R104 Dispatch

Skeleton morphogenesis: Defining the skeletal elementsAndrea Vortkamp

A recent study of mice carrying different combinationsof mutations in the genes for two bone morphogeneticfactors (BMPs), BMP5 and GDF5, indicates that BMPshave specific and synergistic functions in the regulationof skeleton development.

Address: Department of Genetics, Harvard Medical School, 200Longwood Avenue, Boston, Massachusetts 02115, USA.

Electronic identifier: 0960-9822-007-R0104

Current Biology 1997, 7:R104–R107

© Current Biology Ltd ISSN 0960-9822

The vertebrate skeleton consists of a variety of individualelements, each characterized by a specific size and shape.There are two main modes by which these skeletal ele-ments are formed. In one mode, known as ‘intramem-branous ossification’, which occurs for example in thedevelopment of the skull, mesenchymal cells directly dif-ferentiate into bone. In the other mode, known as ‘endo-chondral ossification’, which is used to form the bones ofthe vertebral column, the extremities and the shoulderand pelvic girdles, the skeletal elements are derived fromcartilage precursors that are later replaced by bone tissue.

The first step in endochondral ossification is theaggregation and condensation of mesenchymal cells,which then differentiate into chondrocytes. The cartilage-nous progenitor of a bone can be formed either by thedirect condensation of mesenchymal cells or by branchingand/or splitting of an individual condensation into severalcartilage elements. For example, in the developing limb asingle condensation first branches into a Y-shaped struc-ture and then splits into the precursors of humerus, ulnaand radius, whereas the digits develop as single rod-likecondensations which then segment into the different pha-lanxes [1,2]. Although these processes are morphologicallyvery well described, until recently little was known aboutthe signals that regulate the initiation, branching and split-ting of the early condensations. In the last few years,however, members of the bone morphogenetic protein(BMP) family of secreted proteins have been implicatedin these processes by genetic studies in mice. Mostrecently, by crossing two mouse strains carrying mutationsin two different BMP genes, Storm et al. [3] have demon-strated that BMPs synergystically regulate both the con-densation and splitting of specific cartilage elements.

The BMPs form a subgroup of the transforming growthfactor b (TGFb) family of secreted molecules, and newexamples are being discovered at an increasing rate. They

are synthesized as large precursor proteins containing asignal peptide, an amino-terminal pro-domain and a func-tional carboxy-terminal signalling domain. Mature BMPproteins contain seven conserved cysteine residues andare cleaved from the precursor after a dibasic RXXR motif(where R is arginine and X any amino acid) [4].

The different BMPs are expressed in very complex pat-terns during embryogenesis and have been shown to regu-late the development of a variety of embryonic structures.For example, BMPs have been implicated in the estab-lishment of the dorsal–ventral body axis, left–right asym-metry, the patterning of neural tissue and somites, andthe morphogenesis of the kidney, lung and gut [5].Although the roles of BMPs in some of these processesare well characterized, and in spite of the fact that BMPswere first isolated because of their ability to induceectopic chondrogenesis [6], relatively little is knownabout their function during embryonic development ofthe skeleton.

The recent discovery that two mouse mutants, short ear(se) and brachypodism (bp), are caused by mutations inmembers of the BMP family — BMP5 and growth differ-entiation factor 5 (GDF5), respectively — gave the firsthint that at least some BMPs have specific regulatory rolesin the development of cartilage condensations (Fig. 1)[7,8]. Short ear mice are characterized by having abnor-mally small external ears (pinnae) and several internalskeletal abnormalities, including a change in size andshape of the xiphoid process of the sternum and a slightreduction in the length of the long bones of the limbs [9].The most obvious phenotype of brachypodism mice is ashortening of the long bones of the limbs and a fusion ofthe medial and distal phalanxes in digits II–V of both foreand hind limbs; they also have malformations of severallimb joints [10].

Although the two mutations have different effects on thedevelopment of the skeleton, they have some characteris-tics in common. Developmental studies revealed that, inboth cases, abnormalities of the affected skeletal elementscan first be found at the earliest stages of mesenchymalcondensations. At these stages, both genes are normallyexpressed in the tissues affected by the respective muta-tions. Thus, BMP5 is expressed in early condensation ofthe pinnae and the xiphoid process, and is later found inthe perichondrium surrounding the developing cartilageelements including those of the limb bones [11], whileGDF5 expression can first be detected in those areas ofcartilage condensations where the joints will later form.

This expression of GDF5 precedes any morphologicalsign of joint formation, indicating that GDF5 may have aninductive function in this process [3].

Interestingly, both genes are expressed not only in tissuesaffected in the respective mutations, but in several addi-tional locations of the developing skeleton which are not,or are only mildly, affected in the mutant animals. Thissuggests that neither BMP5 nor GDF5 expression is indis-pensable for the normal development of all the structuresin which they are expressed and that other factors, mostlikely closely related members of the BMP family, mighthave overlapping functions.

That BMPs act in a synergistic way in at least someaspects of skeletal development is implied by the pheno-type of the new se/se; bp/bp double mutants. [3]. Through-out most of the skeleton, the double mutant mice simplydisplay the sum of the individual phenotypes — forexample, they have short ears and short limbs. Strikingly,however, whereas short ear mice show only a mild defectof the xiphoid process of the sternum, and no alterationsof the sternum are found in brachypodism mice, sternumdevelopment is severely affected in the double mutants.

During normal development, the sternum forms from twomesodermal condensations in the lateral body wall, whichfuse above the ventral midline, proliferate and differenti-ate into cartilage. The ribs are derived from independentcondensations which contact the sternal band transiently,inducing its segmentation into individual sternebrae. Thesegmentation process is similar to that which occurs duringdigit formation, but sternebrae and ribs stay connected bya non-mineralized fibrocartilagenous matrix, rather thanbeing completely separated by synovial joints [12].

In the double mutants, the presternal bands condense, butare thinner than normal in the center. This region does notgrow in width, so that this portion of the sternal bands failsto intersect the ribs, no sternebrae form and the ribs growuntil they contact each other. More laterally, where sterne-brae form, the fibrocartilagenous junctions do not remainunmineralized, as they normally do, and the ribs and sterne-brae become connected by ossified tissue. The distal endsof the sternum, however, show no synergistic phenotype.

The phenotype of the double mutant mice clearlydemonstrates that BMP5 and GDF5 act in a synergisticway during the formation of the sternum. GDF5 isexpressed in the early precondensations of the sternumand ribs, later becoming restricted to small bands in theintercostal junctions, whereas BMP5 is expressed in theperichondrium of the sternum and ribs. Interestingly, thisis a rare example where functionally redundant genes areexpressed in different cell populations of the affectedtissue. This may be because they encode proteins that,

while secreted from different sources, can act on the sametarget cells. Alternatively, instead of replacing each other’sfunction in a common signalling pathway, they might acti-vate parallel regulatory pathways which are able to substi-tute for each other by indirect means.

Consistent with the latter possibility, in the singlemutants, loss of BMP5 mainly alters the condensationprocess, whereas GDF5 loss has a major effect on the for-mation of the joints. In addition to missing or malformedjoints, however, GDF5-deficient brachypodism mice arealso characterized by shorter bones. As GDF5 is expressedpredominantly in the joint regions, this combination of

Dispatch R105

Figure 1

A mouse skeleton (a) and a diagram of the sternum (b), coloured toshow the regions affeted by the short ear (se) and brachypodism (bp)mutations. The se and bp mutations affect different parts of the mouseskeleton (coloured red and blue, respectively); se/se; bp/bp doublemutants show the sum of the individual phenotypes and additionallythe sternum is affected (green). The latter result shows that the se andbp gene products, BMP5 and GDF5, respectively, have a synergisticeffect in sternal development.

R106 Current Biology, Vol 7 No 2

malformed joints and shortened long bones raises theinteresting possibility that signals from the joints are nec-essary for the growth of the cartilage elements, and thatthese signals are reduced in the malformed joints of themutant mice. Such an interaction could thus lead to anindirect form of redundancy between BMPs.

The analysis of animals carrying different numbers ofmutated alleles revealed further that BMP5 and GDF5compensate for each other with different efficiency. Se/se;bp/+ mice show only a slightly enhanced sternal pheno-type compared to the short ear mutation. Therefore, oneallele of GDF5 is nearly as efficient as two alleles in com-pensating for the loss of BMP5 in the short ear back-ground. However se/+; bp/bp mice display a range ofdefects in specific sternocostal joints — from ectopic ossi-fications to bone formation continuous with the caudalsternebrae — which are not detected in the brachipodismmouse line. These defects are not as severe as in the se/se;bp/bp double mutant mice, implying that BMP5 compen-sates for GDF5 in a dosage-dependent manner. Thus,GDF5 seems to have a stronger influence on the develop-ment of the sternum and only a normal-strength BMP5signal can fully compensate for loss of GDF5.

The analysis of the different mouse mutations indicatesthat BMPs interact in complex signalling pathways toregulate skeletal development. They can substitute foreach other in the development of some structures, but areindispensable for the organization of others. Thisredundancy of function might help guarantee the develop-ment of a functional skeleton, and in addition allow subtlespecies-specific modification of single skeletal elementswithout impairing the organization of the entire body plan.

A detailed analysis of the expression patterns of thedifferent BMP genes should give further insight intowhich members are involved in common patterningevents. Starting from the described studies [3], the mostpromising candidates might be closely related memberswithin the BMP family. GDF5, GDF6 and GDF7, forexample, form a subgroup of the BMP family, and allthree are expressed in joint regions [3]. BMP2, BMP4,BMP5 and BMP7 are expressed in different early conden-sations and later in the perichondrium surrounding thecondensations, whereas BMP6 is expressed in differenti-ated hypertrophic chondrocytes [13,14]. The analysis ofsingle and double mutant mice should give further insightin the combination of genes responsible for the develop-ment of specific cartilage elements.

Unfortunately, short ear and brachypodism are the onlynatural occurring mutations of genes for BMPs identifiedso far. The generation of ‘knockout’ mice by targetedgene inactivation is a useful tool for analyzing the role ofspecific genes during development, and BMP7 knockout

mice have been generated and found to have skeletalabnormalities, in particular, polydactytly in combinationwith rib and sternal defects [15,16]. As BMPs have impor-tant roles in a variety of early developmental processes,however, loss of many of them might result in lethalitybefore skeletal development starts, as is the case forBMP2 and BMP4. It may be possible to study the role ofthese BMPs by tissue-specific gene inactivation, but thiswill require prior identification of the tissue-specific pro-moter elements.

In addition to signals mediated by BMPs, other factorsare likely to influence the condensation of cartilage ele-ments. Little is known about the mechanisms determin-ing the location and the shape of the cartilagecondensations. Genes likely to be involved in thisprocess include members of the Hox gene family, whichprovide positional information in the early limb bud andalong the anterior–posterior body axis, and studies ofmutant mice have shown Hox genes have a striking, butalso redundant, influence on the skeletal structure (forexample, see [17–19]).

Another question is how the early condensations arerestricted in size. What inhibits BMPs from turning theentire mesenchyme into bone? Misexpression of BMP4during limb development leads to a dramatic increase incartilage formation [20], most likely by overcoming thenatural restriction signals and thereby initiating chondro-genesis throughout the entire limb mesenchyme. Nogginand Chordin are candidate restriction signals in dorso-ventral patterning [21–23] — both have been shown tobind to BMP4 and inhibit its function in dorso-ventral pat-terning. The analysis of the potential role of Noggin andChordin in bone formation could give interesting newinsight into the mechanism of cartilage formation.

The analysis of mice carrying compound BMP5 andGDF5 mutations by Storm et al. [3] has given interestingnew insights into the role of BMPs in skeletal develop-ment and provided an excellent starting point for thefurther analysis of an obviously complex signallingnetwork regulating the development of the skeleton.

AcknowledgementsI would like to thank Cliff Tabin for critical reading of the manuscript andDavid Kingsley for providing the picture of the mouse skeleton.

References1. Oster GF, Shubin N, Murray JD, Alberch P: Evolution and

Morphogenetic Rules: the Shape of the Vertebrate Limb. Ont PhylEvol 1988, 42:862–884.

2. Shubin NH, Alberch P: A Morphogenetic Approach to the Originand Basic Organization of the Tetrapod Limb. Evol Biol 1986,20:319–387.

3. Storm EE, Kingsley DM: Joint patterning defects caused by singleand double mutations in members of the bone morphogeneticprotein (BMP) family. Development 1996, 122:3969–3979.

4. Kingsley DM: The TGF-beta superfamily: new members, newreceptors, and new genetic tests of function in differentorganisms. Genes Dev 1994, 8:133–46.

5. Hogan BLM: Bone morphogenetic proteins: multifunctionalregulators of vertebrate development. Genes Dev 1996,10:1580–1594.

6. Wozney JM, Rosen V, Celeste AJ, Mitsock LM, Whitters MJ, Kriz RW,Hewick RM, Wang EA: Novel regulators of bone formation:molecular clones and activities. Science 1988, 242:1528–1534.

7. Kingsley DM, Bland AE, Grubber JM, Marker PC, Russell LB,Copeland NG, Jenkins NA: The mouse short ear skeletalmorphogenesis locus is associated with defects in a bonemorphogenetic member of the TGF beta superfamily. Cell 1992,71:399–410.

8. Storm EE, Huynh TV, Copeland NG, Jenkins NA, Kingsley DM, LeeSJ: Limb alterations in brachypodism mice due to mutations in anew member of the TGF beta-superfamily. Nature 1994,368:639–43.

9. Green MC: Mechanism of the pleiotropic effects of the short earmutant gene in the house mouse. J Exp Zool 1968, 167:129–150.

10. Gruenberg H, Lee AJ: The anatomy and development ofbrachypodism in the mouse. J Embryol Exp Morph 1973,30:119–141.

11. King JA, Marker PC, Seung KJ, Kingsley DM: BMP5 and themolecular, skeletal, and soft-tissue alterations in short ear mice.Dev Biol 1994, 166:112–122.

12. Chen JM: Studies on the morphogenesis of the mouse sternum.I. Normal embryonic development. J Anat 1952, 86:373–386.

13. Kingsley DM: What do BMPs do in mammals? Clues from themouse short-ear mutation. Trends Genet 1994, 10:16–21.

14. Lyons KM, Hogan BL, Robertson EJ: Colocalization of BMP 7 andBMP 2 RNAs suggests that these factors cooperatively mediatetissue interactions during murine development. Mech Dev 1995,50:71–83.

15. Dudley AT, Lyons KM, Robertson EJ: A requirement for bonemorphogenetic protein-7 during development of the mammaliankidney and eye. Genes Dev 1995, 9:2795–2807.

16. Luo G, Hofmann C, Bronckers AL, Sohocki M, Bradley A, Karsenty G:BMP-7 is an inducer of nephrogenesis, and is also required foreye development and skeletal patterning. Genes Dev 1995,9:2808–2820.

17. Davis AP, Capecchi MR: A mutational analysis of the 5¢ HoxDgenes: dissection of genetic interactions during limbdevelopment in the mouse. Development 1996, 122:1175–1185.

18. Davis AP, Witte DP, Hsieh-Li HM, Potter SS, Capecchi MR: Absenceof radius and ulna in mice lacking hoxa-11 and hoxd-11. Nature1995, 375:791–795.

19. Fromental-Ramain C, Warot X, Messadecq N, LeMeur M, Dolle P,Chambon P: Hoxa-13 and Hoxd-13 play a crucial role in thepatterning of the limb autopod. Development 1996,122:2997–3011.

20. Duprez D, de H. Bell EJ, Richardson MK, Archer CW, Wolpert L,Brickell PM, Francis-West PH: Overexpression of BMP-2 and BMP-4 alters the size and shape of developing skeletal elements in thechick limb. Mech Dev 1996, 57:145–157.

21. Holley S, Neul J, Attisano L, Wrana J, Sasai Y, O’Connor M, DeRobertis E, Ferguson E: The Xenopus dorsalizing factor nogginventralizes Drosophila embryos by preventing DPP fromactivating its receptor. Cell 1996, 86:607–617.

22. Piccolo S, Sasai Y, Lu B, De Robertis E: Dorsoventral patterning inXenopus: inhibition of ventral signals by direct binding of chordinto BMP-4. Cell 1996, 86:589–598.

23. Zimmerman LB, de Jesus-Escobar JM, Harland RM: The Spemannorganizer signal noggin binds and inactivates bonemorphogenetic protein 4. Cell 1996, 86:599–606.

Dispatch R107