UNC5H1 Induces Apoptosis via Its Juxtamembrane Region through an Interaction with NRAGE
2003; Elsevier BV; Volume: 278; Issue: 19 Linguagem: Inglês
10.1074/jbc.m300415200
ISSN1083-351X
AutoresMegan E. Williams, Phyllis Strickland, Ken Watanabe, Lindsay Hinck,
Tópico(s)Zebrafish Biomedical Research Applications
ResumoThe UNC5Hs are axon guidance receptors that mediate netrin-1-dependent chemorepulsion, and dependence receptors that mediate netrin-1-independent apoptosis. Here, we report an interaction between UNC5H1 and NRAGE. Our experiments show that this interaction is responsible for apoptosis induced by UNC5H1, and this level of apoptosis is greater than the amount induced by either UNC5H2 or UNC5H3. We mapped the NRAGE binding domain of UNC5H1 to its ZU-5 domain and show that this region, in addition to an adjacent PEST sequence, is required for UNC5H1-mediated apoptosis. Chimeric UNC5H2 and UNC5H3 receptors, containing the NRAGE binding domain and PEST sequence of UNC5H1, bind NRAGE and cause increased levels of apoptosis. UNC5H1 expression does not induce apoptosis in differentiated PC12 cells, which down-regulate NRAGE, but induces apoptosis in native PC12 cells that endogenously express high levels of NRAGE and in differentiated PC12 cells when NRAGE is overexpressed. Together, these results demonstrate a mechanism for UNC5H1-mediated apoptosis that requires an interaction with the MAGE protein NRAGE. The UNC5Hs are axon guidance receptors that mediate netrin-1-dependent chemorepulsion, and dependence receptors that mediate netrin-1-independent apoptosis. Here, we report an interaction between UNC5H1 and NRAGE. Our experiments show that this interaction is responsible for apoptosis induced by UNC5H1, and this level of apoptosis is greater than the amount induced by either UNC5H2 or UNC5H3. We mapped the NRAGE binding domain of UNC5H1 to its ZU-5 domain and show that this region, in addition to an adjacent PEST sequence, is required for UNC5H1-mediated apoptosis. Chimeric UNC5H2 and UNC5H3 receptors, containing the NRAGE binding domain and PEST sequence of UNC5H1, bind NRAGE and cause increased levels of apoptosis. UNC5H1 expression does not induce apoptosis in differentiated PC12 cells, which down-regulate NRAGE, but induces apoptosis in native PC12 cells that endogenously express high levels of NRAGE and in differentiated PC12 cells when NRAGE is overexpressed. Together, these results demonstrate a mechanism for UNC5H1-mediated apoptosis that requires an interaction with the MAGE protein NRAGE. hemagglutinin nerve growth factor phosphate-buffered saline glutathione S-transferase analysis of variance green fluorescent protein Tdt-mediated dUTP nick-end labeling deleted in colorectal cancer paraformaldehyde Apoptosis plays a critical role in determining the size and shape of the vertebrate nervous system (1Nijhawan D. Honarpour N. Wang X. Annu. Rev. Neurosci. 2000; 23: 73-87Crossref PubMed Scopus (270) Google Scholar). The execution phase of the apoptotic program in neurons is well characterized and, as with most cell types, depends on the activation of intracellular proteases, primarily caspases (2Yuan J. Yankner B.A. Nature. 2000; 407: 802-809Crossref PubMed Scopus (1627) Google Scholar). In contrast, it is not well understood how cues from the environment regulate this process during development of the nervous system. UNC-5 was originally characterized in Caenorhabditis elegans as a gene required for axonal repulsion in netrin/UNC-6 responsive neurons (3Hedgecock E.M. Culotti J.G. Hall D.H. Neuron. 1990; 4: 61-85Abstract Full Text PDF PubMed Scopus (745) Google Scholar, 4Leung-Hagesteijn C. Spence A.M. Stern B.D. Zhou Y. Su M.W. Hedgecock E.M. Culotti J.G. Cell. 1992; 71: 289-299Abstract Full Text PDF PubMed Scopus (349) Google Scholar). The vertebrate members of this family (UNC5H1, 2, and 3) (5Ackerman S.L. Kozak L.P. Przyborski S.A. Rund L.A. Boyer B.B. Knowles B.B. Nature. 1997; 386: 838-842Crossref PubMed Scopus (323) Google Scholar, 6Leonardo E.D. Hinck L. Masu M. Keino-Masu K. Ackerman S.L. Tessier-Lavigne M. Nature. 1997; 386: 833-838Crossref PubMed Scopus (434) Google Scholar), together with C. elegans UNC-5 (4Leung-Hagesteijn C. Spence A.M. Stern B.D. Zhou Y. Su M.W. Hedgecock E.M. Culotti J.G. Cell. 1992; 71: 289-299Abstract Full Text PDF PubMed Scopus (349) Google Scholar) and Drosophila Unc5 (7Keleman K. Dickson B.J. Neuron. 2001; 32: 605-617Abstract Full Text Full Text PDF PubMed Scopus (251) Google Scholar), comprise a subgroup of the Ig superfamily of receptors. The UNC5s contain two Ig and two thrombospondin type-I repeats in the extracellular domain. In addition, their cytoplasmic domains contain regions of homology with other proteins: 1) a ZU-5 domain homologous to Zona Occludens-1, a protein implicated in tight-junction formation (8Schultz J. Milpetz F. Bork P. Ponting C.P. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 5857-5864Crossref PubMed Scopus (3055) Google Scholar); and 2) a C-terminal death domain, a domain first identified as the pro-apoptotic region of tumor necrosis factor receptor-1 (9Tartaglia L.A. Ayres T.M. Wong G.H. Goeddel D.V. Cell. 1993; 74: 845-853Abstract Full Text PDF PubMed Scopus (1173) Google Scholar, 10Hofmann K. Tschopp J. FEBS Lett. 1995; 371: 321-323Crossref PubMed Scopus (103) Google Scholar). In a netrin-1 gradient, a complex of UNC5H1 and DCC mediates repulsion (11Hong K. Hinck L. Nishiyama M. Poo M.M. Tessier-Lavigne M. Stein E. Cell. 1999; 97: 927-941Abstract Full Text Full Text PDF PubMed Scopus (584) Google Scholar), although there is evidence suggesting that short range repulsion by netrin-1 may be mediated by UNC5 alone (3Hedgecock E.M. Culotti J.G. Hall D.H. Neuron. 1990; 4: 61-85Abstract Full Text PDF PubMed Scopus (745) Google Scholar, 7Keleman K. Dickson B.J. Neuron. 2001; 32: 605-617Abstract Full Text Full Text PDF PubMed Scopus (251) Google Scholar). NRAGE (Dlxin-1, MAGE-D1) is a recently identified molecule belonging to the MAGE (melanoma antigen) protein family. There are currently over 25 MAGE proteins in humans, characterized by the presence of a MAGE homology domain. The expression of many MAGE proteins is restricted to cancer cells (12Barker P.A. Salehi A. J Neurosci. Res. 2002; 67: 705-712Crossref PubMed Scopus (269) Google Scholar); however, recent studies have revealed a role for two MAGE proteins in the nervous system. One MAGE family member, necdin, is thought to maintain the differentiated state of post-mitotic neurons by preventing entry into the cell cycle (13Taniura H. Taniguchi N. Hara M. Yoshikawa K. J. Biol. Chem. 1998; 273: 720-728Abstract Full Text Full Text PDF PubMed Scopus (149) Google Scholar, 14Hayashi Y. Matsuyama K. Takagi K. Sugiura H. Yoshikawa K. Biochem. Biophys. Res. Commun. 1995; 213: 317-324Crossref PubMed Scopus (79) Google Scholar). Another MAGE family member, NRAGE, is expressed in the nervous system during early development in proliferative neural populations (15Kendall S.E. Goldhawk D.E. Kubu C. Barker P.A. Verdi J.M. Mech. Dev. 2002; 117: 187-200Crossref PubMed Scopus (58) Google Scholar). Recent studies have reported two major functions for NRAGE, as a transcriptional regulator for the dlx/msx family of transcription factors (16Masuda Y. Sasaki A. Shibuya H. Ueno N. Ikeda K. Watanabe K. J. Biol. Chem. 2001; 276: 5331-5338Abstract Full Text Full Text PDF PubMed Scopus (97) Google Scholar, 17Sasaki A. Masuda Y. Iwai K. Ikeda K. Watanabe K. J. Biol. Chem. 2002; 277: 22541-22546Abstract Full Text Full Text PDF PubMed Scopus (88) Google Scholar) and as a regulator of apoptosis. The first study to implicate NRAGE in apoptosis found that NRAGE binds the nerve growth factor receptor p75NTR, blocks cell cycle progression, and promotes p75NTR-mediated apoptosis (18Salehi A.H. Roux P.P. Kubu C.J. Zeindler C. Bhakar A. Tannis L.L. Verdi J.M. Barker P.A. Neuron. 2000; 27: 279-288Abstract Full Text Full Text PDF PubMed Scopus (257) Google Scholar). Subsequently, it was found to utilize two mechanisms to induce apoptosis in cells. One involves NRAGE-dependent degradation of the survival protein XIAP (X-linked inhibitor of apoptosis) (19Jordan B.W. Dinev D. LeMellay V. Troppmair J. Gotz R. Wixler L. Sendtner M. Ludwig S. Rapp U.R. J. Biol. Chem. 2001; 276: 39985-39989Abstract Full Text Full Text PDF PubMed Scopus (79) Google Scholar) and the other involves NRAGE-dependent activation of the c-Jun N-terminal kinase signaling pathway and caspases (20Salehi A.H. Xanthoudakis S. Barker P.A. J. Biol. Chem. 2002; 277: 48043-48050Abstract Full Text Full Text PDF PubMed Scopus (118) Google Scholar). Recently a role has begun to emerge for netrin-1 receptors in apoptosis. In this paper, we report that UNC5H1 mediates more than twice the amount of cell death as UNC5H2 and UNC5H3, and we identify NRAGE as a specific binding partner for UNC5H1. UNC5H1 binds NRAGEin vitro and can be co-immunoprecipitated from cells that endogenously express both proteins. Both the apoptotic region and NRAGE binding domain of UNC5H1 map to its juxtamembrane region that contains a PEST sequence and ZU-5 domain. We find that chimeric UNC5H2 and UNC5H3 proteins containing the PEST and ZU-5 sequence of UNC5H1 bind NRAGE and induce increased levels of apoptosis. UNC5H1 and NRAGE co-localize at the cell membrane in heterologous cells and are co-expressed in several regions of the developing nervous system. Using PC12 cells, we show that UNC5H1 expression induces apoptosis in native, mitotically active cells that endogenously express high levels of NRAGE but not in differentiated cells that sharply down-regulate NRAGE. We also show that UNC5H1 induces apoptosis in these differentiated PC12 cells when NRAGE is over-expressed. Taken together our data identify a novel signaling mechanism for UNC5H1-mediated apoptosis that requires an interaction with NRAGE. Full-length rat unc5h1and unc5h2 were cloned into pSecTagB (Invitrogen), which contains a C-terminal myc tag as described previously (6Leonardo E.D. Hinck L. Masu M. Keino-Masu K. Ackerman S.L. Tessier-Lavigne M. Nature. 1997; 386: 833-838Crossref PubMed Scopus (434) Google Scholar, 11Hong K. Hinck L. Nishiyama M. Poo M.M. Tessier-Lavigne M. Stein E. Cell. 1999; 97: 927-941Abstract Full Text Full Text PDF PubMed Scopus (584) Google Scholar). All other unc5h mutant constructs were generated in pSecTagB by PCR cloning using the flanking restriction sitesHindIII to XbaI and placed in-frame with themyc tag included within the vector. H2/H1apo and H3/H1apo were constructed using PCR to delete the juxtamembrane region of UNC5H2 (amino acids Asp407-Cys625) and UNC5H3 (amino acids Glu410-Cys612), and this region was replaced with a NotI site. Then, the apoptotic region of UNC5H1 (amino acids Leu391-Cys579) was generated by PCR with flanking NotI sites and inserted into the new NotI site in unc5h2 andunc5h3. unc5h3 was also cloned in-frame with themyc tag of pSecTagB using the EcoRI andXbaI sites. Full-length mouse nrage was cloned into pcDNA3 (Invitrogen) with a 5′ HA1 tag as described previously (16Masuda Y. Sasaki A. Shibuya H. Ueno N. Ikeda K. Watanabe K. J. Biol. Chem. 2001; 276: 5331-5338Abstract Full Text Full Text PDF PubMed Scopus (97) Google Scholar). unc5h1myc and HA-nrage, including a consensus Kozak and start site, were cloned into the sindbis virus vector pSinRep5 (Invitrogen). All PCR products were verified by sequencing. COS cells were maintained in Dulbecco's modified Eagle's medium supplemented with 107 fetal bovine serum and transfected using FuGENE (Roche Molecular Biochemicals). PC12 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 107 horse serum, 57 fetal bovine serum, and 307 PC12 conditioned medium and grown on collagen-coated dishes. Cells were differentiated for 14 days by incubating in serum-free Dulbecco's modified Eagle's medium with 50 ng/ml purified NGF (Sigma). sindbis virus expressing UNC5H1 was packaged in baby hamster kidney cells as described by the manufacturer's protocol (Invitrogen), and undiluted virus was used to infect native and differentiated PC12 cells. Cells were fixed for immunostaining in 47 PFA + 0.17 Triton X-100 for 20 min, rinsed in PBS, and blocked in PBS + 37 heat-inactivated goat serum + 0.17 Triton X-100 for 30 min. Primary antibody in blocking buffer was added for 60 min at room temperature followed by three washes for 5 min each. Fluorescent secondary antibody was diluted and incubated on cells for 40 min and washed three times before coverslipping. To determine the percent of pyknotic nuclei using DAPI stain, cells were incubated in DAPI diluted in PBS to 1 ॖg/ml for 5 min before coverslipping. Cells were visualized under fluorescence, and a total of at least 300 transfected cells were scored for pyknotic nuclei and apoptotic morphology. Each transfection was repeated and scored at least three times in a blind manner by multiple researchers. The UNC5H1 antibody 6E9 was raised against the extracellular domain of UNC5H1. 2S. Faynboym, L. Hinck, E. D. Leonardo, and M. Tessier-Lavigne, unpublished information. The NRAGE rabbit polyclonal was used in Western blotting at 1:200 (16Masuda Y. Sasaki A. Shibuya H. Ueno N. Ikeda K. Watanabe K. J. Biol. Chem. 2001; 276: 5331-5338Abstract Full Text Full Text PDF PubMed Scopus (97) Google Scholar). Anti-myc 9E10 and anti-HA 12CA5 were used for both immunostaining and Western blotting at 1 ॖg/ml. The E18 mouse brain library was made according to manufacturer's instructions using the Stratagene cDNA synthesis kit and ligating into the EcoRI/XhoI sites in pACT2 (Clontech), which contains a Gal4 activating domain. H1ICD was cloned by PCR into theEcoRI/SalI sites of pBTM116 in-frame with the LexA DNA binding domain (11Hong K. Hinck L. Nishiyama M. Poo M.M. Tessier-Lavigne M. Stein E. Cell. 1999; 97: 927-941Abstract Full Text Full Text PDF PubMed Scopus (584) Google Scholar). DNA was transformed into the L40 yeast strain stably transformed with a LexA-driven HIS3 gene and LexA-driven LacZ gene (21Vojtek A.B. Hollenberg S.M. Cooper J.A. Cell. 1993; 74: 205-214Abstract Full Text PDF PubMed Scopus (1717) Google Scholar). Transformants were selected for the ability to grow on minus histidine plates supplemented with 30 mm3-amino1,2,4-triazole and subsequently screened for the ability to produce LacZ. Rip60NRAGE in pcDNA3 was translated and labeled with [35S]methionine using the in vitro translation system (Promega). In vitro translated rip60NRAGE was incubated in buffer (0.57 Nonidet P-40, 10 mm Tris, pH 8.0, 150 mm NaCl, 107 glycerol, 5 mm dithiothreitol, protease inhibitors) with 5 ॖg of GST-H1ICD or GST control bound to glutathione-agarose beads. The samples were rocked for 2 h at 4 °C, washed three times with binding buffer, and fractionated by SDS-PAGE for autoradiography. COS or PC12 cells were immunoprecipitated as described (11Hong K. Hinck L. Nishiyama M. Poo M.M. Tessier-Lavigne M. Stein E. Cell. 1999; 97: 927-941Abstract Full Text Full Text PDF PubMed Scopus (584) Google Scholar). Briefly, cells were incubated in lysis buffer (50 mm Tris, pH 7.5, 150 mm NaCl, 5 mm EDTA, 17 TritonX-100, 107 glycerol) containing aprotinin, leupeptin, and phenylmethylsulfonyl fluoride at 1 ॖg/ml each. The sample was rocked at 4 °C for 40 min and then pelleted in a microfuge at 14,000 rpm (full speed) for 20 min. The supernatant was then incubated with antibody prebound to protein A/G (Santa Cruz Biotechnology, Inc.) for 6 h at 4 °C. Western blots were visualized using ECL detection (Amersham Biosciences). A portion of the ratunc5h1 intracellular domain (1480–2080 bp) was subcloned from cDNA into pBlueScript II. Similarly, the N terminus (1–873 bp) and C terminus (2025–2325 bp) of nrage cDNA were cloned into pBlueScript II. These constructs were used to make digoxigenin-labeled riboprobes. Brain tissue was prepared by immersion in 47 PFA for 2 h at room temperature, infiltrated with 307 sucrose, embedded in tissue freezing medium (Triangle Biomedical Sciences), and frozen at −80 °C. 20-ॖm sections were brought to room temperature and post-fixed in 47 PFA. Next, sections were digested with 10 ॖg/ml proteinase K for 2 min, stopped in 0.27 glycine, and fixed in 47 PFA for 5 min. Tissue was acetylated in acetic anhydride for 10 min and permeabilized with PBS with 0.17 Triton X-100 for 30 min. Prehybridization solution (507 deionized formamide, 4× SSC, 1× Denhardt's, 1 mg/ml tRNA, 0.5 mg/ml herring sperm DNA) was added for 2 h at room temperature. Hybridization solution including either digoxigenin-labeled UNC5H1 (900 ng/ml) or NRAGE (500 ng/ml) was added to serial sections and incubated overnight at 65 °C. Washes in 2× SSC/507 formamide for 1 h and 0.2× SSC for 1 h were followed by standard anti-digoxigenin detection protocols (Roche Molecular Biochemicals) with detection in BM Purple. Statistical analyses including ANOVA with Tukey's post-test for multiple comparisons were performed using R version 1.6.1 and Microsoft Excel. NIH image 1.63 software was used to determine relative protein levels following Western analysis and enhanced chemiluminescence detection. Based on earlier observations that the UNC5Hs may be involved in apoptotic processes (11Hong K. Hinck L. Nishiyama M. Poo M.M. Tessier-Lavigne M. Stein E. Cell. 1999; 97: 927-941Abstract Full Text Full Text PDF PubMed Scopus (584) Google Scholar, 22Llambi F. Causeret F. Bloch-Gallego E. Mehlen P. EMBO J. 2001; 20: 2715-2722Crossref PubMed Scopus (283) Google Scholar), we directly assessed the ability of UNC5H1, UNC5H2, and UNC5H3 to induce apoptosis in heterologous cells. Apoptosis was assayed by immunohistochemistry to identify transfected cells and DAPI labeling to identify pyknotic nuclei, a standard indicator of apoptosis. The identification of pyknotic nuclei is not affected by differences in the efficiency of transient transfections, because cells are observed at single cell resolution and equal numbers of transfected cells are counted. GFP was transfected as a control to determine the basal rate of death due to culturing, transfecting, and overexpressing protein in COS cells. Fig. 1Ashows a representative image of UNC5H1 expressing cells that display condensed, fragmented nuclei and loss of adhesion, indicative of dying cells. This cell morphology is distinctly different compared with cells expressing UNC5H2 and UNC5H3 that maintain extensive cell processes and intact nuclei, indicative of healthy cells (Fig. 1A). Quantification of apoptotic nuclei reveals that UNC5H1 expression causes a dramatic increase in apoptotic cells (697) compared with the GFP control (127) (p = 0.001; Fig. 1B). UNC5H2 (247) and UNC5H3 (167) induce apoptosis over control cells (p = 0.01 for each; Fig. 1B) but at significantly lower levels than UNC5H1. Because UNC5H1 expression induces the most robust apoptosis in our assays, we focused on understanding the mechanism underlying the ability of UNC5H1 to induce apoptosis. To understand the signaling mechanism used by UNC5H1, a yeast two-hybrid screen was performed on a cDNA library generated from E18 murine brain tissue using the intracellular domain of UNC5H1 as bait. From 14 positive clones, we identified one, rip60NRAGE, as the C-terminal half of the pro-apoptotic protein NRAGE (Dlxin-1, MAGE-D1) (16Masuda Y. Sasaki A. Shibuya H. Ueno N. Ikeda K. Watanabe K. J. Biol. Chem. 2001; 276: 5331-5338Abstract Full Text Full Text PDF PubMed Scopus (97) Google Scholar, 18Salehi A.H. Roux P.P. Kubu C.J. Zeindler C. Bhakar A. Tannis L.L. Verdi J.M. Barker P.A. Neuron. 2000; 27: 279-288Abstract Full Text Full Text PDF PubMed Scopus (257) Google Scholar). The NRAGE amino acid sequence contains a repeat region consisting of 25 repeats of a WQXPXX consensus sequence followed by a MAGE homology domain. The rip60NRAGE clone begins at amino acid 411 and continues through the poly(A) tail of NRAGE and contains five WQXPXX repeats, the entire MAGE domain, and the C-terminal tail (Fig.2A). The interacting domains between UNC5H1 and NRAGE were mapped in yeast. The C-terminal tail of NRAGE, the last 100 amino acids, is required for binding the intracellular domain of UNC5H1 (H1ICD), but is not sufficient to interact on its own (Fig. 2A). Neither the WQXPXX repeat nor the MAGE domain alone were sufficient to interact with H1ICD; however, adding the C-terminal tail back onto the MAGE domain restores binding (Fig. 2A). Therefore, the C-terminal tail of NRAGE may directly bind UNC5H1, or the tail is needed for the MAGE domain to fold properly, allowing the MAGE domain to directly bind H1ICD. Similarly, we mapped the NRAGE binding site of UNC5H1. Deletions of H1ICD from the C terminus, including the death domain, interact with rip60NRAGE (Fig. 2A). Increasingly, C-terminal truncations of UNC5H1 interacted less well, possibly because they were subject to degradation. Therefore, we generated several N-terminal truncations of H1ICD and tested them for the ability to interact with rip60NRAGE. We found that removing the ZU-5 domain abolished binding with rip60NRAGE (Fig. 2A) and conclude that this region is necessary for the interaction between UNC5H1 and NRAGE. UNC5H1 possesses a unique ability to induce apoptosis at significantly higher levels than either UNC5H2 or UNC5H3 (Fig. 1B). Therefore, we used the yeast two-hybrid system to test whether the NRAGE interaction is specific to UNC5H1 or whether it interacts with other UNC5Hs. Fig. 2A shows that, in yeast, rip60NRAGE does not interact with either UNC5H2 or UNC5H3 and thus, our results suggest that UNC5H1 interacts specifically with NRAGE through its ZU-5 domain. To confirm that the UNC5H1-NRAGE interaction in yeast is the result of a direct protein-protein interaction, we performed an in vitro GST pull down assay. A purified GST-H1ICD fusion protein revealed a strong interaction with in vitro translated rip60NRAGE, whereas GST alone shows little nonspecific binding, indicating that the interaction between UNC5H1 and NRAGE is direct (Fig. 2B). Next, we tested whether UNC5H1 interacts with NRAGE in cells by co-immunoprecipitation. COS cells were transiently transfected with full-length unc5h1myc and full-length HA-nrage. Fig. 2C shows that immunoprecipitating with anti-Myc pulls down HA-NRAGE only when UNC5H1myc is present and not when cells are transfected with empty vector. Deleting the death domain of UNC5H1 (H1ΔDDmyc) does not affect the interaction with HA-NRAGE in cells, but deleting the entire intracellular domain of UNC5H1 (H1ΔICDmyc) abolishes the interaction (Fig. 2C). These results confirm our observations in yeast and identify the ZU-5 domain as the NRAGE binding domain. To explore the specificity of the interaction between NRAGE and UNC5H1 in cells, we also tested whether UNC5H2myc or UNC5H3myc co-immunoprecipitate with HA-NRAGE. Our results show that UNC5H2myc binds HA-NRAGE but at consistently lower levels than UNC5H1myc (Fig. 2C). Quantification of NRAGE from Western blots shows that UNC5H2myc co-immunoprecipitates ∼707 (±13) less NRAGE than UNC5H1myc, and no interaction between UNC5H3myc and HA-NRAGE is detected (Fig. 2C). Last, we examined the cellular localization of UNC5H1 and NRAGE in COS cells. Previous studies have shown that NRAGE is localized to multiple cellular compartments, including the cytosol, nucleus, and at the cell membrane (16Masuda Y. Sasaki A. Shibuya H. Ueno N. Ikeda K. Watanabe K. J. Biol. Chem. 2001; 276: 5331-5338Abstract Full Text Full Text PDF PubMed Scopus (97) Google Scholar, 18Salehi A.H. Roux P.P. Kubu C.J. Zeindler C. Bhakar A. Tannis L.L. Verdi J.M. Barker P.A. Neuron. 2000; 27: 279-288Abstract Full Text Full Text PDF PubMed Scopus (257) Google Scholar). Because UNC5H1 is a transmembrane protein, we examined whether NRAGE is localized to the membrane in the presence of UNC5H1. Co-expression of UNC5H1myc and HA-NRAGE in COS cells shows strong co-localization between the two proteins at the cell membrane (Fig. 2D). For the interaction between UNC5H1 and NRAGE to be physiologically relevant, the two proteins must be present in the same place at the same time. It is already known that both unc5h1and nrage are highly expressed in the nervous system (6Leonardo E.D. Hinck L. Masu M. Keino-Masu K. Ackerman S.L. Tessier-Lavigne M. Nature. 1997; 386: 833-838Crossref PubMed Scopus (434) Google Scholar, 15Kendall S.E. Goldhawk D.E. Kubu C. Barker P.A. Verdi J.M. Mech. Dev. 2002; 117: 187-200Crossref PubMed Scopus (58) Google Scholar,16Masuda Y. Sasaki A. Shibuya H. Ueno N. Ikeda K. Watanabe K. J. Biol. Chem. 2001; 276: 5331-5338Abstract Full Text Full Text PDF PubMed Scopus (97) Google Scholar, 18Salehi A.H. Roux P.P. Kubu C.J. Zeindler C. Bhakar A. Tannis L.L. Verdi J.M. Barker P.A. Neuron. 2000; 27: 279-288Abstract Full Text Full Text PDF PubMed Scopus (257) Google Scholar, 23Shiras A. Sengupta A. Shepal V. Mol. Cell. Biol. Res. Commun. 2001; 4: 313-319Crossref PubMed Scopus (9) Google Scholar, 24Barrett C. Guthrie S. Mech. Dev. 2001; 106: 163-166Crossref PubMed Scopus (17) Google Scholar). To determine specific sites of overlapping expression, we performed in situ hybridization analysis on serial sections from embryonic brain tissue. unc5h1 andnrage mRNA are found together in the E16 striatum and both are largely excluded from the neighboring subventricular zone (Fig. 3, A and C). The outer layers of the E16 cortex, including the cortical plate and the marginal zone, stain strongly for both unc5h1 andnrage transcripts (Fig. 3, E and F). We have noted several other sites of overlapping expression during development, including the E18 hippocampus (Fig. 3, G andH) and olfactory bulb and E11 motor neurons (data not shown). These data demonstrate that unc5h1 andnrage are co-localized in the nervous system, suggesting they may function together in vivo. The initial observation that UNC5H1 induces apoptosis was made in COS cells. Therefore, if NRAGE is required for UNC5H1-mediated apoptosis, COS cells must express NRAGE endogenously. To test this hypothesis, we performed a Western blot on COS cell lysate using a polyclonal antibody against NRAGE. Fig. 4A shows that NRAGE appears to be abundant in COS cell lysates, and thus we decided to perform a structure/function analysis on the ability of UNC5H1 to induce apoptosis in these cells. To examine whether the interaction between UNC5H1 and NRAGE is required for cell death, we constructed several mutants of UNC5H1 and assayed these constructs for the ability to induce apoptosis. First, we assayed H1Δsac, a truncation that removes much of the intracellular domain, and found that H1Δsac induces significantly less apoptosis than full-length UNC5H1 (p = 0.002; Fig. 4B). Although this mutant removes the NRAGE binding domain, it also removes the death domain, a region of UNC5H1 shown by others to be required for UNC5H1-mediated apoptosis (22Llambi F. Causeret F. Bloch-Gallego E. Mehlen P. EMBO J. 2001; 20: 2715-2722Crossref PubMed Scopus (283) Google Scholar). To address this, we asked whether a construct deleted in only the death domain (H1ΔDD) induces apoptosis. Our results show that H1ΔDD does not significantly impede the ability of UNC5H1 to induce death in COS cells (Fig. 4B). Next, we made an UNC5H1 construct deleted in the DCC binding domain (H1ΔDB), because the UNC5H1 intracellular domain interacts with the netrin-1 receptor DCC (11Hong K. Hinck L. Nishiyama M. Poo M.M. Tessier-Lavigne M. Stein E. Cell. 1999; 97: 927-941Abstract Full Text Full Text PDF PubMed Scopus (584) Google Scholar), which may also be pro-apoptotic (25Mehlen P. Rabizadeh S. Snipas S.J. Assa-Munt N. Salvesen G.S. Bredesen D.E. Nature. 1998; 395: 801-804Crossref PubMed Scopus (368) Google Scholar). One possibility is that cell death induced by UNC5H1 is the result of a signal through endogenous DCC. Our results show that deleting the DCC binding domain has no effect on the ability of UNC5H1 to induce apoptosis (Fig. 4B). Together, these results indicate that the ZU-5/NRAGE binding domain is required for UNC5H1-mediated apoptosis and rule out a role for both the death and DCC binding domains. We dramatically decreased apoptosis induced through UNC5H1 by deleting the NRAGE binding domain; however, we noticed that H1Δsac still causes apoptosis above the level in control cells expressing GFP (Fig.4B). Therefore, we removed the entire intracellular domain of UNC5H1 (H1ΔICD) to determine whether this completely abrogates the death signal. Fig. 4B shows that H1ΔICD eliminates UNC5H1 apoptosis to control levels. Sequence analysis of this 116-amino acid region between the end of the transmembrane domain and the start of the ZU-5 domain revealed the presence of a PEST sequence that is not conserved in either UNC5H2 or UNC5H3 (26Rechsteiner M. Rogers S.W. Trends Biochem. Sci. 1996; 21: 267-271Abstract Full Text PDF PubMed Scopus (1421) Google Scholar, 27Rogers S. Wells R. Rechsteiner M. Science. 1986; 234: 364-368Crossref PubMed Scopus (2096) Google Scholar). When we delete the PEST domain (H1ΔPEST) the UNC5H1 apoptotic signal is eliminated. These results suggest that most of the UNC5H1 death signal requires the presence of the NRAGE binding domain (ZU-5); however, some signaling requires only the unique PEST sequence of UNC5H1. Taken together, we identified the juxtamembrane region, consisting of the adjacent PEST and ZU-5 domains, as the primary signaling region in UNC5H1-mediated apoptosis and NRAGE binding. UNC5H1 expression induces the highest percent of apoptosis (697) in cells whereas UNC5H2 expression induces less than half (247), and UNC5H3 expression induces even less (167) (Fig. 1B). The same pattern is true for the UNC5Hs ability to interact with NRAGE. UNC5H1 directly binds NRAGE whereas UNC5H2 binds NRAGE relatively weakly, and UNC5H3 shows no binding (Fig. 2C). These results suggest that the ability of UNC5Hs to induce apoptosis depends on their interaction with NRAGE. Thus, we asked whether the apoptotic domain of UNC5H1, when present on the homologous receptors, enhances either the apoptotic signal or NRAGE binding. To accomplish this, we constructed two chimeric receptors (H2/H1apo and H3/H1apo) in which the juxtamembrane region including the ZU-5 domain of UNC5
Referência(s)