was isolated from injected animal caps treated with activin and probed for muscle-specific actin (a-actin). Embryos were injected at the 2-cell stage with 1.5 ng of the indicated RNAs, 750 pg/blastomere. Animal caps were isolated from midblastulae (stage 8) and incubated until sibling embryos reached stage 25.

BMPs. Since this receptor blocks signaling by BMP2 and BMP4, but not activin (Graff et al., 1994; Suzuki et al., 1994), our results demonstrate that a functional BMP signal-

ing pathway is required for the ventralizing activity of these metalloproteinases. The simplest explanation is that overexpression of BMP1 and Xld increases the activity of Xenopus TGF-b-related BMPs during blastula and gastrula stages. These may include BMP2, BMP4, BMP7 and a BMP-related gene called antidorsalizing morphogenetic protein (ADMP) (Dale et al., 1992; Nishimatsu et al., 1992; Hawley et al., 1995; Hemmati-Brivanlou and Thomsen, 1995; Moos et al., 1995). In contrast to the results presented here, Lin et al. (1997) have provided evidence that Xenopus BMP1 dorsalizes ventral mesoderm. These authors injected mRNAs for either XBMP1 or the more abundantly expressed XTld splicing variant into both ventral blastomeres at the 4-cell stage, and subsequently isolated the ventral marginal zone (VMZ) from early gastrulae. Whereas XBMP1 had no effect on the differentiation of injected VMZs, XTld converted ventral mesoderm (e.g. blood) into dorsal mesoderm (muscle) and upregulated expression of the dorsal marginal zone markers goosecoid and chordin. These are results expected of a molecule that inhibits BMPs, rather than one that activates them. We have repeated these experiments using XBMP1 and Xld, and obtained no evidence that these metalloproteinases are capable of dorsalizing VMZs (data not shown). One explanation for these different results is that the XTld splicing variant used by Lin et al. (1997), which contains additional CUB and EGF repeats, has a different substrate specificity to the short-splicing variant of BMP1 used in our studies. Our results suggest that XBMP1 and Xld ventralize dorsal mesoderm by enhancing the activity of TGF-b-related BMPs, but how is this achieved? One explanation is that these metalloproteinases interact directly with the TGF-brelated BMPs themselves, as suggested by co-purification of BMP1 with BMP2 and BMP3 (Wozney et al., 1988). Like most members of the TGF-b superfamily BMPs are first synthesized as large inactive precursors from which the active C-terminal dimers are proteolytically cleaved. Perhaps the function of BMP1 and Xld is to serve as the processing enzyme responsible for this cleavage event, although the cleavage sites more closely resemble the consensus sequence (RX[K/R]R) for subtilisin-like serine proteinases (Barr, 1991). While cleavage sites for Xld have yet to be determined, the known cleavage sites for BMP1 do not match this consensus sequence (Kessler et al., 1986; Panchenko et al., 1996). Alternatively, these metalloproteinases could release TGF-b-related BMPs from latent complexes with their own proregions. Although such complexes have not been described for BMPs, they have been well characterized for TGF-b itself and proteinase treatment has been shown to release the active molecule (Lyons et al., 1988). However, XBMP1 and Xld do not enhance the ventralizing activity of coexpressed XBMP4 (data not shown), suggesting that these metalloproteinases do not activate BMP4 directly. A second explanation is that BMP1 and Xld degrade components of the extracellular matrix (ECM), thereby releasing bound TGF-b-related BMPs. ECM proteins are common substrates for metalloproteinases and astacin, the prototypi-

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

AID

DB 8840

/

6x39$$$$$1

02-25-98 12:09:21

dbal

155

BMPs in Xenopus Development

cal enzyme for BMP1, completely degrades the native triple helix of type I collagen (Sto¨cker and Zwilling, 1995). Moreover, we have evidence that SpAN, a sea urchin BMP1related metalloproteinase (Reynolds et al., 1992), degrades fibronectin when expressed in early Xenopus embryos, although this is a property not shared by either XBMP1 or Xld (Wardle and Dale, unpublished). At present we do not know if Xld acts on the ECM, but BMP1 is responsible for cleaving the C-terminal peptides from procollagens I, II and III (Kessler et al., 1996; Li et al., 1996). BMP1 also correctly activates prolysyl oxidase (Panchenko et al., 1996), an enzyme involved in intermolecular cross-linking in collagen fibres, and a loss-of-function mutation in the mouse bmp1 gene disrupts the assembly of collagen fibres during embryogenesis (Suzuki et al., 1996). These results suggest that BMP1 is responsible for the correct assembly of collagen fibres, rather than degradation. However, the presence of multiple BMP1-related genes (Takahara et al., 1996), which might compensate for the loss of BMP1, means that other roles cannot be excluded. In Xenopus fibrillar collagens are not expressed until neurula stages (Su et al., 1991), so it seems likely that maternal BMP1, and Xld, have other substrates. Some of these may be components of the ECM that sequester BMPs. A third explanation is that BMP1 and Xld release TGF-brelated BMPs sequestered by inhibitory binding proteins such as chordin, noggin and follistatin (Piccolo et al., 1996; Zimmerman et al., 1996; Fainsod et al., 1997). We have recently shown that Xld can cleave chordin, but not noggin, in vitro, and provided evidence that this releases active BMPs (Piccolo et al., 1997). Xld also blocks the dorsalizing activity of chordin in vivo, but not that of noggin and follistatin, suggesting that specific cleavage of chordin may also be its in vivo role. Consistent with this, xld is expressed in the inner epidermal layer of the developing tailbud, in close proximity to the chordoneural hinge which expresses chordin (Sasai et al., 1995). At gastrula stages chordin is localized to the organizer (Sasai et al., 1995), while xld is expressed by all cells and may even be more abundant in the organizer. This is different from the situation in Drosophila where Tld is coexpressed with Dpp at the cellular blastoderm stage, and not in cells expressing the chordin homologue Sog (Shimell et al., 1991; Francois et al., 1994). These differences may reflect the fact that chordin directly binds both BMP2 and BMP4 (Piccolo et al., 1996), and the fact that bmp2 is expressed in the organizer (Hemmati-Brivanlou and Thomsen, 1995). In addition, the organizer also expresses both bmp7 and admp (Hawley et al., 1995; Moos et al., 1995), which may also interact with chordin. Although the function of these ‘‘ventralizing’’ genes in the organizer has yet to be elucidated, a requirement for Xld in regulating inhibitory interactions with chordin may explain expression of xld in the organizer. At present we do not know if XBMP1 also cleaves chordin, or whether it cleaves other inhibitory binding proteins such as noggin or follistatin. However, since XBMP1 and Xld do not act synergistically in coinjection experiments, the DAI score of coinjected embryos is similar to

that of singularly injected embryos (data not shown), it is possible that they act on the same substrates. Recently it has become clear that dorsoventral patterning in the early embryo of both Drosophila and Xenopus, is mediated by a conserved system of extracellular signals (reviewed by De Robertis and Sasai, 1996). Dorsoventral fates are specified by the graded activity of two closely related members of the TGF-b family, Dpp (Drosophila) and BMP4 (Xenopus), gradients that may be established by diffusion of the inhibitory binding proteins Sog (Drosophila) and chordin (Xenopus). Our results suggest that an additional layer of regulation may also be conserved between Drosophila and Xenopus, in that BMP1-related metalloproteinases may play an essential role in establishing the activity gradient of Dpp/BMP4.

ACKNOWLEDGMENTS We thank Elizabeth Wang for providing the human BMP1 cDNA, Dylan Edwards for the human collagenase cDNA, Naoto Ueno for the tBR cDNA, John Shuttleworth for the oocyte cDNA library, Jim Smith for human activin A, and Kate Nobes for the 9E10 antibody. This work was funded by the British Medical Research Council and the Wellcome Trust. F.C.W. is a Wellcome Prize student.

REFERENCES Barr, P. J. (1991). Mammalian subtilisins: The long-sought dibasic processing endoproteases. Cell 66, 1–3 Bode, W., Gomis-Ru¨th, F. X., Huber, R., Zwilling, R., and Sto¨cker, W. (1992). Structure of astacin and implications for activation of astacins and zinc-ligation of collagenase. Nature 358, 164–167. Bork, P., and Beckman, G. (1993). The CUB domain. A widespread molecule in developmentally regulated proteins. J. Mol. Biol. 231, 539–545. Dale, L. (1997). Bone morphogenetic proteins in amphibian development. In ‘‘The Astacins: Structure and Function of a New Protein Family’’ (R. Zwilling and W. Sto¨cker, Eds.), Pp. 235–246. Dale, L., Howes, G., Price, B. M. J., and Smith, J. C. (1992). Bone morphogenetic protein 4: A ventralising factor in early Xenopus development. Development 115, 573–585. De Robertis, E. M., and Sasai, Y. (1996). A common plan for dorsoventral patterning in Bilateria. Nature 380, 37–40. Dosch, R., Gawantka, V., Delius, H., Blumenstock, C., and Niehrs, C. (1997). Bmp-4 acts as a morphogen in dorsoventral mesoderm patterning in Xenopus. Development 124, 2325–2334. Fainsod, A., Steinbeisser, H., and De Robertis, E. M. (1994). On the function of BMP-4 in patterning the marginal zone of the Xenopus embryo. EMBO J. 13, 5015–5025. Fainsod, A., Deissler, K., Yelin, R., Marom, K., Epstein, M., Pillemer, G., Steinbeisser, H., and Blum, M. (1997). The dorsalizing and neural inducing gene follistatin is an antagonist of BMP-4. Mech. Dev. 63, 39–50. Ferguson, E. L., and Anderson, K. V. (1992). Localized enhancement and repression of the activity of the TGF-b family member, decapentaplegic, is necessary for dorsal–ventral pattern formation in the Drosophila embryo. Development 114, 583–597. Finelli, A. L., Xie, T., Bossie, C. A., Blackman, R. K., and Padgett,

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

AID

DB 8840

/

6x39$$$$$1

02-25-98 12:09:21

dbal

156

Goodman et al.

R. W. (1995). The tolkin gene is a tolloid/BMP-1 homologue that is essential for Drosophila development. Genetics 141, 271–281. Francois, V., Solloway, M., O’Neill, J. W., Emery, J., and Bier, E. (1994). Dorso-ventral patterning of the Drosophila embryo depends on a putative negative growth factor encoded by the short gastrulation gene. Genes Dev. 8, 2602–2616. Fukagawa, M., Suzuki, N., Hogan, B. L. M., and Jones, C. M. (1994). Embryonic expression of mouse bone morphogenetic protein-1 (BMP-1), which is related to the Drosophila dorsoventral gene tolloid and encodes a putative astacin metalloendopeptidase. Dev. Biol. 163, 175–183. Graff, J. M., Thies, R. S., Song, J. J., Celeste, A. J., and Melton, D. A. (1994). Studies with a Xenopus BMP receptor suggest that ventral mesoderm-inducing signals override dorsal signals in vivo. Cell 79, 169–179. Harland, R. M. (1991). In situ hybridization: An improved whole mount method for Xenopus embryos. Methods Cell Biol. 36, 685–695. Hawley, S. H. B., Wunnenberg-Stapleton, K., Hashimoto, C., Laurent, M. N., Watabe, T., Blumberg, B. W., and Cho, K. W. Y. (1995). Disruption of BMP signals in embryonic Xenopus ectoderm leads to direct neural induction. Genes Dev. 9, 2923–2935. Hemmati-Brivanlou, A., and Melton, D. A. (1997). Vertebrate embryonic cells will become nerve cells unless told otherwise. Cell 88, 13–17. Hemmati-Brivanlou, A., and Thomsen, G. (1995). Ventral mesodermal patterning in Xenopus embryos: Expression patterns and activities of BMP-2 and BMP-4. Dev. Genet. 17, 78–89. Hishida, R., Ishihara, T., Kondo, K., and Katsura, I. (1996). hch-1, a gene required for normal hatching and normal migration of a neuroblast in C. elegans, encodes a protein related to TOLLOID and BMP-1. EMBO J. 15, 4111–4122. Holley, S. A., Jackson, P. D., Sasai, Y., Lu, B., De Robertis, E., Hoffmann, F. M., and Ferguson, E. L. (1996). A conserved system for dorsal–ventral patterning in insects and vertebrates involving sog and chordin. Nature 376, 249–253. Isaacs, H. V., Tannahill, D., and Slack, J. M. W. (1992). Expression of a novel FGF in the Xenopus embryo. A candidate inducing factor for mesoderm formation and anteroposterior specification. Development 114, 711–720. Jones, C. M., Dale, L., Hogan, B. L. M., Wright, C. V. E., and Smith, J. C. (1996). Bone morphogenetic protein-4 (BMP-4) acts during gastrula stages to cause ventralization of Xenopus embryos. Development 122, 1545–1554. Jones, C. M., Lyons, K. M., Lapan, P. M., Wright, C. V. E., and Hogan, B. L. M. (1992). DVR-4 (bone morphogenetic protein-4) as a posterior-ventralizing factor in Xenopus mesoderm induction. Development 115, 639–647. Kao, K. R., and Elinson, R. P. (1988). The entire mesodermal mantle behaves as Spemann’s organizer in dorso-anterior enhanced Xenopus laevis embryos. Dev. Biol. 127, 64–77. Katagiri, C., Maeda, R., Yamashika, C., Mita, K., Sargent, T. D., and Yasumasu, S. (1997). Molecular cloning of Xenopus hatching enzyme and its specific expression in hatching gland cells. Int. J. Dev. Biol. 41, 19–25. Kessler, E., Adar, R., Goldberg, B., and Niece, R. (1986). Partial purification and characterization of a procollagen C-proteinase from culture medium of mouse fibroblasts. Collagen Rel. Res. 6, 249–266. Kessler, E., Takahara, K., Biniaminov, L., Brusel, M., and Greenspan, D. S. (1996). Bone morphogenetic protein-1: The type 1 procollagen C-proteinase. Science 271, 360–362. Lepage, T., Ghiglione, C., and Gache, C. (1992). Spatial and tempo-

ral expression pattern during sea urchin embryogenesis of a gene coding for a protease homologous to the human protein BMP-1 and to the product of the Drosophila dorsal–ventral patterning gene tolloid. Development 114, 147–164. Li, S.-W., Sieron, A. L., Fertala, A., Hojima, Y., Arnold, W. V., and Prockop, D. J. (1996). The C-proteinase that processes procollagens to fibrillar collagens is identical to the protein previously identified as bone morphogenetic protein-1. Proc. Natl. Acad. Sci. USA 93, 5127–5130. Lin, J. J., Maeda, R., Ong, R. C., Kim, J., Lee, L. M., Kung, H. F., and Maeno, M. (1997). XBMP-1B (Xtld), a Xenopus homolog of the dorso-ventral polarity gene in Drosophila, modifies tissue phenotypes of ventral explants. Develop. Growth Diff. 39, 43–51. Liu, Q.-R., Hattar, S., Endo, S., MacPhee, K., Zhang, H., Cleary, L. J., Byrne, J. H., and Eskin, A. (1997). A developmental gene (Tolloid/BMP-1) is regulated in Aplysia neurons by treatments that induce long-term sensitization. J. Neurosci. 17, 755–764. Lyons, R., Keski-Oja, J., and Moses, H. (1988). Proteolytic activations of latent transforming growth factor-b from fibroblast conditioned medium. J. Cell Biol. 106, 1659–1665. Maeno, M., Xue, Y., Wood, T. I., Ong, R. C., and Kung, H. F. (1993). Cloning and expression of cDNA encoding Xenopus laevis bone morphogenetic protein-1 during embryonic development. Gene 134, 257–261. Maeno, M., Ong, R. C., Suzuki, A., Ueno, N., and Kung, H. F. (1994). A truncated bone morphogenetic protein 4 receptor alters the fate of ventral mesoderm to dorsal mesoderm: Roles of animal pole tissue in the development of ventral mesoderm. Proc. Natl. Acad. Sci. USA 91, 10260–10264. Matthews, G., and Colman, A. (1991). A highly efficient, cell-free translation/translocation system prepared from Xenopus eggs. Nucleic Acids Res. 19, 6405–6412. Mohun, T. J., Garrett, N., Stutz, F., and Spohr, G. (1988). A third striated muscle actin gene is expressed during early development in the amphibian Xenopus laevis. J. Mol. Biol. 202, 67–76. Moos, M., Wang, S., and Krinks, M. (1995). Anti-dorsalizing morphogenetic protein is a novel TGF-b homolog expressed in the Spemann organizer. Development 121, 4293–4301. Nguyen, T., Jamal, J., Shimell, M. J., Arora, K., and O’Connor, M. B. (1994). Characterization of tolloid-related-1: A BMP-1-like product that is required during larval and pupal stages of Drosophila development. Dev. Biol. 166, 569–586. Nishimatsu, S., Suzuki, A., Shoda, A., Murakami, K., and Ueno, N. (1992). Genes for bone morphogenetic proteins are differentially transcribed in early amphibian embryos. Biochem. Biophys. Res. Commun. 186, 1487–1495. Padgett, R. W., Wozney, J., and Gelbert, W. M. (1993). Human BMP sequences can confer normal dorsal–ventral patterning in the Drosophila embryo. Proc. Natl. Acad. Sci. USA 90, 2905–2909. Panchenko, M. V., Stetler-Stevenson, W. G., Trubetskoy, O. V., Gacheru, S. N., and Kagan, H. M. (1996). Metalloproteinase activity secreted by fibrogenic cells in the processing of prolysyl oxidase. J. Biol. Chem. 271, 7113–7119. Piccolo, S., Agius, E., Lu, B., Goodman, S., Dale, L., and De Robertis, E. M. (1997). Cleavage of chordin by the Xolloid metalloprotease suggests a role for proteolytic processing in the regulation of Spemann organizer activity. Cell 91, 407–416. Piccolo, S., Sasai, Y., Lu, B., and De Robertis, E. M., (1996). Dorsoventral patterning in Xenopus: Inhibition of ventral signals by direct binding of chordin to BMP-4. Cell 86, 589–598. Reynolds, S. D., Angerer, L. M., Palis, J., Nasir, A., and Angerer, R. C. (1992). Early mRNAs, spatially restricted along the animal– vegetal axis of sea urchin embryos, include one encoding a protein related to tolloid and BMP-1. Development 114, 769–786.

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

AID

DB 8840

/

6x39$$$$$1

02-25-98 12:09:21

dbal

157

BMPs in Xenopus Development

Sampath, T. K., Rashka, K. E., Doctor, J. S., Tucker, R. F., and Hoffman, F. M. (1993). Drosophila transforming growth factor b superfamily proteins induce endochondral bone formation in mammals. Proc. Natl. Acad. Sci. USA 90, 6004–6008. Schmidt, J. E., Suzuki, A., Ueno, N., and Kimelman, D. (1995). Localized BMP-4 mediates dorso-ventral patterning in the early Xenopus embryo. Dev. Biol. 169, 37–50. Shimell, M. J., Ferguson, E. L., Childs, S. R., and O’Connor, M. B. (1991). The Drosophila dorsal–ventral patterning gene tolloid is related to human bone morphogenetic protein 1. Cell 67, 469– 481. Shuttleworth, J., Godfrey, R., and Colman, A. (1988). p40MO15 , a cdc2-related protein kinase involved in negative regulation of meiotic maturation of Xenopus oocytes. EMBO J. 9, 3233–3240. Sto¨cker, W., and Zwilling, R. (1995). Astacin. Methods Enzymol. 248, 305–325. Su, M. W., Suzuki, H. R., Bieker, J. J., Solursh, M., and Ramirez, F. (1991). Expression of two nonallelic type II procollagen genes during Xenopus laevis embryogenesis is characterized by stagespecific production of alternatively spliced transcripts. J. Cell Biol. 115, 565–575. Suzuki, A., Thies, R. S., Yamaji, N., Song, J. J., Wozney, J. M., Murakami, K., and Ueno, N. (1994). A truncated BMP receptor affects dorso-ventral patterning in the early Xenopus embryo. Proc. Natl. Acad. Sci. USA 91, 10255–10259. Suzuki, N., Labosky, P. A., Furata, Y., Hargett, L., Dunn, R., Fogo, A. B., Takahara, K., Peters, D. M. P., Greenspan, D. S., and Hogan, B. L. M. (1996). Failure of ventral body wall closure in mouse embryos lacking procollagen C-proteinase encoded by Bmp1, a mammalian gene related to Drosophila tolloid. Development 122, 3587–3595. Takahara, K., Lyons, G. E., and Greenspan, D. S. (1994). Bone mor-

phogenetic protein-1 and a mammalian Tolloid homologue (mTld) are encoded by alternatively spliced transcripts which are differentially expressed in some tissues. J. Biol. Chem. 269, 32572–32578. Takahara, K., Brevard, R., Hoffman, G. G., Suzuki, N., and Greenspan, D. S. (1996). Characterisation of a novel gene product of (mammalian tolloid-like) with high sequence similarity to mammalian tolloid/bone morphogenetic protein-1. Genomics 34, 157–165. Tucker, A. M., and Slack, J. M. W. (1995). Tail bud determination in the vertebrate embryo. Curr. Biol. 5, 807–813. Thomsen, G. H. (1997). Antagonism within and around the organizer: BMP inhibitors in vertebrate body patterning. Trends Genet. 13, 209–211. Van Wart, H. E., and Birkedal-Hansen, H. (1990). The cysteine switch: A principle of regulation of metalloproteinase activity with potential applicability to the entire matrix metalloproteinase gene family. Proc. Natl. Acad. Sci. USA 87, 5578–5582. Walmsley, M. E., Guille, M. J., Bertwistle, D., Smith, J. C., Pizzey, J. A., and Patient, R. K. (1994). Negative control of Xenopus GATA-2 by activin and noggin with eventual expression in precursors of the ventral blood islands. Development 120, 2519– 2529. Wozney, J. M., Rosen, V., Celeste, A. J., Mitsock, L. M., Whitters, M. J., Kriz, R. W., Hewick, R. M., and Wang, E. A. (1988). Novel regulators of bone formation: Molecular clones and activities. Science 242, 1528–1534. Zimmerman, L. B., De Jesu´s-Escobar, J. M., and Harland, R. M. (1996). The Spemann organizer signal noggin binds and inactivates bone morphogenetic protein 4. Cell 86, 599–606. Received for publication October 10, 1997 Accepted December 8, 1997

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

AID

DB 8840

/

6x39$$$$$1

02-25-98 12:09:21

dbal