During limb outgrowth, signaling by bone morphogeneticproteins (BMPs) must be moderated to maintain the signalingloop between the zone of polarizing activity (ZPA) and theapical ectodermal ridge (AER). Gremlin, an extracellular BMPantagonist, has been proposed to fulfill this function andtherefore be important in limb patterning. We tested thismodel directly by mutating the mouse gene encoding gremlin(Cktsf1b1, herein called gremlin). In the mutant limb, thefeedback loop between the ZPA and the AER is interrupted,resulting in abnormal skeletal pattern. We also show that the
gremlin mutation is allelic to the limb deformity mutation (ld).Although BMPs and their antagonists have multiple roles inlimb development, these experiments show that gremlin is theprincipal BMP antagonist required for early limb outgrowthand patterning.
During development of the vertebrate limb, signals from the AERand the ZPA direct outgrowth and patterning of the limb skeletalelements1–3. The AER is a strip of columnar cells at the distal edgeof the limb bud that secretes multiple fibroblast growth factors
1Department of Molecular and Cell Biology, University of California-Berkeley, 401 Barker Hall, Berkeley, California 94720, USA. 2Present addresses: Renovis, 270Littlefield Avenue, South San Francisco, California 94080, USA (D.H.) and Department of Microbiology and Immunology, Stanford University, California 94305, USA(M.S.D.). 3These authors contributed equally to this work. Correspondence should be addressed to R.M.H. ([email protected]).
Gremlin is the BMP antagonist required for maintenanceof Shh and Fgf signals during limb patterningMustafa K Khokha1,3, David Hsu1–3, Lisa J Brunet1, Marc S Dionne1,2 & Richard M Harland1
NATURE GENETICS VOLUME 34 | NUMBER 3 | JULY 2003 303
NheI
lacZ neoDTA
neolacZ
5' Probe 3' ProbeFRTFRT
FRT FRT
Targetingcassette
Genomicregion
Null allele
XbaI Eco
RV
XmnI
EcoRV
NheI
XbaI
NheI
XmnI
EcoRV
gremlin
XmnI
EcoRV
NheI
EcoRV
NheI
XbaI
XmnI
EcoRV
XmnI
XmnI
EcoRV/5' Probe XmnI/3' Probe
10 kb
4 kb8 kb
6 kb
11 kb
+/+ +/+ +/– +/+ +/–+/–∆neogremlin mRNA LacZ
+/– +/–
XmnI
a
b c
1 2 3 4 5 6
Figure 1 Targeting of the gremlin locus. (a) Mapof the targeting vector, gremlin locus and themutated allele produced by homologousrecombination in ES cells. DTA, diptheria toxincassette for negative selection; neo, the PGK-neoR cassette for positive selection flanked byFRT sites. An EcoRV–NheI fragment was usedas a Southern-blot probe to confirm correct 5′targeting. An EcoRV–XmnI fragment was used toconfirm 3′ targeting. (b) Southern-blot analysisof genomic DNA to identify heterozygotes. The5′ probe detects a 10-kb band in the wild-typeallele (lanes 1,2) and a 4-kb fragment in the nullallele (lane 3) when genomic DNA is digestedwith EcoRV. A small band cross-hybridizes and isdetected in all lanes. For the 3′ end, the 3′ probedetects an 8-kb band in the wildtype allele (lane5) and a 6-kb band in the null allele (lane 6)when genomic DNA is digested with XmnI.When the PGK-neoR cassette is removed by Flprecombinase, the 3′ probe detects a longer(11-kb) fragment (lane 4). (c) Expression ofgremlin in the limbs, somites and flank of theembryo. LacZ staining of heterozygous embryosshows a similar pattern. Embryo heads wereremoved for genotyping.
(Fgfs; refs. 4,5). In particular, the AER secretes Fgf4 and Fgf8,which together are required for limb outgrowth6. The ZPA, locatedin the posterior mesenchyme of the limb bud, secretes Sonic hedge-hog (Shh), which defines antero-posterior pattern in the limb7,8.These two signaling centers can reciprocally induce and maintaineach other’s gene expression9,10. BMPs have a negative effect on thisfeedback loop, and experimental suppression of endogenous BMPactivity prolongs and reinforces AER activity11,12. It has been pro-posed that an endogenous BMP antagonist may fulfill such a role invivo. A number of BMP antagonists are expressed in the limb,including noggin, gremlin, dan, chordin and follistatin13–16. Theexpression of some of these BMP antagonists suggests that theymay have redundant functions. Loss of dan, a highly conservedBMP antagonist, has no effect on limb outgrowth or patterning17.Expression of other BMP antagonists is highly localized. For exam-ple, noggin is expressed in cartilage condensations and is required
for the proper allocation of limb mesenchyme to the cartilage com-partment and for specification of joints13. But noggin is not neededfor early limb outgrowth and patterning. To date, no BMP antago-nist has been shown to be required for maintenance of the Shh–Fgffeedback loop.
The limb bud mesenchyme broadly expresses gremlin, a memberof the dan family of BMP antagonists18–21 (Fig. 1c). Gremlin expres-sion is restricted to the distal limb bud mesenchyme and concen-trated posteriorly. Also, gremlin expression is lost in the ld (limbdeformity) mutant mouse19, in which the AER regresses prema-turely22 and the Shh–Fgf feedback loop is disrupted23,24. These find-ings suggest that gremlin may have an important role in maintainingthe AER and the Fgf–Shh feedback loop (see Fig. 3c; ref. 19). To testthis hypothesis directly, we mutated the mouse gene gremlin, and weshow here that gremlin is required for correct limb patterning andcontinued expression of Fgfs and Shh. We also show that the gremlinmutation is allelic to ld.
To produce a null allele at the gremlin locus, we replaced the openreading frame, which lies on a single exon, with an in-frame fusion oflacZ to the sequence encoding the first five residues of gremlin (Fig. 1a).We confirmed proper targeting by Southern-blot analysis (Fig. 1b).Additionally, the expression of LacZ in heterozygous mice recapitulatesthe expression pattern of gremlin (Fig. 1c). Because the selectablemarker could complicate the interpretation of the mutation, we flankedthe neoR gene with FRT sites. We mated heterozygous mice with miceexpressing Flp recombinase to remove the PGK-neoR cassette and con-firmed this by Southern-blot analysis (Fig. 1b). Phenotypes of mice withor without the PGK-neoR cassette were identical, and we used theminterchangeably. Heterozygotes were fertile and viable, whereas almostall homozygotes died within 48 h of birth owing to absence of kidneys.
Heterozygotes had no observable abnormalities in limb develop-ment when stained with alcian blue and alizarin red (Fig. 2q,s), buthomozygotes had a conspicuous skeletal phenotype that is fully pene-trant (Fig. 2r,t). In the zeugopod of both the forelimb and hindlimb,only a single bone was present. Also, both the forelimb and hindlimbhad fewer digits and abnormal maintenance of interdigital tissue (softtissue syndactyly; data not shown). The cartilage and bone mass werenot excessive, however, and joints appeared normal, in contrast tomice with mutations of noggin13 (Fig. 2q–t).
304 VOLUME 34 | NUMBER 3 | JULY 2003 NATURE GENETICS
a b c d
e f g h
i j k l
m n o p
q r s t
Figure 2 Serial skeletal stains of embryos and newborn pups. Lateralviews of right limbs stained with alcian blue (a–p) or with alcian blue andalizarin red (q–t). Alcian blue stains cartilage and alizarin red stains bone.(a,e,i,m,q) Forelimbs from wild-type or heterozygous embryos at theindicated embryonic days (E) and from newborn pups (NB). (b,f,j,n,r)Forelimbs from mutant embryos. (c,g,k,o,s) Hindlimbs from wild-type orheterozygous embryos. (d,h,l,p,t) Hindlimbs from mutant embryos. Scalebars indicate 1 mm and are accurate across each row of panels.
FGF
Ant
Post
Prox Dist
Shh
Grem
lin
BMPFormin
Fgf4
Fgf8
Shh
Ptch
+/+;+/– –/–
+/+;+/– –/–
Forelimb Hindlimb
a b c
d e
f g h i
k l mj
Figure 3 Shh–Fgf feedback loop analysis. Expression of Fgf4 in wild-type(a) and mutant embryos (b). Expression of Fgf4 was lost in mutant AER butpersisted in the somites (arrowhead). (c) Schematic of the Shh–Fgf signalingloop in the developing limb. Ant, anterior; Dist, distal; Post, posterior; Prox,proximal. (d,e) Expression of Fgf8 in wild-type (d) and mutant limbs (e).(a,b,d,e) Oblique dorsolateral views to show AER staining of the forelimbs.(f–m) Dorsal views of limb buds from E10.5 embryos. (f–i) Expression ofShh in wild-type forelimbs (f), mutant forelimbs (g), wild-type hindlimbs (h)and mutant hindlimbs (i). Arrows indicate persistent expression of Shh innotochord and floorplate. (j–m) Expression of Ptch in wild-type forelimbs (j),mutant forelimbs (k), wild-type hindlimbs (l) and mutant hindlimbs (m).
Figure 5 Expression of Fmn andcomplementation of ld with gremlin.(a–c) Dorsal views of limb buds from E10.5embryos. Expression of Fmn in limb budfrom heterozygous (a) and mutant embryos (b).(c) Fmn sense control on wild-type limbindicating background staining. (d) Genomicorganization of Fmn and gremlin. Fmn spansover 400 kb and only the very 3′ end isdepicted. Vertical bars indicate exons. Grayarrowheads indicate direction of transcription.The top diagram depicts two identified ldmutations. The bottom diagram depicts thegremlin null allele. If the mutations in Fmnlead to a loss of gremlin expression through acis-regulatory mechanism (red arrow) then thisconfiguration will lead to a limb deformitydefect. (e,f) Progeny from a ldJ heterozygote ×gremlin heterozygote mating. Lateral views of right forelimbs from E14.5 embryos stained with alcian blue. (e) Embryo with a wild-type phenotype.(f) Embryo with the ld phenotype that is heterozygous with respect to gremlin (as determined by Southern blotting; data not shown) and, therefore, hasthe genotype ldgrm/ld J. Of 18 embryos scored from ldgrm × ld J matings, 5 had the ld phenotype.
L E T T E R S
To address the development of the single bone in the zeugopod, weanalyzed skeletal stains (using alcian blue) at various stages of develop-ment (Fig. 2). In the forelimb, two bones were initially formed in thezeugopod but these bones subsequently fused. Although the forelimbzeugopod appeared to be correctly patterned initially, the hindlimb zeu-gopod never developed the appropriate pattern; even at embryonic day(E) 12.5, only a single bone was present in the zeugopod. Thus, the out-come, though similar, develops differently in forelimbs and hindlimbs.
To investigate the possible reasons for the alterations in limb patternseen in the gremlin mutant mice, we examined expression of compo-nents of the Fgf–Shh feedback loop9,10,19 (Fig. 3c). Owing to loss ofgremlin activity, unopposed BMP signaling might result in reducedexpression of Fgfs11,12. Although multiple Fgfs are expressed in theAER, we examined only Fgf4 and Fgf8, which together are required forlimb outgrowth6. In limbs from gremlin mutant mice, expression ofFgf4 was undetectable in the AER but persisted in the somites (Fig.3a,b). Expression of Fgf8 was reduced and disorganized (Fig. 3e).These results appear identical in both forelimbs and hindlimbs (Fig. 3and data not shown). In turn, disruption of Fgf signaling should leadto reduced expression of Shh (Fig. 3c). In the forelimb, expression ofShh was markedly lower although it was detectable (Fig. 3f,g). In thehindlimb, however, expression of Shh was nearly undetectable (Fig.3h,i). The near-absence of Shh expression in the hindlimb of gremlinmutant mice may account for the severe phenotype and its similarityto the Shh-null mutant8,25. Thus, the principal consequence of loss ofgremlin in the limb may be loss of Shh expression.
Patched 1 (encoded by Ptch) is a downstream target of Shh26, andreduction of Ptch expression correlated with the loss of Shh expressionin gremlin mutant mice (Fig. 3j–m). This finding provides further evi-dence that the reduction of Shh mRNA expression also correlates witha decrease in Shh signal, resulting in the difference in phenotypebetween the forelimb and hindlimb.
In the Shh–Fgf feedback loop model, excess BMP signal must bereduced to maintain AER activity and the feedback loop. Our experi-ments show that loss of gremlin activity disrupts this Shh–Fgf feed-back loop. To investigate whether the limb bud of gremlin mutant micehas increased BMP signaling, we examined whether expression ofgenes encoding BMPs in the limb was altered (Fig. 4). Expression ofBmp7 seemed unchanged in mutant limbs (Fig. 4c,d), but, consistentwith the loss of the AER, expression of Bmp4 was reduced in mutantlimb buds (Fig. 4a,b). Nevertheless, overall BMP signal may still beincreased in the limbs of gremlin mutant mice owing to the loss ofBMP antagonism by gremlin. We therefore analyzed the expression ofthe BMP signaling targets Msx1 and Msx2 (ref. 12). In gremlin mutantmice, expression of Msx1 was slightly higher, most notably in the distalposterior region of the limb bud (Fig. 4e,f), where expression of grem-lin is normally highest (Fig. 1c). Expression of Msx2 was significantlyhigher in mutant embryos (Fig. 2g,h). These results suggest that BMPsignaling is greater in the limb bud of gremlin mutant mice, resultingin disruption of the Shh–Fgf feedback loop.
The limb deformity (ld) mutant mouse has a disruption in Fmn, thegene encoding formin27,28, and does not maintain either the AER22 or the
NATURE GENETICS VOLUME 34 | NUMBER 3 | JULY 2003 305
+/+;+/– –/– +/+;+/– –/–
Bmp4
Bmp7
Msx1
Msx2
a b
c d
e f
g h
Figure 4 BMPs and BMP-target genes.(a–h) Dorsal views of forelimbs from E10.5embryos. Expression of Bmp4 in wild-type (a)and mutant (b) limb buds. Expression ofBmp7 in wild-type (c) and mutant (d) limbbuds. Expression of Msx1 in wild-type (e)and mutant (f) limb buds. Arrow indicates theposterior distal part of the limb bud whereexpression of Msx1 is absent in wild-type limbbuds but present in gremlin mutant limbs.Expression of Msx2 in wild-type (g) andmutant (h) limb buds.
Shh–Fgf feedback loop19,23,24. In ld mutant mice, gremlin expression islost, suggesting that the Fmn locus acts to induce and maintain gremlinexpression19. The notable similarity between the phenotypes of gremlinmutant mice and ld mutant mice suggests that the wild-type ld locus actsprimarily to regulate gremlin expression. Fmn is thought to be a down-stream target of Shh signaling because Shh mutant mice lack Fmn tran-scripts in the limb bud19. We assayed expression of Fmn in limb buds ofgremlin mutant mice and found that the expression was consistentlyabove background and similar to that in control embryos (Fig. 5a–c). Inaddition, LacZ expression was present in the mutant limb bud, suggestingthat transcription of the gremlin locus is maintained (data not shown).
The relationship between gremlin and Fmn is interesting because oftheir proximity in the mouse genome (Fig. 5d). The Fmn transcriptcomprises a large number of exons with the most 3′ exon locatedapproximately 40 kb from the gremlin open reading frame. Defects inFmn were suggested to be responsible for the phenotypes of ld mutantmice because two transgene-induced deletions (ldTgHD, ldTgBri) and achromosomal translocation (ldIn2) occur near the 3′ end of the Fmnlocus28–30. Many different formin isoforms are expressed duringembryonic development, and a subset of these was lost in the ldmutants28,29. In two other spontaneous alleles, ldJ and ldOR, grossgenomic alterations have not been identified nor have alterations inexpression of formin28,29. Given the apparently identical phenotype ofld mutant mice to gremlin mutant mice and the proximity of Fmn togremlin, we tested whether gremlin would complement ld. gremlin failsto complement ld J and identifies another ld allele (Fig. 5e,f). Althougha complex interaction between Fmn and gremlin is possible, we favorthe idea that the mapped ld mutations affect gremlin expression directlyand that they lie in cis-regulatory elements for gremlin expression.
In summary, BMP antagonism by gremlin is required for correct pat-terning of the vertebrate limb. To produce appropriate limb patterning,a feedback loop involving Fgf, Shh and BMP signaling is established. Fgfactivity in the AER defines outgrowth of the limb and maintains Shhexpression in the ZPA. Expression of genes encoding Fgfs is inhibited byBMPs, which are in turn inhibited by the antagonist gremlin. Thus,gremlin moderates the BMP inhibition of Fgfs. In gremlin mutant mice,the Shh–Fgf feedback loop is disrupted, as the current model would pre-dict. Although other BMP antagonists are expressed in the limb bud,they are not sufficiently redundant with gremlin to rescue normal devel-opment. Mice with null mutations of noggin, another BMP antagonistexpressed in the developing limb, have a very different phenotype13. Inthat case, the cartilage compartment of the limb is greatly expanded,and joints, as well as an early joint marker, are lost. The pattern of digitsand bones in the zeugopod, however, do not seem to be affected. Boththe relative size of the cartilage compartment and the joints in gremlinmutant mice appear essentially unaffected (Fig. 2q–t). Although bothnoggin and gremlin can interchangeably rescue gene expression changesin the limb bud of ld mutant mice19, the physiological roles of these twoBMP antagonists in the limb bud are very different. These distinct rolesare probably due to the differences in the expression patterns of the twomolecules, with noggin expressed in the cartilage precursors and grem-lin nearby in the mesenchyme. The implication of this difference is thatthese antagonists must act over very short distances under physiologicalconditions. Alternatively, the differences may reflect different proteinactivity that has not been identified by previous assays. Although thereare apparently non-overlapping phenotypes in the limbs of mice withmutations of noggin and gremlin, these two BMP antagonists mayhave additional overlapping and redundant roles in the developinglimb. Finally, the discovery that gremlin is allelic to ld provides newdirections for investigating the mechanism by which Shh regulatesgremlin expression.
METHODSProduction of gremlin mutant mice. We isolated a genomic clone from a 129/Svλ phage genomic library (Stratagene). We constructed a targeting cassette thatincluded a flanking diptheria toxin cassette for negative selection against randomintegrants (Fig. 1a). The targeting cassette was electroporated into embryonicstem (ES) cells (E14). We identified two positive clones for homologous recombi-nation by Southern-blot analysis (Fig. 1b) using the probes in Figure 1a. We gen-erated chimeric mice by blastocyst injection of targeted ES cells. We maintainedgremlin heterozygotes on a C57/Bl6 background. We genotyped mice bySouthern-blot analysis or by PCR. We genotyped gremlin mutant embryos usingPCR analysis of DNA prepared from embryo heads. Primer sequences are avail-able on request. All experiments were done in accordance with guidelines set bythe Animal Care and Use Committee at University of California-Berkeley.
Skeletal staining and in situ hybridization. We considered noon on the daythat we found a vaginal plug to be E0.5. We stained newborn skeletons withalcian blue and alizarin red according to established protocols. We carried outRNA localization by whole-mount in situ hybridization according to estab-lished protocols. We bisected heterozygous and wild-type embryos andprocessed them with mutant embryos throughout all stages of the in situhybridization protocol to effectively compare staining intensities. We generateda probe for Fmn by PCR amplification (primer sequences are available onrequest). We cloned this PCR product using the TopoTA kit (Invitrogen) andsequenced it to confirm the identity of the insert. We then subcloned the insertinto pCS107, linearized it with ClaI and generated a digoxigenin-labeled probewith T7 RNA polymerase. We generated the sense probe by linearization withSalI and transcription with Sp6 RNA polymerase. We stained embryos withLacZ according to standard protocols.
ACKNOWLEDGMENTSThe authors thank D. Strong for mouse husbandry; W. Skarnes for help withproducing the gremlin mutant mouse; T. Grammer and J. Wallingford forcomments on the manuscript; and A. McMahon, G. Martin, J. Hebert, M. Scott, P.Sharpe, R. Zeller and S. Dymecki for reagents and useful discussions. M.K.K. wassupported by the Pediatric Scientist Development Program of the National Instituteof Child Health and Human Development and a K08 award from the NationalInstitute of Child Health and Human Development /National Institutes of Health.D.H. was supported by a career development award from the Muscular DystrophyAssociation. R.M.H is supported by the National Institutes of Health, and theMuscular Dystrophy Association.
COMPETING INTERESTS STATEMENTThe authors declare that they have no competing financial interests.
Received 6 March; accepted 15 May 2003Published online 15 June 2003; doi:10.1038/ng1178
1. Saunders, J.W. Jr. The proximo-distal sequence of origin of the parts of the chick wingand the role of the ectoderm. J. Exp. Zool. 108, 363–403 (1948).
2. Saunders, J.W. Jr. & Gasseling, M.T. Ectodermal-mesenchymal interactions in the ori-gin of limb symmetry. in Epithelial-Mesenchymal Interactions in Development (eds.Fleischmajer, R. and Billingham, R.E.) 78–97 (Williams & Wilkins, Baltimore,1968).
3. Niswander, L. Pattern formation: old models out on a limb. Nat. Rev. Genet. 4,133–143 (2003).
4. Niswander, L., Tickle, C., Vogel, A., Booth, I. & Martin, G.R. FGF-4 replaces the api-cal ectodermal ridge and directs outgrowth and patterning of the limb. Cell 75,579–587 (1993).
5. Crossley, P.H., Minowada, G., MacArthur, C.A. & Martin, G.R. Roles for FGF8 in theinduction, initiation, and maintenance of chick limb development. Cell 84, 127–136(1996).
6. Sun, X., Mariani, F.V. & Martin, G.R. Functions of FGF signalling from the apical ecto-dermal ridge in limb development. Nature 418, 501–508 (2002).
7. Riddle, R.D., Johnson, R.L., Laufer, E. & Tabin, C. Sonic hedgehog mediates thepolarizing activity of the ZPA. Cell 75, 1401–1416 (1993).
8. Kraus, P., Fraidenraich, D. & Loomis, C.A. Some distal limb structures develop inmice lacking Sonic hedgehog signaling. Mech. Dev. 100, 45–58 (2001).
9. Niswander, L., Jeffrey, S., Martin, G.R. & Tickle, C. A positive feedback loop coordi-nates growth and patterning in the vertebrate limb. Nature 371, 609–612 (1994).
10. Laufer, E., Nelson, C.E., Johnson, R.L., Morgan, B.A. & Tabin, C. Sonic hedgehog andFgf-4 act through a signaling cascade and feedback loop to integrate growth and pat-terning of the developing limb bud. Cell 79, 993–1003 (1994).
11. Niswander, L. & Martin, G.R. FGF-4 and BMP-2 have opposite effects on limb growth.Nature 361, 68–71 (1993).
306 VOLUME 34 | NUMBER 3 | JULY 2003 NATURE GENETICS
12. Pizette, S. & Niswander, L. BMPs negatively regulate structure and function of thelimb apical ectodermal ridge. Development 126, 883–894 (1999).
13. Brunet, L.J., McMahon, J.A., McMahon, A.P. & Harland, R.M. Noggin, cartilage mor-phogenesis, and joint formation in the mammalian skeleton. Science 280,1455–1457 (1998).
14. Pearce, J.J., Penny, G. & Rossant, J. A mouse cerberus/Dan-related gene family. Dev.Biol. 209, 98–110 (1999).
15. Zhang, D. et al. A role for the BMP antagonist chordin in endochondral ossification.J. Bone Miner. Res. 17, 293–300 (2002).
16. Merino, R. et al. Control of digit formation by activin signalling. Development 126,2161–2170 (1999).
17. Dionne, M.S., Skarnes, W.C. & Harland, R.M. Mutation and analysis of Dan, thefounding member of the Dan family of transforming growth factor beta antagonists.Mol. Cell Biol. 21, 636–643 (2001).
18. Hsu, D.R., Economides, A.N., Wang, X., Eimon, P.M. & Harland, R.M. The Xenopusdorsalizing factor Gremlin identifies a novel family of secreted proteins that antago-nize BMP activities. Mol. Cell 1, 673–683 (1998).
19. Zuniga, A., Haramis, A.P., McMahon, A.P. & Zeller, R. Signal relay by BMP antago-nism controls the SHH/FGF4 feedback loop in vertebrate limb buds. Nature 401,598–602 (1999).
20. Merino, R. et al. The BMP antagonist Gremlin regulates outgrowth, chondrogenesisand programmed cell death in the developing limb. Development 126, 5515–5522(1999).
21. Capdevila, J., Tsukui, T., Rodriquez Esteban, C., Zappavigna, V. & Izpisua Belmonte,J.C. Control of vertebrate limb outgrowth by the proximal factor Meis2 and distal
antagonism of BMPs by Gremlin. Mol. Cell 4, 839–849 (1999).22. Zeller, R., Jackson-Grusby, L. & Leder, P. The limb deformity gene is required for api-
cal ectodermal ridge differentiation and anteroposterior limb pattern formation.Genes Dev. 3, 1481–1492 (1989).
23. Chan, D.C., Wynshaw-Boris, A. & Leder, P. Formin isoforms are differentiallyexpressed in the mouse embryo and are required for normal expression of fgf-4 andshh in the limb bud. Development 121, 3151–3162 (1995).
24. Kuhlman, J. & Niswander, L. Limb deformity proteins: role in mesodermal inductionof the apical ectodermal ridge. Development 124, 133–139 (1997).
25. Chiang, C. et al. Cyclopia and defective axial patterning in mice lacking Sonic hedge-hog gene function. Nature 383, 407–413 (1996).
26. Marigo, V., Scott, M.P., Johnson, R.L., Goodrich, L.V. & Tabin, C.J. Conservation inhedgehog signaling: induction of a chicken patched homolog by Sonic hedgehog inthe developing limb. Development 122, 1225–1233 (1996).
27. Woychik, R.P., Maas, R.L., Zeller, R., Vogt, T.F. & Leder, P. ‘Formins’: proteinsdeduced from the alternative transcripts of the limb deformity gene. Nature 346,850–853 (1990).
28. Mass, R.L., Zeller, R., Woychik, R.P., Vogt, T.F. & Leder, P. Disruption of formin-encoding transcripts in two mutant limb deformity alleles. Nature 346, 853–855(1990).
29. Vogt, T.F., Jackson-Grusby, L., Wynshaw-Boris, A.J., Chan, D.C. & Leder, P. The samegenomic region is disrupted in two transgene-induced limb deformity alleles. Mamm.Genome 3, 431–437 (1992).
30. Wang, C.C., Chan, D.C. & Leder, P. The mouse formin (Fmn) gene: genomic structure,novel exons, and genetic mapping. Genomics 39, 303–311 (1997).
NATURE GENETICS VOLUME 34 | NUMBER 3 | JULY 2003 307
![A Secreted BMP Antagonist, Cer1, Fine Tunes the Spatial … · 2017. 4. 6. · USAG1 may serve as a Bmp7 antagonist in the more advanced kidney [38]. Cerberus encodes a Spemann’s](https://static.documents.pub/doc/80x56/60f8afc81de3454e282e2b8e/a-secreted-bmp-antagonist-cer1-fine-tunes-the-spatial-2017-4-6-usag1-may.jpg)
