Characterization of the cell surface receptor for a multi-lineage colony-stimulating factor (CSF-2 alpha).

125I-Labeled colony-stimulating factor (CSF) 2 alpha (interleukin 3, multi-CSF, and mast cell growth factor) was used to characterize receptors specific for this lymphokine on the cell surface of the factor-dependent cell line FDC-P2. CSF-2 alpha binding to these cells was specific and saturable. Among a panel of lymphokines and growth factors, only unlabeled CSF-2 alpha was able to compete for the binding of 125I-labeled CSF-2 alpha to cells. Equilibrium binding studies revealed that CSF-2 alpha bound to 434 +/- 281 receptors/cell with a Ka of 8.7 +/- 3.9 X 10(9) M-1. Affinity cross-linking experiments with the homobifunctional cross-linking reagents disuccinimidyl suberate, disuccinimidyl tartrate, and dithiobis(succinimidyl propionate) produced a radiolabeled band of Mr = 97,000 on intact cells and in purified cell membranes, while an additional band of Mr = 138,000 was produced upon cross-linking to intact cells only. The relationship between these two bands is discussed. The results indicate that the receptor for CSF-2 alpha on FDC-P2 cells consists at a minimum of a subunit of Mr = 72,500.

In order to more completely understand the mechanisms behind CSF-2a action we have 1251 radiolabeled CSF-2a and used this reagent to characterize receptors specific for this lymphokine on the surface of FDC-P2 cells. Nuclear. Nerve growth factor, fibroblast growth factor, platelet-derived growth factor, and epidermal growth factor were obtained from Bethesda Research Labs. Human follicle-stimulat~ng hormone, human luteinizing hormone, human thyroid-stimulating hormone, human growth hormone, and bovine insulin were obtained from Calbiochem-Behring.
Human recombinant IL-2 was expressed in, and purified from Escherichia coli and provided via a collaborative research agreement between Immunex Corp. and Hoffman-LaRoche. Mouse GM-CSF and CSF-2a were cloned from a cDNA library prepared from RNA extracted from LBR~-33-5A4 cells? The nucleotide sequences of these two cDNA clones were essentially identical to those reported previously (7,9,17). GM-CSF and CSF-2a were produced in a yeast expression system that used the prepro CY factor promoter and leader sequence (18) to direct secretion of the mature forms of the two factors? Using reversed phase high performance liquid chromatography GM-CSF was purified to homogeneity and CSF-2a partially purified from yeast-conditioned m e d i~m .~ Human IL-1 was purified to homogeneity as previously described (19) from media conditioned by activated human macrophages.
CeZi Culture-The factor-dependent cell lines FDC-PZ (13), a CSF-2a-dependent mouse bone marrow-derived line kindly provided by T.

CSF-Zo!
Purifkatwn-CSF-Za was purified to homogeneity from medium conditioned by ph~ohemagglutinin-stimulated LBRM-33-5A4 cells as previously described (2) and summarized below. Protein was first sequentially precipitated from conditioned medium by the stepwise addition of ammonium sulfate to 30, 50, and finally to 80% saturation. The proteins contained in the 80% precipitate were then fractionated by cation exchange chromatography followed by anion exchange chromatography. Purification to homogeneity was achieved by reversed phase high performance liquid chromatography; first, on a C , pBondapak column (30 cm X 3.9 mm) equilibrated in 0.1% D. Cosman, unpublished results. V. Price, unpublished results. D. Urdal, unpublished results.

205
trifluoroacetic acid and eluted with a gradient of acetonitrile, and second on the same column but with a gradient of 1-propanol in 0.9 M acetic acid, 0.2 M pyridine (2). A third high performance liquid chromatography step was used to exchange the 1-propanol solvent for the acetonitrile solvent such that the final CSF-2a sample was contained in 0.1 % trifluoroacetic acid, 48% acetonitrile.
CSF-2a activity was measured by its ability to sustain the proliferation of FDC-P2 cells (13), a factor-dependent cell line isolated from mouse bone marrow. FDC-P2 cells die in the absence of CSF-2a. Units of CSF-2a activity were determined as the reciprocal dilution of a sample which generated 50% of maximal FDCP2 [' HI thymidine incorporation as compared to a laboratory standard (WEHI-3b cell line conditioned medium). For example, if a sample generated 50% of maximal FDC-P2 ~' H l t h~i d i n e i n c o~o r a t i o n at a dilution of l:lO, 1/10 of 100 pi (assay volume) or 10 pl was said to contain 1 unit. The sample would therefore contain loo0 + 10 or 100 units of CSF-2a activity/ml. One pg of CSF-Sa corresponds to 1 X io7 units of activity (2).
Iodination of CSF-2a-CSF-2a was radiolabeled using the Enzymobead radioiodination reagent (Bio-Rad) essentially following the manufacturers specifications. Aliquots (2 X lo6 units in 25 pl) of CSF-2a in acetonitrile and trifluoroacetic acid were combined with 50 pl of 0.2 M sodium phosphate, pH 7.2, and the acetonitrile was evaporated under nitrogen. Fifty p1 of Enzymobead reagent, 20 pl of '%I (2 mCi), and 10 p1 of 2.5% @-D-glUCOSe were added and the mixture incubated at 25 "C for 10 min. Sodium azide (20 pl of 25 mM) and sodium metabisulfite (10 pl of 5 mg/ml) were then added sequentially. After 5 min at 25 "C, iodinated CSF-Sa was separated from free lZ5I by chromatography on a 2-ml Sephadex G-25 column equilibrated in 0.05 M sodium phosphate, pH 7.2, containing 0.01% gelatin. Fractions containing CSF-2a were pooled, bovine serum albumin (0.01% final concentration) and sodium azide (0.02% final concentration) were added, and the pool stored at 4 "C. Bioactivity of '"I-CSF-2a was determined in the FDC-P2 proliferation assay described above. Labeled CSF-2a preparations were analyzed by gel filtration chromatography on Sephadex (3-75 (Pharmacia) to test for the presence of high molecular weight aggregates.
Assay for Binding of 12SI-CSF-2a to Intact Cells-Binding assays were performed by a phthalate oil separation method described previously (21). Cells (2-10 X lo6) were incubated with '"I-CSF-2a in 150 p1 of RPMI-1640 containing 2% bovine serum albumin, 20 mM Hepes buffer, and 0.2% sodium azide, pH 7.2 (binding medium). Incubations were carried out in 96-well microtitre plates maintained on a mini-orbital shaker (Bellco) at 37 -C for 2 h. As previously shown by Palaszynski and Ihle (22), we found this time and temperature to be optimal for achievement of binding equilibrium. Replicate 70-#1 aliquots of incubation mixtures were then transfered to precooled 400-pi polyethyiene centrifuge tubes containing 200 pl of a phthalate oil mixture (1.5 parts dibutylphthalate, 1 part bis(2-ethylhexyl)phthalate), and cells plus bound 'zSI-CSF-2a separated from unbound 1261-CSF-2a by centrifugation for 1.5 min in an Eppendorf Microfuge. Nonspecific binding of '251-CSF-2a was measured in the presence of a 50-fold or greater molar excess of unlabeled CSF-2a. Sodium azide (0.2%) was included in all binding assays to inhibit internalization and degradation of 1261-CSF-2a by cells at 37 "C. To verify that no degradation of ligand was occurring, aliquots were removed from incubation mixtures maintained at 37 "C for up to 3 h and precipitated on Whatman 3MM paper with 10% trichloroacetic acid. No decrease in trichloroacetic acid-precipitable counts was detected over this time period.
Preparation of Plasma Membranes-Crude plasma membrane fractions from FDC-P2 cells were prepared by the basic procedure of Takacs and Staehelin (23) with some modifications. FDC-P2 cells (2-15 X lo8) were washed twice in 0.02 M Hepes, containing 0.85% NaCl and 0.5 mM MgC12, pH 7.5 (lysing medium), and then suspended to a concentration of 8 X 107/ml in the same buffer. For some preparations the lysing medium also contained a mixture of protease inhibitors with the following final concentrations: 2 mM PMSF, 10 pM pepstatin, 10 pM leupeptin, and 2 mM 0-phenanthroline. The cell suspension was then mixed with an equal volume of 0.5 M sucrose and subjected to nitrogen cavitation at 1,400 p.s.i. for 30 min at 4 "C. The resulting cell homogenate was brought to an EDTA concentration of 1 mM and centrifuged at 700 X g for 10 min at 4 "C to pellet nuclei and cell debris. The supernatant was then centrifuged at 30,000 X g for 30 min at 4 "C and the crude membrane pellet washed and resuspended in PBS and stored in aliquots at -70 "C. Membrane protein was determined by the method of Lowry (24).
Cross-linking to Intact Cells-FDC-P2 cells (2-10 X 10') were incubated with '"I-CSF-2a (3-6 X lo-'' M) at 37 'C in 150 pl of binding medium, both in the presence and absence of a 50-fold or greater molar excess of unlabeled CSF-2a. After 2 h the cells were harvested and washed three times in PBS by centrifugation at 200 X g for 10 min at 25 "C. The final cell pellets were resuspended in 150 pl of PBS and chilled on ice. Aliquots of DSS, DSP, or DST dissolved in dimethyl sulfoxide were then added such that the concentration of dimethyl sulfoxide in the incubation did not exceed 2%. The final concentration of cross-linker was in most cases 1 mg/ml. Control incubations were treated in an identical manner except no crosslinker was added. After 1 h at 0 "6, the cells were harvested, washed three times in PBS, and resuspended in 100 pl of PBS/l% Triton containing 2 mM PMSF. The incubations were maintained at 25 "C for 30 min, centrifuged at 12,000 X g for 10 min, and the supematants retained. For analysis by SDS-polyac~lamide gel electrophoresis, supernatant aliquota were dried under vacuum after the addition of 2 p1 of 10% SDS. The dried residue was dissolved in 40 pl of sample buffer (0.06 M Tris-HC1, pH 6.8,2% SDS, 10% glycerol) both in the presence and absence of 5% 2-mercaptoethanol, boiled for 3 min, and subjected to electrophoresis as described below.
Cross-linking to Plasma Membranes-Plasma membranes (100 pg) were incubated with '251-CSF-2a (3-6 X lo-'' M) at 37 "C in 100 pl of binding medium both in the presence and absence of a 50-fold or greater molar excess of unlabeled CSF-2a. After 2 h, the membranes were harvested and washed three times in PBS by centrifugation at 30,000 X g for 15 min at 4 "C. The final membrane pellets were then resuspended in 150 pl of PBS and chilled on ice. Aliquots of DSS, DSP, or DST dissolved in dimethyl sulfoxide were then added such that the concentration of dimethyl sulfoxide in the incubation did not exceed 2%. The final concentration of cross-linker was in most cases 1 mg/ml. Control incubations were treated in an identical manner except no cross-linker was added. After 1 h at 0 "C the membranes were harvested and washed three times in PBS by centrifugation at 30,000 X g for 15 min at 4 "C. Membrane pellets were then dissolved in 40 p1 of sample buffer both in the presence and absence of 5% 2-mercaptoethanol, boiled for 3 min, and then subjected to electrophoresis as described below. SDS-Polyacrylamide Gel Electrophoresis-Samples were prepared as described above and then subjected to electrophoresis on either 8% or linear 5-15% or 10-20% gradient gels according to the stacking gel procedure of Laemmli (25). After electrophoresis, gels were either stained with Coomassie Blue (0.25% in 25% isopropyl alcohol, 10% acetic acid) or treated with ENHANCE (New England Nuclear) according to the manufacturers specifications and then dried. Dried gels were exposed to Kodak X-Omat AR film at -70 "C. Fluorographs were scanned on a Hoefer GS30O scanning densitometer (Hoefer Scientific Instruments) and peak areas integrated by a Hewlett-Packard 3390A integrator.
Data Adysis-Curve fitting of binding and kinetic data was done using RS/l (Bolt, Beranek, and Newman, Boston, MA), a commercially available data processing package running on a VAX 111750 under the VMS operating system. Binding data were analyzed using an equation describing simple bimolecular binding, and inhibition data were analyzed with an equation for competitive inhibition between two ligands for one type of site (26).

RESULTS AND DISCUSSION
CSF-2a, a colony-stimulating factor which is similar if not identical to IL-3 and mast cell growth factor, was radiolabeled with lZ5I and used to characterize the specific receptors for this lymphokine on the surface of FDC-P2 cells. Fig. 1 illustrates an autoradiograph of a typical iodinated CSF-Pa preparation, where the major species after iodination had an apparent molecular weight of 24,500 on analysis by SDSpolyacrylamide gel electrophoresis (SDS-PAGE). Based on a specific activity for CSF-2a of 10 x lo6 units/Ng protein (2), the radiolabeled preparations had estimated specific activities in the range of 1 x 10l6 cpm/mmol, and when tested in a cell proliferation assay, radiolabeled CSF-2a preparations were routinely found to retain >75% of their biological activity. Preparations of lZ5I-CSF-2a were stable for at least 1 month when stored at 4 "C in 0.05 M NaP04, pH 7.2, containing 0.01% bovine serum albumin and 0.02% sodium azide, and X 10l6 cpm/mmol) was boiled for 3 min in sample buffer containing 2% SDS and 5% 2-mercaptoethanol, and 50,000 cpm were applied to a linear 10-20% gradient gel. Electrophoresis and autoradiography were then conducted as described under "Experimental Procedures." specific and nonspecific binding at both 37 and 4 "C were comparable to those reported (22; data not shown).
CSF-2a iodinated by the Enzymobead method was found to retain full binding affinity. Fig. 3 represents the ability of increasing concentrations of unlabeled CSF-2a to block the binding of l2'1-CSF-2a to FDC-P2 cells. The inhibition constant ( K I ) calculated from this curve (7.80 X lo9 f 1.61 X lo9 M-') was not significantly different frm the K,, of '2'I-CSF-2a (8.7 x 109 f 3.9 x lo9 M-').
A number of continuous cell lines of mouse, rat, or human origin were examined for their ability to bind '"I-CSF-2a. As shown in Table I,  exhibited no changes in binding characteristics over this time period. In addition, gel filtration chromatography showed no evidence of formation of high molecular weight aggregates upon storage.
12'I-CSF-2a was shown to exhibit specific binding to FDC-P2 cells that was saturable with increasing concentrations of radioligand. Fig. 2 illustrates typical equilibrium binding data for 12'I-CSF-2a, where Scatchard analysis (27) of the data yielded a straight line, indicating a single class of binding sites for CSF-2a. Nonspecific binding increased linearly with increasing concentration and did not exceed 1% of the total counts/min added. From eight binding experiments using four different radiolabeled preparations, the calculated apparent K, was 8.7 f 3.9 X lo9 M" with 434 * 281 specific binding sites/cell. This compares to a K, of approximately 5 X 10'" M" with 1500-2500 specific binding sites/cell for '251-IL-3 binding to FDC-P1 cells, as reported by Palaszynski and Ihle (22). Although this small difference could be attributable to many causes, the simplest possibility is the dissimilarity between the FDC-P1 and FDC-P2 cell lines, which are known to vary in expression of some surface antigens (28). Similar to the data reported by these investigators, we have found the time required to reach binding equilibrium of '2sI-CSF-2a a t 37 "C to be approximately 120 min, with a very slow subsequent dissociation rate. In addition, the relative levels of  (22), is not known. We also failed to find "'I-CSF-2a binding to the macrophage line P388D1, which these investigators reported to bind lZ5I-IL-3 a t low levels. Considering that we were able to detect binding to the macrophage tumor cell lines 5774 and PU5-1.8, while these investigators detected no binding to another macrophage line they tested (RAW 264.7), it would seem that at least some, but not all, macrophage tumor lines are capable of expressing the CSF-2a receptor. Of the cell lines tested, FDC-P2 exhibited the highest number of binding sites and was therefore used in further experiments designed to examine the physical characteristics of the CSF-2a receptor.
The specificity of '''I-CSF-2a binding was examined by testing a number of purified lymphokines and other polypeptide hormones for their ability to compete with 'Z51-CSF-2a for binding to its receptor on FDC-P2 cells. As shown in Fig.  4, natural CSF-20 eliminated >50% of lZ5I-CSF-2a binding when present in 10-fold excess (column d ) with up to 80% competition a t higher concentrations (columns b and c). Similarly, a 20-fold excess of recombinant CSF-2a eliminated 80% of 12sII-CSF-2a binding (column e). None of the other lymphokines or hormones tested, including GM-CSF, exhibited any ability to compete with Iz5I-CSF-2a binding to FDC-P2, even when present a t concentrations that were 500-1000fold greater (on a molar basis) than that of Iz5I-CSF-2a (columns f-q). Further evidence of this specificity is shown in Aliquots corresponding to 1.5 X lo6 cells were dried under vacuum, dissolved in sample buffer containing 2% SDS and 5% 2-mercaptoethanol, boiled for 3 min, and subjected to SDS-polyacrylamide gel electrophoresis on a linear 5-15% gradient gel.
to cells. Other workers have suggested that GM-CSF and CSF-2a may interact with the same receptor (17). The results reported here would suggest that this is not the case. Indeed, we have recently examined the binding of lZ5I-labeled GM-CSF to cells and found that GM-CSF does not bind to FDC-P2 cells and, on cells that bind both factors, no evidence of receptor cross-reactivity is detected.' These observations do not rule out, however, the possibility that on cells expressing both receptors, the binding of one ligand will result in the down-regulation of receptors for the second ligand.
The CSF-2a receptor is present on FDC-PZ cells in low abundance. One means of characterizing such recc ptors is by affinity labeling (29). Radiolabeled CSF-2a was bound to cells as described above. Then, the cells were washed and the bifunctional cross-linking reagents DSS, DSP, or DST (1 mg/ ml) were added in order to covalently cross-link the Iz5I-CSF-2a to its plasma membrane receptor. Cells treated in such a manner were then extracted with PBS/1% Triton and the soluble fraction analyzed by SDS-PAGE (Fig. 6). Under both reducing (panel A ) and nonreducing (panel B ) conditions two major cross-linked species were evident with M, of approximately 138,000 and 97,000. In addition, some cross-linked material of very high molecular weight was collected at the top of the separating gel. Both the 138,000 and 97,000 species were present whether cross-linking was done with DSS ( lanes  c and d), DSP (lanes e and f ) , or DST (lanes g and h). The results of densitometric analysis of 10 different cross-linking experiments are summarized in Table I1 and suggest that the two bands were present in approximately equal amounts. Controls showed no cross-linked species were found in the absence of cross-linker (lanes a and b ) or in samples containing excess unlabeled CSF-2a (lanes 6, d, f, and h). With increasing concentration of cross-linker, there was, however, an increase in the abundance of the very high molecular weight material present at the top of the separating gel suggesting that this material probably represents aggregates due to the high concentration of cross-linker. The nature and relationship of the two lower cross-linked species was not readily apparent from these experiments. Further experiments showed that identical results were obtained if lZ5I-CSF-2a binding to FDC-P2 cells was done at 4 "C rather than 37 "C. In addition, incubation of lZ5I-CSF-2a bound FDC-P2 cells in 25 mM EDTA for 30 min at 4 "C prior to cross-linking with DSS did not effect the appearance of either of the crosslinked species, suggesting that neither of the cross-linked species was a loosely bound peripheral membrane protein.
The possibility that the 97,000 M , species was a proteolytic cleavage product of the 138,000 M, species was addressed by including in the extraction buffer a mixture of protease inhibitors which contained PMSF (2 mM), pepstatin A (10 PM), leupeptin (10 PM), 0-phenanthroline (2 m~) , and EGTA (2   mM). Both the 138,000 and 97,000 M, species were found to be present in similar amounts under all conditions examined.
In addition to the cross-linking of '251-CSF-2a to its receptor on intact FDC-P2 cells, we also examined cross-linking to FDC-P2 plasma membrane preparations. Conditions for 1251-CSF-2a binding and cross-linking to plasma membranes were identical to those used in experiments with intact cells, except that following the cross-linking step, washed membranes were dissolved directly in SDS containing sample buffer rather than extracted with PBS/1% Triton. Membranes were then harvested, washed, and cross-linked for 1 h at 4 "C with either DSS or DST (1 mg/ml) as described under "Experimental Procedures." Membrane pellets were dissolved in sample buffer containing 2% SDS and 5% 2-mercaptoethanol, boiled for 3 min, and subjected to electrophoresis on an 8% polyacrylamide gel. Lanes a and b, 1 mg/ml DSS; c and d, 1 mg/ml DST, e and f, no cross-linker. Lanes b, d, and f correspond to samples which contained unlabeled CSF-2a during incubation of 1251-CSF-2a with plasma membranes. CSF- 2a (lunes b, d, and f). Identical results were obtained when samples were run on SDS-PAGE under nonreducing conditions. Furthermore, when plasma membranes were prepared from FDC-P2 cells in the presence of the protease inhibitor mixture described above and cross-linking experiments performed, only the 97,000 M , species was detected. This band most likely represents a plasma membrane protein of M , 72,500 that is covalently linked to '251-CSF-2~ (Mr = 24,500). The possibility that the 97,000 band is actually a proteolytic cleavage product of the 138,000 M, band is possible but unlikely. A panel of protease inhibitors had no effect on the appearance of either species in whole cells or plasma membrane preparations. This would then suggest that the 138,000 M, band represents either the M, = 97,000 receptor complex coupled to a third protein of M, = 41,000 or a protein of M , = 113,000 that is directly coupled to CSF-2a. While we cannot yet distinguish between these two possibilities, it is apparent that during the preparation of plasma membranes, the positioning of this putative protein(s), which allows it to be either directly or indirectly cross-linked to CSF-2a, is disrupted. Whether this protein might be part of a specific CSF-2a receptor complex, or is simply a closely associated membrane or cytoskeletal protein remains to be determined.

CSF-BCX
Using CSF-201 radiolabeled to high specific activity with lZ5I, we have investigated the nature of the receptor for CSF-2a on FDC-P2 cells through the use of bifunctional crosslinking reagents. The data detailed herein establish the existence of CSF-2a receptors on FDC-P2 cells and characterize for the first time the molecular nature of the receptor.