Necdin-related MAGE Proteins Differentially Interact with the E2F1 Transcription Factor and the p75 Neurotrophin Receptor*

Necdin is a growth suppressor expressed predominantly in postmitotic neurons and implicated in their terminal differentiation. Necdin shows a moderate homology to the MAGE family proteins, the functional roles of which are largely unknown. Human genes encoding necdin, MAGEL2 (necdin-like 1), and MAGE-G1 (necdin-like 2) are located in proximal chromosome 15q, a region associated with neurodevelopmental disorders such as Prader-Willi syndrome, Angelman syndrome, and autistic disorder. The necdin and MAGEL2 genes are subjected to genomic imprinting and suggested to be involved in the etiology of Prader-Willi syndrome. In this study, we compared biochemical and functional characteristics of murine orthologs of these necdin-related MAGE proteins. The colony formation and bromodeoxyuridine incorporation analyses revealed that necdin and MAGE-G1, but not MAGEL2, induced growth arrest. Necdin and MAGE-G1 interacted with the transcription factor E2F1 via its transactivation domain, repressed E2F1-dependent transcription, and antagonized E2F1-induced apoptosis of N1E-115 neuroblastoma cells. In addition, necdin and MAGE-G1 interacted with the p75 neurotrophin receptor via its distinct intracellular domains. In contrast, MAGEL2 failed to bind to these necdin interactors, suggesting that MAGEL2 has no necdin-like function in developing brain. Overexpression of p75 translocated necdin and MAGE-G1 in the proximity of the plasma membrane and reduced their association with E2F1 to facilitate E2F1-induced death of neuroblastoma cells. These results suggest that necdin and MAGE-G1 target both E2F1 and p75 to regulate cell viability during brain development.

Necdin is a 325-amino acid protein encoded in a cDNA clone isolated from a subtraction library of neurally differentiated mouse embryonal carcinoma cells (1). The mouse necdin gene is expressed predominantly in postmitotic cells such as neuron and skeletal muscle (2)(3)(4). Ectopic expression of necdin suppresses the proliferation of several cell lines (5)(6)(7). Necdin interacts with various cytoplasmic and nuclear proteins (4, 6 -9), suggesting that this protein is multifunctional. Intriguingly, necdin interacts with the transcription factor E2F1, a major cell cycle regulator, and represses E2F1-dependent transcription (6). In addition, necdin interacts with viral oncoproteins SV40 large T antigen and adenovirus E1A. These characteristics resemble those of the retinoblastoma protein (Rb), 1 a principal growth suppressor involved in terminal mitosis and differentiation during neurogenesis (10). Necdin also contributes to the terminal differentiation of sensory neurons that are dependent on nerve growth factor (NGF) (11). Furthermore, necdin associates with the p75 neurotrophin receptor (p75NTR) (12), which also binds to the necdin-homologous MAGE protein NRAGE (MAGE-D1) (13). These observations suggest that necdin induces cell cycle arrest and controls neuronal apoptosis through interactions with E2F1 and p75NTR.
The human necdin gene is mapped to chromosome 15q11-q12, a region deleted in Prader-Willi syndrome (PWS) (14). PWS is a genomic imprinting-associated neurodevelopmental disorder, the major symptoms of which are feeding problems, gross obesity, and hypogonadism, suggesting abnormalities in hypothalamic neurons. The necdin gene is maternally imprinted, transcribed only from the paternal allele, and not expressed in individuals with PWS (15)(16)(17). Disruption of the mouse necdin gene results in early postnatal lethality (18), reduction in specific groups of hypothalamic neurons, and behavioral alterations, which are characteristics of the PWS phenotype (19). These findings suggest that necdin is responsible, at least in part, for the pathogenesis of PWS. On the other hand, the gene encoding MAGEL2 is located near the necdin gene and maternally imprinted (20,21). Thus, it appears that MAGEL2 is also involved in the pathogenesis of PWS. More recently, another gene encoding novel necdin homologous protein MAGE-G1 (also designated necdin-like 2) has been mapped to proximal chromosome 15q (22). The proximal region of human chromosomal 15q is subject to genomic imprinting and implicated in various human neurological and mental disorders including PWS, Angelman syndrome, autism, epilepsy, and schizophrenia, all of which show the parent-of-origin effects (23). These findings suggest that the necdin-related proteins are involved in brain development, and their abnormalities cause neurodevelopmental diseases. However, the biochemical and functional features of MAGEL2 and MAGE-G1 remain totally unknown at present.
Here we characterize MAGEL2 and MAGE-G1 based on the known features of necdin, and demonstrate that MAGE-G1, * This work was supported in part by a grant-in-aid for the National Project on Protein Structure and Functional Analysis from the Ministry of Education, Culture, Sports, Science and Technology of Japan. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ‡ Present address: Laboratory of Molecular Pharmacology, Graduate School of Natural Science and Technology, Kanazawa University, 13 but not MAGEL2, has characteristics similar to those of necdin such as growth suppression and interactions with E2F1 and p75NTR. The present findings suggest that other MAGE proteins with unknown function can be characterized using p75NTR and E2F1.

EXPERIMENTAL PROCEDURES
DNAs and Plasmids-Mouse full-length MAGEL2 coding sequence was synthesized by polymerase chain reaction (PCR) using mouse genomic DNA as a template and sets of primers flanking the coding region (20). Mouse MAGE-G1 cDNA was isolated from a phage library of neurally differentiated P19 embryonal carcinoma cells as one of the cDNA clones that hybridized with human MAGE-F1 cDNA (kindly provided by Dr. B. Stone, Virginia Mason Research Center, Seattle, WA). DNAs encoding MAGEL2 (amino acids 1-490) and MAGE-G1 (amino acids 1-279) were subcloned into pBluescript (Stratagene), 6ϫ Myc tag plasmid (kindly provided by Dr. M. McBurney, University of Ottawa, Ottawa, Ontario, Canada), and pcDNA3.1 (Invitrogen). Mouse necdin cDNA was subcloned into pBluescript, pRc/CMV, and p3xFLAG-CMV14 (1,9). The sequences and homologies were analyzed by Genetyx-Mac version 11 software (Genetyx, Tokyo, Japan).
RNA Blot Analyses-Total RNA was prepared from ICR strain mice at various developmental stages by the method of Chomczynski and Sacchi (24). Standard RNAs were synthesized in vitro with T7 RNA polymerase by transcription of the inserts for necdin, MAGEL2, and MAGE-G1 in pBluescript. For RNA dot blot analysis, known amounts of synthetic RNA and total RNA extracted from mouse tissues were blotted onto Hybond-N ϩ membrane (Amersham Biosciences) and fixed by UV irradiation. For Northern blot analysis, total RNA was electrophoresed on a 1% agarose gel containing 6% formaldehyde and blotted onto Hybond-N ϩ membrane. The membranes were incubated with DNA probes labeled with digoxigenin (DIG-High Prime kit, Roche Molecular Biochemicals). Hybridization was carried out as reported previously (25) except for the hybridization temperature of 50°C. The membrane was incubated with alkaline phosphatase-conjugated anti-digoxigenin antibody (Roche), and detected by CDP-Star chemiluminescence detection system (Roche). The equality and integrity of total RNA were examined by staining RNA on the membrane with ethidium bromide.
Cell Death Analyses-N1E-115 cells were transfected with the expression vectors pRc-E2F1, pRc-necdin, pcDNA3.1-Myc-MAGEL2, pcDNA3.1-Myc-MAGE-G1 (1.5 g each), and pRc-LacZ (1 g) using LipofectAMINE 2000. The amount of plasmid DNA was adjusted to 4 g by adding empty pcDNA3.1 vector. Cells were treated with Me 2 SO to differentiate 24 h after transfection, fixed 24 h later, and stained with the anti-␤-galactosidase antibody and Hoechst 33342 (Sigma) (9). Apoptotic cells carrying condensed or fragmented nuclei were counted. For lactic dehydrogenase (LDH) assay, N1E-115 were transfected with combinations of pRc-necdin (0 , and pcDNA3.1-HA-p75NTR (2.4 g) using LipofectAMINE 2000. The amount of plasmid DNA was adjusted to 4 g by adding empty pcDNA3.1. Transfected cells were cultured in the presence or absence of NGF (100 ng/ml), and induced to differentiate by Me 2 SO treatment 24 h after transfection. The conditioned medium was collected 72 h after transfection, and LDH activity was measured using the MTX LDH assay kit (Kyokuto) as described previously (27). LDH activity (A 560 of diformazan generated for 30 min at 37°C) was obtained by subtracting the value of the culture medium of untransfected cells.
Translocation Analyses-P19 cells treated with retinoic acid were double-stained for necdin with NC243 and chromosomal DNA with Hoechst 33342 as described previously (2,11). The images were captured with the CCD camera attached to the fluorescence microscope. N1E-115 cells grown on coverslips in 35-mm dishes were transfected with expression vectors (2 g each) for p3xFLAG-necdin, Myc-MAGE-G1, Lac-Z, E2F1, and HA-p75NTR using LipofectAMINE 2000. Cells were induced to differentiate 24 h after transfection, fixed 72 h later, and incubated with the antibodies against FLAG, Myc, ␤-galactosidase, E2F1, and p75NTR. Immunofluorescence images were obtained by confocal laser scanning microscopy (LSM5 PASCAL, Carl Zeiss).

Expression of the Necdin-related Genes in Mouse Brain-
Mouse MAGEL2 and MAGE-G1 encoded in the cloned DNAs consist of 490 and 279 amino acids, respectively, and their primary structures are consistent with those reported previously (20,22). MAGE-G1 is more homologous to necdin than MAGEL2; The degrees (percentage of identity) of overall protein sequences are as follows: necdin versus MAGEL2, 23 We examined the expression of mRNAs for necdin, MAGEL2, and MAGE-G1 in the mouse tissues in vivo using their DNAs as hybridization probes (Fig. 1A). These probes hybridized with their corresponding synthetic RNAs with similar intensities, and the detection limits were ϳ10 pg. By the dot blot analysis, necdin and MAGEL2 mRNAs were detectable in the brain at E14 and P70, but MAGEL2 mRNA was undetectable. The abundances of necdin mRNA in several brain samples were 5-10 times those of MAGE-G1 mRNA (data not shown). By Northern blot analysis using these probes, a ϳ1.7-kb necdin mRNA was readily detected in the brain and very weakly in the ovary at P70, whereas a ϳ1.7-kb MAGE-G1 mRNA was detected in all the tissues examined (Fig. 1B). The levels of MAGE-G1 mRNA in the brain, testis, and ovary were higher than those in other organs. We then analyzed the mRNA expression at various stages of brain development (Fig. 1C). The levels of necdin mRNA were relatively high at late embryonic stages (E16 and E18) and decreased thereafter, whereas those of MAGE-G1 mRNA in the brain remained relatively constant. On the other hand, MAGEL2 mRNA, for which the predicted size is ϳ2.5 kb (20), was undetectable by Northern blotting in any tissues at P70 and at any stage of brain development, consistent with the previous report (20). We failed to detect a ϳ4.5-kb MAGEL2 mRNA species reported previously (21).
Growth Suppression by the Necdin-related Proteins-Because necdin exerts a strong suppressive effect on the growth of various cell lines, we employed the SAOS-2 colony formation assay to examine whether MAGEL2 and MAGE-G1 suppress cell growth. Necdin and MAGE-G1 reduced the colony densities to 31 and 35% of the control level (transfected with the empty vector), respectively, whereas MAGEL2 showed no significant reduction (Fig. 2, A and B). We also examined the effects of these proteins on the proliferation of HEK293A cells by BrdUrd incorporation assay (Fig. 2C). Necdin and MAGE-G1 significantly reduced the number of BrdUrd-positive S-phase cells among ␤-galactosidase-positive cells to 37 and 43%, respectively, of the control level (Fig. 2D). In contrast, MAGEL2 exerted no significant effect on cell growth. Together, these results suggest that MAGE-G1, like necdin and unlike MA-GEL2, facilitates growth arrest.
Functional Interactions of the Necdin-related Proteins with E2F1-We next examined the effects of necdin and MAGE-G1 on E2F1-dependent transcriptional activity using a mouse c-Myc reporter in U2OS osteosarcoma cells (Fig. 4). E2F1 markedly increased the transcriptional activity ϳ20 times the control. Necdin strongly repressed the E2F1-stimulated activity in a dose-dependent manner. In contrast, MAGEL2 failed to repress E2F1-stimulated transcription. MAGE-G1, like necdin, repressed the E2F1-stimulated activity in a dose-dependent manner, although its inhibitory activity was weaker than that of necdin. These results suggest that necdin and MAGE-G1 suppress E2F1-induced transactivation through their interactions with the transactivation domain of E2F1.
Physical Interactions of the Necdin-related Proteins with p75NTR-Because necdin associates with p75NTR (12), we tested the interactions of the necdin-related proteins with p75NTR by co-immunoprecipitation assay (Fig. 6A). HA-tagged p75NTR was co-immunoprecipitated with Myc-tagged MAGE-G1 and FLAG-tagged necdin, but not with Myc-tagged MAGEL2, indicating that both necdin and MAGE-G1 bind to p75NTR in vivo. We determined the necdin-and MAGE-G1binding domains on p75NTR by in vitro binding assay using p75NTR mutants lacking intracellular functional domains (31) (Fig. 6B). Necdin interacted with the deletion mutant lacking the death domain (amino acids 1-341) and even weakly with the mutant lacking a COOH-terminal half of the juxtamembrane domain (amino acids 1-311) (Fig. 6C). In contrast, MAGE-G1 bound to the death domain (amino acids 340 -418) but not to the juxtamembrane domain.
Necdin and MAGE-G1 Form Endogenous Complexes with E2F1 and p75NTR in Neurally Differentiated P19 Cells-We attempted to demonstrate that necdin and MAGE-G1 endogenously form stable complexes with E2F1 and p75NTR. Western blotting revealed that P19 embryonal carcinoma cells express necdin, MAGE-G1, and E2F1 during the course of neuronal differentiation (Fig. 7A). The levels of E2F1 (58 kDa) and MAGE-G1 (32 kDa) were the highest in retinoic acidtreated P19 cells, whereas the levels of necdin (43 kDa) and p75NTR (68 and 75 kDa) were the highest in postmitotic neurons. We failed to detect TrkA by Western blotting (data not shown), in agreement with the previous finding that the TrkA gene is not expressed in P19 cells (32). We then examined the distribution of necdin immunoreactivity in P19 cells at different stages (Fig. 7B). A low level of necdin immunoreactivity was detected in the cytoplasm of undifferentiated cells, whereas necdin is readily detected in both the cytoplasm and the nucleus of retinoic acid-treated P19 cells. In postmitotic neurons, necdin was present predominantly in the cytoplasm and neurites, suggesting that necdin interacts with its interactors in the neuronal cytoplasm. We were unable to detect endogenous MAGE-G1 and p75NTR in P19 cells by immunocytochemistry with the antibodies used in the Western blot analysis (data not shown). We then employed immunoaffinity purification using antibodies against necdin and MAGE-G1 for detection of the endogenous complexes with E2F1 and p75NTR in retinoic acid-treated P19 cells and differentiated neurons. Necdin and MAGE-G1 bound to both E2F1 and p75NTR (Fig. 7,  C and D). These results suggest that necdin and MAGE-G1 endogenously form stable complexes with E2F1 and p75NTR in differentiated P19 cells.
E2F1 and p75NTR Translocate Necdin and MAGE-G1-Because necdin and MAGE-G1 interacted with both the nuclear factor E2F1 and the transmembrane protein p75NTR, we investigated whether necdin and MAGE-G1 alter their intracellular distribution when interacted with E2F1 and p75NTR (Fig. 8). N1E-115 neuroblastoma cells were transfected with expression vectors for FLAG-tagged necdin and Myc-tagged MAGE-G1 in combination with those for LacZ, E2F1, and p75NTR and observed by confocal laser microscopy (Fig. 8, A  and C). Necdin and MAGE-G1 were distributed almost evenly in the cytoplasm and the nucleus when co-expressed with ␤-galactosidase (Fig. 8, B and D). These patterns were similar to those in the cells expressing necdin or MAGE-G1 alone (data not shown). E2F1 translocated necdin and MAGE-G1 to the nucleus, whereas p75NTR mainly to the cytoplasm near the plasma membrane. Quantification of the cells showing preferential nuclear and cytoplasmic distribution patterns revealed that necdin and MAGE-G1 were mainly localized to the nucleus (89 and 86%, respectively) when co-expressed with E2F1, whereas they were localized to the cytoplasm (82 and 78%, respectively) when co-expressed with p75NTR (Fig. 8, B and  D). The distribution pattern of necdin or MAGE-G1 was unchanged in N1E-115 cells co-transfected with p75NTR cDNA when treated with NGF (100 ng/ml) (data not shown). These results suggest that necdin and MAGE-G1 alter their subcellular distribution under the influence of their binding partners E2F1 and p75NTR.
p75NTR Reduces the Association of Necdin and MAGE-G1 with E2F1-We then investigated whether the association of necdin and MAGE-G1 with E2F1 is competed by co-expressed p75NTR in differentiated N1E-115 neuroblastoma cells (Fig.  9). A small amount of endogenous p75NTR was detected in differentiated N1E-115 cells (Fig. 9, A and B, leftmost lanes of the bottom panels). When equal amounts of E2F1 and HA-p75NTR cDNAs were transfected, the amounts of necdin and MAGE-G1 in the E2F1 complexes decreased to 60 and 80%, respectively, of the control level (Fig. 9, A and B). When a larger amount of p75NTR expression vector was transfected, the amounts of necdin and MAGE-G1 in the E2F1 complexes decreased to 42 and 35%, respectively, whereas the amounts of necdin and MAGE-G1 immunoprecipitated with p75NTR increased 2.3-and 2.5-fold, respectively. These results suggest that necdin and MAGE-G1 interact with both E2F1 and p75NTR in differentiated N1E-115 cells. Because we failed to detect the NGF effects on the translocation of necdin and MAGE-G1 by immunocytochemistry, we used this immunoprecipitation assay for demonstrating the NGF effects. NGF slightly enhanced the association of necdin and MAGE-G1 with p75NTR but greatly reduced the amounts of necdin and MAGE-G1 in the E2F1 complexes. No appreciable changes of p75NTR and E2F1 binding activities of necdin and MAGE-G1 were noted when transfected cells were treated with NGF at the concentrations ranging from 50 to 200 ng/ml (data not shown).
We then examined whether overexpression of p75NTR influences E2F1-induced apoptosis in differentiated N1E-115 cells (Fig. 9C). The LDH release assay revealed that necdin and MAGE-G1 significantly suppressed the E2F1-induced cell death, whereas MAGEL2 failed to suppress it. p75NTR par- tially but significantly abrogated the necdin-and MAGE-G1induced suppression of apoptosis. Moreover, NGF significantly augmented the effect of p75NTR. These results suggest that co-expression of p75NTR potentiates E2F1-induced cell death by reducing the association of necdin and MAGE-G1 in the E2F1 complexes. DISCUSSION Using the necdin interactors E2F1 and p75NTR, we demonstrated that these MAGE proteins show distinct features. MAGE-G1 resembles necdin in all assays used in the present study, whereas MAGEL2 has no necdin-like activities. In addition, the level of MAGEL2 mRNA in the brain was too low to be detected by Northern blot analysis (Fig. 1). This suggests that the MAGEL2 gene is expressed only at limited regions in the brain. Indeed, it has most recently been reported that MAGEL2 mRNA is detected only in specific hypothalamic nuclei (33), whereas necdin mRNA is expressed throughout the brain, especially in the hypothalamus at the highest level (2,3). Thus, MAGEL2 may be less functional in brain development than necdin, although we cannot exclude the possibility that the function of MAGEL2 is distinct from that of necdin.
Recent studies have shown that necdin homologous proteins classified as type II MAGE proteins (34) such as NRAGE (MAGE-D1) and magphinin (MAGE-D4) suppress cell growth (13,35). However, the mechanism underlying growth suppression induced by these proteins remains unclear. Necdin and MAGE-G1 suppressed the growth of Rb-deficient SAOS-2 os-teosarcoma cells (Fig. 2). This indicates that necdin and MAGE-G1 require no functional Rb for their growth-suppressive effects. Furthermore, MAGE-G1, like necdin, bound to the COOH-terminal transactivation domain of Rb-target E2F1 (Fig. 3) and repressed E2F1-dependent transcriptional activity (Fig. 4). We have previously shown that E2F1 antagonizes necdin-induced growth suppression of undifferentiated neuroblastoma cells (9). These findings together suggest that necdin and MAGE-G1, like Rb, induce growth arrest through the repression of E2F1-dependent processes. Because type II MAGE proteins that have growth suppressive effects possess the homology domain resembling that of necdin (34), we assume that some of the type II MAGE proteins also target E2F1 and exert growth-suppressive effects.
E2F1 has been suggested to serve as a pro-apoptotic molecule in postmitotic neurons (29,36). We have previously reported that necdin suppresses E2F1-induced apoptosis in differentiated neuroblastoma cells (6,9). The present study has shown that MAGE-G1 also suppresses E2F1-induced apoptosis in differentiated neuroblastoma cells (Fig. 5). Necdin, MAGE-G1, and their interactor E2F1 are expressed during neuronal differentiation in vivo and in vitro (Figs. 1 and 7) (29,37), in which many cells undergo apoptosis. These findings suggest that necdin and MAGE-G1 serve as intrinsic anti-apoptotic proteins that prevent neuronal precursors and postmitotic neurons from E2F1-induced death.
We confirmed the previous findings that necdin interacts with the intracellular domain of p75NTR (12,38). The present study has shown that MAGE-G1 also interacts with p75NTR. Thus, it seems likely that p75NTR is a common interactor of necdin and its homologous MAGE proteins including MAGE-G1, MAGE-D1 (NRAGE), and MAGE-H1 (12,13). The necdinbinding site of p75NTR (amino acids 275-311) is close to the binding sites of tumor necrosis factor receptor-associated factors 4 and 6 (amino acids 245-313, 268 -283, respectively), which modulate c-Jun NH 2 -terminal kinase and nuclear factor B activities as well as apoptosis (39,40). NRAGE also interacts with a juxtamembrane region (amino acids 276 -329) of p75NTR (13). Furthermore, the necdin-binding domain encompasses a death domain termed Chopper (amino acids 273-301), which also modulates apoptosis (41). In contrast, MAGE-G1 interacted with the COOH-terminal death domain of p75NTR (amino acids 340 -418), with which necdin failed to interact (Fig. 6, B and C). The previous report has shown that necdin interacts with both the juxtamembrane and death domains by the yeast two-hybrid assay (12). This discrepancy may be the result of different configurations of the necdin protein expressed as fusion proteins in the assay systems. It is noteworthy that the p75NTR interactors SC-1, NRAGE, and NRIF1/ NRIF2 influence the cell cycle (13,42,43). We assume that necdin and MAGE-G1 are novel p75NTR interactors that negatively regulate apoptosis and cell cycle.
The present study has shown that necdin and MAGE-G1 interact with endogenous p75NTR and E2F1 in retinoic acidtreated P19 cells and postmitotic neurons (Fig. 7). Both necdin and MAGE-G1 in transfected neuroblastoma cells were sequestered by p75NTR in the cytoplasm near the plasma membrane and by E2F1 in the nucleus (Fig. 8). In addition, overexpression of p75NTR greatly reduced the association of necdin and MAGE-G1 with E2F1, which may cause deregulation of E2F1 to induce cell death (Fig. 9). NGF modestly increased the association of necdin and MAGE-G1 with p75NTR, whereas it drastically reduced the association with E2F1 (Fig. 9, A and B). This may be because NGF increases the amounts of necdin and MAGE-G1 that weakly bind to p75NTR and are lost after FIG. 9. Overexpression of p75NTR reduces association of necdin and MAGEL2 with E2F1. A, co-immunoprecipitation assay for necdin. N1E-115 cells were transfected with combinations of expression vectors (amounts indicated at the top in g) for FLAG-necdin (FLAG-Necdin), E2F1 (E2F1), and HA-p75NTR (HA-p75NTR), induced to differentiate, and cultured in the presence (ϩ) or absence (Ϫ) of NGF (100 ng/ml). Cell lysates (400 g) were immunoprecipitated (IP) with antibodies against E2F1 (E2F1, top panel) and p75NTR (p75NTR, middle panel) and immunoblotted (IB) with anti-FLAG antibody (FLAG). Signal intensities are shown under the top and middle panels. The aliquots (10 g) were immunoblotted for FLAG, E2F1, and p75NTR (lower three panels). B, co-immunoprecipitation assay for MAGE-G1. N1E-115 cells were transfected with combinations of expression vectors for Myc-MAGE-G1 (Myc-G1), E2F1, and HA-p75NTR and treated as in A except anti-Myc-antibody was used for immunoblotting. C, LDH assay. N1E-115 were transfected with combinations of expression vectors for E2F1 (E2F1), HA-p75NTR (p75NTR), necdin (Necdin), Myc-MAGEL2 (L2), and Myc-MAGE-G1 (G1). Transfected cells were induced to differentiate and cultured in the presence or absence of NGF (100 ng/ ml). The conditioned medium was collected 72 h after transfection for LDH measurement (mean Ϯ S.E., n ϭ 6). *, p Ͻ 0.02; **, p Ͻ 0.05, significantly different from the values of E2F1ϩ/Necdinϩ and E2F1ϩ/p75NTRϩ/Necdin(or G1)ϩ, respectively.
immunoprecipitation. Another possibility is that NGF-activated p75NTR facilitates the association of necdin and MAGE-G1 with their interactors other than p75NTR in the cytoplasm. These findings suggest that NGF-activated p75NTR, at least in part, controls E2F1-dependent processes such as cell cycle and apoptosis by sequestering necdin and MAGE-G1 from the nucleus. Other interpretations are also possible. For example, it has recently been reported that the p75NTR intracellular domain is released by ␥-secretase-mediated cleavage and translocated into the nucleus, where it modulates intracellular events (44,45). Thus, it is tempting to speculate that the soluble p75NTR intracellular domain and E2F1 in the nucleus interact with necdin-related MAGE proteins in a competitive manner. It is also noteworthy that, because most of the present data are based on the ectopic expression in proliferative cell lines, the translocation and function of the endogenous MAGE proteins may be very different in primary neurons and glia.
The present findings that necdin and MAGE-G1 share common biochemical and functional features suggest that these two proteins act complementarily in brain development. Another necdin-like protein, MAGE-G2, which is highly homologous to MAGE-G1, has been reported (GenBank TM accession no. AF319980). This protein may also possess characteristics similar to those of necdin. We speculate that these necdinhomologous proteins compensate for the absence of necdin expression in Prader-Willi syndrome and necdin knockout mice. This idea may account for the inconsistent phenotypes seen in necdin knockout mice (18,19,46,47). Further studies on necdin-homologous MAGE proteins will provide valuable insights into the roles of these family proteins in brain development and neurodevelopmental disorders.