Fatty acid control of growth of human cervical and endometrial cancer cells.

Stearic acid and iodo-stearic and inhibited cell growth in a cervical cancer cell line (HOG-1) in a dose-related manner, with a half maximal effect at 50 microM stearic acid. Addition of oleic acid abrogated the effect of stearic acid. EGF-stimulated DNA synthesis and growth of HOG-1 cells was inhibited in the presence of stearic acid without any apparent effect on EGF receptor number or affinity.

These effects may be explained by the alteration in functional properties of cell membranes caused by changing their fatty acid content (Spector & Burns, 1987). The function of membrane proteins can be modified by the fluidity or microviscosity of surrounding lipids, which may effect the conformation and availability of membrane receptors (Shinitzky, 1984). Mobility within the plane of the cell membrane is necessary for the activation of the epidermal growth factor (EGF) receptor (Schlessinger, 1988), but there is no reported work on the possible effects of fatty acid modulation of cell membranes on EGF receptor function.
Stearic acid (SA) has been shown to inhibit growth of normal and neoplastic rat mammary epithelial cells (Wicha et al., 1979). As effects of fatty acids on neoplasic of the uterine cervix have not been investigated, this work examines the effect of stearic acid (SA) and 9 (10)-iodostearic acid (ISA), which may strongly influence membrane fluidity (Apostolov, 1980), on a human cervical cancer cell line. Fatty acid membrane modulation may affect the characteristics of receptor proteins and, as the importance of EGF and its receptor in the growth of gynaecological tumours has been suggested by Cowley et al. (1984), Kudlow et al. (1986, Sainsbury et al. (1987), Singletary et al. (1987) and Wu et al. (1981), we therefore also investigated the effect of SA on EGFstimulated growth and receptor characteristics in HOG-1 cells.

Materials and methods
Maintenance of cells in culture HOG-1 cells were maintained in Dulbecco's Modified Eagle Medium (DMEM) plus 10% heat inactivated fetal calf serum (FCS) in 5% CO2 at 37°C. Gentamicin (640 1tg per 100 ml) and glutamine (29.2 mg per 100 ml) were added to the medium. As a preliminary step to each experiment, cells were pre-incubated in DMEM containing 10% delipidised fetal calf serum (DLS) for three days. DLS was prepared by adding 50 ml of FCS to 500 ml of 1:1 ethanol:acetone. This was stirred for 4 h and filtered through Watman no. 1 paper with cold ether. The precipitate was dried, reconstituted in deionised water, and the resulting serum was filter sterilised. Growth of HOG-I cells in serum-free defined medium Some experiments were conducted in serum-free chemically defined medium to preclude the possible interaction of SA with unspecified serum growth components. In these experiments cells were plated in medium containing serum and allowed to adhere overnight before changing to DME/ F12 supplemented with insulin (6.24 IL g ml-') transferrin (6.25 sg ml1'), selenium (6.25 ,sg ml-'), bovine serum albumin (BSA) (1.25 mg ml-') and linoleic acid (5.35 sg ml1'). SA was prepared in ethanol as detailed below, and this was added directly to the medium.

Treatment of cells
Six (17.3 ml) or 12 (7.5 ml) multi-well plates were used, and cells were seeded in DMEM/DLS at concentrations of 20-30 x 103 per well (12-well plates) or 100-300 x 103 per well (six-well plates). After cell attachment the medium was removed and replaced by medium containing the appropriate additive. SA, ISA and oleic acid (OA) were supplied by Sigma Chemical Company and were stored at 4°C. Fresh solutions were prepared for each experiment. ISA and OA were heated to 37°C before aliquoting as they are not liquid at 4°C. All substances were dissolved in ethanol which was then added to DLS to achieve the indicated concentration and a final ethanol concentration of not greater than 0.1 % in medium. For experiments using serum-free medium, SA was dissolved in ethanol as usual, and this was added directly to the medium. The same concentration of ethanol was added to control media. Mouse EGF (Sigma) was stored at -70°C. This was reconstituted in phosphate buffered saline (PBS) and added at a concentration of 20 ng ml' of medium.

Measurement of cell responses
The effects of each treatment on growth were determined by measurement of cell numbers and DNA synthesis by 3Hthymidine incorporation (0.5 ltCi methyl thymidine ml -' medium; s.a. 40-60 Ci nmol-') into TCA-precipitable material for the final 6 h of the experiment. Cell counts were conducted in quadruplicate using a haemocytometer, counting a minimum of 200 cells and assessing viability at each time point using Trypan blue. NADH exclusion was also used to assess cell viability (Aldred & Cooke, 1983).
To allow pooling of data, cell counts were expressed as a percentage of the day 0 count. The resulting distribution of data was tested for normality (Royston, 1983) and paired t tests were used to compare results. Measurement of EGF receptor concentration and affinity '25I-EGF was prepared by the lactoperoxidase method using sodium iodide (Amersham) and mouse EGF (Sigma) to a final specific activity of approximately 120 jiCi jig -' (59% incorporation). The iodinated EGF was desalted on a G-25 column. Cells were grown in six-well plates, and after incubation of the treated cells with SA (75 jiM) for 3 days, the medium was removed and the cells washed three times with 1 ml DMEM containing 0.1% BSA. They were then incubated for 1 h at 25°C with '25I-EFG (0.5 nM) in a total volume of I ml and a range of concentrations of cold EGF (0.018 -5.0 nM) as competitor. Triplicate samples were prepared. Non-specific binding was determined in the presence of 100-fold excess of cold EGF. The medium was removed after incubation and the cells washed three times with I ml ice-cold DMEM containing 0.1% BSA. The cells were solubilised in I ml 0.5 N NaOH, the radioactivity determined and analysed by the method of Scatchard (1949).

Results
Cell growth Both SA and ISA significantly inhibited cell growth of HOG-1 cells after 48 h in culture (P <0.005 for days 2, 3 and 4) (Figure 1). There was no difference between the inhibition seen with SA and ISA, both inhibited growth by 30-35% at 48 h. The growth inhibition was dose-dependent in the range of 12.5-100 jiM (Figure 2). At 125 jiM SA, cell death occurred. Differences in growth rates assessed by cell numbers were confirmed by 3H-thymidine incorporation assays; control 42,153 ± 501 c.p.m.; SA (75 j M) 25,166 ± 359 c.p.m. on day 2 (n=6, P<0.05).
The lower cell number at each time point in the SA-treated group was not due to cytocidal effect. This was treated by plating out cells and, after adding SA in the usual manner, performing cell counts at 30 min, 2, 4 and 10 h. There was no difference in cell counts (n = 3) between control and treated cells in this time period. Trypan blue testing (at each timepoint cells were counted) showed no differences between control and treated cells (data not shown).
In serum-free medium the inhibition of growth was less marked than in DMEM/DLS, but by 72 h 23% growth inhibition was seen: control cell no. = 307% of day 0 count (s.d. 19%); SA treated cell no. = 273% of day 0 count (s.d. 10%); n = 3; P = 0.02. SA in desaturated oleic acid, which contains a double bond at the C9-C10 position. However, the growth inhibitory effect is not mediated by OA, as the inhibitory effect of 75 t M SA on HOG-I cells is abrogated by OA 30 jiM at 48 h ( Figure   3).
The effect of SA on HOG-I cells is reversible. After achieving 30% growth inhibition with SA (75 jM) after 48 h in the usual manner, control and treated cells were harvested, counted and tested for viability with Trypan blue and resuspended in DMEM + 10% FCS. Equal numbers of control  Figure 1. Range of day 0 cell numbers: 20-30 x 103 per well; n = 4. Significant difference between SA and SA + OA treatment; P = 0.015. and previously SA-treatd cells were re-seeded into separate 50 ml flasks. Successive flasks were harvested and counted 24, 48 and 72 h following re-seeding, showing logarithmic growth and no difference in cell numbers between control cells and those which had been previously growth-inhibited by SA (75 jM) (Figure 4). Growth inhibition without fall in viability was also seen in an endometrial carcinoma cell line (Ishikawa). There was 23% growth inhibition after 48 h in the treated cells compared with control: control cell no. = 255% of day 0 count (s.d. 28%), SA treated cell no. = 197% of day 0 count (s.d. 23%); n = 3; P = 0.018.
Inhibition of response to EGF HOG-I cells contain membrane EGF receptors, and treatment with EGF (20 ng ml-') for 3 days resulted in a 48% increase in cell numbers, and 27% increase in 3H thymidine incorporation. However, the stimulatory effect of EGF was not seen when cells were treated with 75 jiM SA (P = 0.02; n = 3; Figure 5). The presence of SA did not inhibit the binding of iodinated EGF added to growth medium (data not shown) therefore excluding the possibility of decreased availability of hormone. There was no difference in EGF receptor concentration per cell between control and treated cells (control, 47.3 fmol per million cells; treated, 43.1 fmol per million cells). Scatchard analysis showed no difference in EGF receptor affinity between control cells and those to which SA had been added (Kd control 3.41 nM; treated 5 2.82 nM; Figure 6).

Discussion
The fatty acid content of cell membranes may be altered by changes in the composition of the growth medium, the extent Range of day 0 cell numbers = 16-35 x 10' cells per well.
Significant difference between EGF-treated and EGF + SA treated cells; P = 0.02. of enrichment of a particular FA being dependent on the amount of supplement added (Yorek et al., 1984). Such alterations lead to changes in the physical and functional properties of the membrane such as altered fluidity (Chapman & Quinn, 1976), transport (Spector & Burns, 1987) and enyzme and receptor protein activity (Sandermann, 1978;Shinitzky, 1984). Changes in receptor protein activity may be due to conformational changes induced by alteration in surrounding lipid (Shinitzky, 1984).
Stearic acid (an 18-carbon SFA) effectively inhibits phytohaemagglutinin (PHA)-induced transformation of lymphocytes (Mertin & Hughes, 1975) and '4C-uridine uptake by PHA-stimulated lymphocytes (Weyman et al., 1977). Our results show that SA is capable of reversibly inhibiting cell growth in a cervical squamous carcinoma cell line. Growth is inhibited by 30-35% after 48 h in DMEM/ DLS; and by 22% in serum-free medium after 72 h.
An endometrial cancer cell line shows 23% growth inhibition after 72 h. The finding that OA negates the effect of SA agrees with the results of Weyman et al. (1975)   measured growth of mouse LM cells. Both high and low affinity binding sites for EGF have been reported in many cell lines (King & Cuatrecasas, 1982;Kawamoto et al., 1983). Cell lines expressing only a single class of EGF receptor are also described (Mummery et al., 1983;Fabbro et al., 1986). Using EGF concentrations ranging from 0.018 to 5.0 nm, there was no evidence of two receptor affinities in the HOG-I cell line. Under the conditions of the assay (25°C for I h) greater than 90% of cell-associated radioactivity was removed by acid washing, consisteht with other reports (Imai et al., 1982;Fabbro et al., 1986), suggesting that analysis of the binding data is not complicated by internalisation. The inhibition of EGFstimulated growth in the presence of SA cannot be explained by availability of growth factor or changes in receptor number or affinity. Interference with signal transduction distal to ligand binding, for example an effect on receptor clustering (Schlessinger, 1988) and/or protein tyrosine kinase activity (Northwood & Davis, 1989) may explain the observed effects of SA.
Although previous work has shown effects of lipidmodulated changes in membrane fluidity on receptor proteins (Shinitzky, 1984;Spector & Burns, 1987;Ginsberg et al., 1981) there have been few reports of FA effects on the EGF receptor. Gladhaug et al. (1988) demonstrated that the 4carbon butyrate increases EGF receptor binding in rat hepatocytes, and affects receptor status and cell morphology.
It is possible that modulation of membrane viscosity surrounding the EGF receptor leads to a conformational change in the receptor, possibly inhibiting dimer/oligomer formation. EGF mediates pleotropic effects on cells, including the sitmulation of mitogenesis (Gullick & Waterfield, 1987), and has a demonstrably important role in gynaecological malignancy. In this context, there could possibly be a role for saturated fatty acids in the management of malignancy, possibly as an adjunctive treatment.