APP-BP1, first identified as a protein that interacts with the carboxyl (C) terminus of the amyloid precursor protein (APP), is one-half of the bipartite activating enzyme for the ubiquitin-like protein NEDD8. We report here that APP-BP1 also specifically interacts with apoptosis stimulating protein of p53 ASPP2 in non-transfected cells through the functional predominant N-terminal domain ASPP2(332–483). ASPP2 inhibits the ability of APP-BP1 to rescue the ts41 cell cycle mutation and inhibits APP-BP1 induced apoptosis in primary neurons. ASPP2 reduces the ability of NEDD8 to conjugate to Cullin-1, inhibits APP-BP1-dependent ts41 cell proliferation, and blocks the ability of APP-BP1 to cause apoptosis and to cause DNA synthesis in neurons. We also show that ASPP2 activates nuclear factor-κB (NF-κB) transcriptional activity, which seems to be inhibited by the neddylation pathway since the dominant negative NEDD8 activating enzyme causes enhanced NF-κB activity. Our data provide the first in vivo evidence that ASPP2 is a negative regulator of the neddylation pathway through specific interaction with APP-BP1 and suggest that dysfunction of the APP–BP1 interaction with APP may be one cause of Alzheimer's disease.
sodium dodecyl sulfate–polyacrylamide gel electrophoresis
APP-BP1 was first identified as a protein that interacts with the carboxyl (C) terminus of the amyloid precursor protein (APP) (Chow et al. 1996). Subsequently, we and others showed that APP-BP1 is a cell cycle protein that acts as one half of the bipartite activating enzyme for the ubiquitin-like protein NEDD8 (Osaka et al. 1998; Gong and Yeh 1999; Chen et al. 2000). We also showed that the human APP-BP1 molecule could rescue the ts41 mutation in Chinese hamster lung cells (Chen et al. 2000). This mutation previously had been shown to lead to successive S phases of the cell cycle without intervening G2, M and G1, suggesting that the product of this gene negatively regulates entry into the S phase and positively regulates entry into mitosis (Hirschberg and Marcus 1982; Handeli and Weintraub 1992). We showed (Chen et al. 2000) that expression of APP-BP1 in ts41 cells at the non-permissive temperature drives the cell cycle through the S-M checkpoint, and that this function is mediated by the NEDD8 conjugation (neddylation) pathway. We also demonstrated that overexpression of APP-BP1 in primary neurons causes apoptosis, and that this apoptosis can be blocked by inhibition of neddylation.
APP-BP1 is homologous to the amino (N) terminus of the ubiquitin-activating enzyme E1, but APP-BP1 lacks the conserved cysteine required for E1 ubiquitin activation activity. APP-BP1 forms a heterodimer with hUba3 that is homologous to the C terminus of E1 and contains the conserved cysteine. Together APP-BP1 and hUba3 fulfill E1-like functions for NEDD8 activation. Activated NEDD8 forms a thiol ester bond with the conjugating enzyme hUbc12, which is analogous to E2 in the ubiquitination pathway, and is covalently targeted to proteins at a lysine residue (reviewed in Hochstrasser 2000). All known neddylation targets in mammalian cells are cullin (Cul) family members (Kipreos et al. 1996; Hori et al. 1999). Cul-1 is also a major component of the ubiquitin-ligase known as SCF complex (consisting of the core subunits Skp1, Cul-1, ROC1, and substrate recognition adaptors known as F-box proteins), which is involved in ubiquitination of a multitude of proteins (Deshaies 1999; del Pozo and Estelle 2000). NEDD8 conjugation to Cul-1 catalyzes the ubiquitination of inhibitor of NF-κB (IκBα), β-catenin, p27Kip1, and cyclin E. Cul-2 is a component of a multiprotein complex termed VEC (consisting of pVHL, elongin C and Cul2), which is structurally and functionally similar to SCF complexes. NEDD8 conjugation to Cul-2 catalyzes the ubiquitination of the hypoxia-inducible transcription factor HIF (Ohh et al. 2002).
Numerous studies have demonstrated the in vivo importance of the neddylation pathway in cell cycle regulation. ts41 cells, which have a temperature sensitive mutation in APP-BP1 and therefore lack a functional NEDD8 activating enzyme, show successive S phases of the cell cycle without intervening G2, M and G1 (Hirschberg and Marcus 1982; Handeli and Weintraub 1992). Mice with a deletion of Uba3, the catalytic subunit of the NEDD8 activating enzyme (Tateishi et al. 2001), die in utero due to defects in both mitotic and endoreduplicative cell cycle progression. They also accumulate cyclin E, and β-catenin, a mediator of the Wnt/wingless signaling pathway. The knockout animal data suggest that Cul-1 is required for the developmentally regulated G1 to G0 transition (Kipreos et al. 1996; Dealy et al. 1999; Wang et al. 1999).
Two major considerations prompted us to search for proteins other than Uba3 that may interact with APP-BP1. One is that the studies cited above show effects of neddylation primarily on the cell cycle in dividing cells. However, we observed high levels of APP-BP1 expression in neurons (Chow et al. 1996), suggesting that APP-BP1-mediated neddylation has non-cell cycle related functions in neurons and/or that APP-BP1 participates in other pathways besides neddylation in neurons. The other is that the heterodimeric nature of the NEDD8 activating enzyme suggests that APP-BP1 can be modulated by other proteins, thereby influencing the enzymatic activity of the catalytic subunit hUba3. We therefore screened a brain cDNA expression library to identify proteins that may interact with APP-BP1. Using GST-APP-BP1 as a protein probe, we retrieved a partial cDNA encoding the C terminus of apoptosis stimulating protein of p53 ASPP2.
In the present study, we characterize both biochemically and functionally the interaction of ASPP2 with APP-BP1, and provide the first in vivo evidence that ASPP2 is a negative regulator of the neddylation pathway through specific interaction with the regulatory subunit of the NEDD8 activating enzyme, APP-BP1.
Materials and methods
Expression screening with GST-APP-BP1
Full-length APP-BP1 was inserted in-frame with the glutathione S-transferase (GST) coding sequence in pGSTag (Ron and Dressler 1992), between the Sma1 and SalI sites. The purified GST-APP-BP1 fusion protein was used to screen a human fetal brain library exactly as described (Chow et al. 1996).
Generation of rabbit anti-ASPP2 antibody 636 A
A sequence encoding amino acids 124–692 of ASPP2 was inserted in-frame with the GST coding sequence in pGEX-KG (Guan and Dixon 1991). The resultant fusion protein was expressed in bacteria and purified from inclusion bodies as described (Harlow and Lane 1988). Antibodies were generated in rabbits, and the serum (636A) was preadsorbed against a GST column (Research Genetics, Inc.). The 636 A antibody was then affinity-purified using protein A immobilized on Sepharose 4B fast Flow (Sigma, St Louis, MO, USA) as described (Ey et al. 1978).
All plasmid constructs were made in pcDNA3 and pHSVPrpUC vectors using standard techniques, and were verified by sequence analysis. Protein expression was confirmed by immunoblot analysis using the appropriate antibodies.
Both HEK293 and HELA cells were maintained in Dulbecco's modified Eagle's medium (DMEM; Mediatech, Inc., Herndon, VA, USA) containing 10% fetal bovine serum (Hyclone, Logan, UT, USA) and 1% penicillin–streptomycin–glutamine (Invitrogen, Grand Island, NY, USA) in a 10% CO2, 37°C, humidified incubator.
The ts41 cell line is a V79 Chinese hamster mutant first isolated by Hirschberg and Marcus (1982), who showed that at non-permissive temperature (40°C), cells go through successive S phases without progressing into G2. The ts41 cells were grown as a monolayer in DMEM (Mediatech, Inc.) containing 10% cosmic calf serum (Hyclone Laboratories), 1% penicillin–streptomycin–glutamine (Invitrogen), and 500 µg/L amphotericin B (Sigma) in 10% CO2 at 34°C.
Pregnant adult female Sprague–Dawley rats were used as the source of E18 rat embryos for establishment of primary cortical cell cultures. Before the animals were killed for removal of the embryos, they were rendered unconscious with carbon dioxide and then beheaded using a guillotine. This method was chosen because it is rapid and because it entails no introduction of drugs into the animals, which could interfere with our results. The dissociated cortical cells were plated in poly-d-lysine-coated 24-well plates with glass coverslips (Fisher Scientific, Pittsburgh, PA, USA) for the apoptosis assays and BrdU labeling experiments. For the luciferase assays, the cortical cultures were plated onto poly-d-lysine-coated 60-mm dishes. The cortical cultures were grown in Neurobasal medium supplemented with B27 (Invitrogen), 1% fetal bovine serum, and 1% horse serum (both from Hyclone). Neurons were used for the experiments at 4–5 days in vitro.
Methods for coimmunoprecipitations of transiently transfected HEK293 cells or ts41 cells have been described (Chen et al. 2000). Coimmunoprecipitation of endogenous ASPP2 with APP-BP1 was performed with non-transfected HEK293 cells. Confluent HEK293 cells were harvested with ice-cold phosphate-buffered saline (PBS), pelleted, and lysed for 15 min on ice in a buffer containing 0.2% Triton X-100, 50 mm Tris (pH 8.0), 150 mm NaCl, and 1 mm EDTA plus protease inhibitors. The lysate was then precleared with protein G and normal rabbit serum for 1 h. Primary antibody 636A or its preimmune serum was added to 13 mg of total protein extract and incubated on ice for 2 h. Protein G (30 µL of the 50% slurry) was then added to precipitate the immune complex. The immune complex was then washed once with the lysis buffer and washed four times in a buffer containing 0.1% Triton X-100, 50 mm Tris (pH 8.0), 150 mm NaCl, and 1 mm EDTA plus protease inhibitors. The samples were then mixed with 4X sample loading buffer and electrophoresed through 7.5% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE). The proteins were transferred to nitrocellulose, after which the membrane was blocked with 6% non-fat milk in PBS overnight and probed with rabbit anti APP-BP1 BP339 (Chen et al. 2000).
ts41 cell proliferation assays
These assays were carried out essentially as described previously (Chen et al. 2000). Since only APP-BP1-transfected cells can survive at 40°C, we designated the cell number for this sample as 100% cell proliferation; all the other samples were expressed as a percentage in relation to the APP-BP1 sample. The cell count did not exclude those few large cells that were present in all samples. These large cells represented those that had accumulated DNA content over time at the non-permissive temperature and were on the verge of dying due to the absence of APP-BP1 expression.
Five days post plating, primary neuronal cultures were infected with HSV-APP-BP1 in the presence of 10 µm BrdU (Sigma). Cells were then washed twice with D-PBS (Cellgro) and were fixed in 70% alcohol for 15 min at 4°C. Cells were treated with 2 m HCl for 30 min, washed with D-PBS three times, and then blocked with Tris Buffer (0.1 m Tris, pH 7.4, 0.85% NaCl, 0.1% Triton X-100, 2% bovine serum albumin (BSA), and 10% normal goat serum) for 15 min. Both primary and secondary antibodies were diluted in the Tris Buffer. The cells were incubated with the rabbit anti BrdU antibody (Megabase Research Products, Lincoln, NE, USA; diluted 1 : 100) for 2 h, washed twice for 5 min each with Tris buffer, and then incubated with secondary Alexia 488-conjugated goat anti-rabbit antibody (Molecular Probes) for 1 h. Neurons were identified using the mouse anti-neuronal nuclei (NeuN) antibody (Chemicon International, Inc., Temecula, CA, USA; 1 : 50) together with the secondary Cy5-conjugated goat anti-mouse antibody (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, USA). Transfected cortical cultures were stained using the same protocol. The primary rabbit anti-ASPP2 antibody 636A was used to identify mycASPP2(1–483) transfected cells, while neurons were identified with mouse anti-NeuN (Chemicon). Cells were washed twice with Tris Buffer without normal serum, 10 min each, and then washed three times, 5 min each with phosphate buffer (10 mm phosphate, pH 7.4, 0.85% NaCl). Coverslips were mounted in Gel/Mount (Biomedia corp., Foster City, CA, USA). Images were collected on a Leica confocal microscope.
Assays for neuronal apoptosis and DNA synthesis
Bisbenzimide staining assays for neuronal apoptosis following infection of neurons with HSV vectors have been described in Bursztajn et al. (1998). For analysis of DNA synthesis, BrdU (Sigma) was added to the neuronal cultures at a final concentration of 10 µm at the time of infection with HSV vectors. Infections were done at a multiplicity of infection (moi) of 1; where necessary, the total amount of virus in the infections was adjusted with HSV-LacZ virus. Fourteen hours postinfection, cells were fixed in cold 70% ethanol for 30 min and processed according to the protocol provided in the Zymed BrdU labeling kit. According to the protocol, cells were incubated with biotinylated mouse anti-BrdU antibody (Zymed) for 60 min. Coverslips were then rinsed three times, for 2 min each, with PBS. The cells were incubated with streptavidin peroxidase for 10 min, after which they were rinsed three times, for 2 min each, with PBS. Positive staining was visualized with diaminobenzidine (DAB). The coverslips were mounted onto glass slides for microscopy. Ten random fields of 200–300 cells each were analyzed for each condition. The number of cells with positively stained nuclei relative to the total number of cells per field was calculated and expressed as a percentage.
On the fifth day after plating, rat primary cortical neurons were transfected with 1 µg of the 3X-κB-L/pGL2-Basic luciferase reporter vector (Mitchell and Sugden 1995), 0.35 µg of the pGKβgal reporter vector (Kaye et al. 1996), and 1 µg of the appropriate construct in pcDNA3 vector using LipofectAMINE and Plus Reagent (Invitrogen). Forty-three hours later, neurons were washed once with ice-cold PBS and scraped into cold PBS. Cells were collected by centrifugation and processed using the Luciferase Assay System (Promega). Luciferase activity in 60 µL of the lysate was measured using a DYNEX luminometer, following the Luciferase Assay System protocol for plate-reading luminometers. We also measured β-galactosidase activity in 7 µL of the same lysate using the β-galactosidase Reporter Gene Assay System (Applied Biosystems) and the DYNEX luminometer. Samples were read in duplicate for each assay, and the luciferase activity was normalized to the corresponding β-galactosidase activity to control for variations in transfection efficiency. Transfection of neurons using LipofectAMINE and Plus Reagent results in about 0.8% transfected cells (unpublished data, Y. Chen and R. Neve). To confirm that neuronal cells were transfected in this assay, we used the same protocol to transfect cortical cultures with mycASPP2(1–483)/pcDNA3, and double labeling immunocytochemistry was carried out as described above, using anti-ASPP2 antibody 636A and anti-NeuN. Cells immunostained for the ASPP2 were counted in 10 random fields among which neurons were identified by the NeuN marker. Among a total of 59 transfected cells in these fields, 54% were neurons.
Statistical analysis was performed using analysis of variance (anova). When a significant p-value (p < 0.05) was found, t-tests assuming unequal variance were performed to compare individual groups. anova (also called the F ratio) is designed to establish whether or not a significant (non-chance) difference exists among several sample means. Statistically, a large F ratio, that is when the variance between samples is larger than the variance within samples, usually indicates a non-chance or significant difference. The t-test assuming unequal variance assumes that the variances of both ranges of data are unequal and test whether two sample means are equal. All error bars represent standard error of the mean and are based on the variability within the sample and the sample size.
APP-BP1 interacts with ASPP2
To isolate cDNAs encoding cellular proteins capable of interacting with the APP-binding protein APP-BP1, we screened a human fetal brain expression cDNA library with bacterially produced 32P-labeled APP-BP1. Two independent and identical positive clones were isolated and shown by sequence analysis to encode amino acids 570–1005 of the p53- and Bcl2-binding cell cycle protein 53BP2/Bbp (Iwabuchi et al. 1994; Naumovski and Cleary 1996), now known to be a partial version of ASPP2 (Samuels-Lev et al. 2001). cDNAs encoding the 1005-amino acid Bbp and the 1128-amino acid ASPP2 proteins were obtained from L. Naumovski (Stanford).
We generated an ASPP2 rabbit polyclonal antibody (636A) using GST-ASPP2(124–692) as the immunogen. ts41 cells were transfected with pcDNA3 vector alone (odd-numbered lanes, Fig. 1a) or with pcDNA3/mycBbp (even-numbered lanes, Fig. 1a). To test the specificity of the 636A antibody, immunoblots of the two lysates were probed with anti-myc epitope 9E10, 636A, or the preimmune serum (Fig. 1a). As expected, 636A, but not the preimmune serum, recognized the myc-tagged Bbp.
To test whether the APP–BP1 interaction with the ASPP2 fragment encoded by the partial cDNA isolated in the original screens occurs in vivo, vectors encoding myc-APP-BP1 and HA-ASPP2(693–1128) were cotransfected into HEK293 cells, and lysates of the transfected and control vector-transfected cells were immunoprecipitated with the HA epitope-specific antibody 12CA5. An immunoblot of the lysates (Fig. 1b) was probed with 9E10, revealing that myc-APP-BP1 coprecipitated with HA-ASPP2(693–1128). To test whether endogenous APP-BP1 and ASPP2 interact, we immunoprecipitated HEK293 cell lysates with either the ASPP2 antibody 636A or its preimmune serum (Fig. 1c). The immunoblot of the lysates was probed with the anti-APP-BP1 antibody BP339 (Chen et al. 2000), revealing that APP-BP1 coprecipitated with ASPP2 in non-transfected cells.
To determine which domain of ASPP2 interacts with APP-BP1, we constructed ASPP2 truncation mutants (Fig. 1d). We transfected HEK293 cells with HSV-myc-APP-BP1 together with HSV vectors expressing each of the ASPP2 mutants (Fig. 1e). The lysates were immunoprecipitated with the anti-ASPP2 antibody 636A, and were blotted with the anti-myc antibody 9E10. The results revealed that myc-APP-BP1 coprecipitated with all of the ASPP2 truncation mutants except for ASPP2(124–331). Thus, APP-BP1 appears to interact with two different domains of ASPP2, C-terminal amino acids 693–1128 and a region close to the N terminus, amino acids 332–483. A number of proteins interact with the C terminus of ASPP2, but APP-BP1 is the only protein that has been found to interact with a region of ASPP2 close to the N terminus.
ASPP2 inhibits the ability of APP-BP1 to rescue the ts41 mutation
Expression of wild-type APP-BP1 can rescue the ts41 mutation in Chinese hamster cells at the non-permissive temperature (Chen et al. 2000). It has been reported that ASPP2/Bbp can either induce cell death or inhibit cell cycle progression in dividing cells (30–32). We found that ASPP2 and Bbp both inhibited, to the same extent (31%), cell proliferation mediated by APP-BP1 expression at the non-permissive temperature (Fig. 2a). Two aspects may contribute to the incomplete inhibition of ts41 cell growth by ASPP2: one is that not all proliferating APP-BP1 transfected cells are also transfected by ASPP2; the second is that intermolecular interactions with ASPP2 C terminus may modulate such inhibition because an even stronger inhibition was observed by the ASPP2(124–483) fragment as shown below. To determine the functional significance of the two APP-BP1 binding sites on ASPP2 in this assay, we expressed APP-BP1 together with selected ASPP2 truncation mutants in the ts41 cells. As shown in Fig. 2(a), the N terminus of ASPP2 (amino acids 124–483) inhibited the proliferative effect of APP-BP1 the most (57% inhibition), whereas the C terminus of ASPP2 (amino acids 693–1128, containing the proline-rich domain, ankyrin repeats and SH3 domain) did not. Thus, the dominant inhibitory effect of ASPP2 on ts41 cell growth was present in the APP-BP1 binding region close to the N terminus of ASPP2. Interestingly, ASPP2(1–483) inhibited less the proliferative effect of APP-BP1 than does ASPP2(124–483), suggesting that amino acids 1–123 of ASPP2 might interact with a separate protein that antagonizes the inhibitory effect of amino acids 124–483.
APP-BP1 is the regulatory subunit of the NEDD8 activating enzyme. Therefore, we assessed whether ASPP2 interaction with APP-BP1 affected the neddylation of the NEDD8 target proteins Cul-1 and Cul-2 (Fig. 2b). HeLa cells were transfected with pcDNA3 vectors encoding ASPP2, Bbp, ASPP2(124–483), or HA-ASPP2(693–1128). Lysates of the transfected cells were subjected to immunoblot analysis using rabbit antibodies to Cul-1 or Cul-2. Endogenous Cul-1 in HeLa cells is present primarily as the NEDD8-conjugated form (upper band; see pcDNA3-transfected lane). However, expression in HeLa cells of ASPP2(124–483), the fragment of ASPP2 that inhibited ts41 cell growth, shifted the ratio of NEDD8-conjugated and -non-conjugated forms of Cul-1 so that the latter predominated. However, neddylation of Cul-2 did not seem to be affected by overexpression of ASPP2 or its subfragments. Interestingly, ASPP2 appears to downregulate the total protein level of Cul-1 but not of Cul-2.
ASPP2 protects neurons from APP-BP1 induced apoptosis
We reported previously that APP-BP1 overexpression in neurons caused apoptosis, and that this neuronal death was inhibited by dominant negative mutants of hUba3 (enzymatic partner of APP-BP1) or of the NEDD8-conjugating enzyme hUbc12 (Chen et al. 2000). We therefore tested ASPP2 and its deletion mutants for their effect on APP-BP1-mediated apoptosis in primary rat cortical neurons. HSV-APP-BP1 was coexpressed in primary neurons with HSV vectors expressing different truncation mutants of ASPP2, each at an moi of 1 (Fig. 3a). The cells were then fixed and analyzed for condensed nuclei by bisbenzimide staining (Fig. 3b). The results indicated that the ASPP2 domain near the N terminus (amino acids 331–483) effectively protected neurons against APP-BP1-induced apoptosis. ASPP2 and Bbp exhibited similar protection against APP-BP1-induced apoptosis, but ASPP2(1–331) and ASPP2(1–123) did not.
ASPP2 inhibits APP-BP1-induced DNA synthesis in neurons
Because of the involvement of APP-BP1 in the cell cycle, we tested whether overexpression of APP-BP1 caused DNA synthesis to occur in neurons, surmising that entry into the cell cycle might lead to the apoptosis caused by APP-BP1 overexpression. HSV-APP-BP1 or HSV-LacZ was expressed in primary neurons in the presence of BrdU, and the cells were fixed and analyzed for BrdU immunoreactivity. As shown in Fig. 4(a), overexpression of APP-BP1 does indeed cause an increase in the number of cells undergoing DNA synthesis relative to HSV-LacZ-infected or mock-infected controls (41% versus 26%). Just as the APP-BP1-induced neuronal apoptosis is inhibited by coexpression of a dominant negative mutant of the NEDD8-conjugating enzyme hUbc12 (C111S) (Chen et al. 2000), APP-BP1-induced neuronal DNA synthesis also was inhibited by coexpression of the C111S mutant of hUbc12 (data not shown).
To confirm that the increased DNA synthesis caused by APP-BP1 occurred in neurons, we did double staining experiments in which the cultures were subjected to immunocytochemistry with the neuron-specific NeuN antibody (red fluorescence) and an anti-BrdU antibody (green fluorescence). As seen in Fig. 4(b), the basal level of DNA synthesis in the cultures appears to be due to non-neuronal cells, which comprise approximately 21% of the cells. These non-neuronal cells, typically with large nuclei, were immunopositive for BrdU but not for NeuN. APP-BP1 did not induce apoptosis in these cells (data not shown), consistent with our observation that it causes proliferation rather than apoptosis of ts41 cells. The HSV-APP-BP1-infected cultures contained many double-labeled cells (arrows, Fig. 4a), which represented neurons undergoing DNA synthesis.
The observation that ASPP2 inhibited ts41 cell growth supported by APP-BP1 expression at the non-permissive temperature, suggested that ASPP2 might promote neuronal survival in the presence of APP-BP1 overexpression by prohibiting cell cycle entry in neurons through inhibition of neddylation. HSV-APP-BP1 was coexpressed in primary neurons with HSV vectors expressing ASPP2 or truncation mutants of ASPP2 in the presence of BrdU (Fig. 4a). The results showed that the ASPP2 domain near the N terminus (amino acids 332–483) protected neurons against APP-BP1-induced DNA synthesis.
ASPP2 activates NF-κB
The results described above suggest that ASPP2 protects neurons from APP-BP1-induced apoptosis by inhibiting APP-BP1-induced DNA synthesis via the neddylation pathway. Neddylation of Cul-1 was shown to enhance the activity of the ubiquitin ligase SCF complex, which controls the degradation of a number of proteins, including IκB, through the proteasome pathway (Read et al. 2000). The C terminus of ASPP2 was reported to interact with the NF-κB subunit p65 (Yang et al. 1999). It is also known that NF-κB has neuroprotective effects (Tamatani et al. 1999; Mattson and Camandola 2001). Therefore we examined ASPP2 and its truncation mutant ASPP2(1–483) for their ability to activate NF-κB-mediated transcription in neurons (Fig. 4b). We found that ASPP2 and ASPP2 (1–483) both activated NF-κB-mediated transcription in neurons. We then coexpressed ASPP2 or ASPP2 (1–483) with a dominant negative mutant (C218S) of the NEDD8-activating enzyme hUba3 and assayed for NF-κB transcriptional activity. The results revealed that hUba3 (C218S) enhanced NF-κB activation by ASPP2 or ASPP2(1–483). The dominant negative mutant (C111S) of the NEDD8 conjugating enzyme hUbc12 yielded similar results (data not shown). The same data (not shown) were obtained in ts41 cells and HEK293 cells, except that ASPP2 inhibited cell proliferation in these dividing cells rather than protecting against apoptosis, as it did in neurons. We did not use any known NF-κB activators such as tumor necrosis factor alpha (TNFα) to stimulate the cells in these assays, but instead we examined the endogenous level of NF-κB activity in relation to the interaction of ASPP2 with APP-BP1. Our data suggested that the native NF-κB activity might not depend on Cul-1 neddylation and there is evidence in the literature supporting our view (Read et al. 2000; Ohh et al. 2002). In contrast, our data supported the hypothesis that the native NF-κB activity is inhibited by the neddylation pathway.
We have reported that APP-BP1 interacts with ASPP2 at two different sites, one near the N terminus, within amino acids 332–483, and the second within the C-terminal amino acids 693–1128, which contain a proline-rich domain, the ankyrin repeats, and an SH3 domain. To our knowledge, APP-BP1 is the first protein discovered to interact with ASPP2 in its N-terminal half. The N-terminal APP-BP1-interacting domain of ASPP2 appears to be the locus of the major effect of the ASPP2 holo-protein on the neddylation pathway. This domain inhibits the proliferative effect of APP-BP1 in ts41 cells, and protects neurons from APP-BP1-induced apoptosis, suggesting that this region normally dominates these specific functional interactions of ASPP2 with APP-BP1. Since the effect of ASPP2 or Bbp is less than that of ASPP2(124–483) fragment, the C-terminal portion of ASPP2 may have a modulatory effect due to its interaction with other proteins.
NF-κB activation is associated with neuronal survival (Tamatani et al. 1999; Mattson et al. 2001). We have shown that NF-κB activation by ASPP2 seems to have different outcomes in dividing cells versus postmitotic neurons. In dividing cells, ASPP2 activated NF-κB and inhibited APP-BP1-induced cell proliferation. In neurons, ASPP2 protected neurons from APP-BP1-mediated apoptosis and such protection is associated with NF-κB activation. In both cases, nonetheless, ASPP2 antagonized the action of APP-BP1, which we have shown previously enhances cell proliferation in dividing cells and causes apoptosis in neurons. These seemingly disparate effects of APP-BP1 may have the same mechanism, since we showed in this paper that APP-BP1 caused DNA synthesis in the neurons, which presumably preceded the apoptosis.
We also showed in the present paper that ASPP2(124–483) activation of NF-κB is enhanced by inhibition of neddylation, suggesting that there is a NF-κB activation pathway independent of Cul-1-NEDD8 conjugation and that neddylation normally inhibits such NF-κB activation. Our data suggest that ASPP2 protects neurons from APP-BP1-induced apoptosis by activating NF-κB and that inhibition of neddylation enhances this effect overall. APP-BP1 was initially identified as a protein that interacts with APP. Interestingly, the expression of APP apparently is regulated by NF-κB (Grilli et al. 1995). The APP gene is rapidly transcribed in the brain in response to a number of stress conditions, including head trauma, focal ischemia, neurotoxicity, and heat shock (reviewed in O'Neill and Kaltschmidt 1997; Ghosh et al. 1998; Perkins 2000). APP in turn interacts with APP-BP1, which may suppress further activation of NF-κB by interacting with hUba3 associated with the inflammatory response, or may enhance activation of NF-κB by interacting with ASPP2 associated with neuronal survival.
In summary, we have identified a new APP-BP1 binding partner, ASPP2. The specific interaction between APP-BP1 and ASPP2 modulates the neddylation pathway and results in inhibition of cell proliferation or protection against APP-BP1-mediated apoptosis in neurons. This interaction is in contrast to the interaction between APP-BP1 and hUba3, which promotes cell cycle progression through the S-M checkpoint in dividing cells and induces apoptosis in neurons. To our knowledge, ASPP2 is the first protein shown to modulate neddylation at the activating enzyme step of this cascade of reactions. This modulation appears to be specific to Cul-1, since ASPP2 did not affect NEDD8 conjugation to Cul-2. The cul-1 knockout studies have shown that Cul-1 is essential for controlling cell cycle transition at G1-S and G1-G0 checkpoints (Kipreos et al. 1996; Dealy et al. 1999), but these studies do not relate Cul-1 function to neddylation. We have shown indirectly that de-neddylation of Cul-1 causes inhibition of APP-BP1 dependent ts41 cell growth and protection of neurons from apoptosis. It is not clear why ASPP2 inhibits Cul-1 but not Cul-2 neddylation. One possibility is that there are distinct conjugating enzymes specific for Cul-1 and Cul-2 neddylation, and that the binding of ASPP to APP–BP1 inhibits the interaction between the NEDD8 activating enzyme and the conjugating enzyme that is specific for Cul-1 neddylation.
The involvement of APP-BP1 in the cell cycle via its opposing interactions with hUba3 and ASPP2 is of interest in the light of numerous findings of cell cycle abnormalities in AD (Lee et al. 1992; Pope et al. 1994; Liu et al. 1995; Vincent et al. 1996; Vincent et al. 1997; Busser et al. 1998; Chow et al. 1998). Evidence has been accumulating (Guo et al. 1998; Cotman and Anderson 1995) that some neurons degenerate via apoptotic pathways in Alzheimer's disease. Most recently, Yang et al. (2001) showed directly that a significant number of hippocampal pyramidal and basal forebrain neurons in AD brain have undergone DNA replication. We hypothesize that dysfunction of pathways mediated by APP may be one cause of the reactivation of cell cycle proteins in AD brain. In particular, APP interaction with APP-BP1 may be abnormal in the disease. This may take the form of enhanced APP–BP1 interaction with hUba3 or inhibition of APP–BP1 interaction with ASPP2. Augmentation of ASPP2 interaction with APP-BP1 may be a potential therapeutic target for AD.
We thank Dr Donna McPhie for critical reading of the text, and Lorraine Ng and Sean Bradley for technical assistance. This work was supported by NIH grant AG12954 to RLN. YC is a Rappaport Mental Research Scholar from the Rappaport Charitable Foundation and an American Health Assistance Foundation Investigator for Alzheimer's Disease Research.