Transforming growth factor-beta mRNA expression and growth control of human ovarian carcinoma cells.

The pattern of TGF beta expression and in vitro response to TGF beta has been defined in three ovarian carcinoma cell lines (PEO1, PEO4 and PEO14). Marked differences in both mRNA expression and growth responses were detected between the cell lines. All expressed mRNA for TGF beta 3, PEO1 and PEO4 but not PEO14 expressed mRNA for TGF beta 1, whereas PEO14 but not PEO1 and PEO4 expressed TGF beta 2. Growth of PEO14 cells in culture was markedly inhibited by both TGF beta 1 and beta 2. PEO1 cells were inhibited by TGF beta 1, but not TGF beta 2 whilst growth of PEO4 cells were not affected by exposure to either of these peptides. These data indicate that several elements of potential autocrine loops involving TGF beta's are present within ovarian cancer cells. ImagesFigure 1Figure 2Figure 3

The transforming growth factors are increasingly recognised as important molecules in the regulation of cell growth and differentiation (Nilsen, 1990;Barnard et al., 1990 for reviews). Whilst the role of TGF-x has been relatively widely studied, data on the TGFP's are only now becoming available. These peptides are generating interest in a variety of fields including development (Akhurst et al., 1990;Roberts et al., 1990a), bone remodelling (Noda & Rodan, 1989;Bonewald & Mundy, 1990), extracellular matrix production (Roberts & Sporn, 1989) and the prevention and treatment of cancer (Colletta, 1990). Research into the biological role of TGFP's is complicated by the diversity of the peptide family, with at least three TGF-,B peptides being identified thus far in the human (Derynck et al., 1985;Arrick et al., 1990) and additional forms being present in other species (Roberts et al., 1990b). However, TGFIB has been detected in a wide range of tissues, including transformed cells (Roberts et al., 1981), and dependent upon conditions it may be either stimulatory or inhibitory for cell growth (Roberts et al., 1985). There are multiple binding proteins for TGFPI's (Frolik et al., 1984;Massague & Like, 1985;Massague et al., 1990) and recent data suggest that the growth regulatory effects of TGFP may be dependent upon the presence of specific classes of binding proteins (Roberts, 1991). TGFP appears to play a role in normal ovarian function, particularly in the regulation of granulosa cell functions in response to follicle stimulating hormone (Adashi et al., 1989).
However, little is known of the effects of TGFP's and their expression in ovarian carcinomas. In order to investigate further the role of TGF,B in ovarian carcinoma cells, we have developed cell line models and have examined them for the presence of TGFP mRNA and their growth response to TGF-P.
Effects of TGFI31 and TGF P2 on growth of ovarian carcinoma cell lines Exponentially growing cells were harvested by trypsinisation and plated in 24-well plates (Falcon) at densities of approximately 2 x 104 cells/well (four wells per experimental condition) in RPMI containing 5% FCS. After 24 h, medium was removed, cells were washed with phosphate buffered saline (pH 7.4; PBS) and medium replaced with RPMI containing either 5% double charcoal stripped (DCS), FCS, 0.5% DCS, FCS or HITS (hydrocortisone 10 nm, insulin 5 ,tg ml-', transferrin 1O fig ml-', selenium 30 nm) and incubated for a further 24 h. Cells were then washed with PBS and medium replaced with RPMI with the corresponding additives (as above) with or without human recombinant TGF-P, or porcine TGF-P2 (British Biotechnology) added at concentrations ranging from 0.01 ng ml-' to 1.0 ng ml-'. This time point was designated day 0. Cells were incubated at 37°C for 6 days. Media was replenished on day 3. On days 0, 3, and 6, cells were harvested by trypsinisation and counted using a Coulter Counter.
Effects of TGF132 and TGF P2 on the cell cycle in the PE014 cell line Exponentially growing cells were harvested by trypsinisation and plated in 6-well plates (Falcon) at densities of approximately 2 x 104 cells well (four wells per experimental condition) in RPMI containing 5% FCS . After 24 h, medium was removed, cells were washed with PBS and medium replaced with RPMI containing 5% DCS, FCS with or without added growth factor. Cells were harvested by trypsinisation at time 0 and after 24 and 48 h incubation with TGFP, or TGFP2 at a concentration of 1 ng ml-'. The cell suspension was transferred to plastic tubes containing 0.5 ml FCS (to neutralise trypsin) and cells centrifuged for 4 min at 500 g at room temperature, washed in PBS and repelleted. Ethanol (0.5 ml 70%) was added to the cells and pellets stored at -40'C until required for analysis. Cells were treated with detergent and the DNA stained with propidium iodide (Vindelov et al., 1983). Analysis was performed using a FACScan flow cytometer (Becton Dickinson), with gates set to exclude fragmented or clumped material and doublets. mRNA extraction Exponentially growing cells were harvested from 175 cm2 culture flasks as follows: Cells were washed with ice cold PBS, harvested using a cell scraper, suspended in ice cold PBS (25 ml) and spun down in a bench top centrifuge (1,000g, 10min). The cell pellet was stored at -70°C until used for RNA extraction. Using a sterile pasteur pipette the cell pellet was transferred to a 15 ml tube containing 3 M lithium chloride/6 M urea (6 ml). The homogenate was sonicated twice at 4°C for 30 s and stored overnight at 4°C. The pellet was spun down at 15,000 g, 4°C for 30 min. The supernatant was discarded and the pellet washed with fresh lithium chloride/urea (6 ml) and centrifuged at 15,000 g, 4°C for 30 min. The pellet was then resuspended in Tris-HCl (10 mM pH 7.5, 6 ml) SDS (0.5%), with proteinase K (50 fig ml 1, Boehringer Mannheim) added and the sample incubated at 37°C for 20 min. Following incubation the samples were extracted using 100% phenol (pre-equilibrated with 0.1 M Tris pH 7.4), this extraction was repeated using phenol:chloroform: isoamyl-alcohol (25:24:1 v/v/v) and chloroform:isoamyl-alcohol (24:1 v/v). Following each extraction the sample was centrifuged at 2,000 g at room temperature for 10 min and the aqueous phase recovered. After the final extraction, lithium chloride (300 yl 8 M), absolute alcohol (2.5 volumes) were added and the RNA precipitated overnight at -20°C. RNA was pelleted by centrifugation at 4,000 g, 4°C for 45 min. The supernatant was decanted and the pellet dried and resuspended in diethylpyrocarbonate treated water. Optical density measurements at 260 and 280 nm were taken to assess yield and purity of the RNA preparation.
Synthesis of riboprobes Labelled RNA was prepared from linearised template DNA using a Gemini II system (Promega Ltd, Southampton, UK).
Template DNA was incubated in the presence of an RNAase inhibitor (Human placental RNAsin; Amersham plc), cold ribonucleosides, dithiothreitol and 32P-rCTP with the appropriate RNA polymerase (T3, T7 or SP6) for 1 h at 37°C. The DNA template was then removed by incubation with RQ1 DNAase (Promega Ltd) for 15 min at 37°C. Labelled RNA was precipitated in the presence of added tRNA (Sigma) as carrier and full length transcripts were isolated by polyacrylamide electrophoresis. Following identification of full length transcripts by autoradiography, the bands were excised and labelled RNA eluted from the gel, precipitated under ethanol and resuspended in hybridisation buffer prior to use in RNAase protection assays.

RNAase protection assay
Test RNA (20 rig) was precipitated under ethanol, dried and resuspended in 30 gAl hybridisation buffer (80% formamide, 40 mM Pipes (pH 6.7), 400 mM NaCl, 1 mM EDTA); tRNA was prepared in a similar manner as a negative control. Test probe (106 c.p.m.) plus actin probe (106 c.p.m.) were added to each sample. Samples were incubated at 85°C for 20 min, transferred to a water bath and left to hybridise overnight at 51°C. After hybridisation, single stranded RNA (both labelled and cold) was removed by incubating with single strand specific RNAases A and TI (Boehringer Mannheim) at 37°C for 30 min, followed by incubation with proteinase K in SDS at 37°C for 15 min. Protein was extracted by using phenol/ chloroform-isoamyl alcohol. Double stranded probe: test RNA was precipitated with carrier tRNA (5 fAg) and separated by gel electrophoresis. Full length transcripts for test probes were scored as positive, whilst transcripts for actin were used as an internal control.

Statistics
Cell growth and cell cycle responses in vitro were analysed using Wilcoxon Rank test and significant differences at the P <0.05 level defined.

Results
TGFIp mRNA expression in ovarian carcinoma cell lines TGFiI mRNA expression was observed only in cell lines PEOl and PE04, and not in PEO14 cells (Figure 1). In contrast, expression of mRNA for TGFP2 could not be demonstrated in PEOI or PE04 cell extracts whilst PE014 cells appeared to express this factor ( Figure 2). Finally, TGFP3 mRNA expression was observed in all three cell lines tested (Figure 3). Growth responsiveness of ovarian carcinoma cell lines to TGFp, and TGF132 PEOI TGFI31 was capable of producing significant inhibitory effects on the growth of PEOI cells, but these were small and observed only under certain culture conditions. Thus, as  is shown in Figure 4, significant inhibitory effects were produced at day 6 but not day 3 by addition of 1 ng ml1' TGFPI in HITS and 0.1 and 1 ng ml-' TGF,I3 in 0.5% serum supplemented culture medium. In culture systems containing 5% serum TGFP, produced no significant effect at any time point. No significant effect of TGFP2 on the growth of PEOl cells was observed under any of the conditions tested during the course of these experiments (data not shown).
PE04 Neither TGFPI nor TGFP2 altered the growth of PE04 cells under any growth conditions tested (5% or 0.5% DCS FCS or HITS; Data not shown).
PE014 In contrast to PEOI and PE04, TGFPI3 produced a significant inhibitory effect on the growth of PE014 in each of the culture conditions tested. The effects were dose related and were more pronounced after 6 days of culture. At the highest doses of TGFi, growth was significantly inhibited during the first 3 days of exposure of the cells (P<0.05) whilst by 6 days after initial exposure growth inhibition was produced by all doses above 0.01 ng ml-' reaching a maximum at 1 ng ml-' TGFI (P <0.05; Figure 5). These effects of TGFP, occurred irrespective of the presence or concentration of serum and were consistent in each of three replicate experiments (representative experiment shown; Figure 5).
The effects observed with TGFPi2 on the cell line PE014 were similar but less marked than those found with TGF,I,. Thus higher concentrations of TGFP2 were required for significant effects to be demonstrated and the degree of inhibition was less pronounced, nevertheless the observed effects were consistent in each of three replicate experiments (representative experiment shown; Figure 6).
Effects of TGFJP, and TGF P2 on the cell cycle in the PEOJ4 cell line Short term exposure to either TGFP, or TGFP2 was capable of producing significant effects on the cell cycle distribution of PE014 cells (Figure 7). After 24 h an increase in cell numbers in the GO/GI phase of the cell cycle was observed (P <0.05) in cells treated with either TGF,I or TGFP2 and in the case of TGFP, this was associated with a decrease in cell numbers in S-phase; there were no marked effects upon cell numbers in the G2/M phases of the cycle at this time.
However, after 48 h exposure to TGFP, or TGFP2 the increase in cell numbers in GO/GI compared with untreated cells became more pronounced and was associated not only with decreased cell numbers in S phase, but also a significant reduction in the proportion of cells in the G2/M phases of the cell cycle (P<0.05; Figure 7). Solid bars represent untreated cells. Each point represents mean ± s.e. of quadruplicate points. _ = untreated cells, = cells exposed to 0.1 ng ml-' and = cells exposed to 1 ng ml-TGFP, respectively over either 3 or 6 days. * = Statistically significant difference with respect to time matched control (P< 0.05). a, Cells grown in RPMI containing HITS. b, Cells grown in RPMI containing 0.5% DCS FCS. c, Cells grown in RPMI containing 5% DCS FCS (see text for details).

Discussion
Evidence is presented here to demonstrate that TGFP's inhibit cell growth in some, but not all ovarian carcinoma cell lines in vitro. Whilst both TGFPI and TGFP2 markedly inhibited the growth of PE014 cells ( Figure 5-6), PEOI cells responded only to TGFP, (Figure 4), and PE04 cells were unaffected by either peptide. The responses observed for the ovarian carcinoma cell lines investigated here contrast with those observed previously (Marth et al., 1990) in which all four cell lines tested were inhibited by TGFP2, but the concentrations used were up to 100-fold higher than those used in the present study. None of the inhibitory doses used in the present study would have been effective on these previously tested cell lines (Marth et al., 1990). It would therefore appear that the sensitivity of ovarian carcinoma cell lines to TGFPI's varies considerably between cell lines. Of all ovarian carcinoma cell lines tested to date, PE014, showed the most marked sen- Figure 5 Growth response of PEO14 cells to TGFPi,: Cell counts per well of PEO14 ovarian carcinoma cells treated with TGFPi,.
Solid bars represent untreated cells. Each point represents mean ± s.e. of quadruplicate points. = untreated cells, 1 = cells exposed to 0.01 ng ml-' and m = cells exposed to 0.1 ng ml-; = cells exposed to 0.5 ng ml-' and = = cells exposed to 1.Ongml-' TGFP, respectively over either 3 or 6 days. * = Statistically significant difference with respect to time matched control (P<0.05). a, Cells grown in RPMI containing HITS. b, Cells grown in RPMI containing 0.5% DCS FCS. c, Cells grown in RPMI containing 5% DCS FCS (see text for details). sitivity to TGFP's being over 100 times more sensitive to TGFPI, than other cell lines reported. Treatment with doses of TGFP between of 0.1 and 1 ng ml', reduced cell proliferation by up 50%.
It has recently been suggested that the growth inhibitory effects of the TGFI3 family results in an arrest of cells in the GI phase of the cell cycle (Roberts et al., 1991). The data obtained for the PE014 cell line support this observation. Solid bars represent untreated cells. Each point represents mean ± s.e. of quadruplicate points. = untreated cells, M = cells exposed to 0.01 ng ml-' and E = cells exposed to 0.1 ng ml-'; 1 = cells exposed to 0.5 ng ml' and = cells exposed to 1.0 ng ml-I TGFP2 respectively over either 3 or 6 days. * = Statistically significant difference with respect to time matched control (P< 0.05). a, Cells grown in RPMI containing HITS. b, Cells grown in RPMI containing 0.5% DCS FCS. c, Cells grown in RPMI containing 5% DCS FCS (see text for details).
In addition to the growth inhibitory effects of TGFP's being targeted primarily in the GI phase of the cycle, data exists which suggests that such effects are mediated via type II binding sites for TGFP's on the cell surface (Roberts et al., 1991). It has been demonstrated (Massague et al., 1990) that three separate binding proteins exist for the TGFi family. These proteins, designated class I-III, have yet to be isolated and purified, and therefore have not yet been sufficiently characterised to define them as receptors. The cell lines studied here will provide a useful model for the further investigation of the relative importance of the different TGF,B binding proteins.
Data presented here also demonstrate that mRNA's for the different forms of TGF0 are expressed in ovarian carcinoma cell lines and that the expression patterns of TGFP's vary between the different cell types investigated.  TGFP,. Further studies to define the level of peptide production to establish the presence of TGFP binding proteins and examine the effects of TGFP blocking antibodies would be of interest in determining the relative role of the different forms of TGFPI and their binding proteins in these putative autocrine pathways. TGFPi has been shown to be an autocrine regulator of breast cancer cells (Arteaga et al., 1988) and in lymphocyte activation (Lucas et al., 1990) but this loop had not been previously established for ovarian carcinomas. In breast cancer, secretion of TGFP occurs under steroid hormone regulation (Knabbe et al., 1987;Colletta, 1991), therefore to demonstrate the presence of mRNA for TGFP cells were grown in the presence of non-charcoal stripped serum. The physiological and clinical relevance of these findings remain to be elucidated, however it is known that (1) TGFP may be a marker of progression to steroid insensitivity in breast cancer cells (Daly & Darbre, 1990).
(2) TGFP has been implicated in the autocrine regulation of normal ovarian function (Knecht et al., 1987;Kim & Schomberg, 1989;Magoffin et al., 1989), and (3) TGF beta has been implicated in the process of metastasis (Schwarz et al., 1990). In summary, results from the present study show that TGFi may be a potential autocrine regulator of ovarian carcinoma cell division in vitro, in both oestrogen sensitive (PEOI) and insensitive (PE014) cells. The differential expression patterns for the various forms of TGFP observed suggest that these factors may play distinct roles in the regulation of cell division.