Essential Requirement of Apolipoprotein J (Clusterin) Signaling for IκB Expression and Regulation of NF-κB Activity*

Apolipoprotein J/clusterin is an enigmatic protein highly regulated in inflammation, apoptosis, and cancer. Despite extensive studies, its biological function has remained obscure. Here we show that apolipoprotein J inhibits neuroblastoma cell invasion. Since this function can be regulated by NF-κB, we explored the possibility that apolipoprotein J might interfere with NF-κB signaling. Ectopic apolipoprotein J expression strongly inhibited NF-κB activity in human neuroblastoma cells and murine embryonic fibroblasts by stabilizing inhibitors of NF-κB (IκBs). Steady state levels of IκB proteins are drastically reduced in mouse embryo fibroblasts after disruption of the apolipoprotein J gene. Absence of apolipoprotein J causes reduction of IκB stability, a tumor necrosis factor-dependent increase in NF-κB activity, increased transcription of the NF-κB target gene c-IAP and down-modulation of p53 protein. These results suggest that an unexpected physiological role of apolipoprotein J is to inhibit NF-κB signaling through stabilization of IκBs and that this activity may result in suppression of tumor cell motility.

Apoliprotein J/clusterin (apoJ) 1 is an ubiquitously expressed secreted protein whose presumed functions include cell-cell interactions, inhibition of complement-mediated cytolysis, and chaperone activity (1,2). The product of the apoJ gene consists of a predominant form of about 75 kDa, which is secreted in the extracellular spaces and body fluids. The 449-amino acid primary polypeptide chain of human apoJ is proteolytically cleaved into the ␣ and ␤ chains. Five disulfide bridges link the two subunits to obtain the mature form of the protein. ApoJ is highly conserved in different species, showing a 70 -80% protein homology in mammals (1). Its expression is low in normal conditions but is greatly induced by oxidative, thermal, and apoptotic stimuli, suggesting that apoJ function could be di-rectly or indirectly related to the cell suicide program and/or the stress response.
The biological role of apoJ is still controversial. Several studies have suggested that apoJ has antiapoptotic activity, protecting cultured cell lines against a variety of stress signals. Exogenously supplied apoJ protects tumor cells from cytokineor drug-induced apoptosis and inhibition of apoJ results in the increased sensitivity of cancer cells to chemotherapeutic drugs (3)(4)(5)(6). However, a nuclear apoJ form has been described which marks cells for apoptosis through its association with the DNArepair protein KU70 (7).
The generation of knock-out mice have complicated the picture. Mouse development is not affected by the absence of apoJ (8). However, apoJ null mice show increased sensitivity to autoiimune myocarditis, suggesting a role for apoJ in protecting the heart tissue from postinflammatory destruction (8). In contrast, in the absence of apoJ, mice are partially protected after hypoxic injury, suggesting that it can have a negative role in neuronal survival (9).
The prevailing hypothesis on the biological role of clusterin suggests that it is involved in the clearance of toxic substances from extracellular spaces through its ability to bind to unfolded proteins, cell debris, or immune complexes. ApoJ binds to the endocytic receptor Megalin (LRP-2), and the clearance effect would be achieved by internalization of the receptor-ligand complex and lysosomal degradation of the toxic substances (1,2). ApoJ has been shown to be highly regulated during tumor progression and to be a B-MYB oncoprotein target gene (10,5). In this study we have investigated whether apoJ expression may change tumor cell characteristics by impinging on specific signal transduction pathways.

Cell Cultures and in Vitro Invasion
Assay-The human neuroblastoma cell line LAN5 (11), the amphotropic-packaging cell line Phoenix A (ATCC SD3444), and early passage mouse embryonic fibroblasts (MEFs) were maintained in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fetal calf serum, 2 mM glutamine, and 2 mM penicillin-streptomycin (Invitrogen). For in vitro invasion assays 5 ϫ 10 5 cells were resuspended in 1 ml of medium containing no serum and seeded on top of the invasion assay chamber. The bottom chamber was filled with medium containing 5% fetal calf serum that served as chemo-attractant. After 24 h, cells were collected in the bottom chamber and scored as described in the manufacturer's instructions (BD Biosciences). Retroviral infection of LAN5 and MEFs cells was carried out as described previously (12).
Plasmids-The complete apolipoprotein J cDNA was amplified by reverse transcription (RT)-PCR from LAN5 cells. Correct amplification was assessed by sequencing. The blunt PCR product was cloned into HpaI-digested MIGR1 retroviral vector (12). The apoJ cDNA was cloned into BamHI, EcoRV-digested pCDNA3 to obtain pCDNA3-APOJ. The NF-B luciferase reporter plasmid pNFB-LUC, containing four copies of the NF-B consensus sequence, is from Clontech. A c-Jun-responsive * This work was partially supported by an Institute of Child Health pump-priming grant. 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. reporter vector, containing AP1-binding sites from the c-Jun promoter cloned upstream the luciferase gene, was obtained from Dr. J. Ham. The mutant IB superrepressor cDNA was obtained by mutagenesis of the RT-PCR-amplified product, which was subcloned into the MIGR1 vector. Expression of the construct was verified by WB analysis with IB ␣ antibody, and its inhibitory function was verified by its ability to suppress activity of the pNFB-LUC plasmid in luciferase assays. The retroviral vector encoding for the constitutively active IKK kinase (IK-Kee) was published previously (13).
Luciferase Assays-MEFs and LAN5 cells were transfected with LipofectAMINE 2000 (Invitrogen) with 0.5 g of the pNFB-LUC reporter plasmid in combination with 0.5 g of empty pCDNA3 or pCDNA3-APOJ in 35-mm wells. After 18 h, cells were exposed to 0.5 g/ml doxorubicin or 50 ng/ml of tumor necrosis factor (TNF) for 8 h, harvested, and monitored for luciferase activity with a dual-luciferase assay kit, following the manufacturer's instructions (Promega). Light emission was evaluated with the aid of a luminometer and expressed as light units, which were normalized with a Renilla luciferase reference plasmid.
For semiquantitative RT-PCR, preliminary experiments were carried out to determine the linear range of amplification. c-IAP-1 primers were the following: forward primer, 5Ј-GGACATTAGGAGTCTTCCC; reverse primer, 5Ј-GCTGGGAGTGACAGTGAAC.
Protein Stability and in Vitro Kinase Assays-For protein stability experiments, 293 cells were transfected with the plasmid encoding for the IB mutant (Ser 32/36 ) and incubated with cycloheximide (25 g/ml) for the indicated times with or without protease inhibitor. Protease inhibitors were dissolved in Me 2 SO, which was added (0.2%) to untreated cells as a control. We have assessed IB mutant expression by Western blotting with a hemagglutinin-tag antibody. The in vitro kinase assay was performed as described previously (14).

ApoJ Negatively Regulates Neuroblastoma Cell Invasion-
We wanted to determine whether ectopic apoJ expression interferes with the invasive behavior of neuroblastoma cells. Since human neuroblastoma cells are difficult to transiently transfect, we subcloned the apoJ cDNA into the MIGR1 retroviral vector and we transduced LAN5 cells with the apoJexpressing, or control, retroviruses. The MIGR1 vector contains the enhanced GFP gene and allows simultaneous expression of the inserted cDNA through the presence of an IRES sequence (l2). GFP-positive cells were isolated by FACS sorting and subjected to in vitro invasion assays. ApoJ overexpression drastically reduced the number of cells migrating in the bottom of the well, suggesting that it may suppress cell invasion (Fig. 1A). Cells transduced with a dominant-negative IB (inhibitor of )-␣ molecule, which behaves as a NF-B super-repressor (15), were less prone to invade the matrigel substrate, suggesting that, as in other tumor models (16), NF-B activity is important for neuroblastoma cell-invasive properties (Fig. 1B).
Consistent with this hypothesis, a constitutively active IKK kinase (IKKee), which induces degradation of inhibitors of B (IB) and activation of NF-B, significantly promotes invasion of LAN5 cells in vitro (Fig. 1C). Notably, concurrent apoJ expression drastically reduced the number of IKKee-transfected cells migrating through the matrigel membrane, suggesting that apoJ activity impinges on the NF-B pathway (Fig. 1C). To assess whether apoJ could directly interfere with the IKK kinase, we carried out in vitro kinase assays, which established that apoJ does not inhibit IKK kinase activity (Fig.  1D). The conclusion drawn from these experiment is (a) NF-B is required for in vitro invasion of neuroblastoma cells and (b) apoJ inhibits this function downstream the IKK kinase.
ApoJ Specifically Inhibits NF-B Activity and Enhances Stability of Inhibitors of NF-B in Cancer Cell Lines-To directly assess whether ectopic apoJ expression can modulate NF-B activity, we transfected LAN-5 cells with a CMV-driven apoJ cDNA together with a NF-B-responsive promoter linked to luciferase. Basal or doxorubicin-induced NF-B activity is strongly inhibited in LAN5 cells transfected with the apoJ expression vector (Fig. 2A). Similar results were obtained with another neuroblastoma cell line, SHSY5Y (not shown). The experiments described in the above section (Fig. 1) insinuate that apoJ biological effects could be achieved, at least in part, by interfering with factors downstream the IKK kinase. Indeed, we found that apoJ expression enhances IB-␣/␤, but not tubulin, levels and that doxorubicin down-modulates IB expression more efficiently in control than in apoJ-transduced cells (Fig. 2B). IB proteins play a key role in maintaining the NF-B molecule in an inactive state (17,18).
Interestingly, extrinsic expression of apoJ prolongs the halflife of the IB Ser 32/36 mutant (IB-M, used in the experiment described in the legend to Fig. 1B), which is refractory to signal-dependent (i.e. IKK-dependent) degradation (Fig. 2C). Both signal-dependent and -independent IB degradation can occur through IB ubiquitination, which is followed by proteasome degradation (19). Accordingly, we found that the proteasome inhibitor MG132 and apoJ, but not the calpain inhibitor calpeptin, prolongs ectopically expressed IB-M half-life. (Fig.  2D). These results suggest that apoJ modulates both signal-dependent and -independent IB turnover downstream the IKK kinase.
To assess the specificity of the apoJ-suppressing effect on NF-B activity, we transfected neuroblastoma cells with NFB-or AP-1-responsive promoters, with or without the apoJ expression vector, and carried out luciferase assays. ApoJ induced a ϳ 70% inhibition of the NF-B-responsive promoter, but it decreased only slightly Jun promoter activity, suggesting that apoJ significantly suppresses NFB-dependent, but not AP-1-dependent, transactivation (Fig. 2, E and F). Deletion of the apoJ amino terminus, containing the protein export signal, did not change its activity, indicating that cytoplasmic, but not secreted, apoJ is involved in suppression of NF-B activity (data not shown).
ApoJ Knock-out Causes Enhanced NF-B Response after Genotoxic Stress in Primary Fibroblasts-To explore the role of apoJ signaling in a normal cell context, we assessed NF-kB activity in embryonic fibroblasts (MEFs) from normal or apoJ knock-out mice. Basal activity of the NF-B reporter plasmid was higher in apoJ knock-out MEFs compared with control cells (3.6 Ϯ 0.65 versus 6.8 Ϯ 1.2, Fig. 3). Notably, apoJ null, but not control, MEFs show a peak of NF-B activity after exposure to doxorubicin (Fig. 3, A and B). Reconstitution of apoJ expression results in suppression of the NF-B response in doxorubicin-treated apoJ null MEFs (Fig. 3B). Expression of endogenous apoJ in primary fibroblasts was detectable in control but not in apoJ knock-out MEFs, as expected (Fig. 3C).
Reduced IB Protein Levels in ApoJ Knock-out Fibroblasts-A classical signaling pathway that leads to NF-B activation is that triggered by the tumor necrosis factor receptor. IB-␤, and to a lesser extent IB-␣, protein levels are significantly reduced in fibroblasts from apoJ knock-out embryos in the presence of TNF stimulus (Fig. 4A). Interestingly, IB ␤ protein expression is drastically decreased in apoJ knock-out MEFs, even in the absence of TNF (Fig. 4A). While protease inhibitors did not modulate IB-␤ levels in normal fibroblasts, the proteasome inhibitor MG132, but not the calpain inhibitor calpeptin, rescued IB protein levels in apoJ knock-out cells (Fig. 4B). This experiment demonstrates that, even in the absence of specific stimuli, IB-␤ is more susceptible to proteasomedependent degradation in the apoJ knock-out background, which probably justifies the higher basal NF-B activity observed in apoJ-null cells. Northern analysis shows that IB-␤ mRNA is robustly expressed in apoJ knock-out cells, indicating that endogenous apoJ expression does not affect transcription of the IB-␤ gene (Fig. 4C). Reconstitution of apoJ expression restores, at least in part, IB-␤ protein levels in apoJ knock-out cells (Fig. 4D, compare lanes 2 and 3), confirming that differences in IB expression are related to the absence of apoJ and not due to clonal variations. Overall these results indicate that endogenous levels of apoJ are critically required for steady state IB-␤, and to a less extent IB-␣, protein levels.

NF-B Target Genes Expression Is Altered in ApoJ
Knockout Cells-An important prediction drawn from these experiments is that NF-B target genes expression should be modulated in the absence of apoJ. We have verified this hypothesis by transiently transfecting apoJ knock-out, or control, MEFs with the NF-B reporter vector in the presence of TNF. TNF-dependent, as well as independent, NF-B activity is substantially increased in apoJ knock-out cells, as compared with wild type fibroblasts (Fig. 5A, bottom). We furthered this finding by examining mRNA levels of the NF-B-target gene c-IAP (20). In comparison with wild type cells, c-IAP-1 mRNA levels were increased in both TNFtreated, or untreated, apoJ knock-out fibroblasts (Fig. 5A,  upper side). We have repeated this experiment with two more independent MEF preparations with similar results. It has been shown, by Verma and co-workers (21), that p53 protein stability is negatively regulated by endogenous levels of NF-B and that p53 levels are decreased in mouse embryo fibroblasts with activated NF-B. In agreement with this observation, we found that p53 protein levels are significantly decreased in apoJ knock-out fibroblasts (Fig. 5B). Albeit basal expression of p53 protein is reduced in apoJ knockout fibroblasts, it is detectable after longer Western blot exposures, and it is induced by chemotherapeutic drugs treatment (data not shown). Overall these experiments demonstrate that the gene and protein expression patterns of apolipoprotein J knock-out fibroblasts are consistent with that of cells with activated NF-B. DISCUSSION NF-B is a transcription factor that plays a pivotal role in numerous cellular processes and is constitutively expressed in many cancer cell types. Its activity is regulated through inter- action with inhibitory molecules (IBs) that modulate NF-B function by sequestering it in inactive cellular sites or by suppressing its transcriptional activity (17,18). IBs, are, in turn, subjected to phosphorylation by kinases (IKKs) that mark for their destruction, or inactivation, resulting in the release of active NF-B (17,18). Modulation of NF-B activity is though to be important for cell survival, stress response, immunity, cell motility, proliferation, and transformation (16 -18, 20).
In this investigation we have reported the crucial role of apoJ in controlling IB expression and NF-B activity in normal and tumorigenic cells. Regulation of NF-B activity could satisfactorily explain apoJ opposing effects in different systems. Heart damage is exacerbated in apoJ-deficient mice subjected to experimental autoimmune myocarditis (8). ApoJ expression could be required to suppress excessive NF-B activation, which determines inflammation and cardiomyocyte apoptosis in autoimmune myocarditis in the rat (22). This hypothesis is also suggested by our observation that disruption of the apoJ gene results in increased NF-B response evoked by the inflammatory cytokine TNF (Fig. 5). In contrast, apoJ contributes to brain injury in a study of experimental hypoxia-ischemia, where it has been shown that exogenous clusterin increases cortical neuron death induced by oxygen/glucose deprivation (9). Since NF-B activation is required for the survival of cortical neurons subjected to ischemia (23), suppression of NF-B function could be at the base of clusterin neurotoxicity.
Loss of apoJ expression in cells that depend on NF-B activity for chemoresistance, proliferation, or invasion could lead to tumor progression. In this regard, we show that acutely expressed apoJ can drastically reduce the in vitro invasive properties of neuroblastoma cells (Fig. 1). Consistent with our results, Sivamurthy et al. (24) have recently shown that exogenously supplied apoJ inhibited vascular smooth muscle cell migration in microchemotaxis chambers. TNF-induced vascular smooth cell migration depends on NF-B-mediated regulation of key cytokines (25). In preliminary experiments we have observed that apoJ suppresses the metastatic phenotype of neuroblastoma xenografts in immunocompromised mice. 2 The apoJ gene chromosomal location is at 8p21, a region 2 G. Santilli, J. Anderson, and A. Sala, manuscript in preparation.

FIG. 4.
Endogenous apoJ is critically required for IB-␤ protein stability in MEF fibroblasts. A, two independent primary fibroblast cell lines from wild type (indicated by the ϩ sign) or apoJ knockout (indicated by the Ϫ sign) mice were cultured without or with 50 ng/ml of TNF for the indicated times. Cell lysates were subjected to Western blot analysis with IB-␣ and -␤ antibodies. Equal loading of the lanes was assessed by stripping and reprobing with a ␤-actin antibody. B, wild type (ϩ/ϩ) or apoJ knock-out (Ϫ/Ϫ) MEFs were treated with the indicated protease inhibitors or 0.2% Me 2 SO (DMSO) and harvested 8 h later for Western blot analysis with IB-␤ or actin antibodies. C, total cellular RNA was extracted from wild type or apoJ knock-out MEFs and subjected to Northern blot analysis with a IB-␤ probe. Equal loading of the lanes was assessed by reprobing the blot with GAPDH. 20 g of RNA were loaded per each lane. This experiment was repeated with other independent MEF lines with similar results. D, for reconstitution experiments, MEFs were infected with the control (lanes 1 and 2) or MIGR1-APOJ vector (lane 3). After cell sorting, GFP-positive cells were subjected to Western blot analysis with the indicated antibodies. Note that the apoJ antibody detects only exogenous (human) apoJ. This experiment was repeated twice.

FIG. 5. TNF induces increased NF-B response and transcription of c-IAP in apoJ knock-out MEFs.
A, upper panel, semiquantitative PCR analysis was performed after mixing 1 ⁄10 of the reverse transcription products with the PCR reaction mixtures. Linear ranges of the reaction were determined empirically by analyzing the PCR products after different cycle steps. Cycles numbers in the figure refer to amplification of c-IAP. GAPDH cycles were 20 and 25, respectively. Lanes 1, control MEFs; lanes 2, control MEFs ϩ TNF; lanes 3, apoJ knock-out MEFs; lanes 4, apoJ knock-out MEFs ϩ TNF. Bottom panel histogram, MEFs from control (ϩ/ϩ) or apoJ knockout (Ϫ/Ϫ) mice were transiently transfected with the NF-B luciferase reporter plasmid as described in the legend to Fig. 2. The next day, cells were treated with 50 ng/ml of TNF and harvested after 8 h for luciferase assays. Error bars indicate S.D. values. This experiment was repeated twice with independent MEF preparations. B, p53 levels were detecetd by Western blot analysis of exponentially growing wild type (ϩ/ϩ) or apoJ knockout (Ϫ/Ϫ) MEFs. Similar results were obtained with other MEFs preparations.
commonly deleted in human cancers. We propose that apoJ could be a tumor suppressor protein required to control tumorigenic signals emanating from the NF-B pathway. This could explain why advanced testicular tumors show greatly reduced expression of apoJ compared with lower grade tumors or normal tissue (26). In other cancer types, apoJ expression could be advantageous through reduced expression of NF-B-regulated death receptors and increased resistance to genotoxin-induced apoptosis, as reported previously (27). Deletion of the protein export signal does not change the ability of apoJ to regulate NF-B (data not shown), suggesting that this biological function depends on intracellular apoJ. How apoJ mechanistically regulates IB protein turnover needs to be clarified. We have determined that apoJ does not directly bind to IB ␣ or ␤ (data not shown). The stabilization effect could be achieved by direct regulation of IB-interacting proteins or relevant ubiquitin ligases. In this regard, a recently described new class of RASlike proteins suppress NF-B function by directly binding to IB molecules, resulting in their stabilization (14). Verification of these hypotheses will be the subject for future investigations.