The mutation ROR2^W749X, linked to human BDB, is a recessive mutation in the mouse, causing brachydactyly, mediating patterning of joints and modeling recessive Robinow syndrome

Recessive Robinow syndrome (RRS, MIM 268310) and autosomal dominant brachydactyly type B (BDB, MIM 113000) are two distinct syndromes caused by mutations in the ROR2 gene (Afzal and Jeffery, 2003). RRS is characterized by moderately short stature, hemivertebrae, mesomelic limb shortening, brachydactyly, abnormal facial features and genital hypoplasia(Patton and Afzal, 2002). By contrast, BDB is characterized by hypoplasia of the distal and middle phalanges with variable degrees of distal and proximal symphalangism, often accompanied by nail dysplasia (Schwabe et al., 2000).

ROR2 belongs to a small family of receptor tyrosine kinases containing frizzled domains (Forrester,2002). The extracellular region is characterized by the presence of immunoglobulin (Ig), frizzled-like cystein-rich (Frz or CRD) and kringle(Kr) domains, all postulated to mediate protein-protein interactions. The intracellular region contains a tyrosine kinase (TK) domain, and a conserved domain consisting of two regions rich in serine and threonine (ST1, ST2)flanking a region rich in proline (PR). Based on the homology of its Frz domain to a corresponding domain in the Wnt receptors, the Frizzleds, ROR2 was proposed, and then shown, to act as a receptor for Wnts(Masiakowski and Yancopoulos,1998; Saldanha et al.,1998; Hikasa et al.,2002; Billiard et al.,2005; Oishi et al.,2003; Mikels and Nusse,2006).

The genetic lesions found in RRS patients consist of homozygous missense,nonsense and frameshift mutations affecting the Frz, Kr and TK domains of ROR2. They are predicted to eliminate or severely reduce receptor function and are thus thought to be loss-of-function mutations(Forrester, 2002). Consistent with this hypothesis, knockouts of Ror2 in the mouse cause a developmental phenotype that displays some of the features of RRS, such as short limbs and brachydactyly, as well as genital, cardiac and craniofacial defects (DeChiara et al., 2000; Takeuchi et al., 2000). By contrast, mutations in ROR2 causing autosomal dominant BDB cluster within two segments immediately upstream and downstream of the TK-encoding region, and are all predicted to result in truncations of the intracellular domain (Schwabe et al., 2000). Because neither heterozygous carriers of RRS nor individuals with chromosomal deletions encompassing ROR2 exhibit digit defects(Oldridge et al., 1999), it appears unlikely that the BDB mutations have a loss-of-function effect and it is instead thought that they confer novel gene functions(Forrester, 2002).

In this study, we use a knock-in approach in mice to assess the effect of ROR2^W749X, one of the mutations identified in individuals with severe forms of BDB. Unlike Ror2 knockout mice(Takeuchi et al., 2000; DeChiara et al., 2000), Ror2^{W749FLAG/W749FLAG} animals are viable, allowing the study of the effect of this ROR2 truncation postnatally. This animal model exhibits many of the features of RRS, and uncovers novel phenotypes in fertility and body composition. In addition, a differential phenotypic trait,restricted to the digits, was observed between mutants missing the entire Ror2 coding region versus the Ror2^{TMlacZ/TMlacZ}and Ror2^{W749FLAG/W749FLAG} lines. Our observations point to a direct role of ROR2 cytoplasmic mutations in the modulation of joint specification.

Ror2^W749FLAG mice were generated using VelociGene technology (Valenzuela et al.,2003). Briefly, a bacterial artificial chromosome-based targeting vector (BACVEC) was used to replace the sequence coding for domain G750-D930 of mouse ROR2 with the FLAG epitope, followed by a neomycin resistance cassette. ESC transfection, chimera production and genotyping of offspring were as described (Valenzuela et al.,2003). Embryos were genotyped by allele-specific PCR on yolk sac DNA, using the primers: 5′-CTTCCCACAGCCTCAGTTCATC-3′,5′-TCACCTGGAGCGTGTCATTGTC-3′,5′-GACTACAAAGACGATGACGACAAGC-3′ and 5′-TGGATGTGGAATGTGTGCGAG-3′. RNA was isolated with the RNeasy kit(Qiagen, Valencia, CA, USA) from 15.5 days post-coitum (dpc) embryos preserved in RNAlater (Ambion, Austin, TX, USA). First-strand cDNA and PCR amplification were performed as described (Marie et al.,1998), using the primers: 5′-ATTCACTGCTGCCCATCCGC-3′,5′-GATGAACTGAGGCTGTGGGAAGG-3′ and 5′-GCTTGTCGTCATCGTCTTTGTAGTC-3′.

Immunoprecipitation followed by protein G-Sepharose (Amersham Biosciences,Piscataway, NJ, USA) precipitation was performed on tissue lysates (1% Brij 96 buffer) with either an anti-FLAG antibody (M2, Sigma-Aldrich, St Louis, MI,USA) or a rabbit anti-ROR2 antibody against the 80 carboxy-terminal amino acids. Immunoprecipitates were separated by SDS/PAGE, immunoblotted with antibodies to ROR2, FLAG or phosphotyrosine (4G10, Millipore, Billerica, MA,USA), and visualized by chemiluminescence (SuperSignal West Pico, Pierce,Rockford, IL, USA).

ROR2 constructs were generated with the QuikChange mutagenesis kit(Stratagene, La Jolla, CA, USA), cloned into pcDNA3.1 (Invitrogen, Carlsbad,CA, USA) and transfected into Cos1 cells with ExGene500 (Fermentas,Burlington, Canada). Detection was with an anti-human ROR2 antibody (R&D Systems, Minneapolis, MN, USA) followed with Alexa Fluor 568 anti-human IgG(Molecular Probes, Eugene, OR, USA). Counterstaining was with 4′,6-diamidino-2-phenylindole (DAPI). Images were generated on a Zeiss Axiovert 2000 microscope with ApoTome optics (Zeiss, Jena, Germany).

Body composition and length were measured on anesthetized mice by dual-emission X-ray absorptiometry (pDEXA Sabre X-Ray Bone Densitometer and software, Norland Medical Systems, Ft. Atkinson, WI, USA). Quality assurance measurements were performed prior to each use. Measurements were an average of two consecutive scans per animal.

Ror2^{W749FLAG/W749FLAG} and Ror2^+/W749FLAG littermate males (n=6) were mated with CD-1 virgin females not selected for oestrus. For each trial mating, one female was brought to the male's cage and separated the following day. Testing continued until each male had been mated with 10 females. The percentages of females presented to each male that exhibited a vaginal plug and that became pregnant were calculated as indexes of plugging rate and pregnancy rate.

The content of the vas deferens and cauda epididymis was released into HTF-0.4% BSA (human tubal fluid, Conception Technologies, San Diego, CA, USA)and analyzed using the Hamilton Thorne Biosciences IVOS System with Animal Motility Software (Beverly, MA, USA). For IVF assays, cumulus-oocyte-complexes from superovulated B6C3F1/Crl females were incubated with sperm in 500-μl drops of HTF-BSA at 37°C. Fertilized embryos were transferred 4 to 6 hours later to 75-μl drops of KSOM (Specialty Media, Phillipsburg, NJ, USA) for overnight culture. Two-cell embryos were then evaluated, transferred to fresh KSOM and cultured for 72 hours to the expanded-blastocyst stage. Assays were performed on 3-month-old Ror2^{W749FLAG/W749FLAG},Ror2^+/W749FLAG and control littermates (n=3).

Cleared skeletal preparations were prepared as described(Lufkin et al., 1992; Jegalian and De Robertis,1992). Cell proliferation analysis in vivo was carried out by injection of 5-bromo-deoxyuridine (BrdU, Calbiochem, San Diego, CA, USA) at 50 mg/kg 2 hours before euthanasia. For E18.5 and P5 analysis, limbs were formalin fixed and decalcified (Decal Stat High Speed, Decal Corporation,Tallman, NY, USA). Humeri and femurs were then dissected and embedded in Tissue-Tek OCT (Miles, Elkhart, IN, USA). Staining of frozen sections was done with the Dako EnVision+ System and BrdU antibody (Dako, Glostrup, Denmark). Quantitation of BrdU-positive cells was performed with NIH Image (NIH,Bethesda, MD, USA) on six sections from two mice. For E12.5 analysis, limb sections were stained with the BrdU Labeling and Detection Kit II (Roche Applied Science, Indianapolis, IN, USA) and counterstained with DAPI to visualize nuclei. Proliferation rates were calculated from four sections, as the number of BrdU-positive nuclei relative to the total number of nuclei. For prepubertal testes histology, 10-μm frozen sections were stained with Hematoxylin. All other histology was performed on paraffin-embedded tissues by Charles River Laboratories (Wilmington, MA, USA).

In situ hybridization was performed on 7-μm paraffin sections of limbs,using digoxigenin-labeled riboprobes prepared as described(Stricker et al., 2002). Whole-mount hybridizaton was performed as described(Albrecht et al., 2002). Photography was with a binocular microscope and camera (Leica, Bensheim,Germany). The riboprobes used were: Sox9(Healy et al., 1996), Hoxd11, Hoxd12, Hoxd13, Gdf5, Fgf8 and Shh(Albrecht et al., 2002).

Radiography was performed post-mortem at 30 kV for 20 seconds (MX 20,Faxitron X-ray, Wheeling, IL, USA). For histomorphometry, mice were injected with 20 mg/kg calcein and 50 mg/kg demeclocycline at an interval of 2, 5 or 7 days (for 3-, 8- and 24-week old mice, respectively), and sacrificed 2 days after the demeclocycline injection. Dissected femurs were fixed in 70%ethanol, dehydrated and embedded undecalcified in methyl methacrylate. Longitudinal 5-μm sections were cut on a Microm microtome (Richard-Allan Scientific, Kalamazoo, MI, USA) and stained with 0.1% Toluidine Blue, pH 6.4. Static parameters of bone turnover were measured in a defined area between 725μm and 1270 μm from the growth plate, using OsteoMeasure (Osteometrics,Atlanta, GA, USA). Dynamic histomorphometric parameters were measured as described (Gazzerro et al.,2005). Terminology and units are as recommended by the ASBMR Histomorphometry Nomenclature Committee(Parfitt et al., 1987).

ROR2^W749X, resulting from a 2246G→A transition mapping downstream of the tyrosine kinase domain(Fig. 1A), is one of several heterozygous mutations found in families affected with classical BDB disorder(Oldridge et al., 2000). To investigate the basis for the two different syndromes associated with Ror2 mutations, we generated knock-in mice expressing ROR2 truncated at W749 and tagged at the carboxyl terminus with the FLAG epitope(Ror2^W749FLAG; Fig. 1B). Experimental cohorts were genotyped by PCR(Fig. 1C). Transcription of the truncated ROR2-W749FLAG transcript was verified by RT-PCR(Fig. 1D). Expression and tyrosine phosphorylation of the truncated receptor were confirmed by immunoprecipitation from uterus lysates(Fig. 1E). Like ROR2,ROR2-R747FLAG, a mutein very similar to ROR2-W749FLAG, localizes to the membrane (Fig. 1F,G; identical results have been obtained with ROR2-W749FLAG, not shown).

Homozygous mutant mice were born at the expected Mendelian ratio and were viable. Whereas Ror2^W749X is transmitted as a dominant mutation in BDB patients, heterozygous Ror2^W749FLAG/+ mice were normal and did not display brachydactyly. Homozygous mutant mice appeared smaller (Fig. 2B,D) and exhibited a significant decrease in body weight, apparent before weaning(Fig. 2G). Body composition analysis revealed that Ror2^{W749FLAG/W749FLAG} mice had significantly decreased fat mass and increased lean mass relative to their body weight (Fig. 2I). To determine whether the reduced body mass of Ror2^{W749FLAG/W749FLAG} mice was accompanied by alterations in metabolic rate, we measured food intake, activity, energy expenditure,oxygen consumption and carbon dioxide production. No significant differences between mutant and wild-type littermates were found in any metabolic parameter, or in glucose or insulin tolerance tests (not shown).

Homozygous Ror2^{W749FLAG/W749FLAG} had multiple craniofacial defects. They presented with a shorter snout and epiphora, which deteriorated with age and was often associated with entropion(Fig. 3A). Adult homozygous mutants showed significant shortening and broadening of the nasal bones(Fig. 3B,E,F), as well as significant shortening of the mandible, due to reduced length in the anterior region, from the infradentale to menton landmarks(Fig. 3D,H). The intra-orbital width was significantly increased in mutant mice(Fig. 3B,G). In the postcranial skeleton, all homozygous mice exhibited a reduced longitudinal axis,brachydactyly in the fore and hind paws (detailed analysis below) and, in 5%of the cases, a bifid digit I in the right hind feet. In addition, 25% of the homozygous mutants had one or more tail kinks(Fig. 2B). Body length, but not width, was significantly decreased in adult Ror2^{W749FLAG/W749FLAG} mice (9.5±0.18 cm versus 10.4±0.25 cm, P=0.005). The shortening of the longitudinal axis resulted from hypoplastic vertebral bodies, rather than from a reduced number of vertebrae (Fig. 2D). Tail kinks were caused by hemivertebrae(Fig. 2E, arrows). In addition,fusions of caudal vertebrae were observed in 5% of the mutants(Fig. 2E, arrowhead). Stylopod and zeugopod bones were shorter by 12%-19% in Ror2^{W749FLAG/W749FLAG} mice of weaning age; however, the differences between mutant and controls became less marked as the animal aged(see Table S1 in the supplementary material). In the autopod, most phalanges and metacarpals remained significantly shorter than in controls (up to 27%, P<0.01; see Table S2 in the supplementary material).

ROR2 has been reported to have signaling activity in osteoblastic cells and to promote osteoblastic activity in vitro(Billiard et al., 2005; Liu et al., 2006), whereas Ror2 knockout embryos exhibit delayed ossification of endochondral bones (DeChiara et al., 2000). We were therefore interested in determining whether ROR2-W749FLAG has an effect in the regulation of bone formation. No abnormalities could be detected in bone histomorphometric parameters in femurs from 8- to 24-week-old Ror2^{W749FLAG/W749FLAG} mice(Fig. 4C), suggesting that Ror2 does not exert a significant action on the regulation of adult bone remodeling. In actively growing femurs from 3-week-old Ror2^{W749FLAG/W749FLAG} mice, a 52% reduction in trabecular bone volume was observed associated with a 57% decrease in the number of trabeculae. There was no change in the number of osteoblasts per perimeter(not shown), or in the osteoblast surface/bone surface. A lower value of the calculated number of osteoblasts/area was due to lower bone area. Dynamic histomorphometry indicated that the mineral apposition rate was lower in the mutants than in controls, suggesting a reduced osteoblastic function. However,the mineralizing surface per bone surface was higher in the mutant femurs,resulting in an elevation of the calculated bone formation rate. In summary,the low bone volume and the high bone formation rate observed in 3-week-old animals are accompanied by normal osteoblast number; they resolve later in life, and may therefore represent a delay in the process of endochondral ossification in mutant animals rather than a defect in osteoblast differentiation and function.

In trial matings with CD-1 mice, female Ror2^{W749FLAG/W749FLAG} proved fertile, whereas Ror2^{W749FLAG/W749FLAG} males showed decreased fertility,reflected by a 4% pregnancy rate versus 15% for heterozygous controls(Fig. 5A). Ror2^{W749FLAG/W749FLAG} males had a lower combined testicular weight (118±9 mg versus 165±8 mg, P=0.002,2-month-old animals), which was also significant after correction for their reduced body weight (Fig. 5B). Histologically, early puberty Ror2^{W749FLAG/W749FLAG} testes(16 days) showed a decreased density of seminiferous tubules(Fig. 5D). In addition, most tubules had a reduced diameter and contained a single basal cell layer(Fig. 5F), as opposed to control testes, where all the tubules had a multilayered seminiferous epithelium. This abnormal histology partially resolved in adulthood, when degenerative tubules devoid of germ cells were seen only in focal areas of the testis (Fig. 5G). Numerous sperm were noted in the epididymis and ductus deferens, and accessory sex glands were normal. Consistent with the decreased testicular weight, the number of sperm cells collected from Ror2^{W749FLAG/W749FLAG}testis was decreased (P=0.04; Table 1). However, mutant and control spermatozoa performed equally well in motility and IVF assays(Table 1). Therefore, the Ror2^W749FLAG mutation results in delayed germ cell differentiation during puberty and decreased sperm production in the adult mouse, but does not affect overall sperm function or fertilizing ability.

To analyze the progression of the skeletal phenotype, embryonic skeletal stains were prepared. At E18.5, no abnormalities were observed in the heterozygotes, whereas homozygous embryos displayed defects in the limbs and spine (Fig. 6N,Q,T). We observed delayed ossification of metacarpals, metatarsals and phalanges, and missing P2 (Fig. 6Q,T). Intervertebral discs were shortened, and 20% of the embryos presented one or two misshapen thoracic vertebrae with rib fusion of the associated ribs(Fig. 6N). These defects represent a subset of the abnormalities displayed by Ror2 knockout embryos [described in detail elsewhere(DeChiara et al., 2000; Schwabe et al., 2004), and shown in Fig. 6L,O,R,U for comparison], and are milder. At E14.5, Ror2^{W749FLAG/W749FLAG} embryos displayed hypoplastic skeletons, with a reduction in the size of all the anlagens, including the Meckel's and nasal cartilages (Fig. 6B). Vertebral deformities and tilting were apparent at this stage(Fig. 6E), but, with the exception of impaired segmentation of the digital rays(Fig. 6H), patterning was normal. These abnormalities are reminiscent of those exhibited by Ror2^{TMlacZ/TMlacZ} E14 embryos(Fig. 6C,F,I), but are not as severe. Because the cartilaginous elements of Ror2^{W749FLAG/W749FLAG} embryos displayed anatomical defects consistent with those observed in the skeletons of perinatal and adult mice,we conclude that the function mediated by the carboxy-terminal domain of ROR2 is required prior to E14.5, at the time of cartilaginous condensation.

Histological examination of Ror2^{W749FLAG/W749FLAG} E13.5 embryos revealed a reduced number of cells in the chondrogenic condensations,whereas cell density did not appear to be affected(Fig. 7B,E,F). Cartilage matrix production by the chondrocytes was evaluated histologically using HAB(Hematoxylin, Alcian Blue) or HGF (Hematoxylin, Fast Green, Basic Fuschin),which stain for proteoglycans (blue) and collagen-associated proteoglycans(fuchsia), respectively (Ippolito et al.,1983; Tribioli and Lufkin,1999). There were no differences in the pattern or intensity of the matrix stain in chondrocytic condensations at any stage between E13.5 and E15.5 (Fig. 7C-H).

Once the cartilage anlagens formed, their program of development proceeded normally in Ror2^{W749FLAG/W749FLAG} mutants, except for a delay in the onset of developmental hallmarks, such as hypertrophy and vascularization (Fig. 7G,H),which translated into a slight reduction of bone length at birth(Fig. 7I,J). However, no change could be detected between mutants and controls in the proliferation rates of condensation or growth plate chondrocytes(Fig. 7K-N). These findings,together with the histological observations, indicate that the Ror2^{W749FLAG/W749FLAG} skeletal phenotype is not caused by an altered proliferation or differentiation of chondrocytes, but rather by an impaired recruitment of mesenchymal cells into the chondrogenic condensation.

An intriguing aspect of the phenotypes associated with the Ror2allelic panel is that both Ror2^{W749FLAG/W749FLAG} and Ror2^{TMlacZ/TMlacZ} embryos, expressing truncated ROR2, lack phalanges P2. This stands in marked contrast to Ror2^-/-embryos, where all the elements of the digit are formed(Schwabe et al., 2004)(Fig. 6U), indicating that ROR2 function is dispensable for specification of the distal digit joint, yet truncated ROR2 receptors lacking either part or all of the cytoplasmic domain interfere, in homozygosis, with normal segmentation of the phalangeal rays. To investigate the potential mechanism, in situ hybridization with markers of limb patterning was performed.

Hoxd11-Hoxd13 are essential for normal patterning of the mouse autopod, and targeted disruption of any one of them causes autopod phenotypes reminiscent to Ror2^{W749FLAG/W749FLAG}(Davis and Capecchi, 1994; Davis and Capecchi, 1996). However, Hoxd11, Hoxd12, and Hoxd13 expression is not affected in Ror2^{W749FLAG/W749FLAG} limbs at E12.5, a stage at which phalanges are being specified (see Fig. S1 in the supplementary material). Similarly, the expression patterns of Shh and Fgf8 were not altered in mutant embryos during E11.5-E13.5 (see Fig. S2 in the supplementary material), indicating that the effect of ROR2 truncation on autopod bone length and patterning is independent of the SHH-HOXD signaling pathway and the SHH-FGF feedback loop.

Finally, the expression of Gdf5, an early marker of joint interzones, was examined. E12.5 Ror2^{W749FLAG/W749FLAG}limbs showed the correct pattern of Gdf5 expression expected at this stage; diffuse expression in the inter-digital mesenchyme and more intense expression in the outerzone of condensation and in the presumptive metacarpo-phalangeal joint interzones as the growth of the digital rays advances (Storm and Kingsley,1996) (Fig. 8A,B). At E13.5, expression of Gdf5 in Ror2^{W749FLAG/W749FLAG} limbs was correctly restricted to the region of the presumptive digit joints but it showed a decreased intensity and domain in the presumptive proximal interphalangeal interzone(Fig. 8D, arrow). At later stages, expression was detected in the proximal interphalangeal interzone,whereas the region of the presumptive distal interzone failed to show Gdf5 expression (Fig. 8E,F; see also Fig. S3 in the supplementary material). Thus,ROR2^W749FLAG seems to interfere with the specification of the distal digit interzone and the subsequent formation of the P2/P3 synovial joint.

Mutations in Ror2 are responsible for two classes of rare genetic disorders in humans, BDB and RRS, which display distinct modes of inheritance(autosomal dominant and recessive, respectively) and are associated with different sets of gene mutations. In this study, we generated mice bearing the W749X mutation in Ror2, which in humans results in a severe form of BDB. Our results show that Ror2^W749FLAG does not result in dominant brachydactyly in the mouse. Rather it causes abnormalities similar to those of Ror2 knockout mice, without, however, causing the perinatal lethality observed in these lines. In this respect, it provides an animal model most closely resembling RRS.

In contrast to BDB, which only affects the digits, Ror2^{W749FLAG/W749FLAG} mice exhibit abnormalities in the whole skeleton. Similar to RRS patients and Ror2 knockout mice, Ror2^{W749FLAG/W749FLAG} mice show hemivertebrae, rib fusions, and shortening of the long bones. In the digits, a missing phalanx causes brachydactyly, a feature also described in the Ror2^{TMlacZ/TMlacZ} mice(DeChiara et al., 2000) and found in some RRS families. Furthermore, about 5% of the animals present a duplicated digit I in the right hind limb, a phenotype also observed in 20% of the Ror2^TMlacZ mutants that may be related to the bifid thumbs shown by some RRS patients(Oldridge et al., 2000; Schwabe et al., 2004). The craniofacial defects, characterized by midfacial hypoplasia due to shortened nasal and jaw bones, resemble those of Ror2 knockout mice, albeit manifested to a milder degree. Although hypertelorism, a feature of RRS patients and Ror2^-/- mice(Schwabe et al., 2004), is not detected in Ror2^{W749FLAG/W749FLAG}, a wider intra-orbital space, and entropion with epiphora and dochitis are observed. It is possible that changes in the orbital bones of Ror2^{W749FLAG/W749FLAG}mice affect the architecture of the naso-lacrimal ducts, resulting in obstruction and dochitis (Meyer,1993), as is observed in brachycephalic canine and feline breeds. Overall, in mice, the mutation Ror2^W749FLAG does not phenocopy the human BDB phenotype associated with this mutation, resulting instead in a mouse model resembling RRS.

The different pattern of inheritance of the W749X mutation in humans compared with in mice may reflect differences in skeletal development between the two species (Wilkie,2003), or may be attributed to genetic background effects on the expression of the mutation. Indeed, there are examples of distal phalanx brachydactyly in RRS patients (Patton and Afzal, 2002; Balci et al.,1993), and, conversely, of facial hypertelorism, hypoplastic alae nasi, and high nasal bridge and arched palate in some BDB cases(Hamamy et al., 2006). These observations suggest an overlap in the spectra of clinical features caused by BDB and RRS mutations.

The skeletal hypoplasia observed in Ror2^{W749FLAG/W749FLAG} mice can be traced back in development to the formation of chondrogenic condensations, which are all reduced in size in mutant embryos. Once formed, Ror2^{W749FLAG/W749FLAG} condensations develop normally,except for a delay with respect to controls. However, this is not accompanied by changes in the number or density of BrdU-labeled cells in the cartilage(E12.5, E18.5) or in the growth plates (P5). This is in agreement with observations in Ror2^-/- mice(Schwabe et al., 2004), or after the overexpression of dominant-negative forms of Ror2 in chicken(Stricker et al., 2006). Growth of the cartilaginous anlagen is a complex process that results from cellular recruitment from the mesenchyme underlying the perichondrium, as well as from chondrocyte proliferation in the hyaline cartilage and, especially, in the growth plates of the long bones, once they are formed(Dodds, 1930). We do not observe changes in chondrocyte proliferation in Ror2^{W749FLAG/W749FLAG} mice; however, a reduction in the initial size of all of the chondrogenic condensations, a crucial parameter in determining the dimensions of the final skeletal development(Atchley and Hall, 1991), is observed. These results are consistent with an effect of Ror2^W749FLAG in the recruitment of mesenchymal cells into the chondrogenic condensation, rather than with an effect on elongation of the cartilage anlagen.

Ror2^{W749FLAG/W749FLAG} mice show brachydactyly due to a missing phalanx in digits II to IV, a feature shared with Ror2^{TMlacZ/TMlacZ} mice. Brachydactyly in BDB and RRS patients is thought to result from gain-of-function mutations of ROR2, as individuals carrying deletions encompassing the whole locus do not exhibit defects in the digits(Oldridge et al., 2000). Evidence in support of this interpretation is provided by Ror2^{TMlacZ/TMlacZ} and Ror2-null mice; although both exhibit very similar defects in the skeleton, Ror2-null pups show a normal number of phalanges, whereas Ror2^{TMlacZ/TMlacZ} mice fail to develop the distal interphalangeal (P2/P3) joint. Therefore, although ROR2 is dispensable for normal joint specification, cytoplasmic truncations of ROR2 disrupt the specification of the P2/P3 joint. The skeletal abnormalities of Ror2^{W749FLAG/W749FLAG} or Ror2^{TMlacZ/TMlacZ} mice may thus be interpreted as compound phenotypes, resulting from a combination of the loss of normal ROR2 function(skeletal syndrome observed in all Ror2 mutants) and newly acquired ROR2 functions (interference with distal joint specification, observed in Ror2^W749FLAG and Ror2^TMlacZ, but not in mice lacking Ror2).

Most of the synovial joints in the limb, including those in the digits,form through segmentation of a single pre-existing cartilaginous condensation at the interzone (Archer et al.,2003). Gdf5 expression is crucial for normal joint development and represents one of the earliest markers of interzone formation,preceding any morphological sign of specification(Storm and Kingsley, 1999). No Gdf5 expression can be detected in Ror2^W749FLAGmutants at the site of the presumptive distal interzone, indicating that truncated ROR2 impairs distal joint specification at a very early stage,upstream of GDF5. Wnt is an upstream regulator of GDF5 in chondrocytes and is thought to be important for joint formation through the induction of GDF5(Hartmann and Tabin, 2001; Guo et al., 2004; Tamamura et al., 2005). As ROR2 can interact with several Wnts(Billiard et al., 2005; Mikels and Nusse, 2006; Hamamy et al., 2006; Oishi et al., 2003), it is conceivable that ROR2 cytoplasmic truncations exert an inhibitory effect on Wnt signaling at the developing joint through non-productive interactions with Wnt ligands. Indeed, genetic and biochemical evidence in C. elegansshows that the extracellular domain of CAM-1 (the only C. elegans ROR homolog) is sufficient to antagonize multiple Wnts in a non-cell-autonomous manner, suggesting that CAM-1 buffers Wnt levels through sequestration(Green et al., 2007). The high degree of similarity shared by nematode and mammalian ROR proteins suggests a conserved ability of ROR receptors to modulate Wnt signaling in a cell-membrane-dependent, but kinase-independent, manner.

It has been proposed that during segmentation of the cartilaginous condensations, each new joint is formed at a distance from the previous one by spatial recovery from Noggin-inhibitory signals (Wnt, GDF5, BMPs) that emanate from the previous interzone (Guo et al.,2004). Under this model, the lack of the distal digital joint in Ror2 mutants could be explained by the shorter length of the digital ray condensations, which would not provide enough distance at the distal end to allow concentrations of a joint-inducing activity to recover from the inhibitory field of the proximal joint. However, if this were the mechanism responsible for lack of the P2/P3 joint in the Ror2^{W749FLAG/W749FLAG} digits, we would also expect Ror2^-/- mice to lack the P2/P3 joint, as their developing digits are as short as those of Ror2^{TMlacZ/TMlacZ} and Ror2^{W749FLAG/W749FLAG} mice(Fig. 6H,I,T,U). Yet, Ror2^-/- mice show normal specification of the digital joints. These observations support a direct role of ROR2 cytoplasmic truncations in blockage of formation of the P2/P3 joint.

Survival of the Ror2^W749FLAG homozygous mutant mice makes them useful for studying the role of ROR2 in adult bone. Based on studies using cell lines and mouse calvarial primary cultures, it has been proposed that ROR2 promotes both osteoblast differentiation and the commitment of mesenchymal stem cells to the osteoblastic lineage(Liu et al., 2006). Although at 3 weeks of age Ror2^{W749FLAG/W749FLAG} mice exhibit a reduction in trabecular bone volume, this phenotype is consistent with the delay in embryonic bone development and resolves later in life. Histomorphometric analysis of femoral bones shows that Ror2^W749FLAG has no major role in adult bone homeostasis.

The mutation, however, uncovers previously unrecognized effects of ROR2 in other organs. Ror2^{W749FLAG/W749FLAG} mice display a lean phenotype with a normal metabolic rate, insulin tolerance, and adsorption and elimination of glucose. The reduced adiposity of Ror2^W749FLAG homozygous mice may result from developmental changes affecting the differentiation program of embryonic mesenchymal stem cells into fat and muscle lineages (Taylor and Jones, 1979). Consistent with this interpretation, ROR2 overexpression has been reported to affect the adipogenic differentiation of human mesenchymal stem cells (Liu et al.,2006).

Ror2^{W749FLAG/W749FLAG} males are hypogonadic, as a result of focal degeneration of seminiferous tubules, have a reduced sperm count and show a markedly reduced fertility rate. In testes from 16-day-old mutant mice, all of the seminiferous tubules show a reduced diameter, contain a single layer of spermatogonia, and are devoid of spermatocytes. This suggests a role of ROR2 in the differentiation of germ cells during early puberty. The testicular and reproductive phenotype of Ror2^{W749FLAG/W749FLAG} mice is very similar to that of Bmp8^tm1blh homozygous males, which exhibit infertility and dual defects in spermatogenesis: delay in the initiation of germ cell proliferation during early testicular development; and focal tubular degeneration in the mature testis (Zhao et al., 1996). It will be interesting to investigate whether ROR2 and BMP8B act in the same pathway during spermatogenesis.

We thank all at Regeneron Pharmaceuticals for their assistance, especially the VelociGene and RAF cores for targeting vectors, blastocyst injections, and mouse breeding and genotyping. We also thank S. Statton for graphic support,S. Croll for statistical analysis, T. Rakowski-Anderson for the diagnosis of entropion, and CRL for histopathology and IVF services. In part supported by grants DK 42424 and DK 45227 from the National Institutes of Diabetes,Digestive and Kidney Disease (E.C.), and grant SFB577 from the Deutsche Forschungsgemeinschaft (S.S. and S.M.).