The receptor tyrosine kinase Ror2 associates with the melanoma-associated antigen (MAGE) family protein Dlxin-1 and regulates its intracellular distribution.

The mammalian Ror family receptor tyrosine kinases, Ror1 and Ror2, play crucial roles in developmental morphogenesis. Although the functions of Ror1 and Ror2 are redundant, Ror2 exhibits more specific functions during development. We show that when expressed in mammalian cells, Ror2, but not Ror1, associates with the melanoma-associated antigen (MAGE) family protein, Dlxin-1, which is known to bind to the homeodomain proteins Msx2 and Dlx5 and regulate their transcriptional functions. This association requires the cytoplasmic C-terminal region of Ror2, containing proline-rich and serine/threonine-rich domains, and the C-terminal necdin homology domain of Dlxin-1. Interestingly, the cytoplasmic C-terminal region of Ror2 is missing in patients with brachydactyly type B. Interestingly, transient expression and immunohistochemical analyses reveal that both Dlxin-1 and Msx2 are co-localized in the nuclei in the absence of Ror2. In the presence of Ror2, Dlxin-1 is co-localized with Ror2 at the membranous compartments and Msx2 is retained in the nuclei. It was also found that the majority of cellular Dlxin-1 is retained in the membrane fractions of wild-type but not Ror2-/- mouse embryonic fibroblasts. Furthermore, we show that transcriptional activity of Msx2, irrespective of Ror2 kinase activity, is regulated by ectopic expression of Ror2 using a reporter plasmid containing the WIP element. Thus, Ror2 sequesters Dlxin-1 in membranous compartments, thereby affecting the transcriptional function of Msx2.

tion, differentiation, migration, and death (1). The Ror family of RTKs are orphan RTKs, characterized by the presence of extracellular Frizzled-like cysteine-rich domains, membraneproximal Kringle domains, and intracellular distal proline-rich domains that are assumed to mediate protein-protein interactions (2)(3)(4)(5)(6)(7)(8)(9). Pairs of structurally related Ror family RTKs are found in Drosophila (Dror and Dnrk) and mammals (Ror1 and Ror2). Because the spatiotemporal expression patterns of Ror1 and Ror2 overlap considerably, and both transcripts are detected in the face, limbs, heart, and lungs during mouse embryogenesis (10,11), it has been assumed that the developmental functions of Ror1 and Ror2 may be at least partially redundant.
Previous genetic studies have demonstrated that both mouse Ror2 and Ror1 play important roles in developmental morphogenesis, in particular in skeletal and cardiac development (12)(13)(14). Mice lacking Ror2 exhibit dwarfism, short limbs (with mesomelic dysplasia) and tails, facial abnormalities, ventricular septal defects, and respiratory dysfunction resulting in neonatal lethality (12,14). Ror1-deficient mice also die soon after birth due to respiratory dysfunction, yet they do not show any apparent skeletal or cardiac phenotypes (13). Furthermore, Ror1/Ror2 double mutant mice exhibit markedly enhanced skeletal and cardiac abnormalities compared with Ror2 mutant mice, indicating that Ror1 and Ror2 interact genetically and functionally during the development of these organs (13). These findings demonstrate the pleiotropic and specific, yet partially redundant, functions of Ror2 in mouse development. Interestingly, it has recently been reported that mutations within Ror2 are responsible for brachydactyly type B (BDB), a dominant skeletal disorder characterized by hypoplasia/aplasia of distal phalanges (15,16), and Robinow syndrome, a recessive condition characterized by short stature, limb bone shortening, segmental defects of the spine, and a dysmorphic facial appearance (17,18) in humans. This finding further indicates the crucial function of Ror2 in developmental morphogenesis. However, little is known about the signaling mechanisms mediated by Ror1 and Ror2.
To gain insights into Ror1-and/or Ror2-mediated signaling, yeast two-hybrid screening was employed to identify proteins that interact with Ror1 and/or Ror2, and a cDNA encoding Dlxin-1, a member of the melanoma-associated antigen (MAGE) family (19 -21), was obtained. Dlxin-1 has been shown to associate with the homeodomain proteins Msx2 and Dlx5 and to regulate the transcriptional function of Dlx5 (22). The Rat ortholog of Dlxin-1, designated NRAGE or MAGE-D, has also been characterized (23,24). NRAGE has been shown to bind to the p75 neurotrophin receptor (p75NTR) and to confer nerve growth factor-dependent apoptosis on developing neuronal cells (24). In addition, it has been reported that NRAGE interacts with an inhibitor of apoptosis protein (IAP) and augments interleukin-3 withdrawal-induced apoptosis of the promyeloid leukemic cell line 32D (25). Furthermore, it has recently been shown that NRAGE induces caspase activation and activates a mitochondrial apoptotic cascade through a c-Jun NH 2 -terminal kinase-and c-Jun-dependent pathway (26). Here we show that Ror2, but not Ror1, associates with Dlxin-1 when expressed in mammalian cells. The association of Ror2 with Dlxin-1 requires the cytoplasmic C-terminal region of Ror2, which containing proline-rich and serine/threonine-rich domains. This region is deleted in Ror2 from BDB patients, suggesting a possible role of Dlxin-1 in the pathogenesis of BDB. Interestingly, although Dlxin-1 and Msx2 are co-localized in the nuclei in the absence of Ror2, the subcellular distribution of Dlxin-1, but not Msx2, is altered by the presence of Ror2. These data indicate that Dlxin-1 is an important mediator of Ror2 function and that its subcellular distribution is regulated in turn by Ror2.
Yeast Two-hybrid Screening-Yeast two-hybrid screening was performed following the manufacturer's instructions (Matchmaker, Clontech). Briefly, the bait plasmids, pAS2-1-Ror1DK and pAS2-1-Ror2DK, were constructed by inserting the cDNA fragments encoding the cytoplasmic regions of kinase-dead version of mouse Ror1 and Ror2, respec-tively, in-frame at the NcoI site of pAS2-1 (Clontech). A yeast strain, Y190, was co-transformed with the bait plasmids pAS2-1-Ror1DK or pAS2-1-Ror2DK along with the mouse embryo (17 days post-coital (d.p.c.)) cDNA library carried in the prey plasmid pGAD10 (Clontech). Approximately 5 ϫ 10 6 transformants were selected for their ability to grow on SD plates lacking histidine (containing 50 mM 3-amino-1,2,4triazole). The colonies grown on SD plates were subsequently analyzed for ␤-galactosidase activity by filter assays, and ϳ20 and 30 ␤-galactosidase-positive clones were obtained when pAS2-1-Ror1DK and pAS2-1-Ror2DK were co-transformed, respectively. Among them, three and six independent clones, respectively, were found to possess the prey plasmids containing the cDNA fragments that encode Dlxin-1 protein, as assessed by sequence analysis with an ABI PRISM 310 Genetic Analyzer (PerkinElmer Life Sciences). Prey plasmids recovered from positive yeast colonies were re-examined by co-transformation into Y190 with the original bait plasmid followed by a ␤-galactosidase assay.
Immunoprecipitation and Immunoblotting-Cell lysates were precleared for 1 h at 4°C with protein A-Sepharose (Amersham Biosciences). The precleared supernatants were then immunoprecipitated with anti-FLAG antibody (M2) conjugated to protein A-Sepharose beads for 2 h at 4°C. The immunoprecipitates were washed five times with 1 ml of the above described lysis buffer and eluted with Laemmli sample buffer. The immunoprecipitates or whole cell lysates were separated by SDS-PAGE and transferred to polyvinylidene difluoride membrane filters (Immobilon, Millipore). The membranes were immunoblotted with the respective antibodies, and bound antibodies were visualized with horseradish peroxidase-conjugated antibodies to mouse, rat, or rabbit IgGs using chemiluminescence reagent (Renaissance, PerkinElmer Life Sciences) as described previously (6,32).
Immunofluorescence-Cells grown on coverslips coated by rat-tail collagen were fixed in 4% paraformaldehyde/PBS for 15min at room temperature and then permeabilized with PBS containing 0.1% Triton X-100 for 15 min at room temperature. After blocking in PBS with 10% FCS for 30 min, cells were incubated with primary antibodies, anti-FLAG monoclonal antibody (M2, 1:500), and/or anti-HA monoclonal antibody (3F10, 1:200), in PBS, 10% FCS for 30 min at room temperature. Cells were washed twice with PBS and then incubated with secondary antibodies, Alexa Fluor 546 (anti-mouse IgG antibody, 1:500), and/or Alexa Fluor 488 (anti-rat IgG antibody, 1:200) in PBS, 10% FCS at room temperature for 30 min. After two washes in PBS, the cells were mounted with Pristine Mount (Research Genetics) and analyzed with an uplight fluorescence microscope (Zeiss).
Reporter Gene Assays-HEK293 cells (1.5 ϫ 10 5 cells/well) were seeded onto 24-well plates. Cells were transfected with the pGL2-WIP reporter gene plasmid (27) along with the respective expression vectors as indicated by using LipofectAMINE reagent (Invitrogen). The total amount of DNA (1.0 g) was kept constant with empty vector. For normalization of transfection efficiencies, 0.2 g of Renilla (sea pansy) luciferase expression plasmid (pRL-TK, Promega) was included in transfection experiments. Transfected cells were lysed and subjected to luciferase assays using PicaGene dual sea pansy system (Toyo Ink) following the manufacturer's instructions. The assays were performed in triplicates.
In Situ Hybridization-In situ hybridization analyses were performed essentially as described previously (33). Single strand RNA probes for Ror1, Ror2, and Dlxin1 were synthesized as described previously (11,22).

RESULTS AND DISCUSSION
Identification of Dlxin-1 as a Ror2-interacting Protein-To better understand the roles of Ror1 and Ror2 during development, we performed a yeast two-hybrid screening to identify potential cytoplasmic signaling molecule(s) that can associate with the cytoplasmic regions of Ror1 and Ror2. A whole mouse E17 yeast two-hybrid library was screened using the entire mouse cytoplasmic regions of Ror1 and Ror2 as bait (see "Experimental Procedures"). We identified Dlxin-1, a member of the MAGE family, as a candidate Ror1 and Ror2-interacting protein (22) (data not shown). To test whether Dlxin-1 associ-ates with intact Ror1 and/or Ror2 in mammalian cells, HAtagged Dlxin-1 (full length) and FLAG-tagged Ror1 or Ror2 (full length) were co-expressed in HEK293T cells. As shown in Fig. 1A, HA-tagged Dlxin-1 specifically co-immunoprecipitated with FLAG-tagged Ror2, but not with FLAG-tagged Ror1, indicating that Ror2, but not Ror1, associates with Dlxin-1 in vivo.
We next examined whether or not association of Ror2 with Dlxin-1 requires Ror2 kinase activity. To this end, HA-tagged Dlxin-1 along with either FLAG-tagged wild-type (WT) or a kinase-dead Ror2 mutant (Fig. 1B, Ror2 DK) were expressed in HEK293T cells, and their association was examined by anti-FLAG immunoprecipitation followed by anti-HA immunoblotting. As shown in Fig. 1C (left panel), Ror2 WT, but not Ror2 DK, was phosphorylated on tyrosine residues when overexpressed in 293T cells as revealed by anti-phosphotyrosine immunoblotting of anti-FLAG immunoprecipitates. It was found that both Ror2 WT and Ror2 DK could associate with Dlxin-1 to a similar extent (Fig. 1C, right panel), indicating that this association occurs independent of Ror2 kinase activity. It has been reported that CAM-1, a Caenorhabditis elegans ortholog of the Ror family RTKs, plays several important roles in regulating cellular migration, polarity of asymmetric cell divisions, and axonal outgrowth of neurons during nematode development. It has also been reported that the tyrosine kinase activity of CAN abnormal migration 1 (CAM-1) is required for some of these functions but not for its role in cellular migration (2,34). Thus, it is possible that Ror2-associated Dlxin-1 may mediate the tyrosine kinase-independent function(s) of Ror2.
Expression Patterns of Ror2 and Dlxin-1 Overlap during Mouse Embryogenesis-To verify the physiological significance of the Ror2-Dlxin-1 association observed in HEK293T cells, we compared the embryonic expression patterns of Ror1, Ror2, and Dlxin-1 by whole-mount in situ hybridization analyses on mouse embryos at E10.5 and E12.5 (Fig. 2, data not shown). At E10.5, Ror2 and Dlxin-1 exhibited remarkably similar expression patterns, especially in the pharyngeal arches and limb buds, whereas Ror1 transcripts exhibited a somewhat different pattern of localization (Fig. 2). Both Ror2 and Dlxin-1 were expressed throughout the limbs, whereas Ror1 expression was more restricted to the proximal regions of the limb buds. However, the expression patterns of these three transcripts were similar in the pharyngeal arches. At a later stage (E12.5), Dlxin-1, like Ror2, was expressed in the perichondrium of the digits and the marginal regions of the limbs; this overlapping expression pattern was maintained at E13.5 (Ref. 11; data not shown). On the other hand, Ror1 transcripts were detected in the anterior and posterior portions of the limbs (Ref. 11; data not shown). The results reveal that the expression patterns of Ror2 and Dlxin-1 overlap significantly, especially in the developing face and limbs. This suggests that Ror2 does indeed associate with Dlxin-1 during mouse embryogenesis.
The C-terminal Proline-rich Domain of Ror2 and the C-terminal Necdin Homology Domain (NHD) of Dlxin-1 Are Required for the Association of Ror2 with Dlxin-1-To identify a region(s) within Ror2 that is required for its association with Dlxin-1, we generated two truncated mutants of Ror2: Ror2 ⌬C and Ror2(Rob) (see Fig. 1B). Ror2 ⌬C lacks the C-terminal amino acids 788 -944, containing the Pro-rich and Ser/Thr-rich 2 domains, whereas Ror2(Rob) lacks amino acids 502-944, containing the Pro-rich, Ser/Thr-rich 1, and Ser/Thr-rich 2 domains and most of the kinase domain. We evaluated the abilities of these mutants to associate with Dlxin-1 in HEK293T cells. As shown in Fig. 3A, neither Ror2 ⌬C nor Ror2(Rob) could associate with Dlxin-1, indicating that the C-terminal region of Ror2, containing the Pro-rich and Ser/Thrrich 2 domains, is required for association. Although the cytoplasmic tyrosine kinase domains of Ror1 and Ror2 exhibit an amino acid identity of Ͼ70%, their C-terminal regions exhibit a lower degree of similarity (an amino acid identity of Ͻ30%). It should also be noted that Ror2 isolated from BDB patients exhibits frameshift and non-sense mutations that generate a truncated version of Ror2 lacking either the C-terminal proline-rich and serine/threonine-rich domains or the entire cytoplasmic region (15,16). It has been reported that p75NTR associates with the rat ortholog of Dlxin-1 via the cytoplasmic juxtamembrane domain (amino acids 276 -328) of p75NTR (24). However, no apparent amino acid sequence similarity was found between the cytoplasmic C-terminal region of Ror2 and the cytoplasmic juxtamembrane domain of p75NTR. It will be of interest to test whether Ror2 and p75NTR can compete for association with Dlxin-1 (NRAGE) in cells.
Dlxin-1 belongs to the MAGE family of proteins, characterized by the presence of an NHD in their C-terminal portions (Refs. 22 and 24; see Fig. 3B). Necdin is one of the best characterized proteins in this family, and has been shown to associate with E2 promoter binding factor 1 (E2F1) and p53 and mimic the function of retinoblastoma (pRb) during cell cycle progression (35,36). It therefore seems likely that the NHD is involved in protein-protein interaction. In the central region of Dlxin-1 protein, there is a repeated motif, consisting of the sequence WQXPXX, which is hypothesized to serve as site of interaction with Dlx5 (22, see Fig. 3B). The N-terminal domain of Dlxin-1 does not show significant homology to any protein in the public data bases. We thus examined several deletion mutants of Dlxin-1 for their association with Ror2, as shown in Fig. 3C. We found that HA-tagged Dlxin-1 (⌬N), but not Dlxin-1 (DlxBD) or Dlxin-1 (⌬C), co-immunoprecipitated with FLAG-tagged Ror2. This indicates that the NHD of Dlxin-1 is responsible for its association with Ror2.
Both the WQXPXX Repeat (25 WQXPXX) and the NHD of Dlxin-1 Are Required for Its Association with Msx2-Although Dlxin-1 was originally identified as a Dlx5-interacting protein, it was subsequently shown that Dlxin-1 associates with Dlx5, Dlx7, and Msx2 (22). Dlxin-1 associates with Msx2 more efficiently than with Dlx5 or other Dlx family proteins, including Dlx7 (22, data not shown). We examined which domain(s) of Dlxin-1 is responsible for its interaction with Msx2. FLAGtagged Msx2 and HA-tagged Dlxin-1 (⌬C), Dlxin-1 (DlxBD), or Dlxin-1 (⌬N) were expressed in HEK293T cells, and their as-sociation was examined by anti-FLAG immunoprecipitation followed by anti-HA immunoblotting. As shown in Fig. 3D, HA-tagged Dlxin-1 (DlxBD) and Dlxin-1 (⌬N), but not Dlxin-1 (⌬C), co-immunoprecipitated with FLAG-tagged Msx2, indicating that both the WQXPXX repeat domain (25 WQXPXX) and the NHD of Dlxin-1 are required for association with Msx2. Considering the fact that the NHD of Dlxin-1 is required for association with both Ror2 and Msx2, it was assumed that the association of Ror2 and Msx2 with Dlxin-1 is competitive. Indeed, we found that Msx2 was not co-immunoprecipitated with Ror2 even in the presence of Dlxin-1 (Fig. 3E). On the other hand, the association between Ror2 and Dlxin-1 was almost completely unaffected by the absence or presence of Msx2 (Fig.  3E). This is consistent with competitive binding between Msx2 and Ror2 for Dlxin-1, if one assumes that the interaction between Ror2 and Dlxin-1 predominates to the extent that it prevents any interaction between Msx2 and Dlxin-1.
Subcellular Localization of Dlxin-1 Is Regulated by Msx2 and Ror2-To understand the biological relevance of the observed molecular associations of Ror2 and Msx2 with Dlxin-1, we examined the subcellular distribution of Ror2, Dlxin-1, and of Msx2 proteins in 293 cells by immunofluorescence (see ''Experimental Procedures''). When FLAG-tagged Ror2, HA-tagged Dlxin-1, and FLAG-tagged Msx2 were individually expressed in 293 cells, Ror2 was found to be localized in the membranous compartments (mainly the plasma membrane and endoplasmic reticulum), whereas Dlxin-1 and Msx2 were localized exclusively in the cytoplasm and nuclei, respectively (Fig. 4A). We found that HA-tagged Dlxin-1 was co-localized with either FLAG-tagged Ror2 or Msx2 when expressed in 293 cells (Fig. 4, B and C). This is consistent with our previous biochemical analyses (see Figs. 1A and 3D). Dlxin-1 is co-localized with Msx2 in the nuclei (Fig. 4B), whereas Dlxin-1 is co-localized with Ror2 at the plasma membrane and endoplasmic reticulum (Fig. 4C). On the other hand, when HA-tagged Ror2 and FLAGtagged Msx2 were coexpressed in 293 cells, Ror2 and Msx2 proteins were detected at the membranous compartments and in the nuclei, respectively, and their intracellular distribution did not overlap at all (Fig. 4D). Importantly, when all three molecules were expressed in 293 cells, HA-Dlxin-1 was detected exclusively at the plasma membrane and endoplasmic reticulum (Fig. 4E, green tag), whereas, as expected, HA-Ror2 plus Flag-Msx2 were detected in the membranous compartments and in the nuclei (Fig. 4E, red tag). These results indicate that Dlxin-1 is co-localized with Ror2, but not with Msx2 in the membranous compartments, whereas Msx2 alone is retained in the nuclei. This is consistent with our biochemical analyses showing that Ror2 and Msx2 associate competitively with Dlxin-1, with Ror2 binding predominating. Co-localization of Dlxin-1 with Ror2, but not with Msx2, was further confirmed by a similar immunostaining analysis using HA-tagged Dlxin-1, HA-tagged Ror2, and FLAG-tagged Msx2 expressed simultaneously in 293 cells (data not shown).
The Ror2 Mutant, Ror2 ⌬C, Fails to Sequestrate Dlxin-1 in Intracellular Membranous Compartments-The Ror2 mutant Ror2 ⌬C fails to associate with Dlxin-1 in a co-immunoprecipitation assay (see Fig. 3A). We therefore next examined whether Ror2 ⌬C is co-localized with Dlxin-1. When FLAG-Ror2 ⌬C was expressed by itself in 293 cells, it localized mainly at the plasma membrane and endoplasmic reticulum (Fig. 5A), similar to wild-type Ror2 (Ror2 WT). When FLAG-tagged Ror2 ⌬C and HA-tagged Dlxin-1 were co-expressed, Ror2 ⌬C was detected in the membranous compartments, whereas Dlxin-1 stained diffusely throughout the cytoplasm (Fig. 5B). As expected, when HA-tagged Ror2 ⌬C and FLAG-tagged Msx2 were co-expressed in 293 cells, both proteins were detected in the On the other hand, as shown in B and C, expression of Ror2 and Dlxin-1 extended throughout the limbs. It should also be noted that Ror2 and Dlxin1 were expressed similarly in the pharyngeal arches. membranous compartments and in the nuclei, respectively, and their intracellular distribution showed no overlap at all (Fig. 5C). When HA-tagged Dlxin-1, FLAG-tagged Ror2 ⌬C, and FLAG-tagged Msx2 were co-expressed in 293 cells, HA-Dlxin-1 was detected exclusively in the nuclei (Fig. 5D, green  tag), whereas Ror2 ⌬C plus Msx2 exhibited cell surface, reticular, and nuclear staining (Fig. 5D, red tag). This result suggests that Dlxin-1 co-localizes with Msx2 in the nuclei, whereas Ror2 ⌬C localizes in the membranous compartments. Co-localization of Dlxin-1 with Msx2, but not Ror2 ⌬C, was also verified by a similar immunostaining analysis using HA-tagged Dlxin-1, HA-tagged Ror2 ⌬C, and FLAG-tagged Msx2 co-expressed in 293 cells (data not shown). Taken together, these results indicate that the Ror2 mutant, Ror2 ⌬C, fails to sequester Dlxin-1 in the membranous compartments. As expected, another Ror2 mutant (Ror2(Rob)) bearing more extensive cytoplasmic deletion than Ror2 ⌬C (see Fig. 1B) also failed to sequester Dlxin-1 in the membranous compartments (data not shown).
Subcellular Distribution of Dlxin-1 Is Regulated by Ror2-To further confirm that Ror2 sequesters Dlxin-1 in the membranous compartments, we performed subcellular fractionation of pression of Dlxin-1 did not differ significantly between the wild-type and Ror2 Ϫ/Ϫ MEFs (data not shown), suggesting that expression of Dlxin-1 is unaffected by Ror2. As shown in Fig. 6, the vast majority of cellular Dlxin-1 was localized in the membrane fractions in wild-type MEFs, with somewhat lower levels present in the cytosol. On the other hand, in Ror2 Ϫ/Ϫ MEFs, only a small percentage of cellular Dlxin-1 was detected in the membrane fraction (Fig. 6), whereas the amount of cytosolic Dlxin-1 was comparable between Ror2 Ϫ/Ϫ and wild-type MEFs. The amount of total cellular Dlxin-1 was also comparable between Ror2 Ϫ/Ϫ and wild-type MEFs (Fig. 6). The results indicate that a large proportion of cellular Dlxin-1 in Ror2 Ϫ/Ϫ MEFs may be localized in the nucleocytoskeletal fractions. This result suggests that Ror2 is responsible for the subcellular localization of Dlxin-1 and that it normally sequesters Dlxin-1 in the membrane fractions.
Transcriptional Activity of Msx2 Is Modulated by Ectopic Expression of Ror2-To pursue the functional significance of observed sequestration of Dlxin-1 by Ror2 in the membranous compartments, we examined whether expression of Ror1 WT, Ror2 WT, or Ror2 DK affects the transcriptional activity of Msx2. To this end, luciferase assays were performed with 293 cells transfected with the respective expression vectors along with a reporter plasmid containing a WIP element upstream of the simian virus 40 minimal promoter and a luciferase reporter gene (pGL2-WIP) (27). Experiments were carried out without ectopic co-expression of Dlxin-1, because it was found that Dlxin-1 is expressed endogenously in 293 cells at a relatively high level (data not shown). Consistent with a previous report (27), transfection of the Msx2 expression vector alone resulted in drastic transcriptional repression of the WIP reporter plasmid (Fig. 7). As shown in Fig. 7, this transcriptional repression by Msx2 was cancelled at least partially by ectopic expression of Ror2 WT but not Ror1 WT. Interestingly, it was also cancelled at least partially by ectopic expression of Ror2 DK (Fig.  7), suggesting that Ror2, irrespective of its kinase activity, regulates the transcriptional activity of Msx2 in the nuclei by sequestrating Dlxin-1, a transcriptional co-factor for Msx2, in the membranous compartments.
Dlxin-1 Is a Possible Mediator of Ror2 Function-In mammals, the Ror family RTKs consist of two structurally related proteins, Ror1 and Ror2 (3,7). Although both Ror1 and Ror2 play crucial roles during development, Ror2 exhibits more specific functions compared with Ror1 (13). We have shown that Dlxin-1, a member of the MAGE family of proteins, associates with Ror2, but not with Ror1 (Fig. 1). This suggests that Dlxin-1 may be involved in mediating some if not all of the developmental functions of Ror2. Interestingly, Ror2 and Dlxin-1 exhibit remarkably similar expression patterns in the developing face and limbs, in contrast to Ror1, which shows a different pattern (Fig. 2). This suggests that the molecular association between Ror2 and Dlxin-1 may be biologically significant during development. Furthermore, it has been shown that the cytoplasmic C-terminal region of Ror2, containing proline-rich and serine/threonine-rich domains, and the C-terminal necdin homology domain of Dlxin-1 are required for the association between Ror2 and Dlxin-1 (Fig. 3, A and C). It should also be noted that the cytoplasmic C-terminal region of Ror2 is missing in BDB patients, suggesting a possible involvement of Dlxin-1 in the pathogenesis of BDB. Intriguingly, Dlxin-1 is co-localized with the wild-type Ror2 in the membranous compartments as the result of sequestration by Ror2 (Figs. 4 and 6), whereas the mutant Ror2 fails to sequestrate Dlxin-1 in the membranous compartments (Fig. 5).
Until recently, the functions of the MAGE family of proteins were largely unknown, except for the protein necdin. Necdin was shown to mediate growth arrest of postmitotic neurons, presumably by interacting with E2F1 and p53 (20,35,36). Interestingly, it has recently been shown that Dlxin-1 (and its rat ortholog, NRAGE) is a multifunctional adaptor protein: (i) Dlxin-1 associates with the homeodomain proteins Msx2 and Dlx5 and regulates the transcriptional function of the latter (22); (ii) NRAGE associates with p75NTR and is required for nerve growth factor-dependent apoptosis of developing neurons (24); and (iii) NRAGE associates with the IAPs ITA and XIAP and augments the apoptosis of 32D cells upon growth factor withdrawal (25). Similar to our observation that Ror2 and Msx2 bind to an overlapping region within Dlxin-1 (Fig. 3, C   FIG. 6. Distribution of Dlxin-1 in subcellular fractions from MEFs (wild-type and Ror2 ؊/؊ ). Cytosolic and membrane fractions as well as total cell extracts were prepared from MEFs (wild-type and Ror2 Ϫ/Ϫ ) as described under ''Experimental Procedures.'' The amounts of Dlxin-1 in the cytosolic and membrane fractions as well as total cell extracts from MEFs (wild-type and Ror2 Ϫ/Ϫ ) were determined by immunoblotting with anti-Dlxin-1 antibody (anti-DXN-NHD Ab). The amount of proteins recovered in the cytosolic and membrane fractions as well as in total cell extracts is almost comparable between the wild-type and Ror2 Ϫ/Ϫ MEFs. The results shown are representative of three independent experiments. FIG. 7. Transcriptional suppression activity of Msx2 is cancelled partially by co-expression of Ror2 WT and Ror2 DK, respectively. 293 cells were transfected with the respective expression vectors along with pGL2-WIP reporter plasmid as indicated. Cell extracts were prepared and luciferase assays performed as described under ''Experimental Procedures.'' Luciferase activities are expressed as fold increase relative to cells transfected with the empty vector alone. Expression levels of Ror1 WT, Ror2 WT, and Ror2 DK in the respective transfectants were almost comparable as assessed by anti-FLAG immunoblotting (data not shown). Data are expressed as the mean Ϯ S.D. of relative luciferase activity in three replicate wells from one representative experiment of three. and D), it has been reported that NRAGE and TrkA association with p75NTR are also physically exclusive (24).
Collectively, our findings help elucidate the pathophysiological roles of Ror2 during development. Dlxin-1 has previously been shown to associate with and regulate the transcriptional activity of Dlx5 (22), a member of the Dlx family of homeodomain proteins involved in skeletal development. Therefore, it is likely that the transcriptional activity of Msx2, another homeodomain protein involved in skeletal development (37,38), would be affected similarly by its interaction with Dlxin-1. In fact, we show that transcriptional repression of the WIP reporter plasmid by Msx2 was cancelled partially by ectopic co-expression of Ror2 WT and Ror2 DK, respectively (Fig. 7). The result supports the idea that sequestration of Dlxin-1 by Ror2 at the membranous compartments, irrespective of its kinase activity, indirectly affects the transcriptional function of Msx2 in the nuclei. Furthermore, it is likely that stimulation of Ror2 by its cognate ligand may alter the sequestration of Dlxin-1. Hence, the expression levels of Ror2, Dlxin-1, and of Msx2 in a particular cell may be the critical determinants that regulate the behavior and function of the cell under unstimulated or ligand-stimulated conditions. Although the cognate ligand of Ror2 is currently unknown, its identification may help elucidate the mechanisms regulating the subcellular distribution and function of Dlxin-1 and/or Msx2.