Ligand Binding Characteristics of a ... - Bioscience Reports

Bioscience Reports, Vol. 20, No. 2, 2000

Ligand Binding Characteristics of a Glycosylphosphatidyl Inositol Membrane-Anchored HeLa Cell Folate Receptor Epitope-Related to Human Milk Folate Binding Protein Jan Holm,1,5 Steen Ingemann Hansen,2 Mimi Høier-Madsen,3 Lars Korsbaek,4 Heidi Beckmann,4 and Knud Josefsen4 Received March 6, 2000 The folate receptor (FR) in HeLa cells was characterized as to ligand binding mechanism, antigenic properties and membrane anchor in order to obtain information to be used for the design of biological agents targeting FR in malignant tumors. The receptor displayed the following binding characteristics in equilibrium dialysis experiments (37°C, pH 7.4) with [ 3H] folate: a high-affinity type of binding that exhibited positive cooperativity with a Hill coefficient H1.0 and an upward convex Scatchard plot, a slow radioligand dissociation at pH 7.4 becoming rapid at pH 3.5 and inhibition in the presence of other folates. The molecular size of the receptor was 100 kDa on gel filtration with Triton X-100, or similar to that of high molecular weight human milk folate binding protein (FBP). The latter protein represents a 25 kDa molecule which equipped with a hydrophobic glycosylphosphatidyl inositol (GPI) membrane anchor susceptible to cleavage by phosphatidylinositol specific phospholipase C (PI-PLC) forms micelles of 100 kDa size with Triton X-100. The HeLa cell FR immunoreacted with antibodies against purified human milk FBP in ELISA, and in a fluorescence activated cell sorting system, where HeLa cells exposed to increasing concentrations of antibody showed a dose-dependent response. Exposure to PI-PLC decreased the fraction of immunolabeled cells indicating a linkage of FR to cell membranes by a GPI anchor. HeLa cells incubated with radiofolate showed a continuous uptake with time, however, with a complete suppression of uptake in the presence of an excess of cold folate. Prewash of cells at acidic pH to remove endogenous folate increased the uptake. Binding and uptake of [3H] folate was increased in cells grown in a folate-deprived medium. The HeLa FR seems to be epitope related to human milk FBP. KEY WORDS: Ligand binding; glycosylphosphatidyl inositol anchor; epitope-relatedness to human milk folate binding protein.

INTRODUCTION High-affinity folate binding proteins (FBP) exist in two forms: a glycosyl phosphatidyl inositol (GPI) membrane-anchored folate receptor (FR) and a soluble FBP present in milk and other body fluids [1]. The FBP from bovine milk has been 1

Department of Clinical Chemistry, Herning Hospital, Herning, DK-7400, Denmark. Department of Clinical Chemistry, Central Hospital Hillerød, DK-3400, Denmark. 3 Laboratory for Autoimmune Serology, State Serum Institute, Copenhagen, DK-2300 S, Denmark. 4 Bartholin Institute, Municipal Hospital, Copenhagen, DK-1399, Denmark. 5 To whom correspondence should be addressed. 109 2

0144-8463y00y0400-0109$18.00y0  2000 Plenum Publishing Corporation

Holm et al.

110

characterized as to primary and secondary structure, polymerization phenomena, ionic charge and mechanism of ligand binding which exhibits positive cooperativity, and a remarkably slow dissociation of ligand at neutral pH [2–6]. The latter phenomenon is probably related to profound changes in the conformational protein structure subsequent to ligand binding as indicated by ligand-induced changes in CD-spectrum, stability and polymerization tendency [3, 4]. Two FBP with molecular weights (MW) of 27 kDa and 100 kDa on gel filtration were prepared from Triton X-100 solubilized human milk by a combination of ion exchange chromatography on CM-Sepharose CL-6B in the presence of Triton X100 and affinity chromatography on a methotrexate-AH-Sepharose 4B column desorbed with a pH-gradient [7]. Amino acid sequence analysis of the low MW human milk FBP revealed a high degree of sequence homology with bovine milk FBP [8]. Both receptor proteins had estimated MWs of approximately 30 kDa (including 3% carbohydrate) on the basis of amino acid composition and sequence homology [8]. The two FBP in human milk have identical N-terminal amino acid sequence for 39 cycles suggesting that 27 kDa FBP is a cleavage product of 100 kDa FBP [8]. Recent studies have confirmed this hypothesis in as much as phosphatidylinositol-specific phospholipase C (PI-PLC) cleaves the GPI residue inserted into Triton X-100 micelles from high MW human milk FR thereby decreasing its apparent MW from 100 to 27 kDa [9]. Both FBP in human milk exhibited complex ligand binding mechanisms similar to those of bovine milk FBP and the FR in several human tissues [2, 7, 10–12]. Cytotoxic proteins unable to permeate cell membranes can be covalently attached to folic acid and non-destructively delivered into malignant epithelial cells via folate receptor mediated endocytosis [13, 14]. These in vitro observations emphasize a current need for detailed knowledge of ligand–receptor interactions of the FR to be utilized for the design of therapeutic agents targeting the receptor in human malignancy. As yet, there are, however, few data on the exact ligand binding mechanism of folate in malignant cell lines. The purpose of the present study was to examine radioligand binding characteristics of HeLa cell FR, compare them to those of other human FR, and establish the antigenic identity of HeLa cell FR. METHODS AND MATERIALS Cell Culture HeLa cells were obtained from ATCC (Rockville, MD, U.S.A.) and grown in 90% DMEM (Gibco-BRL, Gaithersburg, MD, U.S.A.), 10% inactivated fetal calf serum (IFCS, Biological Industries Kibbutz Beit, Haemek, Israel) with one weekly passage. Folate-deprived cells were grown to confluency and the medium replaced with 90% custom made folate-free DMEM, 10% IFCS for four days. Radioligand Binding, Inhibition and Dissociation Harvested cells were solubilized in Triton X-100 (10 gyl) containing the proteinase inhibitor, phenylmethane sulphonyl fluoride, PMSF (1 mM; BDH) after

HeLa Cell Folate Receptor

111

repeated freeze–thaw cycles and centrifuged. Supernatants prepared as described below were used for radioligand studies, gel filtration, quantification by ELISA and SDS-PAGE immunoblotting. Cell supernatants were dialyzed against 0.2 M acetate buffer, pH 3.5 at 4°C to remove endogenous folate. Equilibrium dialysis experiments were performed as described previously [10] in 0.17 M TrisyHCl buffer, pH 7.4 (37°C) with the sample in the internal (1000 µl) and [3H] folate ([3H] pteroylglutamate with a specific activity of 30–47 Ciymmol from Amersham International Ltd., Amersham, U.K.) in the external solution (100 ml). Triton X-100 at a concentration of 10 gyl was added to both sides of the dialysis membrane. In inhibition experiments folate analogs were added to the external solution together with the radiolabel. The following analogs were used: pteroylglutamate (folate) and 5-formyltetrahydrofolate, both supplied by Sigma as well as methotrexate (Lederle, 4587–24) purified as previously described [15]. Samples (1000 µl) predialyzed to equilibrium (pH 7.4, 7°C) against 0.1 nM [ 3H] folate were dialyzed against 1000 ml volumes of folate-free buffer according to a previously published procedure [10]. The concentration of protein was determined by a commercial assay (Bio-Rad Protein Assay Kit 500-0001). Enzyme-Linked Immunosorbent Assay (ELISA) for HeLa Cell FR Rabbit antisera against purified human milk FBP were pooled, the immunoglobulins precipitated and employed in a previously described ELISA for human FBP [10, 16]. Gel Filtration The molecular size of HeLa cell FR was estimated by chromatography on a column (5.3 cm2B94 cm) of Ultrogel (IBF) as previously reported [10]. Triton X100 (1 gyl) was added to the elution buffer 0.17 M TrisyHCl, pH 7.4 (5°C). Samples were incubated with 10 nM [ 3H] folate for 3 hr (25°C) prior to column application. SDS-PAGE and Immunoblotting HeLa cell supernatant was analyzed by polyacrylamide gel (12%) electrophoresis in the presence of sodium dodecylsulphate (SDS-PAGE) according to Laemmli and Kyhse-Andersen [17, 18] as described in a previous report [10]. Anti-human milk FBP rabbit serum was used as primary antibody, non-immune rabbit serum as control and peroxidase-conjugated swine-anti-rabbit (total) Immunoglobulin (DAKO, Glostrup, Denmark) as secondary antibody. Immunostaining Antibodies against low molecular human milk FBP (see above) raised in rabbits were used. The cells were trypsinized and rinsed once in phosphate buffered saline PBS (pH 7.4), then incubated with the antibody in concentrations ranging from 1 :100 to 1 :10 on ice for 30 min. The cells were then rinsed twice in PBS before

Holm et al.

112

incubation for another 30 min. on ice with secondary antibody (FITC conjugated swine anti-rabbit immunoglobulin, DAKO, Glostrup, Denmark). The cells were analyzed on a FACStar Plus (Becton Dickinson, Mountain View, CA, U.S.A.) equipped with a 488 nm, 2 W argon laser and 70 µm or 100 µm nozzle tips. For each sample 5000 cells were analyzed. The staining intensity was given as percentage of cells exceeding fluorescence intensity in control experiments using only the secondary antibody. In some experiments HeLa cells were treated with phosphatidylinositol specific phospholipase C (PI-PLC) from Sigma (P8804), 50 milliunits per 105 cells at 37°C for 1 hr prior to immunolabeling. Radioligand Uptake by HeLa Cells HeLa cells (0.5B106) were seeded in tissue culture dishes (60B15, Nunc, Life Technologies, Roskilde, Denmark). The medium was removed after 2 days and cells were washed twice in sterile ice chilled PBSyHCl, pH 4.01 containing 5 mM glucose (acidic prewash) or in PBS buffer, pH 7.4 (no acidic prewash). Finally the cells were rinsed twice in ice chilled PBS. The cells were incubated in 3 ml sterile PBS, 10 nM [ 3H] folate for 60 or 120 min at either 4°C or 37°C. The supernatant was removed and stored at A20°C. The cells were washed twice in PBS on ice, removed from the tissue culture dish using a cell scraper and stored at A20°C until radioactivity measurement. RESULTS Radioligand Binding, Dissociation and Inhibition A saturation curve for radioligand ([ 3H] folate) binding to supernatants of solubilized HeLa cells is shown in Fig. 1. (Scatchard plot inserted). The binding was of a high-affinity type (see Table 1) and displayed apparent positive cooperativity as indicated by a Hill coefficient significantly greater than 1.00 and an upward convex Schatchard plot (Fig. 1 and Table 1). No dissociation of radioligand occurred at pH 7.4 even after dialysis for 24 hr, whereas dissociation was complete within 24 hr of dialysis at pH 3.5 (data not shown). The effect of other folates on radioligand (0.1 nM [ 3H] folate) binding to HeLa cell supernatant was studied in a few experiments. A 97% inhibition of radiofolate binding occurred in the presence of 1 nM folate whereas a 40% inhibition was seen with 5-formyltetrahydrofolate, a reduced folate form (data not shown). Purified solutions of methotrexate (1 nM) had no effect on binding (data not shown). Comparative [ 3H] folate binding experiments were performed with populations of HeLa cells grown in standard media and folate-depleted media. Maximum [ 3H] folate binding determined at two dilution levels (duplicate determinations) was 0.59 and 1.34 nmolyg protein, respectively. ELISA of FR in HeLa Cells The concentration of radioligand-bound FR was determined in supernatants obtained from two populations of HeLa cells, one grown in standard medium and


113

Fig. 1. High-affinity binding of [ 3H] folate to supernatant obtained from Triton X100 treated HeLa cells. Equilibrium dialysis experiments in 0.17 M Tris-HCl buffer (pH 7.4, 37°C). Scatchard plot of binding data (abscissa, bound folate nM; ordinate, boundyfree) given as inset.

one grown in folate-depleted medium. Serial dilutions of supernatants from the two populations with known concentrations of radioligand-bound FR were analyzed in ELISA for human milk FBP calibrated with serial dilutions of a reference solution containing purified human milk FBP. Figure 2 shows complete titration curves in ELISA for HeLa cell FR versus concentration of radioligand-bound FR on the abscissa. The slope of both curves is 0.4 or less than 1.00 which is the theoretical slope of the titration curve for purified human milk

1slope: concentration of radioligand-bound FR Grelative immunoreactivity2 . concentration of immunoreactive FR

Molecular Size of FR Determined by Gel Filtration The gel filtration profile of HeLa cell supernatant pre-exposed to an excess of [ 3H] folate is shown in Fig. 3. One predominant peak of both radioligand-bound and immunoreactive FR elutes at 100 kDa. Table 1. Parameters of High-Affinity Binding of [ 3H] Folate to HeLa Cell Supernatant in Equilibrium Dialysis Experiments (pH 7.4, 37°C). N (nM) HeLa cell supernatant

6.3

S0.5 (nM) 1yS0.5 (M −1) 0.082

1.22B10

10

h (S.D.) 1.52 (0.09)*

The Hill coefficient (h) is significantly higher than 1.00: pF0.001* (calculation of h based on 11 observations). Binding data were analyzed by Scatchard and Hill plots. N, maximum [ 3H] folate bound; S0.5 , the concentration of free folate at half saturation of folate binding. ‘‘Overall’’ affinity expressed as 1yS0.5 .

Holm et al.

114

Fig. 2. ELISA titration curves for immunoreactive FR in serial dilutions of supernatant from Triton X-100 treated HeLa cells using anti human milk FBP antibodies (ordinate). Concentration of radioligand-bound FR in serial dilutions of HeLa cells supernatants (see above) used for ELISA (abscissa). A solution of purified human milk FBP antigen was used as calibratoryreference for ELISA. HeLa cells grown in standard medium (●). HeLa cells grown in folate depleted medium (,). The theoretical slope of the titration curve for human milk FBP calibrator indicated: concentration of immunoreactive FBP G1.0 (relative immunoreactivity). concentration of radioligand-bound FBP

SDS-PAGE Immunoblotting Immunoblotting of HeLa cell supernatants with anti-human milk FBP antibodies revealed bands at 65–70 kDa. No bands were seen in control experiments performed with non-immune rabbit immunoglobulin (data not shown).

FACS Studies Rabbit antibodies against human milk FBP were used for immunolabeling of HeLa cells in a flow cytometric system (FACS), where cells exposed to increasing concentrations of antibody showed a dose-dependent response with FITC-conjugated swine anti-rabbit immunoglobulin as secondary antibody (Fig. 4A, B, D). Exposure to PI-PLC resulted in a significant decrease in the fraction of immunolabeled cells in the fluorescence diagram (Fig. 4B, C, D).


115

Fig. 3. Ultrogel AcA 44 chromatography of supernatant obtained from Triton X-100 treated HeLa cells incubated with [ 3H] folate prior to column application. Left ordinate is cpm in the effluent (●). Right ordinate is OD 490 in ELISA (s).

Uptake of [ 3H] Folate by HeLa Cells The time course of [ 3H] folate uptake by HeLa cells grown in folate-depleted media was studied in a few experiments (Fig. 5). Cells prewashed at acidic pH showed a significantly higher uptake than cells prewashed at near-neutral pH (normal). The uptake depended on the incubation temperature since it was much lower at 4°C than at 37°C. The presence of unlabeled folate (1 µM) in the incubation medium completely suppressed the uptake (data not shown). A few experiments were performed with cells grown in standard media. These cells exhibited a lower uptake of [ 3H] folate than cells grown in folate-deprived media (data not shown).

DISCUSSION The present study has shown that the folate receptor protein in HeLa cells, a malignant cell line, shares antigenic determinants with human milk FBP as evidenced by its ability to immunoreact with antibodies against human milk FBP in ELISA, immunoblotting and FACS. The latter observations indirectly indicate that HeLa cell FR is epitope-related to FR in a number of human tissues, e.g., choroid plexus, proximal kidney tubules and serious ovarian carcinomas, all of which are highly immunoreactive in the presence of anti human milk FBP antibodies [10–12]. Our FACS studies with immunolabeled HeLa cells provided evidence that the receptor is attached to the cell membrane by a GPI anchor since preincubation of cells in the presence of PI-PLC significantly decreased the fraction of immunolabeled cells. The molecular size of HeLa cell FR on gel filtration (100 kDa) agrees fairly well with

116

Holm et al.

Fig. 4. Fluorescence-activated cell sorting (FACS) studies of membrane-linked FR in HeLa cells immunolabeled with antibodies (rabbit immunoglobulin) against human milk FBP. A: Control without primary antibody. B: Primary antibody dilution 1 :10. C: Primary antibody dilution 1 :10 and exposure to PIPLC. D: Titration of HeLa cells with anti-human milk FBP immunoglobulin with or without PI-PLC.

this notion. By analogy to other FBPs [9] the 100 kDa peak most likely represents FR (25 kDa) equipped with a hydrophobic residue (GPI anchor) which inserts into Triton X-100 micelles and forms a large molecular size species on gel filtration. Ligand (folate) binding to HeLa cell FR is of a high-affinity type involving complex mechanisms. Those were positive cooperativity, a slow dissociation at nearneutral pH, competition for binding by folate analogs, in particular oxidized forms, and no inhibition in the presence of purified methotrexate deprived of contaminating inhibitory folates [15]. Hence, our data suggest that ligand binding to the HeLa cell FR displays characteristics similar to those of ligand binding to FR in normal and malignant human tissues [10–12]. This means that HeLa cells or other malignant cell lines could be approximate in vitro models for studies of the mechanisms involved in interaction between folate and human FR. Although few, the present data on uptake of [ 3H] folate by HeLa cells are consistent with the involvement of a high-affinity FR in the process. The uptake is


117

Fig. 5. Uptake of [ 3H] folate by HeLa cells at 4°C (circles) and 37°C (triangles) with (closed symbols) or without (open symbols) acidic prewash at pH 3.5.

thus inhibited by an excess of cold folate, and significantly increased in cells exposed to acidic prewash to remove endogenous folate as compared to control cells preincubated at near neutral pH. Another point of interest is that cells grown in folatedeprived media exhibit a higher concentration of radioligand-bound FR as well as a higher uptake of [ 3H] folate. However, as shown in Fig. 2, folate deprivation had no effect on the relative immunoreactivity of FR in the presence of antibodies against human milk FBP. ACKNOWLEDGEMENTS We appreciate the valuable technical assistance of Mette Wolf and are indebted to Jens Peter Stenvang for the performance of the FACS analysis. REFERENCES 1. Antony, A. C. (1992) Blood 79:2807–2820. 2. Hansen, S. I., Holm, J., Lyngbye, J., Pedersen, T. G., and Svendsen, I. (1983) Arch. Biochem. Biophys. 226:636–642. 3. Kaarsholm, K. C., Kolstrup, A.-M., Danielsen, S. E., Holm, J., and Hansen, S. I. (1993) Biochem. J. 292:921–925. 4. Pedersen, T. G., Svendsen, I., Hansen, S. I., Holm, J., and Lyngbye, J. (1980) Carlsberg Res. Commun. 45:161–166. 5. Svendsen, I., Martin, B., Pedersen, T. G., Hansen, S. I., Holm, J., and Lyngbye, J. (1979) Carlsberg Res. Commun. 44:89–99. 6. Svendsen, I., Hansen, S. I., Holm, J., and Lyngbye, J. (1984) Carlsberg Res. Commun. 49:123–131. 7. Hansen, S. I., Holm, J., and Lyngbye, J. (1983) in: Chemistry and Biology of Pteridines (Blair, J. A., ed.), Walter de Gruyter and Co., Berlin, pp. 729–733. 8. Svendsen, I., Hansen, S. I., Holm, J., and Lyngbye, J. (1982) Carlsberg Res. Commun. 47:371–376. 9. Hansen, S. I., and Holm, J. (1992) Biosci. Rep. 12:87–93. 10. Holm, J., Hansen, S. I., Høier-Madsen, M., and Bostad, L. (1991) Biochem. J. 280:267–271. 11. Hansen, S. I., Høier-Madsen, M., and Bostad, L. (1992) Kidney Int. 41:50–55.

118

Holm et al.

12. Holm, J., Hansen, S. I., Høier-Madsen, M., Birn, H., and Helkjær, P.-E. (1999) Arch. Biochem. Biophys. 366:183–191. 13. Leamon, C. P., and Low, P. S. (1992) J. Biol. Chem. 267:24966–24971. 14. Leamon, C. P., and Low, P. S. (1993) Biochem. J. 291:855–860. 15. Holm, J., Hansen, S. I., and Lyngbye, J. (1980) Biochem. Pharmacol. 29:3109–3112. 16. Høier-Madsen, M., Hansen, S. I., and Holm, J. (1987) Biosci. Rep. 7:553–557. 17. Laemmli, U. K. (1970) Nature 227:680–685. 18. Kyhse-Andersen, J. (1984) J. Biochem. Biophys. Methods 10:203–209.