Activation of the Nuclear Factor κB Pathway by Astrocyte Elevated Gene-1: Implications for Tumor Progression and Metastasis

We recently reported the full-length cloning and functional characterization of an HIV-1-inducible gene, astrocyte elevated gene-1 (AEG-1), which displays slow induction kinetics following HIV infection or gp120 treatment (1–3). AEG-1 mRNA is expressed ubiquitously with higher expression in tissues containing muscular actin and its expression is increased in astrocytes infected with HIV-1 or treated with gp120 or tumor necrosis factor α (TNF-α; ref. 3). The mRNA encodes a single pass transmembrane protein of predicted molecular mass of 64 kDa and pI 9.3 that predominantly localizes in the endoplasmic reticulum and perinuclear region. Intriguingly, AEG-1 expression is elevated in subsets of breast carcinomas, malignant gliomas, and melanomas, and it synergizes with oncogenic Ha-ras to enhance soft agar colony forming ability of nontumorigenic immortalized melanocytes (3). In addition, AEG-1 expression is elevated in adult astrocytes transformed by sequential overexpression of SV40 T/t antigen, telomerase (hTERT), and T24 Ha-ras, thereby displaying an aggressive glioma-like phenotype (3). Although AEG-1 expression is not increased in all cancer cells, these results strongly suggest that AEG-1, in specific cellular contexts, may play a positive role in promoting cancer development and/or maintenance.

The transcription factor nuclear factor κB (NF-κB) regulates the expression of a wide variety of genes involved in cellular events such as inflammation, immune response, proliferation, and apoptosis (4). There are five members of the NF-κB family in mammals, including Rel (c-Rel), RelA (p65), RelB, NF-κB1 (p50 and its precursor p105), and NF-κB2 (p52 and its precursor p100; refs. 5, 6). The most abundant activated form of NF-κB is a heterodimer composed of a p50 and p65 subunit that functions predominantly as a transcriptional activator. NF-κB is present in most cells, where it remains in an inactive form in the cytoplasm bound with an inhibitory protein, IκB, which masks the nuclear localization signal. On stimulation by various NF-κB activating signals, IκB is phosphorylated and degraded through a ubiquitin-dependent process. This process frees NF-κB complex, which then translocates to the nucleus to activate transcription of downstream genes. Several recent studies indicate that NF-κB is activated by various carcinogens and tumor promoters. These include benzo[a]pyrene, 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone, UV radiation, ionizing radiation, and phorbol esters (7–13). For instance, UV radiation has been shown to cause sunburn reactions (swelling, leukocyte infiltration, epidermal hyperplasia, and accumulation of proinflammatory cytokines) leading to skin cancer, and suppression of NF-κB blocks the sunburn-induced damage (14). The oncogenic role of NF-κB in solid cancer has also been described in breast, colon, ovarian, thyroid, and pancreatic cancer (15–17).

TNF-α is one of the major proinflammatory cytokines and it plays an essential role in the pathogenesis of HIV-1-associated dementia by its ability to modulate the neurotransmitter glutamate-induced excitotoxicity to neurons. The excitatory amino acid transporter 2 (EAAT2) prevents glutamate toxicity by removing excess glutamate from the synaptic cleft. Treatment with TNF-α down-regulates EAAT2 mRNA and protein and also inhibits EAAT2-Prom activity in primary human fetal astrocytes (18, 19). The primary pathway by which TNF-α suppresses EAAT2 expression is by activating NF-κB (19, 20) that binds to the EAAT2 promoter and inhibits its activity. Interestingly, ectopic expression of AEG-1 also inhibits EAAT2 promoter activity. The ability of AEG-1 to promote cell transformation and inhibit EAAT2 promoter activity and the involvement of NF-κB in both of these processes prompted us to hypothesize that AEG-1 might also exert its effects by activating NF-κB.

In the present study, we report that AEG-1 markedly increases binding of the transcriptional activator p50/p65 complex of NF-κB. Inhibition of NF-κB by mt32IκBα superrepressor (IκBα-mt32) inhibits AEG-1-induced increased agar cloning efficiency and increased matrigel invasion of HeLa cells. A human NF-κB signaling pathway gene array reveals that Ad.AEG-1 infection specifically induces the expression of NF-κB dependent genes, especially the adhesion molecules involved in tumor metastasis. Furthermore, we show that TNF-α treatment results in nuclear translocation of AEG-1 and p65 wherein they physically interact. Taken together, these data suggest that AEG-1 is critically involved as a positive regulator of NF-κB-induced gene expression, and this activation may provide a molecular basis for the ability of AEG-1 to promote anchorage-independent growth and invasion, prominent contributors to cancer progression.

Cell lines, reagents, and virus infection protocol. The human cervical carcinoma cell line HeLa was cultured as previously described (21). The recombinant replication-incompetent adenovirus expressing AEG-1 (Ad.AEG-1) and the mt32IκBα superrepressor (Ad.IκBα-mt32; ref. 22) was created in two steps as previously described and plaque purified by standard procedures (23). These viruses were constructed as described in ref. 23 by the Massey Cancer Center Virus Vector Shared Resource. Cells were infected with a multiplicity of infection (m.o.i.) of 50 plaque-forming units (pfu)/cell of Ad.vec (control replication-incompetent adenovirus) or Ad.AEG-1 as previously described (23).

Transient transfection and luciferase assay. Cells (2 × 10⁵ per well in 12-well plates) were either uninfected or infected with Ad.vec or Ad.AEG-1 at an m.o.i. of 50 pfu/cell and either untreated or treated with 10 ng/mL of TNF-α. Transient transfection was done 12 hours after infection using LipofectAMINE 2000 transfection reagent (Invitrogen, Carlsbad, CA) and 1.2 μg of plasmid DNA per well that included 1 μg of pGL3Basic, 3κB-Luc, or 3κBmut-Luc plasmids (21) and 0.2 μg of β-galactosidase expression plasmid (pSV40-β-gal; Promega, Madison, WI). Luciferase assays were done 48 hours after transfection using a Luciferase Reporter Gene Assay kit (Promega) according to the protocol of the manufacturer. The β-galactosidase activity was determined using the Galacto-Light Plus kit (Tropix, Bedford, MA). Luciferase activity was normalized by β-galactosidase activity and the data presented are the fold activation relative to pGL3Basic from triplicate determinations.

Cell fractionation and electrophoretic mobility shift assay. Cells were harvested and the cytoplasm and nucleus were fractionated by the modified Schreiber's method as previously described (21). Electrophoretic mobility shift assay (EMSA) using the nuclear extracts was done as previously described (21). The sequences of the consensus and mutated NF-κB probes are 5′-AGTTGAGGGGACTTTCCCAGGC-3′ and 5′-AGTTGAGGCGACTTTCCCAGGC-3′, respectively (Santa Cruz Biotechnology, Inc., Santa Cruz, CA). The antibodies used for supershift analysis were anti-p50 and anti-p65 (rabbit polyclonal antibodies; Santa Cruz Biotechnology).

Reverse transcription-PCR. Total RNA was extracted from cells using the RNeasy mini kit (Qiagen, Hilden, Germany) according to the protocol of the manufacturer. Two micrograms of total RNA were used for reverse transcription-PCR (RT-PCR) according to standard methods. The primers used were as follows: interleukin (IL)-8 sense, 5′-GGTGCAGAGGGTTGTGGAGAA-3′; IL-8 antisense, 5′-GCAGACTAGGGTTGCCAGATT-3′; glyceraldehyde-3-phosphate dehydrogenase (GAPDH) sense, 5′-ATGGGGAAGGTGAAGGTCGGAGTC-3′; and GAPDH antisense, 5′-GCTGATGATCTTGAGGCTGTTGTC-3′.

Western blot analysis. Whole-cell lysates and cytoplasmic as well as nuclear extracts were prepared and Western blotting was done as previously described (21). The primary antibodies used were anti-p65, anti-IκBα (rabbit polyclonal antibody; 1:250; Santa Cruz Biotechnology), anti-EF1α (mouse monoclonal antibody; 1:1,000; Upstate Biotechnology, Lake Placid, NY), and anti-AEG-1 (chicken; 1:5,000).

Immunofluorescence microscopy. Cells cultured on two-well chamber slide were infected with Ad.vec and Ad.AEG-1 at an m.o.i. of 50 pfu/cell. Forty-eight hours later, cells were fixed with 4% formaldehyde in PBS. Double immunofluorescence staining was done with anti-AEG-1 and anti-p65 antibodies as described (23) and images were taken with a confocal laser scanning microscope LSM multiphoton 510 (Zeiss, Thornwood, NY). For TNF-α experiments, cells were treated with TNF-α (10 ng/mL) for 30 minutes and then immunofluorescence staining was done as described.

Anchorage-independent growth assay. Cells were infected with Ad.vec, Ad.AEG-1, Ad.IκBα-mt32 alone (50 pfu/cell), or a combination of Ad.AEG-1 and Ad.IκBα-mt32 (50 + 50 pfu/cell). Anchorage-independent growth assays were done by seeding 1 × 10⁵ cells in 0.4% Noble agar on a 0.8% agar base layer, both of which contained growth medium (23). Colonies were counted 2 weeks after seeding and the data from triplicate determinations were expressed as mean ± SD.

Invasion assay. Invasive ability of the cells infected with Ad.vec, Ad.AEG-1 (50 pfu/cell), or a combination of Ad.AEG-1 and Ad.IκBα-mt32 (50 + 50 pfu/cell) in vitro was determined as previously described (23). Invasion was measured by using 24-well BioCoat cell culture inserts with an 8-μ-porosity polyethylene terepthylate membrane coated with Matrigel basement membrane matrix (100 μg/cm²). Briefly, the Matrigel was allowed to rehydrate for 2 hours at room temperature by adding warm, serum-free DMEM. The wells of the lower chamber were filled with DMEM containing 5% fetal bovine serum. Cells (5 × 10⁴) were seeded in the upper compartment (6.25-mm membrane size) in serum-free DMEM. The invasion assay was done at 37°C in a 5% CO₂ humidified incubator for 48 hours. At the end of the invasion assay, the filters were removed, fixed, and stained with the RAL 555 staining kit (23). Cells on the upper surface of the filters were removed by wiping with a cotton swab and invasion was determined by counting of the cells that had migrated to the lower side of the filter with a microscope at ×100 magnification. Experiments were assayed in triplicate and at least 10 fields were counted in each experiment.

Human NF-κB signaling pathway gene array. The expression levels of NF-κB signaling pathway genes after Ad.vec and Ad.AEG-1 infection were analyzed in total RNA using the Human NF-κB signaling gene array (SuperArray Biosciences Corporation, Frederick, MD) according to the protocol of the manufacturer.

Immunoprecipitation. Cells were lysed in immunoprecipitation buffer [50 mmol/L Tris-HCl (pH 7.5), 150 mmol/L NaCl, and 0.1% NP40]. After centrifugation, the supernatant was incubated with anti-hemagglutinin (HA) antibody (Covance Research Products, Inc., Berkeley, CA), anti-p65 antibody (Santa Cruz Biotechnology), or anti-AEG-1 antibody for at least 1 hour. Immunocomplexes were recovered by adsorption to protein G-Sepharose (Amersham Biosciences, Piscataway, NJ). After being washed four times in immunoprecipitation buffer, the immunoprecipitates were analyzed by immunoblotting.

Statistical analysis. Statistical analysis was done using one-way ANOVA, followed by Fisher's protected least significant difference analysis. P < 0.05 was considered significant.

Ad.AEG-1 infection generates AEG-1 protein. To confirm the authenticity of the Ad.AEG-1 construct, the expression of AEG-1 protein in HeLa cells was determined by Western blot analysis 2 days after infection with Ad.AEG-1 at an m.o.i. of 50 pfu/cell. As shown in Fig. 1A, under basal condition, expression of AEG-1 could be detected in HeLa cells and infection with Ad.AEG-1 significantly increased AEG-1 level in these cells. We have previously reported that TNF-α up-regulates AEG-1 mRNA expression. Here we show that TNF-α treatment (10 ng/mL for 4 days) also up-regulates AEG-1 expression at the protein level (Fig. 1A).

Endogenous AEG-1 is localized mainly at the perinuclear region and in endoplasmic reticulum in immortalized primary human fetal astrocytes. To examine the consequences of overexpression of AEG-1 on the localization of AEG-1 protein, immunofluorescence microscopy using anti-AEG-1 antibody was done in parental HeLa cells and HeLa cells infected with Ad.AEG-1. As shown in Fig. 1B, subcellular localization of endogenous AEG-1 in HeLa cells was similar to that in immortalized primary human fetal astrocytes. However, forced overexpression of AEG-1 resulted in the localization of the protein both in the cytoplasm and in the nucleus in HeLa cells.

Infection with Ad.AEG-1 activates the NF-κB pathway. AEG-1 is overexpressed in multiple cancers and it cooperates with ras to promote transformation of melanocytes (3). AEG-1 is also induced by TNF-α treatment (2). In these contexts, we hypothesized that AEG-1 might activate the NF-κB pathway, a known TNF-α-induced signaling cascade that is activated in multiple tumors. To test whether Ad.AEG-1 infection activates the NF-κB pathway, HeLa cells were either uninfected or infected with Ad.vec or Ad.AEG-1 and then transfected with either empty vector (pGL3Basic), 3κB-Luc, containing three tandem NFκB binding sites upstream of the luciferase gene, or 3κBmut-Luc, containing mutated NF-κB binding sites (21), and luciferase activity was analyzed. The luciferase activity was normalized by β-galactosidase activity, which was further normalized by the activity obtained by transfection with pGL3Basic. Cells transfected with pGL3Basic showed only basal luciferase activity under any experimental condition (data not shown). In control and Ad.vec-infected cells, transfection of 3κB-Luc increased basal activity over transfection of either pGL3Basic or 3κBmut-Luc ∼5-fold, which is most likely a consequence of constitutive NF-κB DNA binding activity in HeLa cells (Fig. 1C). However, infection with Ad.AEG-1 resulted in ∼15-fold induction in relative luciferase activity of 3κB-Luc over 3κBmut-Luc (Fig. 1C). As a positive control, TNF-α treatment also induced 3κB-Luc activity ∼10-fold over 3κBmut-Luc.

NF-κB DNA binding following Ad.AEG-1 infection was analyzed by EMSA using radiolabeled consensus NF-κB binding site as a probe and nuclear extracts from HeLa cells. As shown in Fig. 1D, in uninfected and untreated cells, two shifted bands were observed. After infection with Ad.AEG-1 or treatment with TNF-α, the fast-migrating band disappeared and a marked increase in the intensity of the slow migrating band was apparent. To determine the composition of the DNA-protein complexes observed in EMSA, supershift experiments were done in which DNA-binding reactions were incubated with antibodies directed against NF-κB subunits p50 and p65. The anti-p50 antibody supershifted both the slow- and fast-migrating bands whereas anti-p65 antibody supershifted only the slow-migrating band. These findings show that under basal condition, both p50/p50 homodimer and p50/p65 heterodimer bind to the NF-κB probe. However, on Ad.AEG-1 infection and TNF-α treatment, the binding pattern changed, with the p50/p50 homodimer disappearing and the binding of the p50/p65 heterodimer increasing markedly. These findings indicate that AEG-1 is a potent activator of NF-κB because it promotes increased binding of the potent transcriptional activator p50/p65.

Overexpression of AEG-1 up-regulates NF-κB downstream genes. To confirm the functional end point of NF-κB activation by AEG-1, expression of mRNA of IL-8, a well-known NF-κB target gene, was analyzed by RT-PCR analysis after Ad.AEG-1 infection or TNF-α treatment. A significant increase in the expression of IL-8 mRNA (Fig. 2A) was detected in Ad.AEG-1-infected and TNF-α-treated HeLa cells in comparison with untreated Ad.vec-infected cells.

To extend these findings, expression of 113 genes related to NF-κB-mediated signal transduction was analyzed using a NF-κB signaling pathway microarray (SuperArray Biosciences) in Ad.vec and Ad.AEG-1-infected HeLa cells 2 days after infection. Ad.AEG-1 infection significantly induced the expression of several NF-κB downstream genes such as intercellular adhesion molecule (ICAM)-3, ICAM-2, selectin E, selectin P ligand, selectin L, toll-like receptor (TLR)-4, TLR-5, FOS, JUN, and IL-8 (Fig. 2B). These findings indicate that activation of NF-κB pathway by AEG-1 translates into actual changes in NF-κB-responsive gene expression.

AEG-1 activates NF-κB via IκBα degradation and p65 translocation. Classically, NF-κB is activated following the degradation of the cytoplasmic inhibitor protein IκBα that binds to the NF-κB complex. Degradation of IκBα releases NF-κB, which relocates to the nucleus to bind promoter elements and activates transcription. To determine whether AEG-1-induced NF-κB activation also proceeds through a similar pathway, the expression of IκBα was analyzed by Western blot analysis. The level of IκBα decreased significantly 48 hours after Ad.AEG-1 infection or after TNF-α treatment in comparison with Ad.vec infection in the cytoplasmic extracts of HeLa cells (Fig. 3A). In contrast, the level of the housekeeping protein EF1-α remained unchanged under all experimental conditions.

The level of the p65 subunit of NF-κB was analyzed in cytoplasmic and nuclear extracts after Ad.AEG-1 infection and TNF-α treatment. As shown in Fig. 3B, the levels of p65 protein decreased in cytoplasmic extract and increased in the nuclear extract of cells after Ad.AEG-1 infection and TNF-α treatment in comparison with Ad.vec infection. These findings indicate that Ad.AEG-1 infection and TNF-α treatment resulted in translocation of p65 from the cytoplasm to the nucleus. Immunofluorescence studies using Anti-p65 antibody confirmed the nuclear translocation of p65 following Ad.AEG-1 infection and TNF-α treatment (Fig. 3C). Whereas high cytoplasmic and little nuclear staining for p65 was detected in Ad.vec-infected cells (Fig. 3C,, left), the staining in the nucleus increased significantly in Ad.AEG-1-infected or TNF-α-treated cells with a corresponding decrease in the cytoplasmic staining (Fig. 3C , middle and right, respectively).

Inhibition of AEG-1-induced NF-κB activation blocks AEG-1-mediated anchorage-independent growth and invasion. We next evaluated the functional consequence of AEG-1-induced NF-κB activation. For this purpose, we employed an adenovirus expressing the mt32IκBα superrepressor (IκBα-mt32), Ad.IκBα-mt32, which inhibits IκBα degradation and subsequent NF-κB nuclear translocation (22). As shown in Fig. 4A, infection with Ad.IκBα-mt32 significantly inhibited both basal and Ad.AEG-1-induced NF-κB DNA binding as determined by EMSA. Additionally, infection with Ad.IκBα-mt32 dramatically reduced AEG-1-induced nuclear translocation of p65 (Fig. 4B).

We previously showed that AEG-1 synergizes with oncogenic Ha-ras to enhance soft agar colony-forming ability of nontumorigenic immortalized melanocytes (3). Because NF-κB induces a variety of genes that play an important role in invasion and metastases of cancer cells, we next explored the involvement of the activated NF-κB pathway in AEG-1-induced anchorage-independent growth and invasion. HeLa cells were infected with Ad.AEG-1 and Ad.IκBα-mt32, either alone or in combination, and anchorage-independent growth in soft agar was analyzed. AEG-1-overexpressing cells developed more colonies in soft agar (Fig. 4C) than control or Ad.vec-infected cells. Overexpression of IκBα-mt32 significantly inhibited the AEG-1-induced increase in anchorage-independent growth, indicating a potential involvement of activated NF-κB pathway in this phenomenon.

In correlation with growth in soft agar, overexpression of AEG-1 significantly enhanced the invasive ability of the cells as revealed by matrigel invasion assay (Fig. 5A and B). This increased invasion was inhibited on co-infection with Ad.IκBα-mt32, further strengthening the involvement of the activated NF-κB pathway in AEG-1-induced migration and invasion.

AEG-1 interacts with the p65 component of NF-κB. TNF-α induces nuclear translocation of p65 and up-regulates AEG-1. AEG-1 also induces p65 nuclear translocation and Ad.AEG-1 infection results in the accumulation of AEG-1 protein in the nucleus (Fig. 1B). These findings provided a rationale to investigate whether the localization of AEG-1 is similarly altered by TNF-α treatment. We evaluated the subcellular localization of AEG-1 in HeLa cells by immunofluorescence microscopy with or without TNF-α treatment. As previously reported (3) and as shown in Fig. 6A (middle, top row), AEG-1 localized primarily in the cytoplasm. Following treatment with TNF-α for 30 minutes, a significant level of AEG-1 (Fig. 6A,, middle, bottom row) was observed to accumulate in the nucleus, similar to p65 (Fig. 6A,, left, bottom row). These results suggest that at least a portion of cytoplasmic AEG-1 enters the nucleus in a TNF-α-dependent manner. The nuclear translocation of AEG-1 was further confirmed by Western blot analysis done using cytoplasmic and nuclear extracts of HeLa cells, either untreated or treated with TNF-α, showing an increased level of AEG-1 in the nucleus on TNF-α treatment (Fig. 6B).

Merging of the immunofluorescent images for AEG-1 and p65 showed intense yellow staining in the nucleus on TNF-α treatment (Fig. 6A,, right, bottom row), indicating a potential colocalization of AEG-1 and p65 in the nucleus. These observations prompted us to check whether AEG-1 and p65 interact with each other by immunoprecipitation assay. Immunoprecipitation with anti-AEG-1 antibody did not pull down p65 in untreated cells (Fig. 6C). However, with TNF-α treatment for 30 minutes, a significant level of p65 was immunoprecipitated by anti-AEG-1 antibody, indicating that TNF-α treatment results in nuclear tranlocation of p65 and AEG-1 wherein these two proteins physically associate. These findings also indicate that by interacting with p65, AEG-1 might directly augment the transcriptional activity of the NF-κB complex.

To further confirm the physical interaction between AEG-1 and p65, we transiently transfected HeLa cells with expression constructs encoding HA-AEG-1 and the p65 subunit of NF-κB. Forty-eight hours posttransfection, whole-cell lysates were used in coimmunoprecipitation assays using anti-p65 antibody. After extensive washing, the immunoprecipitates were subjected to SDS-PAGE followed by Western blotting, using anti-HA antibody. Anti-p65 antibody immunoprecipitated both endogenous and exogenous p65 protein (Fig. 6D,, middle). When both p65 and HA-AEG-1 were overexpressed, AEG-1 was coimmunoprecipitated by anti-p65 antibody, indicating that AEG-1 and p65 interact in vivo (Fig. 6D,, top). In addition, using the HA antibody, we successfully immunoprecipitated p65 from AEG-1- and p65-overexpressing cells but not from control cells (Fig. 6D , bottom), further confirming the in vivo interaction of AEG-1 and p65.

We presently show that AEG-1 activates the NF-κB pathway and interacts with the p65 component of NF-κB in the nucleus of HeLa cells. Under basal condition, AEG-1 is a cytoplasmic protein located predominantly in the perinuclear region and in the ER. However, on TNF-α treatment, which up-regulates AEG-1 expression, or following forced overexpression of AEG-1, accumulation of AEG-1 in the nucleus was observed together with translocation of p65 to the nucleus where these two proteins interacted. We did not detect any interaction between p65 and AEG-1 under basal condition when they are located in the cytoplasm, further strengthening the observation that nuclear translocation is necessary to facilitate this interaction. Analysis of the amino acid sequence of AEG-1 reveals the presence of putative, either monopartite or bipartite, nuclear localization signal between amino acids 432 to 451 and amino acids 561 to 580 (Table 1). It is still not clear how AEG-1 is retained in the cytoplasm and how, on TNF-α treatment or on AEG-1 overexpression, the nuclear localization signal is unmasked and the protein translocates into the nucleus. Conceptually, it is possible that AEG-1 interacts with an inhibitory protein like IκBα in the cytoplasm, and AEG-1 itself, as well as TNF-α, might trigger a signaling event that results in the degradation of this interacting protein as well as IκBα that facilitates the unmasking of nuclear localization signal of both AEG-1 and p65. These hypotheses are analogous to the nuclear receptor coactivator SRC-3 (pCIP/ACTR/AIB-1/RAC-3/TRAM-1) that is located mainly in the cytoplasm, translocates from the cytoplasm to the nucleus on TNF-α activation, and enhances NF-κB-mediated gene expression (24). Current studies are in progress to identify AEG-1-interacting molecules to provide a better understanding of the molecular mechanism of AEG-1 action.

A separate group also cloned human AEG-1 as metadherin and the cloning of the mouse (3D3) and rat (lyric) homologues of AEG-1 has been reported. Rat lyric was reported as an overexpressed gene in rat liver and colon tumors (3). As a corollary, human metadherin manifested high expression in breast cancer cells and contains a lung-homing domain that facilitates metastasis of the breast cancer cells to the lungs. We also detected elevated levels of AEG-1 in subsets of breast cancer, glioblastoma multiforme, and melanoma cells, and ectopic expression of AEG-1 promoted colony forming ability of immortalized melanoma cells (3). These findings indicate that in both human and rat, AEG-1 is involved in promoting tumor progression and metastasis, and we now elucidate, for the first time, that NF-κB-dependent gene expression changes play a crucial mechanistic role in mediating augmentation of tumorigenic potential by AEG-1. We show that AEG-1 facilitates IκBα degradation, resulting in an increase in NF-κB DNA binding activity and NF-κB promoter activity in reporter assays. Additionally, we show that inhibition of NF-κB by IκBα-mt32 significantly inhibits AEG-1-induced augmentation in soft agar colony formation and invasion of HeLa cells. Several different tumor cell types, including leukemia, lymphoma, myeloma, melanoma, prostate, colon, breast, pancreas, glioma, and head and neck squamous cell carcinoma cell lines, as well as samples obtained from cancer patients, have been reported to express constitutively active NF-κB (15–17, 25–27). Inhibition of p65 by specific antisense oligonucleotides successfully reduced the growth of cancer cells. It will be interesting to determine whether the constitutive activation of NF-κB in these cells is a result of increased AEG-1 expression and whether inhibition of AEG-1 by an antisense or siRNA strategy might ablate the tumorigenic potential of these cells by inhibiting constitutive NF-κB activation.

NF-κB is a ubiquitously expressed transcription factor for genes involved in cell survival, cell adhesion, inflammation, differentiation, and growth (28–31). In intestinal epithelial cells, the responses to NF-κB activation include the production of cytokines and chemokines (IL-1β, IL-6, IL-8, and macrophage inflammatory protein-2), cell-surface receptors (IL-2 receptor), adhesion molecules (ICAM-1), inflammatory enzymes (inducible nitric oxide synthase and cyclooxygenase 2), stress proteins (complement factors B, C3, and C4), and immunoregulatory molecules (MHC I and MHC II; ref. 32). A human NF-κB signaling pathway gene array revealed that Ad.AEG-1 infection resulted in marked up-regulation of NF-κB-responsive cell adhesion molecules (ICAM-2 and ICAM-3, selectin E, selectin L, and selectin P ligand), TLR4 and TLR5, and cytokines such as IL-8 (Fig. 2B). What is the significance of induction of these genes by AEG-1 in the context of tumorigenesis and HIV infection? ICAM-2 and ICAM-3 are frequently overexpressed in B-cell chronic lymphocytic leukemia (33). ICAM-3 exerts numerous functions in T-cell adhesion, polarization, and activation during the immune response (34–36). Previous studies suggest that ICAM-3 acts as a costimulatory molecule to increase HIV-1 transcription and viral production, a process allowing productive infection in quiescent CD4+ T lymphocytes (37). Our observations with the cytokine array that infection with Ad.AEG-1 markedly upregulated ICAM-3 production (34.44-fold over control) suggest that AEG-1 might act as a molecule in the positive feedback loop to further promote HIV replication. E-selectin was shown to regulate the attachment of a colon carcinoma cell line to endothelial cells in vitro (38) whereas P-selectin has been shown to bind to various types of carcinoma cells (39). High levels of soluble E-selectin have also been reported in melanoma and some epithelial tumors (40). Similarly, L selectin has been identified as a possible predictor of the hematogenous dissemination of murine lymphomas (41). Infection with Ad.AEG-1 produced a marked up-regulation of all of the above-mentioned selectins, uncovering a potential mechanism of AEG-1-mediated enhanced lung metastasis of breast cancer cells and increased adhesion of immortalized melanocytes to lung endothelial cells (data not shown). AEG-1 also induced significant up-regulation of TLR4 (42), which has been implicated in tumor cell immune evasion. Blockade of the TLR4 pathway by either TLR4 short interfering RNA or a cell-permeable TLR4 inhibitory peptide reverses tumor-mediated suppression of T-cell proliferation and natural killer cell activity in vitro and in vivo, delays tumor growth, and thus prolongs the survival of tumor-bearing mice. The observation that AEG-1 induces certain cell adhesion molecules that are intimately involved in tumor metastasis suggests a potential involvement of AEG-1 in these pathologic processes.

In summary, we now document that AEG-1 induces increased anchorage-independent growth and invasiveness of tumor cells and increased expression of adhesion molecules by activating the NF-κB pathway, and we hypothesize that these fundamental physiologic changes promoted by this gene might be important components of tumor progression and metastasis. Moreover, we have shown that AEG-1 can physically interact with p65 and modulate its function in the nucleus. Increased insights into the detailed molecular mechanism(s) of AEG-1 action will help determine its role in the process of tumorigenesis and HIV infection and facilitate development of strategies to counteract these virulent disease states by targeting AEG-1 via an antisense or siRNA approach, or a small-molecule inhibitor, for inactivation.

Grant support: NIH/NS P01 NS31492; the Samuel Waxman Cancer Research Foundation; the Chernow Endowment; and a Joelle Syverson Fellowship from the American Brain Tumor Association (L. Emdad).

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.