SEARCH

SEARCH BY CITATION

Keywords:

  • fjx1;
  • four-jointed;
  • Notch;
  • fjx1 binding sites

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

The mouse fjx1 gene was identified as a homologue to the Drosophila gene four-jointed (fj). Fj encodes a transmembrane type II glycoprotein that is partially secreted. The gene was found to be a downstream target of the Notch signaling pathway in leg segmentation and planar cell polarity processes during eye development of Drosophila. Here, we show that fjx1 is not only conserved in vertebrates, but we also identified the murine fjx1 gene as a direct target of Notch signaling. In addition to the previously described expression of fjx1 in mouse brain, we show here that fjx1 is expressed in the peripheral nervous system, epithelial cells of multiple organs, and during limb development. The protein is processed and secreted as a presumptive ligand. Through the use of an fjx1-AP fusion protein, we could visualize fjx1 binding sites at complementary locations, supporting the notion that fjx1 may function as a novel signaling molecule. Developmental Dynamics 234:602–612, 2005. © 2005 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

The fjx1 gene has been described as the murine homologue of the Drosophila gene four-jointed (fj; Ashery-Padan et al.,1999). Nothing is known thus far about the function of this gene, but in the Drosophila gene, mutations lead to various defects. The most striking phenotype, which was also eponymous for the gene, is the fusion of two tarsal segments in the leg, resulting in the loss of the intermediate joint (Tokunaga and Gerhart,1976). This phenotype is caused most likely by an initial failure in proliferation as well as a failure in initiating segment boundary formation (Villano and Katz,1995). A less-severe phenotype that has been investigated in more detail, however, is a planar cell polarity defect in the eye (Zeidler et al.,1999; Yang et al.,2002). There, approximately 0.3% of the ommatidia exhibit wrong orientation and chirality (Strutt et al.,2004). A more drastic planar cell polarity phenotype is observed in fj mosaic mutants when the endogenous gradient of fj expression is disturbed (Zeidler et al.,1999; Casal et al.,2002). This disruption is seen most obviously in the orientation of the hairs of the abdomen and the wing, where hairs always point to the direction of highest fj expression (Zeidler et al.,2000; Casal et al.,2002). Despite this detailed phenotypic information on mutants, little is known about the molecular function of fj. The Drosophila protein appears to be a type II transmembrane protein that is at least partially secreted (Villano and Katz,1995; Brodsky and Steller,1996). But with the exception of the transmembrane region, the protein completely lacks known protein domains.

In leg and eye development, Notch signaling also plays an important role (Irvine,1999; Rauskolb and Irvine,1999; Strutt and Strutt,1999). Drosophila mutants of the Notch gene show similarities in phenotype with the four-jointed mutants; thus, it was not surprising when fj was identified as a downstream target of Notch (Buckles et al.,2001). Notch encodes a transmembrane receptor that can be activated by its ligands delta (dl) and serrate (ser). Upon activation, the intracellular part of the receptor is cleaved off and translocates to the nucleus, where it binds to the Suppressor of Hairless (Su(H)) protein, thereby converting it from a transcriptional repressor to an activator (Bray and Furriols,2001). In mammals, there are four Notch genes (Notch1–4) and five ligands, namely three delta homologs (Dll1, Dll3, and Dll4) and two serrate homologs (Jagged1 and Jagged2; Lai,2004). The nuclear mediator of Notch signaling in mammals is called RBP-Jκ or CBF-1.

In this work, we show that fjx1 is not only conserved in vertebrates, but we could further identify fjx1 as a downstream target of the mammalian Notch proteins. As described earlier, the murine fjx1 gene is expressed in distinct regions of the brain (Ashery-Padan et al.,1999). We now show that fjx1, furthermore, is expressed predominantly in epithelial structures of different organs. Using an fjx1-AP fusion protein, we visualized binding sites of fjx1, a secreted glycoprotein, to demonstrate its potential for signaling. The localization of this fjx1 binding partner correlates with fjx1 expression in many tissues during embryogenesis and in the adult brain.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Fjx1 Is Conserved in Several Species

The fjx1 gene was first identified in human and mouse as a gene homologous to fj in Drosophila (Ashery-Padan et al.,1999). To deduce conserved and, hence, presumably structurally important parts of the protein, we cloned additional fj-related sequences from Xiphophorus and zebrafish (Fig. 1A). Furthermore, complete fjx1 homologs have since been identified by database search and assembly of trace sequences in zebrafish, medaka, Fugu, Tetraodon, Xenopus, rat, and chimpanzee. A partial fjx1 sequence was found in chicken expressed sequence tag sequences. All fjx1 genes share a single-exon structure encoding polypeptides of 425–450 amino acids. There are in-frame stop codons upstream of the putative start codon in several cases with a promoter predicted by various software algorithms. The amino acid identity between fjx1 proteins of more distant species is approximately 40% (mouse–zebrafish 37%, mouse–Xenopus 44%). The identity between human and mouse or between fish species is much higher (88% and approximately 70%, respectively) reflecting their closer relationship (Fig. 1B). The Drosophila fj gene encodes a type II transmembrane glycoprotein. It is only partially cleaved to release the extracellular portion from cells (Villano and Katz,1995). All vertebrate fjx1 proteins have hydrophobic stretches very close to the amino terminus, which rather suggests a function as signal peptides. Signal peptidase cleavage would shorten the proteins by 22–27 amino acids, depending on species (Fig. 1A). Only the extracellular part of the Drosophila fj protein shows strong similarity to the vertebrate sequences, clustered in two highly conserved domains of currently unknown function. Structure prediction programs suggest that the protein consists essentially of α-helices, but no previously known sequence motifs were identified.

thumbnail image

Figure 1. The four-jointed proteins have been conserved during evolution. A: The fjx1 protein is highly conserved in different species, but only the Drosophila fj is a transmembrane type II glycoprotein. The fjx1 protein in vertebrates is classified as a secreted protein with a signal peptide by most prediction programs. The signal peptide and transmembrane domain, respectively, are underlined. The conserved N-glycosylation sites are framed and marked by an arrow. B: The fjx1/fj gene is found in many vertebrates and Drosophila as a single copy gene, and even in different fish species, there is no evidence for a gene duplication. The corresponding accession numbers or Ensemble Transcript ID are as follows: NM010218.2 (mouse), XM345408.1 (rat), AJ245599.1 (human), Genscan00000067788 (chimpanzee), scaffold462 (http://genome.jgi-psf.org/xenopus0/xenopus0.home.html) (Xenopus), Genscan00000008811 (Fugu), GSCT00010935001 (Tetraodon), AJ849641 (Xiphophorus), scaffold1334_contig12 (http://dolphin.lab.nig.ac.jp/medaka/index.php) (medaka), Genscan00000014345 (zebrafish), and U44904.1 (Drosophila).

Download figure to PowerPoint

Fjx1 Encodes a Secreted and Glycosylated Protein

To characterize the predicted fjx1 protein, we generated a polyclonal fjx1 antiserum. Recombinant fjx1 protein was generated in the baculovirus/Sf9 system by expressing the secreted part of mouse fjx1 fused to an N-terminal His6 tag. The protein proved to be highly insoluble and aggregated during renaturation attempts. Purified denatured protein, therefore, was used to generate polyclonal rabbit antisera, which were affinity purified.

A stably transfected HEK293 cell line was then created that expresses the complete coding region of fjx1 under the control of a tetracycline-regulated promoter. Western blot analysis of this 293-fjx1 cell line showed an accumulation of fjx1 protein in the supernatant, whereas the cellular fraction appeared to be almost devoid of fjx1 protein. This finding suggests that the protein is completely secreted, yielding two bands differing by approximately 4 kDa (Fig. 2A). There are two potential N-glycosylation sites in the murine fjx1 sequence. After treatment with endoglycosidase H to remove N-linked carbohydrates, we again found two protein forms, which were of reduced size, however (Fig. 2B). This finding suggests that fjx1 is not only glycosylated but may undergo partial proteolytic cleavage. To address the latter question, we expressed amino- and carboxy-terminal fusion proteins with alkaline phosphatase (AP) by transient transfection in HEK293 cells. Immunoprecipitation of metabolically labeled proteins from cell culture supernatants with anti-AP antibodies revealed partial processing only of the N-terminal AP fusion protein. The smaller fraction of processed proteins may be due to steric differences in the fusion molecule. The corresponding size alteration in the C-terminal fusion protein would be too small to be resolved under these conditions (Fig. 2C). The proteolytic cleavage of the AP-fjx1 fusion protein was also detectable by Western blot analysis of precipitated cell culture supernatant with the fjx1 antibody (data not shown). These data are consistent with the notion that there is partial proteolytic cleavage at the N-terminus.

thumbnail image

Figure 2. Analysis of the fjx1 protein. A: Fjx1 produced in stably transfected HEK293 cells (293-fjx1) is completely secreted into the culture medium. The protein size detected by Western blot analysis (47.5 kDa) with the fjx1 antibody is slightly smaller than predicted (50.4 kDa), resulting from the cleavage of the signal peptide. There is also a second protein isoform of 43.5 kDa, while no bands are seen in untransfected control cells (HEK293). B: Digestion of fjx1 with endoglycosidase H (EndoHf) reduces protein size, confirming the predicted N-glycosylation. C: Immunoprecipitation of 35S-methionine labeled secreted proteins with alkaline phosphatase (AP) antibodies reveals a partial cleavage at the N-terminus of fjx1 (cleavage site marked with a triangle in the schematic diagram). The size of the cleavage product of AP-fjx1 (arrow in the blot) is approximately 7 kDa (including an approximately 3-kDa linker) larger than secreted AP (SEAP) alone, suggesting a cleavage very closely to the N-terminus. In the AP-fjx1 fusion protein, only the extracellular part of fjx1 starting with amino acid 21 is fused to SEAP, whereas the fjx1-AP fusion protein contains the full-length fjx1, including the signal peptide (cleavage site marked by arrow).

Download figure to PowerPoint

Fjx1 Expression Is Notch Dependent

For Drosophila fj, it is known that its expression is regulated by means of several pathways, among them the Notch pathway (Papayannopoulos et al.,1998; Zeidler et al.,2000). Using an fjx1-luciferase reporter, we tested the effect of different mammalian Notch isoforms on fjx1 promoter activity. The fjx1-luciferase construct containing the luciferase gene under the control of a 2-kb fjx1 promoter fragment was cotransfected with expression vectors for different constitutively active Notch intracellular domains. In COS7 cells, all Notch isoforms led to an increase of luciferase activity (Fig. 3A). The strongest effects were observed with Notch1 and Notch2, which increased luciferase activity by approximately 20-fold (Notch1, 17.1; Notch2, 22.3). As the promoter region of fjx1 contains several potential RBP-Jκ binding sites to mediate Notch signals, we generated a series of fjx1-luciferase deletion constructs (Fig. 3B). When the site at position −260 to −270 (gtgggag) was deleted, a strong reduction in transactivation potential was evident (compare constructs fjx1 Δ2 and fjx1 Δ3). Nevertheless, deletion of this site in the context of the longest promoter construct (fjx1 ΔRBPJ) almost fully restored Notch inducibility, suggesting that more distant target sites may compensate for its loss. Interestingly, this binding site is completely conserved in available mouse, rat, opossum, chimp, and human sequences, arguing for a biological relevance of this site. Notch1 knockout mice hybridized with an fjx1 probe exhibit no significant change in fjx1 expression. Only in Dll1 knockout mice did we sometimes find a diminished fjx1 expression in the presomitic mesoderm, partly depending on the grade of malformation of the embryo (data not shown). The early lethality in both knockout genotypes precluded analysis of later embryonic stages, where higher fjx1 expression levels would provide better chances to observe in vivo regulation. Therefore, clear-cut in vivo proof of Notch-dependent fjx1 expression currently is lacking.

thumbnail image

Figure 3. Analysis of the murine fjx1 promoter region. A: Cotransfection with vectors expressing constitutively active Notch1, Notch2, or Notch3 intracellular domains increase the luciferase activity of an fjx1-luc reporter construct. In comparison, Notch4 has only a minimal effect on the expression of this reporter construct containing the fjx1 promoter region (−1992 to −11 relative to start codon of fjx1, accession number AM000025). B: By consecutively shortening the promoter region, the RBP-Jκ binding site that mediates most of the Notch signal was identified. All putative binding sites are marked by vertical lines. The loss of a single additional RBP-Jκ binding site in construct fjx1Δ3 drastically reduces the effect of Notch cotransfection on the expression of the reporter construct. The loss of this binding site (marked by X) in the context of the full promoter region (fjx1ΔRBPJ) could be compensated by the other binding sites. The effect of the different Notch isoforms on the original fjx1-luc construct is set to one.

Download figure to PowerPoint

Expression of fjx1 During Mouse Development

To gain insight into the role of fjx1 during mouse development, we analyzed its expression pattern by in situ hybridization of whole-mount embryos and paraffin sections particularly with regard to expression areas aside from the already described expression in the developing brain (Ashery-Padan et al.,1999). At embryonic day (E) 12.5, expression is detected in restricted regions of the telencephalon, the ventricles, the diencephalon, and the medulla oblongata; at E14.5, strong staining is visible in the olfactory bulb (Fig. 4A–D,L). More caudally, fjx1 is detected in the entire neural tube with preferential expression in the ependymal layer and the dorsal root ganglia (Fig. 4A–C,J). The expression in the neural tube persists in an identical pattern during later stages of development. Only at E12.5 is there an elevated expression of fjx1 in the floor and roof plate (data not shown). Fjx1 is also expressed in various cranial ganglia, e.g., the trigeminal, the vestibulocochlear, and glossopharyngeal ganglion, as well as in the cochlear ganglion and the olfactory epi- thelium (Fig. 4M–O).

thumbnail image

Figure 4. A–D: RNA in situ hybridization of fjx1 during embryogenesis, at embryonic day (E) 11.5 (D), E12.5 (A), and E14.5 (B,C). E–I: Fjx1 expression can be detected during somitogenesis in the PSM and in the somites at E11.5 (E,F), and it persists in derived structures, e.g., intervertebral discs at E12.5 (G) and E14.5 (H) and sequential stripes (asterisk) in the closing thoracic cage at E12.5 (I). L: Additionally, fjx1 is found in tendon-like structures, e.g., in the tongue and limbs (see also Fig. 5J,K) at E14.5. J,K,M: Strong expression is also found at E14.5 in the neural tube (J), multiple ganglia (J,M) and in the ganglion layer of the eye (K). N,O: Furthermore, fjx1 is expressed in the olfactory epithelium (N) and cochlear ganglion of the inner ear (O). cg, cochlear ganglion; co, cochlea; drg, dorsal root ganglia; epl, ependymal layer; ey, eye; fb, forebrain; gcl, ganglion cell layer; gg, glossopharyngeal ganglion; h, heart; hb, hindbrain; ie, inner ear; ivd, intervertebral disc anlagen; lb, limb buds; li, liver; nt, neural tube; oe, olfactory epithelium; ofb, olfactory bulb, po, pons; PSM, presomitic mesoderm; S0, somite 0; sg, salivary gland; so, somite; t, tooth; tg, trigeminal ganglion; to, tongue; ttr, tubo-tympanic recess; vg, vestibulocochlear ganglion.

Download figure to PowerPoint

Another interesting fjx1 expression domain is observed in the tail region at E11.5. Here, fjx1 is expressed constantly throughout the presomitic mesoderm, but compressed to a small caudal stripe in the newly formed somites. Thereafter, fjx1 expression is reduced and limited to the sclerotome of the somites (Fig. 4E,F). This finding may be related to a subsequent expression in the trunk: a characteristic expression in sequential stripes found in the closing body wall at E12.5 and E13.5 of the developing thoracic cage (Fig. 4I). An additional somite-derived expression area is the strong staining of the developing intervertebral discs (Fig. 4G,H).

Specific expression of fjx1 is also detected in many structures that develop through epithelial–mesenchymal interactions, e.g., in the developing tooth buds and the whisker follicles starting at E13.5 (Fig. 5A,I). At E12.5 and E14.5, expression of fjx1 was found in all epithelia of the pancreas, the thymus, the submandibular glands, the gut endoderm, and the comma- and S-shaped bodies of the developing kidney (Fig. 5B–H). We could not detect fjx1 expression in liver or heart.

thumbnail image

Figure 5. Most organs that develop through mesenchymal epithelial transitions and inductions also express fjx1 in the epithelial compartment. B–H: Examples are pancreas (C), gut (D), salivary gland (B), and the kidney (E,F) at embryonic day (E) 14.5 and E17.5 (G,H). A,I: Epithelial expression of fjx1 is also detectable in the tooth bud at E14.5 (A) and the whisker follicles at E16.5 (I). J,K: Additionally, fjx1 is expressed in the developing joints at E13.5 (J) and E14.5 (K). cb, comma-shaped body; cu, hair cuticle; d, digits; ge, gut epithelium; j, joints; k, kidney; mt, molar tooth; p, pancreas; sg, salivary gland; tp, tendon precursor.

Download figure to PowerPoint

In limb development, fjx1 expression can be detected from E9.5 onward. At E10.5 and E11.5, expression is restricted to the AER (Fig. 4D). From E12.5 on, fjx1 is expressed in regions flanking the developing joints of elbow and digits and later gets restricted to joints and tissue flanking bones that may later develop into tendons (Fig. 5J,K). Such a presumed tendinous expression of fjx1 would also be in line with the distinct staining in the tongue, in the region of the future tendinous aponeurosis linguae (Fig. 4L).

Expression Pattern of fjx1 Binding Sites

Based on partial fj secretion and non–cell-autonomous behavior in Drosophi- la mosaic fj mutants, the protein is expected to function as an intercellular signaling ligand (Zeidler et al.,1999; Casal et al.,2002). This statement is further strengthened by the complete secretion that we find for the vertebrate fjx1 protein. We, therefore, used an fjx1-AP fusion protein to identify fjx1 binding sites in whole-mount embryos (E10.5) and on cryosections. The protein was expressed in 293T cell to allow for glycosylation and specific proteolytic processing.

A strong and highly specific staining pattern could be generated using the fjx1-AP fusion protein on embryos of different ages, while control stainings with unfused alkaline phosphatase alone showed only marginal and nonspecific signals (Fig. 6C′,D′,J′,K′). Throughout embryogenesis the staining pattern for fjx1 obtained by RNA in situ hybridization is similar to that of its presumed interaction partner, visualized by the AP fusion protein. At E10.5, fjx1 binding sites are detected in the telencephalon and the tectum and those structures remain unstained in embryos treated with secreted AP (SEAP) alone (not shown). Later on, fjx1 binding sites are additionally found in the medulla, the pons, and the spinal cord, in the ependymal layer, and the developing ventral horn (Fig. 6B). Throughout the body peripheral nerves and, to a somewhat lesser extent, ganglia, e.g., the dorsal root ganglia, the cervical plexus, and the trigeminal ganglion, are stained with the fjx1-AP fusion protein (Fig. 6B–E).

thumbnail image

Figure 6. Localization of the fjx1 binding sites with an fjx1-alkaline phosphatase (AP) fusion protein. Negative controls with unfused secreted AP (SEAP) are shown for comparison (indicated by quotation marks). A–E: During embryogenesis, fjx1 binding sites are located in the brain at embryonic day (E) 10.5 and the limb buds (A), in dorsal root ganglia and nerve fibers E14.5 (B), the cervical plexus (C), the trigeminal ganglion and nerve (D), and the inner ear epithelium (E). Additionally, fix1 binding sites are found in mesenchymal tissue. F–I: At E14.5, staining with the fjx1-AP fusion protein is detected in the lung (F, enlarged in G) and the kidney (H, enlarged in I). J,K: During limb development, fjx1 binding sites are found in tissue flanking the developing bone at E14.5 (J) and E18.5 (K). L–O: In the adult brain, the fjx1-AP fusion protein prominently stains areas adjacent to fjx1 mRNA expression (L,N) in the hippocampus (M) and the cerebellum (O) in adult mice. ad, adrenal gland; br, bronchus; cep, cervical plexus; co, cochlea; d, digits; drg, dorsal root ganglia; epl, ependymal layer; fb, forebrain; gcl, granular cell layer; gl, granular layer; k, kidney; lb, limb buds; lma, lateral malleolus; lu, lung; mb, midbrain; ml, molecular layer; mma, medial malleolus; mtb, metatarsal bone; nt, neural tube; Pc, Purkinje cells; phb, phalangeal bone; pl, pyramidal layer; sb, S-shaped body; slm, stratum lacunosum-moleculare; sor, stratum oriens; sr, stratum radiatum; tg, trigeminal ganglion; tn, trigeminal nerve; u, ureter.

Download figure to PowerPoint

Similar to the fjx1 expression found by RNA in situ hybridization, fjx1-AP fusion protein staining is found in epi- thelial cells of different organs. Examples are fjx1 binding sites in the epithelium of the inner ear, the lung, and the kidney at E14.5 (Fig. 6E,F–I). In the inner ear, fjx1 binding sites are located at the inner epithelium of the cochlea and the nerve fibers, whereas fjx1 expression is found in the ganglia. In the lung, fjx1 binding sites and fjx1 expression are partially overlapping, although pulmonary fjx1 expression is very weak and more diffuse compared with the fjx1 binding sites, which clearly surround the bronchial tubes (Fig. 6G). Fjx1 binding sites in the kidney are located in the collecting ducts, the comma- and S-shaped bodies, and later on in the tubuli (data not shown). In contrast, fjx1 expression appears to be restricted to the comma- and S-shaped bodies.

As expected from the fjx1 expression, a distinct staining of the limb buds is visible at E10.5 (Fig. 6A). Likewise, at E14.5, we observed a rather clear staining of the developing bones and the adjacent tissue. In older stages, endogenous AP activity of the bones could only partially be inactivated by heat treatment and may overlay the staining by the AP fusion protein to some extent (Fig. 6J,K). In contrast to the fjx1 in situ hybridization, the joints are not stained with the AP fusion protein, but in the bone flanking tissue, where fjx1 is expressed too, fjx1 binding sites were detected.

In adult organs, strong staining with the AP fusion protein is detected especially in the brain. As described previously, fjx1 is expressed in the Purkinje cell layer of the cerebellum, in the olfactory bulb, and in the striatum pyramidale and the granular layer of the hippocampus (Ashery-Padan et al.,1999). Adjacent to these areas of fjx1 expression, fjx1 binding sites are detected in a rather complementary pattern, e.g., in the cortex of the cerebellum, and in the striatum lacunosum, radiatum, and oriens of the hippocampus, respectively (Fig. 6L–O).

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

The human and mouse fjx1 genes have been described as the vertebrate homologs of the Drosophila four-jointed gene. We have now identified counterparts in a larger number of vertebrates either by cloning or database searches. Notably, we could not find closely related sequences in complete or partial sequence assemblies of several lower nonvertebrate species such as Caenorhabditis elegans and Ciona. This finding suggests that four-jointed is conserved during evolution but restricted to higher vertebrates and only a limited range of invertebrates, such as Drosophila and Anopheles. The Drosophila fj gene encodes a type II transmembrane glycoprotein, whereas the vertebrate fjx1 proteins lack the extended intracellular domain, and they may be secreted directly by means of a N-terminal signal peptide. Consequently, it is only the extracellular part of the fj/fjx1 proteins that exhibits strong sequence similarities. Multiple alignment of all available protein sequences identified two blocks of highly conserved amino acid residues that are separated by a nonconserved spacer in Drosophila. Unfortunately, the conserved sequence blocks do not show similarity to known protein modules and, thus, do not provide any hints as to the function of any of these genes or proteins.

Despite the conservation of the carboxy terminal part of the fj and fjx1 proteins, it remains unclear if these have identical sites of action within cells or tissues. The fj protein has been shown to undergo only partial proteolytic cleavage and secretion of the extracellular domain (Villano and Katz,1995; Buckles et al.,2001). Thus, either the secreted or transmembrane form—or both—may be functionally relevant. Recently, Strutt et al. reported that, in Drosophila, cleavage and secretion of fj are not required for its function. Furthermore, they provided evidence that fj is most active in the Golgi apparatus and speculated that it may act in a fringe-like manner to modify other secreted or transmembrane proteins, even if part of the fj protein is secreted (Strutt et al.,2004). Contrary to that, we found that fjx1 is efficiently secreted, when expressed in a stable manner in human HEK293 cells. Only transient transfection of CMV promoter-driven fjx1 expression vectors into 293T cells lead to an accumulation of the protein in the cells and only little secretion (data not shown). This finding may, however, represent an artifact of cellular overloading due to massive overexpression. On the other hand, fringe is efficiently secreted as well, even if it is most active in the Golgi apparatus (Johnston et al.,1997). But in contrast to fringe, fjx1 does not contain any protein domains that would point to an enzymatic activity in the Golgi complex. As long as the function of fjx1 is unclear, it will remain open as to whether its activity in mammals depends on the amount of secreted protein or on perhaps smaller amounts of Golgi-associated protein. Alternatively, fjx1 may even have two functions as known from various secreted glycosyltransferases. One prominent example is the β1,6,-N-acetylglucosaminyl transferase V, which acts in the Golgi apparatus as a glycosyltransferase, and as secreted protein, it stimulates the fibroblast growth factor receptor signaling pathway (Saito et al.,2002).

The Drosophila fj gene appears to be regulated by various pathways because Notch, Unpaired (JAK/STAT), and Wingless signals can modulate its expression (Zeidler et al.,1999). By using a series of luciferase reporter constructs with fjx1 promoter sequences of different lengths, we were able to show that the murine fjx1 promoter is a direct target for activation by the Notch intracellular domain. A proximal, conserved RBP-Jκ binding site could be identified that is capable of mediating most of this Notch response. There are additional upstream binding sites, however, that seem to act redundantly in this in vitro assay and compensate for the loss of the more proximal site. Whereas the activation of luciferase expression by Notch1–3 was relatively high, ranging from 9- to 20-fold, the effect of Notch4 was barely detectable (1.4-fold transactivation). This observation partly may be due to unknown differences in expression levels of the Notch fragments, but it is also in agreement with the expression patterns of Notch and fjx1 in vivo: in the pancreas and the kidney, fjx1 is primarily coexpressed with Notch1 and Notch 2, mostly in epithelial cells, but we never find fjx1 expression in endothelial cells, where Notch4 is expressed (Lammert et al.,2000; Leimeister et al.,2003). In the adult brain, the overlap of Notch expression (Notch1–3) with fjx1 is most evident, e.g., in the cerebellum and the hippocampus (Irvin et al.,2001). We could not observe any differences in the expression of fjx1 in Notch1 knockout embryos, but this finding may be due to the redundant expression of the Notch receptors. In Dll1 knockout mice, fjx1 expression was diminished in the presomitic mesoderm, with some variation depending on the degree of malformation of the embryo. Most of the expression domains were unaltered, however, suggesting that the Notch pathway certainly is probably not the major factor in controlling fjx1 expression.

The secretion of fjx1 proteins and the partly non–cell-autonomous action of fj in Drosophila would predict that fjx1 represents a ligand for an as yet unknown receptor on target cells. By using an fjx1-AP fusion protein, we were able to detect distinct binding sites for fjx1 in several different organs throughout development as well as in adult mice. The localization of these fjx1 binding sites corresponds well to the known expression pattern of fjx1. Embryonic organs such as kidney, the inner ear, or the lung show staining with both the fjx1 in situ probe and the fjx1-AP protein. The staining is partially overlapping, e.g., in the epithelia of the lung and the kidney, but sometimes also distinct as in the inner ear and the limb buds, where fjx1 is expressed adjacent to the fjx1 binding sites (ganglia vs. peripheral nerves and epithelium; joints and tendons vs. developing bones). In the adult brain, fjx1 and its binding sites exhibit a striking complementarity in their expression: In the cerebellum and the hippocampus, both are detected in adjacent cell layers, one in the neuron dense zone and the other preferentially in fiber tracts. These data suggest that fjx1 may act as a short range signal on neighboring cells or directly adjacent tissues.

The nature of the fjx1 interacting partner remains unknown, because it was not possible to identify the presumed receptor either by expression cloning or by coimmunoprecipitation from mouse brain. In Drosophila, fj is supposed to act in a common pathway with two very large cadherins, namely dachsous (ds) and fat (ft) (Yang et al.,2002; Ma et al.,2003), but nothing is known about a direct interaction between fj and ds and/or ft. In the two hybrid interaction screen of the Drosophila proteosome, no interacting partner of fj was identified, while ds and ft are connected by means of an as yet uncharacterized protein CG13139 (Giot et al.,2003). Although two homologs of dachsous and four homologs of fat are known in mammals (Ponassi et al.,1999; Nakajima et al.,2001; Mitsui et al.,2002; Nakayama et al.,2002; Hong et al.,2004), there is nothing known about an interaction of these genes. Indeed only fat1 has been characterized in more detail, and it plays a role in kidney development (Ciani et al.,2003). In the rat, fat1 protein is localized in epithelial cells, for example, in the distal and collecting tubules (Inoue et al.,2001), where we find fjx1 binding sites as well. Likewise, rat fat2 (MEGF1) protein is colocalized with the fjx1 binding sites in the molecular layer of the cerebellum (Nakayama et al.,2002). Unfortunately, nothing is known about the protein localization of the dachsous homologs, whose expression was only determined by Northern blot experiments. The very large size of these cadherins in mammals and their potential redundancy will make interaction studies rather challenging. Nevertheless, it would be very interesting to clarify whether this signaling interaction between fj, ds, and ft is conserved in mammals and, if so, to identify the components involved in this scheme. This discovery would shed more light on the still elusive functions of vertebrate fjx1 proteins.

EXPERIMENTAL PROCEDURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Cloning of Xiphophorus fjx1 and Zebrafish fjx1

Part of Xiphophorus fjx1 was amplified from genomic DNA with degenerated primers (sequences available upon request). After labeling by random priming the polymerase chain reaction product (195 bp) was used as hybridization probe. Hybridization of a Xiphophorus maculatus cosmid library provided by the RZPD (Burgtorf et al.,1998) according to standard protocols identified two fjx1 containing clones: MPMGc120 J24 076 and MPMGc120 D09 076. Appropriate restriction fragments were subcloned and sequenced to verify the Xiphophorus fjx1 sequence. Part of zebrafish fjx1 was originally cloned with degenerated primers, but in the meantime, the full sequence of zebrafish fjx1 is available in Ensemble.

Cell Culture

HEK293, 293-fjx1, and COS7 cells were maintained in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal calf serum (FCS) and a mixture of penicillin/streptomycin (100 U/ml each). The cell lines were cultured at 37°C under 5% CO2. Sf9 cells were maintained in TC-100 medium containing 10% FCS and a mixture of penicillin/streptomycin (50 U/ml each). The cells were cultured at 28°C without CO2. The fjx1 expressing HEK293 cell line was generated by transfection with the tetracycline-regulated pBPSTR-1 vector (Paulus et al.,1996) containing the entire fjx1 coding region (2-kb NotI–BamHI fragment) followed by selection with puromycin.

Protein Purification, Antibody

The extracellular part of fjx1 without the hydrophobic N-terminus was cloned into the pFastbac vector behind a 6xHis-tag. Sf9 cells were transfected with this pFastBac-His6fjx1 construct using the DAC-30 Transfection kit (Gibco BRL) according to the manufacturer's instructions. The supernatant of this transfection was used for the subsequent infections of Sf9 cells. For protein purification infected cells were harvested with lysis buffer (6 M guanidine hydrochloride, 100 mM sodium dihydrogen phosphate, 1% Tween, 10 mM Tris-HCl, pH 8.0) 96 hr after infection. Cell debris was spun down, and the supernatant was loaded on a Ni-NTA column. The column was washed with 10 vol of wash buffer (8 M urea, 100 mM sodium dihydrogen phosphate, 1% Tween, 10 mM Tris-HCl, pH 8.0) and then the protein was eluted with elution buffer (8 M urea, 100 mM sodium dihydrogen phosphate, 1% Tween, 10 mM Tris-HCl, pH 5.9). The purified protein was used for immunization of rabbits. The antisera were affinity purified and used for Western blot analysis according to standard protocols.

Immunoprecipitation

293T cells were transfected with 1.5 μg DNA (fjx1-AP, AP-fjx1, or SEAP) in a six-well plate using the standard calcium phosphate procedure. Cells were washed three times with methionine-free medium 24 hr after transfection and then incubated for 6 hr in labeling solution (DMEM without methionine, 10% dialyzed FCS, and 200 μCi/ml 35S-methionine). The supernatants were collected and concentrated to approximately 200 μl in 30-kDa Centrifugal Filter Tubes (Eppendorf). The AP proteins were then precipitated with anti-AP (Sigma) coupled Sepharose beads for 30 min at room temperature. After several washing steps with TBSN (150 mM NaCl, 25 mM Tris-HCl, pH 8.0, 0.1% NP-40) and modified RIPA buffer (0.5% NP-40, 0.5% sodium desoxycholate, 0.1% sodium dodecyl sulfate, 0.1% sodium azide, 144 mM NaCl, 50 mM Tris-HCl, pH 8.0), the proteins were separated on a 10% polyacrylamide gel and visualized by fluorography.

Reporter Assays

Cells were plated in 24-well plates at a density of 20,000 cells/well 24 hr before transfection (Maier and Gessler,2000). Routinely 0.4μg/well of plasmid DNA was transfected using the Escort reagent (Sigma) according to the manufacturer's instructions. Each transfection included 0.05 μg of reporter plasmid, 0.25 μg of expression plasmid (or empty vector), and 0.1 μg of pCMV-βGal vector as an internal control. Cells were harvested 40–48 hr after transfection, and luciferase activity was measured in a luminometer (Lumat LB9501, Berthold, Wildbach) and normalized to the β-galactosidase activity, which was determined using ONPG as substrate.

RNA In Situ Hybridization

In situ hybridization of whole-mount embryos and paraffin sections was performed as described elsewhere (Leimeister et al.,1998). The fjx1 riboprobe was prepared by linearizing the 2-kb full-length fjx1 plasmid with NotI and transcribing with T3 RNA polymerase in the presence of digoxigenin-UTP (Roche).

Alkaline Phosphatase Staining

Unfixed whole-mount embryos (stage E10.5) were rinsed once with HBAH (Hank's balanced salt solution, 0.5 mg/ml bovine serum albumin, 0.1% sodium azide, 20 mM HEPES, pH 7.0). Cryosections (20 μm) of unfixed tissue were washed for 10 min in HBS (150 mM NaCl, 20 mM HEPES, pH 7.0) and rinsed twice with HBAH (Flanagan and Cheng,2000; Flanagan et al.,2000). Both, whole-mounts and cryosections, were then incubated for approximately 2 hr at room temperature with fjx1-AP fusion protein or SEAP (approximately 800 mU/ml each). Afterward, sections were washed six times with HBAH, embryos at least 10 times for 5 min each. Fixation with acetone/formalin (65% acetone, 8% formalin, 20 mM HEPES, pH 7.0) was then performed for 15 sec (sections) or 2 min (whole-mounts). Sections and embryos were subsequently washed with HBS twice or three times, respectively, for 5 min. Endogenous AP was inactivated at 65°C, 10 min for sections and 20 min for whole-mounts. Tissues were rinsed once with AP-staining buffer (100 mM NaCl, 5 mM MgCl2 and 100 mM Tris-HCl, pH9.5) and stained with BM Purple (Roche) containing 1 mM levamisole. When desired color intensity was reached, staining was stopped with PBS/10 mM ethylenediaminetetraacetic acid. Embryos were then fixed with 4% paraformaldehyde at 4°C overnight and stored in PBS. Sections were mounted with Kaiser's glycerol gelatin (Merck) prior to storing.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

We thank B. Kempkes, U. Lendahl, D. Hayward, and J. Kitajewski for the Notch expression vectors and J. Altschmied for the pLUC1 vector.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES