Effect of Serum Albumin on Estrogen Metabolism in Human Cancer Cell Lines

ecular and t?ellulrr Endocrinology Molecular and Cellular Endocrinology 115(1995) 221-225

Rapid Paper Effect of serum albumin on estrogen metabolism in human cancer cell lines

H. Leon Bradlow”, Felice Arcurib, Livio Blasib, Luigi Castagnettabqc

“Strang-Cornell Cancer Research Laboratory, Cornell University Medical College, New York, ,YY, USA bHormone Biochemistry Laboratory, University of Palermo School of Medicine, Via Marchese Ugo, 56 90141 Palermo, Italy ‘Experimental Oncology and Molecular Endocrinology Units, Palermo Branch of IST-Genoa, c/o ‘M. Ascoli’ Cancer Hospital Center, Palermo. Italy

Received 9 August 1995; accepted 6 September 1995

Abstract

The observation that charcoal-treated fetal bovine serum (ctFBS) was able to modify one of main pathways of estrogens in cancer cells in culture, prompted us to initiate the present study. The active component of serum was isolated using native preparative polyacrylamide gel electrophoresis (PAGE). Under analysis with SDS-PAGE, a Mw of 68 kDa and mobility of authentic bovine serum albumin (BSA) was observed. The addition of BSA to the serum free culture medium of HEC 1A human endometrial cancer cell line, resulted in an alteration of estradiol (EJ metabolism similar to that observed in the presence of ctFBS. BSA in fact, much enhanced 16cr-hydroxylation and significantly reduced 2-hydroxylation of E, in HEC 1A cells. Comparable results were obtained with different endometrial (Ishikawa) and mammary (MCF-7) tumor cell lines having a different metabolic conversion rate of E,. Several albumin preparations from either bovine or human serum had the same effect; besides, BSA activity was unaffected by treatment with dextran-charcoal or heat. In the light of the present results, the inclusion of serum albumin (SA) in the formulation of media for studies evaluating steroid metabolism in cultured cells should be carefully considered.

Keywords: Cancer cell (human); Fetal bovine serum; Albumin; Estradiol metabolism

1. Introduction components of steroid-depleted serum might modulate estradiol (EJ metabolism. The aim of this study was to The use of varying amounts of serum has been investigate which factor(s) was responsible for the ob- considered mandatory to promote substrate adhesion served changes focusing in particular, on 16a- and and growth of cells in culture. However, in studies 2-hydroxylation of E,. We found that serum albumin evaluating steroid metabolism, the presence of serum (SA) is the main active component of FBS. has always been considered critical since it represents a Recent observations have demonstrated that SA is well-known source of hormones. Therefore, many able to alter metabolic reactions and cellular responses efforts have been made to develop steroid-depleted sera, in normal as well as in tumor cells. SA has been but little or no attention has been given to the role of reported to modify pregnenolone production by imma- other components of serum that might af+ct metabolic ture rat Leydig cells in vitro (Melsnert et al., 1989). transformation of steroids. In our studies on human Singh et al. (1992) identified human SA (HSA) as a endometrial and breast cancer cell lines, we have ob- component of a bioactive fraction able to stimulate the served that the addition to the culture medium of reductive conversion of estrone (E,) to E, in MCF-7 charcoal-treated fetal bovine serum (ctFBS) modifies a cancer cell line. In this paper, we investigate the role for major metabolic pathway of estrogen, indicating that SA in the regulation of E, metabolism, namely en- hancement of 16cr- and reduction of 2-hydroxylase * Corresponding author, Tel.: 39 91 666 4345; Fax: 39 91 666 4352. activity.

2. Material and methods exchange and normalized for the amount of DNA, is expressed as percentage of control. 2.1. Chemical Bovine serum albumin (fraction V, fatty acid free, 2.4. Fetal bovine serum fractionation globulin free), human serum albumin (fatty acid free), CtFBS was fractionated by native preparative acry- thyreoglobulin, y-globulin, fetuin, gelatin, dextran, lamide gel electrophoresis (PAGE) using acrylamide dithiothreitol and ethylenediamine tetraacetic acid were concentrations of 7.5% for resolving and 4.0% for obtained from Sigma Chemicals Co. (St. Louis, MO). stacking gels, respectively. Samples of ctFBS were [C16a-3H]E, (SA, 20 Ci/mmol) was prepared as previ- loaded in a 3-mm thick acrylamide gel and run ously described (Schneider et al., 1982). [C2-3H]E, (SA, overnight at 4°C. Dithiothreitol (DTT) and ethylenedi- 15 Ci/mmol) was purchased from New England Nu- aminetetraacetic acid (EDTA), 1 mM each, were in- clear (Boston, MA). cluded in both stacking and resolving gel as protective agents. The following day the gel was cut horizontally 2.2. Cell culture in fractions of 8.0 mm each, and proteins were electroe- HEC 1A and Ishikawa human endometrial cancer luted in dialysis bags (Mw cut-off, 3.5 kDa) in 0.1 M cell lines were kindly supplied by Dr. E. Gurpide sodium-phosphate buffer, pH 7.4, DTT and 1 mM (Mount Sinai School of Medicine, New York, USA). EDTA, at 4°C. The content of the dialysis bags was MCF-7 human breast cancer cell line was obtained concentrated using Microsep 10 K centrifugal concen- from Michigan Cancer Foundation (Detroit, MI). Cells trators (Filtron Technology Co., Northborough, MA), were routinely maintained at 37°C in a humidified assayed for protein following the method of Bradford atmosphere of 95% air, 5% CO, in MEM containing (1976) and tested for bioactivity. Gel electrophoresis of 8.63 mg/l phenol red as pH indicator and supplemented proteins was carried out in 10.5% polyacrylamide in the with 10% FBS, 10 mM L-glutamine, 100 III/ml peni- presence of sodium dodecyl sulfate (SDS) as described cillin, 100 pg/ml streptomycin, 0.25 pug/ml fungizone in by Laemmli (1970). Proteins were stained with silver loo-mm tissue culture plates. according to the procedure of Wray et al. (1981).

2.3. Bioactivity assay 2.5. Statistical analysis The assay is based on the capability of protein prepa- Statistical comparisons were performed using one rations to modify 16cw-and 2-hydroxylation of E, in sample two tailed Student’s t-test. Probability values of cultured cells. The extent of C16a- and C2-hydroxylase less than 0.05 were considered significant. activity was measured by determining 3H,0 formation in cells incubated in the presence either of [C16a-3H]E, 3. Results or [C2-3H]E2. The cells were grown to confluence, trypsinized and seeded in 24-well culture dishes (1 x Fig. 1 shows the effect of ctFBS on the metabolic 10’ cells per well, four wells for each experimental conversion of E, by HEC 1A human endometrial can- condition) in maintenance medium. The next day the cer cell line. The presence of ctFBS in the culture medium was removed and, after washing the cultures medium resulted in a dramatic increase in 16a-hydroxy- three times with HBSS, changed to serum-free, phenol red free, D-MEM medium containing either vehicle 350 1 control or Jest proteins in aqueous solution. Unless ..I I T otherwise indicated, proteins were tested at a concentration of 2.0 mg/ml. After 24 h, the medium was i replaced with fresh medium that contained ‘H-E, at a ;6 250 concentration of 1 x 10 - 9 M in ethanol (final alco- b 200 holic concentration = 0.1%). After 48 h of incubation at & 37°C in a humidified atmosphere of 95% air, 5% CO,, 7c1 150 aliquots of 700 ,ul of medium were diluted to 1.5 ml t with distilled water, lyophilized and the sublimate was k 100 Untreated Control a counted for radioactivity in duplicate in a liquid scintil- 50 lation counter. The 3H released from specifically labeled E, provided an indirect measurement of the regiospe- 0 cific hydroxylation of the steroid (Schneider et al., 16a-hydroxylation 2-hydroxylation 1982). The cell layer was lysed in 0.1% SDS and DNA Fig. I. Elect of ctFBS on 16a- and 2-hydroxylation of EZ in HEC 1A content was determined by a spectrofluorimetric cells. The results are means f SD (16a-hydroxylation, n = 5; 2-hy- method (Hinegardner, 1971). The extent of 16c(- and droxylation, n = 3). ***P = 0.0025; **P = 0.019 (two tailed Student’s 2-hydroxylation of E,, corrected for the non-specific 3H t-test). H.L. Bradlow et al. / Molecular and Cellular Endocrinology 115 (1995) 221-225 223

12.5 225

E 10.0 200 95,000 .F t;al 7.5 175

a’ 68,000 I 5.0 150

2.5 125 43,000 0.0 100 1 2 3 4 5 6 7 6 9 1011 Fraction Number 36,000 Fig. 2. Effect of ctFBS fractions obtained by native polyacrylamide gel electrophoresis on 16a-hydroxylase activity in HEC IA cells. Proteins were tested at the concentration of 0.5 m&ml. 29,000 lase activity (271% &- 56.5 with respect to control; P = 0.0025) and in a significant decrease of 2-hydroxylase (78.6% & 5.2 of control; P = 0.019) as reflected by the 16a-/2_hydroxylation ratio (3.1 -fold of control). Fig. 3. SDS-PAGE of fraction # 10 from native polyacrylamide gel To identify the component of serum responsible for this electrophoresis of ctFBS. The protein standards used (with their M,) effect, ctFBS was fractionated by native preparative were phosphorylase B (95 000 Da), bovine serum albumin (68 000 PAGE electrophoresis and the fractions tested for Da), ovalbumin (43 000 Da), lactate dehydrogenase (36 000 Da) and carbonic anhydrase (29 000 Da). Gel was stained with silver nitrate. bioactivity. As shown in Fig. 2, three (from # 8 to Lane A, protein standards. Lane B, 1 fig of protein from fraction # 10) of the eleven fractions obtained were able to # 10. significantly increase 16a-hydroxylation of E, in HEC 1A cells. SDS-PAGE followed by silver staining showed the presence in fraction # 10 of a single band with an apparent Mw of 68 kDa with the mobility of authentic 2.0 mg/ml of BSA/HSA to the culture medium deter- BSA (Fig. 3). mined an increase on 16a-hydroxylation as well as a We tested the ability of commercial preparations of decrease of 2-hydroxylation demonstrated by the aug- BSA to modify E, metabolism. In HEC-1A cells (Fig. ment of the 16a-/2-hvdroxvlation ratio (2.7-/3.8- and 4) the addition to serum-free culture medium of BSA 3.9-/2.9-fold of control respectively). (Fraction V) resulted in an increase of 16a-hydroxylation (292% + 49.9 of control; P = 0.0046), and in a significant decrease of 2-hydroxylation (78.5% f 1.75 of control P = 0.0022) comparable to those observed 350 in the presence of FBS. This was reflected by the 300 16a-/2-hydroxylation ratio (3.15-fold of control). BSA 5 2 stimulated 16a-hydroxylase activity over the range of 250 5 concentrations tested (0.4-4.0 mg/ml); a similar effect 0 was obtained with albumin preparations obtained from % 200 bovine (fatty acid free and globulin free) as well as from & 150 human serum (globulin free). BSA treated with char- s ! Untraatad Control coal or heated at 60°C for 30 min retained its activity. ; 100 In contrast, other proteins not structurally related to P albumin (y-globulin, thyroglobulin and ovalbumin) and 50 gelatin were unable to modify 16a-hydroxylase activity. 0 Fetuin, a 48-kDa protein of FBS known to migrate as l&x-hydrowylation Z-hydroxylation 68 kDa in SDS-PAGE (Rohrilich, 1981), was also Fig. 4. Effect of BSA (fraction V) on 16rr- and 2-hydroxylation of E, ineffective (data not shown). in HEC IA cells. The results are means k SD (16a-hydroxylation A comparable effect was obtained in experiments n = 4; 2-hydroxylation n = 3). ***P = 0.0046; **P = 0.0022 (two with Ishikawa and MCF-7 cells where the addition of tailed Student’s t-test). 224 H.L. Bradlow et al. 1 Molecular and Cellular Endocrinology 115 (1995) Z-225

4. Discussion acid free) are able to influence Ca2 + currents in plasma membrane of chicken granulosa cells. Caffrey et al. In this paper we show that serum albumin is able to (1979) and Condon and Pate (1981) showed that BSA is modify one of the main pathways of E, metabolism able to enhance steroidogenesis in ovine and bovine increasing 16cl-hydroxylase activity in human cancer luteal cells. Melsert et al. (1989) demonstrated that cell lines regardless of their different estrogen metabolic albumin is the active protein of rat testicular fluid conversion rates (Castagnetta et al., 1986). The concur- enhancing luteinizing hormone-stimulated pregnenolone rent decrease of 2-hydroxylation to a lesser extent was production by immature Leydig cells in vitro. Singh and also observed. This could be simply a consequence of co-workers (1992) identified albumin as the component the increase of 16cl-hydroxylase activity; however a of breast tumor cytosol able to modify the reductive direct action on 2-hydroxylase activity, given the strong pathway of E, in MCF-7 breast cancer cell line. How- reduction of 20H-compound ( > 60%) remarked in ever, only HSA among the commercial preparations MCF-7, cannot be excluded. Formation of both 16~~ tested was able to modify E, to E, conversion, BSA hydroxyestrone (16cr-HOE,) and 2-hydroxyestrone being devoid of any activity. In our study on E, hydrox- (20H-E,) may not have a negligible biological effect. ylation, all the SA preparations tested were active. As The existence of a link between increased 16a-hydroxy- shown in Fig. 4, BSA increased 16cl- and decreased lation of E, and cancer risk has been previously demon- 2-hydroxylation of E, in HEC 1A cell line. These results strated (Schneider et al., 1982; Bradlow et al., 1985; were confirmed in estrogen-responsive endometrial Osborne et al., 1988). Moreover, 16cr-OHE, has been (Ishikawa) and breast (MCF-7) cancer cells. HSA was shown to increase proliferative activity and unsched- also able to increase 16a-hydroxylation in all three cell uled DNA repair synthesis, and induce growth in soft- lines tested, suggesting that the ability of SA to modu- agar in normal mouse mammary epithelial cells (Telang late E, metabolism is not strictly species related. More- et al., 1992). On the other hand, 20H-E,, produced by over, the lack of effect of proteins not structurally 2-hydroxylation of E,, exerts anti-estrogenic and anti- related to SA suggests the absence of non-specific effect. proliferative effects (Schneider et al., 1984). Recently, it It may be argued that minor contaminants which has been demonstrated that 4-hydroxytamoxifen and co-purified with SA could be responsible for the ob- indole 3-carbinol, two compounds known to possess served modification of metabolic pathways of E,. tumor suppressing effect, increase 2-hydroxylation of E, Serum components such as globulins, enzymes, growth in vitro (Bradlow et al., 1991; Telang et al., 1994). factor, fatty acids and bilirubin have all been found in Moreover, Reed and Wiig (1983) showed that in commercial preparations of SA (Peters, 1985). To this DMBA-induced rat mammary tumors the concentra- aim, we have tested samples of SA obtained through tion of SA is higher in tumor than in muscular tissue; different procedures of purification, from different Lea et al. (1987) observed that in breast tumors, extra- sources and treated differently. Even if the hypothesis cellular proteins constitute about 50% of total cytosol cannot be definitely ruled out, it seems to be unlikely proteins, indicating the presence of a high capillary that the observed effects can be attributed to a uniden- leakage in tumor tissues. By analogy with the increase tified contaminant present in SA regardless of source of 16cr-/2-hydroxylation ratio observed in cell culture and process of purification. following the addition of SA, elevated SA levels sur- From these data it can be deduced that SA, a major rounding tumor cells might increase the presence of component of many defined media, is able to alter 16a-OH-E, and lower, at the same time, the concentra- metabolism of estradiol in several long-term cell lines. tion of 20H-E,. Previous studies established a correla- The addition to the culture medium of commercial tion between SA concentration and physiology of preparations of albumin from either bovine or human tumor tissue. Lea et al. (1987) and Soreide et al. (1991) serum, no matter what purification procedures and demonstrated the existence in malignant tumors, of an treatments used, modifies the production rate of Ez inverse correlation between estrogen (ER) and proges- metabolites. In the light of the present results, the terone (PgR) receptor levels and albumin concentra- inclusion of SA in the formulation of media for evalu- tion. This inverse relationship is in agreement with the ating steroid metabolism by cells in culture, should be poor prognosis of ER, PgR negative tumors and is carefully considered. likely to associate also with an altered 16a-/2-hydroxylation ratio. Acknowledgements There are several references in literature about the biological effect of SA. Specific receptors for SA have This study was supported in part by Associazione been identified by Brandes et al. (1982) in rat adipocytes Italiana per la Ricerca sul Cancro (AIRC) and by and by Weisiger et al. (1981) in liver cells. More re- Italian National Research Council (CNR) special pro- cently, Chiang and co-workers (1993) observed that ject ‘Aging’ contract no. 95.010.PF40. .FA is a recipient commercial preparations of BSA (fraction V and fatty of a fellowship from AIRC. H.L. Bradlow et al. / Molecular and Cellular Endocrinology I15 (1995) 221-225 225

References Osborne, M.P., Karmali, R.A. and Bradlow, H.L. (1988) Cancer Invest. 6, 6299631. Peters, T. (1985) Adv. Protein Chem. 37, 161-245. Bradford, M.M. (1976) Anal. Biochem. 72, 248-254. Reed, R.K. and Wiig, H. (1983) Stand. J. Clin. Lab. Invest. 43, Bradlow. H.L., Herscopf, R.J., Martucci, C.P. and Fishman, J. (1985) 503-512. Proc. Natl. Acad. Sci. USA 82, 6295-6299. Rohrilich, S.T. and Rifkin, D.B. (1981) J. Cell. Phys. 109, I 15. Bradlow, H.L., Michnovicz, J.J., Telang, N.T. and Osborne, M.P. Schneider, J., Kinne D., Fracchia, A.. Pierce V., Anderson, K.E., ( I99 I ) Carcinogenesis 12, 157 I- 1574. Bradlow, H.L. and Fishman, J. (1982) Proc. Natl. Acad. Sci. USA Brandes, R., Ockner. R.K., Weisinger, R.A. and Lysenko, N. (1982) 79, 3047-3051. Biochem. Biophys. Res. Commun. 105. 821-827. Schneider, J., Huuh, M.M., Bradlow, H.L. and Fishman, J. (1984) J. Caffrey, J.L., Nett, T.M., Abel., Abe, Jr., J.H. and Niswender, G.D. Biol. Chem. 259, 4840-4845. (1979) Biol. Reprod. 20, 2799287. Singh, A., Ghilchik, M.W., Patel, S.R., Blench, I., Morris, H.R. and Castagnetta, L., Granata, O.M., Lo Casto, M., Miserendino, V., Reed, M.J. (1992) Mol. Cell. Endocrinol. 83, 85-92. Calo’, M. and Carruba, G. (1986) J. Steroid Biochem. 25, 803-809. Soreide, J.A., Lea, O.A. and Kvinnsland, S. (1991) Acta Oncol. 30, Chiang, M., Strong, J.A. and Asem E.K. (1993) Mol. Cell. En- 7977802. docrinol. 94. 27-36. Telang, N.T., Suto, A., Wong, G.Y., Osborne, M.P. and Bradlow, Condon, W.A and Pate, J.L. (1981) Biol. Reprod. 20, 2799287. H.L. (1992) J. Natl. Cancer Inst. 84, 6344638. Hinegardner, R.T. (1971) Anal. Biochem. 39, 197-201. Telang, N.T., Bradlow, H.L. and Osborne, M.P. (1994) Cancer Laemmli, U.K. (1970) Nature 227, 680-685. Detect. Prevent. 18(4), 313-321. Lea, O.A., Thorsen, T. and Kvinnsland, S. (1987) Anticancer Res. 7, Weisiger, R., Gollan, J. and Ockner, R. (1981) Science 211, 1048- 1133118. 1051. Melsert, R.. Hoogerbrugge, J.W. and Rommerts, F.F.G. (1989) Mol. Wray, W., Boulikas, T., Wray. V.P. and Hancock, R. (1981) Anal. Cell. Endocrinol. 64, 35-44. Biochem. 118, 197-203.