Cyclin-Dependent Kinase 1 Gene Expression Is Associated with Poor Prognosis in Gastric Carcinoma

Many basic and clinical studies have indicated that low levels of p27 are associated with high aggressiveness and poor prognosis in a large variety of malignant tumors (1). We previously reported that p27 expression status was an independent prognostic factor for patients with gastric carcinomas (2). p27 is an inhibitor of the protein kinases Cdk2⁴ /cyclin E and Cdk2/cyclin A, which drive cells from the G₁ to S phase in the cell cycle (1), and is mainly regulated in ubiquitin-mediated proteolysis with SCF-type ubiquitin ligase complex (3).

It is widely known that genetic or epigenetic changes in the p27 gene are rare in carcinomas (4). One main mechanism responsible for the decreased level of p27 protein in carcinomas is the increased expression of SCF^Skp2, especially Skp2 (5, 6, 7, 8, 9), which is a substrate-specific receptor of p27 (10). We also reported that Skp2 expression was correlated inversely with p27 expression and could be a prognostic factor in gastric carcinomas, and furthermore, Skp2 can modulate the malignant phenotype of gastric carcinomas possibly via p27 proteolysis (11). However, the reduced expression of p27 seems to be not necessarily caused by Skp2 overexpression in carcinomas (6, 9, 11).

Recently, human Cks1 has been identified as an essential and specific cofactor in the ubiquitination and degradation of p27 by SCF^Skp2 (12, 13). Cks proteins were originally identified as subunits that interacted tightly with cyclin-Cdk complexes (14, 15). Human Cks1 binds to Skp2 and greatly increases the binding of threonine 187-phosphorylated p27 to Skp2 (12). We hypothesized that Cks1 might be involved in p27 degradation in gastric carcinoma cases with low expression levels of both Skp2 and p27.

We therefore examined the relationship between Cks1 and p27 expression in vivo and clarified the clinical significance of Cks1 expression in human gastric carcinomas. We next established Cks1- or Skp2-transfected and Cks1/Skp2-cotransfected human gastric carcinoma cell lines and examined the relationship between Cks1, Skp2, and p27 expression in vitro.

Fresh surgical specimens were obtained from 76 patients with both primary gastric carcinoma tissues (T), and their paired adjacent normal gastric mucosa (N) after written informed consent was obtained. The patients had undergone surgery at the Department of Molecular and Surgical Oncology, Medical Institute of Bioregulation, Kyushu University (Beppu, Japan) from 1993 to 2000. None of these patients received preoperative treatment such as radiation or chemotherapy. Of the 76 patients, 50 received postoperative therapy (i.e., chemotherapy). Data concerning patient outcome, including overall survival and development of metastasis, were available for all 76 patients, and the observation period ranged from 3 to 77 months (the median follow-up period was 36.6 months). Of the 76 patients, 30 died of gastric carcinoma. Paraffin blocks were available from 28 cases and were analyzed by immunohistochemistry. All 76 cases were analyzed with quantitative RT-PCR.

The two different human gastric carcinoma cell lines, AZ521 and MKN28, were obtained from Cell Resource Center for Biomedical Research Institute of Development, Aging and Cancer, Tohoku University, and maintained in RPMI 1640 supplemented with 10% fetal bovine serum at 37°C in a 5% humidified CO₂ atmosphere.

Rabbit polyclonal antibody to Cks1 and mouse monoclonal antibodies to Skp2 and p27 were purchased from Santa Cruz Biotechnology (Santa Cruz, CA), Zymed Laboratories (San Francisco, CA), and Transduction Laboratories (Lexington, KY), respectively.

Frozen tissue specimens or cultured cell lines in a state of subconfluency were homogenized, the total RNA was extracted using the modified acid-guanidine-phenol-chloroform method, and reverse transcriptase reaction was performed. The PCR amplification for quantification of Cks1, Skp2, p27, and GAPDH mRNA in tissues and cell lines was performed in the LightCycler using the LightCycler-FastStart DNA Master Sybr Green I kit (Roche Diagnostics). The amplification conditions of 40 cycles consisted of denaturation at 95°C for 10 s, annealing at 60°C for 10 s, and elongation at 72°C for 20 s. After the amplification, the products were subjected to a temperature gradient from 68°C to 95°C at 0.2°C/s with continuous fluorescence monitoring to produce a melting curve of the products. After proportional background adjustment, the fit point method was used to determine the cycle in which the log-linear signal was distinguished from the background and that cycle number was used as a crossing-point value. The standard curve was produced by measuring the crossing point of each standard value (2-fold serially diluted cDNAs of AZ521) and plotting them against the logarithmic value of concentrations. The concentrations of each sample were then calculated by setting their crossing points to the standard curve. The expression levels were normalized by GAPDH. We classified the cases into two groups; a Cks1 high expression group (T/N < 1; n = 41) and a Cks1 low expression group (T/N ≦ 1; n = 35). The expression level of Skp2 mRNA was classified into two groups; a high group (n = 50) and a low group (n = 26) according to the criteria described previously (11). The following primers were used (all 5′ to 3′ direction): Cks1 (179 bp) sense primer AGCAAACCGAGCGATCATGTC and antisense primer CCCATCCCTGACTCTGCTGAA; Skp2 (292 bp) sense primer GCTGCTAAAGGTCTCTGGTGT and antisense primer AGGCTTAGATTCTGCAACTTG; p27 (225 bp) sense primer ACGGGAGCCCTAGCCTGGAGC and antisense primer TGCCCTTCTCCACCTCTTGCC; and GAPDH (249 bp) sense primer TTGGTATCGTGGAAGGACTCA and antisense primer TGTCATCATATTTGGCAGGTT.

Immunohistochemical studies of Cks1, Skp2 and p27 were performed on specimens available from 28 gastric carcinoma cases using the avidin-biotin-peroxidase method (LSAB2 kit; Dako, Kyoto, Japan) on formalin-fixed, paraffin-embedded tissues. All sections were counterstained with hematoxylin. The primary antibodies against Cks1, Skp2, and p27 were used at dilutions of 1:50, 1:500, and 1:1000, respectively.

p27 scores were measured by observing 1000 cancer cells in at least five high-power fields and were classified as high (staining in >50% of cells) or low (staining in ≦50% of cells) as described previously (2). The scoring was done independently by two observers (T. M. and Y. Y.).

Total protein was extracted from samples with radioimmunoprecipitation assay buffer. Aliquots of total protein were applied to 15% acrylamide gradient gels. After electrophoresis, the samples were electroblotted onto a polyvinylidene membrane (Immobilon; Millipore, Inc., Bedford, MA) at 0.5 A for 30 min at 4°C. Cks1, Skp2, and p27 were detected using the antibodies at dilutions of 1:500, 1:2000, and 1:2500, respectively. The blots were developed with horseradish peroxidase-linked antirabbit or antimouse immunoglobulin (Promega, Inc., Madison, WI). Signals were detected using Supersignal (Pierce, Inc., Rockford, IL).

We previously established stably Skp2-transfected gastric carcinoma cell lines (parent cell lines: AZ521 and MKN28; Ref. 11). Human Cks1 cDNA was generated by RT-PCR and subcloned into a pcDNA3.1Hygro+ expression vector (Invitrogen, Carlsbad, CA) and then transfected into the Skp2-transfected cells and the mock cells using Lipofectamine (Life Technologies, Inc., Tokyo, Japan). Mock vector-transfected clones of each cell line were used for the control.

Associations between the variables were tested by the Fisher’s exact test. Survival curves were drawn according to the Kaplan-Meier method, and the survival analysis was carried out by the Mantel-Cox test when two curves were being compared or the Tarone-Ware test when three curves were being compared. Multivariate analysis to determine the patient prognosis was performed with Cox regression analysis with the forward stepwise model. All differences were deemed significant at the level of P < 0.05. The histological type of gastric carcinomas was classified on the basis of the criteria set by the Japanese Society for Cancer of the Stomach (16).

The gastric carcinoma tissue (T) and normal mucosa (N) showed variable levels of Cks1 and Skp2 mRNA signals by quantitative RT-PCR (Fig. 1, A and B). This analysis revealed that the expression of Cks1 mRNA (Cks1 value) was greater in T (0.1–1.3; median, 0.2) than in N (0.0–1.0; median, 0.1) in 41 of the 76 cases (53.9%). No correlation in mRNA expression level was found between Cks1 and Skp2 in these cases (data not shown). Cks1 T/N ranged from 0.1 to 5.8 (median, 1.1).

Immunohistochemical analysis revealed that Cks1 was predominantly expressed in the gastric carcinoma cells (Fig. 1 C).

As shown in Table 1,A, 12 (70.6%) of 17 Cks1 high cases showed a low expression level of p27 protein and 8 (72.7%) of 11 Cks1 low cases showed a high expression level of p27 protein. Thus, Cks1 mRNA expression was inversely correlated with the expression level of p27 protein in gastric carcinomas (P < 0.05). Next, we classified these 28 cases into three groups: Cks1/Skp2 double-high group (n = 9); Cks1 high and Skp2 low or Cks1 low and Skp2 high group (n = 12); and Cks1/Skp2 double-low group (n = 7), and examined the correlation with the expression level of p27 protein. As shown in Table 1 B, 8 (88.9%) of 9 in the Cks1/Skp2 double-high group showed low expression level of p27 protein, and 6 (85.7%) of 7 in the Cks1/Skp2 double-low group showed a high expression level of p27 protein.

In gastric carcinoma tissues, cells positive for Cks1 and Skp2 showed no or very low p27 protein expression and vice versa, implying that there was an inverse relationship between the expression profiles of Cks1, Skp2 and p27 proteins (Fig. 1 D).

No significant difference was found between the high and low Cks1 expression groups regarding Skp2 expression and clinicopathological factors such as age, sex, histology, serosal invasion, lymph node metastasis, lymphatic or vascular invasion, and postoperative therapy (Table 2). As shown in Fig. 2,A, the Cks1 mRNA high expression group showed a significantly poorer prognosis than the low expression group (P = 0.018). On multivariate analysis for prognosis, Cks1 mRNA expression, Skp2 mRNA expression, lymph node metastasis, lymphatic involvement, and serosal invasion, which were found to be prognostic factors on univariate analysis, were included for the parameters. This analysis demonstrated that Cks1 mRNA expression was an independent prognostic factor (Table 3). Next, the 76 cases were classified into three groups: Cks1/Skp2 double-low group (n = 13); Cks1 or Skp2 high group (n = 35); and Cks1/Skp2 double-high group (n = 28), and survival analysis was performed (Fig. 2 B). The Cks1/Skp2 double-high group showed the poorest prognosis, and the Cks1/Skp2 double-low group showed the best prognosis among the three groups (P = 0.013).

We transfected Cks1 cDNA into the two different Skp2-transfected gastric carcinoma cell lines, and the expression levels of Cks1 and Skp2 were analyzed by both quantitative RT-PCR and Western blot (Fig. 3). The level of p27 mRNA was not altered by Cks1 or/and Skp2 transfection. Skp2-transfected cell lines showed a lower level of expression of p27 protein than the controls, and Cks1/Skp2-cotransfected cell lines showed a markedly lower expression level of p27 protein than the controls. Treatment with proteasome inhibitor (MG132) blocked p27 degradation in Cks1- or/and Skp2-transfected cell lines (data not shown).

In the present clinical study of gastric carcinomas, Cks1 was overexpressed in gastric carcinoma cells, and its expression was inversely correlated with the expression level of p27 protein, and high Cks1 mRNA expression was significantly correlated with poor prognosis. These findings suggest that Cks1, in addition to Skp2 (11), is strongly related to the expression level of p27 in gastric carcinomas. Cks1 could be a potential prognostic factor for gastric carcinomas.

It was worth noting that the Cks1/Skp2 double-high expression group showed a low expression level of p27 protein in 8 (88.9%) of 9 cases, and the Cks1/Skp2 double-low expression group showed a high expression level of p27 protein in 6 (85.7%) of 7 cases. The Cks1/Skp2 double-high group showed the poorest prognosis, and the Cks1/Skp2 double-low group showed the best prognosis among the three groups. In the in vitro study, the Cks1/Skp2-cotransfected gastric carcinoma cells showed a markedly lower expression level of p27 protein than the controls. These findings indicate that both Cks1 and Skp2 overexpression more strongly promote the proteolysis of p27 than either Cks1 or Skp2 overexpression. Thus, Cks1 and Skp2 may cooperate in regulating the expression level of p27 protein in gastric carcinomas. These clinical and in vitro data suggest that, in most gastric carcinomas, the expression level of p27 is mainly regulated by the expression of Cks1 and Skp2 through ubiquitin-mediated proteolysis. To clarify the correlation between the Cks1/Skp2 expression and p27 proteolysis, additional studies using more gastric carcinoma tissues will be required.

It is unclear how Cks1 participates in the p27 ubiquitination by SCF^Skp2. However, one possibility is that Cks1 associates with Skp2 and probably confers an allosteric change in Skp2, resulting in increasing its affinity for the phosphorylated p27 and consequently promoting p27 proteolysis (17, 18).

The mechanism of Cks1 and Skp2 overexpression in gastric carcinoma cells is uncertain. Recently, it was reported that the genomic locus of the Skp2 gene was amplified and the DNA copy number was closely associated with Skp2 expression in small cell lung carcinoma (8). On the other hand, gene amplification of human chromosome 1q21, where the human Cks1 gene is located, has been reported in human epithelial malignancies (19, 20). Cks1 and Skp2 overexpression in carcinomas may be attributable to this gene amplification. We are now examining the genomic alterations of Cks1 and Skp2 and will clarify the mechanism of Cks1 and Skp2 overexpression in gastric carcinomas.

In conclusion, Cks1 expression was correlated inversely with the expression level of p27 protein and could be an independent prognostic factor in gastric carcinomas. Furthermore, Cks1 cooperates with Skp2 to strongly degrade p27 protein, and Cks1/Skp2 coexpression was correlated inversely with the expression level of p27 protein in gastric carcinomas. These findings strongly suggest that Cks1, in addition to Skp2, could play an important role in gastric carcinoma progression and would be a novel molecular target for the treatment of gastric carcinomas, as well as a strong prognostic marker.

We thank Dr. Fumiaki Tanaka, Mitsuhiko Ohta, and Hidetaka Masuda for critical advice and Junko Miyake, Kazue Ogata, and Toshiko Shimooka for technical assistance.