Expression of cyclin D1 correlates with malignancy in human ovarian tumours.

Cyclin D1 is a cell cycle regulator of G1 progression that has been suggested to play a relevant role in the pathogenesis of several human cancer types. In the current study, the expression of cyclin D1 has been investigated in a series of 33 patients, with benign (10 patients), borderline (five patients) and malignant (18 patients) ovarian disease. Cyclin D1 protein and mRNA content were analysed by Western blotting and reverse transcriptase polymerase chain reaction respectively. The levels of cyclin D1 protein were undetectable in patients with benign disease, detectable in the majority of patients with borderline disease and elevated in those with ovarian carcinomas, being significantly related to the degree of malignancy (carcinoma vs benign, P = 0.0001; benign vs borderline, P = 0.0238). A significant relationship between cyclin D1 expression and tumour proliferative activity was also found (P = 0.000001). Moreover, eight benign lesions, two borderline tumours and 11 carcinomas proved to be suitable for the analysis of cyclin D1 transcript, and emerging data demonstrated significant agreement between protein abundance and mRNA expression. Results from the current study suggest that cyclin D1 expression is associated with the degree of transformation and most probably plays a role in the early development of ovarian malignancy. ImagesFigure 3Figure 4Figure 5Figure 6

Neoplastic cells are characterized by altered mechanisms of cell growth and DNA replication. Recent advances in eukaryotic cell cycle research have demonstrated the relevance of cell cycle checkpoints (G,/S and G2/M transitions) in the control of progression along the proliferative cycle (Peeper et al, 1994). The GI/S checkpoint is currently the best understood in mammalian cells; an important group of molecules involved in this checkpoint control has been identified in cyclins and their catalytic partners; these are the so-called cyclin-dependent kinases (CDKs). Several studies have reported that the abnormal expression of certain cyclins may be associated with oncogenesis and cancer progression (Motokura et al, 1993). To date, at least eight mammalian cyclin genes have been identified (A, B, C, D 1-3, E, H) on the basis of their different patterns of expression in the phases of the cell cycle (Hunter and Pines, 1994). Among the members of the cyclin gene family, cyclin Dl is the most strongly implicated in human tumorigenesis (Hinds et al, 1994). Cyclin Dl is a GI-specific protein essential for the progression through GU phase to S-phase (Baldin et al, 1993); its expression and activity reach a peak in GI and gradually decline in S-phase . Cyclin DI associates with the CDK4 subunit to form a complex, activated by phosphorylation, that leads to the transition from G, phase into S-phase (Pines, 1993). The cyclin DI locus has been mapped to the chromosome 11 band q13, and amplifications of the 1 1q13 region, as well as the cyclin Dl gene as a component of this amplicon, have been observed in a variety of human carcinomas (Lammie and Peters, 1991). The indication that the cyclin Dl gene is somehow centrally relevant to cancer, functioning as an oncogene, is supported by several experimental observations (Jiang et al, 1993a;Lovec et al, 1994). Altered expression of cyclin Dl may result from rearrangement, isolated as PRAD-1 in parathyroid adenomas (Rosenberg et al, 1991), translocation, isolated as bcl-l in B-lymphocytic malignancies (Motokura et al, 1991;Withers et al, 1991), and amplification, such as in head and neck, breast and squamous cell carcinomas (Lammie and Peters, 1991;Schuuring et al, 1992).
Investigations on the cyclin Dl gene expression in human ovarian tumours are lacking and, at present, no significant data are available. The natural history and management of ovarian cancer have been extensively studied and different prognostic indicators have been identified (Perez et al, 1991). Biological factors, such as DNA ploidy, cell kinetics and oncogene expression, have been demonstrated to be extremely helpful in predicting tumour aggressiveness, clinical outcome and response to chemotherapeutic agents. Among them, tumour proliferative activity, evaluated by the thymidine labelling index (TLI), represents an important prognostic factor able to estimate tumour aggressiveness and chemosensitivity of ovarian neoplastic cells (Alama et al, 1994).
The aim of the current study is to investigate the expression of cyclin Dl in human ovarian lesions and its relationship with transformed phenotype and tumour kinetics.

MATERIALS AND METHODS Patients and tumour samples
Thirty-three patients who presented at the Department of Obstetrics and Gynecology of the University of Genoa entered this study from December 1994 to December 1995. Tissue samples were obtained at the time of primary surgery. Specimens consisted of 10 benign lesions, five borderline tumours and 18 ovarian carcinomas. Tumour histology is reported in Table 1. FIGO stage  (Ulfelder et al, 1978) was available for 16 of the 18 malignant tumours, and the group consisted of one stage I, one stage II, 11 stage III and three stage IV tumours. Data on histological grade, available for 16 of these patients, identified two well-differentiated, two moderately differentiated and 12 poorly differentiated tumours. Adequate material from fresh ovarian tissues or ascites were immediately processed for cyclin Dl expression and tumour kinetic analyses.
Thymidine labelling index (TLI) Cell kinetics, as the percentage of thymidine-labelled tumour cells in DNA synthesis, was evaluated in surgical samples or ascites. Solid specimens were grossly mechanically disaggregated with scissors and scalpel to obtain single-cell suspensions. Cells were resuspended in RPMI 1640 medium (ICN) supplemented with 10% fetal calf serum (FCS, ICN) and labelled with 10 ,uCi ml-' [3H]thymidine (specific activity 5 Ci mM-', Amersham) at 37'C for 30 min. Radiolabelling was stopped with ice-cold phosphatebuffered saline (PBS), and cells were cytocentrifuged onto slides and fixed in methanol-acetic acid 3:1. Slides were then dipped in NTB-2 Nuclear Track Emulsion (Kodak) and exposed at 4°C for 24 h; haematoxylin-eosin staining was performed after gold-activated autoradiography (Braunschweiger et al, 1976). Cells containing more than five nuclear grains, over background, were scored as labelled, and the proportion of labelled tumour cells expressed as a percentage of the total cell population represents the TLI value.

Western blot analysis
Cell suspensions were washed twice in cold PBS and dissolved in lysis buffer (1% Triton X-100, 0.15 M sodium chloride, 10 mM Tris, pH 7.4) containing protease inhibitors (50 ,ug ml-' phenylmethylsulphonyl fluoride, 2 jig ml-' aprotinin, 2 jig ml-' leupeptin) at 4°C for 30 min. The protein concentration of cell lysates was determined by the Bio-Rad Protein Assay kit using bovine serum albumin (BSA) as a standard. Equal amounts of total proteins (10 gg) were separated on a 12% polyacrylamide gel in the presence of sodium dodecyl sulphate (SDS) (SDS-PAGE). After running, gel was electrotransferred onto a nitrocellulose membrane (Hybond C-Extra, Amersham). Filters were blocked with 5% BSA (Sigma) in PBS at room temperature for 1 h. After washes in PBS-T (0.05% Tween 20), blots were incubated with the cyclin DI monoclonal antibody (Santa Cruz Biotechnology) at 1:1000 dilution in PBS-T at room temperature for 2 h. After three washes in PBS-T, the filter was incubated with a 1:5000 dilution of horseradish peroxidase (HRP)-linked anti-mouse IgG secondary antibody (Dako) at room temperature for 1 h. After three washes, the cyclin DI band (36 kDa) was visualized by chemiluminescent detection (ECL, Amersham), following the supplier's recommended procedures. Prestained low molecular weight markers were used as reference.
Reverse transcriptase polymerase chain reaction (RT-PCR) analysis Specific cyclin DI transcript levels were determined by a semiquantitative RT-PCR amplification. Total RNA from fresh ovarian samples was isolated using the RNAzol B method (Biotecx Lab). Reverse transcription (RT) was carried out on 1 jig of total RNA (at 42°C for 1 h) using the first-strand cDNA synthesis kit (Clontech), according to the manufacturer's suggestions. RT was inactivated by incubating the tubes at 94°C for 5 min. Synthetic primers, to amplify specific mRNAs, were purchased from TIB-MolBiol.
The co-amplification of the cyclin Dl and P-actin cDNAs was carried out under the following conditions: 94°C for 1 min (denaturation), 65°C for 1.5 min (annealing), 72°C for 1.5 min (chain extension) for 30 cycles, followed by a final incubation at 72°C for 15 min to flush-end the amplified fragments. Before the first cycle, tubes were soaked at 94°C for 5 min. RT-PCR amplified fragments (cyclin DI, 514 bp; P-actin, 600 bp) were analysed by 2% agarose British Journal of Cancer (1997) 75(9) . Cancer Research Campaign 1997 gel electrophoresis and visualized by ethidium bromide staining. Resulting bands were quantified and the relative amount of cyclin Dl mRNA in each sample was estimated by normalization to the P-actin mRNA detected in the same sample.

Densitometric analysis
Quantitative determination of the cyclin Dl protein and mRNA bands were performed by scanning densitometry using an Ultroscan XL densitometer (Pharmacia-LKB). The intensities of the bands from radiographs were scanned and the resulting peak areas, related to the absorbance, were determined.
The association between cyclin Dl densitometric values and tumour types was investigated using the Kruskal-Wallis statistic and was tested for significance (P < 0.05).
The correlations between cyclin Dl densitometric scores and TLI, as well as cyclin Dl protein and mRNA levels, were evaluated using the Spearman rank test.
Statistical analyses were carried out using the Statistica 3.Ob program from StatSoft.

RESULTS
Ten samples of non-malignant lesions, five borderline tumours and 18 carcinomas of the ovary were obtained at the time of primary surgery. Table 1 outlines the histological types and the cyclin Dl expression of the patients investigated. The most direct approach to search for cyclin Dl abnormalities is to examine the protein abundance; thus, cyclin Dl expression at the protein level was evaluated by Western blotting analysis. Total cell lysates were obtained from fresh solid tumours or ascites. To quantitate the abundance of cyclin Dl, densitometric scanning of the 36-kDa band was performed. The median densitometric value of cyclin Dl for the benign group was markedly lower (0.001, range 0.001-0.387) than that of the borderline (0.209, range 0.001-0.348) and malignant group (0.388, range 0.027-2.260). The analysis of the correlation between cyclin Dl densitometric values and tumour types was performed by the Kruskal-Wallis non-parametric test. As shown in Figure 1, comparison of the groups indicates that the cyclin Dl protein content was significantly higher in the carcinoma group than in the benign group (P = 0.0001) and higher in the borderline group than in the benign group (P = 0.0238), but the difference did not reach statistical significance when comparing the carcinoma group with the borderline group (P = 0.0624). As cyclin Dl is related to the proliferative status of the cells, a kinetic analysis of tumour samples was carried out. A [3H]thymidine incorporation assay (TLI) was used to estimate the percentage of tumour cells in DNA synthesis. Benign lesions displayed lower TLI values (median TLI 0.20%, range 0.01-0.40) than borderline tumours (median TLI 0.90%, range 0.60-1.70) and carcinomas (median TLI 4.80%, range 0.01-22.50). Furthermore, a significant association between cyclin Dl expression and TLI, as assessed by the Spearman rank test (r = 0.732, P = 0.000001) was found (Figure 2).
When adequate material could be obtained from fresh tumour specimens, cyclin Dl mRNA levels were evaluated amplifying the specific cyclin Dl fragment (514 bp) by RT-PCR analysis. This As shown in Table 1, all the carcinomas exhibited detectable to highly detectable cyclin Dl transcripts compared with the undetectable signals of the benign lesions; in addition, cyclin Dl mRNA expression coincided with or exceeded protein abundance, except in three cases. A significant agreement between the two sets of data wvas detenmined by Spearman statistical analysis (P = 0.0004), suggesting that increased levels of cycin DI transcript and protein might be a feature of ovarian malignancy in this preliminary series of 21 patients.
In order to confirm that overexpression of cyclin Dl in ovarian cancer cells was not only a consequence of the proliferative status in malignant cells, cyclin Dl levels in exponential-growth cultures were compared with those of quiescent cells. Three ovarian cancer cell lines, established from the ascites of three patients included in the present study (patient nos. 16, 17 and 18 as in Table 1), were used in a serum-starvation assay. These cell lines, named OC 314 (patient no. 16), OC 315 (patient no. 17) and OC 316 (patient no. 18), have been extensively characterized as reported by Alama et al (1996).
In addition, the MCF-7 breast cancer cell line, used as reference for cyclin DI overexpression (Buckley et al, 1993), was also included in the experiments. Cell cultures were either grown  Table 2. A marked decrease in DNA synthesis was observed in all three cell lines after 3 days of serum reduction. Collectively, the densitometric values of cyclin DI obtained from mRNA and protein analyses were retained in the range of the well detectable levels (++; range 0.209-0.696); the mRNA expression remained constant in both starved and exponential cultures, while a slight protein content variability, probably because of a different intracellular accumulation and/or degradation of the protein, was reported. These data suggest that altered cyclin Dl overexpression could be a relatively stable feature in malignant cells, at least in these cultures, independent of their proliferative behaviour.

DISCUSSION
Various genetic alterations have been described in human ovarian cancer, such as amplification of the oncogenes HER-2/neu, K-ras and c-myc, as well as the mutation and overexpression of the p53 tumour-suppressor gene (Piver et al, 1991). Furthermore, chromosome aberrations, including translocations (Lee et al, 1990), rearrangements (Pejovic et al, 1992) and aneuploidy (Rodenburg et al, 1988), have also been reported. As a consequence, concurrent genomic abnormalities could result in alterations of the mechanisms controlling cellular replication at the molecular level, leading to tumour promotion and progression. Although amplification and overexpression of the cyclin D1 gene have been demonstrated in various carcinomas (breast, urinary bladder, head and neck, lung and oesophageal) and play a role in multistep carcinogenesis, data concerning cyclin D1 expression in ovarian cancer are presently lacking.
In the current paper, we reported that the malignant lesions studied differed markedly from the benign lesions in their growth behaviour; moreover, the proliferative activity in ovarian tumours was positively related to the levels of cyclin DI protein, with Dl being undetectable in benign lesions and highly expressed in carcinomas. Previous studies on cell proliferative activity and DNA flow cytometry of ovarian cancer have demonstrated that highly proliferating and aneuploid cancers are associated with tumour aggressiveness and poor prognosis (Rodenburg et al, 1988;Alama et al, 1994). Therefore, the concomitant analyses of a cell cyclerelated gene, such as cyclin Dl, with tumour kinetics might provide additional insight into the biology and clinical aggressiveness of this disease.
The relationship between TLI and cyclin DI expression in the 33 patients with ovarian tumours in this study demonstrated that higher TLI values were significantly associated with cyclin Dl abundance in neoplastic cells, while lower proliferative activity and undetectable cyclin Dl levels were observed in histologically benign ovarian tissues.
The observation that expression of cyclin Dl is frequently related to increased DNA synthesis is in agreement with data from the literature reporting that G, cyclins are rate controlling for G1 duration and S progression. Indeed, overexpression of G, cyclins in human fibroblasts 'in vitro' accelerated G, progression, shortening G, phase length and reducing serum requirement for the transition from G, to S . Similar results were reported in a paper by Jiang et al (1993b) in which retrovirustransducted Rat 6 embryo fibroblasts, which stably overexpress the human cyclin Dl cDNA, displayed abnormalities in cell cycle and growth control, together with a decrease in the duration of the G, phase which resulted in shortening their rates of GI to S-phase transition. At the clinical level, some preliminary studies have shown that cyclin Dl could represent a feature of malignancy with prognostic significance. One of these reports showed that aberrant expression of cyclin DI is involved in human hepatocarcinogenesis and, in a subset of five patients with stage IV hepatocellular carcinomas (HCCs), seemed to be associated with faster tumour growth and aggressive behaviour than early-stage disease, for which no alterations of the cyclin Dl gene were detected (Nishida et al, 1994). Moreover, the prognostic significance of cyclin DI abnormalities was investigated in squamous cell carcinomas of the head and neck (HNSCC) in a study by Michalides et al (1995). The study showed that cyclin Dl overexpression, detected by immunohistochemistry in a retrospective series of 47 operable HNSCC, correlated with rapid cancer recurrence and a poor survival. In addition, as no correlations between cyclin Dl expression and other clinical features (tumour size, stage) were reported, it was suggested that cyclin D 1 could represent an independent prognostic marker.
The pathogenetic relevance of cyclin Dl abundance in breast cancer has also been described in detail. One of the first immunohistochemical studies (Bartkova et al, 1994) analysing the cyclin Dl protein in human samples indicated significant differences in the staining intensities between normal tissues and breast carcinomas; 37% of carcinomas revealed a marked detectable cyclin Dl nuclear signal, while normal breast tissues were mainly negative. In addition, a more recent paper by Bartkova et al (1995), including a larger number of tumours (breast, colorectal, uterine, melanoma and soft tissue sarcoma) and their normal counterparts, showed that cyclin Dl immunostaining in carcinomas ranged from weak to high according to the degree of malignancy. The authors concluded that alterations of cyclin Dl expression represented a common feature of malignancy in different human cancers.
The present paper suggests that increased expression of cyclin Dl and high percentage of S-phase cells might be related to the degree of malignancy in ovarian cancer. However, such observations can not exclude that proliferative abnormalities might influence cyclin Dl expression in ovarian neoplastic cells. Although further investigations are needed to precisely understand the role of cyclin Dl in the mechanisms of tumorigenesis, the present study could provide information on cyclin D1 expression as a proliferation marker in benign and borderline ovarian lesions, thus helping identify subsets of patients at increased risk of developing ovarian carcinoma.