Affinity Purification of Human Granulocyte Macrophage Colony-stimulating Factor Receptor a-Chain DEMONSTRATION OF BINDING BY PHOTOAFFINITY LABELING*

The granulocyte macrophage low affin- placenta biotinylated of the placental

The human granulocyte macrophage colony-stimulating factor (GM-CSF) receptor a-chain, a low affinity component of the receptor, was solubilized and affinity-purified from human placenta using biotinylated GM-CSF. Approximately 2 g of the placental membrane extract was subjected to a biotinylated GM-CSF-fixed streptavidin-agarose column, resulting in a single major band at 70 kDa on a silver-stained sodium dodecyl sulfate gel. The radioiodination for the purified material disclosed that the purified protein had an approximate molecular mass of 70 kDa and a p1 of 6.6. Binding activity of the purified material was demonstrated by photoaffinity labeling using HSAB-'2SI-GM-CSF, producing a similar specific band at 70-95 kDa as was demonstrated for the crude protein.
membranes (DiPersio et al., 1988, Park et al., 1989Chiba et al., 1990aChiba et al., , 1990b. A complementary DNA (cDNA) for GM-CSF receptor has recently been cloned screening a placental cDNA library (Gearing et al., 1989). However, biochemical analysis of the receptor proteins has yet been inadequate.
We have recently reported that the human GM-CSF receptor consists of at least two distinct proteins (Chiba et al., 1990b). A protein of around 80 kDa (a-chain) shows a lower affinity (& = 0.74 nM), faster dissociation, and broader expression not only on hematopoietic cells but also on various nonhematopoietic cells and tissues including placenta. In contrast, another protein of 135 kDa (P-chain) is related to higher affinity binding  and has slower dissociation and expression limited to hematopoietic cells. Thus, the placenta contains only the a-chain (Chiba et al., 1990b). The distinction between the two proteins was clearly demonstrated by the different proteolytic cleavage with the use of V-8 protease.' The cloned cDNA probably corresponds to the a-chain.
We report here (i) an assay of solubilized GM-CSF receptor a-chain utilizing polyethylene glycol precipitation, (ii) onestep affinity purification of the a-chain from placental tissues utilizing biotinylated GM-CSF, and (iii) identification of solubilized and purified receptor by photoaffinity labeling.
The human granulocyte macrophage colony-stimulating factor (GM-CSF)' is a glycoprotein active on a variety of hematopoietic cells (Wong et al., 1985;Nicola, 1989) and on nonhematopoietic cells (Bussolino et al., 1989;Baldwin et al., 1989). The characterization of the cellular receptor for GM-CSF has been developed using intact cells or intact cellular  (Bolton and Hunter, 1973) as described elsewhere (Chiba et al., 1990a) and vielded sDecific I radioactivities of 5-10 X 10' cpm/ng of protein. Preparation of photoreactive N-hydioxysuccinimidyl-4-azidobenzoate (HSAB)-'251-GM-CSF was performed as described nreviouslv (Oka et al., 1985)  as a 100,000 x g pellet was solubilized with 1.5% Triton X-100 in the same buffer (solubilizing buffer). The solubilizing mixture was then incubated for 30 min on ice and centrifuged at 100,000 X g for 60 min, and the supernatant was stored at -80 "C.
Binding of ""I-GM-CSF or HSAB-'Y'I-GM-CSF to Solubilized GM-CSF &c&or-Extracts of cellular membranes of placenta were incubated at 4 "C with various concentrations of ""I-GM-CSF or HSAB-'Y"I-GM-CSF in the binding buffer (25 mM HEPES, pH 7.4, 150 mM NaCl, 10 mM KCI, 10 mM CaC12) at a final volume of 200 ~1. Separation of receptor-bound radioactivities from free radioactivities was performed as previously described using polyethylene glycol (Yeung et al., 1986 Genzyme, Boston, MA) at a final concentration of 12.5%. The samples were mixed vigorously and allowed to stand on ice for 15 min, followed by centrifugation at 12,000 x g for 5 min at 4 "C. Phthalate oil was then layered on and centrifuged at 12,000 X g for a further 2 min at 4 "C to wash the pellets and the walls of the tubes. The radioactivity in the pellets was counted with a y-counter (Aloka ARC-300). Nonspecific binding was determined in the presence of a lOOfold molar excess of unlabeled GM-CSF.
The binding data were transposed by the method of Scatchard (Scatchard, 1949). Binding to the affinity-purified material was also carried out by the same protocol.  (1975).

Biochemical and Biological Properties
of HSAB-""I-GM-CSF-The GM-CSF with double modification by '2"I-Bolton-Hunter reagent and HSAB was examined for physiological integrity and biological activity. HSAB-""I-GM-CSF as well as ""I-GM-CSF migrated on a SDS gel as a single band at approximately 15 kDa (Fig. 1). We compared the biological activities of unlabeled and HSAB-""I-GM-CSF. The growth stimulatory effect of HSAB-""I-GM-CSF on TF-1 cells (Kitamura et al., 1989) was very close to that of unlabeled GM-CSF (data not shown) when estimated by a calorimetric assay. The binding property of HSAB-""I-GM-CSF was also tested by Scatchard analysis of its binding to U-937 cells, which has two affinity classes of receptor, yielding biphasic plots (data not shown) similar to those obtained using '251-GM-CSF without HSAB (Chiba et al., 1990a).
Assay of Solubilized GM-CSF Receptor-When we initially mixed y-globulin (final concentration, 0.05%) and PEG (final concentration, 12.5%) with I*?-GM-CSF, and centrifuged the mixture at 12,000 x g, approximately 95% radioactivity was detected in the supernatant. This predicted that the receptorbound form of GM-CSF could be separated from free GM-CSF by PEG. '*"I-GM-CSF bound to solubilized GM-CSF receptor in a specific and saturable manner (Fig. 2B). The recovery of the protein from the membranes through solubilization was 70-80%, but the recovery of GM-CSF binding capacity was more than lOO%, indicating that the GM-CSF binding activity was not decreased but rather increased through the solubilizing procedure. Triton X-100 did not affect the amount of specific binding at the concentration of 0.1-l% (data not shown). Scatchard analysis for the solubilized placental membranes was similar to that for the intact membranes (Fig. 2, A and B plot from binding data of the solubilized sample and HSAB-'*'I-GM-CSF was essentially the same as that obtained by using 1251-GM-CSF (Fig. 2, B and C).

Affinity
Labeling of Intact Placental Membranes with HSAB-125Z-GM-CSF-HSAB-1251-GM-CSF was cross-linked to intact placental membranes by irradiation with ultraviolet rays. The extent of cross-linking depended on the length of time the reaction mixtures were irradiated (Fig. 3A). and o(chain receptor. A faint radioactivity spread around 50-60 kDa and a sharp band running at 48 kDa on lane a probably represent the degradation product by ultraviolet rays, since these bands were variably observed in the cross-linking experiments using the photoaffinity protocol. CSF and the a-chain due to protein destruction by the ultraviolet light (data not shown).

Purification of GM-CSF Receptor a-Chain by NSB-GM-CSF Affinity
Chromatography-The solubilized GM-CSF receptor in the crude sample, the flow-throughs, and the eluate from the NSB-GM-CSF-streptavidin-agarose column were identified in a specific manner using HSAB-'251-GM-CSF followed by application of the irradiation protocol (Fig. 4). A marked reduction in the nonspecific bands was noted in lanes g and h (the eluate) suggesting the purification of the receptors. A band of 70-95 kDa was detected in he g and not detected in the presence of excess unlabeled GM-CSF as is shown in lane h. The purification efficiency is summarized in Table I. When the eluate from the NSB-GM-CSF-streptavidin-agarose column was analyzed by SDS-gel electrophoresis and silver staining under reducing conditions, a single major band was detected approximatley at 70 kDa (Fig. 5, lone a). After radioiodination of the purified material by the Bolton-Hunter method, the 'IO-kDa band as well as other minor lower molecular weight bands could be detected (Fig. 5, lane b). In two-dimensional electrophoresis, a single major radioactive spot was detected around 70 kDa with a p1 of 6.6 (Fig. 6). The crude lysate, the first and the second flow-throughs passing the affinity column, or the eluate were mixed with 3 nM HSAB-'*"I-GM-CSF in the presence (lanes b, d, f, and h) or the absence (lanes a, c, e, and g) of excess unlabeled GM-CSF in the bindine buffer at a final volume of 100 ~1 in a 96-well plate. After 90 rnil of incubation at 4 "C, 3 min bf irradiation was performed with ultraviolet rays. The mixtures were directly added by i/a volumes of 5 x Laemmli sample buffer, boiled for 3 min, and subjected to SDS-polyacrylamide gel electrophoresis. The position of the band which specifically disappeared in the presence of excess unlabeled GM-CSF is shown by an arrow. The samples were loaded on a gel without separating receptor-bound radioactivity from free radioactivity, which thus caused fairly high background. column a Protein amount was estimated by the relative intensities of bands in silver-stained SDS gels.
b The maximal binding was variable probably due to fine alteration of the condition of the precipitation. The value represents a mean of four individual experiments.

DISCUSSION
Previous works on interaction of human GM-CSF and its receptor have been restricted to studies with intact cells (DiPersio et al., 1988;Gearing et al., 1989;Chiba et al., 1990aChiba et al., , 1990bMiyagawa et al., 1990) or intact cellular membranes (Chiba et al., 1990b). In this paper, we first described the binding assay for the solubilized GM-CSF receptor by separating the bound form of iz51-GM-CSF from the free form by using PEG. The solubilized placental cellular membranes were successfully assayed for the GM-CSF receptor, demonstrating a single affinity class of receptor with a Kd of approximately 0.5-0.8 nM. The assay of the solubilized receptor could be used for further studies of the interaction of the ligand and receptor in a cell-free system.
Although a cross-linking technique using DSS worked well to identify the receptor in intact membranes, our initial trial using DSS in order to identify the solubilized GM-CSF receptor yielded very high backgrounds. Therefore, we developed labeling of '251-GM-CSF with a heterobifunctional, photoreactive cross-linker, HSAB, and utilized it as a tool for A 5-d aliquot of the purified and then concentrated material was applied to SDS-8% polyacrylamide gel electrophoresis and stained with silver reagent under reducing conditions (lane a). The same material was radioiodinated by the Bolton-Hunter method and directly electrophoresed with SDSd% polyacrylamide gel (lane b). Some minor bands at approximately 35-40 kDa and smear radioactivities at a lower position were probably due to trace contaminants accumulated in the labeling. Ten ~1 of the material was mixed with a lOO-~1 sample solution for isoelectric focusing containing 8.5 M urea, 2% Nonidet P-40, and 4% Ampholine (pH 4-8, Bio-Rad). The isoelectric focusing gel contained 0.4% acrylamide, 0.02% bis, 8.5 M urea, 2% Nonidet P-40, and 2% Ampholine with the same pH range. The radioactivity at the gel front probably represents again trace contaminants accumulated in the labeling or unincorporated Bolton-Hunter reagent. The arrow indicates the spot of interest. affinity labeling of the receptor. The conjugation of i2Y-GM-CSF with HSAB could be carried out without a detectable change of binding properties of is51-GM-CSF judged from a Scatchard analysis of the binding to intact U-937 cells (data not shown) and to the solubilized placental membranes (Fig.  2, B and C). Very similar cross-linked bands were detected for intact placental membranes by utilizing either HSAB-""I-GM-CSF (either by the irradiation protocol or the protocol using DSS without irradiation; see Fig. 3) or ""I-GM-CSF (by the protocol using DSS (Chiba et al., 1990b)). Although HSAB is incorporated into the amino residues of the protein and