KPL-1 is a human breast carcinoma cell line that was established from the malignant effusion of a breast cancer patient 22. This cell line is estrogen receptor positive, grows rapidly in female athymic mice, and often causes regional lymph node metastasis when inoculated into the mammary fat pad. KPL-1 cells were maintained in Dulbecco's modified Eagle's minimum essential medium (DMEM 0/5921; Sigma, St. Louis, MO, USA) supplemented with 10% fetal bovine serum (FBS; Gibco BRL, Grand Island, NY, USA), and were grown at 37°C in 5% carbon dioxide/95% humidified air.

DHA ethyl ester (purity 97%) was obtained from Bizen Chemical (Okayama, Japan). CDHA was prepared by alkaline treatment 21. Briefly, a solution of potassium hydroxide in ethylene glycol was prepared at a concentration of 21% (weight:weight), and this solution was bubbled with nitrogen gas. Then, 100 g DHA ethyl ester was added to 1000 ml of the 21% potassium hydroxide/ethylene glycol solution. The mixture was bubbled with nitrogen gas and allowed to stand for 10 min at 180°C. The reaction mixture was cooled and 1000 ml methanol was added. The mixture was acidified with 2000 ml of 6 N HCl. After dilution with 2000 ml distilled water, the conjugated fatty acid was extracted with 5000 ml n-hexane. The hexane extract was then washed with 3000 ml of 30% methanol/distilled water, washed with 3000 ml distilled water, and finally evaporated under the nitrogen gas stream to produce CDHA. DHA isomerized with 21% potassium hydroxide for 10 min preferentially forms conjugated pentaene and conjugated hexaene 21. The purity of CDHA used in the following experiments was approximately 65% (35% was unreacted and remained as DHA). DHA and CDHA were stored at -80°C in the dark.

Viable cells exposed to DHA or CDHA were quantified using a colorimetric 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay.

Stock solutions of DHA or CDHA in ethanol were prepared at a concentration of 500 mmol/l. For each experiment test compound was prepared from the stock solution and diluted to a final concentration of 1 μmol/l to 5 × 10² μmol/l with culture medium. Initially, 5 × 10³ KPL-1 cells were seeded in 96-well culture plates and cultured for 24 hours to allow them to adhere to the plates. After preincubation, the culture medium was changed to the experimental medium supplemented with DHA or CDHA. We used four concentrations of each test compound, covering a 3-log range (1 μmol/l to 5 × 10² μmol/l) that was chosen to span the 50% inhibitory concentration (IC₅₀) determined in preliminary assays. The final concentration of ethanol never exceeded 0.2% (vol:vol). Control cells were exposed to test medium supplemented with 0.2% ethanol. After incubation with the test compound for 24, 48 or 72 hours, MTT (Sigma) was added and the plate samples were then read in eight wells at an optical density value of 540 nm. Untreated controls were incubated with culture medium alone. The percentages of surviving cells, as compared with untreated controls, were calculated.

To evaluate the mode of cell death, asynchronous KPL-1 cells were exposed to DHA or CDHA at IC₅₀ for 72 hours, in DMEM supplemented with 10% FBS. To evaluate cell cycle progression, after serum starvation (incubated with FBS-free medium) for 24 hours synchronous KPL-1 cells were exposed to each test compound at the same concentration in DMEM supplemented with 10% FBS for 0, 3, 6, or 24 hours. Both asynchronous and synchronous KPL-1 cells were treated as follows. The adhered and floating cells were mixed and washed twice in phosphate-buffered saline (PBS)(-) followed by centrifugation, and fixed in 70% ethanol. Then, the cells were treated with 1 mg/ml RNase, diluted with PBS(-), and stained with 50 μg/ml propidium iodide in PBS(-). Cells were analyzed by FACScan (Becton Dickson, Mountain View, CA, USA) using Cell Quest software. Cell cycle distribution was quantified with Modfit LT software (Verity Software House, Topsham, ME, USA), using a doublet discrimination module to eliminate the possibility of confusing multiples of G₁ cells with ordinary G₂ cells.

KPL-1 cells were examined for expression of p53, p21^Cip1/Waf1, cyclin D₁, Bax, and Bcl-2 proteins. KPL-1 cells were exposed to DHA or CDHA for 3, 6 and 24 hours, or for 24, 48 and 72 hours (at the IC₅₀ dose for 72 hours) and compared with unexposed cells. Immunoblotting was performed as previously described 23. Briefly, cell pellets were homogenized using lysis buffer (50 mmol/l Tris-HCl [pH 6.8], 2% SDS, 5 mmol/l β-mercaptoethanol, 10% glycerol). Cell lysates were clarified by centrifugation at 13,500 g for 40 min. Protein concentrations were measured by Bio-Rad assay (Bio-Rad, Richmond, CA, USA), and 70 μg protein from each sample was mixed with loading buffer, electrophoresed on 15% SDS polyacrylamide gels, and blotted onto nitrocellulose membrane (Hybond-P; Amersham Pharmacia Biotech, Buckinghamshire, UK). The membranes were blocked with 5% dry milk in Tris-buffered saline–Tween for 1 hour at room temperature and probed with anti-p53 (clone DO-7; Dako, Glostrup, Denmark), anti-p21^Cip1/Waf1 (clone SX-118; BD Biosciences, San Jose, CA, USA), and anti-cyclin D₁ (clone P2D11F11; Novocastra Laboratories, Newcastle upon Tyne, UK) antibodies, and with anti-Bax (BD Biosciences) or anti-Bcl-2 (BD Biosciences) antiserum. Next, the membrane was treated with peroxidase-conjugated goat antimouse or antirabbit IgG antiserum (Envision + system HRP; Dako). Immobilized antigen was detected using ECL plus Western-blotting detection reagents and hyperfilm (Amersham), according to the manufacturer's instructions. Intensity of bands was quantified with the NIH Image1.47 Processing and Analysis program for use with Macintosh computers http://rsb.info.nih.gov/nih-image/.

Female athymic BALB/c-nu/nu mice (age 4 weeks) were purchased from Charles River Japan (Atsugi, Japan). After arrival the mice were randomly assigned to three different dietary groups, with 10 mice in each group and two cages for each group. For basal diet without CDHA, an American Institute of Nutrition (AIN-76 A) standard reference diet 24 was used. This diet consists of 20% casein, 0.3% DL-methionine, 55% cornstarch, 10% sucrose, 5% cellulose, 5% corn oil, 3.5% AIN-76 mineral mix, 1.0% AIN-76 vitamin mix, and 0.2% choline bitartrate. For diets containing 0.2% or 1.0% CDHA, the appropriate amount of CDHA was added to the basal diet. The diets were prepared at Oriental Yeast (Chiba, Japan) and were stored in sealed plastic bags in the dark at 4°C. During the experimental period the animals were housed in plastic cages with sterilized white pine chips for bedding. The animal room was maintained under specific pathogen-free conditions and was controlled for temperature (22 ± 2°C), light (12-hour cycle), and humidity (60 ± 10%). After the animals had become acclimatization to their respective powdered diets over 2 weeks, the semiconfluent KPL-1 cells growing in DMEM supplemented with 10% FBS at 37°C were trypsinized, and 2 × 10⁷ viable cells/0.25 ml of medium were injected subcutaneously into the area of the right thoracic mammary fat pad using a 26-gauge needle.

The local growing tumor was checked every day before it became visible, and was measured once a week after it became visible. Tumor volume was measured with calipers and calculated using a standard formula 25: width² × length × 0.5. All mice were killed when the tumors of the basal diet fed group began to exhibit central necrosis (41 days after tumor cell inoculation), and all animals underwent necropsy. At necropsy tumors were weighed and all organs were checked macroscopically. Abnormal organs, locally growing tumors, and axillary lymph nodes were fixed in 10% neutral buffered formalin and processed for hematoxylin–eosin staining. Axillary lymph nodes were evaluated microscopically for metastases.

All results are expressed as means ± standard error. In all in vitro experiments, significance of differences was determined using the two-tailed Mann–Whitney U-test for unpaired samples after assuring homogeneity of variance. In in vivo experiments, after assurance of homogeneity of variance, analysis was performed using the non-repeated-measures analysis of variance parametric test or the Kruskal–Wallis nonparametic test. If the P value of these pretests was below 0.05, then post hoc analysis was performed using Fisher's protected least significant difference test to compare body weight, tumor volume, tumor weight, and frequency of metastasis. P < 0.05 was considered statistically significant.

The number of viable KPL-1 cells decreased after treatment with DHA and CDHA (Fig. 1). Dose and time dependent cytotoxity (reduction in cell number to below the initial plating density) was observed. For 72 hours of treatment, the IC₅₀s of DHA and CDHA were 270 and 97 μmol/l, respectively. CDHA was more cytotoxic than DHA toward KPL-1 cells.

To determine whether the growth inhibitory activity of DHA and CDHA was related to induction of apoptosis, the effect of both reagents on the cell cycle was analyzed by flow cytometry. Representative DNA histograms of asynchronous KPL-1 cells are shown in Fig. 2. At the IC₅₀ for 72 hours, both DHA and CDHA induced appearance of a sub-G₁ fraction, which is characteristic of apoptosis.

Cells in asynchronous cell populations are in different stages of the cell cycle at any given time, making it difficult to interpret changes in cell cycle progression precisely. To examine inhibition of cell growth by DHA or CDHA, synchronous KPL-1 cells were treated with 270 μmol/l DHA or 97 μmol/l CDHA (IC₅₀ concentration for 72 hours). Cell cycle was synchronized by serum starvation for 24 hours followed by refeeding. Cell cycle fraction was recorded after 3, 6, and 24 hours of treatment with each test compound, and was compared with cells incubated in medium alone. Representative DNA histograms for synchronous cells and data from three independent experiments are shown in Fig. 3. Among synchronous cells incubated in medium alone, 71 ± 2% of cells were in G₀/G₁ phase and 19 ± 2% were in S phase. At 3 hours after refeeding, the percentage of cells in G₀/G₁ decreased to 37 ± 1% and the percentage in S phase increased to 43 ± 5%. At 6 hours the percentage of cells in G₂/M phase increased to 22 ± 0%. In cells treated with DHA or CDHA, cell cycle progression was comparable with that in cells in medium alone until 6 hours after re-feeding. However, after 24 hours of incubation with DHA or CDHA, the percentage of cells in S phase was lower (15 ± 7% and 12 ± 0%, respectively) and the percentage of cells in G₀/G₁ was higher (53 ± 10% and 76 ± 4%, respectively), as compared with cells incubated in medium alone, suggesting G₀/G₁ arrest. After 24 hours of incubation, the percentage of cells in G₂/M phase was high (32 ± 3%) among cells treated with DHA but not among cells treated with CDHA (17 ± 5%), as compared with cells incubated in medium alone (12 ± 1%). DHA may have induced cell arrest in the early stage of G₀/G₁ phase, leading to accumulation of cells in G₂/M phase.

Cell cycle regulatory proteins, including p53, p21^Cip1/Waf1, and cyclin D₁, were examined. Representative Western blotting results are shown in Fig. 4a, and data from three independent experiments using individually prepared lysates are summarized in Fig. 5a. Density of immunoreactive bands in DHA-treated or CDHA-treated samples is expressed as percentage of the density of corresponding bands in control samples, which were designated 100%. In KPL-1 cells, both DHA and CDHA at IC₅₀ for 72 hours increased levels of p53 and p21^Cip1/Waf1, and decreased levels of cyclin D₁ at 3, 6 and 24 hours, as compared with untreated control cells. Changes in the apoptosis accelerating protein Bax and the apoptosis suppressing protein Bcl-2 were examined 24, 48 and 72 hours after treatment. Although the level of Bax was not altered at these time points, a decrease in Bcl-2 level was observed (Figs 4b and 5b).

There was no evidence of gross toxicity resulting from the 0.2% or 1.0% CDHA diets. There were no significant differences in body weight change among groups (Fig. 6) and no deaths occurred. However, growth of KPL-1 cells, as evaluated by tumor volume, was always less in CDHA-fed mice than in CDHA-unfed controls (Fig. 7). At the end of the study tumor volume was significantly reduced in 1.0% CDHA-treated mice (855 ± 88 mm³) compared with CDHA-unfed controls (1314 ± 223 mm³; P < 0.05); the 0.2% CDHA diet (1193 ± 138 mm³) had no significant effect. Although a 26% reduction in tumor weight was observed in 1.0% CDHA treated mice compared with CDHA-unfed controls (1170 ± 158 and 1580 ± 261 mg, respectively), the difference was not significant; tumor weight was not reduced in 0.2% CDHA treated mice (1581 ± 157 mg). Axillary lymph node metastasis was observed in 60% (6/10) of control mice, 50% (5/10) of 0.2% CDHA-treated mice, and 40% (4/10) of 1.0% CDHA-treated mice (i.e. metastasis gradually decreased with increasing amount of ingested CDHA). In each dietary group metastasis tended to correlate with the size of the primary tumor (Fig. 8). Further experiments are needed to determine whether metastasis per se is affected by CDHA.