Nuclear Factor-κB-dependent Expression of Metastasis Suppressor KAI1/CD82 Gene in Lung Cancer Cell Lines Expressing Mutant p53¹

KAI1/CD82, a protein with four transmembrane domains that belongs to the so-called TM4SF³ (1), has been shown to be a metastasis suppressor for human prostate and pancreatic cancer and possibly for NSCLC(2, 3, 4, 5). In addition, a recent study reported by Mashimo et al. (6) revealed that wild-type tumor suppressor p53 can directly activate KAI1/CD82 gene expression. On the other side, immunohistochemical analysis of tumor tissue samples from prostate cancer patients showed that among p53-negative tumor samples, 39% turned out to be KAI1/CD82-positive(6), suggesting the possibility that KAI1/CD82 expression in cancer cells is regulated by other pathways in vivo. However, the response of KAI1/CD82 gene expression in cancer cells to exogenous stimulants has not been investigated.

TNF, produced by various cell types including macrophages, endothelial cells, and cancer cells themselves during inflammatory reactions or tumor growth, is a central mediator of inflammation (7). TNF has an ability to activate leukocytes, trigger local production of inflammatory cytokines, and enhance adherence of neutrophils,monocytes, and tumor cells to the intercellular matrix and endothelial cells; therefore, TNF mediates many of the cellular responses associated with the cytotoxicity of leukocytes and the invasion and metastasis of cancer cells (8).

A pleiotropic transcription factor complex NF-κB, which participates in the regulation of genes coding for cytokines, cytokine receptors,MHC antigens, adhesion molecules, and viruses including HIV-1,plays a major role in the TNF-mediated activation of gene expression(9). The active NF-κB complex is composed of two subunits designated p50 and p65 (10). In resting cells,NF-κB exists as an inactive form in the cytoplasm associated with the inhibitory molecule IκBα (11). NF-κB can be activated by extracellular signals such as pro-inflammatory cytokines,including TNF (12). When cells are stimulated by these signals, IκBα is phosphorylated and then proteolytically degraded and separates from NF-κB. Then NF-κB migrates into the nucleus,where it activates target genes (13).

The present study was designed to determine whether TNF can affect KAI1/CD82 expression in lung cancer cells and, if so,whether NF-κB is involved in the mechanism of regulation of KAI1/CD82 gene expression. For this, we examined the expression of KAI1/CD82 protein and mRNA in lung cancer cell lines with or without TNF-α treatment by FACS analysis and RT-PCR. We then tested the effect of a specific inhibitor of NF-κB, IκBαSR (mutant IκBα, NF-κB super-repressor; Refs. 14  and 15) gene transfer, on KAI1/CD82expression.

Here, we present evidence to show the potential roles of NF-κB activation in KAI1/CD82 expression in lung cancer cells.

The amphotropic retrovirus packaging cell line PA317 was cultured in DMEM supplemented with 10% FBS. PC-14 (adenocarcinoma) was kindly supplied by Dr. Saijo (National Cancer Institute, Tokyo, Japan). RERF-LC-KJ (adenocarcinoma) and RERF-LC-AI (squamous cell carcinoma)were from Dr. Akiyama (Radiation Effects Research Foundation,Hiroshima, Japan). RERF-LC-OK (adenocarcinoma), RERF-LC-MS(adenocarcinoma), and A549 (adenocarcinoma) were obtained from American Type Culture Collection. These human lung cancer cell lines were grown in RPMI 1640 supplemented with 10% FBS.

Production and characterization of murine mAb 53H5, which binds to human CD82, were described previously (16). Anti-ICAM-1 antibody was purchased from Cosmo Bio Co. (Tokyo, Japan). Anti-MHC class I and anti-CD9 antibody were from Immunotech (Marseille, France)and anti-IκBα rabbit polyclonal antibody IκB-α-(C-21) was from Santa Cruz Biotechnology (California). TNF-α (specific activity, 3.25 × 10⁵ Japanese reference units/mg protein) was a gift from the Hayashibara Institute(Okayama, Japan).

Cells (10, 6) were resuspended in PBS supplemented with 10% human pooled AB serum to prevent nonspecific antibody binding. After incubation for 30 min at 4°C, the cells were washed once and incubated for 30 min at 4°C in 2% FBS-PBS containing 10 μg/ml each 53H5, anti-MHC class I, anti-ICAM-1,anti-CD9 mAb, or mouse control serum (Tago, Inc., Burlingame,CA). The cells were then washed with PBS, and fluorescein-conjugated goat antimouse IgG (H+L; Immunotech, Marseille, France) was added as a second antibody. After a 30-min incubation at 4°C, the cells were washed again and the fluorescence intensity was measured with a FACScan flowcytometer (Becton Dickinson, Mountain View, CA). Data files were analyzed using CellQuest software (Becton Dickinson), and the channel numbers were converged to linear values to compare the relative intensities of logarithmically amplified signals. The relative MFI of KAI1/CD82 was defined as the MFI of gated live cells stained with 53H5/the MFI of cells stained with mouse control serum.

PC-14 variant cells incubated with 20 units/ml TNF-α for the indicated period and untreated RERF-LC-OK cells were lysed in ISOGEN(Nippon Gene, Tokyo, Japan), a mixture of guanidinium isothiocyanate,and phenol. Total cellular RNA was then extracted with chloroform and precipitated with isopropanol. Continuous RT-PCR using avian myeloblastosis virus reverse transcriptase and Taq DNA polymerase was performed with 1 μg of total RNA using an mRNA-selective PCR kit using the nucleotide analogue deoxynucleotide triphosphate (TaKaRa, Tokyo, Japan). On the basis of the nucleotide sequence of KAI1/CD82, 5′-ATGGGCTCAGCCTGTATCAA-3′ was used as the sense primer, and 5′-ATAGCAGCTGCCTCAGTACT-3′ was used as the antisense primer. This primer pair amplifies an 816-bp fragment(nucleotides 166–981; Ref. 17). In the case of p53, 5′-CCTTCCCAGAAAACCTACCA-3′ was used as the sense primer and 5′-TCATAGGGCACCACCACACT-3′ as the antisense primer; these complementary sequences were located in exons 4 and 6, respectively,and the amplification product was a 371-bp fragment (18).β-Actin DNA amplification was used as the internal PCR control; the sense primer was 5′-AAGAGAGGCATCCTCACCCT-3′ and the antisense primer was 5′-TACATGGCTGGGGTGTTGAA-3′ (19). The reaction mixture(50 μl) was subjected to 30 PCR amplification cycles of 30 s at 85°C, 30 s at 40°C, and 90 s at 72°C. Five μl of the amplified DNA samples were run on a 2% agarose gel, and bands were visualized with ethidium bromide.

The oligomers used for the present study were the NF-κB binding site of the immunoglobulin κ light chain gene(cagaGGGACTTTCCgaga; Ref. 20) and the NF-κB-like sequence of the KAI1/CD82 gene (−371–362 bp, cagaGGGAGGgCCCgaga; Ref.21). Cancer cells were stimulated by the addition of 20 units/ml TNF-α, or not stimulated, for 2 h. Nuclear proteins were extracted according to the method described by Dignam et al. (22). Nuclear proteins (8 μg) were incubated with a ³²P-labeled probe (2 × 10⁴ cpm/reaction) and 0.5 mg/ml poly(dI-dC) · Epoly(dI-dC) in 20 μl of binding buffer [20 mm HEPES (pH 7.9), 60 mmKCl, 4 mm MgCl2, 0.2 mmEDTA, 1 mm DTT, 10% (v/v) glycerol, and 2%(v/v) polyvinyl alcohol] for 20 min at 25°C. In some experiments,nuclear extracts were incubated with a 25- or 100-fold molar excess of unlabeled oligomers for 10 min at 4°C before radiolabeled probe and poly(dI-dC) · poly(dI-dC) were added. Samples were loaded onto 6%polyacrylamide gels (acrylamide/N,N′-methylene bisacrylamide, 30:1) in 0.25× Tris borate buffer. After electrophoresis, gels were dried and analyzed using an image analyzer (BAS 2000; Fuji Film Co., Tokyo,Japan).

EMSA using a ³²P-labeled oligomer identical to the p53 binding site of the KAI1/CD82 gene (−896 to −863 bp; Ref. 6) was also performed.

A deletion mutant of IκBα missing the NH₂-terminal 36 amino acids, including two phosphorylation sites was used as IκBαSR (15) and cloned into pLXSN. The resulting plasmid was termed pLXSN/IκBαSR. DNA transfection was carried out by the high-efficiency calcium phosphate coprecipitation method. PA317 cells were transfected with pLXSN/IκBαSR or pLXSN and then selected with 600 μg/ml of G418. The resulting G418-resistant cells were pooled and used as retrovirus-producing cells.

PC-14 or RERF-LC-OK cells were plated at 5 × 10⁴ cells/100-mm dish in medium containing 6 μg/ml of polybrene (Aldrich, Milwaukee, WI), and transduced with LXSN/IκBαSR or LXSN retrovirus-containing cell supernatant. The transduced cells were then selected with 600 μg/ml of G418. The resulting G418-resistant cells were pooled and used as transduced cells. LXSN/IκBαSR-transduced cells and LXSN-transduced cells were termed PC-14/IκBαSR or OK/IκBαSR and PC-14/LXSN or OK/LXSN,respectively.

Cellular proteins were extracted according to the method described previously (23). Samples of 60 μg of cellular lysates were resolved by SDS-PAGE using a 10–20% gradient gel and transferred to a nitrocellulose membrane. The membrane was blocked in PBS with 0.1% Tween 20 containing 5% nonfat milk for 6 h at 4°C. After blocking, the membrane was incubated with a 1:1000 dilution of IκB-α-(C-21) for 1 h at room temperature. The immunoblots were incubated for 45 min with a 1:1000 dilution of horseradish peroxidase-linked sheep antirabbit IgG antibody (Amersham) and then with an enhanced chemiluminescence reagent.

Differences in results obtained in cytofluorometry were evaluated by Student’s two-tailed t test. In all determinations,differences were considered significant at P < 0.05.

First, we examined the levels of expression of KAI1/CD82 in human lung cancer cell lines by indirect staining with 53H5, and then the levels were expressed as relative MFI compared with control. As shown in Fig. 1 A, human lung cancer cell lines constitutively expressed KAI1/CD82 at a wide range of levels, and the expression in RERF-LC-OK(OK) was highest among these cell lines.

We next examined the effect of TNF-α on the expression of KAI1/CD82 in the same human lung cancer cell lines by incubating these cells in medium with 40 units/ml cytokine for 3 days before FACS analysis and found that KAI1/CD82 in lung cancer cells could be augmented by TNF-α, and that PC-14 was most responsive to TNF-α treatment (Fig. 1 B). Time-kinetic analysis showed that KAI1/CD82 expression in PC-14 cells treated with TNF-α reached a maximal level by 24 h after stimulation, and that the induction by TNF-α was dose-dependent and reached a plateau at 20 units/ml (data not shown).

We then examined the effects of TNF-α on KAI1/CD82 mRNA expression. As shown in Fig. 2 A, TNF-a could induce the expression of high level of KAI1/CD82 mRNA in PC-14 cells. The induction reached a maximum after 16 h of stimulation. Consistent with the low level of the spontaneous expression of KAI1/CD82 protein in PC-14 cells,KAI1/CD82 mRNA was detectable in unstimulated PC-14 cells when the gel was loaded with a sufficient amount of PCR product or when the numbers of PCR amplification cycles were increased (data not shown).

RT-PCR analysis of p53 mRNA was also performed to evaluate the association between expression of the p53 and KAI1/CD82 genes in PC-14 cells. However, increased amounts of p53 mRNA were not detectable in PC-14 cells treated with TNF-α, in which an increased level of KAI1/CD82 mRNA was induced (Fig. 2 B).

Nuclear extracts from unstimulated PC-14 cells or PC-14 cells stimulated with TNF-α were analyzed by a EMSA using ³²P-labeled oligonucleotides containing the immunoglobulin κ light-chain NF-κB consensus sequence or the NF-κB-like sequence of the KAI1/CD82 gene. Increased amounts of nuclear NF-κB were clearly detectable in cells treated with TNF-α (Fig. 3, Lanes 1 and 2). The specificity of the NF-κB mobility shift was confirmed by its prevention with the addition of an excess of the unlabeled immunoglobulin κ light-chain NF-κB probe (Fig. 3, Lanes 3 and 4). The same mobility shift was unaffected by the addition of an excess of the unlabeled oligonucleotides containing the NF-κB-like sequence of the KAI1/CD82 gene (Fig. 3, Lanes 5 and 6). Moreover, a specific mobility-shifted band was not detected when the ³²P-labeled KAI1/CD82 NF-κB-like probe was used (Fig. 3, Lanes 7 and 8).

When a 2-fold larger amount of nuclear proteins from unstimulated cells was incubated with ³²P-labeled immunoglobulin κlight-chain NF-κB probe, a small amount of mobility-shifted band was detected (data not shown).

We examined the effect of PDTC, which has the ability to block NF-κB activation independently of the inducing agent and cell line(24), on the up-regulation of KAI1/CD82 expression induced by TNF-α. PC-14 cells were treated with 100 mm PDTC for 24 h, supplemented with TNF-α (20 units/ml), and then cultured for another 24 h. As shown in Table 1, PDTC could block the augmentation of expression of not only MHC class I molecules and ICAM-1, for which NF-κB acts as a transcription factor, but also the expression of KAI1/CD82 on PC-14 cells induced by TNF-α. On the other hand, the expression of CD9, another member of the TM4SF, was slightly suppressed by TNF-α, and PDTC did not affect this suppression.

Becasue PDTC has not only an anti-NF-κB effect, but also metal-ion chelating activities, we tested the effect of a specific inhibitor of NF-κB, IκBαSR, on the KAI1/CD82 expression in PC-14 cells using a recombinant retroviral vector. After G418 selection, transduced cells were cloned by limiting dilution, and their ability to express IκBαSR was evaluated. Because IκBαSR is missing the NH₂-terminal 36 amino acids, the molecular weight is smaller than that of endogenous IκBα. The transduced cells,therefore, showed two bands in Western blot analysis using anti-IκBα antibody IκB-α-(C-21), which binds to the COOH-terminal sequence. As shown in Fig. 4,A, we isolated two clones with the ability to express high(clone 3) and intermediate (clone 10) levels of IκBαSR, and then the mixed population, which expressed a low level of IκBαSR, and two clones were further characterized. As shown in Fig. 4,B, IkBαSR gene transfer could inhibit the activation of NF-κB induced by TNF-α in correlation with the level of IκBαSR expression, and the amounts of NF-κB in the nucleus also correlated well with those of KAI1/CD82 mRNA (Fig. 4,C) and protein expression (Fig. 5 and Table 2) in PC-14 cells. Unlike MHC class I molecules, ICAM-1 and KAI1/CD82,CD9 expression was unaffected by IκBαSR gene transfer (Table 2).

Because KAI1/CD82-high-expressing cell line RERF-LC-OK constitutively expressed MHC class I molecules and ICAM-1 at high levels, NF-κB activation was expected in this cell line. Therefore,we also tested the effect of IκBαSR on the expression of these proteins on RERF-LC-OK cells. Although the transduction efficiency was very low, IκBαSR gene transfer inhibited spontaneous expression of MHC class I molecules, ICAM-1 and KAI1/CD82 on RERF-LC-OK cells, as expected (Table 3).

The experiments presented here show that TNF-α augments the KAI1/CD82 expression in lung cancer cell lines, and that NF-κB activation is involved in the mechanism of regulation of KAI1/CD82 gene expression. Transfer of the gene for IκBαSR, a specific inhibitor of NF-κB, inhibited the activation of NF-κB induced by TNF-α, and the amount of NF-κB in the nucleus correlated well with KAI1/CD82 gene and protein expression in PC-14 cells (Fig. 4, Table 2). In addition, IκBαSRgene transfer inhibited the spontaneous expression of this protein on KAI1/CD82-high-expressing RERF-LC-OK cells (Table 3).

The NF-κB activation in PC-14 cells induced by TNF-α was not completely blocked by IκBαSR gene transfer, and weak KAI1/CD82 mRNA and protein induction were still observed (Fig. 4, B and C), perhaps because endogenous wild-type IκBα in the mutant-transfected cell lines(Fig. 4 A) was degraded following TNF-α treatment, leading to NF-κB activation despite the expression of the mutant IκBα.

A previous study by Mashimo et al. (6) using human prostatic cancer cell lines identified the p53-responsive element in the 5′ upstream region of the KAI1 gene (−892 to −868)and revealed that wild-type p53, for which NF-κB seems to be a transcription factor (25), can directly activate KAI1/CD82 gene expression by interacting with this site. However, the p53 of PC-14 is a so-called contact mutant with a mutation of the Arg248 residue, which is directly involved in DNA-binding(26, 27), and increased amounts of p53 mRNA were not detectable in TNF-α-treated PC-14 cells, which showed increased KAI1/CD82 mRNA expression (Fig. 2 B). In addition, EMSA using a p53-related oligomer did not show p53 protein accumulation in TNF-α-treated PC-14 cells (data not shown). Moreover, the p53 of RERF-LC-OK has been reported to be a functionally mutant type as shown by functional analysis of separated alleles in yeast (28). These data indicate that p53 does not play a major role in KAI1/CD82 promoter activation in these lung cancer cell lines.

The promoter region of the KAI1/CD82 gene has no TATA or CCAAT box and, like other genes with TATA-less promoters, has many putative binding motifs for various transcription factors, including nine Sp1 sites and five AP2 sites (21). However, the KAI1/CD82 promoter sequence does not include a putative NF-κB binding site, and EMSA indicated that NF-κB cannot bind the NF-κB-like sequence of the promoter region of the KAI1/CD82 gene (−371 to −362; Ref. 21; Fig. 3). A possible explanation is that NF-κB acts as a transcription regulatory factor for another protein that directly activates KAI1/CD82 gene expression or increases the mRNA stability, or that some yet-undetected physical interaction between NF-κB and some other transcription factor(s), a phenomenon termed cross-coupling (29, 30), is required for their cooperative activation of KAI1/CD82 gene.

One characteristic common to target genes of NF-κB is their rapid induction. Precursors of the activating complex are present in the cytoplasm, and simple dissociation from the inactive complex does not require new protein synthesis (31). However, the effect of TNF-α on KAI1/CD82 mRNA induction in PC-14 cells was not rapid (Fig. 2 A). These results suggest the possibility that some unidentified factor mediates the relationship between the NF-κB-stimulation and KAI1/CD82 mRNA induction.

NF-κB is a transcription factor that has been shown to be associated with increased cell survival in many tumor cells. Numerous studies have revealed that the mechanisms in tumor cells that inhibit the apoptotic response induced by chemotherapy (32, 33, 34), radiation(35, 36), or TNF (37, 38) involve NF-κB activation. These data suggest that the NF-κBfamily could be considered to be proto-oncogenes. On the other hand,IκBα is a possible tumor suppressor, and antisense IκBαtreatment results in oncogenic transformation (39). In the case of IκBαSR gene-transduced RERF-LC-OK cells, after G418 selection for 2 weeks, several G418-resistant colonies that showed decreased KAI1/CD82expression were obtained (Table 3); however, stable transfectants could not be obtained by additional culturing, suggesting that the inhibition of NF-κB is toxic for cells in which NF-κB is constitutively activated.

MRP-1, which is identical to CD9, also belongs to the TM4SF and suppresses cell motility and metastasis (40). Although its precise biological functions remain unknown, several reports have shown that MRP-1/CD9 and KAI1/CD82 seem to act by similar mechanisms (41, 42, 43, 44). In the present study, however,TNF-α slightly suppressed MRP-1/CD9 expression, in contrast to the KAI1/CD82 induction, and an NF-κB inhibitor did not affect the TNF-α-inducible suppression or spontaneous expression of MRP-1/CD9 (Tables 1 and 2). These observations indicate that MRP-1/CD9 and KAI1/CD82 gene expression are regulated differently and independently of each other.

It is not evident as yet whether in vivo NF-κB status of NSCLC affects KAI1/CD82 expression and prognosis. At present, mechanisms by which KAI1/CD82 may act as a metastasis suppressor for NSCLC remains unclear. Recently, there is accumulating evidence that anti-inflammatory cytokines, such as IL-4 and IL-10, negatively control NF-κB activation and inhibit inflammatory cytokine production (45, 46), and that these cytokines are detectable at NSCLC lesions in significant concentration (47). In view of these results and the data presented here on the NF-κB dependent pathway for KAI1/CD82 expression, one possible explanation for reduced expression of KAI1/CD82 in human NSCLC, which is associated with a poor prognosis (5), may be that anti-inflammatory cytokines in the tumor microenvironment reduce KAI1/CD82expression through NF-κB inhibition.

In summary, our results revealed potential roles of NF-κB activation in KAI1/CD82 expression in lung cancer cells. These findings suggest that the level of in vivo KAI1/CD82expression may change during inflammatory reactions or tumor progression. Because NF-κB can up-regulate other factors that contribute to the malignant phenotype, this transcription factor may have dual roles in human lung cancer metastasis. Because several clinical reports have indicated that KAI1/CD82 expression cannot necessarily prevent metastasis (48), additional studies concerning the relationship between the p53 status and NF-κB activation may be necessary as part of the evaluation of the clinical effects of KAI1/CD82.