Cloning of Chicken Ror2
Expressed sequence tag (EST) clone 369c3 containing the partial coding sequence for cRor2 was obtained from the BBSRC (Boardman et al.,2002). Using whole chicken embryo day 5 (Hamburger and Hamilton stage [HH] 27) cDNA and genomic DNA, we cloned the full-length coding sequence (cds) of cRor2 with a combination of degenerate reverse transcriptase-polymerase chain reaction (RT-PCR), genomic walking, and rapid amplification of cDNA ends (RACE). The cRor2 cds consists of 2,805 nucleotides, compared with 2,835 in mouse (Oishi et al.,1999) and 2,832 in human (Masiakowski and Carroll,1992). The deduced amino acid sequence of cRor2 shows a high degree of sequence identity with mouse and human (83.5% and 85.8%, respectively). The main differences are found in the leader peptide sequence. The extracellular domains as well as the intracellular TK and PST domains are highly conserved (Fig. 1A). In a phylogenetic tree derived from a multisequence alignment using all available vertebrate Ror2 sequences and D. melanogaster Ror as an outgroup, cRor2 is appropriately positioned between X. laevis Ror2 and the mammalian Ror2 group (Fig. 1B). To preclude influences of the divergent signal peptide sequences, an alignment without the signal peptides was done, which resulted in an identical tree.
Figure 1. A: Alignment of Gallus gallus (chicken) Ror2 with Homo sapiens (human) and Mus musculus (mouse) Ror2. Conserved amino acids between the three species are shaded black, and conserved amino acids between only two species are shaded gray. Positions of Brachydactyly type B (BDB) and Robinow syndrome (RS) mutations used in this study are indicated by an “x” below the alignment. Position where the human 1398insA frameshift was fused to the chicken sequence (at position 1374) is indicated by an arrow. B: Phylogenetic tree from an alignment of the Ror sequences indicated. Accession numbers: Mus musculus, BAA75481; Rattus norvegicus, XP_225181; Homo sapiens, Q01974; Canis familiaris, XP_541309; Xenopus laevis, AB087137; Danio rerio, XP_689681; Drosophila melanogaster, Q24488.
Download figure to PowerPoint
Embryonic Expression of cRor2
The embryonic expression of the Ror2 gene has been extensively studied in the mouse, but so far no expression pattern for a nonmammalian vertebrate Ror2 has been described in detail. To gain an overall insight into the expression of cRor2, we first performed whole-mount in situ hybridization (WMISH). At HH23–27, we found expression of cRor2 in the dorsal root ganglia (Fig. 2A–D, arrows), the branchial arches, craniofacial mesenchyme, and the otic vesicle (arrowheads). Overall, the expression pattern showed a high overlap with the expression seen in the mouse at comparable stages (Matsuda et al.,2001). Vibratome sections at HH25 demonstrated expression of cRor2 predominantly in the dorsal–lateral layers of the dorsal root ganglia (drg, Fig. 2E,F). Section ISH also revealed expression of cRor2 in dorsal–medial layers of the drg (Fig. 2G). Strong expression was also seen in the limb buds (Fig. 2H,I). In early limb buds, the expression of cRor2 was detected all over the limb bud, with stronger expression in anterior and posterior areas. At later stages, the expression of cRor2 in central mesenchyme was further weakened, while the expression in the anterior and posterior margins of the limb stayed intense, a pattern also observed in the mouse (Matsuda et al.,2001). Vibratome cross-sections demonstrated strong expression of cRor2 in mesenchymal cells directly beneath the ectoderm in the trunk dermomyotome and in the limbs but not within the ectoderm (Fig. 2J).
Figure 2. Expression of cRor2 in whole chick embryos. Hamburger and Hamilton (HH) stages are indicated. A–D: Whole-mount in situ hybridization (ISH) showing expression of cRor2 in craniofacial mesenchyme, branchial arches, otic vesicle (arrowheads), dorsal root ganglia (drg, arrows), and limbs. H,I: Magnification of forelimbs (FL) and hindlimbs (HL) from stages 23 and 25 are shown. E,F: Vibratome sections from HH25 embryos hybridized with cRor2 in planes as indicated by dotted lines in B. G: Section ISH showing signal in drg. J: Vibratome section demonstrating cRor2 expression in superficial mesenchymal cell layers in the limb; ectoderm shows no expression (boxed region shown as magnification below). Ec, ectoderm.
Download figure to PowerPoint
To gain insight into the expression of cRor2 during organogenesis, we analyzed the distribution of cRor2 mRNA and protein on embryonic HH29 (day 6) sections. Specificity of signals was confirmed by using a sense probe for cRor2, which yielded no signal (not shown). We found expression of cRor2 mRNA in a variety of organs (Fig. 3A,B) Strongest expression was seen in the nervous system. Signals were detected in the spinal chord and in the dorsal root ganglia (Fig. 3G,H). In the head, cRor2 was expressed in the cerebral vesicle, the optic lobe, the nasal sac, and Meckel's cartilage (Fig. 3K). Additional strong expression was observed in the otic vesicle, the basilar plate, the acoustic ganglion, and the cochleovestibular ganglion (Fig. 3K,L). We also detected cRor2 in muscles, such as limb muscles and myocardium (Fig. 3A,C). In the digestive system, cRor2 showed strong expression in the gizzard (Fig. 3D) and in the intestinal loops (Fig. 3I, arrows). Expression was also seen in the liver (Fig. 3D, arrowhead) and in the lung (Fig. 3J, arrows). Chicken Ror2 was also detected in the mesonephric tubules (Fig. 3F). Additionally, we found cRor2 expression in the cartilaginous condensations of the ribs (Fig. 3E, arrow).
Figure 3. Expression of cRor2 mRNA and protein on sections from Hamburger and Hamilton stage (HH) 29 chicken embryos. A–L: In situ hybridizations. M–S: Immunohistochemistry. A,B: Sagittal and parasagittal sections, respectively, giving an anatomical overview as well as an overview of cRor2 expression. C–J: Magnification of sections shown in A,B show expression of cRor2 in the heart (C), gizzard and liver (D, arrowhead), rib condensations (E, arrow), mesonephros (F), spinal chord (G), dorsal root ganglia (H), intestinal loops (I, arrows), and lung (J, arrows). K: cRor2 expression in the head. L: Magnification from the boxed area in K. M–S: Immunohistochemistry on cryosections adjacent to those shown in A,B confirms expression of cRor2 in the heart (M), gizzard (N), intestine (O), mesonephros (P), liver (Q), lung (R), and dorsal root ganglia and spinal chord (S). ag, acoustic ganglion; at, atrium; bp, basilar papilla; cv, cerebral vesicle; cvg, cochleovestibular ganglion; drg, dorsal root ganglia; mc, Meckel's cartilage; ns, nasal sac; ol, optic lobe; ov, otic vesicle; ve, ventricle.
Download figure to PowerPoint
For the immunolabeling on HH29 sections, anti-human ROR2 (R&D Systems) was used, which gave a clear signal for the chick Ror2 constructs, but no cross-reactions on Western blots made from chicken cell extracts (see Fig. 5B). The antibody staining largely confirmed the in situ hybridizations. Chicken Ror2 protein was found in the heart (Fig. 3M), the gizzard (Fig. 3N), and the intestine (Fig. 3O). In the mesonephric tubules, the protein was predominantly localized to the tubular lumen (Fig. 3P). Expression of cRor2 was also confirmed in liver, lung, and dorsal root ganglia (Fig. 3Q–S).
Figure 5. Schematic depiction of the different cRor2 mutants used for replication-competent retroviral vector–mediated overexpression. A: Constructs were generated corresponding to two different Brachydactyly type B (BDB) and one Robinow syndrome (RS) mutation, respectively. Chicken Ror2 R433X and E459X truncate the protein at positions corresponding to frameshifts/stops set by the BDB mutations 1321-1325del(5) and 1398insA. A heterologous construct was generated, in which the human frameshift (FS) caused by the 1398insA mutation was fused to the proximal part of cRor2, thus exactly copying the BDB mutation (c/hRor2 1374insA, gray box). As a construct corresponding to a Robinow syndrome mutation (Q502X), a stop mutation was introduced at the corresponding position 494 in the chick. B: Protein expression from all four viral constructs used in this study and green fluorescent protein (GFP) -control virus in DF1 cells. As positive control, protein extract from HEK293 cells transfected with mouse Ror2 full-length was used (right lane). β-actin was used as a loading control.
Download figure to PowerPoint
Overall, the expression pattern of chicken Ror2 is strikingly similar to the expression of Ror2 in the mouse (Al-Shawi et al.,2001; Matsuda et al.,2001; Schwabe et al.,2004). Thus there is a high evolutionary conservation of the embryonic expression pattern of Ror2, indicating functional conservation. This finding makes the chick a valuable model for functional analysis of Ror2.
Human mutations in ROR2 as well as the inactivation of Ror2 in the mouse lead to severe defects in limb development. As the chicken is an important developmental model system and is frequently used for studying limb development, we analyzed the expression of cRor2 in this process. As shown above, in the early limb bud, cRor2 is expressed in superficial limb mesenchyme. As the differentiation of limb mesenchymal cells proceeds, different lineages of tissues arise within the limb, such as tendons, connective tissue and cartilage. At HH27, cRor2 could be detected in the early cartilaginous condensations of the proximal skeletal elements of stylopod and zeugopod (Fig. 4A), while shortly after at HH30, it was also expressed in the condensations of the autopod (Fig. 4B). In all cases, the presumptive joints are excluded from cRor2 expression.
Figure 4. Expression of cRor2 in the developing limb cartilage. A: At Hamburger and Hamilton stage (HH) 27, cRor2 can be detected in the condensations of the proximal limb cartilage elements. B: At HH30, expression can be seen in the condensations of the phalanges. Presumptive joints are excluded (arrows). C: In the tibia growth plate at HH35, cRor2 is expressed in undifferentiated, nonhypertrophic chondrocytes and in the perichondrium as determined by radioactive in situ hybridization. Top: Brightfield picture of section stained with toluidine blue. Bottom: Darkfield image. D,E: A similar pattern was found in the mouse by in situ hybridization (D) and beta-galactosidase staining on limbs from Ror2lacZ/+ embryos (E). Note absence of beta-galactosidase staining in joint region (arrows). Hc, hypertrophic chondrocytes.
Download figure to PowerPoint
The initial cartilaginous condensations of the limb skeleton start to expand by proliferation and undergo a series of differentiation steps ultimately resulting in the formation of hypertrophic cartilage. In a process called endochondral ossification, hypertrophic cartilage is eventually degraded and becomes replaced by bone. The tight regulation of proliferation on one hand and differentiation on the other ensures the proper shape and growth of the skeletal anlagen until they reach their final size. During this process, expression of cRor2 was observed throughout the area of undifferentiated chondrocytes, but was excluded from hypertrophic cartilage (Fig. 4C). Additionally, we performed section ISH on E15.5 mouse limbs to compare the expression of cRor2 in the growth plate to the expression in the mouse. As in the chick, mRor2 was expressed strongly in all chondrocytes, except hypertrophic cartilage (Fig. 4D). This finding was also corroborated by lacZ staining on forelimbs of Ror2lacZ/+ mouse embryos (DeChiara et al.,2000) (Fig. 4E). Overall, expression of Ror2 in cartilaginous condensations of the ribs and the limb cartilage elements has been found in the mouse in a pattern equal to what we observed in the chick (Schwabe et al.,2004), confirming the evolutionary conservation of Ror2 expression as we have observed it in other organs.
Skeletal Defects Caused by Overexpression of Truncated Ror2 Variants
Skeletal malformations are a major finding in RS (mesomelic limb shortening, hemivertebrae) as well as in BDB (missing phalanges). To gain insight into the functional consequences of the truncating mutations in both conditions on limb cartilage development, we used the RCAS replication competent retrovirus. We chose three different mutations that in humans lead to either BDB or RS. All mutations truncate the protein within a range of approximately 60 amino acids C-terminal of the transmembrane domain. We generated four different constructs (Fig. 5A). Two mutations in chick Ror2 truncated the protein at positions corresponding to two different human BDB mutations (Schwabe et al.,2000) by introducing a stop codon. However, human mutations at these sites result in frameshifts, adding additional amino acids to the polypeptide until a premature stop is reached. Thus we made a chimeric construct carrying the human frameshift caused by the 1398insA mutation (Schwabe et al.,2000) fused to the chicken Ror2 sequence at the corresponding position (c/hRor2 1374insA). The fourth mutation (Q494X) truncates the protein at a position identical to the human RS mutation Q502X (Afzal et al.,2000). As cloning into the RCAS virus is limited to a maximal size of approximately 2.4 kb (Morgan and Fekete,1996), the further C-terminal truncations causing BDB and RS or the full-length wild-type cRor2 could not be used in this system. All truncated variants of cRor2 in RCAS are strongly expressed when transfected into chicken DF-1 cells. Proteins were of correct size and were stable (Fig. 5B).
To analyze the functional consequences of truncating mutations in Ror2 in vivo and to get an insight into the function of Ror2 in the chick, we overexpressed the different mutant constructs in developing chicken limbs. RCAS–green fluorescent protein was used as a control and elicited no effect on limb patterning and outgrowth (not shown). To assess the function of mutant Ror2, the truncated constructs were injected into HH10 forelimb or hindlimb fields, which yields high infection throughout the limb mesenchyme. First, embryos were harvested after a total of 7 days of incubation (HH32). At that time, the proximal skeletal elements of the limbs are already established and start to differentiate, while the cartilaginous elements of the autopod are still in the process of condensation. When the embryos were harvested at this early stage, overexpression of the different constructs resulted in comparable phenotypes, thus representative data are shown for construct E459X only (Fig. 6A). Overexpression of truncated forms of Ror2 has no obvious effect on skeletal patterning, but the cartilage elements in the wing as well as in the leg appear shorter and thicker when compared with the uninfected contralateral control limbs. Of interest, the proximal elements of the zeugopod and stylopod were more severely affected than the autopod. Humerus and femur were affected most severely, with the femur showing a reduction in size of up to 70%.
Figure 6. Consequences of overexpression of different cRor2 mutants in vivo. A: Representative data demonstrating the effect of cRor2-E459X overexpression in chicken fore- and hindlimbs. Limbs were harvested after 7.5 days of development (Hamburger and Hamilton stage [HH] 32), the limb skeleton is stained with Alcian blue. Note the shorter and thicker appearance of distal condensations (arrows). B: Comparison of overexpression phenotypes caused by the different mutations. Limbs were harvested after 11 days (HH37) and stained with Alcian blue and alizarin red. Note the absence of a brachydactyly phenotype in all cases. C: Hematoxylin and eosin–stained sections through tibia and metatarsal of HH32 injected and control limbs. Note the lack of flattened chondrocytes in the infected tibia and the disarrayed growth plate in the metatarsal. D: In situ hybridization with cSox9 and cIhh on tibia sections of HH32 injected and control limbs, demonstrating a block in chondrocyte differentiation. E: Unchanged proliferation rate (Histone-H3 labeling) on strongly affected HH32 tibia and mildly affected HH37 femur versus contralateral controls. Nuclei are stained with 4′,6-diamidine-2-phenylidole-dihydrochloride (DAPI, blue), Histone-H3 is stained with Alexa Fluor 568 (red). Proliferation rate was assessed as percentage of Histone-H3–stained nuclei relative to DAPI-stained nuclei within the growth plate. Columns represent mean values from eight different sections. Error bars represent standard deviations. The differences are insignificant as determined by Student's t-test (P > 0.2).
Download figure to PowerPoint
Concomitant with the onset of hypertrophic differentiation and ossification, functional differences between the different mutants became apparent. Figure 6B shows representative examples from injections of the truncated constructs harvested at day 11 of development (HH37; n > 8 for each). Overall, the strongest effects were observed for the BDB mutations E459X and 1374insA. In both cases, the differentiation of the cartilaginous elements in zeugopod and stylopod was severely delayed, whereas the autopod was only mildly affected. In the proximal elements, ossification was strongly delayed or completely prevented at HH37, indicating a delay/block in chondrocyte differentiation. Infection with the most proximal BDB truncation (R433X) resulted in a reproducibly weaker phenotype when compared with the other BDB mutants and also the RS mutant Q494X.
BDB is a dominant condition. It was thus speculated that the truncating mutations causing BDB act as dominant-negatives, whereas RS is most likely to be caused by loss of protein function (Schwabe et al.,2000,2004). However, in our experimental system, the RS mutation Q494X causes a phenotype comparable to the BDB mutants, albeit weaker than the E459X and 1374insA mutations.
Of interest, all human mutations leading to BDB that truncate ROR2 before the TK domain are frameshift mutations. This finding raises the possibility that the presence of an additional C-terminal peptide might have an influence on the pathogenic role of ROR2. We, therefore, fused the human frameshift mutation 1398insA (Schwabe et al.,2000) to the proximal part of chicken Ror2 at the identical position (1374; see Fig. 5). As described above, overexpression of this construct showed an identical result as overexpression of the variant E459X, which is truncated at the same position. Most strikingly, a brachydactyly phenotype was never observed (Fig. 6B). The smaller size and the delay in ossification observed in the autopod cartilage elements indicated viral infection in this area, thus excluding lack of viral spread in the autopod. In all cases, a mesomelic shortening of cartilaginous elements with delay/block of ossification was seen. Thus none of the BDB constructs, although the 1374insA variant was bearing an exact copy of the human mutation 1398insA, was able to phenocopy the human phenotype. A possible explanation might be that the observed discrepancies are due to species differences. In humans, BDB is a dominant condition characterized by missing medial and distal phalanges, but the respective mouse model (DeChiara et al.,2000) shows a considerably milder phenotype with missing medial phalanges, which is penetrant only in the homozygous condition. This condition might be one example where the human limb is more susceptible to alterations in signaling balances than the mouse or the chick. Similar findings have recently been reported for loss-of-function mutations in Wnt7a, which cause mild dorsoventral polarity phenotypes in the mouse (Parr and McMahon,1995), but in humans WNT7A loss of function causes a range of severe limb malformation syndromes (Woods et al.,2006).
Truncated Forms of Ror2 Delay Chondrocyte Differentiation
Once initial cartilaginous condensations are formed, a fine balance of chondrocyte proliferation and differentiation controls the growth of the skeletal elements. As shown before, after overexpression in vivo, the BDB mutant E459X and 1374insA displayed the strongest phenotype. Thus subsequent experiments were performed with the construct E459X. Hematoxylin and eosin–stained sections from infected and control tibia at HH32 demonstrate presence of small, rounded chondrocytes throughout the infected cartilage element, whereas the control tibia shows the normal chicken growth plate array of undifferentiated, flattened, and hypertrophic chondrocytes (Fig. 6C). In the metatarsal bones, where the skeletal preparations had generally shown a milder phenotype, some degree of chondrocyte differentiation occurred. However, the normal growth plate architecture was severely disturbed (Fig. 6C). The undifferentiated appearance of chondrocytes after overexpression of cRor2-E459X was confirmed by in situ hybridization for cSox9, a marker for early chondrocytes. In the wild-type, the diaphysial region shows no expression of cSox9, whereas in the infected limb, cSox9 was expressed throughout the cartilage (Fig. 6D). Conversely, the wild-type tibia shows strong expression of cIhh, a marker for prehypertrophic chondrocytes (Vortkamp et al.,1996), which is completely lacking in the infected tibia (Fig. 6D). Again, in the metatarsals, the delay of differentiation is less pronounced. Expression of cIhh occurs in infected elements, however, it is markedly down-regulated (not shown). Thus truncated Ror2 appears to severely disrupt growth plate architecture and chondrocyte differentiation at or before the onset of prehypertrophic differentiation.
To analyze alterations in chondrocyte proliferation, we performed phospho-Histone-H3 labeling (Galli et al.,2004) on sections of infected and control limbs. We applied this method to the severely affected tibias infected with the E459X mutant used above and also to femurs of less severe affected embryos, which were significantly shortened (40%), but still showed an intact growth plate morphology. As demonstrated in Figure 6E, we did not observe differences in chondrocyte proliferation in either case.
Taken together, the overexpression of truncated forms of cRor2 led to a severe defect in cartilage differentiation, predominantly in the proximal elements, without affecting proliferation. This finding is in agreement with previous findings in the Ror2 null mouse, where Schwabe et al. (2004) ascribed the mesomelic limb shortening to a defect in chondrocyte differentiation but not proliferation. Thus the effect we observed after overexpression of BDB mutants mirrors a loss of function of Ror2, which could be explained by a dominant-negative action of the mutant constructs over the endogenous protein. However, the phenotype we observed is stronger than in Ror2−/− embryos. This observation indicates that also other pathways may be negatively influenced. Of interest, double knockout of Ror2 and Ror1 increases the phenotypic severity to a level comparable with our chicken experiments (Nomi et al.,2001), indicating that mutant Ror2 constructs might also interfere with Ror1 signaling. In addition, other pathways could be influenced by the mutant constructs. There is growing evidence that Ror2 can interact with several of the Wnt signaling factors (Hikasa et al.,2002; Oishi et al.,2003; Billiard et al.,2005). Recently it was shown that Ror2 can act as a coreceptor for Wnt5a, thereby leading to inhibition of canonical Wnt signaling (Mikels and Nusse,2006). Moreover, it was shown previously, that Ror2 can also interact with and inhibit the bone morphogenetic protein-receptor type 1b (BmpR1b) pathway, an activity lost in truncated Ror2-mutants (Sammar et al.,2004). Interestingly, both Bmp and Wnt/beta-catenin pathways can interact on different levels in a variety of processes (Hoffmann and Gross,2001, and references therein; Church and Francis-West,2002; Grotewold and Ruther,2002; Hartmann,2006). This might implicate Ror2 at a central position in a regulatory network integrating distinct pathways by modulating their activity and potentially also their cross-talk. It will be a challenging task to elucidate the relationships of Ror2 and the different aspects of canonical and noncanonical Wnt signaling as well as Bmp signaling.