2447 Reduced Expression of Cyclin-Dependent Kinase Inhibitor p27Kip1 Is an Indicator of Malignant Behavior in Oral Squamous Cell Carcinoma Yasusei Kudo, D.D.S.1 Takashi Takata, D.D.S., Ph.D.1 Wataru Yasui, M.D., Ph.D.2 Ikuko Ogawa, D.D.S., Ph.D.3 Mutsumi Miyauchi, D.D.S., Ph.D.1 Toshitsugu Takekoshi, D.D.S.1 Eiichi Tahara, M.D., Ph.D.2 Hiromasa Nikai, D.D.S., Ph.D.1 1 Department of Oral Pathology, Hiroshima University School of Dentistry, Hiroshima, Japan. 2 First Department of Pathology, Hiroshima University School of Medicine, Hiroshima, Japan. 3 Clinical Laboratory, Hiroshima University Dental Hospital, Hiroshima, Japan. Presented at the 39th Annual Meeting of the Japanese Association for Oral Biology, Kitakyushu, Japan, October 1–2, 1997. Supported in part by Grant No. 07457429 from the Ministry of Education, Japan. Address for reprints: Yasusei Kudo, D.D.S., Department of Oral Pathology, Hiroshima University School of Dentistry, 1-2-3 Kasumi, Minami-ku, Hiroshima 734, Japan. Received February 23, 1998; revision received April 27, 1998; accepted April 27, 1998. © 1998 American Cancer Society BACKGROUND. Reduced expression of the cyclin-dependent kinase inhibitor p27Kip1 has been reported to correlate with poor survival in cohorts of breast and colorectal carcinoma patients. Posttranslational ubiquitin-mediated proteasomal proteolysis is related to p27Kip1 protein levels. However, to the authors’ knowledge, no previous study has examined the expression of p27Kip1 in oral squamous cell carcinoma (OSCC). METHODS. To examine the expression of p27Kip1 and its clinicopathologic roles in OSCC, the authors studied the expression of p27Kip1 protein by immunohistochemistry in deparaffinized tissue sections of 20 normal oral mucosa specimens, 22 epithelial dysplasia specimens, and 70 OSCCs, and analyzed its correlation with clinicopathologic parameters. They also studied the expression of p27Kip1 mRNA and protein in six OSCC cell lines by Northern blot and Western blot analysis. To examine the mechanism of reduced expression of p27Kip1, OSCC cell lines were treated with the proteasome inhibitor LLnV. RESULTS. All the normal oral mucosa specimens and 73% (16 of 22) of the oral epithelial dysplasia specimens expressed p27Kip1 at high levels, whereas 87% of the OSCCs (61 of 70) showed reduced expression of p27Kip1. Furthermore, the levels of expression of this protein were significantly lower in carcinomas with metastasis than those without metastasis. Although OSCC cell lines expressed p27Kip1 mRNA at various levels, most of them expressed p27Kip1 protein at lower or undetectable levels. LLnV induced the expression of p27Kip1 protein in HSC2 cells, in which p27Kip1 protein was originally undetectable. CONCLUSIONS. These findings suggest that 1) reduced expression of p27Kip1 may correlate with the development and progression of OSCC and can be an indicator of malignant behavior of this neoplasm, and 2) increased proteasome-mediated degradation may play an important role in the reduction of p27Kip1 protein expression. Cancer 1998;83:2447–55. © 1998 American Cancer Society. KEYWORDS: oral squamous cell carcinoma, p27Kip1, reduced expression, proteasome-mediated degradation, ubiquitin. A bnormalities in cell cycle regulators allow uncontrolled cell growth and division, which may have a role in carcinogenesis.1,2 Multiple cyclins and cyclin-dependent kinases (CDKs) are positive regulators of progress of the cell cycle. Cyclin/CDK complexes are activated by phosphorylation by the CDK-activating kinase (CAK), whereas cyclin/CDK complexes are negatively regulated by a number of CDK inhibitors.3–11 CDK inhibitors belong to two large families based on their structural and functional properties. The Ink4 family, which includes p16Ink4a, p15Ink4b, p18Ink4c, and p19Ink4d, consists of tandem repeats of an ankyrin-like sequence, whereas the Cip/Kip 2448 CANCER December 15, 1998 / Volume 83 / Number 12 TABLE 1 Age and Gender Distribution of the Patients Age (yrs) Normal oral mucosa Epithelial dysplasia OSCC Gender No. of cases Mean Range Male Female 20 22 70 55.85 59.91 58.67 31–80 29–83 24–88 12 14 36 8 8 34 OSCC: oral squamous cell carcinoma. family, which includes p21Cip1/Waf1, p27Kip1, and p57Kip2, has a homologous amino-terminal domain that contains contiguous cyclin and CDK binding regions. p27Kip1 is associated predominantly with cyclin D/CDK4, but has the ability to inhibit various cyclin/ CDK complexes in vitro.7,12 p27Kip1 in combination with newly formed cyclin E/CDK2 complexes blocked cell cycle progression.7 The cyclin E/CDK2 complex shows strong kinase activity shortly before cells enter the S phase and leads to further phosphorylation of the pRb protein.13–15 Cyclin E is overexpressed in certain types of human tumors.16 –18 p27Kip1 mediates G1 arrest induced by transforming growth factor-␤ (TGF␤), contact inhibition, or serum deprivation in epithelial cell lines.6 Although the role of p27Kip1 protein in cancer cells has not been elucidated, reduced expression of p27Kip1 has recently been shown to correlate with poor prognosis in breast and colorectal carcinomas.19 –22 Moreover, high levels of cyclin E and low levels of p27Kip1 have been shown to be strongly predictive of increased mortality, both before and after adjustment for other clinical and pathologic characteristics.19 Both in vivo and in vitro, p27Kip1 was found to be degraded by the ubiquitin-proteasome pathway.23 In various carcinomas, p27Kip1 mRNA levels are preserved, but protein levels are reduced.20,24 As carcinomas with low or absent p27Kip1 protein display enhanced proteolytic activity specific to p27Kip1, it has been suggested that low p27Kip1 expression is brought about by increased proteasome-mediated degradation rather than altered gene expression.21 To our knowledge, however, there has been no previous study of p27Kip1 expression in oral squamous cell carcinoma (OSCC). Therefore, to study the value of p27Kip1 expression in determining the prognoses of OSCC patients, we examined the immunohistochemical expression of p27Kip1 in OSCC and its correlation with clinicopathologic findings. We also investigated whether a degradation pathway of p27Kip1 by a ubiquitin-proteasome complex is present in OSCC. MATERIALS AND METHODS Tissue Samples Tissue samples of 20 normal oral mucosa specimens, 22 epithelial dysplasia specimens, and 70 OSCCs from the years 1976 –1997 were retrieved from the Surgical Pathology Registry of Hiroshima University Dental Hospital. For the current analysis, only biopsied specimens from the tongue, taken before radiochemotherapy, were selected to avoid possible influences of the lesional sites and treatment modalities on data. The age and gender distributions of the patients are summarized in Table 1. Of the 70 OSCC patients, follow-up data were available for 51 (25 men and 26 women). At the time of diagnosis, their ages ranged from 24 to 87 years (mean, 57.14 years). The mean follow-up period for these OSCC patients was 79.96 months (range, 4 –185 months). Tissues were fixed in 10% buffered formalin and embedded in paraffin. The number of cases of each degree of epithelial dysplasia and histologic grade of OSCC is listed in Table 1. Histologic grade and stage of tumor were classified according to the criteria of the Japan Society for Head and Neck Cancer.25 Stage grouping was based on the TNM classification as follows: Primary tumor (T): no evidence of primary tumor (T0); the greatest dimension of primary tumor less than 2 cm (T1), 2– 4 cm (T2), or more than 4 cm (T3); massive tumor more than 4 cm in greatest dimension with deep invasion to involve the antrum, the pterygoid muscles, the root of the tongue, or the skin of the neck (T4). Lymph node involvement (N): No clinically positive lymph nodes (N0); a single clinically positive ipsilateral lymph node less than 3 cm in greatest dimension (N1); a single clinically positive ipsilateral lymph node 3– 6 cm in greatest dimension, or multiple clinically positive ipsilateral lymph nodes, none over 6 cm in greatest dimension (N2); massive ipsilateral lymph nodes, bilateral lymph nodes, or contralateral lymph nodes (N3). Distant metastasis (M): No distant metastasis (M0), distant metastasis at presentation (M1). Stage was grouped as follows: Stage Reduced Expression of p27Kip1 in OSCC/Kudo et al. I, including T1, N0, M0; Stage II, including T2, N0, M0; Stage III, including T3, N0, M0 or T1–T3, N1, M0; Stage IV, including T4, N0 –1, M0 or any T, N2–3, M0 or any T, any N, M1. Cell Culture Six OSCC cell lines (HSC2, HSC3, HSC4, Ca9-22, Ho1-U-1, and Ho-1-N-1) were examined. All the cell lines were provided by the Japanese Cancer Research Resources Band (JCRB). They were routinely maintained in RPMI-1640 (Kyokuto Pharmaceutical Industrial Co., Tokyo, Japan) and supplemented with 10% heat-inactivated fetal bovine serum (Boehringer Mannheim K. K., Australia) and 100 U/mL penicillin-streptomycin (Gibco BRL, Grand Land, NY) under conditions of 5% CO2 in air at 37°C. For experiments, they were grown to subconfluence in this medium. Immunohistochemistry Immunohistochemical detection of p27Kip1 was performed on 20 normal oral mucosa specimens, 22 epithelial dysplasia specimens, and 70 OSCCs of various stages to analyze the relation with clinicopathologic parameters of OSCC patients. We also studied immunoexpression of cyclin E to examine the correlation between expression of p27Kip1 and cyclin E in 51 OSCC cases with follow-up data. Immunostaining was performed using an antihuman p27Kip1 mouse monoclonal antibody (K25020, Transduction Laboratories, Lexington, KY), an anti– cyclin E monoclonal antibody (14591A, Pharmingen, San Diego, CA), and a streptavidin-biotin kit (Nichirei, Tokyo, Japan). Deparaffinized tissue sections were immersed in methanol containing 0.3% hydrogen peroxide for 30 minutes to block endogenous peroxidase activity. Microwave pretreatment in citrate buffer was performed for 10 minutes 3 times to retrieve the antigenicity. The sections were then incubated with normal rabbit serum for 30 minutes to block nonspecific antibody binding sites. The sections were treated consecutively at 4°C with an anti-p27Kip1 antibody (diluted 1:100) or an anti– cyclin E antibody (diluted 1:200) overnight, biotinylated antimouse immunoglobulin G rabbit serum for 30 minutes, and a streptavidin-biotinylated horseradish peroxidase complex for 30 minutes. Peroxidase staining was performed for 5–10 minutes using a solution of 3,3⬘-diaminobenzidine tetrahydrochloride in 50 mM Tris-HCl (pH 7.5) containing 0.001% hydrogen peroxide. The sections were slightly counterstained with Mayer’s hematoxylin. Nuclear staining of p27Kip1 was scored on a semiquantitative scale (negative, medium, and high) by evaluating the percentage of stained nuclei within representative areas of each tumor. For superficial carci- 2449 nomas, stained sections were observed throughout the lesion. For advanced large tumors, at least 10 fields, including superficial, central, and deep invasive areas, were observed, and the number of stained cells and staining intensity were evaluated. In each field, we counted at least more than 300 cells, using an eyepiece graticule to prevent recounting. Although qualitative differences in staining intensity were observed with considerable intratumoral heterogeneity, all positive cases showed obvious nuclear staining, at least focally. The expression of p27Kip1 was graded as high (over 30% of tumor cells showed strong or diffuse immunopositivity), medium (5–30% of tumor cells showed moderate or patchy immunopositivity), or negative (less than 5% of tumor cells showed weak or focal immunopositivity or no staining). The expression of cyclin E was classified as strongly positive (over 30% of tumor cells showed immunopositivity) or weakly positive/negative. Northern Blot Analysis The expression of p27Kip1 mRNA was examined in OSCC cell lines by Northern blot analysis. RNAs were extracted by using the QuickPrep Micro mRNA Purification Kit (Pharmacia Biotech Inc., CA). Two g of poly(A)⫹-selected RNA were electrophoresed on 1.0% agarose/formaldehyde gels and blotted onto nitrocellulose filter membranes. The filters were baked for 2 hours at 80°C under vacuum and hybridized with 32Plabeled probes using the random hexamer priming method. After hybridization, the filters were washed under stringent conditions and exposed for autoradiography to Fuji RX films with intensifying screens at ⫺80°C.26 A cDNA probe for p27Kip1 (a 0.69 kb Kpn I and BamH I p27Kip1 fragment) was kindly provided by Dr. J. Massague (Howard Hughes Medical Institute, Memorial Sloan-Kettering Cancer Center, New York, NY). A glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA probe was used as an internal control. Western Blot Analysis We examined the expression of p27Kip1 protein in OSCC cell lines by Western blot analysis. The protein samples were prepared, and Western blotting was carried out as we described previously.27 Fifty g of protein was subjected to 10% polyacrylamide gel electrophoresis followed by electroblotting onto a nitrocellulose filter. The antibody was the same as for immunohistochemistry. For detection of the immunocomplex, the ECL Western blotting detection system (Amersham, Aylesbury, UK) was used. 2450 CANCER December 15, 1998 / Volume 83 / Number 12 FIGURE 1. Immunostaining of p27Kip1 is shown in oral normal mucosa, epithelial dysplasia, and oral squamous cell carcinoma. (a) Oral normal mucosa, original magnification ⫻150. (b) Severe dysplasia, original magnification ⫻100. Most epithelial cells showed high levels of p27Kip1 expression. (c) Moderately differentiated carcinoma, original magnification ⫻150. Five percent to 30% of tumor cells showed immunopositivity. (d) Well differentiated carcinoma, original magnification ⫻100. Most tumor cells were negative for p27Kip1. Infiltrating lymphocytes were p27Kip1 positive. Treatment with the Proteosomal Inhibitor NCarbobenzoxy-L-leucyl-L-leucyl-L-norvalinal (LLnV) In human leukemic HL60 cells, proteasomal function was blocked by LLnV and a clear accumulation of p27Kip1 protein was caused by the addition of LLnV.28 Therefore, we used the specific proteasome inhibitor LLnV to examine the proteasome-mediated degradation activity in OSCC cell lines. LLnV was obtained from the Peptide Institute, Osaka. The compound was dissolved in the amount of dimethyl sulfoxide required to establish a stock solution of 10 mM. LLnV was added to the OSCC cells at a final concentration of 50 M, and incubation was continued for 3, 6, 12, and 24 hours. The p27Kip1 levels were examined in lysates of OSCC cells after incubation with LLnV by Western blot analysis. The signal intensity was measured by densitometric scanning, and relative expression levels were presented using that of the control, with 0 h as a standard (1.0). Statistical Analysis Patients survival data were used to determine whether there was a correlation between p27Kip1 expression levels and disease free survival time. Curves for survival were drawn according to the Kaplan–Meier method, and differences between the survival rates of three groups with different p27Kip1 protein and cyclin E expression were examined. The statistical significance of these data was determined by the Mantel– Cox test. A possible correlation between variables of the analyzed tumor samples were tested for association by Fisher’s exact test. A P value of ⬍0.05 was required for significance. Reduced Expression of p27Kip1 in OSCC/Kudo et al. 2451 TABLE 2 Expression of p27 in Normal Oral Mucosa, Epithelial Dysplasia, and OSCC and Its Correlation with Clinicopathologic Parameters Expression of p27b Normal Epithelial dysplasia Mild or moderate Severe OSCC Histologya Well differentiated Moderately differentiated Poorly differentiated Stagea 1 2 3 4 Metastasis Negative Positive Cyclin E expression Weakly positive/negative Strongly positive P valuec No. of cases High Medium Negative 20 20 (100%) 0 (0%) 0 (0%) 16 6 70 13 (81%) 3 (50%) 9 (13%) 3 (19%) 3 (50%) 33 (47%) 0 (0%) 0 (0%) 28 (40%) 30 33 7 5 (17%) 4 (12%) 0 (0%) 16 (53%) 15 (45.5%) 2 (29%) 9 (30%) 14 (42.5%) 5 (71%) 15 19 2 34 3 (20%) 3 (16%) 1 (50%) 2 (6%) 10 (67%) 11 (58%) 1 (50%) 11 (32%) 2 (13%) 5 (26%) 0 (0%) 21 (62%) 34 36 6 (17.5%) 3 (8%) 21 (62%) 12 (33%) 7 20.5%) 21 (58%) 0.0055 38 13 2 (5%) 3 (23%) 20 (53%) 3 (23%) 16 (42%) 7 (54%) 0.0704 ⬍0.0001 0.3360 0.0149 OSCC: oral squamous cell carcinoma. a According to the criteria of the Japanese Classification of Head and Neck Cancer.25 Stage grouping as follows: Stage I includes T1, N0, M0; Stage II includes T2, N0, M0; Stage III includes T3, N0, M0 or T1–T3, N1, M0; Stage IV includes T4, N0–1, M0, any T, N2–3, M0, or any T, any N, M1. b Grades of p27 positive cells were classified as high (⬎30%), medium (5–30%), or negative (⬍5%). c Correlation was analyzed by Fisher’s exact test, and P values are shown. P values less than 0.05 were regarded as statistically significant. RESULTS Immunohistochemical Analysis of p27Kip1 in OSCC In the normal oral epithelium, cells in the prickle cell and granular cell layers showed strongly positive staining for p27Kip1 in their nuclei, but cells in the basal layer did not (Fig. 1a). p27Kip1 protein was also consistently positive in inflammatory cells, fibroblasts, muscle cells, and endothelial cells. p27Kip1 staining in these components could be regarded as an internal control for the immunohistochemistry. The incidence of p27Kip1 expression in normal oral mucosa, epithelial dysplasia, and OSCC is summarized in Table 2. In epithelial dysplasia, 81% of mild-to-moderate dysplasia specimens (13 of 16) and 50% of severe dysplasia specimens (3 of 6) expressed p27Kip1 at high levels (Table 2, Fig. 1b). Three cases (19%) of mild-tomoderate dysplasia and half of severe dysplasia cases showed decreased p27Kip1 immunoexpression. However, none of the dysplasia cases were negative. In contrast to epithelial dysplasia, 40% of OSCCs (28 of 70) were negative and only 9 (13%) showed p27Kip1 staining at high levels. High p27Kip1 expression was frequently found in early stage OSCCs without metastasis. Reduced expression of p27Kip1 was associated with ad- vanced stage and metastasis, and there was a significant correlation between them (P ⬍ 0.05) (Table 2, Fig. 1c,d). In comparison with well-differentiated carcinomas, poorly differentiated carcinomas tended to show reduced expression of p27Kip1, but there was no statistically significant correlation. Though we also examined the influence of age, gender, and smoking habits of OSCC patients on the expression of p27Kip1, there were no correlations among them. The expression of p27Kip1 in histologic sections was heterogenous in some cases, and there was a tendency to express less p27Kip1 in the invasive front. Metastatic lesions also expressed less p27Kip1 (data not shown). As we reported previously, for gastric carcinoma there was an inverse correlation between the expression of p27Kip1 and that of cyclin E.24 Therefore, we also studied the expression of cyclin E in OSCC (Table 3). Strong expression of cyclin E was found in 13 (25%) of 51 OSCCs. Expression of cyclin E was not found in most normal oral mucosa specimens (data not shown). Compared with poorly differentiated carcinomas, well-differentiated carcinomas tended to show strong expression of cyclin E, but the expression of cyclin E was not associated with advanced stage and 2452 CANCER December 15, 1998 / Volume 83 / Number 12 TABLE 3 Expression of Cyclin E in OSCC and Its Correlation with Clinicopathologic Parameters Expression of cyclin Eb Histologya Well differentiated Moderately differentiated Poorly differntiated Stagea I II III IV Metastasis Negative Positive No. of cases Strongly positive Weakly positive/ negative 22 25 4 7 (32%) 6 (24%) 0 (0%) 15 (68%) 19 (76%) 4 (100%) 9 15 1 26 3 (33%) 2 (20%) 0 (0%) 7 (27%) 6 (67%) 12 (80%) 1 (100%) 19 (73%) 24 27 6 (25%) 7 (26%) 18 (75%) 20 (74%) OSCC: oral squamous cell carcinoma. a According to the criteria of the Japanese Classification of Head and Neck Cancer.25 b Grades of cyclin E expression were classified as strongly positive (⬎30%). metastasis (Table 3). There was no inverse correlation between p27Kip1 and cyclin E in OSCC (Table 2). The correlation between p27Kip1 protein expression and the survival rates of 51 OSCC patients with follow-up data was also examined. Figure 2 shows the Kaplan–Meier survival curves of the patients grouped by the immunoreactivity of p27Kip1 in their tumors. The cumulative survival rate for patients who were negative for expression of p27Kip1 was lower than that for patients with medium-to-high expression of p27Kip1, and there was a statistical significance between p27Kip1 expression and the survival rate. High levels of cyclin E and low levels of p27Kip1 were strongly predictive of increased mortality among patients with breast carcinoma.19 However, in the current study, the survival rate for patients who had strongly positive expression of cyclin E and who were negative for expression of p27Kip1 was higher than the overall rate for patients who were negative for expression of p27Kip1. Expression of p27Kip1 in OSCC Cell Lines Expression of p27Kip1 mRNA and protein in 6 OSCC cell lines is shown in Figure 3a. The same filter was reprobed with a GAPDH cDNA probe. Most of these cell lines expressed p27Kip1 mRNA at high levels. We examined expression of p27Kip1 protein in the 6 cell lines by Western blot analysis (Fig. 3b). Ca9-22 and HSC4 cells expressed p27Kip1 protein at higher levels, and HSC3 and Ho-1-U-1 cells expressed p27Kip1 protein at lower levels. p27Kip1 protein was not found in HSC2 or Ho-1-N-1. The expression levels of p27Kip1 protein did not correspond with the results of Northern blot analysis. We also searched for macroscopic alterations of the p27Kip1 genes in these cell lines by Southern blot analysis. Neither gene amplification nor rearrangement of the p27Kip1 genes was found in any of these cell lines (data not shown). Accumulation of p27Kip1 after Treatment with the Proteosomal Inhibitor LLnV in HSC2 and Ca9-22 We examined the p27Kip1 levels in HSC2 and Ca9-22 cells after incubation with LLnV by Western blot analysis. As shown in Figure 3, HSC2 cells lacked p27Kip1 protein and Ca9-22 cells preserved p27Kip1 protein, whereas both expressed p27Kip1 mRNA at high levels. LLnV induced the expression of p27Kip1 protein in HSC2 cells, whereas it did not do so in Ca9-22 cells (Fig. 4). In HSC2 cells, the signal intensity of p27Kip1 protein became larger with time after the treatment with LLnV, and at 24 hours was about 5 times as much as that at 0 h. DISCUSSION The proliferation and progression of cancer cells may be caused by abnormalities of various positive and negative cell cycle regulators.1,2 Reduced expression of p27Kip1 was recently found in various cancers,19 –22,24 and so we focused on the expression of p27Kip1 in OSCC. In this study, we demonstrated the reduced expression of p27Kip1 in 87% of OSCCs as described in breast, gastric, and colorectal carcinomas.19 –22,24 The reduced expression of p27Kip1 was observed in 56% of breast carcinomas, 70% of colorectal carcinomas, and 74% of gastric carcinomas.20,21,24 The reduced expression of p27Kip1 may be a common event in various carcinomas. Normal oral mucosa specimens and 73% of epithelial dysplasia specimens expressed p27Kip1 at high levels, but only 13% of OSCCs expressed this protein immunohistochemically. Moreover, 81% of mild-tomoderate dysplasia specimens showed high p27Kip1 expression in comparison with 50% of severe dysplasia specimens. However, negative expression was not found in epithelial dysplasia. These findings suggest that the reduction of p27Kip1 protein expression may begin at an early stage in the development of OSCC, but it is necessary to confirm this in a large number of epithelial dysplasia cases. Epithelial dysplasia with reduced expression of p27Kip1 may indicate the possibility of progression to cancer. The most interesting finding of the current study is that the immunoexpression of p27Kip1 protein was closely associated with tumor stage and metastasis. Reduced Expression of p27Kip1 in OSCC/Kudo et al. 2453 FIGURE 2. The relation of p27Kip1 and cyclin E expression to survival is shown. Kaplan–Meier plots for disease free survival show the association of survival and expression of cyclin E and p27Kip1 in patients with oral squamous cell carcinoma. The prognoses for patients who were negative for expression of p27Kip1 (n ⫽ 23; median survival ⫽ 57 months) was poorer than the prognoses for patients with medium-to-high expression of p27Kip1 (n ⫽ 28; median survival ⫽ 110 months) or those with high expression of cyclin E who were negative for expression of p27Kip1 (n ⫽ 7; median survival ⫽ 79 months). The statistical significance of these data was measured by the Mantel–Cox test. FIGURE 3. Expression of p27Kip1 mRNA and protein in oral squamous cell carcinoma cell lines is shown. (a) Northern blot analysis: Two g of poly (A)⫹-selected RNA was subjected to Northern blot analysis as described in “Materials and Methods.” A Glyceraldehyde-3-phosphate dehydrogenase cDNA probe was applied as an internal control. Most of these cell lines expressed p27Kip1 mRNA at various levels. (b) Fifty g of protein was subjected to Western blot analysis as described in “Materials and Methods.” Ca9-22 and HSC4 cells expressed p27Kip1 at higher levels, but the rest showed reduced expression of p27Kip1 protein. The degree of reduction is increased in line with tumor progression. The incidence of p27Kip1 positive tumors was significantly lower in Stage IV cases and cases with metastasis. Furthermore, most metastatic tumors also showed the same or reduced immunoreactivity as that of primary tumors (data not shown). It has been reported that antisense oligonucleotide–mediated down-regulation of p27Kip1 in EMT-6 mammary tumor cell spheroids reduced intercellular adhesion and increased cell proliferation.29 We suppose that reduced expression of p27Kip1 may be related to the loss of intercellular adhesion and result in metastasis. Interaction between expression of p27Kip1 and adhesion molecules was suggested, but the mechanism of p27Kip1-mediated metastasis is unknown.30 Figure 2 shows that the prognoses of patients with negative expression of p27Kip1 were poorer than those of patients with medium-to-high expression of p27Kip1. The prognoses of OSCC patients with high expression of cyclin E and negative expression of p27Kip1 were not poor, and this result was different from the result for breast carcinoma.19 In breast carcinoma, most of high grade tumors were found to show cyclin E overexpression, and overexpression of cyclin E correlated with increasing tumor stage and grade.18 The overexpression of cyclin E did not correlate with various parameters for malignancy of OSCC (Table 3), and it seemed to be a major reason for this result. The correlation between p27Kip1 and cyclin E in OSCC cells should be examined in a large number of cases. The reduced expression of p27Kip1 protein was detected in OSCC cell lines, whereas most of them preserved the expression of p27Kip1 mRNA and none of them had macroscopic alterations of the p27Kip1 gene (Fig. 3). This was consistent with the findings reported for gastric, breast, and colorectal carcinomas.19 –22,24 As reduced expression of p27Kip1 protein was not brought about by reduced expression of p27Kip1 mRNA or gene alteration, we examined proteasome-mediated degradation of p27Kip1 protein. We showed that accumulation of p27Kip1 protein was originally found after treatment with proteasome inhibitor LLnV in 2454 CANCER December 15, 1998 / Volume 83 / Number 12 8. 9. 10. 11. 12. 13. FIGURE 4. Treatment of HSC2 and Ca9-22 cells with the proteasome inhibitor LLnV is shown. Fifty g of protein was obtained from the cells after treatment with 50 M LLnV for the indicated periods and analyzed as in Figure 3. LLnV induced the expression of p27Kip1 protein in HSC2 cells, whereas it did not do so in Ca9-22 cells. 14. HSC2 cells without p27Kip1 protein expression (Fig. 4). Our results suggest that the discrepancy between the expression of mRNA and protein is associated with increased proteasome-mediated degradation at posttranslational levels. We should examine whether the reduced expression of p27Kip1 is really caused by proteasome-mediated degradation in OSCC tissues. Reduced expression of p27Kip1 could predict malignant behavior in OSCC. As it is difficult to predict the survival of patients with OSCC, we should examine the significance of p27Kip1 as a prognostic factor. In OSCC, inhibition of proteasome-mediated degradation of p27Kip1 protein may be used as a possible target of novel cancer therapy. 17. REFERENCES 1. 2. 3. 4. 5. 6. 7. Sherr CJ. G1 phase progression: cycling on cue. Cell 1994; 79:515–55. Hunter T, Pines J. Cyclins and cancer II: cyclin D and CDK inhibitors come of age. Cell 1994;79:573– 82. Sherr CJ. Mammalian G1 cyclins. Cell 1993;73:1059 – 65. El.-Deiry WAS, Tokino T, Velculescu VE, Levy DB, Persons R, Trent JM, et al. WAF1, a potential mediator of p53 tumor suppression. Cell 1993;75:817–25. Harper JW, Adami GR, Wei N, Keyomarsi K, Elledge SJ. The p21 Cdk-interacting protein Cip1 is a potent inhibitor of G1 cyclin-dependent kinases. Cell 1993;75:805–16. Polyak K, Kato JY, Solomon MJ, Sherr CJ, Massague J, Roberts JM, et al. p27Kip1, a cyclin-Cdk inhibitor, links transforming growth factor-␤ and contact inhibition to cell cycle arrest. Genes Dev 1994;8:9 –22. Toyoshima H, Hunter T. p27, a novel inhibitor of G1 cyclin- 15. 16. 18. 19. 20. 21. 22. 23. 24. 25. Cdk protein kinase activity, is related to p21. Cell 1994;78: 67–74. Kamb A, Gruis NA, Weaver-Feldhaus J, Liu Q, Harshman K, Tavtigian SV, et al. A cell cycle regulator potentially involved in the genesis of many tumor types. Science 1994;264:436 – 40. Hannon GJ, Beach D. p15INK4B is a potent effector of TGF␤-induced cell cycle arrest. Nature 1994;371:257– 61. Nobori T, Miura K, Wu DJ, Lois A, Takabayashi K, Carson DA. Delations of the cyclin-dependent kinase-4 inhibitor gene in multiple human cancers. Nature 1994;368:753– 6. Serrano M, Hannon GJ, Beach D. A new regulatory motif in cell cycle control causing specific inhibition of cyclin D/CDK4. Nature 1993;366:704 –7. Polyak K, Lee MH, Erdjument-Bromage H, Tempst P, Massague J. Cloning of p27Kip1, a cyclin-Cdk inhibitor and potential mediator of extracellular antimitogenic signals. Cell 1994;78:59 – 66. Hatakeyama M, Brill JA, Fink GR, Weinberg RA. Collaboration of G1 cyclins in the functional inactivation of the retinoblastoma protein. Genes Dev 1994;8:1759 –71. Hinds P, Mittnacht S, Dulic V, Arnold A, Reed SI, Weinberg RA. Regulation of retinoblastoma protein functions by ectopic expression of human cyclins. Cell 1992;70:993–1006. Weinberg RA. The retinoblastoma protein and cell cycle control. Cell 1995;81:323–30. Keyomarsi K, Pardee AB. Redundant cyclin overexpression and gene amplification in breast cancer cells. Proc Natl Acad Sci U S A 1993;90:1112– 6. Leach FS, Elledge SJ, Sherr CJ, Willson JKV, Markowitz S, Kinzler KW, et al. Amplification of cyclin genes in colorectal carcinomas. Cancer Res 1993;53:1986 –9. Keyomarsi K, O’Leary N, Molnar G, Less E, Fingert HJ, Pardee AB. Cyclin E, a potential prognostic marker for breast cancer. Cancer Res 1994;54:380 –5. Porter PL, Malone KE, Heagerty PJ, Alexander GM, Gatti LA, Firpo EJ, et al. Expression of cell-cycle regulators p27Kip1 and cyclin E, alone and in combination, correlates with survival in young breast cancer patients. Nat Med 1997;2:222–5. Catzavelos C, Bhattacharya N, Ung YC, Wilson JA, Roncari L, Sandhu C, et al. Decreased levels of the cell-cycle inhibitor p27Kip1 protein: prognostic implications in primary breast cancer. Nat Med 1997;2:227–30. Loda M, Cukor B, Tam SW, Lavin P, Fiorentino M, Draetta GF, et al. Increased proteasome-dependent degradation of the cyclin-dependent kinase inhibitor p27 in aggressive colorectal carcinomas. Nat Med 1997;2:231– 4. Fredersdorf S, Burns J, Milne AM, Packham G, Fallis L, Gillett CE, et al. High level expression of p27Kip1 and cyclin D1 in some human breast cancer cells: inverse correlation between the expression of p27Kip1 and degree of malignancy in human breast and colorectal cancers. Proc Natl Acad Sci U S A 1997;94:6380 –5. Pagano M, Tam SW, Theodoras AM, Beer-Romero P, Del Sal G, Chau V, et al. Role of the ubiquitin-proteasome pathway in regulating abundance of the cyclin-dependent kinase inhibitor p27. Science 1995;269:682–5. Yasui W, Kudo Y, Semba S, Yokozaki H, Tahara E. Reduced expression of cyclin-dependent kinase inhibitor p27Kip1 is associated with advanced stage and invasiveness of gastric carcinomas. Jpn J Cancer Res 1997;88:625–9. Japan Society for Head and Neck Cancer. General rules for clinical and pathological studies on head and neck cancer. 2nd edition. Tokyo: Kanehara, 1991. Reduced Expression of p27Kip1 in OSCC/Kudo et al. 26. Yasui W, Kuniyasu H, Akama Y, Kitahara K, Nagafuchi A, Ishihara S, et al. Expression of E-cadherin, a- and b-catenins in human gastric carcinomas: correlation with histology and tumor progression. Oncol Rep 1995;2:111–17. 27. Yasui W, Ayhan A, Kitadai Y, Nishimura K, Yokozaki H, Ito H, et al. Increased expression of p34cdc2 and its kinase activity in human gastric and colonic carcinomas. Int J Cancer 1993; 53:36 – 41. 28. Drexler HCA. Activation of the cell death program by inhi- 2455 bition of proteasome function. Proc Natl Acad Sci U S A 1997;94:855– 60. 29. St Croix S, Florenes VA, Rak JW, Flanagan M, Bhattacharya N, Slingerland JM, et al. Impact of the cyclin-dependent kinase inhibitor p27Kip1 on resistance of tumor cells to anticancer agents. Nat Med 1996;2:1204 –10. 30. Fang F, Orend G, Watanabe N, Hunter T, Ruoslahti E. Dependence of Cyclin E-CDK2 kinase activity on cell anchorage. Science 1996;271:499 –502.