Nuclear Factor-κB Activation Correlates with Better Prognosis and Akt Activation in Human Gastric Cancer

It is believed that human cancers, including gastric carcinoma, develop through the accumulation of genetic alterations, such as oncogene activation and tumor suppressor gene loss (1–3). Because acceleration of proliferation and changes in the apoptotic pathway can contribute to carcinogenesis, it is important to identify genetic alterations that facilitate cell proliferation and apoptosis to enable us to understand the behaviors of malignant tumors. Although gastric cancer is one of the most common malignancies worldwide and a major cause of cancer mortality in Asia, the underlying molecular mechanisms of its initiation and progression are largely unknown.

The activation of transcription factor nuclear factor-κB (NF-κB) regulates various genes involved in the proliferation, invasion, angiogenesis, and metastasis of cancer cells. Constitutive activation of NF-κB has been observed and positively related to tumor progression in various cancers, including renal cancer, cervical cancer, and esophageal cancer (4–6). Thus, activated NF-κB was suggested as a therapeutic target for the treatment of these tumors. However, the role of NF-κB activation in tumor progression, cell growth, and apoptosis may differ according to species and cell type (7–9). With respect to gastric cancer, Sasaki et al. (10) reported the constitutive activation of NF-κB and its relations with clinicopathologic features, such as tumor progression, but its significance as a prognostic marker was unclear.

It was reported recently that the expression of tumor suppressor genes correlates with gastric cancer prognosis, and in particular, a tumor metastasis suppressor gene, kangai 1 (KAI1), was found to influence tumor invasion and progression in gastric carcinoma (11, 12). It was also shown that NF-κB activation positively regulates KAI1 expression in lung cancer cells (13). However, the correlation between NF-κB activation and tumor suppressor gene expression in gastric carcinoma has not been reported.

Akt/protein kinase B is an important mediator of many cell survival signaling pathways, and recently, Akt was reported to regulate NF-κB activation (14–18). However, contradictory findings suggesting NF-κB activation independent of Akt activation were also reported in HeLa cells and ovarian cancer cells (19, 20). With respect to gastric carcinoma, we reported previously that pAkt expression might have clinical significance in gastric carcinoma. However, its association with NF-κB activation in gastric carcinoma has not been described previously.

Although the significance of constitutively active NF-κB has been studied in gastric cancer (10), information on the clinical significance of NF-κB activation is limited. In the current study, we investigated the correlation between NF-κB activation and the prognosis as well as expanded clinicopathologic variables in 290 human gastric carcinomas by immunohistochemically studying tissue array slides. We also analyzed the correlation between NF-κB activation and Akt activation or the expressions of tumor suppressor genes. Our results show that, in contrast to those of Sasaki et al. (10), NF-κB activation is significantly correlated with better prognosis in human gastric cancer. In addition, NF-κB activation was found to be regulated, at least in part, by and regulates Akt.

Patients and tissue samples. The files of 290 surgically resected gastric cancer cases examined at the Department of Pathology, Seoul National University College of Medicine (Seoul, Korea) from January 1 to June 30, 1995 were analyzed (21). Age, sex, tumor location, gross type, tumor size, lymphatic invasion, and pathologic tumor-node-metastasis (pTNM) stage (22) were evaluated by reviewing medical charts and pathologic records. Mean patient age was 54.8 years, and 93.3% of the patients had undergone curative resection (R0 according to the International Union against Cancer guideline). The study included 208 advanced and 82 early-stage gastric carcinomas. According to the pTNM classification, 120 cases were in stage I, 57 in stage II, 68 in stage III, and 45 in stage IV. No patient had received preoperative chemotherapy or radiotherapy. Glass slides were reviewed to determine histologic type according to WHO and Lauren's classification. This series included 109 intestinal types, 151 diffuse types, and 30 cases of mixed types. Clinical outcomes were followed from the date of surgery to either the date of death or December 1, 2000, resulting in a mean follow-up period of 54 months ranging from 1 to 72 months. Cases that were not followed and those resulting in death from any cause other than gastric cancer were censored for survival rate analysis.

Tissue array methods. Six array blocks containing a total of 290 cases were used as described previously (Superbiochips Laboratories, Seoul, Korea; ref. 23). Core tissue biopsies (2 mm in diameter) were taken from individual paraffin-embedded gastric tumors (donor blocks) and arranged in a new recipient paraffin block (tissue array block) using a trephine apparatus. As we have reported previously, the staining results of different intratumoral areas of gastric carcinomas in these tissue array blocks show excellent agreement (21). A core was chosen from each case for analysis. An adequate case was defined as a tumor occupying >10% of the core area. Each block contained internal controls consisting of nonneoplastic gastric mucosa from body, antrum, and intestinal metaplasia. Sections of 4 μm were cut from each tissue array block, deparaffinized, and dehydrated.

Immunohistochemistry. Immunohistochemical staining for RelA (p65) was done using a streptavidin peroxidase procedure after autoclave-based antigen retrieval. Anti-p65 polyclonal antibody from Santa Cruz Biotechnology (Santa Cruz, CA) was used as primary antibody. Other antibodies were purchased from the following companies: anti-phospho-Akt (Ser473) and anti-Akt from New England Biolabs (Beverly, MA); anti-p16 from PharMingen (San Diego, CA); anti–adenomatous polyposis coli, anti-Smad4, and anti-KAI1 from Santa Cruz Biotechnology; and anti-Bcl-2 from DAKO (Carpinteria, CA). Other chemicals were purchased from Sigma (St. Louis, MO). Immunostaining results were considered positive if ≥10% of the neoplastic cells were stained.

Cell culture. The human gastric cancer cell line SNU-216 was obtained from the Korean Cell Line Bank (Seoul, Korea). SNU-216 cells were cultured in RPMI medium containing 10% fetal bovine serum in a 37°C humidified incubator in a mixture of 95% air and 5% CO₂.

Retroviral vector construction and retroviral infection of human gastric cancer cells. The retroviral vector MFG.EGFP.IRES.puro has been described (24), and the retroviral vector containing constitutively active Akt (CA-Akt) was constructed by replacing the enhanced green fluorescence protein coding sequence of MFG.EGFP.IRES.puro with the CA-Akt coding sequence obtained by the PCR amplification of pUSEamp-CA-Akt (Upstate Biotechnology, Lake Placid, NY). The retroviral vector MFG.IκBαM.IRES.puro, which encodes a supersuppressive mutant form of IκBα, was kindly provided by Dr. Hee-Yong Chung. MFG.CA-Akt.IRES.puro, MFG.IκBαM.IRES.puro, and MFG.EGFP.IRES.puro control retroviruses were generated and infected into SNU-216 cells as described previously (25). Pooled puromycin-resistant cells were used for further analysis.

Transfection and luciferase reporter assay. The NF-κB-luciferase reporter plasmid (Stratagene, La Jolla, CA) contains a 5× NF-κB response element fused to luciferase. To determine whether the transcriptional activity of NF-κB can be controlled by Akt, SNU-216 cells were transiently transfected with pNF-κB-luciferase or control plasmid pFC-MEKK (Stratagene) using LipofectAMINE Plus (Life Technologies, Grand Island, NY) according to the manufacturer's instructions. After 24 hours of transfection, SNU-216 cells were treated with the phosphoinositide 3-kinase inhibitors, LY294002 (50 μmol/L) or wortmannin (200 nmol/L), for 24 hours, and the activity of firefly luciferase was determined using a Dual-Luciferase Reporter Assay System (Promega, Madison, WI). For SNU-216.CA-Akt cells, cells were transiently transfected with pNF-κB-luciferase or pFC-MEKK, and luciferase activity was measured after 24 hours of transfection. Relative light units were measured using an AutoLumat LB9505c luminometer (Berthold Analytical Instruments, Germany).

Preparation of nuclear and cytoplasmic extracts. Cells were lysed in 100 μL buffer A [10 mmol/L Tris (pH 8.0), 60 mmol/L NaCl, 1 mmol/L EDTA, 1 mmol/L DTT, 0.1% NP40, 1 mmol/L phenylmethylsulfonyl fluoride], incubated on ice for 5 minutes, and centrifuged (pulsing for 5 seconds at 4°C), and the cytoplasmic extracts obtained (the supernatant) were transferred to fresh tubes. Glycerol was then added to 20%, and the extracts were stored at −80°C until required. The pelleted nuclei were immediately washed in 1 mL buffer A without NP40, spun as described above, and resuspended in 50 μL buffer B [200 mmol/L HEPES (pH 7.9), 0.75 mmol/L spermidine, 0.15 mmol/L spermine, 0.2 mmol/L EDTA, 2 mmol/L EGTA, 2 mmol/L DTT, 20% glycerol, 1 mmol/L phenylmethylsulfonyl fluoride, 0.4 mol/L NaCl]. They were then extracted on ice for 10 minutes with occasional vortexing and centrifuged, and the supernatant was collected as nuclear extract and stored at −80°C until required.

Immunoblot analysis. Equal amounts of protein were loaded into 8% to 10% SDS-polyacrylamide gels. Proteins were electrophoretically transferred to a nitrocellulose membrane, which was blocked with 7.5% nonfat dry milk in PBS-Tween 20 (0.1%, v/v) at 4°C overnight, and membranes were incubated with primary antibodies (diluted according to the manufacturer's instructions) for 3 hours. Horseradish peroxidase–conjugated anti-rabbit or anti-mouse IgG was used as secondary antibodies for 1 hour. Immunoreactive protein was visualized by enhanced chemiluminescence (Amersham, Arlington Heights, IL).

Reverse transcription-PCR. Total RNA was extracted using TRIzol (Life Technologies) following the manufacturer's instructions and converted to cDNA using Moloney murine leukemia virus reverse transcriptase (Promega) with oligo(dT) primers (Novagen, Milwaukee, WI). To detect Akt1, Akt2, and Akt3 mRNA, 5 μL of the resultant cDNA were added to 50 μL PCR mixture containing 1× PCR buffer, 2.5 unit Taq DNA polymerase, 1.5 mmol/L MgCl₂, 200 μmol/L deoxynucleotide triphosphates, and 20 pmol of each specific primer. The following specific primers were used: Akt1, 5′-GCTGGACGATAGCTTGGA-3′ and 5′-GATGACAGATAGCTGGTG-3′; Akt2, 5′-GGCCCCTGATCAGACTCTA-3′ and 5′-TCCTCAGTCGTGGAGGAGT-3′; Akt3, 5′-GCAAGTGGACGAGAATAAGTCTC-3′ and 5′-ACAATGGTGGGCTCATGACTTCC-3′; and glyceraldehyde-3-phosphate dehydrogenase, 5′-TGCCGTCTAGAAAAACCTGC-3′ and 5′-ACCCTGTTGCTGTAGCCAAA-3′. The PCR cycling conditions were as follows: 1 cycle at 94°C for 3 minutes followed by 30 cycles at 94°C for 20 seconds, 57°C for 30 seconds, and 68°C for 45 seconds and a final extension for 10 minutes at 72°C. Products were separated in 1.5% agarose gel containing 0.5 μg/mL ethidium bromide.

Statistical analysis. Survival curves were estimated using the Kaplan-Meier product-limit method, and the significances of differences between the survival curves were determined using the log-rank test. The χ² test or Fisher's exact test (two-sided) was used to determine the nature of correlations between p65 expression status and the clinicopathologic characteristics. Results were considered statistically significant for Ps < 0.05. All statistical analyses were conducted using SPSS version 11.0 software (SPSS, Chicago, IL).

Expression of RelA in gastric carcinoma tissues. To investigate whether NF-κB is activated in gastric carcinomas, immunohistochemistry was done using an antibody recognizing an epitope of RelA. Consecutive gastric carcinoma tissues (n = 290) were analyzed using a tissue array method; representative pictures are shown in Fig. 1. In nonneoplastic gastric mucosa, RelA expression was observed throughout the nucleus of the proliferative zone of gastric glands (Fig. 1A). Nuclear RelA expression was detected in intestinal metaplasia with or without cytoplasmic expression (Fig. 1B). Gastric tumor cells expressed RelA in the nucleus (Fig. 1C) and/or in the cytoplasm (Fig. 1D). We counted the number of tumor cells showing nuclear staining regardless of cytoplasmic staining, only cytoplasmic staining or negative staining (Fig. 1E), respectively. Cells showing RelA nuclear staining regardless of cytoplasmic staining were regarded as showing constitutive NF-κB activation.

Association between nuclear factor-κB activation and clinicopathologic features. Data representing the correlation between NF-κB activation and the clinicopathologic features of the 290 gastric cancer cases are summarized in Table 1. Nuclear expression of NF-κB was found in 18% of tumors and nuclear RelA expression was more likely in early-stage pTNM (P = 0.019). Seventy-six percent of nuclear NF-κB-positive tumors were pTNM stage I and II compared with 24% in stages III and IV. Moreover, we found a negative association between the nuclear staining of RelA and lymphatic invasion (P = 0.034) and lymph node metastasis (P = 0.055). No association was found between NF-κB activation and age, gender, tumor location, Lauren's classification, or distant metastasis.

Nuclear factor-κB activation in relation to prognosis. To determine whether NF-κB activation is a significant prognostic factor for the survival of patients with surgically resected gastric carcinoma, we used a log-rank test with Kaplan-Meier estimates. Of the 284 patients analyzed, those with nuclear RelA expression (51 cases) had a significantly higher survival rate than those with only cytoplasmic RelA expression or negative staining (P = 0.0228; Fig. 2). These results indicate that NF-κB activation is correlated positively with patient survival and suggest that NF-κB activation is a valuable biomarker of prognosis.

Correlation between nuclear factor-κB activation and Akt activation in gastric carcinoma. Because NF-κB activation is regulated by Akt (16, 17), we examined whether a correlation exists between nuclear RelA staining and pAkt expression in gastric carcinoma tissues, and nuclear RelA expression was found to be positively correlated with pAkt expression (P = 0.047; Table 2). By comparison, no association was found between nuclear RelA expression and Akt expression.

To confirm the relationship between NF-κB activation and Akt activity in gastric cancer cells, we investigated whether the inhibition of Akt activity by phosphoinositide 3-kinase inhibitors alters NF-κB activation in SNU-216 human gastric carcinoma cell line. We reported previously that Akt activity in SNU-216 cells is effectively blocked by phosphoinositide 3-kinase inhibitor treatment [i.e., LY294002 (50 μmol/L) or wortmannin (200 nmol/L; ref. 25)]. As shown in Fig. 3A, NF-κB transcriptional activity was attenuated in SNU-216 cells treated with LY294002 (50 μmol/L) or wortmannin (200 nmol/L) versus the untreated control. On the other hand, CA-Akt overexpression in SNU-216 cells increased the transcriptional activity of NF-κB (Fig. 3B).

To investigate whether NF-κB has a role in Akt activation, we produced a stable SNU-216 cell line overexpressing IκBαM. The overexpression of IκBαM was found to inhibit the nuclear translocation of RelA and the expressions of pAkt compared with control cells infected with an empty vector (Fig. 4A), suggesting a positive regulatory loop between NF-κB activation and Akt activation. In addition, down-regulation of NF-κB activation by IκBαM overexpression suppressed the expression of Akt protein (Fig. 4A). Thus, to investigate why NF-κB down-regulation suppresses the expression of Akt protein, we did reverse transcription-PCR with the specific primers for Akt family members, Akt1, Akt2, and Akt3, using a SNU-216 cell line overexpressing IκBαM (Fig. 4B). The overexpression of IκBαM resulted in reduced Akt3 mRNA level but not in the mRNA levels of Akt1 and Akt2. These findings indicate that down-regulation of NF-κB activation suppresses Akt3 expression at the transcriptional level and subsequently decreases Akt protein expression.

Activation of nuclear factor-κB or Akt in relation to the expression of tumor suppressor genes and Bcl-2. The associations between NF-κB or Akt activity and the expression of tumor suppressor genes and of Bcl-2 are shown in Table 3. NF-κB activation was found to be highly correlated with p16 (P = 0.004), adenomatous polyposis coli (P < 0.001), Smad4 (P = 0.002), and KAI1 (P < 0.001) expression. In comparison, no association was found between NF-κB activation and Bcl-2 expression. Similarly, Akt activation also showed positive correlation with p16 (P = 0.004) and KAI1 (P < 0.001).

Activation of nuclear factor-κB/Akt in relation to prognosis. We examined whether the combined evaluation of nuclear RelA and pAkt expression is correlated with gastric cancer patient survival. Kaplan-Meier estimate and the log-rank test showed that the survival rates of patients either nuclear NF-κB-positive and/or pAkt-positive immunoreactivity were higher than that of patients with a nuclear NF-κB-negative, pAkt-negative phenotype pattern (P < 0.0001; Fig. 5). In addition, among those negative for pAkt, those with nuclear NF-κB-positive immunoreactivity had a higher survival rate than those with nuclear NF-κB-negative immunoreactivity (P = 0.0003; Fig. 5A). These results indicate that the activations of NF-κB and/or Akt are significantly correlated with patient survival.

The present study confirms constitutively active NF-κB in gastric carcinoma, which agrees with the findings of Sasaki et al. (10). However, we also found that NF-κB activation is more prominent in early-stage gastric cancer and that this is negatively associated with lymphatic invasion. Moreover, gastric cancer patients positive for NF-κB activation showed better prognosis than those negative for NF-κB activation. In addition, this study shows, for the first time, a positive reciprocal relationship between NF-κB activation and Akt activation in gastric cancer cells.

In the present study, nuclear RelA expression was mainly observed in the proliferative zone of normal gastric glands and in intestinal metaplasia, suggesting that NF-κB activation is associated with cell proliferation. In gastric carcinoma, NF-κB was found to be activated in 18% of specimens, and NF-κB activation was more prominent in early-stage pTNM tumors than in late-stage tumors. We also found that NF-κB activation is negatively associated with both lymphatic invasion (P = 0.034) and lymph node metastasis (P = 0.055) and that nuclear RelA expression is positively associated with an improved postsurgery survival rate (P = 0.0228). However, our observations contrast with a report by Sasaki et al. (10), which suggested that NF-κB activity is a late event in carcinogenesis and contributes to the lymphatic invasion of gastric carcinoma by transcriptionally up-regulating urokinase-type plasminogen activator (uPA). Extracellular matrix degradation is an essential step in the processes of invasion and metastasis, and tumor-associated proteases, such as uPA, are thought to contribute to tumor growth, invasion, and metastasis. Although it has been reported that the enzyme activity of uPA in gastric carcinoma tissues is prognostically significant, the immunohistochemical expression of uPA in stromal cells, but not in cancer cells, was significantly correlated with uPA activity in tumor tissues (26, 27). However, Sasaki et al. observed concomitant NF-κB activation and uPA expression, but not uPA activation, in gastric cancer cells rather than in stromal cells. Thus, the high uPA expression in gastric cancer specimens observed by Sasaki et al. does not necessarily confirm the association between invasiveness and NF-κB activity in gastric carcinoma. In addition, they speculated that NF-κB activation was associated with poor prognosis in 64 surgically resected gastric cancer patients. However, the significance of NF-κB activation as a prognostic factor was inconsistent, and it did not correlate with the prognosis of 48 curative resection patients. We speculate that these discrepancies between the results of Sasaki et al. and ours may, at least in part, stem from differences in the numbers of tumor cases analyzed (n = 64 versus n = 290 in the present study). Taken together, it seems that NF-κB activation, at least in part, is required for cell growth and proliferation during early-stage human gastric carcinoma, during which tumors do not usually show lymphatic invasion.

With respect to the mechanism regulating NF-κB activation, it has been suggested that NF-κB activation is regulated through the activation of IκB kinase by Akt (16, 17). In the present study, we found that NF-κB activation positively correlates with pAkt expression in the gastric carcinoma and that, in the human gastric cancer cell line SNU-216, Akt activity regulates the transcriptional activity of NF-κB. Our in vitro study further showed that the inhibition of NF-κB activity by overexpressing IκBαM reduced not only Akt activation but also Akt expression, which was possibly caused by the suppression of Akt3 mRNA level but not Akt1 and Akt2 mRNA levels. These results suggest that there is a positive reciprocal regulatory loop between NF-κB and Akt and that NF-κB activation increases the Akt expression via differential regulation of Akt isoforms. However, our data obtained from gastric carcinoma tissues showed no association between nuclear RelA expression and Akt expression, which is in contrast to results obtained in gastric cancer cell line. Therefore, our data show that Akt activation, which might play a more important role in the apoptosis and cell cycle progression of tumor cells than Akt protein expression, is positively correlated with NF-κB activation in both gastric cancer tissues and gastric cancer cell line. To the best of our knowledge, this is the first report on cross-talk between NF-κB and Akt in the regulation of human gastric cancer cells.

We reported previously on the positive correlation between Akt activation and tumor suppressor genes, such as adenomatous polyposis coli and Smad4 but not Bcl-2 (23). In the present study, we also examined its associations with p16 (P = 0.004) and KAI1 (P < 0.001). Thus, we found that NF-κB activation is positively correlated with the expression of tumor suppressor genes, such as p16 (P = 0.004), adenomatous polyposis coli (P < 0.001), Smad4 (P = 0.002), and KAI1 (P < 0.001), demonstrating a similarity between Akt and NF-κB in terms of their correlation with tumor suppressor genes. Moreover, the selective expression of these suppressor genes in early-stage gastric carcinoma may partly explain their association with NF-κB or Akt activation as described above. However, the biological roles of NF-κB or Akt activation in relation to the expressions of tumor suppressor genes remain to be elucidated.

Akt activation, which is involved in the development of some human malignancies, has been reported to be correlated with poor outcome in several cancers, including breast and pancreatic cancers (28, 29). On the contrary, it was reported recently that Akt has no prognostic significance in the subgroup of breast cancer and that Akt activation is responsible for frequent and early involvement in lung cancer (30, 31). In gastric carcinoma, we reported previously that Akt is highly activated in the early pTNM stage and that Akt activation positively correlates with better patient outcome (23). The present study shows that NF-κB activity in gastric cancer cells is controlled by and controls Akt and that, in gastric carcinoma, positive correlations of NF-κB activation with expression of tumor suppressor genes, early pTNM stage, and better prognosis are similar to those of Akt as described in our previous report (23). Therefore, the combined status of NF-κB activation and Akt activation was assessed in relation to survival, and the nuclear NF-κB-negative and pAkt-negative group showed poorer outcome than the other groups examined.

The assessment of biological prognostic factors is of clinical importance, especially for a disease with poor outcome, such as gastric cancer. The widely used TNM staging has a high prognostic power, but of course, it cannot predict perfectly the outcome for a particular individual. The outcome of cancer patients may be influenced by interpatient variability in tumor biology. Thus, tumors with similar clinical or pathologic characteristics frequently show a different clinical outcome.

Although results of the present study suggest that NF-κB activity may be used as a prognostic biomarker of gastric cancer, it is less reliable than TNM criteria, which is the most important prognostic factor in gastric carcinoma. One of the merits of NF-κB activation and related tumor suppressor genes is that the endoscopic biopsy specimens for the analysis are easily accessible, whereas TNM classification needs surgery materials that can be obtained after very invasive procedures. The evaluation of NF-κB activation in the biopsy specimen may be helpful for predicting cancer stage in patients before radical surgery or in patients impossible to undergo surgery. Currently, endoscopic mucosal resection without lymph node dissection has been done for early-stage cancer patients, but it is insufficient to determine the TNM stage accurately. In this case, the combined evaluation of resection specimen and NF-κB activation in gastric cancer tissues aids in predicting the clinical prognosis of gastric cancer patients more correctly. Moreover, understanding the relationship between expression of these prognostic markers and effect of anticancer drugs will provide an indication of subsequent treatment responses of gastric cancer patients.

In conclusion, our results show that NF-κB activation is frequent in the early-stage gastric carcinoma, negatively associated with lymphatic invasion, and significantly associated with better prognosis. In addition, it was positively correlated with several tumor suppressor genes and Akt activation. The present findings suggest that NF-κB activation plays an important role probably via inducing tumor cell proliferation, which is more conspicuous in the early stage of gastric carcinoma. Further investigations are needed to clarify the mechanism of NF-κB activation, which is a candidate prognostic biomarker in the gastric cancer.

We thank S.P. Kim and Superbiochips Laboratories for their technical assistance.