Failure of ventral closure and axial rotation in embryos ... - Development

4863

Development 125, 4863-4876 (1998) Printed in Great Britain © The Company of Biologists Limited 1998 DEV4090

Failure of ventral closure and axial rotation in embryos lacking the proprotein convertase Furin Anton J. M. Roebroek1,*, Lieve Umans2, Ilse G. L. Pauli1, Elizabeth J. Robertson3, Fred van Leuven2, Wim J. M. Van de Ven1 and Daniel B. Constam3,* 1Laboratory

for Molecular Oncology, 2Experimental Genetics Group, Center for Human Genetics, University of Leuven and Flanders Interuniversity Institute for Biotechnology (VIB), Herestraat 49, B-3000 Leuven, Belgium 3Department of Molecular and Cellular Biology, Harvard University, 16 Divinity Ave, Cambridge, MA 02138, USA *Authors for correspondence (e-mail: [email protected] and [email protected])

Accepted 2 October; published on WWW 12 November 1998

SUMMARY We have examined the role of Furin in postimplantationstage mouse embryos by analyzing both the expression pattern of fur mRNA and the developmental consequences of a loss-of-function mutation at the fur locus. At early stages (day 7.5), fur mRNA is abundant in extraembryonic endoderm and mesoderm, anterior visceral endoderm, and in precardiac mesoderm. 1 day later fur is expressed throughout the heart tube and in the lateral plate mesoderm, notochordal plate and definitive gut endoderm. Embryos lacking Furin die between days 10.5 and 11.5, presumably due to hemodynamic insufficiency associated with severe ventral closure defects and the failure of the heart tube to fuse and undergo looping morphogenesis. Morphogenesis of the yolk sac vasculature is also

INTRODUCTION Specific proteolytic modifications are a powerful means by which to regulate the biological activities of proteins. Also, during embryonic development, proteases control the activation and turnover of many important key regulatory molecules. For example, in Drosophila the syncytial blastoderm is patterned along the dorsoventral axis by a gradient of mature Spätzle protein, which is established by the ventrally localized proteolytic activity of Easter (Chasan and Anderson, 1989; Stein and Nusslein-Volhard, 1992). In the ventral blastoderm, Spätzle signaling leads to repression of several target genes, including tolloid (Kirov et al., 1993). Thus, localized expression of tolloid in the dorsal blastoderm locally enhances the dorsalizing activity of DPP, thereby allowing specification of dorsalmost cell fates (Shimell et al., 1991). Mechanistic insights into the function of the metalloproteases encoded by tolloid and its Xenopus homologue xolloid have come from studies demonstrating that they selectively degrade the DPP/BMP antagonists SOG and Chordin, respectively (Marqués et al., 1997; Piccolo et al., 1997). Similarly, in both arthropods and vertebrates, a key regulatory step in the control of the Delta-Notch pathway

abnormal, although blood islands and endothelial precursors form. Analysis of cardiac and endodermal marker genes shows that while both myocardial precursors and definitive endoderm cells are specified, their numbers and migratory properties are compromised. Notably, mutant embryos fail to undergo axial rotation, even though Nodal and eHand, two molecular markers of left-right asymmetry, are appropriately expressed. Overall, the present data identify Furin as an important activator of signals responsible for ventral closure and embryonic turning. Key words: Development, Turning, Heart morphogenesis, Endoderm, TGFβ, Processing, Mouse

responsible for cell-fate selection is proteolysis of the Notch receptor (reviewed in Greenwald, 1998). Recent data show that the cytoplasmic domain of Notch is cleaved and, upon ligand binding of the extracellular domain, translocates to the nucleus (Kopan et al., 1996; Schroeter et al., 1998). Genetic and biochemical evidence have implicated the metalloprotease encoded by kuz as a potential candidate for Notch processing and turnover (reviewed in Greenwald, 1998); however, in vitro Notch can also be cleaved by Furin or related proteases (Logeat et al., 1998). Furin is the mammalian prototype of a family of serine proteases that have catalytic domains resembling the bacterial protease subtilisin (Roebroek et al., 1986; for review see Van de Ven et al., 1993). Three members of this protease gene family, including PC1/3, PC2 and PC4, are expressed exclusively in endocrine tissues or in testis, respectively. Loss of function mutations have shown that PC1/3 and PC2 are both required for generating physiological amounts of mature insulin (Furuta et al., 1997; Jackson et al., 1997). The lack of PC2 also impairs the production of mature glucagon (Furuta et al., 1997). By contrast, PC4 is uniquely required for normal sperm fertility (Mbikay et al., 1997). These findings demonstrate that the more ubiquitous, closely related activities of Furin, PACE4, PC5/6 or

4864 A. J. M. Roebroek and others PC7 cannot effectively compensate for the loss of PC1/3, PC2 or PC4. The physiological roles of Furin, PACE4, PC5/6 and PC7 remain poorly understood since their normal in vivo substrates are unknown, as is the extent to which their substrate specificities overlap under physiological conditions. Coexpression experiments and in vitro studies have implicated Furin as a major player in the processing of proproteins that are specifically cleaved after R-X-X-R sequences (Molloy et al., 1992), such as insulin-like growth factor I (Duguay et al., 1997), insulin receptor (Bravo et al., 1994), hepatocyte growth factor receptor (Komada et al., 1993), α-integrins (Lehmann et al., 1996) and matrix metalloproteinases (Pei and Weiss, 1995). Similarly, Furin has been shown to efficiently process the precursor forms of TGFβ-related molecules, including TGFβ1 (Dubois et al., 1995), activin A (Roebroek et al., 1993), Mullerian inhibiting substance (Nachtigal and Ingraham, 1996), BMP4 (Cui et al., 1998; D. B. C. and E. J. R., unpublished) and Nodal (D. B. C. and E. J. R., unpublished). Under physiological conditions, the partially overlapping activities of PACE4, PC5/6 and PC7 may also contribute to the processing of at least some of these precursor proteins. In keeping with this suggestion, a previous analysis of their expression patterns during embryonic development led us to propose that the overlapping activities of Furin, PACE4, PC5(6) and PC7 may locally control the availability of mature BMP ligands (Constam et al., 1996). Unlike PACE4 and PC5/6, Furin mRNA has not been documented to be locally restricted to specific tissues in the mouse embryo. In rat embryos, however, fur mRNA is expressed specifically in the extraembryonic endoderm and mesoderm during early primitive streak stages (Zheng et al., 1994). During late somite stages, localized expression of fur is also seen in the cardiovascular system (Zheng et al., 1994), raising the possibility that Furin is specifically required in these tissues during early embryonic development. Here we have performed a detailed analysis of the expression pattern of fur in the early mouse embryo using mRNA whole-mount in situ hybridization. To explore its potential function during embryogenesis, we have also generated a loss-of-function mutation at the fur locus. Furindeficient embryos die between 10.5 and 11.5 dpc, probably due to hemodynamic insufficiency resulting from severe ventral closure defects. In mutant embryos, the cardiogenic mesoderm either fails to fuse at the ventral midline, resulting in cardia bifida, or forms a single linear tube that fails to undergo looping morphogenesis. Interestingly, Furindeficient embryos also fail to turn, and their allantois does not attach to the chorion. To further characterize the turning and ventral closure defects, we analyzed the expression of a nodallacZ reporter allele, and a panel of cardiac- and endoderm-specific molecular markers. Based on our findings we propose that Furin activities promote the migration and/or proliferation of definitive endoderm, and also contribute to the activation of cardiogenic signals.

MATERIALS AND METHODS Whole-mount in situ hybridization and histological analyis Whole-mount in situ hybridization using digoxygenin-labelled RNA probes was performed as described (Wilkinson, 1992). The riboprobes used were as described: shh (Echelard et al., 1993), HNF3β (Sasaki

and Hogan, 1993), MLC2V (Lyons et al., 1995a) and -2A (Kubalak et al., 1994), eHand (Biben and Harvey, 1997), GATA4 (Arceci et al., 1993) and Flk-1 (Yamaguchi et al., 1993). In addition, a 130 bp fragment of the α-cardiac actin gene (provided by R. Harvey) was transcribed using T3 RNA polymerase to synthesize an antisense RNA probe. The Furin riboprobe comprised nucleotides 2014-2774, corresponding to the 3′ end of the Furin coding sequence (gb:L26489). The Nkx2.5 probe comprised nucleotides 820-1488 of murine Nkx2.5/Csx (gb:X75415; kindly provided by S. Izumo). Following photography, embryos were dehydrated in a series of graded ethanol and in xylene, embedded in paraffin wax and sectioned at 10 µm. For histological analysis, embryos were embedded as described, sectioned at 7 µm, and stained with Haematoxylin and Eosin. Derivation of mutant mice A cosmid containing the 5′ region of the mouse fur gene was isolated from a 129/Sv genomic cosmid library by colony hybridization using the mouse fur cDNA (Creemers et al., 1992) as a molecular probe. A 6 kb NotI fragment, comprising exon 1A to intron 6 sequences, was subcloned into pGEM-5Zf(+). A 1.8 kb BglII fragment encoding the hygromycin B phosphotransferase gene, fused to the phosphoglycerate kinase promoter, was cloned into a KpnI site in exon 4 via KpnI-BglII adaptors. The adaptor sequence GTACTAGCTTTCGTTGCCGGATC contained a stop codon (underlined) in-frame with the fur coding sequence in exon 4, and disrupts the KpnI site present in exon 4. The insert of the resulting plasmid was excised with NotI and electroporated into E14 embryonic stem cells (Doetschman et al., 1987). Hygromycin B-resistant colonies were analyzed by Southern blotting using KpnI-digested genomic DNA. Probes were obtained by PCR amplification of nucleotides 298-763 (probe L), 764-1231 (probe R1) and 970-1231 (probe R2) of the fur cDNA, corresponding to exon 2 to exon 4, exon 5 to exon 8, or exon 7 to exon 8 sequences, respectively. An additional probe used to confirm correct targeting comprised a 529 bp fragment of the hygromycin B phosphotransferase gene amplified by PCR using primers 5′-CAGCGAGAGCCTGACCTATTGC-3′ and 5′-CGATCCTGCAAGCTCCGGATG-3′ (probe H). Correctly targeted ES cell clones were injected into C57Bl/6 using standard procedures. The resulting chimeras were mated to C57Bl/6 mice, and transmission of the mutant fur allele was confirmed by Southern blotting and PCR analysis of tail DNA samples derived from F1 progeny (Fig. 2). Heterozygous offspring carrying the mutant allele were backcrossed to C57Bl/6 mice for three generations to rederive the pathogen-free progeny used for the subsequent analysis. PCR genotyping was carried out using the hygromycin-specific primers described above to detect the mutant allele. Primers 5′-CGGTGACTATTACCACTTCTGGCACAGAGC-3′ and 5′-AAACAGAAGAAGCCAGGGTGAGCCTCATCC-3′ were used to amplify a wild-typespecific 373 bp fragment extending from exon 3 into the fourth intron of the fur gene. No product was amplified from the targeted allele using the second primer pair, presumably because the hygromycin cassette interfered with the amplification under the conditions used. To analyze fur mRNA expression in mutant tissues, total RNA from mouse liver was isolated using TRIzol Reagent (Life Technologies) and analyzed by northern blot hybridization as described previously (Roebroek et al., 1993) using probes L1, R2 or H, respectively. Equal loading of RNA in each lane was confirmed by rehybridizing the membranes with a hamster α-actin cDNA probe. To analyze fur expression in mutant embryos, total RNA isolated from 9.5 dpc embryos was subjected to RT-PCR using the Perkin Elmer Geneamp RNA PCR kit. The first strand cDNA synthesis was primed with an oligonucleotide corresponding in sequence to positions 1212-1231 of the fur cDNA complementary to exon 8. In the subsequent PCR reaction, a fragment comprising exon 3 to exon 6 sequences was amplified using primers 5′-CGGTGACTATTACCACTTCTGGCACAGAGC-3′ and 5′-TGTCATTCATCTGTGTGTACCGAGGCTGTG-3′.

Role of Furin in ventral morphogenesis 4865 β-galactosidase staining of embryos carrying the nodallacZ allele The nodallacZ allele was maintained on a 129/Sv background, and animals were genotyped as described (Collignon et al., 1996). For analysis of β-galactosidase expression in a fur null background, heterozygous fur+/− males were mated to nodallacZ/+ females. The resulting doubly heterozygous offspring were crossed to fur+/− partners. Embryos were dissected between 7.5 and 8.5 dpc and genotyped using yolk sac DNA. Staining of β-galactosidase activity was as described (Hogan et al., 1994).

RESULTS Expression of fur mRNA between embryonic days 7.5 and 8.5 Using whole-mount in situ hybridization, we examined the expression of fur in mouse embryos between 7.5 and 8.5 dpc. The probe was derived from coding sequences 3′ to the catalytic domain to exclude regions that are most conserved among related proprotein convertases. At the earliest stage examined (7.5 dpc), fur transcripts are detected in extraembryonic tissues, and in the proximal region of the embryo (Fig. 1A). By late gastrulation, a diffuse signal is associated with the node (Fig. 1A). In the extraembryonic tissues, fur mRNA is localized to both the mesoderm and primitive endoderm, but is excluded from the most proximal primitive endoderm contacting the extraembryonic ectoderm (Fig. 1B). At more distal levels, fur mRNA is strongly expressed in the most anterior visceral endoderm, and in the underlying precardiac mesoderm, as well as in ectodermal cells in the antero-proximal region of the epiblast (Fig. 1C). By headfold stage, fur is strongly expressed in the allantois, and in the anterior intestinal portal (AIP) both within mesodermal and endodermal cells subjacent to the headfolds (Fig. 1D). In the node region, high levels of fur mRNA are apparent in the periphery of the ventral notochordal plate (Fig. 1E). From the 4-6 somite stage onward (8.5 dpc), fur is strongly expressed in the primitive heart, with particularly high levels apparent in the inflow tract and in the AIP, where ventral closure is proceeding in a caudal direction. High expression levels persist in the allantois (Fig. 1F). Furin mRNA is also detected throughout the foregut endoderm (Fig. 1G,H), and in the lateral plate mesoderm of the trunk (Fig. 1I). In the tailbud, Fig. 1. Expression of fur mRNA in 7.5-8.5 dpc mouse embryos analyzed by whole-mount in situ hybridization. (A) Whole-mount view of a 7.5 dpc embryo. (B) Transverse sections through the extraembryonic and (C) embryonic regions of a similar 7.5 dpc embryo. Dashed lines in A indicate the level of sectioning in B and C, respectively. (D) Frontal-lateral and posterior (E) whole-mount view of a headfold stage embryo (8.0 dpc). (F) Whole-mount view and transverse sections through (G) the heart, (H) the AIP, (I) the trunk, and (J,K) through the tailbud of an 8.5 dpc embryo which has formed 3-4 pairs of somites. aip, anterior intestinal portal; al, allantois; ave, anterior visceral endoderm; cf, chorionic fold; cip, caudal intestinal portal; cp, notochordal plate; cv, cardinal veins; e and ee, embryonic and extraembryonic region; em, extraembryonic mesoderm; en, endocardium; ep, epiblast; fg, foregut; h, heart; hg, hindgut; lp, lateral plate; m, mesoderm; mc, prospective myocardium; n, node; nf, neural fold; np, neural plate; so, somite; vys, visceral yolk sac. In A-D and F, anterior is to the left; in G-K, the left side of the embryo is to the right.

fur is most strongly expressed in the notochordal plate (Fig. 1J), and more caudally in the hindgut and caudal intestinal portal (cip; Fig. 1K).

4866 A. J. M. Roebroek and others Targeted inactivation of fur by homologous recombination The targeting construct used to inactivate the fur locus consisted of a 6 kb NotI fragment comprising part of exon 1A through exon 6. A PGKhygro cassette was inserted into exon 4 together with adapter sequences which abolish a resident KpnI site and provide an in-frame stop codon to disrupt the fur coding sequence (Fig. 2A). Following electroporation into E14 ES cells, individual drug-resistant colonies were analyzed by Southern blotting using a cDNA probe corresponding to exons 7-8 (probe R2). Of 44 drug-resistant colonies analyzed, seven correctly targeted ES cell lines were recovered. Two out of

seven correctly targeted ES cell lines (ROE-1 and -2, Fig. 2B) gave rise to germline chimeras. Heterozygous (C57Bl/6 ×129) F1 offspring derived from line ROE-1 were intercrossed, and viable offspring were genotyped. Of 231 progeny, 78 (34%) were wild type, 153 (66%) were heterozygous, but no homozygous mutants were recovered. Similarly, of 190 newborn F2 progeny derived from ROE-2 intercrosses, none were homozygous mutant. To determine the timing of developmental arrest, embryos from heterozygous intercross matings were genotyped by PCR. No homozyogous mutant embryos were recovered from 11.5 dpc onwards, although at this stage, 17 (27%) out of 63 decidua contained remnants of

Fig. 2. Targeted inactivation of the fur gene by homologous recombination in ES cells. (A) Diagram showing the relevant regions of the fur locus and targeting vector. Due to the insertion of a hygromycin resistance gene (hygB) and adaptor sequences providing an in-frame stop codon, the KpnI restriction site in exon 4 is abolished in the mutant allele. Solid black and shaded bars (bottom) indicate the relative positions of the probes used to screen for homologous recombination events. E, EcoRI; H, HindIII; K, KpnI; N, NotI. (B) Southern blot analysis of KpnIdigested genomic DNA isolated from parental (E14) or targeted (E14-Roe-1 and E14-Roe-2) ES cell lines, respectively. (C) Representative PCR genotyping analysis of embryos collected at 9.5 dpc from fur+/− intercross matings. Appropriate PCR controls were included as indicated. (D) Northern blot analysis of total liver RNA from wild-type (+/+) and heterozygous (+/−) mice using the fur cDNA probe L (upper panels) or an α-actin-specific probe (lower panel). The upper panels show the same hybridization exposed either for 3 days (upper panel) or for 14 days (middle panel). The arrow indicates the presence of a transcript derived specifically from the targeted allele. (E) RT-PCR analysis of fur mRNA expression in 9.5 dpc embryos. The upper panel shows PCR products comprising exon 3 to exon 6 sequences, whereas in the lower panel, these products were digested with KpnI. Resistance to KpnI is indicative of cryptic splice events (see text for details). Genotypes and control PCRs were as indicated.

Role of Furin in ventral morphogenesis 4867 resorbed tissues. In contrast at 10.5 dpc, 39 (24%) of 160 embryos analyzed were homozygous mutant (Fig. 2C), suggesting that fur−/− embryos die between 10.5 and 11.5 dpc. To confirm that the targeted fur allele is non-functional, total liver RNA was isolated from wild-type and heterozygous siblings and analyzed by northern blotting. In all samples tested, the fur cDNA probe L hybridized to a 4.3 kb transcript corresponding to wild-type fur mRNA. The abundance of this transcript was reduced by approximately 50% in RNA from heterozygotes compared to wild type (Fig. 2D). In heterozygotes, an additional 6.0 kb transcript was detected that also hybridizes with probes R2 and H (data not shown), suggesting that the targeted allele is transcribed at a reduced level and gives rise to a mRNA where the Furin coding region is disrupted by insertion of the PGKhygro cassette. To further analyze transcripts arising from the mutant allele, a cDNA fragment comprising exon 3 to exon 6 sequences was amplified by RT-PCR using RNA from 9.5 dpc embryos. In both wild-type and heterozygous embryos we detected a 391 bp product together with a significantly less abundant 300 bp fragment (Fig. 2E). Unexpectedly, small amounts of similar products were also detected in homozygous mutant embryos. To test whether any of these fragments are derived from low levels of functional mRNA produced from the targeted fur allele, they were digested with KpnI. As shown in Fig. 1E (lower panel) the majority of the RT-PCR product obtained from wild-type or heterozygous embryos was cleaved by KpnI, giving rise to two closely migrating fragments of 184 and 207 bp, respectively. However, both the 300 bp and a minor fraction of the 391 bp fragment were resistant to KpnI treatment, as were the products amplified from homozygous mutant embryos, suggesting that these PCR products likely represent low levels of aberrant splicing events. To confirm this, the KpnI-resistant fragments obtained from fur−/− embryos were further amplified in a second round of PCR. Direct sequencing of the larger fragment revealed that a cryptic splice donor present within the rudimentary KpnI site (GGTAC) of exon 4 had been used to splice around the PGKhygro cassette, resulting in a transcript where exon 4 is truncated by seven nucleotides. This truncation causes a frame shift and disrupts the open reading frame of Furin after amino acid 129. Moreover, sequence analysis of the KpnI-resistant fragments showed that the same cryptic splice site is also used in wildtype embryos. Finally, sequencing of the smaller RT-PCR fragment revealed that it was derived from transcripts devoid of exon 4 sequences. Thus, exon 3 is spliced in-frame to exon 5. An analogous splice event has been observed in a mutant allele of the closely related PC2 gene (Furuta et al., 1997). However, as in the case of PC2, a fur transcript devoid of exon 4 can only give rise to a protein which lacks the multibasic motif required for autoactivation of the zymogenic pro-Furin. Therefore, it is highly unlikely that any functional protein can be generated from the targeted fur allele. Multiple tissue abnormalities in Furin-deficient embryos While homozygous mutant embryos survive until 10.5 dpc, they display several overt abnormalities. Most noticeably, morphogenesis of the yolk sac vasculature is disrupted, and the embryos fail to undergo axial rotation. As shown in Fig. 3, a well organized vascular network consisting of both capillaries

and vitelline vessels has developed throughout the yolk sac of normal 10.5 dpc embryos (Fig. 3A,E), whereas in mutant littermates, the capillaries are poorly organized and vitelline vessels do not form (Fig. 3B,F). Fetal red blood cells are clearly recognizable, but are pooled in a single patch on one side of the yolk sacs (Fig. 3B). Occasionally, the endodermal surface of mutant yolk sacs is ruffled in appearance, presumably due to failure of the vasculature to expand (Fig. 3F). Blood islands can also clearly be seen in mutants 1 day earlier (9.5 dpc, Fig. 3I), but their number is reduced (Fig. 3J,K) compared to control littermates (Fig. 3G). However, the cellular architecture of both visceral endoderm and mesoderm appear to be morphologically normal (Fig. 3I). The presence of Flk-1 transcripts further suggests that extraembryonic mesoderm is competent to give rise to endothelial precursors (Fig. 3C,D; Yamaguchi et al., 1993). During early somite stages (4-6 somites), normal embryos initiate the process of axial rotation. While wild-type or heterozygous embryos have completed this process within 24 hours (Fig. 3G), mutant littermates consistently fail to turn (Fig. 3J,K), and by 9.5 dpc they are approximately 30% smaller than controls. Noteably, the gut endoderm and the body wall have failed to properly fuse at the ventral midline (Fig. 3KM,O). This ventral fusion defect is frequently associated with cardia bifida (77%; n=26), and as development proceeds, with increasingly severe disorganization of all ventral tissues (Fig. 3O). In less severe cases (Figs 3K, 5I,J) a single heart tube occasionally forms, but these consistently fail to undergo looping morphogenesis. Only in rare instances (9%; n=45) were these rudimentary hearts found to be beating at the time of dissection (9.5-10.5 dpc). However, this is not due to the absence of myocytes, since both the epimyocardium and an endothelial layer lining the inside of the heart tube are readily distinguishable (Fig. 3L). The midgut region is also highly abnormal in fur mutants. The ventralmost aspect of the embryo normally occupied by the gut tube comprises the cardinal veins, which are significantly enlarged and engorged with nucleated red blood cells (Fig. 3K,M). Presumably, cardiac insufficiency limits the blood flow between the yolk sac and the embryonic vasculature, and this is likely to be responsible for the developmental arrest and death occurring between 10.5 to 11.5 dpc. Additional defects seen in mutant embryos at 9.5 dpc include exencephaly and kinking of the neural plate and neural tube (Fig. 3K). The anteroposterior body axis is frequently shortened in the trunk and tail region, and the somites are abnormally compact and small (Fig. 3J). Interestingly, in spite of continued growth, the allantois fails to fuse with the chorion and instead becomes highly vacuolated by 9.5 dpc, and is displaced laterally as opposed to projecting towards the chorion (Figs 3J,N, 5M, 7E).

nodallacZ expression is unaffected by the lack of Furin During normal development, axial rotation is preceded by the asymmetric expression of the TGFβ family members Nodal, Lefty-1 and Lefty-2 (Collignon et al., 1996; Lowe et al., 1996; Meno et al., 1996, 1997), consistent with the idea that the direction of heart looping and embryonic turning is likely to be controlled by the overlapping activities of several TGFβ-related signals. To test whether the left-right signaling

4868 A. J. M. Roebroek and others

Fig. 3. Histological examination of fur−/− embryos between 9.5 and 10.5 dpc. (A) Control and (B-D) homozygous mutant yolk sacs. (C) Whole-mount view and (D) section through a mutant yolk sac (9.5 dpc) stained for Flk-1 mRNA. (E) Section through control 10.5 dpc yolk sac, compared to (F) yolk sac of a mutant litter mate. (G) Transverse section of a 9.5 dpc control embryo and (H), at higher magnification, of its yolk sac. (J) Parasagittal and (K,N) transverse sections through mutant littermates of the embryo shown in G, and (I) higher magnification of a blood island in the mutant yolk sac of the embryo shown in K. Note that the mutant embryos have not turned (J,K), and their allantoides failed to fuse with the chorion (J,N). In the mutants, the neural tube is kinked and fails to close (K), and the fusion of the heart tube and of the gut epithelium at the ventral midline is highly abnormal (L,M), leading to the formation of a heart that fails to loop (J-L), or of two separate heart tubes, which by 10.5 dpc are highly disorganized (O). al, allantois; am, amnion; bc, blood cell; bi, blood island; bw, body wall; ca, common atrial chamber; ch, chorion; cv, cardinal veins; da, dorsal aorta; dm, dorsal mesocardium; ec, endothelial cell; em, extraembryonic mesoderm; en, endocardium; ep, epithelial layer; fg, foregut; hg, hindgut; ht, heart tube; lht/rht, left and right heart tube; luv, left umbilical vein; mc, myocardium; np, neural plate; nt, neural tube; sh, right sinus horn; so, somite; ve, visceral endoderm; vv, vitelline vessel; vys, visceral yolk sac.

Role of Furin in ventral morphogenesis 4869

Fig. 4. The lack of Furin does not perturb the asymmetric expression of nodal. (A) Control and (B-D) fur−/− embryos normally express a nodallacZ reporter allele in the left lateral plate mesoderm at 8.5 dpc (B), and in the node between 7.5 (D) and 8.5 dpc (C). Note that even in a fur−/− background, a single copy of functional nodal is sufficient for normal development at least until 8.5 dpc. h, heart; hf, headfold; lp, lateral plate; n, node; so, somite; anterior is to the left.

Fig. 5. Defective ventral closure in fur−/− homozygotes. (A) In control 9.5 dpc embryos, expression of shh marks midline structures, including the floorplate and the notochord (nc), as well as the primitive gut (g). (B) Homozygous mutant exhibiting cardia bifida. No gut tube has formed, but shh-positive cells are situated at the ventral surface on each side of the embryo. (C,D) Frontal (C) and posterior view (D) of a normal embryo at 8.5 dpc stained for HNF3β mRNA. (E) Ventrolateral view of a stage-matched mutant littermate of the embryo shown in C,D. No HNF3β-positive cells are detected in the AIP, and the hindgut has not yet formed. (F) α-cardiac actin expression throughout the heart tube of an 8.5 dpc control embryo that is about to turn. (G) In mutant littermates, no heart tube has formed. (H-K) Expression of α-cardiac actin is reduced in the rudimentary hearts of fur−/− embryos (I-K) compared to control littermates (H). Note the absence of a normally looped heart tube. (L) Normal embryo (9.5 dpc) expressing myosin light chain 2V throughout the heart tube (h). (M) Bifid heart of a mutant embryo in which MLC2V expression marks two separate heart tubes. A,B,DF,H,J,K: lateral view; C,G,I,M: frontal view. aip, anterior intestinal portal; al, allantois; g, gut; h, heart; hg, hindgut; n, node; nc, notochord; np, neural plate.

4870 A. J. M. Roebroek and others pathway responsible for the sidedness of Nodal expression is disturbed in Furin-deficient embryos, a nodallacZ allele was introduced into the fur−/− background. Since insertion of a βgeo cassette in this allele disrupts nodal function (Collignon et al., 1996), this experiment also allowed us to investigate whether a single functional copy of nodal becomes haploinsufficient in a fur null background. Males doubly heterozygous for fur and nodallacZ were mated to fur+/− females. Embryos were recovered at 8.5 dpc, genotyped and stained to examine nodallacZ expression. Of 99 embryos analyzed, 49 were nodallacZ heterozygotes and expressed βgalactosidase, and 26 were homozygous mutant for fur. Of these, 14 (54%) were heterozygous for the nodallacZ allele but appeared morphologically normal, suggesting that even in a fur null background, a single copy of functional nodal is sufficient for normal development at least up to this stage. Moreover, β-galactosidase expression was unperturbed either in the node or the left lateral plate mesoderm of Furindeficient embryos (Fig. 4B,C) compared to stage-matched wild-type litter mates (Fig. 4A). NodallacZ was also found to be normally expressed in the node of mutant embryos at 7.5 dpc (n=2; Fig. 4D). Together, these results suggest that the failure to undergo axial rotation is not associated with deregulation of nodal expression. Ventral closure and cardiac defects Ventral morphogenesis is initiated as a rostral to caudal and lateral to ventral displacement of embryonic mesoderm subjacent to the headfolds, which is fated to fuse at the ventral midline and to form a linear heart tube (Kaufman, 1992). Concomitant with these cell movements, the most anterior definitive endoderm involutes to form the foregut pocket, which then becomes extended caudally during midgut formation. While the molecular basis of endoderm specification and its subsequent migration, growth and patterning remain largely unknown, limited insight into the regulation of ventral closure has come from functional studies of the transcription factor GATA4. Loss of GATA4 results in failure of precardiac mesoderm to fuse at the ventral midline. Consequently, two separate heart tubes develop in the majority of mutant embryos (Kuo et al., 1997; Molkentin et al., 1997). To characterize the ventral closure defects of fur mutants in more detail, we analyzed the expression of a number of marker genes by whole-mount mRNA in situ hybridization. Specifically, we assessed expression of sonic hedgehog (shh) and HNF3β, which identify midline structures and definitive endoderm. At 9.5 dpc, shh is expressed in the notochord and in the overlying floorplate of the neural tube both in control and mutant embryos (Fig. 5A,B), indicating that these structures develop normally. By contrast, in the midgut region of mutant embryos, shh-positive endoderm cells are ectopically situated at the ventrally exposed surface of the primitive veins (Figs 5B, 6B) and are undetectable at more caudal levels (Fig. 6C), with the exception of the CIP (Fig. 6D) and hindgut diverticulum (Fig. 6E). Similar to shh, HNF3β staining did not reveal any defects in midline structures of embryos analyzed at 8.5 dpc (compare the mutant in Fig. 5E to the similarly sized control embryo in Fig. 5C,D). By contrast, in all of the mutant embryos analyzed (n=5), HNF3β-positive cells are significantly reduced in number both in the AIP and CIP region compared to control littermates, and a hindgut has not yet

formed, suggesting that the absence of Furin results in a significant delay in gut morphogenesis. To characterize the basis of the observed heart defects in Furin-deficient embryos, we analyzed the expression of a panel of cardiac markers. In normal embryos, α-cardiac actin is detected at 8.5 dpc in the future myocardial layer throughout the heart tube (Fig. 5F; Sassoon et al., 1988). α-Cardiac actin is also expressed in mutant embryos, indicating the presence of myocardial precursors (Fig. 5G). However, the field of αcardiac actin positive cells is reduced in size in the mutants (n=3) compared to control littermates, and it has not fused at the ventral midline. Even 1 day later (9.5 dpc; n=4), cardiogenic mesenchyme cells expressing α-cardiac actin appeared to be reduced in numbers in mutants compared to controls, and they failed to form a looping heart tube (Fig. 5HK). Next, we examined expression of myosin light chain (MLC)-2V, which by 9.5 dpc is normally restricted to the ventricular and proximal outflow tract of the primitive heart (Figs 5L, 6F; O’Brien et al., 1993). High levels of MLC-2V mRNA were also detected in mutant embryos (n=4), although their heart tubes were abnormally short and bifid (Figs 5M, 6G). Similar results were obtained using a probe for MLC-2A, which is expressed throughout the heart prior to 11.5 dpc (Kubalak et al., 1994; data not shown). Together, these results confirm that cardiac myocytes are specified in mutant embryos, but that a severe ventral closure defect has a major impact on heart tube morphogenesis. We next analyzed expression of eHand, a transcription factor expressed in the heart, pericardium, and in vitelline and umbilical vessels (Biben and Harvey, 1997; Firulli et al., 1998). As expected within the heart tube, eHand was expressed in the future left ventricle both in wild-type and fur+/− embryos (Fig 6H). Interestingly, eHand is also asymmetrically expressed in the rudimentary heart tube(s) of fur−/− embryos (Fig. 6I,J). Thus, both in bifid hearts as well as in cases where a single heart tube forms, eHand is expressed preferentially on the left side (n=5; Fig. 6I,J shows an intermediate situation where a heart tube has partly fused). Caudal to the heart, strong bilateral expression of eHand is observed in tissues fated to form the ventral body wall and the vitelline and umbilical vessels, respectively. However, the lateral mesodermal tissues fail to fuse at the ventral midline, and they are severely disorganized, making it difficult to distinguish the vitelline and umbilical vessels in transverse sections (Fig. 6K). The cardia bifida defects that develop in fur−/− mutants are strikingly similar to those seen in embryos lacking the transcription factor GATA4 (Kuo et al., 1997; Molkentin et al., 1997). On the other hand, fur−/− embryos forming a single heart tube more closely resemble embryos lacking the homeobox transcription factor Nkx2.5, where the rudimentary heart tube fails to loop and instead bends in a ventral as opposed to lateral direction (Lyons et al., 1995a). In light of these similarities, it was of interest to examine whether abnormalities in the expression patterns of either of these transcription factors contribute to the heart defects in fur mutants. At early primitive streak stages, GATA4 is widely expressed in visceral endoderm and nascent mesoderm. As the definitive endoderm migrates to displace the primitive visceral endoderm, GATA4 mRNA becomes confined to the most proximal domain abutting the extraembryonic tissue (Fig. 7A-C). At the advanced primitive streak stage, GATA4 mRNA is expressed in a crescent of

Role of Furin in ventral morphogenesis 4871 mesoderm subjacent to the headfold (Fig. 7A, right-most embryo). By 8.5 dpc, GATA4 transcripts are highly abundant throughout the AIP and the heart tube, as well as in the allantois (Fig. 7D,E). In all 28 embryos collected from intercross matings and examined at 7.5 dpc, the expression pattern of GATA4 appeared normal, indicating that the loss of Furin is unlikely to interfere with the expression of GATA4. Likewise, GATA4 expression appeared to be normal in the AIP of mutant 8.5 dpc embryos (n=4, Fig. 7D). Only in the most anterior aspect of bifid mutant hearts were GATA4 expression levels reduced, probably due to a delay in the migration of precardiac mesoderm (Fig. 7D). Interestingly, GATA4 expression was consistently absent in the allantois of mutant embryos (Fig. 7E), even though twist mRNA, a marker for mesodermal cells in the allantois, was normally expressed (Fuchtbauer, 1995; data not shown). In control embryos (8.0-8.5 dpc), high levels of Nkx2.5 transcripts are detected in cardiogenic mesoderm subjacent to the headfold, and throughout the newly formed heart tube (Fig. 7F,G). In comparison, Nkx2.5 mRNA expression was absent or dramatically reduced in early headfold stage embryos lacking Furin (n=9, Fig. 7F), and detectable only at reduced levels during subsequent stages (n=4; Fig. 7H), indicating that either the numbers of cardiomyocytes or Nkx2.5 transcripts per cell are compromised. DISCUSSION In the early post-implantation rat embryo, fur mRNA has been shown to be expressed in both embryonic and extraembryonic lineages (Zheng et al., 1994). Consistent with these data, we here report that in the mouse embryo fur mRNA is expressed in the extraembryonic endoderm and mesoderm during primitive streak stages. We have also documented strong fur expression in the anterior visceral endoderm and the underlying cardiogenic mesoderm. 1 day later (8.5 dpc), fur mRNA is abundantly expressed in the myocardial and endocardial layers of the heart, in definitive gut endoderm, lateral plate mesoderm, the allantois, and in the ventral layer of the node. While the physiological substrates of Furin in these tissues have not yet been identified, we note a striking degree of overlap in the distribution of fur transcripts with that of candidate substrates belonging to the TGFβ family. For example during gastrulation, both BMP2 and BMP7 are expressed in anterior visceral endoderm (Lyons et al., 1995b; Winnier et al., 1995; Zhang and Bradley, 1996). At the time of ventral closure, BMP2 as well as TGFβ2 transcripts are found within cardiogenic mesoderm (Dickson et al., 1993; Winnier et al., 1995; Zhang and Bradley, 1996), and TGFβ2 mRNA also has been detected in the adjacent pharyngeal endoderm (Dickson et al., 1993). At slightly later stages (8.5 dpc), TGFβ2, BMP4 and BMP7 mRNAs are abundant in the myocardium throughout the heart tube (Jones et al., 1991; Dickson et al., 1993; Lyons et al., 1995b), whereas the endocardial layer expresses TGFβ1 (Dickson et al., 1993). Interestingly, Nodal and BMP7 are both expressed in the node and notochordal plate, whereas BMP4, -5, -7, Nodal and Lefty2 transcripts are present in lateral plate mesoderm (Collignon et al., 1996; Lowe et al., 1996; Meno et al., 1996, 1997; Dudley and Robertson, 1997; M. Solloway and E. J. R., unpublished).

Finally, we note that multiple BMPs including BMP2, -4, -5, -6 and -7 are all coexpressed with Furin in the allantois (Jones et al., 1991; Lyons et al., 1995a; Winnier et al., 1995; M. Solloway and E. J. R., unpublished). Together with biochemical data showing that Furin and related proteases enhance cleavage of TGFβ-related precursor proteins such as TGFβ1 (Dubois et al., 1995), activin A (Roebroek et al., 1993), Mullerian inhibiting substance (Nachtigal and Ingraham, 1996), BMP4 (Cui et al., 1998; D. B. C. and E. J. R., unpublished), and Nodal (D. B. C. and E. J. R., unpublished), these overlapping expression patterns indicate that Furin may contribute to the proteolytic activation of at least a sub-set of TGFβ-related growth factors. To test whether Furin is required for normal development of the mouse embryo, we generated a loss of function mutation at the fur locus using homologous recombination. Homozygous mutant embryos survive until 10.5 dpc, but die shortly thereafter, presumably due to hemodynamic insufficiency. The yolk sacs of mutant embryos fail to develop large blood vessels, although patches of blood islands containing embryonic nucleated red blood cells form, and endothelial precursors expressing the molecular marker Flk-1 are present. A similar defect has been observed in embryos deficient in either the transcription factor SCL/tal-1 or TGFβ1 (Dickson et al., 1995; Robb et al., 1995; Visvader et al., 1998), raising the possibility that Furin may be similarly required for the terminal differentiation of endothelial cells. We also found that the allantois, a tissue that expresses abundant levels of fur mRNA, fails to fuse with the chorion and instead becomes highly vacuolated by 9.5 dpc in mutant embryos. Interestingly, chorioallantoic fusion is also abolished in embryos lacking both BMP5 and BMP7 (M. Solloway and E. J. R., unpublished), consistent with the idea that a decrease in BMP signaling may contribute to this defect in Furin-deficient embryos. Chorioallantoic fusion also is inhibited in embryos lacking either VCAM-1 or α4-integrin (Gurtner et al., 1995; Kwee et al., 1995; Yang et al., 1995), two cell-surface proteins expressed within the allantois or chorion, respectively, which interact with each other to mediate cell-cell adhesion (Elices et al., 1990). Possibly, Furin activities are required to maintain the expression of VCAM-1. Alternatively, Furin may play a role in the proliferation or migration of primitive streak-derived mesodermal cells which form the allantois. However, mutant allantoides are not overtly reduced in size compared to wild type, and they appropriately express twist mRNA, which identifies a population of mesodermal cells (Fuchtbauer, 1995), indicating that proliferation and migration of these cells are relatively normal. In marked contrast, GATA4 expression is absent in mutant allantoides, suggesting perhaps that cells which would normally express GATA4 fail to adopt their normal fate. Alternatively, in this cell population GATA4 may be a downstream target of factors activated by Furin. The most profound defect in fur−/− embryos is their failure to undergo ventral closure, i.e. to form a looping heart tube and a coherent primitive gut. While definitive endoderm, identified by the expression of HNF3β or shh mRNA, is formed, fusion of this sheet of cells at the ventral midline is delayed and fails to proceed in the midgut region. As a result, only the foregut and hindgut diverticula develop in mutant embryos, whereas the midgut is missing. Furin activities therefore seem to be intimately involved in either the proliferation or migration of

4872 A. J. M. Roebroek and others

Fig. 6. Transverse tissue sections obtained from embryos stained by whole-mount in situ hybridization. (A-E) Sections of the shh-stained embryos shown in Fig. 5A,B. At the entrance to the foregut, shh-positive cells have failed to involute in the mutant (B, open arrow), but not in the normal embryo (A). (C) In the trunk region of the same mutant embryo shown in B, shh mRNA highlights the absence of gut endoderm. (D) At the level of the caudal intestinal portal and of (E) the hindgut, shh expression marks the presence of both the notochord and definitive endoderm. (F,G) MLC2V mRNA staining in sections through normal (F) or mutant (G) hearts, respectively (9.5 dpc). (H-K) Expression of eHand in normal (H) or mutant (I-K) 9.5 dpc embryos in cells of the rudimentary left ventricle (H-J, open arrows), and in the trunk and tail region. Note that in the mutant embryo, the body wall has not fused at the ventral midline (K). a, common atrial chamber; bc, bulbus cordis; bw, body wall; cip, caudal intestinal portal; da, dorsal aorta; en, endocardium; fg, foregut; fp, floorplate; g, gut; hg, hindgut; ht, heart tube; lp, lateral plate; lv, left ventricle; mc, epimyocardium; mg, midgut; nc, notochord; nt, neural tube; rv, right ventricle; so, somite; uv, umbilical vein; v, common ventricular chamber; va, vitelline artery.

definitive endoderm cells. In keeping with this hypothesis, high levels of fur mRNA expression are seen in the node and notochordal plate, and in the definitive endoderm, particularly in the regions of the AIP and CIP where ventral fusion normally

occurs. Within tissues fated to form the heart, cardiomyocytes identified by the expression of α-cardiac actin, MLC-2V or MLC-2A are eventually specified, although their numbers appear to be reduced, and they fail to fuse at the ventral midline

Role of Furin in ventral morphogenesis 4873

Fig. 7. Whole-mount in situ hybridization of GATA4 and Nkx2.5 expression in embryos derived from fur+/− intercross matings. (A) GATA4 mRNA expression in four representative 7.5 dpc embryos. (B,C) Transverse serial sections from the proximal embryonic region of a 7.5 dpc embryo stained for GATA4 mRNA. The level of sectioning in B is next to the extraembryonic region, whereas the section in C is derived from a slightly more distal level. (D) Frontal view of embryos stained for GATA4 at 8.5 dpc. Note the bifid hearts in the mutants. (E) Ventral view of embryos (8.5 dpc) showing GATA4 expression in the allantois (al) of control embryos, but not in that of fur−/− mutants. (F) Embryos (8.0 dpc) stained for Nkx2.5 mRNA. Nkx2.5 expression is reduced in mutant embryos (n=9) compared to control litter mates. (G,H) During early somite stages, Nkx2.5 mRNA expression is also reduced in fur−/− mutants (n=4, H) compared to control littermates (G). aip, anterior intestinal portal; al, allantois; en, visceral endoderm; ep, epiblast; ht, heart tube; m, embryonic mesoderm. In A-C and E, anterior is to the left.

during the process of ventral closure, leading to cardia bifida in the majority of mutant embryos. Interestingly, a very similar phenotype develops in embryos lacking GATA4 (Kuo et al., 1997; Molkentin et al., 1997). Since both Furin and GATA4 are expressed in the anterior regions of visceral endoderm and in the underlying cardiogenic mesoderm, one possibility is that they act in a common pathway. However, with the exception of the allantois, GATA4 mRNA expression is largely normal in Furin-deficient embryos, suggesting that it acts independently from Furin to regulate ventral fusion of precardiac mesoderm. Moreover, chimera analysis has shown that embryos largely composed of GATA4−/− cells develop normally in conjunction with wild-type visceral endoderm (Narita et al., 1997), suggesting that GATA4 activities are required exclusively in the primitive visceral endoderm lineage to direct ventral morphogenesis. Most recently, chimera analysis has also been highly informative in delineating distinct functions for the transcription factor HNF3β in the extraembryonic visceral endoderm and in embryonic lineages (Dufort et al., 1998). Thus, HNF3β is required during gastrulation in the visceral endoderm for normal elongation of the primitive streak, whereas at a later stage it is essential in the embryo proper for the specification of node and notochord cell fates. Interestingly, in conjunction with wild-type visceral endoderm, chimeric embryos largely composed of HNF3β−/− cells fail to form definitive fore- and midgut endoderm, and they fail to undergo heart looping morphogenesis and axial rotation. We are currently examining whether Furin activities are essential only in the visceral endoderm, or whether this protease is required in both extraembryonic and embryonic lineages, as is the case for HNF3β. In a small subset of Furin-deficient embryos, a single heart tube occasionally develops, but subsequently fails to undergo looping morphogenesis. A similar heart defect is seen in embryos deficient for either Nkx2.5 or MEF2C (Lyons et al., 1995a; Lin et al., 1997). Interestingly, Nkx2.5 mRNA, marking cardiomyocyte lineages, is expressed at significantly reduced levels in fur mutants, accounting perhaps in part for the failure of heart looping. This reduction in the expression levels of Nkx2.5 is consistent with the idea that Furin activities may contribute to the activation of BMPs in the heart field, since studies in both chick and Drosophila embryos have implicated TGFβ/BMP-related signals in the specification of cardiac lineages (Frasch, 1995; Schultheiss et al., 1997; Andree et al., 1998). Thus, the cardiac markers GATA4 and Nkx2.5 are ectopically induced in anterior paraxial mesoderm treated with recombinant BMP2 or BMP4 (Schultheiss et al., 1997; Andree et al., 1998), whereas cardiac myogenesis assayed by the expression of Nkx2.5 is inhibited in tissue explants treated with the BMP antagonist Noggin, strongly arguing for a role of endogenous BMP signaling in the specification of cardiac myocytes (Schultheiss et al., 1997). Tissue recombination experiments in chick and frog embryos suggest that in addition to BMP activities, inductive signals derived from definitive endoderm are also required for normal heart development (Nascone and Mercola, 1995; Schultheiss et al., 1995). The delay in the migration of definitive endoderm observed in fur mutants is therefore likely to contribute to the heart defects. In support of this conclusion, similar heart defects occur in HNF3β chimeras that lack fore- and midgut endoderm (Dufort et al., 1998).

4874 A. J. M. Roebroek and others In addition to defects in ventral closure, Furin-deficient embryos consistently fail to undergo axial rotation, a process that is normally preceded by the rightward looping of the heart tube. The inability of fur mutants to turn is unlikely to result solely from abnormal heart looping morphogenesis, since similar heart defects do not interfere with axial rotation in embryos lacking Nkx2.5 or MEF2C (Lyons et al., 1995a; Lin et al., 1997). Likewise, impaired yolk sac vasculogenesis is unlikely to account for the failure of axial rotation, since embryos with similar vascular defects owing to the loss of TGFβ1 or the transcription factor SCL/tal-1 turn normally (Dickson et al., 1995; Robb et al., 1995; Visvader et al., 1998). Possibly, the inability to turn results from mechanical constraints associated with the endoderm defects. However, apart from a delay in hindgut formation, fur mutants show no overt indications of mechanical obstruction at early somite stages when embryonic turning is normally initiated, and their morphology and size are strikingly normal up to this stage. Moreover, the left-right axis is established since two molecular markers of left-right asymmetry, Nodal and eHand, are expressed appropriately in fur−/− embryos. Given the observation that fur is coexpressed with both nodal and lefty2 in the left lateral plate mesoderm, it is possible that Furin is required to generate physiological amounts of these ligands. In support of this hypothesis, overexpression of Furin leads to a dramatic increase in the efficiency of Nodal precursor cleavage in COS cells (D. B. C. and E. J. R., unpublished). However, the turning defect is unlikely solely to reflect inefficient processing of Nodal or Lefty, since in iv/iv embryos the absence of asymmetric nodal and lefty expression does not abolish, but rather randomizes the direction of axial rotation (Lowe et al., 1996; Meno et al., 1996). Thus, Furin seems to be required for the activation of additional precursor proteins, potentially including BMP4, -5 and -7, all of which are strongly expressed in the lateral plate (Dudley and Robertson, 1997; M. Solloway and E. J. R., unpublished). While biochemical evidence and the striking overlap in expression patterns support the notion that Furin may be responsible for activating TGFβ-related growth factors, loss of Furin has less severe consequences than mutations in components of these signaling pathways, including nodal (Zhou et al., 1993; Conlon et al., 1994), bmp4 (Winnier et al., 1995), or receptors for TGFβ family members (Mishina et al., 1995; Gu et al., 1998) and their downstream targets (Sirard et al., 1998; Waldrip et al., 1998). This clearly suggests that the activation of TGFβ/BMP signaling pathways involves convertase activities other than Furin. However, the phenotype of fur mutants may in part reflect impaired function of TGFβrelated signals in tissues that strongly express Furin, and is in keeping with our previous hypothesis that the overlapping activities of several related convertases including Furin, PACE4, PC5(6) and/or PC7 may promote the production of mature TGFβ-related ligands to control their local concentration (Constam et al., 1996). In future studies, it will be important to examine the processing efficiency of these and other candidate Furin substrates in cells derived from fur−/− embryos that specifically lack Furin activities. This work was supported by grants from the Fonds voor Wetenschappelijk Onderzoek-Vlaanderen (FWO), by a grant of the Interuniversity-network for Fundamental Research (IUAP) of the

Belgian government, by the action-program for Biotechnology of the Flemish government (VLAB/COT-008), and the National Institutes of Health. L. U. received a post-doctoral research fellowship from the K. U. Leuven research fund. D. B. C. is the recipient of fellowship 823A46642 from the Swiss National Science Foundation.

REFERENCES Andree, B., Duprez, D., Vorbusch, B., Arnold, H. H. and Brand, T. (1998). BMP-2 induces ectopic expression of cardiac lineage markers and interferes with somite formation in chicken embryos. Mech. Dev. 70, 119-131. Arceci, R. J., King, A. A., Simon, M. C., Orkin, S. H. and Wilson, D. B. (1993). Mouse GATA-4: a retinoic acid-inducible GATA-binding transcription factor expressed in endodermally derived tissues and heart. Mol. Cell. Biol. 13, 2235-2246. Biben, C. and Harvey, R. P. (1997). Homeodomain factor Nkx2-5 controls left/right asymmetric expression of bHLH gene eHand during murine heart development. Genes Dev. 11, 1357-1369. Bravo, D. A., Gleason, J. B., Sanchez, R. I., Roth, R. A. and Fuller, R. S. (1994). Accurate and efficient cleavage of the human insulin proreceptor by the human proprotein-processing protease furin. Characterization and kinetic parameters using the purified, secreted soluble protease expressed by a recombinant baculovirus. J. Biol. Chem. 269, 25830-25837. Chasan, R. and Anderson, K. V. (1989). The role of easter, an apparent serine protease, in organizing the dorsal-ventral pattern of the Drosophila embryo. Cell 56, 391-400. Collignon, J., Varlet, I. and Robertson, E. J. (1996). Relationship between asymmetric nodal expression and the direction of embryonic turning. Nature 381, 155-158. Conlon, F. L., Lyons, K. M., Takaesu, N., Barth, K. S., Kispert, A., Herrmann, B. and Robertson, E. J. (1994). A primary requirement for nodal in the formation and maintenance of the primitive streak in the mouse. Development 120, 1919-1928. Constam, D. B., Calfon, M. and Robertson, E. J. (1996). SPC4, SPC6, and the novel protease SPC7 are coexpressed with bone morphogenetic proteins at distinct sites during embryogenesis. J. Cell Biol. 134, 181-191. Creemers, J. W. M., Roebroek, A. J. M., van den Ouweland, A. M., van Duijnhoven, H. L. and Van De Ven, W. J. M. (1992). Cloning and functional expression of a 4.3 kbp mouse fur cDNA: Evidence for differential expression. Mol. Biol. 11, 127-138. Cui, Y., Jean, F., Thomas, G. and Christian, J. L. (1998). BMP4 is proteolytically activated by Furin and/or PC6 during vertebrate embryonic development. EMBO J. (in press) Dickson, M. C., Martin, J. S., Cousins, F. M., Kulkarni, A. B., Karlsson, S. and Akhurst, R. J. (1995). Defective haematopoiesis and vasculogenesis in transforming growth factor-beta 1 knock out mice. Development 121, 1845-1854. Dickson, M. C., Slager, H. G., Duffie, E., Mummery, C. L. and Akhurst, R. J. (1993). RNA and protein localisations of TGF beta 2 in the early mouse embryo suggest an involvement in cardiac development. Development 117, 625-639. Doetschman, T., Gregg, R. G., Maeda, N., Hooper, M. L., Melton, D. W., Thompson, S. and Smithies, O. (1987). Targeted correction of a mutant HPRT gene in mouse embryonic stem cells. Nature 10, 576-578. Dubois, C. M., Laprise, M. H., Blanchette, F., Gentry, L. E. and Leduc, R. (1995). Processing of transforming growth factor beta 1 precursor by human furin convertase. J. Biol. Chem. 270, 10618-10624. Dudley, A. T. and Robertson, E. J. (1997). Overlapping expression domains of bone morphogenetic protein family members potentially account for limited tissue defects in BMP7 deficient embryos. Dev. Dyn. 208, 349362. Dufort, D., Schwartz, L., Harpal, K. and Rossant, J. (1998). The transcription factor HNF3beta is required in visceral endoderm for normal primitive streak morphogenesis. Development 125, 3015-3025. Duguay, S. J., Milewski, W. M., Young, B. D., Nakayama, K. and Steiner, D. F. (1997). Processing of wild-type and mutant proinsulin-like growth factor-IA by subtilisin-related proprotein convertases. J. Biol. Chem. 272, 6663-6670. Echelard, Y., Epstein, D. J., St-Jacques, B., Shen, L., Mohler, J., McMahon, J. A. and McMahon, A. P. (1993). Sonic hedgehog, a member of a family of putative signaling molecules, is implicated in the regulation of CNS polarity. Cell 75, 1417-1430.

Role of Furin in ventral morphogenesis 4875 Elices, M. J., Osborn, L., Takada, Y., Crouse, C., Luhowskyj, S., Hemler, M. E. and Lobb, R. R. (1990). VCAM-1 on activated endothelium interacts with the leukocyte integrin VLA-4 at a site distinct from the VLA4/fibronectin binding site. Cell 60, 577-584. Firulli, A. B., McFadden, D. G., Lin, Q., Srivastava, D. and Olson, E. N. (1998). Heart and extra-embryonic mesodermal defects in mouse embryos lacking the bHLH transcription factor Hand1. Nat. Genet. 18, 266-270. Frasch, M. (1995). Induction of visceral and cardiac mesoderm by ectodermal Dpp in the early Drosophila embryo. Nature 374, 464-467. Fuchtbauer, E. M. (1995). Expression of M-twist during postimplantation development of the mouse. Dev. Dyn. 204, 316-322. Furuta, M., Yano, H., Zhou, A., Rouille, Y., Holst, J. J., Carroll, R., Ravazzola, M., Orci, L., Furuta, H. and Steiner, D. F. (1997). Defective prohormone processing and altered pancreatic islet morphology in mice lacking active SPC2. Proc. Natl. Acad. Sci. USA 94, 6646-6651. Greenwald, I. (1998). LIN-12/Notch signaling: lessons from worms and flies. Genes Dev. 12, 1751-1762. Gu, Z., Nomura, M., Simpson, B. B., Lei, H., Feijen, A., van den Eijndenvan Raaij, J., Donahoe, P. K. and Li, E. (1998). The type I activin receptor ActRIB is required for egg cylinder organization and gastrulation in the mouse. Genes Dev. 12, 844-857. Gurtner, G. C., Davis, V., Li, H., McCoy, M. J., Sharpe, A. and Cybulsky, M. I. (1995). Targeted disruption of the murine VCAM1 gene: essential role of VCAM-1 in chorioallantoic fusion and placentation. Genes Dev. 9, 1-14. Hogan, B., Beddington, R., Costantini, F. and Lacy, E. (1994). Manipulating the Mouse Embryo. A Laboratory Manual. Cold Spring Harbor Laboratory Press, New York. Jackson, R. S., Creemers, J. W., Ohagi, S., Raffin-Sanson, M. L., Sanders, L., Montague, C. T., Hutton, J. C. and O’Rahilly, S. (1997). Obesity and impaired prohormone processing associated with mutations in the human prohormone convertase 1 gene. Nat. Genet. 16, 303-306. Jones, C. M., Lyons, K. M. and Hogan, B. L. M. (1991). Involvement of bone morphogenetic protein-4 (BMP-4) and Vgr-1 in morphogenesis and neurogenesis in the mouse. Development 111, 531-542. Kaufman, M. H. (1992). The Atlas of Mouse Development. Academic Press, London. Kirov, N., Zhelnin, L., Shah, J. and Rushlow, C. (1993). Conversion of a silencer into an enhancer: evidence for a co-repressor in dorsal-mediated repression in Drosophila. EMBO J. 12, 3193-3199. Komada, M., Hatsuzawa, K., Shibamoto, S., Ito, F., Nakayama, K. and Kitamura, N. (1993). Proteolytic processing of the hepatocyte growth factor/scatter factor receptor by furin. FEBS Lett. 328, 25-29. Kopan, R., Schroeter, E. H., Weintraub, H. and Nye, J. S. (1996). Signal transduction by activated mNotch: importance of proteolytic processing and its regulation by the extracellular domain. Proc. Natl. Acad. Sci. USA 93, 1683-1688. Kubalak, S. W., Miller-Hance, W. C., O’Brien, T. X., Dyson, E. and Chien, K. R. (1994). Chamber specification of atrial myosin light chain-2 expression precedes septation during murine cardiogenesis. J. Biol. Chem. 269, 16961-16970. Kuo, C. T., Morrisey, E. E., Anandappa, R., Sigrist, K., Lu, M. M., Parmacek, M. S., Soudais, C. and Leiden, J. M. (1997). GATA4 transcription factor is required for ventral morphogenesis and heart tube formation. Genes Dev. 11, 1048-1060. Kwee, L., Baldwin, H. S., Shen, H. M., Stewart, C. L., Buck, C., Buck, C. A. and Labow, M. A. (1995). Defective development of the embryonic and extraembryonic circulatory systems in vascular cell adhesion molecule (VCAM-1) deficient mice. Development 121, 489-503. Lehmann, M., Rigot, V., Seidah, N. G., Marvaldi, J. and Lissitzky, J. C. (1996). Lack of integrin alpha-chain endoproteolytic cleavage in furindeficient human colon adenocarcinoma cells LoVo. Biochem. J. 317, 803809. Lin, Q., Schwarz, J., Bucana, C. and Olson, E. N. (1997). Control of mouse cardiac morphogenesis and myogenesis by transcription factor MEF2C. Science 276, 1404-1407. Logeat, F., Bessia, C., Brou, C., LeBail, O., Jarriault, S., Seidah, N. G. and Israel, A. (1998). The Notch1 receptor is cleaved constitutively by a furinlike convertase. Proc. Natl. Acad. Sci. USA 95, 8108-8112. Lowe, L. A., Supp, D. M., Sampath, K., Yokoyama, T., Wright, C. V., Potter, S. S., Overbeek, P. and Kuehn, M. R. (1996). Conserved left-right asymmetry of nodal expression and alterations in murine situs inversus. Nature 381, 158-161. Lyons, I., Parsons, L. M., Hartley, L., Li, R., Andrews, J. E., Robb, L. and

Harvey, R. P. (1995a). Myogenic and morphogenetic defects in the heart tubes of murine embryos lacking the homeo box gene Nkx2-5. Genes Dev. 9, 1654-1666. Lyons, K. M., Hogan, B. L. and Robertson, E. J. (1995b). Colocalization of BMP7 and BMP2 RNAs suggests that these factors cooperatively mediate tissue interactions during murine development. Mech. Dev. 50, 71-83. Marqués, G., Musacchio, M., Shimell, M. J., Wünnenberg-Stapleton, K., Cho, K. W. Y. and O’Connor, M. B. (1997). Production of a DPP activity gradient in the early Drosophila embryo through the opposing action of the SOG and TLD proteins. Cell 91, 417-426. Mbikay, M., Tadros, H., Ishida, N., Lerner, C. P., De Lamirande, E., Chen, A., El-Alfy, M., Clermont, Y., Seidah, N. G., Chretien, M., Gagnon, C. and Simpson, E. M. (1997). Impaired fertility in mice deficient for the testicular germ-cell protease PC4. Proc. Natl. Acad. Sci. USA 94, 68426846. Meno, C., Ito, Y., Saijoh, Y., Matsuda, Y., Tashiro, K., Kuhara, S. and Hamada, H. (1997). Two closely-related left-right asymmetrically expressed genes, lefty-1 and lefty-2: their distinct expression domains, chromosomal linkage and direct neuralizing activity in Xenopus embryos. Genes Cells 2, 513-524. Meno, C., Saijoh, Y., Fujii, H., Ikeda, M., Yokoyama, T., Yokoyama, M., Toyoda, Y. and Hamada, H. (1996). Left-right asymmetric expression of the TGF beta-family member lefty in mouse embryos. Nature 381, 151155. Mishina, Y., Suzuki, A., Ueno, N. and Behringer, R. R. (1995). Bmpr encodes a type I bone morphogenetic protein receptor that is essential for gastrulation during mouse embryogenesis. Genes Dev. 9, 3027-3037. Molkentin, J. D., Lin, Q., Duncan, S. A. and Olson, E. N. (1997). Requirement of the transcription factor GATA4 for heart tube formation and ventral morphogenesis. Genes Dev. 11, 1061-1072. Molloy, S. S., Bresnahan, P. A., Leppla, S. H., Klimpel, K. R. and Thomas, G. (1992). Human furin is a calcium-dependent serine endoprotease that recognizes the sequence Arg-X-X-Arg and efficiently cleaves anthrax toxin protective antigen. J. Biol. Chem. 267, 16396-16402. Nachtigal, M. W. and Ingraham, H. A. (1996). Bioactivation of Mullerian inhibiting substance during gonadal development by a kex2/subtilisin-like endoprotease. Proc. Natl. Acad. Sci. USA 93, 7711-7716. Narita, N., Bielinska, M. and Wilson, D. B. (1997). Wild-type endoderm abrogates the ventral developmental defects associated with GATA-4 deficiency in the mouse. Dev. Biol. 189, 270-274. Nascone, N. and Mercola, M. (1995). An inductive role for the endoderm in Xenopus cardiogenesis. Development 121, 515-523. O’Brien, T. X., Lee, K. J. and Chien, K. R. (1993). Positional specification of ventricular myosin light chain 2 expression in the primitive murine heart tube. Proc. Natl. Acad. Sci. USA 90, 5157-5161. Pei, D. and Weiss, S. J. (1995). Furin-dependent intracellular activation of the human stromelysin-3 zymogen. Nature 375, 244-247. Piccolo, S., Agius, E., Lu, B., Goodman, S., Dale, L. and De Robertis, E. M. (1997). Cleavage of Chordin by Xolloid metalloprotease suggests a role for proteolytic processing in the regulation of Spemann organizer activity. Cell 91, 407-416. Robb, L., Lyons, I., Li, R., Hartley, L., Kontgen, F., Harvey, R. P., Metcalf, D. and Begley, C. G. (1995). Absence of yolk sac hematopoiesis from mice with a targeted disruption of the scl gene. Proc. Natl. Acad. Sci. USA 92, 7075-7079. Roebroek, A. J. M., Creemers, J. W. M., Pauli, I. G. L., Bogaert, T. and Van de Ven, W. J. M. (1993). Generation of structural and functional diversity in furin-like proteins in Drosophila melanogaster by alternative splicing of the Dfur1 gene. EMBO J. 12, 1853-1870. Roebroek, A. J. M., Schalken, J. A., Bussemakers, M. J., van Heerikhuizen, H., Onnekink, C., Debruyne, F. M., Bloemers, H. P. and Van de Ven, W. J. M. (1986). Characterization of human c-fes/fps reveals a new transcription unit (fur) in the immediately upstream region of the proto-oncogene. Mol. Biol. Rep. 11, 117-125. Sasaki, H. and Hogan, B. L. (1993). Differential expression of multiple fork head related genes during gastrulation and axial pattern formation in the mouse embryo. Development 118, 47-59. Sassoon, D. A., Garner, I. and Buckingham, M. (1988). Transcripts of alphacardiac and alpha-skeletal actins are early markers for myogenesis in the mouse embryo. Development 104, 155-164. Schroeter, E. H., Kisslinger, J. A. and Kopan, R. (1998). Notch-1 signalling requires ligand-induced proteolytic release of intracellular domain. Nature 393, 382-386. Schultheiss, T. M., Burch, J. B. and Lassar, A. B. (1997). A role for bone

4876 A. J. M. Roebroek and others morphogenetic proteins in the induction of cardiac myogenesis. Genes Dev. 11, 451-462. Schultheiss, T. M., Xydas, S. and Lassar, A. B. (1995). Induction of avian cardiac myogenesis by anterior endoderm. Development 121, 42034214. 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. Sirard, C., de la Pompa, J. L., Elia, A., Itie, A., Mirtsos, C., Cheung, A., Hahn, S., Wakeham, A., Schwartz, L., Kern, S. E., Rossant, J. and Mak, T. W. (1998). The tumor suppressor gene Smad4/Dpc4 is required for gastrulation and later for anterior development of the mouse embryo. Genes Dev. 12, 107-119. Stein, D. and Nusslein-Volhard, C. (1992). Multiple extracellular activities in Drosophila egg perivitelline fluid are required for establishment of embryonic dorsal-ventral polarity. Cell 68, 429-440. Van de Ven, W. J., Roebroek, A. J. and Van Duijnhoven, H. L. (1993). Structure and function of eukaryotic proprotein processing enzymes of the subtilisin family of serine proteases. Crit. Rev. Oncog. 4, 115-136. Visvader, J. E., Fujiwara, Y. and Orkin, S. H. (1998). Unsuspected role for the T-cell leukemia protein SCL/tal-1 in vascular development. Genes Dev. 12, 473-479. Waldrip, W. R., Bikoff, E. K., Hoodless, P. A., Wrana, J. L. and Robertson, E. J. (1998). Smad2 signaling in extraembryonic tissues

determines anterior-posterior polarity of the early mouse embryo. Cell 92, 797-808. Wilkinson, D. G. (1992). Whole-mount in situ hybridization of vertebrate embryos. In In situ Hybridization: A Practical Approach, pp. 75-83. Oxford: IRL Press. Winnier, G., Blessing, M., Labosky, P. A. and Hogan, B. L. M. (1995). Bone morphogenetic protein-4 is required for mesoderm formation and patterning in the mouse. Genes Dev. 9, 2105-2116. Yamaguchi, T. P., Dumont, D. J., Conlon, R. A., Breitman, M. L. and Rossant, J. (1993). flk-1, an flt-related receptor tyrosine kinase is an early marker for endothelial cell precursors. Development 118, 489-498. Yang, J. T., Rayburn, H. and Hynes, R. O. (1995). Cell adhesion events mediated by alpha 4 integrins are essential in placental and cardiac development. Development 121, 549-560. Zhang, H. and Bradley, A. (1996). Mice deficient for BMP2 are nonviable and have defects in amnion/chorion and cardiac development. Development 122, 2977-2986. Zheng, M., Streck, R. D., Scott, R. E., Seidah, N. G. and Pintar, J. E. (1994). The developmental expression in rat of proteases furin, PC1, PC2, and carboxypeptidase E: implications for early maturation of proteolytic processing capacity. J. Neurosci. 14, 4656-4673. Zhou, X., Sasaki, H., Lowe, L., Hogan, B. L. M. and Kuehn, M. R. (1993). Nodal is a novel TGF-beta-like gene expressed in the mouse node during gastrulation. Nature 361, 543-547.