In Vivo Evidence That BMP Signaling Is Necessary for Apoptosis in the Mouse Limb

Home , Mesoderm

Developmental Biology 249, 108–120 (2002) doi:10.1006/dbio.2002.0752 In Vivo Evidence That BMP Signaling Is Necessary for Apoptosis in the Mouse Limb

Udayan Guha,*,†,1 William A. Gomes,* Tatsuya Kobayashi,‡ Richard G. Pestell,† and John A. Kessler§ *Department of Neuroscience, Albert Einstein College of Medicine, Bronx, New York 10461; ‡Endocrine Unit, Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts 02115; †Division of Hormone-Dependant Tumor Biology, The Albert Einstein Cancer Center, Departments of Medicine, Developmental, and Molecular Biology, Albert Einstein College of Medicine, Bronx, New York 10461; and §Department of Neurology, Northwestern University Medical School, Chicago, Illinois 60611

To determine the role of Bone morphogenetic protein (BMP) signaling in murine limb development in vivo, the keratin 14 promoter was used to drive expression of the BMP antagonist Noggin in transgenic mice. Phosphorylation and nuclear translocation of Smad1/5 were dramatically reduced in limbs of the transgenic animals, conﬁrming the inhibition of BMP signaling. These mice developed extensive limb soft tissue syndactyly and postaxial polydactyly. Apoptosis in the developing limb necrotic zones was reduced with incomplete regression of the interdigital tissue. The postaxial extra digit is also consistent with a role for BMPs in regulating apoptosis. Furthermore, there was persistent expression of Fgf8, suggesting a delay in the regression of the AER. However, Msx1 and Msx2 expression was unchanged in these transgenic mice, implying that induction of these genes is not essential for mediating BMP-induced interdigital apoptosis in mice. These abnormalities were rescued by coexpressing BMP4 under the same promoter in double transgenic mice, suggesting that the limb abnormalities are a direct effect of inhibiting BMP signaling. © 2002 Elsevier Science (USA) Key Words: BMP; Noggin; limb development; apoptosis; Msx; Hox; FGF8; keratin 14.

INTRODUCTION oping vertebrate limb is also sculpted by programmed cell death of mesenchymal cells at the anterior and posterior Vertebrate limb buds originate from the embryonic flank margins of the limb bud (anterior necrotic zone, ANZ; and with contributions from both the lateral plate and the posterior necrotic zone, PNZ) and in the interdigital ne- somatic mesoderm. Bones, cartilage, and tendons in the crotic zones (INZ) between the developing digits. limb are derived from the lateral plate mesoderm, while The BMPs are a large family of secreted ligands within muscles, vasculature, and nerves are derivatives of the the TGF␤ superfamily that play essential roles in embry- somatic mesoderm. Early limb bud development is regu- onic development (Hogan, 1996). BMP signaling is received lated by epithelial–mesenchymal interactions between a by two types of serine threonine kinase receptors, type I highly specialized pseudostratified columnar epithelium at (BMPRI) and type II (BMPRII). On ligand binding, the type II the tip of the limb bud, called the apical ectodermal ridge receptor phosphorylates a type I recptor which then triggers (AER), and mesenchymal cells underlying the AER and at an intracellular signaling cascade involving proteins of the the posterior margin of the limb bud, in a region called the SMAD family. BMP2 and BMP4, members of the drosophila zone of polarizing activity (ZPA) (Dudley and Tabin, 2000). decapentaplegic (dpp) family of the BMPs, and BMP7 (also Mesenchymal cells aggregate to form prechondrogenic con- known as OP-1) are expressed in the undifferentiated mes- densations that subsequently ossify to form the bony skel- enchyme, interdigital mesenchyme, and the AER of the eton of the limb (Pizette and Niswander, 2001). The devel- developing limb (Laufer et al., 1997; Lyons et al., 1995). BMP receptors also show specific temporal and spatial 1 To whom correspondence should be addressed. Fax: (718) 430- patterns of expression in developing limbs in the mouse and 8674. E-mail: [email protected]. chick (Dewulf et al., 1995; Yi et al., 2000; Zou et al., 1997).

FIG. 1. Generation of K14-Noggin transgenic mice. (A) Diagram of the linearized transgene construct that was injected. (B) Southern blot analysis of genomic DNA from the tail digested with KpnI and probed with random primed ␣P32-labeled Noggin probe. The 5-kb band is the transgene, and the lower band is the endogenous gene. (C) Noggin protein is overexpressed in the transgenic adult back skin. GDI has been used as loading control. (D) Whole-mount in situ hybridization for Noggin (a, b) and Bmp4 (c, d) expression of E12.5 embryos shows that both Noggin and Bmp4 are misexpressed in all the limbs of transgenic (b, d) compared with wild type embryos (a, c). FL, forelimb; HL, hindlimb; W, whisker. FIG. 2. BMP signaling is reduced in K14-Noggin limbs. P-Smad1/5 immunoreactivity is signiﬁcantly reduced in K14-Noggin interdigital mesenchyme (B) compared with the wild type (A).

BMP functions are antagonized by a large number of se- others) that bind BMPs extracellularly thereby preventing creted proteins, such as Noggin, Chordin, Follistatin, and binding of the BMPs to their receptors. Noggin (Brunet et the DAN family members (Gremlin, Cerberus, DAN, and al., 1998; Capdevila and Johnson, 1998; Pizette and Niswan-

FIG. 3. K14-Noggin mice have abnormalities in limb development. (A) Scanning EM micrograph of the ventral side of E14.5 forelimb (a, b) and hindlimb (c, d); anterior is to the left. (B) Newborn forelimbs dorsal side (a, c), ventral side (e), and hindlimbs dorsal side (b, d) and ventral side (f). (C) Forelimbs (a–c) and hindlimbs (d–f) of 1-month-old mice in Fvb inbred background (c, f) and Fvb ϫ CB6F1, outbred background (a, b, d, e) show distinct abnormalities in the transgenic (b, c, e, f). (D) Alizarin red and Alcian blue staining of postnatal day 2 (P2) forelimb shows proportionally shortened digits and the postaxial extra digit (VЈ) in the transgenic animals.

der, 1999), Gremlin (Capdevila et al., 1999; Merino et al., 2000). BMPs also mediate interdigital cell death in the 1999b), DAN (Pearce et al., 1999), Chordin (Francis-West et chick (Kawakami et al., 1996; Macias et al., 1997; Yokouchi al., 1999), Follistatin (Merino et al., 1999a), and Drm et al., 1996; Zou and Niswander, 1996) and induce cell (Pearce et al., 1999) are all expressed in the developing limb. death in murine interdigital explant cultures (Tang et al., BMP signaling is necessary for the proliferation and differ- 2000). However, a role for the BMPs in murine interdigital entiation of precartilaginous mesenchyme of the phalangeal apoptosis in vivo has not been established. Homozygous region in mice (Baur et al., 2000; Yi et al., 2000) and also deletion of Bmp2 or Bmp4 causes lethality in early devel- chondrocyte differentiation in the chick (Duprez et al., opment, so their role in limb apoptosis has not been studied 1996; Macias et al., 1997; Merino et al., 1998). BMPs in mice (Winnier et al., 1995; Zhang and Bradley, 1996). regulate the growth and regression of the AER in the chick Bmp7 knockout mice exhibit mild patterning defects in the (Ganan et al., 1998; Pizette and Niswander, 2000) and limb in the form of preaxial polydactyly (Luo et al., 1995). specify anteroposterior digital identity (Dahn and Fallon, Since BMP2, BMP4, and BMP7 are all expressed in the

© 2002 Elsevier Science (USA). All rights reserved. BMP Signaling in Murine Limb Development 111 interdigital mesenchyme before and during the onset of Western Blot Analysis interdigital apoptosis, loss of the ligand in single knockouts Transgene expression in the adult was determined by Western is also likely to be compensated by other members of the blot analysis of lysates prepared from adult back skin or embryonic family. Similarly, Bmpr1a knockout mice die on embryonic limb homogenates with a rat monoclonal Noggin antibody, RP57- day 9.5 (E9.5) of gestation (Mishina et al., 1995). Recently, a 16, a gift from Regeneron pharmaceuticals. To verify equal loading, conditional mutant of Bmpr1a in the limb demonstrated polyclonal anti-guanine nucleotide dissociation inhibitor (GDI) (a severe limb phenotype due to interference with the forma- gift from Dr. Perry Bickel, Washington University, St. Louis, MO) tion of the AER and dorsoventral patterning of the limb, but was used. a role of BMP signaling in limb necrotic zone apoptosis was not apparent (Ahn et al., 2001). Bmpr1b knockout mice Analysis of Apoptosis show phalangeal defects, but there is no change in the Cell death was detected by vital dye staining with Nile blue apoptosis of autopodial elements (Yi et al., 2000), raising sulfate (Sigma N-5632) as previously described (Tone, 1983). questions about the proapoptotic role of the BMPs in Briefly, embryos were collected in cold phosphate-buffered saline mouse. (PBS), then incubated in prewarmed DMEM containing 0.001% In order to study the role of BMP signaling in limb Nile blue sulfate at 37°C for 1 h, washed in PBS for5hat4°C, and development in vivo, the keratin 14 (K14) promoter was photographed. Apoptotic cells were detected by TUNEL assay on used to drive expression of the BMP antagonist Noggin in 4% paraformaldehyde fixed serial cryosections of E13.5 and E14.5 transgenic mice. These mice were found to have extensive limbs by using the Apoptag–Rhodamine kit (Intergen, Purchase, soft tissue syndactyly in both the forelimb and hindlimb as NY) according to the manufacturer’s protocol. Activated caspase-3 was detected by Cleaved Caspase-3 (Asp175) Antibody (Cell Sig- well as postaxial polydactyly. There was decreased apopto- naling, Beverly, MA). sis in the necrotic zones in the developing limb leading to incomplete regression of the interdigital tissue and possibly a fate change in the region of the posterior necrotic zone to X-Ray Analysis and Skeleton Preparations an extra digit. The specific limb abnormalities generated by Radiographs of anesthetized adult mice were obtained using a inhibiting BMP signaling could be rescued almost com- Faxitron at 35 kV for 30secs. Neonatal and adult mice were stained pletely in the hindlimb and partially in the forelimb by with Alcian Blue and Alizarin red as described (Hogan, 1994). creating a double transgenic misexpressing both Noggin and BMP4 under the K14 promoter, suggesting that the Scanning EM limb abnormalities are a direct result of inhibiting BMP E14.5 embryos were fixed in 2.5% glutaraldehyde, 0.1 M sodium signaling. cacodyalate pH 7.4, dehydrated through a graded series of ethanol and critical point dried using liquid carbon dioxide in a Tousimis Samdri 790 Critical Point Drier (Rockville MD). The dried embryos were then sputter coated with gold-palladium in a Denton Vacuum MATERIALS AND METHODS Desk-1 Sputter Coater (Cherry Hill, NJ) and examined in a JEOL JSM6400 Scanning Electron Microscope (Peabody MA), using an accelerating voltage of 10kV. Generation of K14-Noggin Transgenic and K14- Noggin/BMP Doubly Transgenic Mice In Situ Hybridization and Immunohistochemistry Murine Noggin cDNA was PCR amplified from E19 whole brain In situ hybridization was performed on 4% paraformaldehyde RNA, and the sequence was verified and subcloned into p- fixed whole-mount E10.5–E12.5 embryos as described (Hogan, Bluescript (pBKS-Noggin). The noggin fragment was then sub- 1994). A Noggin antisense probe was prepared by linearizing cloned into the K14-hGH poly(A) plasmid (Cheng et al., 1992) to pBKS-Noggin plasmid and digoxigenin-labeled riboprobe tran- make the K14-Noggin-hGH poly(A) plasmid, which contains the scribed with T3 polymerase. The Fgf8 and Shh probes were kind keratin-14 promoter and the human growth hormone exon/intron gifts from Dr. Gail. R. Martin and Dr. A. P. McMahon, respectively. and poly(A) sequences. The vector was digested, and the K14- The Msx1 and Msx2 probes were obtained from Dr. Cory Abate- Noggin-poly(A) fragment was injected into one-cell FVB embryos at Shen, HoxD10 and 12 probes from Dr. Denis Dubuole, and the Albert Einstein College of Medicine transgenic facility. Two HoxD11 and 13 probes from Dr. Mario R. Capecchi. A monoclonal transgenic founders were identified by Southern blot analysis, and antibody, 4G1, that recognizes both Msx1 and Msx2 (Liem et al., further screening was performed by PCR with a transgene-specific 1995) (DSHB, Iowa City, IA), was used to detect Msx1/2 protein primer pair. The BMP4 cDNA, also amplified from mouse RNA, expression by immunohistochemistry, and a Biorad confocal mi- was similarly subcloned into the K14-hGH poly(A) plasmid and croscope was used to take the images. An antibody against Cyclin injected into CB6F1 one-cell embryos to generate the K14-BMP4 D1 (Neomarkers, Fremont, CA) was used to detect CyclinD1 mice. To generate double transgenic mice, K14-Noggin heterozy- protein expression by immunohistochemistry. To detect phospho- gous males were bred with K14-BMP4 heterozygous females. For Smad1/5 in tissue sections, 4% paraformaldehyde fixed and experiments with timed embryos, noon of the day on which a paraffin-embedded sections were processed for antigen retrieval, vaginal plug was found was considered embryonic day 0.5 (E0.5). using an antigen unmasking solution (Vector, Burlingame, CA) and Amniotic membranes and embryonic tail were used for genotyping immunostained with an antibody that recognizes only phosphory- of E10.5–E14.5 embryos. lated forms of Smad1 and Smad5 (a kind gift of Dr. Peter ten Dijke).

RESULTS in Fig. 3C). The gross morphology as well as radiographic analysis (see Fig. 8B) of the limbs showed that soft tissue Generation of Noggin Misexpressing Mice syndactyly also resulted in increased flexion of the digits of The BMP ligands, BMP2, BMP4, and BMP7, and their both the forelimb and the hindlimb. Cleared skeletal prepa- receptors have overlapping expression domains in the de- rations of postnatal day 2 mice showed the proportionally veloping limb. We inhibited BMP signaling by misexpress- shortened digits and the postaxial extra digit (Fig. 3D). The ing the secreted BMP antagonist Noggin in the ectoderm K14-Noggin transgenic mice were similar to wild type in before the onset of apoptosis in the limb. The K14 pro- weight at birth, but gained less weight thereafter. Various moter, which may drive expression as early as E9.5 in the abnormalities in whisker and hair follicles were observed in ectoderm (Byrne and Fuchs, 1993; Vassar and Fuchs, 1991), the noggin transgenic mice. These will be described else- was used to express Noggin in these transgenic mice (Fig. where. 1A). Two transgenic founder lines were derived by the pronuclear injection of Fvb one-cell-stage embryos and Prevention of Apoptosis in the Limb Necrotic confirmed by genomic Southern analyses (Fig. 1B). Both Zones of Noggin Transgenic Mice lines developed limb abnormalities. The noggin transgene In order to understand the mechanism of soft tissue was overexpressed in both the forelimbs and the hindlimbs syndactyly and postaxial polydactyly in these mice, we first at E11.5 (data not shown), and expression increased at E12.5 investigated cell death in the developing limb bud at E13.5 and thereafter (Fig. 1D, a and b). Noggin protein was also and E14.5. Vital dye staining with Nile blue sulfate showed overexpressed in the limbs at E14.5 (data not shown) and that cell death in the necrotic zones was suppressed in continued to be overexpressed in the skin in the adult by forelimbs and hindlimbs in the transgenic at E13.5 (Fig. 4A) Western blot analysis (Fig. 1C). This expression pattern and E14.5 (data not shown). Accordingly, the TUNEL- makes these transgenic mice particularly suitable for study labeled apoptotic cells were markedly reduced in the inter- of the role of BMP signaling in interdigital apoptosis and digital spaces (Fig. 4B). There was also decreased activated regression of the AER since the misexpression occurs before caspase 3 staining in the transgenic interdigital spaces (Fig. the onset of apoptosis in the limb. In addition, since the 4C). Hence, these transgenic mice provide a mouse model noggin is overexpressed in the skin, the vital structures of where the role of BMP signaling in promoting apoptosis in the embryo are not affected. Similarly, in the BMP4 trans- the various necrotic zones of developing mouse limb can be genic line, the BMP4 transgene was misexpressed under the studied in vivo. same K14 promoter at E12.5 (Fig. 1D, c and d) and expression persisted in the adult (data not shown). In order to determine whether BMP signaling is inhibited in the limbs Msx Expression Is Unaltered in Noggin Transgenic of the Noggin transgenic mice, we analyzed phosphoryla- Mice tion of the BMP responsive Smads in sections of E14.5 Msx2 has been shown to be concomitantly expressed in limbs by using an antibody that recognizes only phosphor- BMP4-mediated apoptotic cell death of neural crest cells in ylated forms of Smad1 and Smad5. Phospho-Smad1/5 im- odd numbered rhombomeres (Graham et al., 1994), in the munoreactivity was significantly reduced in the K14- enamel knot of the tooth (Jernvall et al., 1998), and in other Noggin interdigital spaces (Fig. 2B) compared with the wild systems like cultured P19 cells (Marazzi et al., 1997). Msx1 type (Fig. 2A), indicating that BMP signaling is indeed and Msx2 are also expressed in the necrotic zones of the reduced by misexpression of Noggin in the transgenic mice. developing limbs. Hence, we studied the expression patterns of Msx1 and Msx2 in whole-mount and section in situ Limb Abnormalities and General Phenotype of hybridization in the Noggin transgenic limbs. Interestingly, K14-Noggin Mice there was no significant change in expression patterns of Msx1 and Msx2 transcripts in E12.5 limbs by whole-mount Scanning electron micrographs of E14.5 limb buds demar- in situ hybridization (Fig. 5A). There was also no significant cated a decreased regression of the interdigital tissue in all change in Msx1 and Msx2 transcripts in sections of E14.5 four limbs of Noggin transgenic mice (Fig. 3A). In the transgenic limbs (Fig. 5B, b and d) compared with the wild newborn forelimbs and the hindlimbs, the digits were type (Fig. 5B, a and c). Msx1/2 protein expression was also incompletely separated compared with the wild type (Fig. similar as revealed by immunohistochemistry with an Ј 3B). A postaxial extradigit (labeled V ) was seen in both the antibody that recognizes both Msx1 and Msx2 in E13 forelimb and the hindlimb in the transgenic (Fig. 3B). An transgenic limb sections compared with the wild type (Fig. Ј extra preaxial digit rudiment (labeled I ) was sometimes 5C). seen in the transgenic hindlimb. The interdigital webbing and the postaxial sixth digit were better seen in the adult hindlimb. In the adult forelimb, the syndactyly was so Altered Expression of Fgf8 and Cyclin D1, but Not pronounced that the digits could not be separated (Fig. 3C). Shh in Noggin Transgenic Mice The limb abnormalities were similar in both the Fvb inbred We investigated the expression pattern of genes involved and Fvb ϫ CB6F1 outbred background (compare b, e and c, f in anterior–posterior patterning of the limb and the main-

expression. FGF8 is the most important fibroblast growth factor in the AER required for the initial specification and outgrowth of the limb bud (Moon and Capecchi, 2000). When the limb bud is formed, Fgf8 is expressed along with other Fgfs in the AER, and these are responsible for the proximodistal growth of the limb (Martin, 1998). We examined whether overexpression of noggin in the ectoderm altered expression of Shh in the ZPA and Fgf8 in the AER. There was no significant change in Shh expression in the developing limbs, considering the variation of the size of individual limbs at E11.5 (Fig. 6A). There was no difference in the expression of Fgf8 at E10.5 (data not shown) and E11.5 (Fig. 6B) in the Noggin transgenic compared with the wild type. However, at E12.5, expression of Fgf8 decreased in the wild type forelimb as the AER regressed, whereas it persisted in the transgenic, suggesting that regression of the AER was delayed in the transgenic (Fig. 6B). The BMPs serve the dual roles of inducing apoptosis of undifferentiated mesoderm as well as promoting growth and differentiation of prechondrogenic blastemas (Merino et al., 1998). Proliferation and differentiation of uncommit- ted mesoderm are likely to be controlled by cell cycle regulation. Therefore, we examined the expression pattern of cyclin D1 as a marker of proliferating mesoderm. Inter- estingly, in E14.5 wild type limbs, cyclin D1 was primarily expressed in the precartilagenous mesenchymal condensations. The pattern of cyclin D1 staining in the Noggin transgenic differed from the wild type pattern (Fig. 6C). Cyclin D1-expressing mesenchymal cells in the wild type were predominantly found in the periphery of the condensation and were tightly packed. In the Noggin transgenics, cyclin D1-expressing cells were loosely packed and spread diffusely throughout the mesenchymal condensation. The condensations also appeared smaller in the transgenic at this stage. FIG. 4. Reduced apoptosis in transgenic limbs at E13.5–E14.5. (A) Nile blue sulfate staining in E13.5 forelimbs (a, b) and hindlimbs (c, Similar Expression Pattern of Hoxd10–13 in the d) shows reduced dead cells in the transgenic limbs (b, d) compared Noggin Transgenic Limb with wild type limbs (a, c). (B) Reduced TUNEL-labeled nuclei in E14.5 interdigital tissue of Noggin transgenic hindlimb (b) com- The nested expression pattern of the Hox genes, which pared with wild type (a). (C) Reduced activated caspase-3 staining of encode homeodomain transcription factors, in the develop- E14.5 transgenic hindlimb interdigit (b) compared with wild type ing limb, along with targeted mutagenesis studies of the (a). members of the HoxD complex (from Hoxd9 to Hoxd13) reveal their essential roles in patterning of the limb. Ho- mozygous null mutants of Hoxd11 (Davis and Capecchi, 1996), Hoxd12 (Davis and Capecchi, 1996; Kondo et al., tenance of the AER. Sonic hedgehog (Shh) is the principal 1996, 1997), and Hoxd13 (Dolle et al., 1993) as well as molecule maintaining the activity of the polarizing region transheterozygotes of various combinations of these genes that determines the anteroposterior pattering of the limb (Davis and Capecchi, 1996) result in various abnormalities and the maintenance of the AER (Laufer et al., 1994; in the autopodial elements including the carpal, metacarpal Niswander, 1994). The BMP antagonist Gremlin has been and phalangeal bones. In a number of these mutants, a shown to relay the Shh signal from the ZPA to the AER significant abnormality is the presence of a postaxial extra (Zuniga et al., 1999). It has also been shown that constitu- digit found with variable penetrance. Furthermore, tive expression of Gremlin in the chick limb causes down- HOXD13 mutation in humans causes type II synpolydac- regulation of Shh expression followed by truncations in tyly (SPD), a syndrome that includes postaxial polydactyly distal skeletal elements (Capdevila et al., 1999), suggesting among other abnormalities (Muragaki et al., 1996; Akarsu that BMP signaling is important for maintenance of Shh et al., 1996). We examined whether the postaxial extra digit

FIG. 5. Msx1 and Msx2 expression pattern is unchanged in Noggin transgenic limb. (A) Whole-mount in situ hybridization in E12.5 forelimbs and hindlimbs shows that Msx1 and Msx2 expression is unaltered in K14-Noggin transgenic (b, d and f, h) compared with the wild type (a, c and e, g); anterior is to the left. (B) Section in situ hybridization showing Msx1 and Msx2 expression in E14.5 hindlimb. (C) Msx1/2 protein expression in E13 hindlimb. Propidium iodide (PI) has been used to counterstain nuclei. MC, mesenchymal condensation. © 2002 Elsevier Science (USA). All rights reserved. BMP Signaling in Murine Limb Development 115

is a result of a patterning defect caused by altered expression pattern of the HoxD genes. Hoxd10, Hoxd11, Hoxd12, and Hoxd13 expression was not altered in E12 Noggin transgenic (Figs. 7B, 7D, 7F, and 7H) compared with wild type (Figs. 7A, 7C, 7E, and 7G, respectively) limbs by whole-mount in situ hybridization, suggesting that BMP signaling does not directly regulate expression of these Hox genes.

Rescue of Limb Abnormalities of Noggin Transgenic Mice by BMP4 Misexpression We further wanted to determine whether the limb abnormalities in K14-Noggin mice were the consequence of inhibition of BMP signaling. For this purpose, we misexpressed BMP4 using the same K14 promoter. The K14- BMP4 heterozygous mice do not exhibit overt limb abnormalities. Heterozygous K14-Noggin and K14-BMP4 transgenic mice were mated and double transgenics were obtained that had nearly complete rescue of the hindlimb abnormalities and partial rescue of forelimb abnormalities (Figs. 8A and 8B). In the hindlimb, only a small web between digits II and III remained (arrowhead in Fig. 8A, f). The soft tissue syndactyly and the postaxial extra digit were abrogated. In the forelimb, the digits were clearly separated (compare c and e in Fig. 8A). The digital ﬂexion of the K14-Noggin transgenic was also reversed in the double transgenic (compare c, e and d, f in Fig. 8B).

DISCUSSION

The characterization of limb abnormalities in K14- Noggin transgenic mice coupled with the rescue in the K14-Noggin-BMP4 double transgenic directly demonstrates that BMP signaling is required for regression of interdigital tissue and determination of the fate of necrotic tissues in mammals in vivo. The various targeted mutations of BMP ligands and receptors do not demonstrate the role of BMPs in interdigital mesenchymal apoptosis in the mouse in vivo FIG. 8. Rescue of limb abnormalities in K14-Noggin/BMP4 either due to lethality of the mutations (Bmp2, Bmp4, and double transgenics. (A) Ventral side of forelimbs and hindlimbs of Bmpr1a) or due to the absence of any change in the 2-month-old wild type Fvb mice (a, b), Noggin-transgenic (c, d), and interdigital apoptosis (Bmpr1b and Bmp7). Furthermore, Noggin/BMP4 double transgenic (e, f). (B) Dorsal views of radio- despite many similarities in the mechanisms regulating graphs of 2-month-old wild type Fvb (a, b), Noggin-transgenic (c, d), tetrapod limb development and digit specification in mam- and Noggin/BMP4 double transgenic (e, f) limbs show the excessive flexion of digits in the Noggin-transgenic mice and its partial mals and avians, there are significant differences in the fate rescue in the double transgenic mice. of interdigital tissues. For example, FGF4, which inhibits

FIG. 6. Altered expression of Fgf8 and Cyclin D1, but not Shh, in Noggin-transgenic animals. (A) Shh expression domain is relatively normal in E11.5 transgenic limbs. (B) Unaltered Fgf8 expression in E11.5 transgenic AER but persistent expression in the forelimbs of E12.5 transgenic limbs; posterior is to the left. (C) Differences in cyclin D1 expression pattern in the mesenchymal condensation of proximal phalange of E14.5 wild type and Noggin transgenic hindlimbs. FIG. 7. HoxD10–13 expression is not altered in K14-Noggin transgenic limbs. Whole-mount in situ hybridization of E12 limbs with antisense probes against HoxD10, HoxD11, HoxD12, and HoxD13 shows that the expression of these genes is unchanged in the K14-Noggin transgenic (B, D, F, H) compared with the wild type (A, C, E, G).

© 2002 Elsevier Science (USA). All rights reserved. BMP Signaling in Murine Limb Development 117 the formation of ectopic digits in the chick, induces digit apoptosis between digits 2 and 5 of the mouse Hammertoe bifurcation in the mouse, and TGF␤ similarly has divergent mutant that resulted in soft tissue syndactyly of all four actions on digit formation (Ngo-Muller and Muneoka, limbs, but polydactyly is not reported (Zakeri et al., 1994). 2000). Clarification of the role of BMPs in mammalian limb Similarly homozygous deletion of Apaf1 (Cecconi et al., development therefore requires examination of their func- 1998) leads to interdigital webbing without polydactyly, as tion in the mouse in vivo. We used the strategy of misex- does double deletion of both Bax and Bak (Lindsten et al., pression of the BMP inhibitor Noggin in the developing 2000). These observations, along with reduced activated limb before the onset of necrotic zone apoptosis to inhibit caspase-3 staining in the noggin transgenic limbs, suggest BMP2, BMP4, and BMP7, which are coexpressed in the that BMP signaling directly or indirectly regulates the necrotic zones. expression of these proapoptotic genes. Contrary to this, in Apoptosis in the developing limb necrotic zones was Bmp7 homozygous and Bmp4 heterozygous mutants, inter- decreased in the noggin transgenic animals with incom- digital regression is not affected, but there is preaxial plete regression of interdigital tissue. Further, coexpression polydactyly with incomplete penetrance analogous to the of BMP4 in double transgenic animals restored the inter- rudimentary extra preaxial digit observed in Noggin trans- digital apoptosis. Thus, BMP signaling is causally impli- genic mice. Hence, it is unclear whether the axial polydac- cated in the apoptotic process in mammals, consistent with tyly in the various Bmp mutants along with the predomi- prior findings in avians (Yokouchi et al., 1996; Zou and nant postaxial polydactyly in the Noggin transgenic mice Niswander, 1996). However, there was no significant are due only to inhibition of programmed cell death; in the change in expression pattern of Msx1 and Msx2 in the Bmp7 knockout, for example, there is altered pattern of Noggin transgenic limbs, although Msx2 has been impli- Hoxd13 expression (Luo et al., 1995). Moreover, postaxial cated in BMP mediated interdigital apoptosis in the chick. polydactyly is a feature of various mutants of Hoxd11–13 in BMPs have been implicated to regulate Msx2 expression the mouse and HOXD13 in humans. But Hoxd10–13 ex- since misexpression of dominant negative BMP receptor pression is not significantly altered in the Noggin trans- can reduce Msx2 expression (Zou and Niswander, 1996), genic, suggesting that postaxial polydactyly and preaxial and treatment of cultured sympathetic neuroblasts with digit rudiment in the Noggin transgenic may be a direct BMP4 increases Msx2 expression (Gomes and Kessler, result of apoptosis inhibition by Noggin misexpression. 2001). The fact that Msx1 and 2 expression is unaltered in These data also indicate that Noggin-mediated inhibition of the Noggin transgenic mice may indicate that Msx1/2 are BMP signaling may be involved in a fate change in the not downstream of the BMPs in directly mediating apopto- posterior and anterior necrotic zones which respond to sis but rather may simply be required as a cofactor for some additional cues to form bony elements. BMP-mediated apoptosis to occur. Alternatively, Msx2 may By contrast, interdigital webbing induced by inhibition of mediate apoptosis by maintaining the expression of some of genes that may act upstream of the BMPs is accompanied the BMPs. Ectopic expression of Msx2 in the chick can by polydactyly. For example, single and double mutants of induce Bmp4 (Ferrari et al., 1998). In Msx1 and Msx2 double retinoic acid receptor genes have various degrees of inter- mutant mice, where there is inhibition of interdigital digital webbing due to decreased apoptosis in the INZ apoptosis, there is also downregulation of Bmp4, but not (Dupe et al., 1999) as well as preaxial polydactyly. Retinoic Bmp2 and Bmp7 (Chen and Zhao, 1998; and unpublished acid can stimulate apoptosis in interdigital regions of cul- observations). This may indicate some inherent differences tured limbs (Lussier, 1993) and appears to promote apopto- in the role of Msx1/2 in BMP-mediated apoptosis in the sis by inducing BMPs like BMP7 and BMP4 in the interdigi- mouse as opposed to the chick. Another explanation of tal mesenchyme (Rodriguez-Leon et al., 1999), thereby unaltered Msx1/2 expression in the Noggin transgenic is acting upstream of BMP signaling. Extra digit formation that there is partial inhibition of BMP signaling by Noggin, following retinoic acid antagonist-mediated repression of and the residual signaling is sufficient to mediate Msx gene interdigital apoptosis suggests that retinoic acid is neces- expression, although insufficient to mediate complete in- sary for repressing BMP mediated chondrogenesis in the terdigital apoptosis. In this regard, it is noteworthy that interdigital tissue. there is some phospho-Smad1/5 labeling in the interdigital There is also persistent expression of Fgf8 in the AER of mesenchyme in the Noggin transgenic, although signifi- transgenic animals, suggesting that there is a delay in cantly lower than that in the wild type (Fig. 2). Alterna- regression of the AER, thus providing an opportunity for tively, Msx gene expression could be induced in the Noggin continued proliferation of distal mesenchyme and perhaps transgenic mice by a ligand other than Noggin-inhibited prolonged cartilage formation. However, it is interesting to BMPs (BMP2, 4, and 7). note that this did not lead to an increase in length or There were additional abnormalities in the noggin trans- number of the phalangeal elements, instead the digits were genic animals, including formation of an extra postaxial proportionally shortened. Nuclear cyclin D1 is a marker of digit, rudiments of an extra preaxial digit, and a delay in the condensing mesenchyme and cyclin D1 immunoreactive regression of the AER. These abnormalities do not always mesenchyme was less condensed in Noggin transgenic accompany a reduction in interdigital apoptosis and syn- mice than in wild type mice. BMP signaling is necessary for dactyly in mice. For example, there is decreased interdigital the proliferation of chondrocytes that have differentiated

© 2002 Elsevier Science (USA). All rights reserved. 118 Guha et al. from precartilaginous mesenchyme of the phalangeal region Brunet, L. J., McMahon, J. A., McMahon, A. P., and Harland, R. M. in mice (Baur et al., 2000; Yi et al., 2000) and for their (1998). Noggin, cartilage morphogenesis, and joint formation in further differentiation (Pizette and Niswander, 2000). the mammalian skeleton. Science 280, 1455–1457. Hence, the less condensed cyclin D1 immunoreactive mes- Byrne, C., and Fuchs, E. (1993). Probing keratinocyte and differen- enchyme in the transgenic animals suggests either a delay tiation speciﬁcity of the human K5 promoter in vitro and in in chondrocyte differentiation or reduced proliferation of transgenic mice. Mol. Cell. Biol. 13, 3176–3190. the condensed mesenchyme. Thus, the requirement for Capdevila, J., and Johnson, R. L. (1998). Endogenous and ectopic BMP signaling in chondrogenesis may mask the potential of expression of noggin suggests a conserved mechanism for regulation of BMP function during limb and somite patterning. Dev. persistent FGF signaling from the AER to affect phalangeal Biol. 197, 205–217. growth or pattern. Capdevila, J., Tsukui, T., Rodriquez Esteban, C., Zappavigna, V., In summary, this study has used a strategy of misexpres- and Izpisua Belmonte, J. C. (1999). Control of vertebrate limb sion of the BMP antagonist Noggin to directly demonstrate outgrowth by the proximal factor Meis2 and distal antagonism of the role of BMP signaling in murine limb development in BMPs by Gremlin. Mol. Cell 4, 839–849. vivo. It further demonstrates the utility of the double Cecconi, F., Alvarez-Bolado, G., Meyer, B. I., Roth, K. A., and transgenic misexpressors to dissect the biologic role of Gruss, P. (1998). Apaf1 (CED-4 homolog) regulates programmed these growth factors. These animals will also be helpful for cell death in mammalian development. Cell 94, 727–737. deﬁning other effects of the BMPs, such as their roles in the Chen, Y., and Zhao, X. (1998). Shaping limbs by apoptosis. J. Exp. development of hair follicles and of the peripheral nerve Zool. 282, 691–702. innervation of developing skin. Cheng, J., Turksen, K., Yu, Q. C., Schreiber, H., Teng, M., and Fuchs, E. (1992). Cachexia and graft-vs.-host-disease-type skin changes in keratin promoter-driven TNF alpha transgenic mice. ACKNOWLEDGMENTS Genes Dev. 6, 1444–1456. Dahn, R. D., and Fallon, J. F. (2000). Interdigital regulation of digit We thank Dr. Lee A. Niswander and Dr. Henry M. Kronenberg identity and homeotic transformation by modulated BMP signal- for helpful discussions and comments on the manuscript. We are ing. Science 289, 438–441. also grateful to Dr. Frank Macaluso for help with the scanning EM Davis, A. P., and Capecchi, M. R. (1996). A mutational analysis of and the transgenic facility of Albert Einstein College of Medicine the 5Ј HoxD genes: Dissection of genetic interactions during for transgene injections. We thank Drs. Elaine Fuchs for the limb development in the mouse. Development 122, 1175–1185. K14-promoter plasmid; A. P. McMahon, Gail R. Martin, Cory Dewulf, N., Verschueren, K., Lonnoy, O., Moren, A., Grimsby, S., Abate-Shen, Mario R. Capecchi, and Denis Duboule for probes; Dr. Vande Spiegle, K., Miyazono, K., Huylebroeck, D., and Ten Aris Economides and Regeneron pharmaceuticals for the Noggin Dijke, P. (1995). Distinct spatial and temporal expression pat- antibody; Dr. Perry Bickel for the GDI antibody; and Dr. Peter ten terns of two type I receptors for bone morphogenetic proteins Dijke for p-Smad1/5 antibody. The monoclonal antibody, 4G1, during mouse embryogenesis. Endocrinology 136, 2652–2663. against Msx1/2 was obtained from the Developmental Studies Dolle, P., Dierich, A., LeMeur, M., Schimmang, T., Schuhbaur, B., Hybridoma Bank developed under the auspices of the NICHD and Chambon, P., and Duboule, D. (1993). Disruption of the HoxD13 maintained by The University of Iowa, Department of Biological gene induces localized heterochrony leading to mice with neo- Sciences, Iowa City, IA 52242. This work was supported by grants tenic limbs. Cell 75, 431–441. from the NIH (RO1 NS20013 and RO1 NS20778) (to J.A.K.), and by Dudley, A. T., and Tabin, C. J. (2000). Constructive antagonism in the Davee Foundation and the Feinberg Clinical Neuroscience limb development. Curr. Opin. Genet. Dev. 10, 387–392. Institute. The work was also supported by NIH Grants (RO1 Dupe, V., Ghyselinck, N. B., Thomazy, V., Nagy, L., Davies, P. J. A., CA70896, RO1 CA75503 and RO1 CA86072) (to R.G.P.) and Chambon, P., and Mark, M. (1999). Essential roles of retinoic acid awards from the Susan G. Komen Breast Cancer Foundation and Breast Cancer Alliance Inc. R.G.P. is a Monique Weill-Caulier and signaling in interdigital apoptosis and control of BMP-7 expres- Irma T. Hirschl Scholar. sion in mouse autopods. Dev. Biol. 208, 30–43. Duprez, D., Bell, E. J., Richardson, M. K., Archer, C. W., Wolpert, L., Brickell, P. M., and Francis-West, P. H. (1996). Overexpression REFERENCES of BMP-2 and BMP-4 alters the size and shape of developing skeletal elements in the chick limb. Mech. Dev. 57, 145–157. Ferrari, D., Lichtler, A. C., Pan, Z. Z., Dealy, C. N., Upholt, W. B., Ahn, K., Mishina, Y., Hanks, M. C., Behringer, R. R., and Cren- and Kosher, R. A. (1998). Ectopic expression of Msx-2 in posterior shaw, E. B., 3rd. (2001). BMPR-IA signaling is required for the formation of the apical ectodermal ridge and dorsal-ventral limb bud mesoderm impairs limb morphogenesis while inducing patterning of the limb. Development 128, 4449–4461. BMP-4 expression, inhibiting cell proliferation, and promoting Akarsu, A. N., Stoilov, I., Yilmaz, E., Sayli, B., and Sarfarazi, M. apoptosis. Dev. Biol. 197, 12–24. (1996). Genomic structure of HOXD 13 gene: A nine polyalanine Francis-West, P. H., Parish, J., Lee, K., and Archer, C. W. (1999). duplication causes synpolydactyly in two unrelated families. BMP/GDF-signalling interactions during synovial joint develop- Hum. Mol. Genet. 5, 945–952. ment. Cell Tissue Res. 296, 111–119. Baur, S. T., Mai, J. J., and Dymecki, S. M. (2000). Combinatorial Ganan, Y., Macias, D., Basco, R. D., Merino, R., and Hurle, J. M. signaling through BMP receptor IB and GDF5: Shaping of the (1998). Morphological diversity of the avian foot is related with distal mouse limb and the genetics of distal limb diversity. the pattern of msx gene expression in the developing autopod. Development 127, 605–619. Dev. Biol. 196, 33–41.

Gomes, W. A., and Kessler, J. A. (2001). Msx-2 and p21 mediate the Marazzi, G., Wang, Y., and Sassoon, D. (1997). Msx2 is a transcrip- pro-apoptotic but not the anti-proliferative effects of BMP4 on tional regulator in the BMP4-mediated programmed cell death cultured sympathetic neuroblasts. Dev. Biol. 237, 212–221. pathway. Dev. Biol. 186, 127–138. Graham, A., Francis-West, P., Brickell, P. M., and Lumsden, A. Martin, G. R. (1998). The role of FGFs in the early development of (1994). The signaling molecule BMP4 mediates apoptosis in the vertebrate limbs. Genes Dev. 12, 1571–1586. rhombencephalic neural crest. Nature 372, 684–686. Merino, R., Ganan, Y., Macias, D., Economides, A. N., Sampath, Hogan, B. L. (1996). Bone morphogenetic proteins in development. K. T., and Hurle, J. M. (1998). Morphogenesis of digits in the Curr. Opin. Genet. Dev. 6, 432–438. avian limb is controlled by FGFs, TGFbetas, and noggin through Hogan, B. L., Beddington, R., Constantini, F., and Lacy, E. (1994). BMP signaling. Dev. Biol. 200, 35–45. “Manipulating the Mouse Embryo: A Laboratory Manual.” Cold Merino, R., Macias, D., Ganan, Y., Rodriguez-Leon, J., Economides, Spring Harbor Laboratory Press, Cold Spring Harbor, NY. A. N., Rodriguez-Esteban, C., Izpisua-Belmonte, J. C., and Hurle, Jernvall, J., Aberg, T., Kettunen, P., Keranen, S., and Thesleff, I. J. M. (1999a). Control of digit formation by activin signalling. Development 126, 2161–2170. (1998). The life history of an embryonic signaling center: BMP-4 Merino, R., Rodriguez-Leon, J., Macias, D., Ganan, Y., Economides, induces p21 and is associated with apoptosis in the mouse tooth A. N., and Hurle, J. M. (1999b). The BMP antagonist Gremlin enamel knot. Development 125, 161–169. regulates outgrowth, chondrogenesis and programmed cell death Kawakami, Y., Ishikawa, T., Shimabara, M., Tanda, N., Enomoto- in the developing limb. Development 126, 5515–5522. Iwamoto, M., Iwamoto, M., Kuwana, T., Ueki, A., Noji, S., and Mishina, Y., Suzuki, A., Ueno, N., and Behringer, R. R. (1995). Nohno, T. (1996). BMP signaling during bone pattern determina- Bmpr encodes a type I bone morphogenetic protein receptor that Development tion in the developing limb. 122, 3557–3566. is essential for gastrulation during mouse embryogenesis. Genes Kondo, T., Dolle, P., Zakany, J., and Duboule, D. (1996). Function Dev. 9, 3027–3037. of posterior HoxD genes in the morphogenesis of the anal Moon, A. M., and Capecchi, M. R. (2000). Fgf8 is required for sphinctor. Development 122, 2651–2659. outgrowth and patterning of the limbs. Nat. Genet. 26, 455–459. Kondo, T., Zakany, J., Innis, J. W., and Duboule, D. (1997). Control Muragaki, Y., Mundlos, S., Upton, J., and Olsen, B. (1996). Altered of colinearity in AbdB genes of the mouse HoxD complex. Mol. growth and branching patterns in synpolydactyly caused by Cell 1, 289–300. mutations in HOXD13. Science 272, 548–551. Krabbenhoft, K. M., and Fallon, J. (1992). Talpid2 limb bud meso- Niswander, L., Jefrey, S., Martin, G. R., and Tickle, C. (1994). A derm does not express Ghox-8 and has analteredexpression positive feedback loop coordinates growth and patterning in the pattern of GHox-7. Dev. Dyn. 194, 52–62. vertebrate limb. Nature 371, 609–612. Laufer, E., Nelson, C. E., Johnson, R. L., Morgan, B. A., and Tabin, Pearce, J. J., Penny, G., and Rossant, J. (1999). A mouse cerberus/ C. (1994). Sonic hedgehog and Fgf-4 act through a signaling Dan-related gene family. Dev. Biol. 209, 98–110. cascade and feedback loop to integrate growth and patterning of Pizette, S., and Niswander, L. (1999). BMPs negatively regulate the developing limb bud. Cell 79, 993–1003. structure and function of the limb apical ectodermal ridge. Laufer, E., Pizette, S., Zou, H., Orozco, O. E., and Niswander, L. Development 126, 883–894. (1997). BMP expression in duck interdigital webbing: A reanaly- Pizette, S., and Niswander, L. (2000). BMPs are required at two sis. Science 278, 305. steps of limb chondrogenesis: Formation of prechondrogenic Liem, K. F., Jr., Tremml, G., Roelink, H., and Jessell, T. M. (1995). condensations and their differentiation into chondrocytes. Dev. Dorsal differentiation of neural plate cells induced by BMP- Biol. 219, 237–249. mediated signals from epidermal ectoderm. Cell 82, 969–979. Pizette, S., and Niswander, L. (2001). Early steps in limb patterning Lindsten, T., Ross, A. J., King, A., Zong, W. X., Rathmell, J. C., and chondrogenesis. Novartis Found. Symp. 232, 23–36. Shiels, H. A., Ulrich, E., Waymire, K. G., Mahar, P., Frauwirth, Rodriguez-Leon, J., Merino, R., Macias, D., Ganan, Y., Santesteban, K., Chen, Y., Wei, M., Eng, V. M., Adelman, D. M., Simon, M. C., E., and Hurle, J. M. (1999). Retinoic acid regulates programmed Nat. Cell Biol. Ma, A., Golden, J. A., Evan, G., Korsmeyer, S. J., MacGregor, cell death through BMP signalling. 1, 125–126. Tang, M. K., Leung, A. K., Kwong, W. H., Chow, P. H., Chan, J. Y., G. R., and Thompson, C. B. (2000). The combined functions of Ngo-Muller, V., Li, M., and Lee, K. K. (2000). Bmp-4 requires the proapoptotic Bcl-2 family members bak and bax are essential for presence of the digits to initiate programmed cell death in limb normal development of multiple tissues. Mol. Cell 6, 1389–1399. interdigital tissues. Dev. Biol. 218, 89–98. Luo, G., Hofmann, C., Bronckers, A. L., Sohocki, M., Bradley, A., Tone, S., Kanaka, S., and Kato, Y. (1983). The inhibitory effect of and Karsenty, G. (1995). BMP-7 is an inducer of nephrogenesis, 5-bromodeoxyuridine on the programmed cell death in the chick and is also required for eye development and skeletal patterning. limb. Dev. Growth Differ. 25, 381–391. Genes Dev. 9, 2808–2820. Vassar, R., and Fuchs, E. (1991). Transgenic mice provide new Lussier, M., Canoun, C., Ma, C., Sank, A., and Shuler, C. (1993). insights into the role of TGF-alpha during epidermal develop- Interdigital soft tissue seperation induced by retinoic acid in ment and differentiation. Genes Dev. 5, 714–727. mouse limbs cultured in vitro. Int. J. Dev. Biol. 37, 555–564. Winnier, G., Blessing, M., Labosky, P. A., and Hogan, B. L. (1995). Lyons, K. M., Hogan, B. L., and Robertson, E. J. (1995). Colocaliza- Bone morphogenetic protein-4 is required for mesoderm formation of BMP 7 and BMP 2 RNAs suggests that these factors tion and patterning in the mouse. Genes Dev. 9, 2105–2116. cooperatively mediate tissue interactions during murine devel- Yi, S. E., Daluiski, A., Pederson, R., Rosen, V., and Lyons, K. M. opment. Mech. Dev. 50, 71–83. (2000). The type I BMP receptor BMPRIB is required for chondro- Macias, D., Ganan, Y., Sampath, T. K., Piedra, M. E., Ros, M. A., genesis in the mouse limb. Development 127, 621–630. and Hurle, J. M. (1997). Role of BMP-2 and OP-1 (BMP-7) in Yokouchi, Y., Sakiyama, J., Kameda, T., Iba, H., Suzuki, A., Ueno, programmed cell death and skeletogenesis during chick limb N., and Kuroiwa, A. (1996). BMP-2/-4 mediate programmed cell development. Development 124, 1109–1117. death in chicken limb buds. Development 122, 3725–3734.

Zakeri, Z., Quaglino, D., and Ahuja, H. S. (1994). Apoptotic cell the formation and differentiation of cartilage. Genes Dev. 11, death in the mouse limb and its suppression in the hammertoe 2191–2203. mutant. Dev. Biol. 165, 294–297. Zuniga, A., Haramis, A. P., McMahon, A. P., and Zeller, R. (1999). Zhang, H., and Bradley, A. (1996). Mice deﬁcient for BMP2 are Signal relay by BMP antagonism controls the SHH/FGF4 feed- nonviable and have defects in amnion/chorion and cardiac de- back loop in vertebrate limb buds. Nature 401, 598–602. velopment. Development 122, 2977–2986. Zou, H., and Niswander, L. (1996). Requirement for BMP signaling in Received for publication January 23, 2002 interdigital apoptosis and scale formation. Science 272, 738–741. Revised May 16, 2002 Zou, H., Wieser, R., Massague, J., and Niswander, L. (1997). Accepted June 11, 2002 Distinct roles of type I bone morphogenetic protein receptors in Published online July 31, 2002