Set down in boneJoan Marsh

People, and animals, come in all shapes and sizes, the basis

of which is the skeleton. For most of the bones of the ske-

leton, the pattern is first laid down as cartilaginous con-

densations, which form from undifferentiated mesenchyme

(though for the so-called membrane bones of the skull, there

is no cartilage model). The molecular biology of skeletogen-

esis and how these molecular events relate to skeletal

growth and patterning was the subject of a recent meeting.1

Limb cartilage initially forms a continuous element exten-

ding proximodistally, which later branches and segments. In

the chick, the bone morphogenetic proteins, Bmp2, Bmp4

and Bmp7, are expressed in the mesenchyme in overlapping

patterns, both before formation of precartilaginous conden-

sations and during their aggregation, in cells adjacent to the

condensations. The receptors are also expressed in pre-

chondrogenic condensations and in immature chondrocytes.

S. Pizette (Sloan Kettering Institute, New York) found that

widespread misexpression of the secreted Bmp antagonist,

Noggin, prior to cartilage formation resulted in short limbs

completely lacking cartilage. More restricted expression of

Noggin caused loss of only newly forming elements while

those remaining were unaffected, which shows that Noggin is

active in chondrogenesis rather than in subsequent pattern-

ing steps.

Joints play a crucial role in linking elements of the

skeleton and in allowing movement. In certain regions,

condensed chondroprogenitors down-regulate genes speci-

fic to the chondrogenic lineage and express new genes, such

as the Growth and Differentiation Factor genes, Gdf5, Gdf6

and Gdf7, in distinctive stripes, marking the sites of

segmentation. Knock-outs of Gdf5 and Gdf6 show defects

in the relevant joints. Implantation of beads soaked in Gdf5,

however, stimulates cartilage production, which suppresses

joint formation (D. Kingsley, Stanford University). Another

gene expressed in developing mouse joints is ank, mutations

in which cause progressive ankylosis. This disorder man-

ifests as a post-natal arthritic phenotype with calcification

and fusion of joints and the erosion of articular cartilage.

During evolution, a variety of tetrapods have lost one or

both pairs of limbs and in many cases this process seems to

be associated with trunk elongation. In work on the python, a

primitive snake that retains rudimentary hind limbs, M. Cohn

(University of Reading) has shown that most of the body

comprises thoracic vertebrae, with accompanying extended

expression of HoxC8. The loss of the forelimb is due to an

expanded domain of expression of HoxB5, whose normal

expression range in the chick embryo is limited to a region of

the flank between fore- and hind-limb buds, along the whole

``thorax''. The python limb bud has no visible apical

ectodermal ridge (AER) and does not express Sonic hedge-

hog (Shh), but its posterior mesenchyme retains polarizing

activity. (Shh is functional elsewhere in the python, suggest-

ing that the gene has a limb-specific regulatory element that

is missing in the snake.) This suggests that the python

ectoderm is unable to respond to the inducing signal and that

failure to form an AER is the direct, visible consequence.

The premise underlying many studies on animal models is

that through the association of specific genes with particular

processes they will help us to understand the defects seen in

heritable disorders of the human skeleton. As pointed out by

S. Mundlos (University Kinderklinik Mainz), however, the

interpretations have not always proved that easy because

the genetics is often complicated by dosage effects. In the

mouse mutant, Brachydactly Type B, for instance, the

heterozygote lacks distal phalanges but the homozygote

has severely truncated limbs, along with spinal defects and

sacral agenesis. Many humans with synpolydactyly (extra

and fused digits) have been found to have an inserted

polyalanine sequence 5 0 of the HOXD13 gene. The severity

of the heterozygous phenotype varies with the length of the

insert and homozygotes show an additional dosage effect,

with short or missing digits. In the mouse, simply knocking

out Hoxd13 does not produce this phenotype; the whole

Hoxd cluster has to be eliminated. This suggests, in turn, that

the human HOXD13 mutation is interfering with expression

of other genes in the cluster. The Short digits mutation in the

mouse (Dsh) is similar to human brachydactyly A1; homozy-

gotes have a fused humerus-ulna and no hand. This pheno-

type resembles that of the Shh knock-out and it has been

proposed that the deletion in Dsh affects regulation of Shh.

Limb anomalies are often observed in patients with

craniosynostosis syndromes (premature fusion of cranial

sutures), which suggests that the developmental pathways

for limb morphogenesis and cranial sutures share some

signalling molecules. Several members of the FGF receptor

family are expressed in the developing sutures (G. Morriss-

Kay, University of Oxford). Interestingly, the equivalent point

mutation in three different FGF receptors produces different

phenotypes: Pfeiffer's syndrome for FGFR1, Apert's syn-

402 BioEssays 22.4 BioEssays 22:402±403, ß 2000 John Wiley & Sons, Inc.

1 Rigault Road, Fulham, London SW6 4JJ, UK.

ÐÐÐÐ1The Molecular Basis of Skeletogenesis, held at The

Novartis Foundation, London, 9±11 November 1999.

Meetings

drome for FGFR2 and mild coronal systosis for FGFR3

(A. Wilkie, Institute of Molecular Medicine). Similarly, dif-

ferent single amino acid substitutions along FGFR2 cause

distinct phenotypes. A mutation in the immunoglobulin-like

loop IIIc (Crouzon's syndrome) leads to dimerization,

preventing ligand binding, whilst a mutation in the linker

region (Apert's syndrome) reduces the off-rate for the ligand,

thereby prolonging signalling.

How are differentiation and morphology related geneti-

cally? Indian hedgehog (Ihh) is produced by prehypertrophic

and hypertrophic chondrocytes and stimulates the produc-

tion of parathyroid hormone-related protein (PTHrP) by

perichondrial and early chondrocytic cells. The Ihh-produc-

ing cells, however, are some distance from those that

synthesize PTHrP and since ``hedgehogs do not move very

far'' (H. Kronenberg, Massachusetts General Hospital), it is

not clear whether Ihh acts directly or via a cascade of

signalling molecules. PTHrP maintains the chondrocytes in a

proliferative, less differentiated state, thereby delaying their

production of Ihh in a negative feedback loop. The roles of

these two proteins were investigated by generating mice with

growth plates chimaeric for wild-type cells and those lacking

the PTH/PTHrP receptor. The results indicate that the

negative feedback loop betweeen Ihh and PTHrP synchro-

nizes and determines the pace of differentiation of chon-

drocytes in the growth plate.

Chondrocyte differentiation is also regulated by retinoic

acid, acting through its family of receptors (RARs). The abs-

ence of one or more of these receptors leads to deficiencies

in cartilage formation at some sites and to the formation of

ectopic cartilage at others. Misexpression of an RAR in devel-

oping limbs disrupts chondrogenesis, leading to skeletal mal-

formations: for example, a constitutively active RARa results

in a shortened tibia and absent fibula. In vitro culture of limb

mesenchyme from these mice showed that there were fewer

cartilage-forming nodules because cells expressing the

transgene were excluded from such nodules (M. Underhill,

University of Western Ontario). This effect, however, could be

reversed by the addition of Bmp2 or Bmp4 to the culture.

Skeletogenesis is also strongly affected by the molecules

of the extracellular matrix. Collagen defects have been

known for some time to underlie a variety of osteochon-

drodysplasias. Such defects tend to be more severe if the

mutation produces an abnormal protein than if the protein is

absent. Mutations in cartilage oligomeric matrix protein,

which associates with collagen IX, result in inclusion bodies,

consisting of unsecreted protein, within the rough endoplas-

mic reticulum, and an abnormal matrix that disrupts both

growth plate architecture and articular cartilage stability.

Other mutations affect enzymes in the pathways that are

responsible for the extensive post-translational processing

of matrix proteins. Diastrophic dysplasia is an autosomal

recessive severe osteochondrodysplasia that is caused by

defects in a sulphate transporter. The same gene has been

linked to two lethal disorders and a recessive form of multiple

epiphyseal dysplasia. This, again, illustrates that a range of

phenotypes can arise from mutations in a single gene,

although in this case there is a direct correlation between the

severity of the phenotype and that of the transport deficit.

A question posed by B. Hall (Dalhousie University) was

whether resorption is part of the normal skeletogenic

developmental programme. R. Poole (Shriners Hospital for

Children) described studies on maturationally distinct popu-

lations of chondrocytes which showed that there are sepa-

rate phases of matrix assembly, controlled turnover and then

partial resorption during hypertrophy. The resorption is

mediated by collagenase 3, expression of which is regulated

by the transcription factor Cbfa1. Thus, the decreased

hypertrophy seen in Cbfa1-deficient mice is due to lack of

stimulation of collagenase 3. Cbfa1 is the only gene known to

control osteoblast development in vivo (G. Karsenty, Baylor

College of Medicine). Since osteoporosis is the most

frequent disease in the Western hemisphere, understanding

the cellular and molecular biology of bone remodelling and

the mechanisms that maintain constant bone mass is highly

important from the standpoint of public health. Current

treatment, however, is still palliative rather than curative,

comprising oestrogen replacement or the use of bispho-

sphonates (G. Russell, University of Sheffield). The latter

adsorb strongly to minerals in the bone and are then taken up

by osteoclasts and incorporated into cytotoxic analogues of

ATP, leading to apoptosis of these bone-degrading cells.

Identification of the osteoclast differentiation factor (ODF),

which normally binds to RANK, a TNF receptor family

member expressed on osteoclast precursors, and osteoclas-

togenesis inhibitory factor, which acts as a decoy receptor,

was described by Gerard Karsenty as the ``most exciting

discovery in the bone field in the last three years''.

T. Suda (Showa University) showed how interleukin 1a,

in addition to its known stimulation of osteoclast formation

from progenitor cells, acts directly on mature osteoclasts to

stimulate their function. Most recently, Suda's laboratory has

found that TNFa can induce osteoclast formation via other

TNF-family receptors, independently of the ODF-RANK

interaction. The resulting cells cannot resorb bone, however,

unless interleukin 1 is also present. An important question for

the future is which pathwayÐthat involving ODF or that

employing TNFa and interleukin 1Ðis working during normal

bone remodelling in the adult and which is deficient in

degenerative diseases such as arthritis.

Altogether, this was a fascinating meeting that raised

as many questions as it answered. The synergy between

studies on animal models, human genetics and molecular

biology has given us a clear picture of many of the processes

involved in skeletogenesis and promises to deliver even

greater understanding in the future.

Meetings

BioEssays 22.4 403