Mutations in ROR2 cause dominant brachydactyly type B (BDB1) or recessive Robinow syndrome (RRS), each characterized by a distinct combination of phenotypic features. We here report a novel nonsense mutation in ROR2 (c.1324C>T; p.R441X) causing intracellular protein truncation in a patient exhibiting features of RRS in conjunction with severe recessive brachydactyly. The mutation is located at the same position as a previously described frame shift mutation causing dominant BDB1. To investigate the apparent discrepancy in phenotypic outcome, we analysed ROR2 protein stability and distribution in stably transfected cell lines expressing exact copies of several human RRS and BDB1 intracellular mutations. RRS mutant proteins were less abundant and retained intracellularly, although BDB1 mutants were stable and predominantly located at the cell membrane. The p.R441X mutation showed an intermediate pattern with membrane localization but also high endoplasmic reticulum retention. Furthermore, we observed a correlation between the severity of BDB1, the location of the mutation, and the amount of membrane-associated ROR2. Membrane protein fraction quantification revealed a gradient of distribution and stability correlating with the clinical phenotypes. This gradual model was confirmed by crossing mouse models for RRS and BDB1, yielding double heterozygous animals that exhibited an intermediate phenotype. We propose a model in which the RRS versus the BDB1 phenotype is determined by the relative degree of protein retention/degradation and the amount of mutant protein reaching the plasma membrane.
The majority of genetic disorders are caused by mutations in a single gene, but mutations in different genes that often participate in a common pathway may also result in a single phenotype. On the other hand, distinct mutations in one gene can cause phenotypically different conditions. For example, dominant brachydactyly type B1 (BDB1, MIM 113000) and recessive Robinow syndrome (RRS, MIM 268310), are caused by mutations in a single gene, ROR2 ( 1–4 ). BDB1 is characterized by hypoplasia and/or aplasia of distal phalanges and nails in hands and feet resulting in an amputation-like phenotype. In less severe cases the distal phalanges may be present but they are frequently fused with the middle phalanges (distal symphalangism). RRS, on the other hand, is a completely distinct condition exhibiting features such as short stature, mesomelic limb shortening, hemivertebrae, genital hypoplasia and a characteristic facial dysmorphism ( 5 ). ROR2 encodes a receptor tyrosine kinase possessing extracellular immunoglobulin-like, cysteine-rich frizzled-like and kringle domains; intracellularly, the tyrosine kinase domain is followed by a serine–proline–threonine rich region that is unique for ROR2 and its closely related paralog ROR1 ( 6 ).
All BDB1 associated changes known to date are frame shift and nonsense mutations that cluster in two mutational hot-spots located either immediately N-terminal or C-terminal of the tyrosine kinase domain. Although the mutations located N-terminal of the tyrosine kinase domain are small deletions or duplications resulting in frame shifts, nonsense as well as frame shift mutations have been reported in the C-terminal location. These mutations are predicted to lead to the expression of truncated proteins that always lack the C-terminal serine–proline–threonine rich region but may or may not contain the tyrosine kinase domain. On the basis of the dominant inheritance and the peculiar mutation sites, the mutated protein has been suggested to act as a gain of function, or as dominant negative allele ( 5 ).
Interestingly, the C-terminal truncations result in a more severe phenotype than the N-terminal mutations ( 3 ). The cause for this is so far unknown. In contrast to the BDB1 mutations, the RRS associated changes are scattered throughout the ROR2 molecule consisting of missense mutations, or premature termination of the polypeptide chain (nonsense or frame shift) at extracellular or intracellular positions ( 7 and references therein). It was suggested that mutations in RRS most likely result in a complete loss of function (LOF) ( 1 , 4 , 5 ). This hypothesis is supported by gene inactivation experiments in the mouse. Similar to humans, Ror2-null mice show malformations of the craniofacial skeleton, the vertebrae and ribs, severe mesomelic limb shortening, as well as anomalies of the heart and lung ( 8 ). Chen et al . ( 9 ) have shown that proteins carrying missense mutations in the extracellular domains lead to intracellular retention. It can thus be assumed that all RRS associated ROR2 mutations result in a LOF, presumably through protein degradation, although the dominant BDB1 mutations should be able to reach the cell membrane in order to interfere with normal signalling. However, the intracellular truncations that have been reported in RRS, some of which are in close proximity to the BDB1 mutations, have not been analysed so far.
Here we present a novel recessive nonsense mutation in ROR2 (c.1324C>T/p.R441X) that is located at the identical position as a previously reported dominant BDB1 frame shift mutation (c.1321-25_del5/p.R441fsX15) ( 3 ). Remarkably, the patient who carries the p.R441X mutation shows combined features of RRS and brachydactyly. This argues for a more sophisticated model in which the phenotypic outcome is not so much determined by a simple loss versus gain of function but rather by a model that integrates intermediate/overlapping phenotypes. To this end, we quantitatively analysed the stability and intracellular distribution of different truncated mutant isoforms of ROR2 leading to RRS, BDB1 or causing overlapping phenotypes. We demonstrate that an intricate balance between LOF by intracellular retention and amount of protein that escapes intracellular retention appears to be critical for the phenotypic outcome.
A novel recessive mutation in ROR2 exhibiting features of RRS and severe brachydactyly
The patient was born at term from an unrelated young Omani couple. Both grandmothers were, however, distantly related and originated from the same tribal background. Both parents are healthy and of average height. The patient exhibited a mildly dysmorphic facial appearance with hypertelorism (Fig. 1 A). The limbs showed acromesomelic shortening, dislocated ulna and bilateral, almost symmetric malformations of the hands and feet (Fig. 1 B and F). The hands were broad with short digits, especially affecting fingers 1, 2, 4 and 5. There was preaxial polydactyly of both hands, syndactyly of fingers two and three and a broad-based third finger. Fingernails were short and hypoplastic and were absent on fingers 3 (Fig. 1 C). There was oligodactily of feet with absent toes 2–4. The large toes were of normal size, but the fifth toes were rudimentary. Toenails of the first toes were dysplastic and there was complete absence of the nails of fifth toes (Fig. 1 D).
Radiological examination of the hands at 9 months of age revealed absent carpal bone ossification, five metacarpal bones, and a duplication of the distal elements of both thumbs. The proximal phalanges of digit 2 were bifurcated resulting in a duplication that was fused to the first digit on the radial side and the third digit on the ulnar side. The middle and distal phalanges of digits 4 and 5 were missing completely (Fig. 1 E). Radius and ulna were short and abnormally modelled (Fig. 1 E). Radiology of the feet showed complete absence of the phalanges of toes 2, 3 and 4, one rudimentary phalanx of toe 5 and large toe of the left foot, and two rudimentary phalanges of the large toe on the right foot (Fig. 1 F). Spinal X-rays showed severe abnormalities of the vertebral column with cleft vertebrae and hemivertebrae in the thoracolumbar region (Fig. 1 G).
ROR2 mutation analysis was performed as reported previously ( 3 ). We identified a nonsense mutation in the coding sequence of ROR2, c.1324C>T, which is expected to result in the translation of a truncated protein (p.R441X). The parents were heterozygous for the c.1324C>T mutation and showed no abnormalities.
The c.1324C>T/p.R441X mutation exhibits an intermediate cellular distribution
The finding that a mutation either truncating the protein by a frame shift or a nonsense mutation at the same position result in different phenotypes and modes of inheritance is incompatible with the view that only the position of the mutation determines the phenotypic outcome. Furthermore, it makes expression constructs that do not exactly copy the human mutations, or that are altered by e.g. the addition of a protein tag, inappropriate for mutant analysis. We therefore decided to use the human full-length ROR2 cDNA and introduced the exact patient mutations into this construct. We generated constructs for wild-type (wt) ROR2 and seven human mutations causative for either BDB1 or RRS (Fig. 2 A).
Protein distribution was first analysed in stably transfected Cos-1 cells by immunocytochemical detection of ROR2 with a specific antibody and by co-staining for BAP31 as a marker protein for the endoplasmic reticulum (ER). As expected for a receptor, wt ROR2 was only detected at the cell membrane. ROR2 with the C-terminal truncation (p.W749X) also exhibited almost complete localization to the membrane. We demonstrate that two so far not investigated intracellular truncations causing RRS (p.Q502X and p.W720X) show clear intracellular retention. Similar results were obtained for the mutation p.N620K, which has been reported previously by Chen et al . ( 9 ). The N-terminal truncating mutations resulting in the less severe form of BDB1 (p.Q467fsX57, p.R441fsX15) also showed a clear membrane staining. However, a large fraction of mutant protein was retained in the ER, thus indicating an intermediate phenotype between the C-terminal BDB1 and the RRS mutations. Interestingly, the novel p.R441X mutation showed a protein localization comparable with the N-terminal BDB1 mutants.
For further verification we analysed co-localization of the ROR2 protein with additional markers in transiently transfected Cos-1 cells ( Supplementary Material, Figure S1 ). Membrane labelling with Gap43-GFP demonstrated surface localization of wt ROR2, the BDB1 constructs and also the p.R441X mutant, but not for the RRS mutants. Co-staining with PDI as an alternative ER marker corroborated the intracellular retention of the RRS variants. In addition, we can show partial overlap between the cis -Golgi marker GM130 and the RRS mutants, the p.R441X mutant and also the BDB1 mutant p.R441fsX15 ( Supplementary Material, Figure S1 ). Finally, we analysed if the mutant ROR2 protein located to the lysosomal compartment by co-staining for H4B4. No significant overlap was identified ( Supplementary Material, Figure S1 ).
Quantification of protein levels and distribution reveal a gradient model for BDB1 versus RRS
Analysis of protein distribution by immunocytochemistry demonstrated that the mutant ROR2 protein R441X was partially retained intracellularly, but significant localization to the cell membrane was also observed. Similar results were obtained for the BDB1 mutation p.R441fsX15. Thus, the question remained why these two mutant proteins showed the same intracellular distribution given the difference in clinical phenotypes and inheritance. To address this question we decided to determine the amount of ROR2 protein retained intracellularly versus the protein that located to the cell membrane. To obtain an inducible system with moderate and constant expression levels, we generated stable HEK293 cell lines expressing either wt ROR2 or six different human mutations using the a Flp-in system system (HEK293 FlpIn T-Rex). Using this system we obtained an approx. 20-fold overexpression compared with endogenous ROR2 (data not shown). Immunocytochemistry for ROR2 and the ER marker BAP31 in these cells showed a protein distribution similar to that observed in Cos-1 cells (Fig. 3 A).
First, we analysed the total amount of ROR2 protein in the cells. Protein measurements were normalized to mRNA expression for each individual construct to eliminate variation of transgene expression in the different cell lines. Figure 3 B shows normalized total protein amounts for each construct in comparison with wt, which was set as 1. The BDB1 mutations p.W749X and p.Q467fsX57 showed a weak increase in total ROR2 protein content although the BDB1 construct p.R441fsX15 showed no difference to the wt. In contrast, the p.R441X mutant was significantly less abundant than the wt and the BDB1 mutations. The RRS mutations p.Q502X and p.N620K were reduced even to a higher extent. These results indicate that BDB1 mutant proteins exhibit a higher stability than RRS mutants.
We used a surface biotinylation assay to unequivocally determine the fraction of membrane-localized ROR2 versus intracellular ROR2 (Fig. 3 C). The amount of membrane-localized relative to intracellular ROR2 was determined for each construct and wt ROR2 was set as 1. All samples were normalized to β-actin as loading control for cell lysates. As shown in Fig. 3 C, we observed a gradual decrease of membrane localization when comparing the different mutants. The BDB1 mutant p.W749X was similar to the wt, but the N-terminal BDB1 mutants p.Q467fsX57 and p.R441fsX15 showed a significantly higher retention than wt or the p.W749X mutant. The RRS mutants p.Q502X and p.N620K were almost completely retained intracellularly. The p.R441X mutant was comparable with the N-terminal BDB1 mutations. To exclude global changes in protein trafficking we examined relative distribution of an unrelated protein, pan-cadherin, which was present at equal levels on the surface of all cell lines analysed ( Supplementary Material, Figure S2 ).
This indicated that the intracellular processing of ROR2 mutants is dependent on two parameters: protein stability and protein distribution. To better discriminate the BDB1 from the RRS mutations, we plotted the relative membrane localization of ROR2 versus the total protein amount on a double logarithmic scale (Fig. 3 D). Using this form of presentation, BDB1 mutations cluster in the left upper quadrant whereas the RRS mutations localize to the left lower quadrant. Again, the p.R441X mutation appears in an intermediate position.
For further clarification we aimed at determining the absolute amount of membrane-localized ROR2 for each construct/cell line. Values for total ROR2 protein quantification obtained in three individual experiments were used as a factor and multiplied with the surface biotinylation values obtained from the same experiment. Mean values with standard errors from the combined three experiments are depicted in Figure 3 E. This calculation demonstrates a continuous gradual decrease in surface ROR2 protein along the allelic series tested in our experiments. Again, the mutations exhibiting intermediate phenotypes in homozygous human patients are placed at intermediate positions between the other BDB1 and RRS mutations. Concordant with the higher total ROR2 protein amount observed for the frame shift mutation p.R441fsX15, this mutation also shows a statistically significant higher total surface ROR2 amount than the nonsense mutation p.R441X.
An allelic series of ROR2 mutations in the mouse confirms BDB1 versus RRS phenotype gradient
The Ror2 −/− mouse ( 10 ) exhibits complete loss of Ror2 function and shows numerous features of RRS including craniofacial and vertebral malformations, heart defect and severe mesomelic limb shortening ( 8 ). Recently, a novel mouse mutant for Ror2 that carries the exact copy of a human BDB1 mutation (Ror2 W749X ) was generated. This mouse mutant exhibits recessive brachydactyly with complete lack of the middle phalanges (p2). In the homozygous state, this mouse also shows features of RRS, albeit to a weaker degree than the Ror2 −/− mouse ( 11 ). We used these mouse mutants to challenge our gradient model in vivo .
The mesomelic limb shortening and p2 shortening were used as readout for RRS and BDB1 features, respectively (Fig. 4 ). The radius of newborn Ror2 W749X/W749X mice showed moderate shortening compared with their wt littermates, whereas Ror2 −/− mice showed a severe reduction in size (Fig. 4 A). Consistent with the biochemical data presented above, mice carrying one truncating and one functional null allele (Ror2 W749X/− ), thus possessing a reduced amount of membrane-bound truncated Ror2, displayed an intermediate phenotype. The Ror2 W749X/W749X mouse showed the most severe brachydactyly phenotype exhibiting an aplasia of p2, although Ror2 −/− mice had only a minor hypoplasia of p2 (Fig. 4 B). Altogether, Ror2 W749X/− mice had a more severe phenotype than Ror2 −/− mice demonstrating a gain of function of the remaining W749X allele on a null background.
The allelic series presented here supports our in vitro findings demonstrating that the p.W749X mutant functions as a gain of function allele and as such induces a brachydactyly phenotype in a dose dependent manner. The mesomelic shortening, in contrast, is caused by a loss of Ror2 function. Figure 4 C illustrates the observed reciprocal gradient, in which the RRS phenotype increases towards the Ror2 −/− mutant and the BDB1 phenotype increases towards the Ror2 W749X/749X mutant.
Although there are numerous examples for mutations in a single gene causing different recessive or dominant conditions, the mechanisms for this phenomenon can be quite different. Often, the type of mutation (e.g. truncating versus missense) or its location (e.g. affecting different functional domains) will determine the phenotype and/or the mode of inheritance. In BDB1 and RRS the position of the mutation is of importance, on the basis of the finding that only truncating mutations immediately N-terminal or C-terminal of the tyrosine kinase domain lead to BDB1, whereas the RRS associated mutations are scattered throughout the molecule. In addition, the type of mutation appears to play a role, since all N-terminal BDB1 mutations reported so far are frame shift mutations. In contrast, nonsense and frame shift mutations have been reported to be associated with the C-terminal BDB1 mutations. The situation is further complicated by the finding that nonsense mutations that are located within the tyrosine kinase domain, only 35 amino acids apart from the N-terminal and 29 amino acids apart from the C-terminal BDB1 mutations, result in RRS.
We here report the rare finding that mutations at the same amino acid position, one resulting in a termination of the polypeptide chain, the other causing a frame shift eventually also leading to a premature stop, cause different phenotypes with distinct modes of inheritance. A patient homozygous for the dominant BDB1 mutation p.R441fsX15 has been described previously. Although the parents showed BDB1 with missing distal phalanges, this individual had almost complete absence of phalanges and nails in the hands and feet ( 3 ). In addition he had mesomelic limb shortening and multiple vertebral malformations but no facial dysmorphism, thus displaying some but not all of the hallmarks of RRS.
The novel nonsense mutation we describe here (p.R441X) terminates the ROR2 polypeptide chain at the same position, but without adding additional amino acids. In contrast to the p.R441fsX15 mutation, both parents had normal hands and feet. In concurrence with the recessive inheritance the patient showed features of RRS including mild facial dysmorphism, vertebral malformations and mesomelic limb shortening. In addition a brachydactyly phenotype was seen that is somewhat similar to the p.R441fsX15 mutation in the feet (absence of toes 2–5), but the hands show a more complex picture with polydactyly, syndactyly and distal hypoplasia. Thus, the frame shift mutation causes a dominant BDB1 phenotype in which the homozygous state results in severe limb reduction defects and Robinow-like features, whereas the nonsense mutation is recessive, also with Robinow-like features and a limb phenotype that is intermediate between BDB1 and homozygosity for the BDB1 frame shift mutation.
We hypothesized that one explanation for the observed discrepancy in phenotypes may be altered cellular response to the mutated proteins. Receptors like ROR2 are generally processed through the ER, folded and thereafter transported to the Golgi apparatus where post-translational modification takes place. Finally, an elaborate trafficking system transports the receptor to the cell membrane. Mutated and misfolded protein is frequently recognized in the ER and subsequently retained and subjected to protein degradation. In general, mutations occurring in highly conserved parts of proteins and/or within complexly folded domains are more prone to misfolding (and thus degradation) than mutations found in regions displaying a looped structure. This may contribute to the low protein abundance and also minimal membrane trafficking observed for the intracellular RRS mutations p.Q502X and p.N620K. The major difference between the p.R441X mutation and the mutations associated with full RRS appears to be their intracellular distribution. Although all classical RRS mutants are almost completely retained in the ER, a significant fraction of the R441X protein is able to reach the cell membrane. Again, this is likely to be caused by less protein misfolding due to the mutation being located in a loop region versus in a conserved domain. This may lead to an escape from the intracellular recognition machinery and from protein retention and degradation.
Our results show that the p.R441X mutant has a similar intracellular distribution but a significantly lower total protein level than the p.R441fsX15 mutant. Thus, protein stability appears to be a major difference between those mutants ultimately leading to lesser membrane-associated R441X protein. How the 15 amino acid peptide resulting from the p.R441fsX15 mutation affect protein stability remains to be determined.
Our findings also provide a possible explanation for the different degrees of severity observed in BDB1 patients ( 3 ). In general, mutations N-terminal of the tyrosine kinase domain cause a less severe phenotype than those located C-terminally. Our results show that protein levels are comparable between BDB1 mutants, but the N-terminal mutations show a higher degree of intracellular retention than the C-terminal p.W749X mutation thus resulting in a higher degree of LOF and, consequently, a milder BDB1 phenotype.
Nonsense-mediated RNA decay (NMD) is widely accepted as a mechanism by which the cell inhibits the production of truncated proteins ( 12 ). Van Bokhoven and colleagues ( 4 ) reported lower amplification efficiency of mRNA from the W720X allele suggesting that NMD is implicated in the pathogenesis of the RRS LOF phenotype. Contrasting this, Ben-Shachar et al . ( 13 ) recently demonstrated NMD for the extracellular truncating mutations, but specifically not the p.W720X mutation, again leaving the question open how this mutation causes RRS. The massive ER localization with lacking membrane localization we demonstrate (Fig. 2 B) argues that protein retention and degradation plays the major role in this mutant, drastically reducing the amount of ROR2 protein. In our system effects exerted by NMD are unlikely, since we used the full-length cDNA for expression and normalized protein amounts measured by western blot to mRNA levels quantified by real-time PCR. Generally, the constructs were variably expressed but this did not correlate with the type of mutation, and can thus be attributed to intrinsic differences between the different cell lines.
Our study also highlights that small alteration of the polypeptide chains of protein constructs may result in significant changes in protein stability and intracellular distribution. As a consequence, protein tags might influence protein stability and/or trafficking and ultimately distort quantification measurements and other procedures. Indeed, C-terminally FLAG-tagged ROR2-constructs showed an overall higher amount of intracellular retention than the untagged constructs (not shown). Similar observations demonstrating an influence of tags on protein distribution or stability have been reported before ( 14–16 ). To further stabilize the experimental situation we used the Flp-in system, which allows single-copy integration and thus a moderate level of expression. We have also generated stable cell lines from transfected Cos-1 cells and used them for surface biotinylation. Although these assays showed the same general tendency as the experiments performed in HEK293 FlpIn T-Rex cells, the results were more variable and difficult to interpret indicating that single- or low-copy expression systems are preferable. This has to be taken into account when designing studies analysing the consequence of human mutations on protein stability and/or trafficking. Thus, for proper analysis untagged constructs exactly copying human mutations that are expressed at low levels in a controlled system should be used.
The fact that FLAG-tagged ROR2 constructs in cell culture showed a higher degree of intracellular retention also offers a possible explanation for the recessive nature of the BDB1 phenotype observed in the Ror2 W749X mouse ( 11 ), which carries a FLAG-tag fused to the truncated Ror2. It is possible that the total truncated Ror2 protein amount reaching the cell membrane is too low to cause a dominant phenotype but is high enough to cause a BDB1-like phenotype in the homozygous situation, thus reflecting the situation observed in the patient harbouring the p.R441X mutation.
In summary our study provides a quantitative biochemical explanation for the variable severity seen in BDB1 and for the appearance of overlapping features of BDB1 and RRS in single patients (Fig. 5 ). In dominant BDB1 the appearance of RRS-like features in precluded by the presence of one wt allele. Truncating alleles found in both conditions appear to have dual functions: mutant proteins exhibit a partial LOF due to intracellular retention and decreased protein stability, but can also act via a gain of function whenever they are able to reach the cell membrane.
One interesting feature of truncated ROR2 proteins is that they seem to be able to fulfil a residual ‘normal’ function, as both homozygous patients carrying the p.R441fsX15 and p.R441X mutations show some but not all features of RRS. In concordance the Ror2 W749X/W749X mouse mutant shows a less severe RRS-like phenotype than the Ror2 −/− mutant. This could be attributed to functions of ROR2 that are independent of the presence of the intracellular part as it was demonstrated in Caenorhabditis elegans ( 17 , 18 ). A possible function of a membrane-tethered variant of ROR2 lacking the intracellular domains may be to serve as a co-receptor as it has been suggested for Wnt signalling in different contexts ( 19 ). Our results argue for a model in which the phenotypic outcome of the ROR2 mutations is determined by two threshold levels: the degree of degradation/retention determines the RRS phenotype, whereas the amount of truncated protein that reaches the cell membrane determines the severity of the BDB1 phenotype. A mixture of both effects can result in a balance of gain- and loss of function and, consequently, overlapping phenotypes.
MATERIALS AND METHODS
Patients, mutation analysis
The family described here originates from the Sultanate of Oman, a country with a population of 2 million and a comprehensive healthcare system with specialized clinics based in the two major hospitals in the capital Muscat. The affected individuals as well as the parents were examined by A.R. and S.M. Blood was taken after informed consent and DNA extracted according to standard procedures. Mutation analysis was performed as described previously ( 3 ).
The human ROR2 coding sequence was amplified from HEK293-derived cDNA and cloned into pcDNA3.1(+) and pcDNA5/FRT/TO vectors (Invitrogen) using a PCR-based strategy. All mutations were introduced employing the QuickChange Site-Directed Mutagenesis Kit (Stratagene) according to the manufacturer's instructions. The pCS2/GAP43-GFP construct was kindly provided by E. M. De Robertis ( 20 ).
Cell culture and transfection
All cell lines were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (Biochrome), 1% Pen/Strep and 2 m ml -glutamine. Culture medium for the HEK293 FlpIn T-Rex cell line (Invitrogen) was additionally supplemented with 100 µg/ml Zeocin as recommended. For transient expression experiments Cos-1 cells were transfected with pcDNA3.1(+) hROR2 constructs using PolyFect (Qiagen). Inducible expressing HEK293 FlpIn T-Rex cell lines were produced by co-transfection of each pcDNA5/FRT/TO vector with the Flp recombinase expression plasmid pOG44 followed by selection with 200 µg/ml hygromycin (FlpIn system, Invitrogen). Twenty-four hours prior to each experiment, the expression of the hROR2 variants was induced with 1 µg/ml tetracycline.
Twenty-four hours post-transfection or induction, Cos-1/HEK293 FlpIn T-Rex cells were fixed in ice-cold methanol and blocked in 10% fetal bovine serum (Biochrome). Anti-ROR2 (1:500, AF2064, R&D Systems) and either anti-BAP31 (1:500, ALX-804-601, Alexis), anti-PDI (1:1000, SPA-891, Stressgen), anti-GM130 (1:1000, 610822, BD Biosciences) or anti-H4B4 (1:500, Developmental Studies Hybridoma Bank) antibodies were applied in indicated dilutions. Primary antibodies were detected with Alexa Fluor 564 donkey anti-goat and Alexa Fluor 488 donkey anti-mouse secondary antibodies (Invitrogen).
Total RNA was isolated from HEK293 FlpIn T-Rex cell lines using peqGOLD TriFast (Peqlab) and reversely transcribed (Taqman Reverse Transcription, Applied Biosystems). The quantitative PCR reaction was performed on ABI Prism 7900 HT (Applied Biosystems) by using SYBR Green PCR Master Mix (Applied Biosystems) and following primers: rt-hROR2-F 5′-CATGGCAGACAGGGCAGC-3′, rt-hROR2 R 5′-TTCTGTGTGTCATCAGCGCC-3′, rt-hACTB-F 5′-TCAAGATCATTGCTCCTCCTGAG-3′, rt-hACTB-R 5′-ACATCTGCTGGAAGGTGGACA-3′. The expression levels of the ROR2 mutants were calculated relative to actin and were normalized to ROR2 wt .
For immunoblotting, induced HEK293 FlpIn T-Rex cells were lysed (50 m m HEPES pH 7.4, 50 m m NaCl, 10 m m EDTA, 10% glycerine, 1% Triton-100, 5 µ m phenylmethanesulfonyl fluoride (P7626, Sigma), 5 µ m sodium fluoride (S7920, Sigma), 5 µ m activated sodium orthovanadate (pH10, S7920, Sigma), 1 µg/ml Aprotinin (A6279, Sigma), 10 µ m Pepstatin (P4265, Sigma), 10 µ m Leupeptin (L2884, Sigma), subjected to SDS-PAGE and transferred to PVDF membranes. The membranes were simultaneously incubated with anti-ROR2 and anti-actin (A2066, Sigma-Aldrich) antibodies followed by subsequent incubation with mouse anti-goat-IgG HRP (DC08L, Calbiochem) and goat anti-rabbit-IgG HRP (DC03L, Calbiochem). Signals were detected using chemiluminiscence substrate (Roti-Lumin, Carl Roth) by the LAS-4000 Imaging System (Fuji) and quantified with the AIDA Image Analyzer (Raytest). The protein levels of the ROR2 mutants were calculated relative to actin and were normalized to ROR2 wt.
Material for RNA and protein isolation was harvested at the same day. In order to obtain a measure for the stability of mutant ROR2 protein in the cell, the ratio of protein and RNA was determined in three independent experiments, each experiment was averaged and a standard deviation was calculated.
For statistical analysis the individual experiments were averaged and the standard error was calculated from the individual standard deviations according to ( 21 ) and statistical significance was calculated by using a Student's t test.
Surface biotinylation, quantification of membrane-bound ROR2
HEK293 FlpIn T-Rex cell lines were cultivated in 6-well plates, washed with PBS and treated with 1 mg/ml EZ-Link Sulfo-NHS-LC-Biotin Reagent (Pierce) for 30 min at 4°C. Excess reagent was neutralized with 100 m m glycine/PBS prior to lysis in 200 µl lysis buffer (as described above). Twenty microlitre of cell lysate was retained as Ror2 expression control, whereas 140 µl were subjected to 50 µl immobilized streptavidin (Pierce) for 1 h. After incubation, 40 µl supernatant was reserved for the determination of intracellular Ror2, and the streptavidin beads were boiled in SDS sample buffer. The complete fraction containing biotinylated protein, the supernatant and the untreated lysate were loaded on an SDS-PAGE gel and treated as described above. Representative original data from two experiments are shown in Supplementary Material, Figure S2 . In order to calculate the relative fraction of ROR2 at the plasma-membrane, the ratios of biotinylated ROR2 and the corresponding intracellular fraction were determined in six independent experiments. The resulting values were normalized to the ROR2 wt ratio, averaged and statistical significance was calculated as described above. As control, the relative amount of membrane-localized endogenous pan-cadherin was determined in three independent experiments as described. The protein was detected using anti-pan-cadherin (Abcam, ab6528) and goat anti-mouse total Ig HRP (Calbiochem, DC08L).
Mouse strains, skeletal preparations
This work was supported by a grant from the Deutsche Forschungsgemeinschaft (DFG; SFB 577) to S.S. and S.M.
We acknowledge the expert technical assistance from Kathrin Seidel. We like to thank Peter N. Robinson for helpful discussions and help with statistical analysis.
Conflict of Interest statement . None declared.