Characterization of the human granulocyte-macrophage colony ...

THEJOURNAL OF BIOLOGICAL CHEMISTRY

Vol. 263, No. 4, Issue of February 5, pp. 1834-1841, 1988 Printed in U.S.A.

Q 1988 by The American Society for Biochemistry and Molecular Biology, Inc.

Characterization of the Human Granulocyte-Macrophage Colonystimulating Factor Receptor* (Received for publication, June 8, 1987)

John DiPersiol,Peter Billing, Susan Kaufman, Pirooz Eghtesady, Robert E. Williams#, and Judith C. Gasson From the Division of Hematology-Oncology, Department of Medicine,UCLA School of Medicine and $Departmentof Microbiology, Molecular Biology Institute, UCLA, Los Angeles, California 90024

Human granulocyte-macrophage colony-stimulating stimulates theproliferation of myeloid stem cells (1-4) as well factor (GM-CSF) is a cytokine derived from activated as enhancing differentiated functionsof mature neutrophils, T cells, endothelial cells, fibroblasts, and macrophages. such aschemotaxis and superoxide generation in response to It stimulates myeloid anderythroidprogenitorsto chemoattractant oligopeptides and C5a (5,6). Recently, GMform colonies in semisolid medium in vitro, as well as CSF hasbeen shown to induce tumoricidal activity in human enhancing multiple differentiated functions of mature monocytes and macrophages (7) as well as to enhance the neutrophils, macrophages, and eosinophils. We have examined the bindingof human GM-CSF to a variety phagocytosis of Staphylococcus aureus and antibody-dependof responsivehumancells and celllines. The most ent cell-mediated cytotoxicityof various transformed cell lines mature myelomonocytic cells, specifically human neu- by human neutrophils (8, 9). GM-CSF also primes mature eosinophils to enhance the killing of Schistosoma mansoni trophils,macrophages, and eosinophils, express the highest numbersof a single classof high affinity recep- larvae and to increase the synthesis and extracellular export tors (Kd -37 pM, 293-1000 sites/cell). HL-60 and KG- of leukotriene C4 in response to the ionophore A23187 (10). Our initial studies using radioiodinated biosynthetic (re1 cells exhibit an increase in specific binding at high concentrations of GM-CSF; computer analysis of the combinant)humanGM-CSF suggested the presence of a data is nonetheless consistent with a single class of single class of high affinity binding sites on responsive cells a Kd 4 3 PM and 20high affinity binding sites with (11).The apparentK d of 15-30 pM is in good agreement with 450 sites/cell. Dimethyl sulfoxide induces a 3-10-fold the concentrationsof GM-CSF which elicit biological activity increase in high affinity receptors expressed in HL-60 i n uitro (1-100 p ~ )Saturation . of binding occurs at approxicells, coincident with terminal neutrophilic differenmately 100 PM, the concentrationat which maximal biological tiation. Finally, binding of lZ6I-GM-CSFto fresh peeffects are seen (11).The number of high affinity binding ripheral blood cells from six patients with chronic myelogenous leukemia was analyzed. In three of six sites on responding cells is extremely low (20-100 sites/cell). cases, binding was similar to the nonsaturable bindingWalker and Burgess (12) have recently examined the speobserved with HL-60 and KG-1 cells. GM-CSF binding cific binding of radioiodinated murine GM-CSF to murine was low, or in some cases, undetectable on myeloblasts hematopoietic cells. Of interest is thatall murine cells tested obtained from eight patients with acute myelogenous express two distinct classes of GM-CSF binding sites: one leukemia. The observed affinities of the receptor for high affinity ( K d 20-60 pM; -10-80 sites/cell) and a second GM-CSF are consistent with all known biological ac- low affinity ( K d 700-1200pM; 300-500 sites/cell). crosstivities. linking studies using murine ’251-GM-CSFand homobifuncAffinity labeling of both normal neutrophils and di- tional cross-linking agents indicated that the murine GMmethyl sulfoxide-induced HL-60 cells with unglycosy- CSF receptor is a single polypeptide chain with an apparent lated lZ6I-GM-CSF yieldeda band of 98 kDa, implying molecular weight of 51,000 (12). It is not clear whether this a molecular weight of -84,000 for the human GM-CSF represents the high or low affinity binding site. The concenreceptor. tration of murine GM-CSF required for the stimulation of myeloid colony formation is 1-50 PM and probably involves binding to high affinity receptors. The activation of murine Human GM-CSF’ is a 22,000-dalton glycoprotein which neutrophils by GM-CSF has been reported to require much higher concentrations of GM-CSF (400-800 PM) and hasbeen * This work was supported by United States Public Health Service postulated toinvolve binding to low affinity receptors (13). Grants CA 40163,30388, and 32737, American Cancer Society Grant In this report,we examine the binding of human GM-CSF, JFRA 134 (to J. C. G.), and Leukemia Society of America Special Fellowship A860220. The costs of publication of this article were over a wide range of concentrations, to normal human periphdefrayed in part by the payment of page charges. This article must eral blood cells, bone marrow, acuteand chronic myeloid therefore be hereby marked “aduertisement” in accordance with 18 leukemic cells, and a number of established human myeloid U.S.C. Section 1734 solelyto indicate this fact. and non-myeloid cell lines. Scatchard analysis of the equilib$ To whom correspondence should be addressed. rium binding data isperformed to ascertain thepresence and ’The abbreviations used are: GM-CSF, granulocyte-macrophage binding affinities of GM-CSF receptors. colony-stimulating factor; MeZSO,dimethyl sulfoxide; FCS, fetal calf

-

-

serum; PBS, phosphate-buffered saline; Hepes, 4-(2-hydroxyethyl)1-piperazineethanesulfonic acid; CSF, colony-stimulating factor; CML, chronic myelogenous leukemia; AML, acute myelogenous leukemia; SDS, sodium dodecyl sulfate; PMA, 4P-phorbol 12-myristate 13-acetate; G-CSF, granulocyte colony-stimulating factor; DSS, disuccinimidyl suberate; CFU-GM, colony-forming unit granulocytemacrophage.

MATERIALS AND METHODS

Cell LinesandCultures All cell lines were maintained at 37 “C in Iscove’s modified Dulbecco’s medium (Irvine Scientific), supplemented with 10% fetal calf serum (FCS), 1%glutamine, and antibiotics. Human cell lines had

1834

CharacterizationHuman of the been generated in this department (KG-1, HL-60 subclones, SLB-I, J-LB-I) or were obtained from the ATCC or the laboratory in which they were isolated. Cell lines used include acute myelogenous leukemia cell lines KG-1 (14); GM (15); U937 (16); the human promyelocytic cell line, HL-60 (17); andits neutrophilic and eosinophilic subclones, HL-60-neut and HL-60-eo, respectively (18); giant cell tumorGCT(19); and two human T cell leukemia virus-infected lymphoid cell lines, SLB-I, J-LB-I (20,21); HSB-2; and theGM-CSF unresponsive variant of KG-1, KG-la (22). GM-CSF-producing cell lines, GCT, S-LB-I, and J-LB-I,were washed twice with PBS (Dulbecco's phosphate-buffered saline without calcium and magnesium salts) and incubated a t 37 "C in binding buffer (Iscove's modified Dulbecco's medium containing 25 mM Hepes and 2 mg/ml bovine serum albumin, pH adjusted to 7.4) to eliminate unlabeled CSF prior to binding '"I-GM-CSF. Preparation of Bone Marrow and Peripheral BloodCells Purification of Neutrophils-Peripheral blood granulocytes and bone marrow were obtained from normal donors with appropriate informed consent. Peripheral blood and bone marrow was collected in preservative-free heparin. Neutrophils were isolated by either Ficoll hypaque gradient centrifugation, followed by 3% dextran sedimentation, orby a one-stepFicoll method using mono-poly resolving medium (Flow Laboratories). Both methods yielded populations of cells with 295% mature granulocytes. Purification of Eosinophils-Purified populations of human eosinophils and neutrophils were prepared by a modification of the method of Roberts and Gallin (23). Peripheral blood from normal volunteers was collected in 50-ml tubes with 0.3% EDTA. fMet-Leu-Phe (final concentrations lo-' M) was added, and thesuspension was incubated in a 37 "C rocking water bath for 30 min. Discontinuous Percoll gradients were prepared as previously described, except that 55 and 65% Percoll solutions were utilized. Band I (layering over the top 55% Percoll layer)contained approximately 90% neutrophils and 10% lymphocytes, monocytes, and basophils. Band I1 (layering over the 65% Percoll layer) contained 90-96% eosinophils and 5-10% contaminating neutrophils. Viability of both neutrophil- and eosinophil-rich fractionswas >98%. Cells from each layer were diluted with PBS, pelleted, and resuspended in the appropriate concentrations in binding buffer. Purification of Monocytes-Monocytes were prepared by taking the lymphocyte-monocyte-rich layer after Ficoll hypaque gradient centrifugation. Cells were pelleted, washed twice with PBS, and resuspended a t 1 X 10' cells/ml in Iscove's medium with 20% fetal calf serum. Cells were allowed to adhere to plastic 250-ml flasks for 2 h a t 37 'C. The nonadherent cells were removed, and theadherent cells were washed extensively with Iscove's medium containing 20% FCS. The adherent cells were removed with a rubber policeman, washed with PBS twice a t room temperature, and resuspended in binding buffer. Preparation of monocytes in such a fashion yielded approximately 8045% monocytes, 5% lymphocytes, and 15% neutrophils. Purification of NormalHuman Bone Marrow-Normal human bone marrow was separated on Ficoll hypaque gradients. The low density lymphocyte-monocyte layer contained lymphoid-appearing cells, all measurable CFU-GM (colony-forming cells), and 2-5% contaminating granulocytes. The red cell pellet contained mature granulocytes and immature myeloid elements including promyeloc y t e ~metamyelocytes, , and bands. Purification of Patient Leukemic Cells-Patients with Philadelphia chromosome positive chronic myelogenous leukemia (CML)in chronic phase with white blood cell counts greater than 50,00O/pl were chosen as a source of CML cells. All patients were being treated with oral alkylating agents. The buffy coat was removed, and the cells were pelleted and resuspended a t 2 X lo7 cells/ml in binding buffer prior to initiating binding experiments. Fresh peripheral blood myeloblasts were obtained from patients with acute myelogenous leukemia (AML) or CML in myeloid blast crisis by allowing the blood obtained in a heparinized tube to settle a t room temperature.Greater than 95% of the buffy coat were myeloblasts. These cells were washed twice with PBS at room temperature and resuspended at 2 X lo7 cells/ml in binding buffer. Preparation of Radioiodinated Biosynthetic GM-CSF Biosynthetic (recombinant)humanGM-CSF was prepared by transfection of COS-cells (COS GM-CSF) with the full-length GMCSF cDNA clone in the 91023(B) expression vector as previously described (24). GM-CSF was purified to homogeneity from serum-

GM-CSF Receptor

1835

free medium of transiently transfected COS cells as previously described (1, 25). Purified GM-CSF expressed in Escherichia coli (unglycosylated)was kindly provided by Sandoz. These two purified GMCSFs migrated as single bands on SDS-polyacrylamide gel electrophoresis, and the protein concentration of each was estimated by amino acid analysis. One to 5 micrograms of purified GM-CSF derived from COS or E. coli were iodinated using one of three separate methods: 1) modification of the two-phase method of Tejedor and Ballesta (26) previously described (11); 2) lactoperoxidase method (27); and 3) Bolton-Hunter reagent method (28). To determine the specific activity, self-displacement curves were generated as previously described (11). Parallel curves are generated when unlabeled and iodinated GM-CSFs compete in an equal fashion for available GM-CSF receptors, implying that labeled GM-CSF has not been damaged by the iodination procedure. COS GM-CSF could be iodinated to high specific activity (19 X 10' cpm/ng) by the method of Tejedor and Ballesta (26). This preparation was used for the majority of binding experiments presented below. E. coli GM-CSF could be effectively iodinated using either the Bolton-Hunter reagent or the lactoperoxidase method, yielding parallel self-displacement curves with specific activities of 0.5-5.2 X 10' cpm/ng. Although the method of Tejedor and Ballesta could also be utilized to iodinate E. coli GM-CSF, the specific activities were so low that it could not be used for equilibrium binding experiments. COS GM-CSF iodinated by the method of Tejedor and Ballesta, and E. coli GM-CSF iodinated utilizing the Bolton-Hunter reagent or the lactoperoxidase method, were utilized in all the equilibrium binding experiments presented below. Both gave equivalent numbers and affinities of binding sites in all cell types tested. Equilibrium binding data was analyzed according to Scatchard (29). Binding of"'GM-CSF

to Cell Lines and Peripheral Blood Leukocytes

Cells used for binding were washed once and then resuspended in binding buffer. Usually 4 X lo6 cells in 400 pl of binding buffer containing various concentrations of '261-labeledGM-CSF, with or without a 50-fold excess of unlabeled GM-CSF, were incubated for 2 h at 23 "C. After incubation, the cells were resuspended and transferred onto 0.75 ml of an ice-cold mixture of 75% fetal bovine serum in binding buffer. The cells were centrifuged for 2 min in a microcentrifuge, the supernatantwas aspirated, and thecell pellets were sliced off with a razor blade for determination of their radioactivity. Specific binding was defined as theamount of binding blocked by competition with a 50-fold excess unlabeled GM-CSF. Similar results were obtained when a 200-fold excess unlabeled GM-CSF was used. Depending upon the iodination method used, specific binding to neutrophils is 60-90% of total binding at 100 pM '"I-GM-CSF. Data from binding experiments were analyzed by weighted nonlinear least squares curve fitting developed by Munson and Rodbard (30). Objective statistical criteria(F test,extra sum squares principle) were used to evaluate goodness of fit and for discriminating between models. Curves from multiple experiments were analyzed both individually and simultaneously using constrained curve fitting to obtain precision of parameter estimates. Nonspecific binding was treated as a parametersubject to error and was fitted simultaneously with other parameters. Induction of High Affinity GM-CSF Receptorson HL-60 Cells with Dimethyl Sulfoxide (Me2SO), 4~-Phorbol12-Myristate 13Acetate (PMA), and Retinoic Acid HL-60 cells (2 X lo5 cells/ml) were incubated with 1.25% MeZSO (0.23 M), 100 nM PMA, or 1 p M retinoic acid at an initial concentration of 5 X 10' cells/ml. Aliquots of the treated populations were removed at 20,60, 120, and 160 h. Cells were washed free of inducer with PBS andresuspended in binding buffer at 2 X 10' cells/ml. Zero to 90 PM '"1-GM-CSF (8.2 X 10' cpm/ng) was added to 4 X lo6 HL60 cells. Binding experiments were condwted asdescribed above. Effect of Purified RecombinantGranulocyte CSF (G-CSF)on GM-CSF Receptor Number and Affinity on Normal Human Neutrophils and HL-60 Cells Neutrophils were incubated at 2 X lo7 cells/ml in binding buffer with the diluent control, 100 nM fMet-Leu-Phe (Sigma), or either 200 p~ or 2 nM purified recombinant G-CSF (kind gift of Dr. L. Souza, Amgen Corp., Thousand Oaks, CA) for 8 h at 37 'C. The cells were washed and resuspended at 2 X lo7 cells/ml in binding buffer.

Characterization Human of the

1836

Binding of 'Z51-GM-CSFto these cells was performed as described above. HL-60 cells were incubated a t 2 x IO5cells/ml in Iscove's medium with 20% FCS for 8 days at 37 "C in the presence of diluent, 1.25% Me2S0, 200 PM or 2 nM G-CSF. The cells were then harvested, washed, and resuspended in binding buffer. Binding of 'Z61-GM-CSF to these cells was performed as previously described above.

Cross-linking'"I-GM-CSFtoNormal Human Neutrophils and Me,SO-induced HL-60 Cells: Immunoprecipitation of lZ5Z-GMCSF Receptor Complex Peripheral blood neutrophils were isolated from normal donors using mono-poly resolving medium (Flow Laboratories) as described above. Unglycosylated '261-GM-CSF (specific activity 6 X lo5 cpm/ ne) was added to 3.0 X 10' neutrophils and 3 X 10' HL-60 cells grown for 7 days in 1.25% MezSO in 1.5 mlof binding buffer to a final concentration of 100 PM. One pg of unlabeled GM-CSF was added to duplicate tubes. Both sets of tubes were incubated for 90 min at 23 "C, placed immediately on ice, and washed twice with ice-cold binding buffer and PBS. The cells were then resuspended in 1 ml of cold PBS. One hundred mM disuccinimidyl suberate(DSS) was prepared in Me2S0. DSS (final concentration of 1 mM) was added to paired tubes with and without unlabeled GM-CSF. All four samples were incubated a t 4" C for 30 min. The reaction was quenched by the addition of Tris, afterwhich the samples were washed twice with cold PBS. To reduce background nonspecific binding and uncross-linked '"I-GM-CSF, the cellular pellets were resuspended in 70 mM sodium acetate, 50 mM NaC1, pH 4.0, and allowed to dissociate for 5 min a t 0 "C. Cells were then washed twice with cold PBS, and the cellular pellets were resuspended in 2.0 ml of PBS containing 1.0% Triton, 2 mM phenylmethylsulfonyl fluoride, 10 KM pepstatin, 1 mg/ml aprotinin, 2 mM EDTA, 10 pM leupeptin (Buffer A). The cell suspensions were gently stirred at 4 'C for 20 min and centrifuged at 12,000 X g for 3 min to remove nuclei and cellular debris. The supernatantswere incubated overnight at 4 "C with 50 p1 of rabbit anti-human GM-CSF antisera generated in our laboratory (31). Twelve hours later, 500 p1 of 10% staphylococcal A protein (Pansorbin, Pharmacia LKB Biotechnology Inc.) was added to each sample and incubated for an additional 4 h a t 4 "C. The samples were centrifuged a t 12,000 X g and the pellets were washed three times with ice-cold Buffer A, suspended in 250 pl of SDS sample buffer (1.5% SDS, 60 mM TrisHCl, pH 6.8, 5% 2-mercaptoethanol, 10% glycerol), and boiled for 5 min. Aliquots of thesupernatants were applied to a 10% SDSpolyacrylamide gel (32). Autoradiographs were performed by incubating the dried gel a t -70 "C for 14 days with an intensifying screen.

GM-CSF Receptor RESULTS

Binding of lZ5I-GM-CSFto Normal Neutrophils, Eosinophils, and Monocytes-We havepreviously demonstrated high affinity binding of '251-GM-CSFto normal human neutrophils. Subsequently, human monocytes and eosinophils have been shown to be responsive to GM-CSF by increased functional activities (1-10).We were interested in comparing the number of GM-CSF receptors on these three types of peripheral blood cells as well as the presence or absence of low affinity binding sites. Table I shows results of numerous experiments characterizing high affinity binding to purified populations of peripheral bloodcells. All three cell types displayed low numbers of a single class of high affinity binding sites with a K d of approximately 40 PM. The reasons for the variability in receptor number between individual donors is presently unknown but likely relates to thephysiological state of the cells.' In spite of this variability, it is apparent from Table I that the number of GM-CSF receptors is greatest on the most differentiated myelomonocytic cells,i.e. neutrophils, monocytes, and eosinophils. A typical experiment demonstrating specific binding of '"IGM-CSF to purified normal human neutrophils as represented by equilibrium binding andScatchard analysis is shown in Fig. 1. In concordance with physiological studies demonstrating maximal biological effects of GM-CSF at -100 PM, there appears to be no increased binding of '"I-GM-CSF above approximately 150 pM and no evidence of a low affinity GM-CSF receptors on normal neutrophils when as much as 8 nM '251-GM-CSF wasadded to purified human neutrophils. In this particular experiment, normal neutrophils expressed 1182 receptors, with a dissociation constant of 82 PM. In light of the marked variability in the number of high affinity GM-CSF binding sites expressed on the surface of neutrophils from differing donors, efforts were madeto isolate highly enriched populations of neutrophils and eosinophils from the same individual, so that a valid comparison of the J. DiPersio, P. Billing, S. Kaufman, P. Eghtesady, R. E. Williams, and J. C. Gasson, unpublished observations.

TABLEI Comparison of the number of human GM-CSFreceptors present on various cell types Cell type

Number of experiments

Number of binding sites (mean")

Kd

(Mean)

A. GM-CSF-resuonsivecells

Mature peripheralblood cells Normal neutrophils Normal eosinophils Normal monocytes Bone marrow Bone marrow lymphocyte-monocyte layer Bone marrow Ficoll pellet Fresh leukemic cells and leukemic cell lines CML (peripheral buffy coat) AML blasts HL-60 KG1 HL-60 eosinophilic clone HL-60 neutrophilic clone GM (32-62) u937

22 6 3

542 (293-982) 772 (207-1007) 450 (250-650)

3 3

111 (87-336)

45 (40-50) 45 (30-60)

6 8 7 7 3 3 1 2

131 (111-360) 9 (0-40) 322 (114-369) 102 (20-142) 41 (28-54) 34 (22-95) 8 38

45 (15-90) 39 (10-68) 33 (22-61) 24 (20-82) 28 (18-38) 30 (20-42) 12 20

35 (10-65)

B. Cells without known response to GM-CSF

KG-la GCT SLB-I JLB-I HSB-2 (I

Pm 51 (12-100) 23 (15-50) 37 (30-44)

Number in parentheses indicates the range of values.

1837

Characterization of the H u m a n GM-CSF Receptor

n. u)

"

0.2

..

0.40.8 0.6

10.0

125 I-GM-CSF (nM)

FIG. 2. Binding of '"I-GM-CSF to HL-60 ( o " 0 ) and KGcells. '251-GM-CSFat concentrations shown on the abscissa was added to 2 X lo6 HL-60 and 4 X lo6 KG-1 cells in the presence or absence of a 50-fold excess unlabeled GM-CSF. Binding was performed as described under "Materialsand Methods." Specific binding was defined as the amount of binding competed by a 50-fold excess of unlabeled GM-CSF. 1 )-(

GM-CSF CONCENTRATION ( nM)

FIG. 1. Binding of 'auI-GM-CSFto normal human neutrophils. Increasing amounts of "'1-GM-CSF (specific activity, 7.6 X 10' cpm/ng) were added to 4 X lo6 normal humanneutrophils. Concentration of the 'Z51-GM-CSFadded is shown as the abscissa. Specific binding is shown on the ordinate and represents total bound radioactivity minus that bound in the presence of a 50-fold excess of unlabeled GM-CSF. Scatchard analysis of the equilibrium binding data is represented inthe inset.

TABLE II Comparison of the number of high affinity GM-CSF receptors expressed on purified neutrophilsand eosinophils from the same donor High affinity

Experiment Donor

sites/cell

Kd

PM

1

Neutrophils Eosinophils

67 207

10 17

2

Neutrophils Eosinophils

124 664

12 16

3

Neutrophils Eosinophils

100 211

50 50

w w

.I*

0.12

a

$! number of receptors on neutrophils and eosinophils could be made. The number and classof receptors on neutrophils and eosinophils from three normal individual patients are shown in Table 11. Neutrophils and eosinophils were separated by 700 1400 density gradient centrifugation after incubation with M e t CSF RECEPTORS (sitedcall) Leu-Phe. We have shown that preincubation of neutrophils with Met-Leu-Phe down-regulates the expression of the GMFIG. 3. Equilibrium bindingof 'a61-GM-CSFbinding to KGCSF receptor (Fig. 5).3 Thus, the number of GM-CSF recep- 1 ( A ) , HL-60 ( B ) , andneutrophils (C)as analyzedbythe tors expressed on neutrophils listed in Table I1 is spuriously method of Scatchard (29). Lines of best fit were constructed low. The possible effect of fMet-Leu-Phe on GM-CSFrecep- according to the Ligand program (30). PMN, polymorphonuclear. tor expression by eosinophils is unknown. We were not able to separate pure populationsof eosinophils without utilizing altered bindingof lZ61-GM-CSFto KG-1 and HL-60 cells, the Met-Leu-Phe and gradient centrifugation. Thus, the estidata from 14 separate experiments statistically best fit the mate of the numberof GM-CSF receptors oneosinophils may "single site model" suggesting the presence of a single high also be low. 30 pM) (p =